Commits on December 20, 2025

1. Fix -F newline splitting and flag interactions

   - Split -F patterns on newlines (each line is an OR alternative)
   - Add count_subpatterns() and split_pattern_on_newlines() helpers
   - Use compiled%count for split pattern count in match_line
   - Fix -c with -l: only print filename when count > 0
   - Fix -o with -c: count mode takes priority over only-matching

   mfwolffe committed 4 months ago
   7d332b5
2. Implement backreference support and Latin-1 case folding

   Backreferences:
   - Add has_backrefs flag to detect patterns with \1-\9
   - Implement backtracking matcher for patterns with backreferences
   - Track captured group start/end positions during matching
   - Support case-insensitive backreference matching
   
   Case folding:
   - Extend to_lower_char() to handle Latin-1 characters (192-222)
   - Add UTF-8 continuation byte handling for Latin Extended-A
   - Enables case-insensitive matching for accented characters (é, ñ, etc.)

   mfwolffe committed 4 months ago
   a3c0ed6
3. Fix -f pattern file reading

   - Change from formatted to unformatted stream access for byte-by-byte reading
   - Fixes issue where only first pattern was read from pattern files
   - Properly handle newlines across different platforms

   mfwolffe committed 4 months ago
   4dc60e3
4. Fix symlink handling for -r option

   - Add is_symlink() function using lstat to detect symbolic links
   - Skip symlinks in collect_files when follow_links is false (-r)
   - Symlinks are followed only with -R (dereference-recursive)

   mfwolffe committed 4 months ago
   be9dcf3
5. Preserve CRLF line endings like grep

   - Remove CR stripping from mmap line reading
   - Remove CR stripping from regular file I/O
   - Grep preserves carriage returns in output; ferp now matches

   mfwolffe committed 4 months ago
   f31f4ea
6. Fix error exit codes and directory handling

   - Add has_error flag to track errors during file processing
   - Return exit code 2 when errors occur (matching grep behavior)
   - Print error message for directories when -d read is used
   - Add has_error to OpenMP reduction clause for thread safety

   mfwolffe committed 4 months ago
   d312f5c
7. Fix -T tab alignment and line number width

   - Only print tab when there's a prefix (filename, line number, or byte offset)
   - Add line_number_width field to grep_options for dynamic width calculation
   - Pad line numbers based on file size for proper -T alignment

   mfwolffe committed 4 months ago
   71ca2f8
8. test fixtures, extended grep tests

   mfwolffe committed 4 months ago
   5b04ac6
9. Enable whitespace pattern test

   The "edge: spaces" test case now passes after fixing whitespace
   pattern handling throughout the codebase.

   mfwolffe committed 4 months ago
   56543cc
10. Fix alternation pattern parsing with null terminators

    Add null terminators when storing parsed alternatives in the
    Aho-Corasick optimization path. Without this, alternatives like
    "hello" stored in fixed-length arrays would have incorrect lengths.

    mfwolffe committed 4 months ago
    3259a58
11. Update matcher to use pattern_len for -w/-x options

    Use explicit pattern lengths when applying word boundary (-w) and
    line regexp (-x) transformations. This prevents null terminators
    from being included in the transformed patterns.

    mfwolffe committed 4 months ago
    6a04d1a
12. Update regex modules to use pattern_len

    Use pattern_len() instead of len_trim() across regex modules to properly
    handle whitespace patterns. Update Makefile dependencies to ensure
    ferp_kinds is compiled before regex modules that use pattern_len.

    mfwolffe committed 4 months ago
    5263050
13. Preserve trailing whitespace in output

    Remove trim() calls when printing matched lines to preserve trailing
    whitespace. This fixes cases where whitespace patterns would not
    display correctly in output.

    mfwolffe committed 4 months ago
    e34a89e
14. Store patterns with null terminators for exact length tracking

    Append null terminators when storing patterns to preserve their exact
    length. This enables correct handling of whitespace-only patterns that
    would otherwise be space-padded in Fortran's fixed-length strings.

    mfwolffe committed 4 months ago
    3df9d92
15. Add pattern_len() function for null-terminated pattern handling

    Fortran's len_trim() returns 0 for whitespace-only strings, which breaks
    patterns like "  " (two spaces). This function respects null terminators
    to track exact pattern length, falling back to full string length when
    no terminator is present.

    mfwolffe committed 4 months ago
    b43bcce
16. show ferp command in addition to grep

    mfwolffe committed 4 months ago
    0d61ac5
17. don't track this

    mfwolffe committed 4 months ago
    a90aacc
18. update fixtures, make tests more robust

    mfwolffe committed 4 months ago
    429310c
19. update tests files

    mfwolffe committed 4 months ago
    7614d9b
20. fixes

    mfwolffe committed 4 months ago
    c5ced7d
21. fixes

    mfwolffe committed 4 months ago
    7373e0f
22. integration testing

    mfwolffe committed 4 months ago
    ed71a95
23. test fixtures

    mfwolffe committed 4 months ago
    4811614