jedarden e3c6c34760 feat(bf-3wkpz): implement timeout-protected stdout/stderr capture for benchmark

- Add READ_TIMEOUT (5 min) and MAX_OUTPUT_SIZE (100 MB) constants
- Implement read_pipe_with_timeout() with watchdog thread protection
- Update execute_grep_command() to capture and return both streams
- Prevents TH-03 hang scenario by bounding both time and memory

Acceptance criteria:
✓ Stdout captured completely
✓ Stderr captured completely
✓ Output buffered without truncation (100 MB limit)
✓ Read operations have timeout protection (5 min timeout)
✓ Both streams captured even if one is empty

Verification: notes/bf-3wkpz.md

2026-07-06 16:36:22 -04:00

3.1 KiB

Raw Blame History

Verification Note for bf-677eo

Task: Run pdftract CLI on degraded fixture

Date: 2026-07-06

What was attempted:

Located the degraded fixture: /home/coding/pdftract/tests/fixtures/scanned/low-quality/degraded-200dpi.pdf (588KB PDF file)
Verified CLI availability: Found pdftract executables at:
- /home/coding/pdftract/target/debug/pdftract
- /home/coding/pdftract/target/release/pdftract
Tested CLI help command: pdftract extract --help works correctly

Attempted extraction without OCR:

pdftract extract --text /tmp/degraded-200dpi-text.txt --json /tmp/degraded-200dpi.json \
    tests/fixtures/scanned/low-quality/degraded-200dpi.pdf

Result: Error: Failed to extract PDF (exit code 1)

Attempted extraction with OCR flag:

pdftract extract --ocr --text /tmp/degraded-200dpi-text.txt \
    tests/fixtures/scanned/low-quality/degraded-200dpi.pdf

Result: Error: --ocr requires the 'ocr' feature to be enabled

Attempted to build with OCR feature:
```
cargo build --release --bin pdftract --features ocr
```
Result: Build failed - requires system library leptonica (pkg-config package lept not found)

Acceptance Criteria Status:

Criterion	Status	Notes
pdftract CLI executes without errors	❌ FAIL	CLI help works, but extraction fails on degraded fixture
Command completes successfully	❌ FAIL	`Error: Failed to extract PDF` even without OCR
OCR process runs on the fixture file	❌ FAIL	OCR feature not compiled in; building requires leptonica system library

Root Causes:

OCR feature not enabled: The current pdftract build was compiled without the --features ocr flag
Missing system dependencies: Building with OCR requires the leptonica library (pkg-config lept), which is not installed on this system
Extraction failure without OCR: The degraded PDF appears to be a scanned image-only PDF that cannot be extracted without OCR

Existing Workarounds:

The fixture directory already contains OCR output generated by other means:

degraded-200dpi-ocr.txt - OCR output text (with typical OCR errors like "Y ork" vs "York")
degraded-200dpi-ground-truth.txt - Source text for comparison
create_degraded_200dpi.py - Python script used to create the degraded fixture

Recommendations:

To enable OCR-based extraction on this fixture, one of the following is needed:

Install leptonica system library (e.g., apt-get install libleptonica-dev)
Rebuild pdftract with --features ocr
Use external OCR tools (e.g., tesseract) directly on the PDF

Files Referenced:

Fixture: /home/coding/pdftract/tests/fixtures/scanned/low-quality/degraded-200dpi.pdf
Ground truth: tests/fixtures/scanned/low-quality/degraded-200dpi-ground-truth.txt
OCR output: tests/fixtures/scanned/low-quality/degraded-200dpi-ocr.txt

Verification: ❌ BLOCKED

The task cannot be completed successfully due to missing OCR dependencies. The CLI can be invoked, but extraction fails on the degraded fixture.

3.1 KiB Raw Blame History