From d6e8f17a55d3de7f210789c8331843f04f5e665f Mon Sep 17 00:00:00 2001
From: Wouter Wakker <wouter.wakker@outlook.com>
Date: Wed, 3 Jun 2026 22:40:36 +0200
Subject: [PATCH 01/15] Update OECD links

---
 test.py | 0
 1 file changed, 0 insertions(+), 0 deletions(-)
 create mode 100644 test.py

diff --git a/test.py b/test.py
new file mode 100644
index 0000000..e69de29

From a5eeb172a85cda5b49ffd758a9ebfe8f4907a3fb Mon Sep 17 00:00:00 2001
From: Wouter Wakker <wouter.wakker@outlook.com>
Date: Wed, 3 Jun 2026 22:41:00 +0200
Subject: [PATCH 02/15] Update OECD links

---
 .DS_Store                 | Bin 0 -> 6148 bytes
 .gitignore                |   2 +-
 .vscode/settings.json     |   4 ++
 CLAUDE.md                 | 100 ++++++++++++++++++++++++++
 iopy/.DS_Store            | Bin 0 -> 6148 bytes
 iopy/core/config.py       | 143 ++++++++++++++++----------------------
 iopy/core/exiobase.py     |   9 +--
 iopy/core/figaro.py       |   9 +--
 iopy/core/oecd.py         |  19 +++--
 iopy/core/utils.py        |  19 +++++
 iopy/dev-requirements.txt |   1 +
 iopy/requirements.txt     |   1 +
 test.py                   |   3 +
 13 files changed, 203 insertions(+), 107 deletions(-)
 create mode 100644 .DS_Store
 create mode 100644 .vscode/settings.json
 create mode 100644 CLAUDE.md
 create mode 100644 iopy/.DS_Store

diff --git a/.DS_Store b/.DS_Store
new file mode 100644
index 0000000000000000000000000000000000000000..e393e44e60e428aaa0bb5c3151f1dbb932e86b57
GIT binary patch
literal 6148
zcmeHKL2uJA6n<_4O&ODVVA3u~k+_c1G7TYd>AF^95(8;LZ~#=ArEC#dTs7$`R8`6u
z{sVu3D}Mz33n%!V?T)71u7HqVWq;1|_ni21@>~;<Xph4tQH_XP6h`L;s!NRfxvW^p
zc-DYIjj>N%dP2`=pXw`Y>#zz~1^zV!_}jfnd-Rxg=_$V3-{0>cbq->A9K<pJ2MI!a
z0@i;jQ`2xLB1uE~P~>5xD|o>q9h&jJ2E#Z?vqs~)D3zUS*DG$tU31^~r*h(F{cN1J
z`=eKU^<2t0n78}Ei)b+E)o$&{B<n}XV59<~Z~)5dmr)YRNn4JSFjcX>?r^JawO8Am
zPVet*Z+Q=!vn_AhY&3Szey}~8Ro#s{_Z}Vf-p6N&{HSI|0{KkYy1_ZH1{d=V)8jb6
z^Ks{e^V}aLu}n_j<3ZNvA|WLX=mfvJ@Fet(<dSS(z+RNMf>e5drRic}eCih2;>QxZ
zN=IPsQil$KbvmQ47*XOUS{%PjBRo-BWQp5LESFNSG8$u{>#K`h`nJTcfQwO4U&2<e
z#^;4s2*K9=mICG-cR2@GvkF)RtOAP)aDMPm7+VH&jVg7ZQbz!w1Gh4?`7FUPp23#E
zTqAm5LWcr%s4zzip~K<#46kJ{*Qmoun8SxKcNXS^BGlb6zNhFUS{iL_6|f2v6)5Y=
z8t?zztIz+%AbVvMunPQF3J9m;ciNbe*;`j8$9t`Z@)m`SdAUXvg34^is=-_FCW<oj
ZdE5ZD4CWf)f!RL-N(Nh51^%c4zW^arvjYGC

literal 0
HcmV?d00001

diff --git a/.gitignore b/.gitignore
index bf2e2f2..f58a6e3 100644
--- a/.gitignore
+++ b/.gitignore
@@ -9,4 +9,4 @@
 .pip/
 **/__pycache__/
 **/.temp_data/
-
+test.py
diff --git a/.vscode/settings.json b/.vscode/settings.json
new file mode 100644
index 0000000..4b5a294
--- /dev/null
+++ b/.vscode/settings.json
@@ -0,0 +1,4 @@
+{
+    "python-envs.defaultEnvManager": "ms-python.python:conda",
+    "python-envs.defaultPackageManager": "ms-python.python:conda"
+}
\ No newline at end of file
diff --git a/CLAUDE.md b/CLAUDE.md
new file mode 100644
index 0000000..3b047e2
--- /dev/null
+++ b/CLAUDE.md
@@ -0,0 +1,100 @@
+# CLAUDE.md
+
+This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
+
+## What this is
+
+`iopy` is a Python package for loading inter-country input-output (IO) data from three
+public databases (OECD ICIO, Eurostat Figaro, EXIOBASE) and running economic analyses on
+them — primarily Leontief demand shocks and Ghosh supply shocks. See [README.md](README.md)
+for the economic theory (Leontief/Ghosh derivations) and the full public API table.
+
+## Commands
+
+```bash
+# Install for development (editable)
+pip install -e .
+pip install -r iopy/dev-requirements.txt   # runtime deps + pytest, pytest-cov
+
+# Run the full test suite
+pytest iopy/tests
+
+# Run a single test file / test
+pytest iopy/tests/test_oecd.py
+pytest iopy/tests/test_oecd.py::TestOECD::test_leontief
+
+# Coverage (config in iopy/tests/.coveragerc)
+pytest --cov=iopy iopy/tests
+```
+
+Note: the tests **download real data over the network** (tens to hundreds of MB) and
+instantiate the loader classes, so they are slow and require internet. Downloaded files are
+cached under `iopy/.temp_data/`, so reruns are faster. `test_download` forces a re-download
+with `refresh=True`.
+
+## Architecture
+
+The core abstraction is a **two-layer design**: a shared analysis engine and per-database loaders.
+
+### `IO` base class — [iopy/core/base_io.py](iopy/core/base_io.py)
+The engine. Its `__init__` asserts that the subclass has already set a fixed contract of
+attributes (`Z`, `X`, `V`, `FD`, `FD_REGION`, `ADD`, `rs`, `regions`, `sectors`,
+`sector_name_mapping`, `unit`, `demand_items`), then derives the coefficient and inverse
+matrices from them:
+- `A` = technical coefficients (`Z / X`), `L` = Leontief inverse `(I − A)⁻¹`
+- `B` = allocation coefficients, `G` = Ghosh inverse `(I − B)⁻¹`
+
+It provides the public methods `leontief_demand_shock`, `ghosh_supply_shock`, and
+`get_imports_exports`, plus the private `_shock` / `_shock_to_df` / `_plot_shock` helpers.
+Zeros in `X` are replaced with 1 before division so the matrices stay invertible.
+
+### Loader classes — `OECD`, `Figaro`, `ExioBase`
+[oecd.py](iopy/core/oecd.py), [figaro.py](iopy/core/figaro.py),
+[exiobase.py](iopy/core/exiobase.py). Each subclasses `IO` and follows the **same lifecycle**
+inside `__init__`:
+1. Validate `version`/`year`/`kind` against `config`.
+2. Resolve the download URL from `config`, extract a `_file_id` via the version's `regex_id`,
+   download to `DATA_FOLDER` if not already cached (or `refresh=True`).
+3. Load the raw file into `self.df`, slice it into the contract matrices (`Z`, `FD`, `X`,
+   `V`, `FD_REGION`, plus database-specific extras in the `ADD` dict).
+4. Call `super().__init__()` **last** to build `A`/`L`/`B`/`G`.
+
+When adding a new data version or year, the change is almost always **data-only**: add the
+download links and `num_regions`/`num_sectors`/`regex_id` to
+[iopy/core/config.py](iopy/core/config.py). Each loader has a module-level `process_df`
+helper that converts the raw dataframe index/columns into `(region, sector)` tuples — the
+three differ because each source formats its labels differently (OECD maps ISO alpha-3 →
+alpha-2 via `ALPHA3_TO_ALPHA2`; EXIOBASE uses MultiIndex columns).
+
+### `Matrix` — [iopy/core/matrix.py](iopy/core/matrix.py)
+A thin `numpy.ndarray` subclass carrying `info` (label), `rows`, and `columns` metadata
+(lists of `(region, sector)` tuples), plus `.I` (inverse) and `.T` (transpose that also swaps
+row/column labels). All IO matrices are `Matrix` instances; row/column selection in
+`get_imports_exports` relies on these label lists.
+
+### Supporting modules
+- [config.py](iopy/core/config.py) — nested dict of `{database: {version: {links, regex_id,
+  num_regions, num_sectors}}}`. The single source of truth for what data exists.
+- [mappings.py](iopy/core/mappings.py) — sector-code → human-name dicts and demand-item lists
+  per database; also the OECD 2022→2021 sector-code remapping.
+- [globals.py](iopy/core/globals.py) — `DATA_FOLDER` (`iopy/.temp_data`), `FILES_LOG`,
+  `IS_WINDOWS`.
+- [utils.py](iopy/core/utils.py) — `assert_is_subset` (the standard validation used
+  throughout), `remove_downloaded_files`, and the `ALPHA3_TO_ALPHA2` country-code map.
+
+### Download cache and cleanup
+Downloaded archives live in `DATA_FOLDER` and every download appends a `db_name;path` line to
+`FILES_LOG` (`_files_log.txt`). `remove_downloaded_files(database=...)` reads that log to
+delete files for one database or `'all'`. This is exposed as `iopy.remove_downloaded_files`
+and as a `remove_downloaded_files` static method on each loader.
+
+## Conventions
+
+- The public surface is re-exported from [iopy/\_\_init\_\_.py](iopy/__init__.py): `OECD`,
+  `Figaro`, `ExioBase`, `remove_downloaded_files`, `get_size_data_folder`. Keep new
+  user-facing entry points exported there.
+- Validation is done with `assert_is_subset` (raises `ValueError` listing what's missing) and
+  plain `assert` for argument-shape checks — match this style rather than introducing custom
+  exception types.
+- Version strings are dict keys (e.g. `'2021'`, `'2022-extended'`, `'2025-regular'`), not
+  numbers; `kind` is `'industry-by-industry'` or `'product-by-product'` (OECD has no `kind`).
diff --git a/iopy/.DS_Store b/iopy/.DS_Store
new file mode 100644
index 0000000000000000000000000000000000000000..c71720f927fd2fd7fe1ab504465f460efd4e9fca
GIT binary patch
literal 6148
zcmeHK%}&BV5dIc1LGi%BgBO#&fF!;_C3x`61E8hEkd#&_5WV&8!}xrj^qbiYp#@Dm
zXrh@(X21QJnPtCeb{PO}uu3n19)J#2uyw>{kI1^{j<mvzEIN(xfH6|sVM#Ul3k*o^
zj?rhvrkG<Ze`z_%lA_3%*!>3U(IU@^{GRJ#bHdb*nf3$)RtzVjJ_A1AWW^-Oii9@&
z%G^1z|E|Iw&{fMBa?HN)dTDsIG2uQ7OzY8c)v!C|=Qzb07h;P?{<4|&<{ahRuhgSI
zVOfvk-WqXw-}7$?E8C|$RDw&{vAz>t8<rfJ7z4(DG4Rt2$hA%BbUj*Y3>X8(z=i?2
zA0nz?9I*1JR|gv{0f=o*tFSM%g!n`O<A9Y%u27tJC3;sEN(|@SIUfbPIAG<`yTgUz
z!-bt)s8F2l&iIj<!^Ix0H3p1<Is*rO*pv7F`t$m~?qnllz!>;f47m1iG#v0q;ceYH
woV>LW^_D6kc9qAC3p@H0(^j71GpY*bBk2(1fR#tiQ0$L@(qN4-@T&}b0BA{aT>t<8

literal 0
HcmV?d00001

diff --git a/iopy/core/config.py b/iopy/core/config.py
index ecf9912..22f67f8 100644
--- a/iopy/core/config.py
+++ b/iopy/core/config.py
@@ -3,102 +3,81 @@
 **Authors**: W. Wakker
 
 """
+
+
+def _chunk_links(base, chunks):
+    """Expand ``{filename: (start_year, end_year)}`` into ``{year: url}``.
+
+    OECD distributes each ICIO edition as a handful of multi-year zip archives
+    (e.g. ``2016-2022_SML.zip`` holds one CSV per year), so every year in a range
+    resolves to the same download URL.
+    """
+    return {year: base + filename
+            for filename, (start, end) in chunks.items()
+            for year in range(start, end + 1)}
+
+
 config = {
     'oecd':
         {'2021':
-             {'links':
-                  {1995: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=91d8e84b-7406-46b9-af5f-ec096242755c',
-                   1996: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=91d8e84b-7406-46b9-af5f-ec096242755c',
-                   1997: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=91d8e84b-7406-46b9-af5f-ec096242755c',
-                   1998: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=91d8e84b-7406-46b9-af5f-ec096242755c',
-                   1999: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=91d8e84b-7406-46b9-af5f-ec096242755c',
-                   2000: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=8adf89dd-18b4-40fe-bc7f-c822052eb961',
-                   2001: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=8adf89dd-18b4-40fe-bc7f-c822052eb961',
-                   2002: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=8adf89dd-18b4-40fe-bc7f-c822052eb961',
-                   2003: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=8adf89dd-18b4-40fe-bc7f-c822052eb961',
-                   2004: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=8adf89dd-18b4-40fe-bc7f-c822052eb961',
-                   2005: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=fe218690-0a3b-44aa-a82c-b3e3da6d24db',
-                   2006: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=fe218690-0a3b-44aa-a82c-b3e3da6d24db',
-                   2007: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=fe218690-0a3b-44aa-a82c-b3e3da6d24db',
-                   2008: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=fe218690-0a3b-44aa-a82c-b3e3da6d24db',
-                   2009: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=fe218690-0a3b-44aa-a82c-b3e3da6d24db',
-                   2010: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=2c2f499f-5703-4034-9457-2f7518e8f2fc',
-                   2011: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=2c2f499f-5703-4034-9457-2f7518e8f2fc',
-                   2012: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=2c2f499f-5703-4034-9457-2f7518e8f2fc',
-                   2013: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=2c2f499f-5703-4034-9457-2f7518e8f2fc',
-                   2014: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=2c2f499f-5703-4034-9457-2f7518e8f2fc',
-                   2015: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=59a3d7f2-3f23-40d5-95ca-48da84c0f861',
-                   2016: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=59a3d7f2-3f23-40d5-95ca-48da84c0f861',
-                   2017: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=59a3d7f2-3f23-40d5-95ca-48da84c0f861',
-                   2018: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=59a3d7f2-3f23-40d5-95ca-48da84c0f861'},
-              'regex_id': r'[A-Za-z0-9]{8}-[A-Za-z0-9]{4}-[A-Za-z0-9]{4}-[A-Za-z0-9]{4}-[A-Za-z0-9]{12}',
+             {'links': _chunk_links('https://webfs-sti.oecd.org/files/STI-PIE/ICIO/2021/', {
+                  'ICIO2021_1995-1999.zip': (1995, 1999),
+                  'ICIO2021_2000-2004.zip': (2000, 2004),
+                  'ICIO2021_2005-2009.zip': (2005, 2009),
+                  'ICIO2021_2010-2014.zip': (2010, 2014),
+                  'ICIO2021_2015-2018.zip': (2015, 2018),
+              }),
+              'regex_id': r'ICIO2021_[0-9]{4}-[0-9]{4}',
               'num_regions': 71,
               'num_sectors': 45
               },
          '2022-extended':
-             {'links':
-                  {1995: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=4a2d6739-f717-48ac-a97b-d95f96984c55',
-                   1996: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=4a2d6739-f717-48ac-a97b-d95f96984c55',
-                   1997: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=4a2d6739-f717-48ac-a97b-d95f96984c55',
-                   1998: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=4a2d6739-f717-48ac-a97b-d95f96984c55',
-                   1999: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=4a2d6739-f717-48ac-a97b-d95f96984c55',
-                   2000: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=8eb6cf87-1899-4547-b337-d76dd4ef608c',
-                   2001: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=8eb6cf87-1899-4547-b337-d76dd4ef608c',
-                   2002: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=8eb6cf87-1899-4547-b337-d76dd4ef608c',
-                   2003: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=8eb6cf87-1899-4547-b337-d76dd4ef608c',
-                   2004: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=8eb6cf87-1899-4547-b337-d76dd4ef608c',
-                   2005: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=f9b6dd9a-e64b-47fb-832e-bd1628c43b72',
-                   2006: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=f9b6dd9a-e64b-47fb-832e-bd1628c43b72',
-                   2007: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=f9b6dd9a-e64b-47fb-832e-bd1628c43b72',
-                   2008: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=f9b6dd9a-e64b-47fb-832e-bd1628c43b72',
-                   2009: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=f9b6dd9a-e64b-47fb-832e-bd1628c43b72',
-                   2010: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=35cfb172-2cc4-4d62-98e5-9e5f1c51d6c9',
-                   2011: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=35cfb172-2cc4-4d62-98e5-9e5f1c51d6c9',
-                   2012: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=35cfb172-2cc4-4d62-98e5-9e5f1c51d6c9',
-                   2013: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=35cfb172-2cc4-4d62-98e5-9e5f1c51d6c9',
-                   2014: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=35cfb172-2cc4-4d62-98e5-9e5f1c51d6c9',
-                   2015: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=e5368118-b274-4e42-820e-33dacbfb94ed',
-                   2016: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=e5368118-b274-4e42-820e-33dacbfb94ed',
-                   2017: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=e5368118-b274-4e42-820e-33dacbfb94ed',
-                   2018: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=e5368118-b274-4e42-820e-33dacbfb94ed',
-                   2019: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=e5368118-b274-4e42-820e-33dacbfb94ed',
-                   2020: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=92539517-1661-4077-a9df-136abc39b858'},
-              'regex_id': r'[A-Za-z0-9]{8}-[A-Za-z0-9]{4}-[A-Za-z0-9]{4}-[A-Za-z0-9]{4}-[A-Za-z0-9]{12}',
+             {'links': _chunk_links('https://webfs-sti.oecd.org/files/STI-PIE/ICIO/2023/', {
+                  '1995-2000_EXT.zip': (1995, 2000),
+                  '2001-2005_EXT.zip': (2001, 2005),
+                  '2006-2010_EXT.zip': (2006, 2010),
+                  '2011-2015_EXT.zip': (2011, 2015),
+                  '2016-2020_EXT.zip': (2016, 2020),
+              }),
+              'regex_id': r'2023/[0-9]{4}-[0-9]{4}_EXT',
               'num_regions': 81,
               'num_sectors': 45},
          '2022-small':
-             {'links':
-                  {1995: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=8f69e3c5-8bc0-4c7b-aad2-5ef776c119ea',
-                   1996: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=8f69e3c5-8bc0-4c7b-aad2-5ef776c119ea',
-                   1997: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=8f69e3c5-8bc0-4c7b-aad2-5ef776c119ea',
-                   1998: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=8f69e3c5-8bc0-4c7b-aad2-5ef776c119ea',
-                   1999: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=8f69e3c5-8bc0-4c7b-aad2-5ef776c119ea',
-                   2000: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=43e7a690-c1ea-4839-b0fb-907e0aa79523',
-                   2001: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=43e7a690-c1ea-4839-b0fb-907e0aa79523',
-                   2002: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=43e7a690-c1ea-4839-b0fb-907e0aa79523',
-                   2003: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=43e7a690-c1ea-4839-b0fb-907e0aa79523',
-                   2004: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=43e7a690-c1ea-4839-b0fb-907e0aa79523',
-                   2005: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=9942ca81-9137-4926-adcf-dbe19fa2bcb6',
-                   2006: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=9942ca81-9137-4926-adcf-dbe19fa2bcb6',
-                   2007: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=9942ca81-9137-4926-adcf-dbe19fa2bcb6',
-                   2008: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=9942ca81-9137-4926-adcf-dbe19fa2bcb6',
-                   2009: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=9942ca81-9137-4926-adcf-dbe19fa2bcb6',
-                   2010: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=832d60ce-8346-456b-8ab8-cafd81c2f054',
-                   2011: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=832d60ce-8346-456b-8ab8-cafd81c2f054',
-                   2012: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=832d60ce-8346-456b-8ab8-cafd81c2f054',
-                   2013: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=832d60ce-8346-456b-8ab8-cafd81c2f054',
-                   2014: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=832d60ce-8346-456b-8ab8-cafd81c2f054',
-                   2015: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=ac018f80-6899-4395-af52-5c21134c51b3',
-                   2016: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=ac018f80-6899-4395-af52-5c21134c51b3',
-                   2017: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=ac018f80-6899-4395-af52-5c21134c51b3',
-                   2018: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=ac018f80-6899-4395-af52-5c21134c51b3',
-                   2019: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=ac018f80-6899-4395-af52-5c21134c51b3',
-                   2020: 'https://stats.oecd.org/wbos/fileview2.aspx?IDFile=60d244bd-5b40-4be3-bd93-0dc35c210ece'},
-              'regex_id': r'[A-Za-z0-9]{8}-[A-Za-z0-9]{4}-[A-Za-z0-9]{4}-[A-Za-z0-9]{4}-[A-Za-z0-9]{12}',
+             {'links': _chunk_links('https://webfs-sti.oecd.org/files/STI-PIE/ICIO/2023/', {
+                  '1995-2000_SML.zip': (1995, 2000),
+                  '2001-2005_SML.zip': (2001, 2005),
+                  '2006-2010_SML.zip': (2006, 2010),
+                  '2011-2015_SML.zip': (2011, 2015),
+                  '2016-2020_SML.zip': (2016, 2020),
+              }),
+              'regex_id': r'2023/[0-9]{4}-[0-9]{4}_SML',
               'num_regions': 77,
               'num_sectors': 45},
+        '2025-extended':
+             {'links': _chunk_links('https://webfs-sti.oecd.org/files/STI-PIE/ICIO/2025/', {
+                  '1995-2000_EXT.zip': (1995, 2000),
+                  '2001-2005_EXT.zip': (2001, 2005),
+                  '2006-2010_EXT.zip': (2006, 2010),
+                  '2011-2015_EXT.zip': (2011, 2015),
+                  '2016-2022_EXT.zip': (2016, 2022),
+              }),
+              'regex_id': r'2025/[0-9]{4}-[0-9]{4}_EXT',
+              'num_regions': 85,
+              'num_sectors': 50},
+         '2025-regular':
+             {'links': _chunk_links('https://webfs-sti.oecd.org/files/STI-PIE/ICIO/2025/', {
+                  '1995-2000_SML.zip': (1995, 2000),
+                  '2001-2005_SML.zip': (2001, 2005),
+                  '2006-2010_SML.zip': (2006, 2010),
+                  '2011-2015_SML.zip': (2011, 2015),
+                  '2016-2022_SML.zip': (2016, 2022),
+              }),
+              'regex_id': r'2025/[0-9]{4}-[0-9]{4}_SML',
+              'num_regions': 81,
+              'num_sectors': 50}
 
