The re module is a wrapper around boost::regex, intended as a full replacement for Lua’s built in regular expressions. It has two main advantages over Lua’s:

  1. Full Unicode support. Lua regular expressions operate on bytes rather than characters, which frequently causes problems with multibyte characters.
  2. A more powerful and flexible syntax. Properly speaking, Lua does not support regular expressions; rather it has a basic pattern matching language that supports a small subset of what can be done with regular expressions. boost::regex, on the other hand, supports perl-compatible regular expressions.

Usage

Import this module with re = require 'aegisub.re'{:.language-lua}.

See boost.regex’s documentation for information about the regular expression syntax. In general any resources on the web that refer to Perl regular expressions or PCRE will apply to this module’s regular expressions.

Match Tables

Several of the functions below return Match Tables, which are tables containing the following fields:

str (string)
The text matched by a pattern or capturing expression
first (number)
The start index of str in the original string which had a regular expression applied to it. Note that this index is one-based and is in bytes, rather than characters, to match Lua’s string indexing.
last (number)
The end index of str in the original string which had a regular expression applied to it. Note that this index is one-based, inclusive, and is in bytes, rather than characters, to match Lua’s string indexing.

{::template name=“luabox”}

re.match(“b”, “abc”) { { [“str”] = “b”, [“first”] = 2, [“last”] = 2 } } {:/}

Flags

The following flags may be passed to all of the static functions (including re.compile). Flags must come after all supplied non-flag arguments, but optional arguments can be skipped.

re.ICASE
Ignore case when matching.
re.NOSUB:
Don’t set backreferences and capture groups. Can improve performance when they aren’t needed.
re.NEWLINE_ALT:
Treat newline characters as the alternation operator (|).
re.NO_MOD_M:
^ and $ only match the beginning and end of the string rather than newlines.
re.MOD_S:
Treat newlines as normal characters, matched by ‘.’.
re.MOD_X:
Ignore unescaped whitespace in the expression, making it possible to write regular expressions that aren’t write-only.
re.NO_EMPTY_SUBEXPRESSION:
Don’t match empty expressions/alternatives.

{::template name=“luabox”}

re.match(“a”, “A”) nil re.match(“a”, “A”, re.ICASE, re.NOSUB) { { [“str”] = “A”, [“first”] = 1, [“last”] = 1 } } {:/}

re.compile

Synopsis: expr = re.compile(pattern, [FLAGS]){:.language-lua}

Compile a regular expression. Reusing a compiled regular expression is faster than recompiling it each time it is used, and is usually more readable as well.

@pattern (string)
Regular expression to compile.
expr (table)
A table with all of the functions listed below, except without the pattern and flags arguments.

{::template name=“luabox”}

expr = re.compile(“a”) expr:split(“banana”) { “b”, “n”, “n” } {:/}

re.split

Synopsis: chunks = re.split(str, pattern, skip_empty=false, max_splits=0){:.language-lua}

Split the string at each of the occurrences of pattern.

@str (string)
String to split.
@pattern (string)
Regular expression to split the string on. Capturing groups in the pattern are ignored.
@skip_empty (boolean)
Do not include zero-length chunks in the results.
@max_splits (number)
If greater than zero, the maximum numbers of times to split the string (i.e. #chunks will be at most max_splits + 1).
chunks (table)
A table containing each of the sections of str between the matches of pattern.

{::template name=“luabox”}

re.split(",", “a,,b,c”) { “a”, “”, “b”, “c” } {:/} {::template name=“luabox”} re.split(",", “a,,b,c”, true) { “a”, “b”, “c” } {:/} {::template name=“luabox”} re.split(",", “a,,b,c”, false, 1) { “a”, “,b,c”, } {:/}

re.gsplit

Synopsis: iter = re.gsplit(str, pattern, skip_empty=false, max_splits=0){:.language-lua}

Iterator version of re.split.

@str (string)
String to split.
@pattern (string)
Regular expression to split the string on. Capturing groups in the pattern are ignored.
@skip_empty (boolean)
Do not include zero-length chunks in the results.
@max_splits (number)
If greater than zero, the maximum numbers of times to split the string (i.e. #chunks will be at most max_splits + 1).
iter (iterator over strings)
An iterator over each of the sections of str between the matches of pattern.

