`5263050`

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.

Authored by mfwolffe <wolffemf@dukes.jmu.edu> 4 months ago

SHA: 5263050a3cf26fd45d0f2da7f877f9cc64ea19a0
Parents: e34a89e
Tree: 7ec0955

5 changed files

Status	File	+	-
M	`Makefile`	6	6
M	`src/regex/aho_corasick.f90`	5	4
M	`src/regex/pcre_api.f90`	12	6
M	`src/regex/regex_api.f90`	3	2
M	`src/regex/regex_lexer.f90`	2	1

Makefilemodified

  $(BUILD_DIR)/%.o: $(SRC_DIR)/%.c | $(BUILD_DIR)
  	$(CC) $(CFLAGS) -c $< -o $@
 -# Regex module dependencies
 +# Regex module dependencies (note: some depend on ferp_kinds for pattern_len function)
  $(BUILD_DIR)/regex_charclass.o:
  $(BUILD_DIR)/regex_types.o: $(BUILD_DIR)/regex_charclass.o
 -$(BUILD_DIR)/regex_lexer.o: $(BUILD_DIR)/regex_types.o
 +$(BUILD_DIR)/regex_lexer.o: $(BUILD_DIR)/regex_types.o $(BUILD_DIR)/ferp_kinds.o
  $(BUILD_DIR)/regex_parser.o: $(BUILD_DIR)/regex_types.o
  $(BUILD_DIR)/regex_nfa.o: $(BUILD_DIR)/regex_types.o $(BUILD_DIR)/regex_charclass.o $(BUILD_DIR)/regex_parser.o
  $(BUILD_DIR)/regex_engine.o: $(BUILD_DIR)/regex_types.o
 -$(BUILD_DIR)/aho_corasick.o:
 -$(BUILD_DIR)/regex_optimizer.o: $(BUILD_DIR)/regex_types.o $(BUILD_DIR)/regex_charclass.o $(BUILD_DIR)/aho_corasick.o
 -$(BUILD_DIR)/regex_api.o: $(BUILD_DIR)/regex_types.o $(BUILD_DIR)/regex_lexer.o $(BUILD_DIR)/regex_parser.o $(BUILD_DIR)/regex_nfa.o $(BUILD_DIR)/regex_engine.o $(BUILD_DIR)/regex_optimizer.o
 -$(BUILD_DIR)/pcre_api.o:
 +$(BUILD_DIR)/aho_corasick.o: $(BUILD_DIR)/ferp_kinds.o
 +$(BUILD_DIR)/regex_optimizer.o: $(BUILD_DIR)/regex_types.o $(BUILD_DIR)/regex_charclass.o $(BUILD_DIR)/aho_corasick.o $(BUILD_DIR)/ferp_kinds.o
 +$(BUILD_DIR)/regex_api.o: $(BUILD_DIR)/regex_types.o $(BUILD_DIR)/regex_lexer.o $(BUILD_DIR)/regex_parser.o $(BUILD_DIR)/regex_nfa.o $(BUILD_DIR)/regex_engine.o $(BUILD_DIR)/regex_optimizer.o $(BUILD_DIR)/ferp_kinds.o
 +$(BUILD_DIR)/pcre_api.o: $(BUILD_DIR)/ferp_kinds.o
  # Main module dependencies
  $(BUILD_DIR)/ferp_options.o: $(BUILD_DIR)/ferp_kinds.o