          },
+
     'figaro':
         {'2022':
             {'links':
diff --git a/iopy/core/exiobase.py b/iopy/core/exiobase.py
index bec7f33..8314655 100644
--- a/iopy/core/exiobase.py
+++ b/iopy/core/exiobase.py
@@ -15,7 +15,7 @@
 from iopy.core.base_io import IO
 from warnings import warn
 from iopy.core.globals import DATA_FOLDER, FILES_LOG
-from iopy.core.utils import remove_downloaded_files
+from iopy.core.utils import remove_downloaded_files, download_file
 
 db_name = os.path.basename(__file__).rstrip('.py')
 
@@ -170,13 +170,8 @@ def _load_data(self):
         return z_raw, fd_raw, x_raw, metadata, sector_codes, FD_codes
 
     def _download_data(self):
-        import requests
-
         try:
-            r = requests.get(self._url, stream=True)
-            with open(self._data_file, "wb") as f:
-                for chunk in r.iter_content(1024 * 5):
-                    f.write(chunk)
+            download_file(self._url, self._data_file)
             with open(FILES_LOG, 'a') as files_log:
                 files_log.write(db_name + ';' + self._data_file + '\n')
         except Exception as e:
diff --git a/iopy/core/figaro.py b/iopy/core/figaro.py
index 19ae9e3..2d0f840 100644
--- a/iopy/core/figaro.py
+++ b/iopy/core/figaro.py
@@ -15,7 +15,7 @@
 from warnings import warn
 from iopy.core.base_io import IO
 from iopy.core.globals import DATA_FOLDER, FILES_LOG
-from iopy.core.utils import remove_downloaded_files
+from iopy.core.utils import remove_downloaded_files, download_file
 
 db_name = os.path.basename(__file__).rstrip('.py')
 
@@ -139,13 +139,8 @@ def _load_data(self):
         return df
 
     def _download_data(self):
-        import requests
-
         try:
-            r = requests.get(self._url, stream=True)
-            with open(self._data_file, "wb") as f:
-                for chunk in r.iter_content(1024 * 5):
-                    f.write(chunk)
+            download_file(self._url, self._data_file)
             with open(FILES_LOG, 'a') as files_log:
                 files_log.write(db_name + ';' + self._data_file + '\n')
         except Exception as e:
diff --git a/iopy/core/oecd.py b/iopy/core/oecd.py
index d723dfd..a821490 100644
--- a/iopy/core/oecd.py
+++ b/iopy/core/oecd.py
@@ -15,7 +15,7 @@
 import os
 from iopy.core.config import config
 from iopy.core.base_io import IO
-from iopy.core.utils import replace_if_exists, remove_downloaded_files
+from iopy.core.utils import replace_if_exists, remove_downloaded_files, download_file
 from iopy.core.globals import DATA_FOLDER, FILES_LOG
 from warnings import warn
 from functools import partial
@@ -65,7 +65,7 @@ def __init__(self,
         self.year = year
         self.version = version
         self._url = config['oecd'][version]['links'][year]
-        self._file_id = re.search(config['oecd'][version]['regex_id'], self._url).group(0)
+        self._file_id = re.search(config['oecd'][version]['regex_id'], self._url).group(0).replace('/', '_')
         self._data_file = os.path.join(DATA_FOLDER, self._file_id + '.zip')
         file_exists = os.path.exists(self._data_file)
         download = not file_exists or refresh
@@ -143,22 +143,21 @@ def _load_data(self):
         if self.version == '2021':
             filename = f'ICIO2021_{self.year}.csv'
         elif self.version == '2022-extended':
-            filename = f'{self.year}.CSV'
+            filename = f'{self.year}.csv'
         elif self.version == '2022-small':
-            filename = f'{self.year}SML.CSV'
+            filename = f'{self.year}_SML.csv'
+        elif self.version == '2025-extended':
+            filename = f'{self.year}.csv'
+        elif self.version == '2025-regular':
+            filename = f'{self.year}_SML.csv'
         with ZipFile(self._data_file, 'r') as zf:
             with zf.open(filename, 'r') as csv_file:
                 df = pd.read_csv(csv_file, index_col=0)
         return df
 
     def _download_data(self):
-        import requests
-
         try:
-            r = requests.get(self._url, stream=True)
-            with open(self._data_file, "wb") as f:
-                for chunk in r.iter_content(1024 * 5):
-                    f.write(chunk)
+            download_file(self._url, self._data_file)
             with open(FILES_LOG, 'a') as files_log:
                 files_log.write(db_name + ';' + self._data_file + '\n')
         except Exception as e:
diff --git a/iopy/core/utils.py b/iopy/core/utils.py
index 26fca1b..3cd103b 100644
--- a/iopy/core/utils.py
+++ b/iopy/core/utils.py
@@ -13,6 +13,25 @@ def assert_is_subset(subset, superset):
         raise ValueError(f'Not found: {set(subset).difference(superset)}')
 
 
+def download_file(url, dest):
+    """Stream-download ``url`` to the local path ``dest``.
+
+    Uses ``curl_cffi`` with browser (TLS) impersonation so downloads succeed even
+    when the host sits behind Cloudflare's bot challenge. The OECD file server
+    (``webfs-sti.oecd.org``) rejects plain ``requests``/``urllib`` traffic with
+    HTTP 403; impersonating a real browser passes the challenge and is harmless
+    for the non-protected Eurostat/EXIOBASE hosts.
+    """
+    from curl_cffi import requests
+
+    r = requests.get(url, stream=True, impersonate='chrome')
+    if not r.ok:
+        raise ConnectionError(r.reason or f'HTTP {r.status_code}')
+    with open(dest, 'wb') as f:
+        for chunk in r.iter_content():
+            f.write(chunk)
+
+
 def replace_if_exists(x, mapping):
     """Replace if x exists in mapping, otherwise return x
 
diff --git a/iopy/dev-requirements.txt b/iopy/dev-requirements.txt
index dd8ec32..25c21b6 100644
--- a/iopy/dev-requirements.txt
+++ b/iopy/dev-requirements.txt
@@ -1,4 +1,5 @@
 requests
+curl_cffi
 numpy
 pandas>=1.1.5
 matplotlib
diff --git a/iopy/requirements.txt b/iopy/requirements.txt
index 48761f3..3c710c3 100644
--- a/iopy/requirements.txt
+++ b/iopy/requirements.txt
@@ -1,4 +1,5 @@
 requests
+curl_cffi
 numpy
 pandas>=1.1.5
 matplotlib
diff --git a/test.py b/test.py
index e69de29..dd633bc 100644
--- a/test.py
+++ b/test.py
@@ -0,0 +1,3 @@
+from iopy import OECD
+
+io = OECD(version='2025-extended', year=2022, refresh=True)

From 93d4f88d5bde9df298d395424bab44969d605c4b Mon Sep 17 00:00:00 2001
From: Wouter Wakker <wouter.wakker@outlook.com>
Date: Wed, 3 Jun 2026 22:53:55 +0200
Subject: [PATCH 03/15] oecd working

---
 iopy/core/mappings.py   | 54 +++++++++++++++++++++++++++++++++++++++++
 iopy/core/oecd.py       | 14 ++++++-----
 iopy/tests/test_oecd.py |  4 +++
 3 files changed, 66 insertions(+), 6 deletions(-)

diff --git a/iopy/core/mappings.py b/iopy/core/mappings.py
index 30b2c54..c482d57 100644
--- a/iopy/core/mappings.py
+++ b/iopy/core/mappings.py
@@ -143,6 +143,60 @@
                                    '94T96': 'Other service activities',
                                    '97T98': 'Households'}
 
+# OECD ICIO 2025 edition uses a 50-industry ISIC Rev.4 breakdown (codes differ from the
+# 45-sector classification above). Source: ReadMe_ICIO_*.xlsx "Area_Activities" sheet.
+oecd_sector_name_mapping_2025 = {'A01': 'Agriculture and hunting',
+                                 'A02': 'Forestry and logging',
+                                 'A03': 'Fishing and aquaculture',
+                                 'B05': 'Mining of coal and lignite',
+                                 'B06': 'Extraction of crude petroleum and natural gas',
+                                 'B07': 'Mining of metal ores',
+                                 'B08': 'Other mining and quarrying',
+                                 'B09': 'Mining support service activities',
+                                 'C10T12': 'Manufacture of food products; beverages and tobacco products',
+                                 'C13T15': 'Manufacture of textiles, wearing apparel, leather and related products',
+                                 'C16': 'Manufacture of wood and of products of wood and cork',
+                                 'C17_18': 'Manufacture of paper and paper products; Printing and reproduction of recorded media',
+                                 'C19': 'Manufacture of coke and refined petroleum products',
+                                 'C20': 'Manufacture of chemicals and chemical products',
+                                 'C21': 'Manufacture of basic pharmaceutical products and pharmaceutical preparations',
+                                 'C22': 'Manufacture of rubber and plastic products',
+                                 'C23': 'Manufacture of other non-metallic mineral products',
+                                 'C24A': 'Manufacture of basic iron and steel',
+                                 'C24B': 'Manufacture of basic precious and other non-ferrous metals',
+                                 'C25': 'Manufacture of fabricated metal products',
+                                 'C26': 'Manufacture of computer, electronic and optical products',
+                                 'C27': 'Manufacture of electrical equipment',
+                                 'C28': 'Manufacture of machinery and equipment n.e.c.',
+                                 'C29': 'Manufacture of motor vehicles, trailers and semi-trailers',
+                                 'C301': 'Building of ships and boats',
+                                 'C302T309': 'Manufacture of other transport equipment',
+                                 'C31T33': 'Manufacture of furniture; other manufacturing; repair and installation of machinery and equipment',
+                                 'D': 'Electricity, gas, steam and air conditioning supply',
+                                 'E': 'Water supply; sewerage, waste management and remediation activities',
+                                 'F': 'Construction',
+                                 'G': 'Wholesale and retail trade; repair of motor vehicles and motorcycles',
+                                 'H49': 'Land transport and transport via pipelines',
+                                 'H50': 'Water transport',
+                                 'H51': 'Air transport',
+                                 'H52': 'Warehousing and support activities for transportation',
+                                 'H53': 'Postal and courier activities',
+                                 'I': 'Accommodation and food service activities',
+                                 'J58T60': 'Publishing, Motion picture, video, television programme production and broadcasting activities',
+                                 'J61': 'Telecommunications',
+                                 'J62_63': 'Computer programming and information service activities',
+                                 'K': 'Financial and insurance activities',
+                                 'L': 'Real estate activities',
+                                 'M': 'Professional, scientific and technical activities',
+                                 'N': 'Administrative and support service activities',
+                                 'O': 'Public administration and defence; compulsory social security',
+                                 'P': 'Education',
+                                 'Q': 'Human health and social work activities',
+                                 'R': 'Arts, entertainment and recreation activities',
+                                 'S': 'Other service activities',
+                                 'T': 'Activities of households as employers; undifferentiated goods- and services-producing '
+                                      'activities of households for own use'}
+
 oecd_demand_items = {'HFCE': 'Household Final Consumption Expenditure',
                      'NPISH': 'Non-Profit Institutions Serving Households',
                      'GGFC': 'General Government Final Consumption',
diff --git a/iopy/core/oecd.py b/iopy/core/oecd.py
index a821490..36cddf1 100644
--- a/iopy/core/oecd.py
+++ b/iopy/core/oecd.py
@@ -3,7 +3,8 @@
 **Authors**: W. Wakker
 
 """
-from iopy.core.mappings import oecd_sector_name_mapping, oecd_demand_items, oecd_sector_2022_2021_mapping
+from iopy.core.mappings import oecd_sector_name_mapping, oecd_sector_name_mapping_2025, oecd_demand_items, \
+    oecd_sector_2022_2021_mapping
 from iopy.core.matrix import Matrix
 from functools import lru_cache
 import numpy as np
@@ -26,15 +27,15 @@
 def process_df(df):
     if df.shape[0] > 1 and df.shape[1] > 1:
         return (df,
-                [(ALPHA3_TO_ALPHA2[r] if r in ALPHA3_TO_ALPHA2 else r, s) for r, s in df.index.str.split('_')],
-                [(ALPHA3_TO_ALPHA2[r] if r in ALPHA3_TO_ALPHA2 else r, s) for r, s in df.columns.str.split('_')])
+                [(ALPHA3_TO_ALPHA2[r] if r in ALPHA3_TO_ALPHA2 else r, s) for r, s in df.index.str.split('_', n=1)],
+                [(ALPHA3_TO_ALPHA2[r] if r in ALPHA3_TO_ALPHA2 else r, s) for r, s in df.columns.str.split('_', n=1)])
     elif df.shape[0] == 1:
         return (df,
                 df.index.to_list(),
-                [(ALPHA3_TO_ALPHA2[r] if r in ALPHA3_TO_ALPHA2 else r, s) for r, s in df.columns.str.split('_')])
+                [(ALPHA3_TO_ALPHA2[r] if r in ALPHA3_TO_ALPHA2 else r, s) for r, s in df.columns.str.split('_', n=1)])
     else:
         return (df,
-                [(ALPHA3_TO_ALPHA2[r] if r in ALPHA3_TO_ALPHA2 else r, s) for r, s in df.index.str.split('_')],
+                [(ALPHA3_TO_ALPHA2[r] if r in ALPHA3_TO_ALPHA2 else r, s) for r, s in df.index.str.split('_', n=1)],
                 df.columns.to_list())
 
 
@@ -128,7 +129,8 @@ def __init__(self,
             self.regions = list(sorted(np.unique([r for r, s in self.Z.rows])))
             self.sectors = list(sorted(np.unique([s for r, s in self.Z.rows])))
             self.unit = 'Million USD'
-            self.sector_name_mapping = oecd_sector_name_mapping
+            self.sector_name_mapping = oecd_sector_name_mapping_2025 if version in ('2025-extended', '2025-regular') \
+                else oecd_sector_name_mapping
             self.demand_items = oecd_demand_items
             self.reference = f'OECD ({self.version[:4]}), OECD Inter-Country Input-Output Database, http://oe.cd/icio'
             self.contact = 'ICIO-TiVA.Contact@oecd.org, mentioning ICIO'
diff --git a/iopy/tests/test_oecd.py b/iopy/tests/test_oecd.py
index 51afe0a..927bdfb 100644
--- a/iopy/tests/test_oecd.py
+++ b/iopy/tests/test_oecd.py
@@ -28,6 +28,10 @@ def test_load(self):
         assert set(o.sectors).issubset(o.sector_name_mapping)
         o = OECD(version='2022-small', year=2018)
         assert set(o.sectors).issubset(o.sector_name_mapping)
+        o = OECD(version='2025-extended', year=2022)
+        assert set(o.sectors).issubset(o.sector_name_mapping)
+        o = OECD(version='2025-regular', year=2022)
+        assert set(o.sectors).issubset(o.sector_name_mapping)
 
     def test_matrices(self):
         for attr in ['Z', 'A', 'B', 'L', 'G', 'V', 'FD', 'X']:

From dac796e6413720095c8a1cfd0d00e9adbb0cc5ab Mon Sep 17 00:00:00 2001
From: Wouter Wakker <wouter.wakker@outlook.com>
Date: Wed, 3 Jun 2026 23:17:18 +0200
Subject: [PATCH 04/15] figaro 2025 working

---
 iopy/core/config.py       | 49 ++++++++++++++++++++++++++++++++++++++-
 iopy/core/figaro.py       |  4 ++--
 iopy/tests/test_figaro.py | 14 +++++------
 3 files changed, 57 insertions(+), 10 deletions(-)

diff --git a/iopy/core/config.py b/iopy/core/config.py
index 22f67f8..4f685dd 100644
--- a/iopy/core/config.py
+++ b/iopy/core/config.py
@@ -17,6 +17,16 @@ def _chunk_links(base, chunks):
             for year in range(start, end + 1)}
 
