regex¶

basic-regex matcher for grep and sed (//#use regex, ;#use regex)
NAME
    regex - basic-regex matcher for grep and sed (//#use regex, ;#use regex)

SYNOPSIS
    //#use regex        C:   cc splices /lib/lib_regex.c
    ;#use regex         asm: mkasm.sh appends /lib/lib_regex.inc

DESCRIPTION
    The classic tiny Thompson/Pike matcher (Kernighan's "Beautiful Code"
    dialect) with the cheap single-character quantifiers added:

        .   any single character
        *   zero or more of the preceding character (or '.')
        +   one  or more of the preceding character (or '.')
        ?   zero or one  of the preceding character (or '.')
        ^   anchor to the start of the line (only as the first char)
        $   anchor to the end   (only as the last char)

    Everything else is literal.

FUNCTIONS
    match(re, t)        1 if re matches ANYWHERE in t (honours a leading ^)
    matchhere(re, t)    1 if re matches a PREFIX of t; on success sets the
                        global `rend` past the match, so a caller (sed) can
                        compute the matched length = rend - start

    One self-recursive matchhere() -- deliberately no forward declaration
    / mutual recursion, since the native p8cc.c bootstrap rejects a
    standalone prototype. Quantifiers are non-greedy (fewest first) --
    fine for grep (yes/no) and adequate for sed's s///.

LIMITS
    Character classes [..] and \ escapes do not fit grep's host build yet
    -- blocked on the p8cc codegen-size work (see BACKLOG). The asm twin
    keeps re/t in memory words, saved on the hardware stack across the
    recursion.

SEE ALSO
    glob, grep, sed