regexp9

Regular expressions based on Plan9 code.

Changes from original code by Rob Pike

• Fully supports UTF8 (transformed old Rune-type).
• Made reentrant: moved global parser variables to stack.
• Added support for escaped ctrl chars in expressions: tab, newline etc. \t \n \r \v \f
• Added support for shorthand character classes and inverse: \d \D \s \S \w \W
• Added support for word boundary meta character and inverse: \b \B
• Added support for inline "single line"/dotAll mode (from insertion point): (?s)
• Added support for POSIX char classes [:alnum:] [:alpha:] [:blank:] [:cntrl:] [:digit:] [:graph:] [:lower:] [:print:] [:punct:] [:space:] [:upper:] [:word:] [:xdigit:]
• Removed obsolete rregexec9() and rregsub9(), and the rather pointless regcomplit9().
• Constified (const char*) all references to input strings.
• Formatting changes: tabs to space, etc.
• Optimizations: malloc usage and shorter code. Fast UTF8 code.
• Compiles with C99, C++.
• Reduced total source code size from about 1600 to 1200 lines.

Example

#include "regexp9.h"
#include <stdio.h>

int main() {
    const char* pattern = "hell.([ \\t]w.rld)+";
    const char* input = "hell😀 w😀rld\tworld wxrld";

    enum {N=5};
    Resub rs[N] = {0};

    Reprog *p = regcomp9(pattern);
    if (regexec9(p, input, rs, N))
        printf("regexp9: '%s' => matched: %s\n", input, pattern);
    else
        printf("regexp9: No match\n");
    regfree9(p);
}