 
+def _circabc_links(kind_tag, ids):
+    """Build CIRCABC anonymous download URLs for Figaro from ``{year: node_id}``.
+
+    Eurostat distributes the Figaro 2025 edition through CIRCABC; every file has a
+    stable, auth-free download URL of the form ``/sd/a/{node_id}/{filename}``.
+    """
+    return {year: f'https://circabc.europa.eu/sd/a/{nid}/matrix_eu-ic-io_{kind_tag}_25ed_{year}.csv'
+            for year, nid in ids.items()}
+
+
 config = {
     'oecd':
         {'2021':
@@ -110,7 +120,44 @@ def _chunk_links(base, chunks):
                 },
                 'regex_id': r'[A-Za-z0-9]{8}-[A-Za-z0-9]{4}-[A-Za-z0-9]{4}-[A-Za-z0-9]{4}-[A-Za-z0-9]{12}',
                 'num_regions': 46,
-                'num_sectors': 64}
+                'num_sectors': 64},
+         '2025':
+             {'links':
+                  {'product-by-product': _circabc_links('prod-by-prod', {
+                  2010: 'f6d2007e-34c9-420a-a106-8c5ea836c49d',
+                  2011: 'd27f7d7c-4006-4c36-8820-7d9990679323',
+                  2012: '7cea8b73-f166-47ff-9c27-f240794b57af',
+                  2013: 'd56c2ed9-0c4f-40a3-9bbc-1aee49f5bd45',
+                  2014: 'b5aa0e38-3d1b-496b-b180-bbd9bab3e153',
+                  2015: 'd037b772-8bf9-424f-97a4-9274d261e6cf',
+                  2016: '66369d21-8262-47d6-ad8a-dc536a2466d0',
+                  2017: 'd6848c38-9569-4848-bb68-4144fed38c71',
+                  2018: '557b9483-023e-497c-96c5-44ce807bd444',
+                  2019: 'fa0e9127-7a2b-4b74-bbad-1bc9e717197b',
+                  2020: '46f5665e-d2e1-4270-95a1-970da3d70d32',
+                  2021: 'a7e2919d-f084-4c35-8b21-e960a854e2bd',
+                  2022: '31484cee-43fe-45d4-a546-209e5898c6dd',
+                  2023: 'e213892a-afac-4d34-83e9-45e8a324e7e8',
+                  }),
+                   'industry-by-industry': _circabc_links('ind-by-ind', {
+                  2010: 'fc80f855-d144-476e-b4bf-5cfba946819c',
+                  2011: '1bcb2624-04ed-43e1-8588-df6680ed352a',
+                  2012: '399671ad-cbb3-493e-ad5f-83e989f1eecc',
+                  2013: 'a2b4746d-1d11-4a44-ab1c-50ac956f0849',
+                  2014: 'beba57b2-2696-497a-b92f-2a5beca724c7',
+                  2015: '1a194b8c-6ea1-4bec-9c73-0cd599febcc3',
+                  2016: '2cdd74fc-0bce-4ae0-8bf2-34d34546d86d',
+                  2017: '4a11c796-4186-4cce-a02f-12d353fc5e59',
+                  2018: '7a57a374-2200-498c-bb5f-cee24202b0b8',
+                  2019: 'c3467617-8a00-44a0-9b6b-ccad8a2ab58d',
+                  2020: '4df668e1-2a8a-4e84-ae57-00309d8bc760',
+                  2021: '6736dea8-da14-450f-b212-a791baf238c8',
+                  2022: 'b20c339d-984f-413c-a499-54ff76beb90c',
+                  2023: '21557f49-1e94-431c-8523-d972fec020b8',
+                  })},
+              'regex_id': r'matrix_eu-ic-io_[a-z-]+_25ed_[0-9]{4}',
+              'num_regions': 50,
+              'num_sectors': 64},
         },
     'exiobase':
         {'3.81':
diff --git a/iopy/core/figaro.py b/iopy/core/figaro.py
index 2d0f840..3613082 100644
--- a/iopy/core/figaro.py
+++ b/iopy/core/figaro.py
@@ -46,8 +46,8 @@ def __init__(self,
         """
 
         Args:
-            version: Edition (year of publication), e.g. '2022'
-            year: Year from 2010 to 2020
+            version: Edition (year of publication), e.g. '2025'
+            year: Year; availability depends on the edition (e.g. 2010-2023 for '2025')
             kind: industry-by-industry (default) or product-by-product
             refresh: Download the data even if it exists on the hard drive
         """
diff --git a/iopy/tests/test_figaro.py b/iopy/tests/test_figaro.py
index 03e2ae9..021db1c 100644
--- a/iopy/tests/test_figaro.py
+++ b/iopy/tests/test_figaro.py
@@ -9,7 +9,7 @@
 import matplotlib
 import matplotlib.pyplot as plt
 
-f = Figaro(version='2022', year=2018, kind='industry-by-industry')
+f = Figaro(version='2025', year=2018, kind='industry-by-industry')
 
 custom_shock_vector = np.random.uniform(size=f.rs, low=-10, high=10).reshape(-1, 1)
 
@@ -19,12 +19,12 @@
 class TestFigaro:
 
     def test_download(self):
-        Figaro(version='2022', year=2018, refresh=True)
+        Figaro(version='2025', year=2018, refresh=True)
 
     def test_load(self):
-        fi = Figaro(version='2022', year=2018, kind='industry-by-industry')
+        fi = Figaro(version='2025', year=2018, kind='industry-by-industry')
         assert set(fi.sectors).issubset(fi.sector_name_mapping)
-        fi = Figaro(version='2022', year=2018, kind='product-by-product')
+        fi = Figaro(version='2025', year=2018, kind='product-by-product')
         assert set(fi.sectors).issubset(fi.sector_name_mapping)
 
     def test_matrices(self):
@@ -107,19 +107,19 @@ def test_get_imports_exports(self):
                                                 export_regions='AU',
                                                 import_sectors=None,
                                                 export_sectors=None,
-                                                use_type='intermediate'), 86280.8749)
+                                                use_type='intermediate'), 72202.248)
 
         assert np.isclose(f.get_imports_exports(import_regions='CN',
                                                 export_regions='AU',
                                                 import_sectors=None,
                                                 export_sectors=None,
-                                                use_type='final'), 13770.405)
+                                                use_type='final'), 8212.555)
 
         assert np.isclose(f.get_imports_exports(import_regions=['CN'],
                                                 export_regions='AU',
                                                 import_sectors=None,
                                                 export_sectors=None,
-                                                use_type='both'), 86280.8749 + 13770.405)
+                                                use_type='both'), 72202.248 + 8212.555)
 
         f.get_imports_exports(import_regions=['CN'],
                               export_regions='AU',

From 6b66ee52b35b601189e7a88b042b854d15aac0b0 Mon Sep 17 00:00:00 2001
From: Wouter Wakker <wouter.wakker@outlook.com>
Date: Fri, 5 Jun 2026 21:58:58 +0200
Subject: [PATCH 05/15] Removed figaro 2022

---
 iopy/core/config.py | 33 +--------------------------------
 1 file changed, 1 insertion(+), 32 deletions(-)

diff --git a/iopy/core/config.py b/iopy/core/config.py
index 4f685dd..1867450 100644
--- a/iopy/core/config.py
+++ b/iopy/core/config.py
@@ -89,38 +89,7 @@ def _circabc_links(kind_tag, ids):
          },
 
     'figaro':
-        {'2022':
-            {'links':
-                {'product-by-product': {
-                    2010: 'https://ec.europa.eu/eurostat/documents/51957/12789261/matrix_eu-ic-io_prod-by-prod_2010.csv/bc2b60d9-32f2-1c80-56e4-f839ca2f06fc?t=1655180575504',
-                    2011: 'https://ec.europa.eu/eurostat/documents/51957/12789261/matrix_eu-ic-io_prod-by-prod_2011.csv/74e6f603-1740-116c-3469-4db7ebe97aa2?t=1655182400093',
-                    2012: 'https://ec.europa.eu/eurostat/documents/51957/12789261/matrix_eu-ic-io_prod-by-prod_2012.csv/17b1bf95-daa9-a506-9d8b-932edf8f3950?t=1655184005469',
-                    2013: 'https://ec.europa.eu/eurostat/documents/51957/12789261/matrix_eu-ic-io_prod-by-prod_2013.csv/d32be770-9f71-e28c-e8ac-11a04d8d885c?t=1655185487140',
-                    2014: 'https://ec.europa.eu/eurostat/documents/51957/12789261/matrix_eu-ic-io_prod-by-prod_2014.csv/815e7a8e-a1c7-89f4-e621-ce3971e79c4b?t=1655186769271',
-                    2015: 'https://ec.europa.eu/eurostat/documents/51957/12789261/matrix_eu-ic-io_prod-by-prod_2015.csv/3c69e27a-8f45-29e9-cdef-59d8a402e927?t=1655186809149',
-                    2016: 'https://ec.europa.eu/eurostat/documents/51957/12789261/matrix_eu-ic-io_prod-by-prod_2016.csv/f82150d8-74b3-b0a3-4764-567eab545b5a?t=1655188856261',
-                    2017: 'https://ec.europa.eu/eurostat/documents/51957/12789261/matrix_eu-ic-io_prod-by-prod_2017.csv/7ec909a5-0faa-e45d-2173-43f6c1b59689?t=1655188931368',
-                    2018: 'https://ec.europa.eu/eurostat/documents/51957/12789261/matrix_eu-ic-io_prod-by-prod_2018.csv/c44abd15-b354-6690-24dc-f186fe147bb2?t=1655196163454',
-                    2019: 'https://ec.europa.eu/eurostat/documents/51957/12789261/matrix_eu-ic-io_prod-by-prod_2019.csv/0ab0c866-8dd8-92d6-1ac7-51ffe78fd5dc?t=1655196217023',
-                    2020: 'https://ec.europa.eu/eurostat/documents/51957/12789261/matrix_eu-ic-io_prod-by-prod_2020.csv/3b838d62-884e-c1dc-9693-314b6460af0d?t=1655196273194',
-                },
-                    'industry-by-industry': {
-                        2010: 'https://ec.europa.eu/eurostat/documents/51957/12789261/matrix_eu-ic-io_ind-by-ind_2010.csv/3db6f05f-8343-a45e-2012-14db4607716b?t=1655180547507',
-                        2011: 'https://ec.europa.eu/eurostat/documents/51957/12789261/matrix_eu-ic-io_ind-by-ind_2011.csv/5ce2f4ab-f4c1-e1dd-2aac-68813c714414?t=1655182368997',
-                        2012: 'https://ec.europa.eu/eurostat/documents/51957/12789261/matrix_eu-ic-io_ind-by-ind_2012.csv/a1ad1f56-3a20-fac1-ee79-9a80b6d25a46?t=1655183982037',
-                        2013: 'https://ec.europa.eu/eurostat/documents/51957/12789261/matrix_eu-ic-io_ind-by-ind_2013.csv/eec78482-56e2-053e-5486-15061be568e6?t=1655185412542',
-                        2014: 'https://ec.europa.eu/eurostat/documents/51957/12789261/matrix_eu-ic-io_ind-by-ind_2014.csv/35eb67c4-0550-f2ae-b827-6ac240b5f877?t=1655186676930',
-                        2015: 'https://ec.europa.eu/eurostat/documents/51957/12789261/matrix_eu-ic-io_ind-by-ind_2015.csv/af93fe1f-a9fd-c094-dd18-a15d390e0f9c?t=1655186725710',
-                        2016: 'https://ec.europa.eu/eurostat/documents/51957/12789261/matrix_eu-ic-io_ind-by-ind_2016.csv/8322aa36-8d5c-5d52-07e5-e3d527875d06?t=1655188541728',
-                        2017: 'https://ec.europa.eu/eurostat/documents/51957/12789261/matrix_eu-ic-io_ind-by-ind_2017.csv/30ad9b86-689f-d5a1-c80c-6765ea0f5421?t=1655188598849',
-                        2018: 'https://ec.europa.eu/eurostat/documents/51957/12789261/matrix_eu-ic-io_ind-by-ind_2018.csv/a0369f43-bf98-a362-9054-f7075cc01704?t=1655195970852',
-                        2019: 'https://ec.europa.eu/eurostat/documents/51957/12789261/matrix_eu-ic-io_ind-by-ind_2019.csv/c1354cbb-b3e6-6da5-780a-5be3319ad6d4?t=1655196037947',
-                        2020: 'https://ec.europa.eu/eurostat/documents/51957/12789261/matrix_eu-ic-io_ind-by-ind_2020.csv/6dc6df43-0d95-856c-897f-18ba2ca053f1?t=1655196109971',
-                    }
-                },
-                'regex_id': r'[A-Za-z0-9]{8}-[A-Za-z0-9]{4}-[A-Za-z0-9]{4}-[A-Za-z0-9]{4}-[A-Za-z0-9]{12}',
-                'num_regions': 46,
-                'num_sectors': 64},
+        {
          '2025':
              {'links':
                   {'product-by-product': _circabc_links('prod-by-prod', {

From ba0a542c4c8bf72015e3414d8707dfa3ba76caa4 Mon Sep 17 00:00:00 2001
From: Wouter Wakker <wouter.wakker@outlook.com>
Date: Fri, 5 Jun 2026 22:08:32 +0200
Subject: [PATCH 06/15] Added proxy and verify

---
 iopy/core/exiobase.py | 11 +++++++++--
 iopy/core/figaro.py   | 11 +++++++++--
 iopy/core/matrix.py   | 16 ++++++++++++++++
 iopy/core/oecd.py     | 11 +++++++++--
 iopy/core/utils.py    | 18 ++++++++++++++++--
 5 files changed, 59 insertions(+), 8 deletions(-)

diff --git a/iopy/core/exiobase.py b/iopy/core/exiobase.py
index 8314655..6631fb1 100644
--- a/iopy/core/exiobase.py
+++ b/iopy/core/exiobase.py
@@ -33,7 +33,9 @@ def __init__(self,
                  version: str,
                  year: int,
                  kind: str = 'industry-by-industry',
-                 refresh: bool = False):
+                 refresh: bool = False,
+                 proxy=None,
+                 verify=True):
         """
 
         Args:
@@ -41,6 +43,9 @@ def __init__(self,
             year: Year from 1995 to 2022
             kind: industry-by-industry (default) or product-by-product
             refresh: Download the data even if it exists on the hard drive
+            proxy: Optional proxy for downloading; a URL string (applied to http and
+                   https) or a ``{scheme: url}`` dict
+            verify: Verify the server's TLS certificate (``False`` to skip, or a CA bundle path)
         """
 
         assert kind in {'industry-by-industry', 'product-by-product'}
@@ -57,6 +62,8 @@ def __init__(self,
         self.version = version
         self.year = year
         self.kind = kind
+        self._proxy = proxy
+        self._verify = verify
         self._url = config['exiobase'][version]['links'][kind][year]
         self._file_id = re.search(config['exiobase'][version]['regex_id'], self._url).group(0)
         self._data_file = os.path.join(DATA_FOLDER, self._file_id + '.zip')
@@ -171,7 +178,7 @@ def _load_data(self):
 
     def _download_data(self):
         try:
-            download_file(self._url, self._data_file)
+            download_file(self._url, self._data_file, proxy=self._proxy, verify=self._verify)
             with open(FILES_LOG, 'a') as files_log:
                 files_log.write(db_name + ';' + self._data_file + '\n')
         except Exception as e:
diff --git a/iopy/core/figaro.py b/iopy/core/figaro.py
index 3613082..6f4f870 100644
--- a/iopy/core/figaro.py
+++ b/iopy/core/figaro.py
@@ -42,7 +42,9 @@ def __init__(self,
                  version: str,
                  year: int,
                  kind='industry-by-industry',
-                 refresh: bool = False):
+                 refresh: bool = False,
+                 proxy=None,
+                 verify=True):
         """
 
         Args:
@@ -50,6 +52,9 @@ def __init__(self,
             year: Year; availability depends on the edition (e.g. 2010-2023 for '2025')
             kind: industry-by-industry (default) or product-by-product
             refresh: Download the data even if it exists on the hard drive
+            proxy: Optional proxy for downloading; a URL string (applied to http and
+                   https) or a ``{scheme: url}`` dict
+            verify: Verify the server's TLS certificate (``False`` to skip, or a CA bundle path)
         """
         assert kind in {'industry-by-industry', 'product-by-product'}
 
@@ -65,6 +70,8 @@ def __init__(self,
         self.version = version
         self.year = year
         self.kind = kind
+        self._proxy = proxy
+        self._verify = verify
         self._url = config['figaro'][version]['links'][kind][year]
         self._file_id = re.search(config['figaro'][version]['regex_id'], self._url).group(0)
         self._data_file = os.path.join(DATA_FOLDER, self._file_id + '.csv')
@@ -140,7 +147,7 @@ def _load_data(self):
 
     def _download_data(self):
         try:
-            download_file(self._url, self._data_file)
+            download_file(self._url, self._data_file, proxy=self._proxy, verify=self._verify)
             with open(FILES_LOG, 'a') as files_log:
                 files_log.write(db_name + ';' + self._data_file + '\n')
         except Exception as e:
diff --git a/iopy/core/matrix.py b/iopy/core/matrix.py
index c818b74..5dbd21e 100644
--- a/iopy/core/matrix.py
+++ b/iopy/core/matrix.py
@@ -4,6 +4,7 @@
 
 """
 import numpy as np
+import pandas as pd
 
 
 class Matrix(np.ndarray):
@@ -68,3 +69,18 @@ def to_numpy(self):
             numpy array
         """
         return np.asarray(self)
+
+    def to_pandas(self):
+        """Convert to pandas DataFrame
+
+        Returns:
+            pandas DataFrame
+        """
+        def _to_index(labels):
+            if len(labels) and isinstance(labels[0], tuple):
+                return pd.MultiIndex.from_tuples(labels, names=['region', 'sector'])
+            return pd.Index(labels)
+
+        return pd.DataFrame(np.asarray(self),
+                            index=_to_index(self.rows),
+                            columns=_to_index(self.columns))
diff --git a/iopy/core/oecd.py b/iopy/core/oecd.py
index 36cddf1..f6c8077 100644
--- a/iopy/core/oecd.py
+++ b/iopy/core/oecd.py
@@ -45,13 +45,18 @@ class OECD(IO):
     def __init__(self,
                  version: str,
                  year: int,
-                 refresh: bool = False):
+                 refresh: bool = False,
+                 proxy=None,
+                 verify=True):
         """
 
         Args:
             version: Publication version of the data; '2021', '2022-small' or '2022-extended'
             year: Year
             refresh: Download the data even if it exists on the hard drive
+            proxy: Optional proxy for downloading; a URL string (applied to http and
+                   https) or a ``{scheme: url}`` dict
+            verify: Verify the server's TLS certificate (``False`` to skip, or a CA bundle path)
         """
 
         if version not in config['oecd'].keys():
@@ -65,6 +70,8 @@ def __init__(self,
 
         self.year = year
         self.version = version
+        self._proxy = proxy
+        self._verify = verify
         self._url = config['oecd'][version]['links'][year]
         self._file_id = re.search(config['oecd'][version]['regex_id'], self._url).group(0).replace('/', '_')
         self._data_file = os.path.join(DATA_FOLDER, self._file_id + '.zip')
@@ -159,7 +166,7 @@ def _load_data(self):
 
     def _download_data(self):
         try:
-            download_file(self._url, self._data_file)
+            download_file(self._url, self._data_file, proxy=self._proxy, verify=self._verify)
             with open(FILES_LOG, 'a') as files_log:
                 files_log.write(db_name + ';' + self._data_file + '\n')
         except Exception as e:
diff --git a/iopy/core/utils.py b/iopy/core/utils.py
index 3cd103b..8a470ba 100644
--- a/iopy/core/utils.py
+++ b/iopy/core/utils.py
@@ -13,7 +13,7 @@ def assert_is_subset(subset, superset):
         raise ValueError(f'Not found: {set(subset).difference(superset)}')
 
 
-def download_file(url, dest):
+def download_file(url, dest, proxy=None, verify=True):
     """Stream-download ``url`` to the local path ``dest``.
 