src/regex/aho_corasick.f90modified

    !> Aho-Corasick automaton for multi-pattern string matching
    !> Matches all patterns in a single pass O(n + m + z)
    !> where n=text length, m=total pattern length, z=matches
 +  use ferp_kinds, only: pattern_len
    implicit none
    private
      ac%ignore_case = ignore_case
      ac%num_patterns = num_patterns
 -    ! Allocate pattern lengths
 +    ! Allocate pattern lengths (use pattern_len to preserve whitespace patterns)
      allocate(ac%pattern_lengths(num_patterns))
      do i = 1, num_patterns
 -      ac%pattern_lengths(i) = len_trim(patterns(i))
 +      ac%pattern_lengths(i) = pattern_len(patterns(i))
      end do
      ! Initial capacity - estimate based on total pattern length
      ac%capacity = 1
      do i = 1, num_patterns
 -      ac%capacity = ac%capacity + len_trim(patterns(i))
 +      ac%capacity = ac%capacity + pattern_len(patterns(i))
      end do
      ac%capacity = max(ac%capacity, 256)
      allocate(ac%nodes(ac%capacity))
      ! Phase 1: Build trie from patterns
      do i = 1, num_patterns
        state = 1  ! Start at root
 -      do j = 1, len_trim(patterns(i))
 +      do j = 1, pattern_len(patterns(i))
          ch = patterns(i)(j:j)
          if (ignore_case) then
            c = to_lower_code(ichar(ch))

src/regex/pcre_api.f90modified

    !> PCRE2 library bindings for Perl-compatible regular expressions
    !> Uses iso_c_binding for C interoperability with libpcre2-8
    use, intrinsic :: iso_c_binding
 +  use ferp_kinds, only: pattern_len
    implicit none
    private
      logical, intent(in), optional :: ignore_case
      integer, intent(out) :: ierr
 -    character(len=len_trim(pattern)+1, kind=c_char) :: c_pattern
 +    integer :: plen
 +    character(len=:), allocatable :: c_pattern
      integer(c_int) :: options, errorcode
 -    integer(c_size_t) :: erroroffset, pattern_len
 +    integer(c_size_t) :: erroroffset, pcre_pattern_len
      ierr = 0
      re%compiled = .false.
      re%error_code = 0
      re%error_msg = ''
 +    ! Get actual pattern length (preserving whitespace patterns)
 +    plen = pattern_len(pattern)
++
      ! Set options - enable UTF-8 and Unicode properties by default
      options = ior(PCRE2_UTF, PCRE2_UCP)
      if (present(ignore_case)) then
        if (ignore_case) options = ior(options, PCRE2_CASELESS)
      end if
 -    ! Prepare pattern as C string
 -    c_pattern = trim(pattern) // c_null_char
 -    pattern_len = int(len_trim(pattern), c_size_t)
 +    ! Prepare pattern as C string (use exact length, not trim)
 +    allocate(character(len=plen+1) :: c_pattern)
 +    c_pattern = pattern(1:plen) // c_null_char
 +    pcre_pattern_len = int(plen, c_size_t)
      ! Compile pattern
 -    re%code = pcre2_compile_8(c_pattern, pattern_len, options, &
 +    re%code = pcre2_compile_8(c_pattern, pcre_pattern_len, options, &
                                 errorcode, erroroffset, c_null_ptr)
      if (.not. c_associated(re%code)) then

src/regex/regex_api.f90modified

    use regex_nfa
    use regex_engine
    use regex_optimizer
 +  use ferp_kinds, only: pattern_len
    implicit none
    private
      if (present(is_ere)) extended = is_ere
      re%is_ere = extended
 -    ! Handle empty pattern
 -    if (len_trim(pattern) == 0) then
 +    ! Handle empty pattern (use pattern_len to preserve whitespace patterns)
 +    if (pattern_len(pattern) == 0) then
        call re%nfa%init()
        re%nfa%start_state = re%nfa%add_state()
        re%nfa%accept_state = re%nfa%add_state()

src/regex/regex_lexer.f90modified

    !> Regex pattern tokenizer for FERP
    !> Handles both BRE (Basic) and ERE (Extended) regex dialects
    use regex_types
 +  use ferp_kinds, only: pattern_len
    implicit none
    private
      ierr = 0
      call tokens%init()
 -    n = len_trim(pattern)
 +    n = pattern_len(pattern)  ! Use pattern_len to preserve whitespace patterns
      i = 1
      in_bracket = .false.