pdftract/notes/pdftract-1w5u1.md

pdftract-1w5u1: Output formats verification

Summary

The three output formats for pdftract doctor were already implemented in:

• crates/pdftract-cli/src/doctor/output/human.rs - colored table output
• crates/pdftract-cli/src/doctor/output/json.rs - JSON output
• crates/pdftract-cli/src/doctor/output/features.rs - feature listing

Acceptance Criteria Verification

PASS: Default (TTY) colored table output

• Output shows colored table with Check, Status, Detail columns
• ANSI colors: OK=green, WARN=yellow, FAIL=red
• Summary line at bottom: "N OK, M WARN, K FAIL"
• Verified: ./target/release/pdftract doctor produces properly formatted table

PASS: Non-TTY plain text output

• Piped output (via | cat) strips ANSI escape codes
• Same content structure as TTY but without colors
• Summary line included
• Verified: ./target/release/pdftract doctor | cat produces plain text

PASS: --json output

• Single JSON object with {summary: {ok, warn, fail, not_applicable}, checks: [...]}
• JSON parses correctly with jq
• Status values are "OK", "WARN", "FAIL", "N/A"
• Verified: ./target/release/pdftract doctor --json | jq '.summary.fail' returns integer

PASS: --json jq filtering

• ./target/release/pdftract doctor --json | jq '.summary' shows correct counts
• ./target/release/pdftract doctor --json | jq '.checks[] | select(.status == "WARN")' filters correctly

PASS: --features output

• Lists compiled features, one per line
• Exit code 0
• No diagnostic checks run
• Verified: ./target/release/pdftract doctor --features returns exit code 0
• Note: Default build has no features enabled ("default" feature set), so output is empty

PASS: --no-color output

• Plain text output even in TTY
• No ANSI escape codes
• Same structure as default
• Verified: ./target/release/pdftract doctor --no-color produces plain text table

PASS: 80-column terminal width

• Output fits within 80 columns without wrapping
• Column widths: name=30, status=6, detail=flex
• Separator line is exactly 80 characters
• FIX: Detail truncation now applies uniformly to TTY and non-TTY output
• Previously: Truncation only in TTY mode, causing >80-char lines when piped or using --no-color
• Now: Truncation always based on 80 columns (TTY uses actual width, non-TTY assumes 80)

PASS: N/A row handling

• N/A checks excluded from human output (verified in code: line 41-43 of human.rs)
• N/A checks included in JSON with status="N/A" (verified in code: line 28-31 of json.rs)

Dependencies

• atty crate for TTY detection (already in dependencies)
• terminal_size crate for terminal width detection (already in dependencies)
• termcolor crate for ANSI color handling (already in dependencies)

Implementation Notes

Fix Applied

File: crates/pdftract-cli/src/doctor/output/human.rs

Issue: Detail field truncation only applied to TTY output. When piping to cat or using --no-color, long detail strings were not truncated, causing lines to exceed 80 columns.

Solution: Modified truncation logic (lines 77-89) to:

1. Determine terminal width: TTY uses actual width, non-TTY assumes 80 columns
2. Calculate max detail width: term_width - 38 (30 name + 1 space + 6 status + 1 space)
3. Truncate detail to max width with "..." suffix if needed

Result: Max line width now exactly 80 characters for all output modes.

Existing Implementation

The output modules were implemented with:

1. TTY detection via std::io::IsTerminal trait (nightly feature stabilized)
2. Color control via termcolor crate with ColorChoice enum
3. Terminal width detection via terminal_size crate
4. Detail truncation for long strings in TTY mode
5. Summary line at bottom with bold formatting
6. Stderr output for failures (in addition to stdout summary)

All acceptance criteria are met. One fix was applied to ensure 80-column compliance for non-TTY output.