     Uses ``curl_cffi`` with browser (TLS) impersonation so downloads succeed even
@@ -21,10 +21,24 @@ def download_file(url, dest):
     (``webfs-sti.oecd.org``) rejects plain ``requests``/``urllib`` traffic with
     HTTP 403; impersonating a real browser passes the challenge and is harmless
     for the non-protected Eurostat/EXIOBASE hosts.
+
+    Args:
+        url: Source URL.
+        dest: Local file path to write to.
+        proxy: Optional proxy. Either a single URL string (e.g.
+               ``'http://user:pass@host:port'``) applied to both http and https,
+               or a ``{scheme: url}`` dict passed straight through.
+        verify: Verify the server's TLS certificate. Set to ``False`` to skip
+                verification (e.g. behind a TLS-intercepting proxy), or pass a
+                path to a CA bundle.
     """
     from curl_cffi import requests
 
-    r = requests.get(url, stream=True, impersonate='chrome')
+    kwargs = {}
+    if proxy is not None:
+        kwargs['proxies'] = {'http': proxy, 'https': proxy} if isinstance(proxy, str) else proxy
+
+    r = requests.get(url, stream=True, impersonate='chrome', verify=verify, **kwargs)
     if not r.ok:
         raise ConnectionError(r.reason or f'HTTP {r.status_code}')
     with open(dest, 'wb') as f:

From 057dc615dd9636364a6926dc62fc5116fbae55a1 Mon Sep 17 00:00:00 2001
From: Wouter Wakker <wouter.wakker@outlook.com>
Date: Fri, 5 Jun 2026 22:38:59 +0200
Subject: [PATCH 07/15] Renamed iopy to iotables

---
 .gitignore                                    |   6 ++
 CHANGELOG.MD                                  |   9 ++-
 CLAUDE.md                                     |  58 +++++++++++-------
 README.md                                     |  18 +++---
 iopy/.DS_Store                                | Bin 6148 -> 0 bytes
 iopy/__about__.py                             |   6 --
 iopy/core/__init__.py                         |   7 ---
 iopy/core/globals.py                          |  10 ---
 iopy/dev-requirements.txt                     |   8 ---
 iopy/requirements.txt                         |   6 --
 iopy/tests/.coveragerc                        |   5 --
 iopy/tests/test_iopy.py                       |  15 -----
 pyproject.toml                                |  37 +++++++++++
 setup.py                                      |  32 ----------
 {iopy => src/iotables}/__init__.py            |  23 ++++---
 {iopy/core => src/iotables}/base_io.py        |   4 +-
 {iopy/core => src/iotables}/config.py         |   0
 {iopy/core => src/iotables}/exiobase.py       |  10 +--
 {iopy/core => src/iotables}/figaro.py         |  12 ++--
 src/iotables/globals.py                       |  14 +++++
 {iopy/core => src/iotables}/mappings.py       |   0
 {iopy/core => src/iotables}/matrix.py         |   0
 {iopy/core => src/iotables}/oecd.py           |  14 ++---
 {iopy/core => src/iotables}/utils.py          |   2 +-
 test.py                                       |   2 +-
 {iopy/tests => tests}/__init__.py             |   0
 {iopy/tests => tests}/results/__init__.py     |   0
 .../results/adjust_source_in_covxml.py        |   2 +-
 {iopy/tests => tests}/test_exiobase.py        |   2 +-
 {iopy/tests => tests}/test_figaro.py          |   2 +-
 tests/test_iotables.py                        |  15 +++++
 {iopy/tests => tests}/test_matrix.py          |   4 +-
 {iopy/tests => tests}/test_oecd.py            |   2 +-
 33 files changed, 166 insertions(+), 159 deletions(-)
 delete mode 100644 iopy/.DS_Store
 delete mode 100644 iopy/__about__.py
 delete mode 100644 iopy/core/__init__.py
 delete mode 100644 iopy/core/globals.py
 delete mode 100644 iopy/dev-requirements.txt
 delete mode 100644 iopy/requirements.txt
 delete mode 100644 iopy/tests/.coveragerc
 delete mode 100644 iopy/tests/test_iopy.py
 create mode 100644 pyproject.toml
 delete mode 100644 setup.py
 rename {iopy => src/iotables}/__init__.py (56%)
 rename {iopy/core => src/iotables}/base_io.py (99%)
 rename {iopy/core => src/iotables}/config.py (100%)
 rename {iopy/core => src/iotables}/exiobase.py (97%)
 rename {iopy/core => src/iotables}/figaro.py (95%)
 create mode 100644 src/iotables/globals.py
 rename {iopy/core => src/iotables}/mappings.py (100%)
 rename {iopy/core => src/iotables}/matrix.py (100%)
 rename {iopy/core => src/iotables}/oecd.py (95%)
 rename {iopy/core => src/iotables}/utils.py (99%)
 rename {iopy/tests => tests}/__init__.py (100%)
 rename {iopy/tests => tests}/results/__init__.py (100%)
 rename {iopy/tests => tests}/results/adjust_source_in_covxml.py (83%)
 rename {iopy/tests => tests}/test_exiobase.py (93%)
 rename {iopy/tests => tests}/test_figaro.py (99%)
 create mode 100644 tests/test_iotables.py
 rename {iopy/tests => tests}/test_matrix.py (96%)
 rename {iopy/tests => tests}/test_oecd.py (99%)

diff --git a/.gitignore b/.gitignore
index f58a6e3..8492abb 100644
--- a/.gitignore
+++ b/.gitignore
@@ -10,3 +10,9 @@
 **/__pycache__/
 **/.temp_data/
 test.py
+
+# packaging build artifacts
+build/
+dist/
+*.egg-info/
+src/*.egg-info/
diff --git a/CHANGELOG.MD b/CHANGELOG.MD
index e55e8da..f582eed 100644
--- a/CHANGELOG.MD
+++ b/CHANGELOG.MD
@@ -8,9 +8,16 @@ and this project adheres to [Semantic Versioning](http://semver.org/).
 
 ### Added
 - Introduce structure for different dataset-specific tools
-- ...
+- `proxy` and `verify` options on the loaders for downloading behind a proxy / custom TLS
+- `Matrix.to_pandas()` to convert a matrix to a labelled DataFrame
+- OECD ICIO 2025 edition (`2025-regular`, `2025-extended`) and Eurostat Figaro 2025 edition (`2025`, years 2010-2023)
 ### Changed
+- Renamed the package from `iopy` to `iotables` and adopted a `src/` layout with `pyproject.toml`
+- Downloads now use `curl_cffi` (browser TLS impersonation) so Cloudflare-protected OECD files work
+- Re-pointed OECD/Figaro download URLs to current hosts (OECD `webfs`, Figaro CIRCABC)
+- Cached downloads now live in a user cache dir (`~/.cache/iotables`, override with `IOTABLES_DATA`)
 ### Fixed
+- OECD `2025-regular` file-id regex and in-zip CSV filenames; cross-edition cache-id collisions
 ___
 
 
diff --git a/CLAUDE.md b/CLAUDE.md
index 3b047e2..091818f 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -4,39 +4,42 @@ This file provides guidance to Claude Code (claude.ai/code) when working with co
 
 ## What this is
 
-`iopy` is a Python package for loading inter-country input-output (IO) data from three
+`iotables` is a Python package for loading inter-country input-output (IO) data from three
 public databases (OECD ICIO, Eurostat Figaro, EXIOBASE) and running economic analyses on
 them — primarily Leontief demand shocks and Ghosh supply shocks. See [README.md](README.md)
 for the economic theory (Leontief/Ghosh derivations) and the full public API table.
 
+The package uses a `src/` layout (`src/iotables/`) with packaging configured in
+[pyproject.toml](pyproject.toml); tests live in the top-level [tests/](tests/) directory.
+
 ## Commands
 
 ```bash
-# Install for development (editable)
-pip install -e .
-pip install -r iopy/dev-requirements.txt   # runtime deps + pytest, pytest-cov
+# Install for development (editable, with test deps)
+pip install -e ".[dev]"
 
 # Run the full test suite
-pytest iopy/tests
+pytest
 
 # Run a single test file / test
-pytest iopy/tests/test_oecd.py
-pytest iopy/tests/test_oecd.py::TestOECD::test_leontief
+pytest tests/test_oecd.py
+pytest tests/test_oecd.py::TestOECD::test_leontief
 
-# Coverage (config in iopy/tests/.coveragerc)
-pytest --cov=iopy iopy/tests
+# Coverage (config in pyproject.toml [tool.coverage])
+pytest --cov=iotables
 ```
 
 Note: the tests **download real data over the network** (tens to hundreds of MB) and
 instantiate the loader classes, so they are slow and require internet. Downloaded files are
-cached under `iopy/.temp_data/`, so reruns are faster. `test_download` forces a re-download
-with `refresh=True`.
+cached under a user cache dir (`~/.cache/iotables`, override with the `IOTABLES_DATA`
+environment variable), so reruns are faster. `test_download` forces a re-download with
+`refresh=True`.
 
 ## Architecture
 
 The core abstraction is a **two-layer design**: a shared analysis engine and per-database loaders.
 
-### `IO` base class — [iopy/core/base_io.py](iopy/core/base_io.py)
+### `IO` base class — [src/iotables/base_io.py](src/iotables/base_io.py)
 The engine. Its `__init__` asserts that the subclass has already set a fixed contract of
 attributes (`Z`, `X`, `V`, `FD`, `FD_REGION`, `ADD`, `rs`, `regions`, `sectors`,
 `sector_name_mapping`, `unit`, `demand_items`), then derives the coefficient and inverse
@@ -49,8 +52,8 @@ It provides the public methods `leontief_demand_shock`, `ghosh_supply_shock`, an
 Zeros in `X` are replaced with 1 before division so the matrices stay invertible.
 
 ### Loader classes — `OECD`, `Figaro`, `ExioBase`
-[oecd.py](iopy/core/oecd.py), [figaro.py](iopy/core/figaro.py),
-[exiobase.py](iopy/core/exiobase.py). Each subclasses `IO` and follows the **same lifecycle**
+[oecd.py](src/iotables/oecd.py), [figaro.py](src/iotables/figaro.py),
+[exiobase.py](src/iotables/exiobase.py). Each subclasses `IO` and follows the **same lifecycle**
 inside `__init__`:
 1. Validate `version`/`year`/`kind` against `config`.
 2. Resolve the download URL from `config`, extract a `_file_id` via the version's `regex_id`,
@@ -61,36 +64,45 @@ inside `__init__`:
 
 When adding a new data version or year, the change is almost always **data-only**: add the
 download links and `num_regions`/`num_sectors`/`regex_id` to
-[iopy/core/config.py](iopy/core/config.py). Each loader has a module-level `process_df`
+[src/iotables/config.py](src/iotables/config.py). Each loader has a module-level `process_df`
 helper that converts the raw dataframe index/columns into `(region, sector)` tuples — the
 three differ because each source formats its labels differently (OECD maps ISO alpha-3 →
 alpha-2 via `ALPHA3_TO_ALPHA2`; EXIOBASE uses MultiIndex columns).
 
-### `Matrix` — [iopy/core/matrix.py](iopy/core/matrix.py)
+### `Matrix` — [src/iotables/matrix.py](src/iotables/matrix.py)
 A thin `numpy.ndarray` subclass carrying `info` (label), `rows`, and `columns` metadata
 (lists of `(region, sector)` tuples), plus `.I` (inverse) and `.T` (transpose that also swaps
 row/column labels). All IO matrices are `Matrix` instances; row/column selection in
 `get_imports_exports` relies on these label lists.
 
 ### Supporting modules
-- [config.py](iopy/core/config.py) — nested dict of `{database: {version: {links, regex_id,
+- [config.py](src/iotables/config.py) — nested dict of `{database: {version: {links, regex_id,
   num_regions, num_sectors}}}`. The single source of truth for what data exists.
-- [mappings.py](iopy/core/mappings.py) — sector-code → human-name dicts and demand-item lists
+- [mappings.py](src/iotables/mappings.py) — sector-code → human-name dicts and demand-item lists
   per database; also the OECD 2022→2021 sector-code remapping.
-- [globals.py](iopy/core/globals.py) — `DATA_FOLDER` (`iopy/.temp_data`), `FILES_LOG`,
+- [globals.py](src/iotables/globals.py) — `DATA_FOLDER` (`~/.cache/iotables`, `IOTABLES_DATA` override), `FILES_LOG`,
   `IS_WINDOWS`.
-- [utils.py](iopy/core/utils.py) — `assert_is_subset` (the standard validation used
-  throughout), `remove_downloaded_files`, and the `ALPHA3_TO_ALPHA2` country-code map.
+- [utils.py](src/iotables/utils.py) — `assert_is_subset` (the standard validation used
+  throughout), `remove_downloaded_files`, the `ALPHA3_TO_ALPHA2` country-code map, and
+  `download_file` (the single shared downloader used by all three loaders).
+
+### Downloading
+All three loaders download via `utils.download_file`, which uses **`curl_cffi`** with browser
+TLS impersonation (`impersonate='chrome'`). This is required because the OECD file server
+(`webfs-sti.oecd.org`) sits behind Cloudflare and rejects plain `requests`/`urllib` with HTTP
+403; impersonation is harmless for the Eurostat (CIRCABC) and EXIOBASE (Zenodo) hosts. Each
+loader's `__init__` accepts `proxy=` (URL string or `{scheme: url}` dict) and `verify=`
+(TLS verification toggle / CA-bundle path), threaded through to `download_file`.
 
 ### Download cache and cleanup
 Downloaded archives live in `DATA_FOLDER` and every download appends a `db_name;path` line to
 `FILES_LOG` (`_files_log.txt`). `remove_downloaded_files(database=...)` reads that log to
-delete files for one database or `'all'`. This is exposed as `iopy.remove_downloaded_files`
+delete files for one database or `'all'`. This is exposed as `iotables.remove_downloaded_files`
 and as a `remove_downloaded_files` static method on each loader.
 
 ## Conventions
 
-- The public surface is re-exported from [iopy/\_\_init\_\_.py](iopy/__init__.py): `OECD`,
+- The public surface is re-exported from [src/iotables/\_\_init\_\_.py](src/iotables/__init__.py): `OECD`,
   `Figaro`, `ExioBase`, `remove_downloaded_files`, `get_size_data_folder`. Keep new
   user-facing entry points exported there.
 - Validation is done with `assert_is_subset` (raises `ValueError` listing what's missing) and
diff --git a/README.md b/README.md
index 1e223ce..b3e9a25 100644
--- a/README.md
+++ b/README.md
@@ -1,4 +1,4 @@
-# iopy
+# iotables
 
 ## Description
 
@@ -14,7 +14,7 @@ Currently the following databases are supported:
 
 ## Installation
 ```bash
-pip install git+https://github.com/WWakker/iopy.git
+pip install iotables
 ```
 
 ## Structure of input-output data
@@ -176,16 +176,16 @@ Creating an instance of the OECD class loads the OECD data and gives access to c
 An instance can be create as follows, specifiying a year between 1995 and 2018.
 
 ```python
-import iopy 
-oecd = iopy.OECD(version='2021', year=2018, refresh=False)
+import iotables 
+oecd = iotables.OECD(version='2021', year=2018, refresh=False)
 ```
 
 Similary, an instance can be created for other data, for example Figaro and ExioBase as
 
 ```python
-import iopy 
-figaro = iopy.Figaro(version='2022', year=2020, kind='industry-by-industry')
-exio = iopy.ExioBase(version='3.81', year=2022, kind='industry-by-industry')
+import iotables 
+figaro = iotables.Figaro(version='2025', year=2020, kind='industry-by-industry')
+exio = iotables.ExioBase(version='3.81', year=2022, kind='industry-by-industry')
 ```
 
 Creating an instance of a database class downloads and loads the data into memory, creates standard input-output matrices, and gives access to the following attributes and methods:
@@ -223,9 +223,9 @@ All matrices are extended `numpy.ndarray`'s with attributes `info`, `rows` and `
 
