Module type Signatures.GLOB

module type GLOB = sig .. end

A self-contained module implementing extended shell glob patterns who have an expressive power equal to boolean combinations of regular expressions.

A globber is a boolean combination of basic expressions indented to work on pathnames. Known operators are or, and and not, which may also be written |, & and ~. There are also constants true and false (or 1 and 0). Expression can be grouped using parentheses.

true matches anything,
false matches nothing,
basic or basic matches strings matching either one of the basic expressions,
basic and basic matches strings matching both basic expressions,
not basic matches string that don't match the basic expression,
basic matches strings that match the basic expression.

A basic expression can be a constant string enclosed in double quotes, in which double quotes must be preceded by backslashes, or a glob pattern enclosed between a < and a >,

"string" matches the literal string string,
<glob> matches the glob pattern glob.

A glob pattern is an anchored regular expression in a shell-like syntax. Most characters stand for themselves. Character ranges are given in usual shell syntax between brackets. The star * stands for any sequence of characters. The joker '?' stands for exactly one, unspecified character. Alternation is achieved using braces {.

glob1glob2 matches strings who have a prefix matching glob1 and the corresponding suffix matching glob2.
a matches the string consisting of the single letter a.
{glob1,glob2} matches strings matching glob1 or glob2.
? any one-letter string, excluding the slash.
* matches all strings not containing a slash, including the empty one.
**/ the empty string, or any string ending with a slash.
/** any string starting with a slash, or the empty string.
/**/ any string starting and ending with a slash.
[c1-c2c3-c4...] matches characters in the range c1 to c2 inclusive, or in the range c3 to c4 inclusive. For instance [a-fA-F0-9] matches hexadecimal digits. To match the dash, put it at the end.

type globber

The type representing globbers. Do not attempt to compare them, as they get on-the-fly optimizations.

val parse : ?dir:string -> string -> globber

parse ~dir pattern will parse the globber pattern pattern, optionally prefixing its patterns with dir.

exception Parse_error of string

A descriptive exception raised when an invalid glob pattern description is given.

val eval : globber -> string -> bool

eval g u returns true if and only if the string u matches the given glob expression. Avoid reparsing the same pattern, since the automaton implementing the pattern is optimized on the fly. The first few evaluations are done using a time-inefficient but memory-efficient algorithm. It then compiles the pattern into an efficient but more memory-hungry data structure.