feat: add parquet_file_metadata and parquet_page_index to datafusion-cli

[jay-dee7]

Which issue does this PR close?

• Part of datafusion-cli: expose bloom filter, page index and sorting-column metadata in parquet_metadata, plus file-level metadata #25499

Rationale for this change

datafusion-cli can't show a parquet file's footer metadata (writer, key-value metadata, footer size) or its page index. Checking bloom filter sizing, page index cost or writer metadata currently means using DuckDB or a custom tool. This PR covers items 2 and 3 of #25499; item 1 is in #25638.

What changes are included in this PR?

• New parquet_file_metadata(path) table function, one row per file: created_by, version, num_rows, num_row_groups, key_value_metadata (a map, so key_value_metadata['ARROW:schema'] works) and footer_length.
• New parquet_page_index(path) table function, one row per data page: the page location from the offset index (page_ordinal, first_row_index, offset, compressed_page_size) and the column index min_value, max_value and null_count. Files without a page index return no rows.
• The path argument parsing is now shared by all three parquet functions.
• Docs for both functions in docs/source/user-guide/cli/functions.md.

What is the testing strategy for this PR?

Snapshot tests test_parquet_file_metadata_works and test_parquet_page_index_works in datafusion-cli/src/main.rs, run against parquet-testing files. They cover an all-null page (NULL min/max) and UTF8 min/max shown as strings. The page values match the parquet-cli output in int32_with_null_pages.md, and footer_length matches the file's raw footer bytes.

Are there any user-facing changes?

Yes: two new datafusion-cli table functions, documented in the CLI functions guide. parquet_metadata output is unchanged and there are no breaking changes.

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 82.51%. Comparing base (cee15b7) to head (b76fba2).
⚠️ Report is 62 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #25640      +/-   ##
==========================================
- Coverage   82.51%   82.51%   -0.01%     
==========================================
  Files        1141     1141              
  Lines      439773   439773              
  Branches   439773   439773              
==========================================
- Hits       362881   362872       -9     
- Misses      54950    54956       +6     
- Partials    21942    21945       +3     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[jay-dee7]

@kumarUjjawal can you please review this one? or should I split this one to two PRs to make it easier to review?

[kumarUjjawal]

Thank you @jay-dee7

[alamb]

this is very cool -- thank you

[alamb]

Thanks @jay-dee7 nd @kumarUjjawal -- this looks very useful

[alamb]

As a follow on, we could also potentially use this:
https://docs.rs/parquet/latest/parquet/arrow/arrow_reader/statistics/struct.StatisticsConverter.html#method.row_group_mins

To convert the min/max values and then call the arrow cast kernel to turn them into strings

There is similar code for data page mins here:
https://docs.rs/parquet/latest/parquet/arrow/arrow_reader/statistics/struct.StatisticsConverter.html#method.data_page_mins

That would handle things like min/max dates better (I think this is just going to show the raw values rather than formatted as a date)

[jay-dee7]

@alamb thanks for sharing this, should I include it here? or in a follow-up PR?

[kumarUjjawal]

I think a follow up would be better.

[jay-dee7]

sure, will do, and thanks for being pro-active :)