 When running a Leontief or Ghosh shock, the percentage shock to final demand/primary inputs in countries and sectors can be specified as
 ```python
-import iopy
+import iotables
 
-oecd = iopy.OECD(version='2021', year=2018)
+oecd = iotables.OECD(version='2021', year=2018)
 
 df_l = oecd.leontief_demand_shock(shock=-10, regions=['FR', 'DE], sectors=['01T02', '35'])
 df_g = oecd.ghosh_supply_shock(shock=-10, regions=['FR', 'DE'], sectors=['01T02', '35'])
diff --git a/iopy/.DS_Store b/iopy/.DS_Store
deleted file mode 100644
index c71720f927fd2fd7fe1ab504465f460efd4e9fca..0000000000000000000000000000000000000000
GIT binary patch
literal 0
HcmV?d00001

literal 6148
zcmeHK%}&BV5dIc1LGi%BgBO#&fF!;_C3x`61E8hEkd#&_5WV&8!}xrj^qbiYp#@Dm
zXrh@(X21QJnPtCeb{PO}uu3n19)J#2uyw>{kI1^{j<mvzEIN(xfH6|sVM#Ul3k*o^
zj?rhvrkG<Ze`z_%lA_3%*!>3U(IU@^{GRJ#bHdb*nf3$)RtzVjJ_A1AWW^-Oii9@&
z%G^1z|E|Iw&{fMBa?HN)dTDsIG2uQ7OzY8c)v!C|=Qzb07h;P?{<4|&<{ahRuhgSI
zVOfvk-WqXw-}7$?E8C|$RDw&{vAz>t8<rfJ7z4(DG4Rt2$hA%BbUj*Y3>X8(z=i?2
zA0nz?9I*1JR|gv{0f=o*tFSM%g!n`O<A9Y%u27tJC3;sEN(|@SIUfbPIAG<`yTgUz
z!-bt)s8F2l&iIj<!^Ix0H3p1<Is*rO*pv7F`t$m~?qnllz!>;f47m1iG#v0q;ceYH
woV>LW^_D6kc9qAC3p@H0(^j71GpY*bBk2(1fR#tiQ0$L@(qN4-@T&}b0BA{aT>t<8

diff --git a/iopy/__about__.py b/iopy/__about__.py
deleted file mode 100644
index caf4018..0000000
--- a/iopy/__about__.py
+++ /dev/null
@@ -1,6 +0,0 @@
-__title__ = "iopy"
-__about__ = "A Python package to easily load and work with input-output data"
-__version__ = '0.0.1'
-__authors__ = 'Wouter Wakker'
-__email__ = "wouter.wakker@outlook.com"
-__url__ = ""
diff --git a/iopy/core/__init__.py b/iopy/core/__init__.py
deleted file mode 100644
index 07a74f5..0000000
--- a/iopy/core/__init__.py
+++ /dev/null
@@ -1,7 +0,0 @@
-"""  Created on 22/11/2022::
-------------- __init__.py -------------
-**Authors**: W. Wakker
-
-"""
-
-from iopy.core.globals import *
diff --git a/iopy/core/globals.py b/iopy/core/globals.py
deleted file mode 100644
index 1be3caf..0000000
--- a/iopy/core/globals.py
+++ /dev/null
@@ -1,10 +0,0 @@
-"""  Created on 22/11/2022::
-------------- globals -------------
-**Authors**: W. Wakker
-
-"""
-import os
-
-DATA_FOLDER = os.path.join(os.path.dirname(os.path.dirname(__file__)), '.temp_data')
-IS_WINDOWS = os.name == 'nt'
-FILES_LOG = os.path.join(DATA_FOLDER, '_files_log.txt')
diff --git a/iopy/dev-requirements.txt b/iopy/dev-requirements.txt
deleted file mode 100644
index 25c21b6..0000000
--- a/iopy/dev-requirements.txt
+++ /dev/null
@@ -1,8 +0,0 @@
-requests
-curl_cffi
-numpy
-pandas>=1.1.5
-matplotlib
-tqdm
-pytest
-pytest-cov
diff --git a/iopy/requirements.txt b/iopy/requirements.txt
deleted file mode 100644
index 3c710c3..0000000
--- a/iopy/requirements.txt
+++ /dev/null
@@ -1,6 +0,0 @@
-requests
-curl_cffi
-numpy
-pandas>=1.1.5
-matplotlib
-tqdm
diff --git a/iopy/tests/.coveragerc b/iopy/tests/.coveragerc
deleted file mode 100644
index 2a8a768..0000000
--- a/iopy/tests/.coveragerc
+++ /dev/null
@@ -1,5 +0,0 @@
-[run]
-omit = res/*
-       **/__about__.py
-       **/adjust_source_in_covxml.py
-       **/update_badge.py
diff --git a/iopy/tests/test_iopy.py b/iopy/tests/test_iopy.py
deleted file mode 100644
index 004c5e6..0000000
--- a/iopy/tests/test_iopy.py
+++ /dev/null
@@ -1,15 +0,0 @@
-"""  Created on 22/11/2022::
-------------- test_iopy -------------
-**Authors**: W. Wakker
-
-"""
-import iopy
-
-
-class TestIopy:
-
-    def test1(self):
-        iopy.remove_downloaded_files(database='figaro')
-
-    def test_folder_size(self):
-        assert isinstance(iopy.get_size_data_folder(), str)
diff --git a/pyproject.toml b/pyproject.toml
new file mode 100644
index 0000000..7e462b9
--- /dev/null
+++ b/pyproject.toml
@@ -0,0 +1,37 @@
+[build-system]
+requires = ["setuptools>=64", "wheel"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "iotables"
+version = "0.0.1"
+description = "A Python package to easily load and work with inter-country input-output data"
+readme = "README.md"
+license = { file = "LICENSE" }
+authors = [{ name = "Wouter Wakker", email = "wouter.wakker@outlook.com" }]
+requires-python = ">=3.8"
+dependencies = [
+    "curl_cffi",
+    "numpy",
+    "pandas>=1.1.5",
+    "matplotlib",
+    "tqdm",
+]
+classifiers = [
+    "Programming Language :: Python :: 3",
+    "Operating System :: OS Independent",
+    "License :: OSI Approved :: MIT License",
+]
+
+[project.optional-dependencies]
+dev = ["pytest", "pytest-cov"]
+
+[tool.setuptools.packages.find]
+where = ["src"]
+
+[tool.pytest.ini_options]
+testpaths = ["tests"]
+
+[tool.coverage.run]
+source = ["iotables"]
+omit = ["**/adjust_source_in_covxml.py", "**/update_badge.py"]
diff --git a/setup.py b/setup.py
deleted file mode 100644
index 2e11119..0000000
--- a/setup.py
+++ /dev/null
@@ -1,32 +0,0 @@
-import setuptools
-
-with open("README.md", "r") as f:
-    long_description = f.read()
-
-about = {}
-with open("iopy/__about__.py") as f:
-    exec(f.read(), about)
-
-with open("iopy/requirements.txt") as f:
-    requirements = f.read().splitlines()
-
-setuptools.setup(
-    name="iopy",
-    version=about['__version__'],
-    author=about['__authors__'],
-    author_email=about['__email__'],
-    description=about['__about__'],
-    url=about['__url__'],
-    license='MIT',
-    long_description=long_description,
-    long_description_content_type="markdown",
-    packages=setuptools.find_packages(),
-    install_requires=requirements,
-    classifiers=[
-        "Programming Language :: Python :: 3",
-        "Operating System :: OS Independent",
-    ],
-    python_requires='>=3.6',
-    include_package_data=False,
-    package_data={},
-)
diff --git a/iopy/__init__.py b/src/iotables/__init__.py
similarity index 56%
rename from iopy/__init__.py
rename to src/iotables/__init__.py
index aa7b0fd..1e7c8b2 100644
--- a/iopy/__init__.py
+++ b/src/iotables/__init__.py
@@ -1,11 +1,17 @@
 import os
-from iopy.core.globals import DATA_FOLDER as __DATA_FOLDER
-from iopy.core.globals import IS_WINDOWS as __IS_WINDOWS
-from iopy.core.globals import FILES_LOG as __FILES_LOG
-from iopy.core.oecd import OECD
-from iopy.core.figaro import Figaro
-from iopy.core.exiobase import ExioBase
-from iopy.core.utils import remove_downloaded_files
+from importlib.metadata import version as _version, PackageNotFoundError as _PackageNotFoundError
+from iotables.globals import DATA_FOLDER as __DATA_FOLDER
+from iotables.globals import IS_WINDOWS as __IS_WINDOWS
+from iotables.globals import FILES_LOG as __FILES_LOG
+from iotables.oecd import OECD
+from iotables.figaro import Figaro
+from iotables.exiobase import ExioBase
+from iotables.utils import remove_downloaded_files
+
+try:
+    __version__ = _version("iotables")
+except _PackageNotFoundError:  # package not installed (e.g. running from source tree)
+    __version__ = "0.0.0"
 
 
 def get_size_data_folder():
@@ -34,7 +40,6 @@ def human(size):
     return get_size(__DATA_FOLDER)
 
 
-if not os.path.exists(__DATA_FOLDER):
-    os.mkdir(__DATA_FOLDER)
+os.makedirs(__DATA_FOLDER, exist_ok=True)
 
 del os
diff --git a/iopy/core/base_io.py b/src/iotables/base_io.py
similarity index 99%
rename from iopy/core/base_io.py
rename to src/iotables/base_io.py
index 6549b00..de177a8 100644
--- a/iopy/core/base_io.py
+++ b/src/iotables/base_io.py
@@ -4,8 +4,8 @@
 
 """
 from warnings import warn
-from iopy.core.matrix import Matrix
-from iopy.core.utils import assert_is_subset
+from iotables.matrix import Matrix
+from iotables.utils import assert_is_subset
 import matplotlib.pyplot as plt
 from typing import Union, Iterable, Optional
 import numpy as np
diff --git a/iopy/core/config.py b/src/iotables/config.py
similarity index 100%
rename from iopy/core/config.py
rename to src/iotables/config.py
diff --git a/iopy/core/exiobase.py b/src/iotables/exiobase.py
similarity index 97%
rename from iopy/core/exiobase.py
rename to src/iotables/exiobase.py
index 6631fb1..5ac2435 100644
--- a/iopy/core/exiobase.py
+++ b/src/iotables/exiobase.py
@@ -3,7 +3,7 @@
 **Authors**: S. Boldrini
 """
 
-from iopy.core.matrix import Matrix
+from iotables.matrix import Matrix
 from functools import lru_cache
 import numpy as np
 import pandas as pd
@@ -11,11 +11,11 @@
 from zipfile import ZipFile
 import re
 import os
-from iopy.core.config import config
-from iopy.core.base_io import IO
+from iotables.config import config
+from iotables.base_io import IO
 from warnings import warn
-from iopy.core.globals import DATA_FOLDER, FILES_LOG
-from iopy.core.utils import remove_downloaded_files, download_file
+from iotables.globals import DATA_FOLDER, FILES_LOG
+from iotables.utils import remove_downloaded_files, download_file
 
 db_name = os.path.basename(__file__).rstrip('.py')
 
diff --git a/iopy/core/figaro.py b/src/iotables/figaro.py
similarity index 95%
rename from iopy/core/figaro.py
rename to src/iotables/figaro.py
index 6f4f870..d63df0a 100644
--- a/iopy/core/figaro.py
+++ b/src/iotables/figaro.py
@@ -3,19 +3,19 @@
 **Authors**: W. Wakker
 
 """
-from iopy.core.mappings import figaro_sector_name_mapping_pxp_2022, figaro_sector_name_mapping_ixi_2022, figaro_demand_items
-from iopy.core.matrix import Matrix
+from iotables.mappings import figaro_sector_name_mapping_pxp_2022, figaro_sector_name_mapping_ixi_2022, figaro_demand_items
+from iotables.matrix import Matrix
 from functools import lru_cache
 import numpy as np
 import pandas as pd
 from tqdm import tqdm
 import re
 import os
-from iopy.core.config import config
+from iotables.config import config
 from warnings import warn
-from iopy.core.base_io import IO
-from iopy.core.globals import DATA_FOLDER, FILES_LOG
-from iopy.core.utils import remove_downloaded_files, download_file
+from iotables.base_io import IO
+from iotables.globals import DATA_FOLDER, FILES_LOG
+from iotables.utils import remove_downloaded_files, download_file
 
 db_name = os.path.basename(__file__).rstrip('.py')
 
diff --git a/src/iotables/globals.py b/src/iotables/globals.py
new file mode 100644
index 0000000..6f06481
--- /dev/null
+++ b/src/iotables/globals.py
@@ -0,0 +1,14 @@
+"""  Created on 22/11/2022::
+------------- globals -------------
+**Authors**: W. Wakker
+
+"""
+import os
+
+# Downloaded source files are cached in a user-level cache directory rather than inside
+# the installed package. Override with the IOTABLES_DATA environment variable; otherwise
+# fall back to XDG_CACHE_HOME (or ~/.cache) per platform convention.
+_cache_root = os.environ.get('XDG_CACHE_HOME') or os.path.join(os.path.expanduser('~'), '.cache')
+DATA_FOLDER = os.environ.get('IOTABLES_DATA', os.path.join(_cache_root, 'iotables'))
+IS_WINDOWS = os.name == 'nt'
+FILES_LOG = os.path.join(DATA_FOLDER, '_files_log.txt')
diff --git a/iopy/core/mappings.py b/src/iotables/mappings.py
similarity index 100%
rename from iopy/core/mappings.py
rename to src/iotables/mappings.py
diff --git a/iopy/core/matrix.py b/src/iotables/matrix.py
similarity index 100%
rename from iopy/core/matrix.py
rename to src/iotables/matrix.py
diff --git a/iopy/core/oecd.py b/src/iotables/oecd.py
similarity index 95%
rename from iopy/core/oecd.py
rename to src/iotables/oecd.py
index f6c8077..5991410 100644
--- a/iopy/core/oecd.py
+++ b/src/iotables/oecd.py
@@ -3,21 +3,21 @@
 **Authors**: W. Wakker
 
 """
-from iopy.core.mappings import oecd_sector_name_mapping, oecd_sector_name_mapping_2025, oecd_demand_items, \
+from iotables.mappings import oecd_sector_name_mapping, oecd_sector_name_mapping_2025, oecd_demand_items, \
     oecd_sector_2022_2021_mapping
-from iopy.core.matrix import Matrix
+from iotables.matrix import Matrix
 from functools import lru_cache
 import numpy as np
 import pandas as pd
-from iopy.core.utils import ALPHA3_TO_ALPHA2
+from iotables.utils import ALPHA3_TO_ALPHA2
 from tqdm import tqdm
 from zipfile import ZipFile
 import re
 import os
-from iopy.core.config import config
-from iopy.core.base_io import IO
-from iopy.core.utils import replace_if_exists, remove_downloaded_files, download_file
-from iopy.core.globals import DATA_FOLDER, FILES_LOG
+from iotables.config import config
+from iotables.base_io import IO
+from iotables.utils import replace_if_exists, remove_downloaded_files, download_file
+from iotables.globals import DATA_FOLDER, FILES_LOG
 from warnings import warn
 from functools import partial
 
diff --git a/iopy/core/utils.py b/src/iotables/utils.py
similarity index 99%
rename from iopy/core/utils.py
rename to src/iotables/utils.py
index 8a470ba..915382d 100644
--- a/iopy/core/utils.py
+++ b/src/iotables/utils.py
@@ -3,7 +3,7 @@
 **Authors**: W. Wakker
 
 """
-from iopy.core.globals import FILES_LOG
+from iotables.globals import FILES_LOG
 from collections import defaultdict
 import os
 
diff --git a/test.py b/test.py
index dd633bc..06c3584 100644
--- a/test.py
+++ b/test.py
@@ -1,3 +1,3 @@
-from iopy import OECD
+from iotables import OECD
 
 io = OECD(version='2025-extended', year=2022, refresh=True)
diff --git a/iopy/tests/__init__.py b/tests/__init__.py
similarity index 100%
rename from iopy/tests/__init__.py
rename to tests/__init__.py
diff --git a/iopy/tests/results/__init__.py b/tests/results/__init__.py
similarity index 100%
rename from iopy/tests/results/__init__.py
rename to tests/results/__init__.py
diff --git a/iopy/tests/results/adjust_source_in_covxml.py b/tests/results/adjust_source_in_covxml.py
similarity index 83%
rename from iopy/tests/results/adjust_source_in_covxml.py
rename to tests/results/adjust_source_in_covxml.py
index 153a10f..5d6ea26 100644
--- a/iopy/tests/results/adjust_source_in_covxml.py
+++ b/tests/results/adjust_source_in_covxml.py
@@ -10,7 +10,7 @@
 with open(f'{parent}/cov.xml', 'r+') as f:
     covxml = f.read()
     covxml = re.sub(r'<source>.*</source>',
-                    r'<source>/builds/.../.../iopy/iopy</source>',
+                    r'<source>/builds/.../.../src/iotables</source>',
                     covxml)
     f.seek(0)
     f.write(covxml)
diff --git a/iopy/tests/test_exiobase.py b/tests/test_exiobase.py
similarity index 93%
rename from iopy/tests/test_exiobase.py
rename to tests/test_exiobase.py
index 0815b05..d7841fd 100644
--- a/iopy/tests/test_exiobase.py
+++ b/tests/test_exiobase.py
@@ -3,7 +3,7 @@
 **Authors**: W. Wakker
 
 """
-from iopy import ExioBase
+from iotables import ExioBase
 
 
 class TestExioBase:
diff --git a/iopy/tests/test_figaro.py b/tests/test_figaro.py
similarity index 99%
rename from iopy/tests/test_figaro.py
rename to tests/test_figaro.py
index 021db1c..6371edf 100644
--- a/iopy/tests/test_figaro.py
+++ b/tests/test_figaro.py
@@ -4,7 +4,7 @@
 
 """
 import pytest
-from iopy import Figaro
+from iotables import Figaro
 import numpy as np
 import matplotlib
 import matplotlib.pyplot as plt
diff --git a/tests/test_iotables.py b/tests/test_iotables.py
new file mode 100644
index 0000000..c0110eb
--- /dev/null
+++ b/tests/test_iotables.py
@@ -0,0 +1,15 @@
+"""  Created on 22/11/2022::
+------------- test_iotables -------------
+**Authors**: W. Wakker
+
+"""
+import iotables
+
+
+class TestIotables:
+
+    def test1(self):
+        iotables.remove_downloaded_files(database='figaro')
+
+    def test_folder_size(self):
+        assert isinstance(iotables.get_size_data_folder(), str)
diff --git a/iopy/tests/test_matrix.py b/tests/test_matrix.py
similarity index 96%
rename from iopy/tests/test_matrix.py
rename to tests/test_matrix.py
index 3dd379a..9f3df9e 100644
--- a/iopy/tests/test_matrix.py
+++ b/tests/test_matrix.py
@@ -3,11 +3,11 @@
 **Authors**: W. Wakker
 
 """
-from iopy.core.matrix import Matrix
+from iotables.matrix import Matrix
 import pandas as pd
 import numpy as np
 import pytest
-from iopy.core.utils import ALPHA3_TO_ALPHA2
+from iotables.utils import ALPHA3_TO_ALPHA2
 
 
 def process_df(df):
diff --git a/iopy/tests/test_oecd.py b/tests/test_oecd.py
similarity index 99%
rename from iopy/tests/test_oecd.py
rename to tests/test_oecd.py
index 927bdfb..8762427 100644
--- a/iopy/tests/test_oecd.py
+++ b/tests/test_oecd.py
@@ -4,7 +4,7 @@
 
 """
 import pytest
-from iopy import OECD
+from iotables import OECD
 import numpy as np
 import matplotlib
 import matplotlib.pyplot as plt

From c2a2dff2244d823823215fa96150fa90085f87a8 Mon Sep 17 00:00:00 2001
From: Wouter Wakker <wouter.wakker@outlook.com>
Date: Sat, 6 Jun 2026 13:44:06 +0200
Subject: [PATCH 08/15] Stop tracking .vscode and CLAUDE.md

---
 .vscode/settings.json |   4 --
 CLAUDE.md             | 112 ------------------------------------------
 2 files changed, 116 deletions(-)
 delete mode 100644 .vscode/settings.json
 delete mode 100644 CLAUDE.md

diff --git a/.vscode/settings.json b/.vscode/settings.json
deleted file mode 100644
index 4b5a294..0000000
--- a/.vscode/settings.json
+++ /dev/null
@@ -1,4 +0,0 @@
-{
-    "python-envs.defaultEnvManager": "ms-python.python:conda",
-    "python-envs.defaultPackageManager": "ms-python.python:conda"
-}
\ No newline at end of file
diff --git a/CLAUDE.md b/CLAUDE.md
deleted file mode 100644
index 091818f..0000000
--- a/CLAUDE.md
+++ /dev/null
@@ -1,112 +0,0 @@
-# CLAUDE.md
-
-This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
-
-## What this is
-
-`iotables` is a Python package for loading inter-country input-output (IO) data from three
-public databases (OECD ICIO, Eurostat Figaro, EXIOBASE) and running economic analyses on
-them — primarily Leontief demand shocks and Ghosh supply shocks. See [README.md](README.md)
-for the economic theory (Leontief/Ghosh derivations) and the full public API table.
-
-The package uses a `src/` layout (`src/iotables/`) with packaging configured in
-[pyproject.toml](pyproject.toml); tests live in the top-level [tests/](tests/) directory.
-
-## Commands
-
-```bash
-# Install for development (editable, with test deps)
-pip install -e ".[dev]"
-
-# Run the full test suite
-pytest
-
-# Run a single test file / test
-pytest tests/test_oecd.py
-pytest tests/test_oecd.py::TestOECD::test_leontief
-
-# Coverage (config in pyproject.toml [tool.coverage])
-pytest --cov=iotables
-```
-
-Note: the tests **download real data over the network** (tens to hundreds of MB) and
-instantiate the loader classes, so they are slow and require internet. Downloaded files are
-cached under a user cache dir (`~/.cache/iotables`, override with the `IOTABLES_DATA`
-environment variable), so reruns are faster. `test_download` forces a re-download with
-`refresh=True`.
-
-## Architecture
-
-The core abstraction is a **two-layer design**: a shared analysis engine and per-database loaders.
-
-### `IO` base class — [src/iotables/base_io.py](src/iotables/base_io.py)
-The engine. Its `__init__` asserts that the subclass has already set a fixed contract of
-attributes (`Z`, `X`, `V`, `FD`, `FD_REGION`, `ADD`, `rs`, `regions`, `sectors`,
-`sector_name_mapping`, `unit`, `demand_items`), then derives the coefficient and inverse
-matrices from them:
-- `A` = technical coefficients (`Z / X`), `L` = Leontief inverse `(I − A)⁻¹`
-- `B` = allocation coefficients, `G` = Ghosh inverse `(I − B)⁻¹`
-
-It provides the public methods `leontief_demand_shock`, `ghosh_supply_shock`, and
-`get_imports_exports`, plus the private `_shock` / `_shock_to_df` / `_plot_shock` helpers.
-Zeros in `X` are replaced with 1 before division so the matrices stay invertible.
-
-### Loader classes — `OECD`, `Figaro`, `ExioBase`
-[oecd.py](src/iotables/oecd.py), [figaro.py](src/iotables/figaro.py),
-[exiobase.py](src/iotables/exiobase.py). Each subclasses `IO` and follows the **same lifecycle**
-inside `__init__`:
-1. Validate `version`/`year`/`kind` against `config`.
-2. Resolve the download URL from `config`, extract a `_file_id` via the version's `regex_id`,
-   download to `DATA_FOLDER` if not already cached (or `refresh=True`).
-3. Load the raw file into `self.df`, slice it into the contract matrices (`Z`, `FD`, `X`,
-   `V`, `FD_REGION`, plus database-specific extras in the `ADD` dict).
-4. Call `super().__init__()` **last** to build `A`/`L`/`B`/`G`.
-
-When adding a new data version or year, the change is almost always **data-only**: add the
-download links and `num_regions`/`num_sectors`/`regex_id` to
-[src/iotables/config.py](src/iotables/config.py). Each loader has a module-level `process_df`
-helper that converts the raw dataframe index/columns into `(region, sector)` tuples — the
-three differ because each source formats its labels differently (OECD maps ISO alpha-3 →
-alpha-2 via `ALPHA3_TO_ALPHA2`; EXIOBASE uses MultiIndex columns).
-
-### `Matrix` — [src/iotables/matrix.py](src/iotables/matrix.py)
-A thin `numpy.ndarray` subclass carrying `info` (label), `rows`, and `columns` metadata
-(lists of `(region, sector)` tuples), plus `.I` (inverse) and `.T` (transpose that also swaps
-row/column labels). All IO matrices are `Matrix` instances; row/column selection in
-`get_imports_exports` relies on these label lists.
-
-### Supporting modules
-- [config.py](src/iotables/config.py) — nested dict of `{database: {version: {links, regex_id,
-  num_regions, num_sectors}}}`. The single source of truth for what data exists.
-- [mappings.py](src/iotables/mappings.py) — sector-code → human-name dicts and demand-item lists
-  per database; also the OECD 2022→2021 sector-code remapping.
-- [globals.py](src/iotables/globals.py) — `DATA_FOLDER` (`~/.cache/iotables`, `IOTABLES_DATA` override), `FILES_LOG`,
-  `IS_WINDOWS`.
-- [utils.py](src/iotables/utils.py) — `assert_is_subset` (the standard validation used
-  throughout), `remove_downloaded_files`, the `ALPHA3_TO_ALPHA2` country-code map, and
-  `download_file` (the single shared downloader used by all three loaders).
-
-### Downloading
-All three loaders download via `utils.download_file`, which uses **`curl_cffi`** with browser
-TLS impersonation (`impersonate='chrome'`). This is required because the OECD file server
-(`webfs-sti.oecd.org`) sits behind Cloudflare and rejects plain `requests`/`urllib` with HTTP
-403; impersonation is harmless for the Eurostat (CIRCABC) and EXIOBASE (Zenodo) hosts. Each
-loader's `__init__` accepts `proxy=` (URL string or `{scheme: url}` dict) and `verify=`
-(TLS verification toggle / CA-bundle path), threaded through to `download_file`.
-
-### Download cache and cleanup
-Downloaded archives live in `DATA_FOLDER` and every download appends a `db_name;path` line to
-`FILES_LOG` (`_files_log.txt`). `remove_downloaded_files(database=...)` reads that log to
-delete files for one database or `'all'`. This is exposed as `iotables.remove_downloaded_files`
-and as a `remove_downloaded_files` static method on each loader.
-
-## Conventions
-
-- The public surface is re-exported from [src/iotables/\_\_init\_\_.py](src/iotables/__init__.py): `OECD`,
-  `Figaro`, `ExioBase`, `remove_downloaded_files`, `get_size_data_folder`. Keep new
-  user-facing entry points exported there.
-- Validation is done with `assert_is_subset` (raises `ValueError` listing what's missing) and
-  plain `assert` for argument-shape checks — match this style rather than introducing custom
-  exception types.
-- Version strings are dict keys (e.g. `'2021'`, `'2022-extended'`, `'2025-regular'`), not
-  numbers; `kind` is `'industry-by-industry'` or `'product-by-product'` (OECD has no `kind`).

From 8ff5473bc151a5786a6ee35f792ca5ef1b3d434c Mon Sep 17 00:00:00 2001
From: Wouter Wakker <wouter.wakker@outlook.com>
Date: Sat, 6 Jun 2026 13:44:26 +0200
Subject: [PATCH 09/15] Add .vscode and CLAUDE.md to gitignore

---
 .gitignore | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/.gitignore b/.gitignore
index 8492abb..5e78aaa 100644
--- a/.gitignore
+++ b/.gitignore
@@ -16,3 +16,7 @@ build/
 dist/
 *.egg-info/
 src/*.egg-info/
+
+# editor / tooling
+.vscode/
+CLAUDE.md

From 81c7cadef8a8da892eaf4f6c16ba47bb3ca8a003 Mon Sep 17 00:00:00 2001
From: Wouter Wakker <wouter.wakker@outlook.com>
Date: Sat, 6 Jun 2026 13:46:38 +0200
Subject: [PATCH 10/15] Stop tracking test.py and .DS_Store

---
 .DS_Store  | Bin 6148 -> 0 bytes
 .gitignore |   2 ++
 test.py    |   3 ---
 3 files changed, 2 insertions(+), 3 deletions(-)
 delete mode 100644 .DS_Store
 delete mode 100644 test.py

diff --git a/.DS_Store b/.DS_Store
deleted file mode 100644
index e393e44e60e428aaa0bb5c3151f1dbb932e86b57..0000000000000000000000000000000000000000
GIT binary patch
literal 0
HcmV?d00001

literal 6148
zcmeHKL2uJA6n<_4O&ODVVA3u~k+_c1G7TYd>AF^95(8;LZ~#=ArEC#dTs7$`R8`6u
z{sVu3D}Mz33n%!V?T)71u7HqVWq;1|_ni21@>~;<Xph4tQH_XP6h`L;s!NRfxvW^p
zc-DYIjj>N%dP2`=pXw`Y>#zz~1^zV!_}jfnd-Rxg=_$V3-{0>cbq->A9K<pJ2MI!a
z0@i;jQ`2xLB1uE~P~>5xD|o>q9h&jJ2E#Z?vqs~)D3zUS*DG$tU31^~r*h(F{cN1J
z`=eKU^<2t0n78}Ei)b+E)o$&{B<n}XV59<~Z~)5dmr)YRNn4JSFjcX>?r^JawO8Am
zPVet*Z+Q=!vn_AhY&3Szey}~8Ro#s{_Z}Vf-p6N&{HSI|0{KkYy1_ZH1{d=V)8jb6
z^Ks{e^V}aLu}n_j<3ZNvA|WLX=mfvJ@Fet(<dSS(z+RNMf>e5drRic}eCih2;>QxZ
zN=IPsQil$KbvmQ47*XOUS{%PjBRo-BWQp5LESFNSG8$u{>#K`h`nJTcfQwO4U&2<e
z#^;4s2*K9=mICG-cR2@GvkF)RtOAP)aDMPm7+VH&jVg7ZQbz!w1Gh4?`7FUPp23#E
zTqAm5LWcr%s4zzip~K<#46kJ{*Qmoun8SxKcNXS^BGlb6zNhFUS{iL_6|f2v6)5Y=
z8t?zztIz+%AbVvMunPQF3J9m;ciNbe*;`j8$9t`Z@)m`SdAUXvg34^is=-_FCW<oj
ZdE5ZD4CWf)f!RL-N(Nh51^%c4zW^arvjYGC

diff --git a/.gitignore b/.gitignore
index 5e78aaa..cafa253 100644
--- a/.gitignore
+++ b/.gitignore
@@ -10,6 +10,8 @@
 **/__pycache__/
 **/.temp_data/
 test.py
+.DS_Store
+**/.DS_Store
 
 # packaging build artifacts
 build/
diff --git a/test.py b/test.py
deleted file mode 100644
index 06c3584..0000000
--- a/test.py
+++ /dev/null
@@ -1,3 +0,0 @@
-from iotables import OECD
-
-io = OECD(version='2025-extended', year=2022, refresh=True)

From b26f4de07ffa26e327cf5ca0b7741b2dc74d5865 Mon Sep 17 00:00:00 2001
From: Wouter Wakker <wouter.wakker@outlook.com>
Date: Sat, 6 Jun 2026 14:27:32 +0200
Subject: [PATCH 11/15] Bug fixes

---
 CHANGELOG.MD             |  8 ++++++-
 README.md                | 10 ++++----
 pyproject.toml           | 14 +++++------
 src/iotables/base_io.py  | 13 ++++++----
 src/iotables/exiobase.py | 12 +++++-----
 src/iotables/figaro.py   |  7 +++---
 src/iotables/matrix.py   |  6 +++--
 src/iotables/oecd.py     |  4 +---
 src/iotables/utils.py    | 19 +++++++++++----
 tests/test_figaro.py     |  2 +-
 tests/test_iotables.py   | 51 ++++++++++++++++++++++++++++++++++++++++
 tests/test_matrix.py     |  2 +-
 tests/test_oecd.py       |  2 +-
 13 files changed, 111 insertions(+), 39 deletions(-)

diff --git a/CHANGELOG.MD b/CHANGELOG.MD
index f582eed..1901d03 100644
--- a/CHANGELOG.MD
+++ b/CHANGELOG.MD
@@ -4,7 +4,7 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](http://keepachangelog.com/)
 and this project adheres to [Semantic Versioning](http://semver.org/).
 
-## [Unreleased] - yyyy-mm-dd
+## [0.1.0] - 2026-06-06
 
 ### Added
 - Introduce structure for different dataset-specific tools
@@ -16,8 +16,14 @@ and this project adheres to [Semantic Versioning](http://semver.org/).
 - Downloads now use `curl_cffi` (browser TLS impersonation) so Cloudflare-protected OECD files work
 - Re-pointed OECD/Figaro download URLs to current hosts (OECD `webfs`, Figaro CIRCABC)
 - Cached downloads now live in a user cache dir (`~/.cache/iotables`, override with `IOTABLES_DATA`)
+- User-facing argument validation now raises `ValueError` instead of `assert` (so it survives `python -O`)
 ### Fixed
 - OECD `2025-regular` file-id regex and in-zip CSV filenames; cross-edition cache-id collisions
+- `remove_downloaded_files(database=...)` deleted the wrong database's files
+- `ExioBase` download-failure handler raised `AttributeError` instead of the intended warning
+- Loader `db_name` derivation used `str.rstrip('.py')` (a character set, not a suffix)
+- Dropped `lru_cache` on `_load_data`, which pinned every loader instance (and its data) in memory
+- Downloads are now atomic (written to a `.part` file and moved into place on completion), so an interrupted download no longer leaves a truncated file in the cache
 ___
 
 
diff --git a/README.md b/README.md
index b3e9a25..c91e530 100644
--- a/README.md
+++ b/README.md
@@ -20,8 +20,9 @@ pip install iotables
 ## Structure of input-output data
 
 Input-data model the economy in matrix form. Here, we explain input-output data using OECD data as an example.
-THE OECD input-output tables contain inputs and outputs in current million USD for 45 sectors and 66 countries as well 
-as rest-of-world. China and Mexico data are split into CN1, CN2, MX1, and MX2.
+The OECD input-output tables contain inputs and outputs in current million USD for a set of sectors and countries (the
+exact counts depend on the edition; see `config.py`) as well as rest-of-world. China and Mexico data are split into CN1,
+CN2, MX1, and MX2.
 
 ![OECD sctructure](res/OECD_structure.png)
 *Source*: OECD
@@ -173,7 +174,8 @@ This is the Ghosh equation, which defines the relation between production output
 ### Create an instance of the OECD class
 
 Creating an instance of the OECD class loads the OECD data and gives access to convenient attributes and methods. 
-An instance can be create as follows, specifiying a year between 1995 and 2018.
+An instance can be created as follows, specifying an available year for the chosen version (e.g. 1995-2018 for the
+`2021` version, up to 2022 for the `2025` versions).
 
 ```python
 import iotables 
@@ -227,7 +229,7 @@ import iotables
 
 oecd = iotables.OECD(version='2021', year=2018)
 
-df_l = oecd.leontief_demand_shock(shock=-10, regions=['FR', 'DE], sectors=['01T02', '35'])
+df_l = oecd.leontief_demand_shock(shock=-10, regions=['FR', 'DE'], sectors=['01T02', '35'])
 df_g = oecd.ghosh_supply_shock(shock=-10, regions=['FR', 'DE'], sectors=['01T02', '35'])
 ```
 
diff --git a/pyproject.toml b/pyproject.toml
index 7e462b9..858c3be 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -4,18 +4,18 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "iotables"
-version = "0.0.1"
+version = "0.1.0"
 description = "A Python package to easily load and work with inter-country input-output data"
 readme = "README.md"
 license = { file = "LICENSE" }
 authors = [{ name = "Wouter Wakker", email = "wouter.wakker@outlook.com" }]
-requires-python = ">=3.8"
+requires-python = ">=3.9"
 dependencies = [
-    "curl_cffi",
-    "numpy",
-    "pandas>=1.1.5",
-    "matplotlib",
-    "tqdm",
+    "curl_cffi>=0.5",
+    "numpy>=1.21,<3",
+    "pandas>=1.3",
+    "matplotlib>=3.4",
+    "tqdm>=4.0",
 ]
 classifiers = [
     "Programming Language :: Python :: 3",
diff --git a/src/iotables/base_io.py b/src/iotables/base_io.py
index de177a8..4dda7b6 100644
--- a/src/iotables/base_io.py
+++ b/src/iotables/base_io.py
@@ -53,7 +53,7 @@ def __init__(self):
                         self.Z.rows,
                         self.Z.columns)
 
-        self.G = Matrix('Output inverse',
+        self.G = Matrix('Ghosh inverse',
                         (np.eye(self.rs) - self.B).I,
                         self.Z.rows,
                         self.Z.columns)
@@ -80,7 +80,8 @@ def _shock(self,
         if custom_shock_vector is not None:
             shock_vector = np.array(custom_shock_vector).reshape(self.rs, 1)
         else:
-            assert shock and regions and sectors, "Must supply parameters: 'shock', 'regions', 'sectors'"
+            if shock is None or regions is None or sectors is None:
+                raise ValueError("Must supply parameters: 'shock', 'regions', 'sectors'")
 
             assert_is_subset(regions, self.regions)
             assert_is_subset(sectors, self.sectors)
@@ -171,7 +172,8 @@ def _plot_shock(self,
         Returns:
             fig, ax
         """
-        assert by in {'region', 'sector'}, "plot_by must be 'region' or 'sector'"
+        if by not in {'region', 'sector'}:
+            raise ValueError("plot_by must be 'region' or 'sector'")
         assert_is_subset(regions, self.regions)
 
         df = self._shock_to_df(x_new)
@@ -255,7 +257,7 @@ def ghosh_supply_shock(self,
         """Executes a Ghosh supply shock
 
         Args:
-            shock: Shock in percentage of original demand
+            shock: Shock in percentage of original primary inputs (value added)
             regions: List of regions to be shocked
             sectors: List of sectors to be shocked
             custom_shock_vector: Vector of length regions * sectors with percentage shocks, overrides all other shock
@@ -297,7 +299,8 @@ def get_imports_exports(self,
         Returns:
             float: Sum of trade flow from exporting region-sectors to importing region-sectors
         """
-        assert use_type in {'intermediate', 'final', 'both'}, "use_type must be 'intermediate', 'final' or 'both'"
+        if use_type not in {'intermediate', 'final', 'both'}:
+            raise ValueError("use_type must be 'intermediate', 'final' or 'both'")
 
         if import_sectors is not None and use_type in {'final', 'both'}:
             warn('Note that import_sectors only apply to intermediate use, for final use only import_regions is used')
diff --git a/src/iotables/exiobase.py b/src/iotables/exiobase.py
index 5ac2435..7f1ded3 100644
--- a/src/iotables/exiobase.py
+++ b/src/iotables/exiobase.py
@@ -4,7 +4,6 @@
 """
 
 from iotables.matrix import Matrix
-from functools import lru_cache
 import numpy as np
 import pandas as pd
 from tqdm import tqdm
@@ -17,7 +16,7 @@
 from iotables.globals import DATA_FOLDER, FILES_LOG
 from iotables.utils import remove_downloaded_files, download_file
 
-db_name = os.path.basename(__file__).rstrip('.py')
+db_name = os.path.splitext(os.path.basename(__file__))[0]
 
 
 def process_df(df):
@@ -48,7 +47,8 @@ def __init__(self,
             verify: Verify the server's TLS certificate (``False`` to skip, or a CA bundle path)
         """
 
-        assert kind in {'industry-by-industry', 'product-by-product'}
+        if kind not in {'industry-by-industry', 'product-by-product'}:
+            raise ValueError("kind must be 'industry-by-industry' or 'product-by-product'")
 
         if version not in config['exiobase'].keys():
             raise ValueError(
@@ -100,7 +100,8 @@ def __init__(self,
 
             # Create matrices
             pbar.set_description('Creating matrices...')
-            assert self._Z_raw.shape[0] == self._Z_raw.shape[1]
+            if self._Z_raw.shape[0] != self._Z_raw.shape[1]:
+                raise ValueError('Intermediate-use matrix Z is not square; the downloaded file may be corrupt')
             self.rs = config['exiobase'][version]['num_regions'][kind] * config['exiobase'][version]['num_sectors'][kind]
             self.Z = Matrix('Intermediate use',
                             *process_df(self._Z_raw))
@@ -143,7 +144,6 @@ def __init__(self,
             pbar.update()
             pbar.set_description('Done')
 
-    @lru_cache()
     def _load_data(self):
         folder = f'IOT_{self.year}_{"ixi" if self.kind == "industry-by-industry" else "pxp"}'
         with ZipFile(self._data_file, 'r') as zf:
@@ -183,7 +183,7 @@ def _download_data(self):
                 files_log.write(db_name + ';' + self._data_file + '\n')
         except Exception as e:
             warn(f"Couldn't download the data. Try downloading manually from {self._url} "
-                 f"and save the csv file as {self._file_id}.csv in {self._data_folder}")
+                 f"and save the zip file as {self._file_id}.zip in {DATA_FOLDER}")
             raise e
 
     @staticmethod
diff --git a/src/iotables/figaro.py b/src/iotables/figaro.py
index d63df0a..cfe25a7 100644
--- a/src/iotables/figaro.py
+++ b/src/iotables/figaro.py
@@ -5,7 +5,6 @@
 """
 from iotables.mappings import figaro_sector_name_mapping_pxp_2022, figaro_sector_name_mapping_ixi_2022, figaro_demand_items
 from iotables.matrix import Matrix
-from functools import lru_cache
 import numpy as np
 import pandas as pd
 from tqdm import tqdm
@@ -17,7 +16,7 @@
 from iotables.globals import DATA_FOLDER, FILES_LOG
 from iotables.utils import remove_downloaded_files, download_file
 
-db_name = os.path.basename(__file__).rstrip('.py')
+db_name = os.path.splitext(os.path.basename(__file__))[0]
 
 
 def process_df(df):
@@ -56,7 +55,8 @@ def __init__(self,
                    https) or a ``{scheme: url}`` dict
             verify: Verify the server's TLS certificate (``False`` to skip, or a CA bundle path)
         """
-        assert kind in {'industry-by-industry', 'product-by-product'}
+        if kind not in {'industry-by-industry', 'product-by-product'}:
+            raise ValueError("kind must be 'industry-by-industry' or 'product-by-product'")
 
         if version not in config['figaro'].keys():
             raise ValueError(
@@ -140,7 +140,6 @@ def __init__(self,
             pbar.update()
             pbar.set_description('Done')
 
-    @lru_cache()
     def _load_data(self):
         df = pd.read_csv(self._data_file, index_col=0)
         return df
diff --git a/src/iotables/matrix.py b/src/iotables/matrix.py
index 5dbd21e..cc10f67 100644
--- a/src/iotables/matrix.py
+++ b/src/iotables/matrix.py
@@ -11,8 +11,10 @@ class Matrix(np.ndarray):
 
     def __new__(cls, info, input_array, rows, columns):
         obj = np.asarray(input_array).view(cls)
-        assert len(obj.shape) == 2, "Array must be 2-dimensional"
-        assert obj.shape == (len(rows), len(columns)), "Rows and columns do not have the shape of the array"
+        if len(obj.shape) != 2:
+            raise ValueError("Array must be 2-dimensional")
+        if obj.shape != (len(rows), len(columns)):
+            raise ValueError("Rows and columns do not have the shape of the array")
         obj.info = info
         obj.rows = rows
         obj.columns = columns
diff --git a/src/iotables/oecd.py b/src/iotables/oecd.py
index 5991410..f891855 100644
--- a/src/iotables/oecd.py
+++ b/src/iotables/oecd.py
@@ -6,7 +6,6 @@
 from iotables.mappings import oecd_sector_name_mapping, oecd_sector_name_mapping_2025, oecd_demand_items, \
     oecd_sector_2022_2021_mapping
 from iotables.matrix import Matrix
-from functools import lru_cache
 import numpy as np
 import pandas as pd
 from iotables.utils import ALPHA3_TO_ALPHA2
@@ -21,7 +20,7 @@
 from warnings import warn
 from functools import partial
 
-db_name = os.path.basename(__file__).rstrip('.py')
+db_name = os.path.splitext(os.path.basename(__file__))[0]
 
 
 def process_df(df):
@@ -147,7 +146,6 @@ def __init__(self,
             pbar.update()
             pbar.set_description('Done')
 
-    @lru_cache()
     def _load_data(self):
         if self.version == '2021':
             filename = f'ICIO2021_{self.year}.csv'
diff --git a/src/iotables/utils.py b/src/iotables/utils.py
index 915382d..88fe8a0 100644
--- a/src/iotables/utils.py
+++ b/src/iotables/utils.py
@@ -41,9 +41,20 @@ def download_file(url, dest, proxy=None, verify=True):
     r = requests.get(url, stream=True, impersonate='chrome', verify=verify, **kwargs)
     if not r.ok:
         raise ConnectionError(r.reason or f'HTTP {r.status_code}')
-    with open(dest, 'wb') as f:
-        for chunk in r.iter_content():
-            f.write(chunk)
+
+    # Download to a temporary file and atomically move it into place only once the
+    # stream completes, so an interrupted download never leaves a truncated file in
+    # the cache (which would otherwise load as a wrong-shaped, silently corrupt matrix).
+    tmp = dest + '.part'
+    try:
+        with open(tmp, 'wb') as f:
+            for chunk in r.iter_content():
+                f.write(chunk)
+        os.replace(tmp, dest)
+    except BaseException:
+        if os.path.exists(tmp):
+            os.remove(tmp)
+        raise
 
 
 def replace_if_exists(x, mapping):
@@ -86,7 +97,7 @@ def remove_downloaded_files(database: str = 'all',
             print(f'no files found for {database}, only for {list(files.keys())}')
             return
         other_files = {k: v for k, v in files.items() if k != database}
-        files = files[db]
+        files = files[database]
     else:
         files = {item for sublist in files.values() for item in sublist}
     for path in files:
diff --git a/tests/test_figaro.py b/tests/test_figaro.py
index 6371edf..76222c6 100644
--- a/tests/test_figaro.py
+++ b/tests/test_figaro.py
@@ -52,7 +52,7 @@ def test_shock(self):
                               ((np.eye(f.rs) - f.A).I @ (
                                       f.FD * (custom_shock_vector / 100))) + f.X)
 
-        with pytest.raises(AssertionError):
+        with pytest.raises(ValueError):
             f._shock(model='leontief')
 
         shock_vector = np.array([-.1 if r in EA and s == 'A01' else 0 for r, s in f.X.rows]).reshape(-1, 1)
diff --git a/tests/test_iotables.py b/tests/test_iotables.py
index c0110eb..a4cfc36 100644
--- a/tests/test_iotables.py
+++ b/tests/test_iotables.py
@@ -3,7 +3,9 @@
 **Authors**: W. Wakker
 
 """
+import os
 import iotables
+from iotables import utils
 
 
 class TestIotables:
@@ -13,3 +15,52 @@ def test1(self):
 
     def test_folder_size(self):
         assert isinstance(iotables.get_size_data_folder(), str)
+
+
+class TestRemoveDownloadedFiles:
+    """Offline tests for remove_downloaded_files (no network)."""
+
+    def _setup_log(self, tmp_path, monkeypatch):
+        oecd_file = tmp_path / 'oecd.zip'
+        figaro_file = tmp_path / 'figaro.csv'
+        oecd_file.write_text('x')
+        figaro_file.write_text('y')
+        log = tmp_path / '_files_log.txt'
+        # Order matters: oecd is written first so the last line is figaro. The
+        # earlier wrong-database bug used the leftover loop variable (last line),
+        # which would have made remove(database='oecd') target figaro instead.
+        log.write_text(f'oecd;{oecd_file}\nfigaro;{figaro_file}\n')
+        monkeypatch.setattr(utils, 'FILES_LOG', str(log))
+        return oecd_file, figaro_file, log
+
+    def test_remove_single_database_keeps_others(self, tmp_path, monkeypatch):
+        oecd_file, figaro_file, log = self._setup_log(tmp_path, monkeypatch)
+
+        utils.remove_downloaded_files(database='oecd', verbose=False)
+
+        assert not oecd_file.exists()       # requested database removed
+        assert figaro_file.exists()         # other database untouched
+        assert log.exists()                 # log rewritten, not deleted
+        assert log.read_text().strip() == f'figaro;{figaro_file}'
+
+    def test_remove_all(self, tmp_path, monkeypatch):
+        oecd_file, figaro_file, log = self._setup_log(tmp_path, monkeypatch)
+
+        utils.remove_downloaded_files(database='all', verbose=False)
+
+        assert not oecd_file.exists()
+        assert not figaro_file.exists()
+        assert not log.exists()
+
+    def test_remove_unknown_database_is_noop(self, tmp_path, monkeypatch):
+        oecd_file, figaro_file, log = self._setup_log(tmp_path, monkeypatch)
+
+        utils.remove_downloaded_files(database='nope', verbose=False)
+
+        assert oecd_file.exists()
+        assert figaro_file.exists()
+        assert log.exists()
+
+    def test_no_log_is_noop(self, tmp_path, monkeypatch):
+        monkeypatch.setattr(utils, 'FILES_LOG', str(tmp_path / 'missing.txt'))
+        utils.remove_downloaded_files(database='all', verbose=False)  # must not raise
diff --git a/tests/test_matrix.py b/tests/test_matrix.py
index 9f3df9e..b2971af 100644
--- a/tests/test_matrix.py
+++ b/tests/test_matrix.py
@@ -39,7 +39,7 @@ def test1(self):
             Matrix('something', *process_df(df))
 
     def test2(self):
-        with pytest.raises(AssertionError):
+        with pytest.raises(ValueError):
             Matrix('something', [1, 2, 3], ['something'], ['something'])
 
     def testI(self):
diff --git a/tests/test_oecd.py b/tests/test_oecd.py
index 8762427..ad12d85 100644
--- a/tests/test_oecd.py
+++ b/tests/test_oecd.py
@@ -58,7 +58,7 @@ def test_shock(self):
                               ((np.eye(oecd.rs) - oecd.A).I @ (
                                       oecd.FD * (custom_shock_vector / 100))) + oecd.X)
 
-        with pytest.raises(AssertionError):
+        with pytest.raises(ValueError):
             oecd._shock(model='leontief')
 
         shock_vector = np.array([-.1 if r in EA and s == '35' else 0 for r, s in oecd.X.rows]).reshape(-1, 1)

From 5cc31cd4778ce874fda9ecae3023e8c7cd1d2548 Mon Sep 17 00:00:00 2001
From: Wouter Wakker <wouter.wakker@outlook.com>
Date: Sat, 6 Jun 2026 15:05:20 +0200
Subject: [PATCH 12/15] Harden download integrity, add CI, and polish packaging

- download_file: verify body size against Content-Length (when present and
  unencoded) and raise on a truncated download; offline tests added
- tests: move loaders into fixtures and mark network tests so the suite collects
  and the offline tier runs without a connection (pytest -m "not network")
- CI: GitHub Actions running offline tests on Python 3.9-3.13 + build/twine check
- packaging: PEP 639 SPDX license (clean License-Expression: MIT), py.typed,
  [project.urls], Python-version classifiers; drop stale coverage omit
- cleanups: real module docstrings, IO attribute contract includes
  FD_GRAN/reference/contact, .sum() instead of .sum(0), drop unused EXIOBASE
  metadata load and the del os half-measure
---
 .github/workflows/ci.yml | 44 ++++++++++++++++++++++
 CHANGELOG.MD             |  4 ++
 pyproject.toml           | 24 ++++++++++--
 src/iotables/__init__.py |  6 +--
 src/iotables/base_io.py  | 11 +++---
 src/iotables/config.py   |  6 +--
 src/iotables/exiobase.py | 17 +++------
 src/iotables/figaro.py   |  8 +---
 src/iotables/globals.py  |  6 +--
 src/iotables/mappings.py |  6 +--
 src/iotables/matrix.py   |  6 +--
 src/iotables/oecd.py     |  8 +---
 src/iotables/py.typed    |  0
 src/iotables/utils.py    | 19 +++++++---
 tests/test_exiobase.py   |  8 ++--
 tests/test_figaro.py     | 37 +++++++++---------
 tests/test_iotables.py   | 81 +++++++++++++++++++++++++++++++++++++---
 tests/test_matrix.py     | 14 +++----
 tests/test_oecd.py       | 37 +++++++++---------
 19 files changed, 226 insertions(+), 116 deletions(-)
 create mode 100644 .github/workflows/ci.yml
 create mode 100644 src/iotables/py.typed

diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
new file mode 100644
index 0000000..20a2e1d
--- /dev/null
+++ b/.github/workflows/ci.yml
@@ -0,0 +1,44 @@
+name: CI
+
+on:
+  push:
+    branches: [main]
+  pull_request:
+  workflow_dispatch:
+
+jobs:
+  test:
+    name: Offline tests (Python ${{ matrix.python-version }})
+    runs-on: ubuntu-latest
+    strategy:
+      fail-fast: false
+      matrix:
+        python-version: ["3.9", "3.10", "3.11", "3.12", "3.13"]
+    steps:
+      - uses: actions/checkout@v4
+      - uses: actions/setup-python@v5
+        with:
+          python-version: ${{ matrix.python-version }}
+      - name: Install package with dev deps
+        run: |
+          python -m pip install --upgrade pip
+          pip install -e ".[dev]"
+      - name: Run offline tests
+        # Network tests download hundreds of MB of real data; they are excluded from CI
+        # and run manually. Everything else must pass on every supported Python version.
+        run: pytest -m "not network"
+
+  build:
+    name: Build & check distribution
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.12"
+      - name: Build sdist and wheel
+        run: |
+          python -m pip install --upgrade pip build twine
+          python -m build
+      - name: Validate metadata and README rendering
+        run: twine check dist/*
diff --git a/CHANGELOG.MD b/CHANGELOG.MD
index 1901d03..471f7ce 100644
--- a/CHANGELOG.MD
+++ b/CHANGELOG.MD
@@ -24,6 +24,10 @@ and this project adheres to [Semantic Versioning](http://semver.org/).
 - Loader `db_name` derivation used `str.rstrip('.py')` (a character set, not a suffix)
 - Dropped `lru_cache` on `_load_data`, which pinned every loader instance (and its data) in memory
 - Downloads are now atomic (written to a `.part` file and moved into place on completion), so an interrupted download no longer leaves a truncated file in the cache
+- `download_file` now verifies the downloaded size against `Content-Length` (when advertised and unencoded) and raises on a truncated body
+### Other
+- Added a `py.typed` marker, `[project.urls]`, Python-version classifiers, and a GitHub Actions CI (offline tests across Python 3.9-3.13 + build/twine check)
+- Network tests are marked `network` and moved into fixtures, so the suite collects and the offline tests run without a connection (`pytest -m "not network"`)
 ___
 
 
diff --git a/pyproject.toml b/pyproject.toml
index 858c3be..612820b 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -1,5 +1,5 @@
 [build-system]
-requires = ["setuptools>=64", "wheel"]
+requires = ["setuptools>=77"]
 build-backend = "setuptools.build_meta"
 
 [project]
@@ -7,7 +7,8 @@ name = "iotables"
 version = "0.1.0"
 description = "A Python package to easily load and work with inter-country input-output data"
 readme = "README.md"
-license = { file = "LICENSE" }
+license = "MIT"
+license-files = ["LICENSE"]
 authors = [{ name = "Wouter Wakker", email = "wouter.wakker@outlook.com" }]
 requires-python = ">=3.9"
 dependencies = [
@@ -19,19 +20,34 @@ dependencies = [
 ]
 classifiers = [
     "Programming Language :: Python :: 3",
+    "Programming Language :: Python :: 3.9",
+    "Programming Language :: Python :: 3.10",
+    "Programming Language :: Python :: 3.11",
+    "Programming Language :: Python :: 3.12",
+    "Programming Language :: Python :: 3.13",
     "Operating System :: OS Independent",
-    "License :: OSI Approved :: MIT License",
 ]
 
+[project.urls]
+Homepage = "https://github.com/WWakker/iotables"
+Repository = "https://github.com/WWakker/iotables"
+Changelog = "https://github.com/WWakker/iotables/blob/main/CHANGELOG.MD"
+
 [project.optional-dependencies]
 dev = ["pytest", "pytest-cov"]
 
 [tool.setuptools.packages.find]
 where = ["src"]
 
+[tool.setuptools.package-data]
+iotables = ["py.typed"]
+
 [tool.pytest.ini_options]
 testpaths = ["tests"]
+markers = [
+    "network: test downloads real data over the network (deselect with '-m \"not network\"')",
+]
 
 [tool.coverage.run]
 source = ["iotables"]
-omit = ["**/adjust_source_in_covxml.py", "**/update_badge.py"]
+omit = ["**/adjust_source_in_covxml.py"]
diff --git a/src/iotables/__init__.py b/src/iotables/__init__.py
index 1e7c8b2..7030d56 100644
--- a/src/iotables/__init__.py
+++ b/src/iotables/__init__.py
@@ -1,4 +1,4 @@
-import os
+import os as _os
 from importlib.metadata import version as _version, PackageNotFoundError as _PackageNotFoundError
 from iotables.globals import DATA_FOLDER as __DATA_FOLDER
 from iotables.globals import IS_WINDOWS as __IS_WINDOWS
@@ -40,6 +40,4 @@ def human(size):
     return get_size(__DATA_FOLDER)
 
 
-os.makedirs(__DATA_FOLDER, exist_ok=True)
-
-del os
+_os.makedirs(__DATA_FOLDER, exist_ok=True)
diff --git a/src/iotables/base_io.py b/src/iotables/base_io.py
index 4dda7b6..2c111ce 100644
--- a/src/iotables/base_io.py
+++ b/src/iotables/base_io.py
@@ -1,8 +1,4 @@
-"""  Created on 18/10/2022::
-------------- io -------------
-**Authors**: W. Wakker
-
-"""
+"""Shared analysis engine: derives the coefficient/inverse matrices and runs shocks."""
 from warnings import warn
 from iotables.matrix import Matrix
 from iotables.utils import assert_is_subset
@@ -22,6 +18,7 @@ def __init__(self):
                            'X',
                            'V',
                            'FD',
+                           'FD_GRAN',
                            'ADD',
                            'FD_REGION',
                            'rs',
@@ -29,7 +26,9 @@ def __init__(self):
                            'regions',
                            'sectors',
                            'unit',
-                           'demand_items']
+                           'demand_items',
+                           'reference',
+                           'contact']
         assert_is_subset(necessary_attrs, dir(self))
 
         # Coefficients matrix, replace 0 with 1 to allow inversion
diff --git a/src/iotables/config.py b/src/iotables/config.py
index 1867450..b6686dc 100644
--- a/src/iotables/config.py
+++ b/src/iotables/config.py
@@ -1,8 +1,4 @@
-"""  Created on 08/10/2022::
-------------- config -------------
-**Authors**: W. Wakker
-
-"""
+"""Single source of truth for the data that exists: download links and shapes per database."""
 
 
 def _chunk_links(base, chunks):
diff --git a/src/iotables/exiobase.py b/src/iotables/exiobase.py
index 7f1ded3..e043cf0 100644
--- a/src/iotables/exiobase.py
+++ b/src/iotables/exiobase.py
@@ -1,7 +1,4 @@
-"""  Created on 15/11/2022::
-------------- exiobase -------------
-**Authors**: S. Boldrini
-"""
+"""Loader for EXIOBASE inter-country input-output data."""
 
 from iotables.matrix import Matrix
 import numpy as np
@@ -79,7 +76,7 @@ def __init__(self,
             # Load
             pbar.set_description('Loading data...')
             self.df = None
-            self._Z_raw, self._FD_raw, self._X_raw, self._metadata, self._sector_codes, self._FD_codes = self._load_data()
+            self._Z_raw, self._FD_raw, self._X_raw, self._sector_codes, self._FD_codes = self._load_data()
 
             exiobase_sector_name_mapping = self._sector_codes.reset_index().set_index('CodeNr')['Name'].to_dict()
             exiobase_FD_name_mapping = self._FD_codes.reset_index().set_index('CodeNr')['Name'].to_dict()
@@ -127,10 +124,11 @@ def __init__(self,
                                      columns=[r for r, s in self.FD_GRAN.columns]).T
             fd_region.index.name = 'region'
 
+            fd_region = fd_region.groupby('region').sum().T
             self.FD_REGION = Matrix('Final demand by region',
-                                    fd_region.groupby('region').sum(0).T,
+                                    fd_region,
                                     rows=self.Z.rows,
-                                    columns=fd_region.groupby('region').sum(0).T.columns.to_list())
+                                    columns=fd_region.columns.to_list())
 
             self.regions = list(sorted(np.unique([r for r, s in self.Z.rows])))
             self.sectors = list(sorted(np.unique([s for r, s in self.Z.rows])))
@@ -171,10 +169,7 @@ def _load_data(self):
             with zf.open(f'{folder}/finaldemands.txt', 'r') as csv_file:
                 FD_codes = pd.read_csv(csv_file, sep='\t', index_col=1)
 
-            with zf.open(f'{folder}/metadata.json', 'r') as json_file:
-                metadata = pd.read_json(json_file)
-
-        return z_raw, fd_raw, x_raw, metadata, sector_codes, FD_codes
+        return z_raw, fd_raw, x_raw, sector_codes, FD_codes
 
     def _download_data(self):
         try:
diff --git a/src/iotables/figaro.py b/src/iotables/figaro.py
index cfe25a7..1868f7b 100644
--- a/src/iotables/figaro.py
+++ b/src/iotables/figaro.py
@@ -1,8 +1,4 @@
-"""  Created on 14/10/2022::
-------------- figaro -------------
-**Authors**: W. Wakker
-
-"""
+"""Loader for Eurostat Figaro inter-country input-output data."""
 from iotables.mappings import figaro_sector_name_mapping_pxp_2022, figaro_sector_name_mapping_ixi_2022, figaro_demand_items
 from iotables.matrix import Matrix
 import numpy as np
@@ -121,7 +117,7 @@ def __init__(self,
             # Create region level FD
             fd_region = pd.DataFrame(self.FD_GRAN, columns=[r for r, s in self.FD_GRAN.columns]).T
             fd_region.index.name = 'region'
-            fd_region = fd_region.groupby('region').sum(0).T
+            fd_region = fd_region.groupby('region').sum().T
             self.FD_REGION = Matrix('Final demand by region',
                                     fd_region,
                                     rows=self.Z.rows,
diff --git a/src/iotables/globals.py b/src/iotables/globals.py
index 6f06481..347aa47 100644
--- a/src/iotables/globals.py
+++ b/src/iotables/globals.py
@@ -1,8 +1,4 @@
-"""  Created on 22/11/2022::
-------------- globals -------------
-**Authors**: W. Wakker
-
-"""
+"""Module-level constants: the download cache location and bookkeeping paths."""
 import os
 
 # Downloaded source files are cached in a user-level cache directory rather than inside
diff --git a/src/iotables/mappings.py b/src/iotables/mappings.py
index c482d57..b70c7d8 100644
--- a/src/iotables/mappings.py
+++ b/src/iotables/mappings.py
@@ -1,8 +1,4 @@
-"""  Created on 06/09/2022::
-------------- mappings -------------
-**Authors**: W. Wakker
-
-"""
+"""Per-database sector-code → human-name maps, demand-item lists, and the OECD 2022→2021 remap."""
 
 oecd_sector_2022_2021_mapping = {'A01_02': '01T02',
                                  'A03': '03',
diff --git a/src/iotables/matrix.py b/src/iotables/matrix.py
index cc10f67..b9c48cb 100644
--- a/src/iotables/matrix.py
+++ b/src/iotables/matrix.py
@@ -1,8 +1,4 @@
-"""  Created on 03/10/2022::
-------------- matrix -------------
-**Authors**: W. Wakker
-
-"""
+"""A labelled 2-D ``numpy.ndarray`` subclass carrying region/sector row and column metadata."""
 import numpy as np
 import pandas as pd
 
diff --git a/src/iotables/oecd.py b/src/iotables/oecd.py
index f891855..44ad4a0 100644
--- a/src/iotables/oecd.py
+++ b/src/iotables/oecd.py
@@ -1,8 +1,4 @@
-"""  Created on 06/09/2022::
-------------- oecd -------------
-**Authors**: W. Wakker
-
-"""
+"""Loader for OECD ICIO inter-country input-output data."""
 from iotables.mappings import oecd_sector_name_mapping, oecd_sector_name_mapping_2025, oecd_demand_items, \
     oecd_sector_2022_2021_mapping
 from iotables.matrix import Matrix
@@ -126,7 +122,7 @@ def __init__(self,
             # Create region level FD
             fd_region = pd.DataFrame(self.FD_GRAN, columns=[r for r, s in self.FD_GRAN.columns]).T
             fd_region.index.name = 'region'
-            fd_region = fd_region.groupby('region').sum(0).T
+            fd_region = fd_region.groupby('region').sum().T
             self.FD_REGION = Matrix('Final demand by region',
                                     fd_region,
                                     rows=self.Z.rows,
diff --git a/src/iotables/py.typed b/src/iotables/py.typed
new file mode 100644
index 0000000..e69de29
diff --git a/src/iotables/utils.py b/src/iotables/utils.py
index 88fe8a0..a596715 100644
--- a/src/iotables/utils.py
+++ b/src/iotables/utils.py
@@ -1,8 +1,4 @@
-"""  Created on 11/10/2022::
-------------- utils -------------
-**Authors**: W. Wakker
-
-"""
+"""Shared helpers: validation, the downloader, cache cleanup, and country-code maps."""
 from iotables.globals import FILES_LOG
 from collections import defaultdict
 import os
@@ -47,9 +43,22 @@ def download_file(url, dest, proxy=None, verify=True):
     # the cache (which would otherwise load as a wrong-shaped, silently corrupt matrix).
     tmp = dest + '.part'
     try:
+        written = 0
         with open(tmp, 'wb') as f:
             for chunk in r.iter_content():
                 f.write(chunk)
+                written += len(chunk)
+
+        # Guard against a silently truncated body (server returns 200 then closes the
+        # stream early). Only enforce when the server advertised a length and did not
+        # transform the bytes -- e.g. CIRCABC serves gzip/chunked with no Content-Length,
+        # where the written size legitimately differs from any advertised length.
+        content_encoding = (r.headers.get('Content-Encoding') or '').lower()
+        expected = r.headers.get('Content-Length')
+        if expected is not None and content_encoding in ('', 'identity') and written != int(expected):
+            raise ConnectionError(
+                f'Incomplete download from {url}: got {written} bytes, expected {expected}')
+
         os.replace(tmp, dest)
     except BaseException:
         if os.path.exists(tmp):
diff --git a/tests/test_exiobase.py b/tests/test_exiobase.py
index d7841fd..a285c61 100644
--- a/tests/test_exiobase.py
+++ b/tests/test_exiobase.py
@@ -1,11 +1,9 @@
-"""  Created on 17/11/2022::
-------------- test_exiobase -------------
-**Authors**: W. Wakker
-
-"""
+"""Network tests for the ExioBase loader (download real data; run with the 'network' marker)."""
+import pytest
 from iotables import ExioBase
 
 
+@pytest.mark.network
 class TestExioBase:
 
     def test_load(self):
diff --git a/tests/test_figaro.py b/tests/test_figaro.py
index 76222c6..2c212cf 100644
--- a/tests/test_figaro.py
+++ b/tests/test_figaro.py
@@ -1,21 +1,24 @@
-"""  Created on 19/10/2022::
-------------- test_figaro -------------
-**Authors**: W. Wakker
-
-"""
+"""Network tests for the Figaro loader (download real data; run with the 'network' marker)."""
 import pytest
 from iotables import Figaro
 import numpy as np
 import matplotlib
 import matplotlib.pyplot as plt
 
-f = Figaro(version='2025', year=2018, kind='industry-by-industry')
+EA = ['AT', 'BE', 'CY', 'DE', 'EE', 'ES', 'FI', 'FR', 'GR', 'HR', 'IE', 'IT', 'LT', 'LU', 'LV', 'MT', 'NL', 'PT', 'SI', 'SK']
 
-custom_shock_vector = np.random.uniform(size=f.rs, low=-10, high=10).reshape(-1, 1)
 
-EA = ['AT', 'BE', 'CY', 'DE', 'EE', 'ES', 'FI', 'FR', 'GR', 'HR', 'IE', 'IT', 'LT', 'LU', 'LV', 'MT', 'NL', 'PT', 'SI', 'SK']
+@pytest.fixture(scope="module")
+def f():
+    return Figaro(version='2025', year=2018, kind='industry-by-industry')
+
+
+@pytest.fixture(scope="module")
+def custom_shock_vector(f):
+    return np.random.uniform(size=f.rs, low=-10, high=10).reshape(-1, 1)
 
 
+@pytest.mark.network
 class TestFigaro:
 
     def test_download(self):
@@ -27,24 +30,24 @@ def test_load(self):
         fi = Figaro(version='2025', year=2018, kind='product-by-product')
         assert set(fi.sectors).issubset(fi.sector_name_mapping)
 
-    def test_matrices(self):
+    def test_matrices(self, f):
         for attr in ['Z', 'A', 'B', 'L', 'G', 'V', 'FD', 'X']:
             assert hasattr(f, attr)
             for attr_attr in ['info', 'rows', 'columns', 'I']:
                 assert attr_attr in dir(getattr(f, attr))
 
-    def test_leontief(self):
+    def test_leontief(self, f):
         fd = (np.eye(f.rs) - f.A) @ f.X
         assert np.isclose(f.FD, fd, atol=.001).all()
 
         x = (np.eye(f.rs) - f.A).I @ f.FD
         assert np.isclose(f.X, x, atol=.001).all()
 
-    def test_ghosh(self):
+    def test_ghosh(self, f):
         x = f.G.T @ f.V.T
         assert np.isclose(f.X, x, atol=.001).all()
 
-    def test_shock(self):
+    def test_shock(self, f, custom_shock_vector):
         assert np.array_equal(f._shock(model='ghosh', custom_shock_vector=custom_shock_vector),
                               (f.G.T @ (f.V.T * (custom_shock_vector / 100))) + f.X)
 
@@ -72,17 +75,17 @@ def test_shock(self):
         with pytest.raises(ValueError):
             f._shock(model='leontief', shock=-10, regions=EA + ['something'], sectors=['A01'])
 
-    def test_leontief_shock(self):
+    def test_leontief_shock(self, f):
         assert np.array_equal(
             f.leontief_demand_shock(shock=-10, regions=EA, sectors=['A01']).x_new.values.reshape(-1, 1),
             f._shock(model='leontief', shock=-10, regions=EA, sectors=['A01']))
 
-    def test_ghosh_shock(self):
+    def test_ghosh_shock(self, f):
         assert np.array_equal(
             f.ghosh_supply_shock(shock=-10, regions=EA, sectors=['A01']).x_new.values.reshape(-1, 1),
             f._shock(model='ghosh', shock=-10, regions=EA, sectors=['A01']))
 
-    def test_plot(self):
+    def test_plot(self, f):
         fig, ax = f.ghosh_supply_shock(shock=-10, regions=EA, sectors=['A01'], plot_regions=EA, plot=True, show=False)
         assert isinstance(fig, matplotlib.figure.Figure)
         assert isinstance(ax, plt.Axes)
@@ -101,7 +104,7 @@ def test_plot(self):
             f.ghosh_supply_shock(shock=-10, regions=EA, sectors=['35'],
                                  plot=True, show=True, plot_by='region')
 
-    def test_get_imports_exports(self):
+    def test_get_imports_exports(self, f):
 
         assert np.isclose(f.get_imports_exports(import_regions=['CN'],
                                                 export_regions='AU',
@@ -142,5 +145,5 @@ def test_get_imports_exports(self):
                               import_sectors='A01',
                               use_type='both')
 
-    def test_remove_local_files(self):
+    def test_remove_local_files(self, f):
         f.remove_downloaded_files()
diff --git a/tests/test_iotables.py b/tests/test_iotables.py
index a4cfc36..2107f62 100644
--- a/tests/test_iotables.py
+++ b/tests/test_iotables.py
@@ -1,9 +1,5 @@
-"""  Created on 22/11/2022::
-------------- test_iotables -------------
-**Authors**: W. Wakker
-
-"""
-import os
+"""Offline tests for top-level helpers and the shared downloader."""
+import pytest
 import iotables
 from iotables import utils
 
@@ -64,3 +60,76 @@ def test_remove_unknown_database_is_noop(self, tmp_path, monkeypatch):
     def test_no_log_is_noop(self, tmp_path, monkeypatch):
         monkeypatch.setattr(utils, 'FILES_LOG', str(tmp_path / 'missing.txt'))
         utils.remove_downloaded_files(database='all', verbose=False)  # must not raise
+
+
+class _FakeResponse:
+    def __init__(self, chunks, headers, ok=True, status_code=200):
+        self._chunks = chunks
+        self.headers = headers
+        self.ok = ok
+        self.status_code = status_code
+        self.reason = 'OK'
+
+    def iter_content(self):
+        return iter(self._chunks)
+
+
+class TestDownloadFile:
+    """Offline tests for download_file's atomic + integrity behaviour (no network)."""
+
+    def _patch(self, monkeypatch, response):
+        import curl_cffi
+        monkeypatch.setattr(curl_cffi.requests, 'get', lambda *a, **k: response)
+
+    def test_truncated_body_raises_and_leaves_no_files(self, tmp_path, monkeypatch):
+        # Server advertises 100 bytes but only streams 50: a silent truncation.
+        resp = _FakeResponse([b'x' * 50], {'Content-Length': '100'})
+        self._patch(monkeypatch, resp)
+        dest = tmp_path / 'data.zip'
+
+        with pytest.raises(ConnectionError):
+            utils.download_file('http://example/data.zip', str(dest))
+
+        assert not dest.exists()
+        assert not (tmp_path / 'data.zip.part').exists()
+
+    def test_complete_body_succeeds(self, tmp_path, monkeypatch):
+        resp = _FakeResponse([b'ab', b'cd'], {'Content-Length': '4'})
+        self._patch(monkeypatch, resp)
+        dest = tmp_path / 'data.zip'
+
+        utils.download_file('http://example/data.zip', str(dest))
+
+        assert dest.read_bytes() == b'abcd'
+        assert not (tmp_path / 'data.zip.part').exists()
+
+    def test_length_check_skipped_for_encoded_body(self, tmp_path, monkeypatch):
+        # gzip/chunked (e.g. CIRCABC): advertised length differs from decoded bytes,
+        # so the check must be skipped rather than false-positive.
+        resp = _FakeResponse([b'x' * 50], {'Content-Length': '100', 'Content-Encoding': 'gzip'})
+        self._patch(monkeypatch, resp)
+        dest = tmp_path / 'data.csv'
+
+        utils.download_file('http://example/data.csv', str(dest))
+
+        assert dest.read_bytes() == b'x' * 50
+
+    def test_no_content_length_succeeds(self, tmp_path, monkeypatch):
+        resp = _FakeResponse([b'hello'], {})
+        self._patch(monkeypatch, resp)
+        dest = tmp_path / 'data.csv'
+
+        utils.download_file('http://example/data.csv', str(dest))
+
+        assert dest.read_bytes() == b'hello'
+
+    def test_http_error_raises(self, tmp_path, monkeypatch):
+        resp = _FakeResponse([], {}, ok=False, status_code=504)
+        resp.reason = 'Gateway Timeout'
+        self._patch(monkeypatch, resp)
+        dest = tmp_path / 'data.zip'
+
+        with pytest.raises(ConnectionError):
+            utils.download_file('http://example/data.zip', str(dest))
+
+        assert not dest.exists()
diff --git a/tests/test_matrix.py b/tests/test_matrix.py
index b2971af..c23a7f9 100644
--- a/tests/test_matrix.py
+++ b/tests/test_matrix.py
@@ -1,8 +1,4 @@
-"""  Created on 03/10/2022::
-------------- test_matrix -------------
-**Authors**: W. Wakker
-
-"""
+"""Offline tests for the labelled Matrix subclass."""
 from iotables.matrix import Matrix
 import pandas as pd
 import numpy as np
@@ -13,15 +9,15 @@
 def process_df(df):
     if df.shape[0] > 1 and df.shape[1] > 1:
         return (df,
-                [(ALPHA3_TO_ALPHA2[r] if r in ALPHA3_TO_ALPHA2 else r, s) for r, s in df.index.str.split('_')],
-                [(ALPHA3_TO_ALPHA2[r] if r in ALPHA3_TO_ALPHA2 else r, s) for r, s in df.columns.str.split('_')])
+                [(ALPHA3_TO_ALPHA2[r] if r in ALPHA3_TO_ALPHA2 else r, s) for r, s in df.index.str.split('_', n=1)],
+                [(ALPHA3_TO_ALPHA2[r] if r in ALPHA3_TO_ALPHA2 else r, s) for r, s in df.columns.str.split('_', n=1)])
     elif df.shape[0] == 1:
         return (df,
                 df.index.to_list(),
-                [(ALPHA3_TO_ALPHA2[r] if r in ALPHA3_TO_ALPHA2 else r, s) for r, s in df.columns.str.split('_')])
+                [(ALPHA3_TO_ALPHA2[r] if r in ALPHA3_TO_ALPHA2 else r, s) for r, s in df.columns.str.split('_', n=1)])
     else:
         return (df,
-                [(ALPHA3_TO_ALPHA2[r] if r in ALPHA3_TO_ALPHA2 else r, s) for r, s in df.index.str.split('_')],
+                [(ALPHA3_TO_ALPHA2[r] if r in ALPHA3_TO_ALPHA2 else r, s) for r, s in df.index.str.split('_', n=1)],
                 df.columns.to_list())
 
 
diff --git a/tests/test_oecd.py b/tests/test_oecd.py
index ad12d85..fc75462 100644
--- a/tests/test_oecd.py
+++ b/tests/test_oecd.py
@@ -1,25 +1,28 @@
-"""  Created on 03/10/2022::
-------------- test -------------
-**Authors**: W. Wakker
-
-"""
+"""Network tests for the OECD loader (download real data; run with the 'network' marker)."""
 import pytest
 from iotables import OECD
 import numpy as np
 import matplotlib
 import matplotlib.pyplot as plt
 
-oecd = OECD(version='2021', year=2018)
+EA = ['AT', 'BE', 'CY', 'DE', 'EE', 'ES', 'FI', 'FR', 'GR', 'HR', 'IE', 'IT', 'LT', 'LU', 'LV', 'MT', 'NL', 'PT', 'SI', 'SK']
 
-custom_shock_vector = np.random.uniform(size=oecd.rs, low=-10, high=10).reshape(-1, 1)
 
-EA = ['AT', 'BE', 'CY', 'DE', 'EE', 'ES', 'FI', 'FR', 'GR', 'HR', 'IE', 'IT', 'LT', 'LU', 'LV', 'MT', 'NL', 'PT', 'SI', 'SK']
+@pytest.fixture(scope="module")
+def oecd():
+    return OECD(version='2021', year=2018)
+
+
+@pytest.fixture(scope="module")
+def custom_shock_vector(oecd):
+    return np.random.uniform(size=oecd.rs, low=-10, high=10).reshape(-1, 1)
 
 
+@pytest.mark.network
 class TestOECD:
 
     def test_download(self):
-        o = OECD(version='2021', year=2018, refresh=True)
+        OECD(version='2021', year=2018, refresh=True)
 
     def test_load(self):
         o = OECD(version='2021', year=2018)
@@ -33,24 +36,24 @@ def test_load(self):
         o = OECD(version='2025-regular', year=2022)
         assert set(o.sectors).issubset(o.sector_name_mapping)
 
-    def test_matrices(self):
+    def test_matrices(self, oecd):
         for attr in ['Z', 'A', 'B', 'L', 'G', 'V', 'FD', 'X']:
             assert hasattr(oecd, attr)
             for attr_attr in ['info', 'rows', 'columns', 'I']:
                 assert attr_attr in dir(getattr(oecd, attr))
 
-    def test_leontief(self):
+    def test_leontief(self, oecd):
         fd = (np.eye(oecd.rs) - oecd.A) @ oecd.X
         assert np.isclose(oecd.FD, fd, atol=.001).all()
 
         x = (np.eye(oecd.rs) - oecd.A).I @ oecd.FD
         assert np.isclose(oecd.X, x, atol=.001).all()
 
-    def test_ghosh(self):
+    def test_ghosh(self, oecd):
         x = oecd.G.T @ oecd.V.T
         assert np.isclose(oecd.X, x, atol=.001).all()
 
-    def test_shock(self):
+    def test_shock(self, oecd, custom_shock_vector):
         assert np.array_equal(oecd._shock(model='ghosh', custom_shock_vector=custom_shock_vector),
                               (oecd.G.T @ (oecd.V.T * (custom_shock_vector / 100))) + oecd.X)
 
@@ -78,17 +81,17 @@ def test_shock(self):
         with pytest.raises(ValueError):
             oecd._shock(model='leontief', shock=-10, regions=EA + ['something'], sectors=['35'])
 
-    def test_leontief_shock(self):
+    def test_leontief_shock(self, oecd):
         assert np.array_equal(
             oecd.leontief_demand_shock(shock=-10, regions=EA, sectors=['35']).x_new.values.reshape(-1, 1),
             oecd._shock(model='leontief', shock=-10, regions=EA, sectors=['35']))
 
-    def test_ghosh_shock(self):
+    def test_ghosh_shock(self, oecd):
         assert np.array_equal(
             oecd.ghosh_supply_shock(shock=-10, regions=EA, sectors=['35']).x_new.values.reshape(-1, 1),
             oecd._shock(model='ghosh', shock=-10, regions=EA, sectors=['35']))
 
-    def test_plot(self):
+    def test_plot(self, oecd):
         fig, ax = oecd.ghosh_supply_shock(shock=-10, regions=EA, sectors=['35'], plot_regions=EA, plot=True, show=False)
         assert isinstance(fig, matplotlib.figure.Figure)
         assert isinstance(ax, plt.Axes)
@@ -107,7 +110,7 @@ def test_plot(self):
             oecd.ghosh_supply_shock(shock=-10, regions=EA, sectors=['35'],
                                     plot=True, show=True, plot_by='region')
 
-    def test_get_imports_exports(self):
+    def test_get_imports_exports(self, oecd):
 
         assert np.isclose(oecd.get_imports_exports(import_regions=['CN1', 'CN2'],
                                                    export_regions='AU',

From 57fa156d5744facbebab84fb5ea6444e1f9ac0f4 Mon Sep 17 00:00:00 2001
From: Wouter Wakker <wouter.wakker@outlook.com>
Date: Sat, 6 Jun 2026 15:12:02 +0200
Subject: [PATCH 13/15] Add PyPI deploy workflow (manual dispatch)

---
 .github/workflows/deploy.yml | 32 ++++++++++++++++++++++++++++++++
 1 file changed, 32 insertions(+)
 create mode 100644 .github/workflows/deploy.yml

diff --git a/.github/workflows/deploy.yml b/.github/workflows/deploy.yml
new file mode 100644
index 0000000..a1171fe
--- /dev/null
+++ b/.github/workflows/deploy.yml
@@ -0,0 +1,32 @@
+name: deploy
+
+on: workflow_dispatch
+
+jobs:
+  deploy:
+    environment: Deploy
+    runs-on: ubuntu-latest
+    steps:
+      - run: echo "Starting deployment to PyPi"
+      - uses: actions/checkout@v5
+      - name: Set up Python 3.10
+        uses: actions/setup-python@v6
+        with:
+          python-version: "3.10"
+      - name: Install helpers
+        run: |
+          python -m pip install --upgrade pip
+          pip install build twine
+      - name: Build
+        # This project uses a pyproject.toml build backend (no setup.py), so build
+        # with the PEP 517 frontend rather than `setup.py sdist bdist_wheel`.
+        run: python -m build
+      - name: Check distribution metadata
+        run: twine check dist/*
+      - name: Deploy to PyPi
+        env:
+          TWINE_USERNAME: ${{ secrets.TWINE_USERNAME }}
+          TWINE_PASSWORD: ${{ secrets.TWINE_PASSWORD }}
+        run: |
+          twine upload dist/*
+      - run: echo "Successfully deployed to PyPi"

From ed3252e01e2547eb85279a527eafc905d4a18518 Mon Sep 17 00:00:00 2001
From: Wouter Wakker <wouter.wakker@outlook.com>
Date: Sat, 6 Jun 2026 15:13:53 +0200
Subject: [PATCH 14/15] Switch deploy workflow to PyPI Trusted Publishing
 (OIDC)

---
 .github/workflows/deploy.yml | 11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)

diff --git a/.github/workflows/deploy.yml b/.github/workflows/deploy.yml
index a1171fe..a9f97a5 100644
--- a/.github/workflows/deploy.yml
+++ b/.github/workflows/deploy.yml
@@ -6,6 +6,10 @@ jobs:
   deploy:
     environment: Deploy
     runs-on: ubuntu-latest
+    permissions:
+      # Required for PyPI Trusted Publishing (OIDC); no API token/secret needed.
+      id-token: write
+      contents: read
     steps:
       - run: echo "Starting deployment to PyPi"
       - uses: actions/checkout@v5
@@ -24,9 +28,6 @@ jobs:
       - name: Check distribution metadata
         run: twine check dist/*
       - name: Deploy to PyPi
-        env:
-          TWINE_USERNAME: ${{ secrets.TWINE_USERNAME }}
-          TWINE_PASSWORD: ${{ secrets.TWINE_PASSWORD }}
-        run: |
-          twine upload dist/*
+        # Trusted Publishing: PyPI verifies this repo/workflow/environment via OIDC.
+        uses: pypa/gh-action-pypi-publish@release/v1
       - run: echo "Successfully deployed to PyPi"

From 2db2ed7c7262889bd0a1ea3558c96cd703bac0ef Mon Sep 17 00:00:00 2001
From: Wouter Wakker <wouter.wakker@outlook.com>
Date: Sat, 6 Jun 2026 22:27:14 +0200
Subject: [PATCH 15/15] Minor fixes

---
 .github/workflows/ci.yml | 2 +-
 src/iotables/utils.py    | 6 ------
 2 files changed, 1 insertion(+), 7 deletions(-)

diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index 20a2e1d..8c4c5f8 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -35,7 +35,7 @@ jobs:
       - uses: actions/checkout@v4
       - uses: actions/setup-python@v5
         with:
-          python-version: "3.12"
+          python-version: "3.10"
       - name: Build sdist and wheel
         run: |
           python -m pip install --upgrade pip build twine
diff --git a/src/iotables/utils.py b/src/iotables/utils.py
index a596715..2b45d22 100644
--- a/src/iotables/utils.py
+++ b/src/iotables/utils.py
@@ -12,12 +12,6 @@ def assert_is_subset(subset, superset):
 def download_file(url, dest, proxy=None, verify=True):
     """Stream-download ``url`` to the local path ``dest``.
 
-    Uses ``curl_cffi`` with browser (TLS) impersonation so downloads succeed even
-    when the host sits behind Cloudflare's bot challenge. The OECD file server
-    (``webfs-sti.oecd.org``) rejects plain ``requests``/``urllib`` traffic with
-    HTTP 403; impersonating a real browser passes the challenge and is harmless
-    for the non-protected Eurostat/EXIOBASE hosts.
-
     Args:
         url: Source URL.
         dest: Local file path to write to.