{::template name=“luabox”}

for str in re.gsplit(",", “a,,b,c”) do print str end a

b c {:/} {::template name=“luabox”}

for str in re.gsplit(",", “a,,b,c”, true) do print str end a b c {:/} {::template name=“luabox”} for str in re.gsplit(",", “a,,b,c”, false, 1) do print str end a ,b,c {:/}

re.find

Synopsis: matches = re.find(str, pattern){:.language-lua}

Find all non-overlapping substrings of str which match pattern.

@str (string)
String to search for the pattern in.
@pattern (string)
Pattern to search for. Capturing groups in the pattern are ignored.
matches (table or nil)
A table of Match Tables for all matches, or nil if there were none.

{::template name=“luabox”}

re.find(".", “☃☃”) { { [“str”] = “☃”, [“first”] = 1, [“last”] = 3 }, { [“str”] = “☃”, [“first”] = 4, [“last”] = 6 } } {:/} {::template name=“luabox”} function contains_an_a(str) if re.find(“a”, str) print “Has an a” else print “Doesn’t have an a” end end contains_an_a(“abc”) Has an a contains_an_a(“def”) Doesn’t have an a {:/}

re.gfind

Synopsis: iter = re.gfind(str, pattern){:.language-lua}

Iterate over all non-overlapping substrings of str which match pattern.

@str (string)
String to search for the pattern in.
@pattern (string)
Pattern to search for. Capturing groups in the pattern are ignored.
iter (iterator over string, number, number)
An iterator which produces three values at each step: a matched string, the started index of the match in the source string, and the inclusive end index of the match in the source string.

{::template name=“luabox”}

for str, start_idx, end_idx in re.gfind(".", “☃☃”) do print string.format("%d-%d: %s", start_idx, end_idx, str) end 1-3: ☃ 4-6: ☃ {:/}

re.match

Synopsis: matches = re.match(str, pattern){:.language-lua}

Match a pattern against a string. This differs from find in that find returns all matches and does not capture subgroups, while this returns only a single match along with the captured subgroups.

@str (string)
String to search for the pattern in.
@pattern (string)
Pattern to search for.
matches (table or nil)
nil if the pattern did not match the string. Otherwise, a table containing a Match Table for the full match, followed by a Match Table for each capturing subexpression in the pattern (if any).

{::template name=“luabox”}

re.match("(\d+) (\d+) (\d+)", “{250 1173 380}Help!”) { { [“str”] = “250 1173 380”, [“first”] = 2, [“last”] = 13 }, { [“str”] = “250”, [“first”] = 2, [“last”] = 4 }, { [“str”] = “1173”, [“first”] = 6, [“last”] = 9, }, { [“str”] = “380” [“first”] = 11, [“last”] = 13 } } {:/}

re.gmatch

Synopsis: iter = re.gmatch(str, pattern){:.language-lua}

Iterator version of re.match.

@str (string)
String to search for the pattern in.
@pattern (string)
Pattern to search for.
matches (iterator over table)
An iterator which returns a table containing a Match Table for the full match (if it matched), followed by a Match Table for each capturing subexpression in the pattern (if any).

re.sub

Synopsis: out_str, rep_count = re.sub(str, replace, pattern, max_count=0){:.language-lua}

Replace each occurrence of pattern in str with replace.

@pattern (string)
Pattern to search for.
@replace (string or function)
Replacement for matches. This may be either a string which is inserted, or a function which is called for each match.

If replace is a string, it may contain references to the matches. & and \0 are replaced with the full match, and \<number> is replaced with the appropriate captured subexpression.

If replace is a function, it is called for either the entire match (if there are no capturing subexpressions), or for each captured subexpression. It is passed the match string, start index of the match, and end index of the match. If it returns a string, the match is replaced with the return value. If it returns anything else, then the source string is left unchanged.

@max_count (number)
If greater than zero, the maximum number of replacements to make.
out_str (string)
The input string, with replacements applied.
rep_count (number)
The number of replacements that were made.

Example

Replace all instances of \k with \kf:

>>> re.sub("{\\k10}a{\\k15}b{\\k30}c", "\\\\k", "\\kf")
{\kf10}a{\kf15}b{\kf30}c
Example

Replace all instances of \k and \K with \kf:

>>> re.sub("{\\K10}a{\\K15}b{\\k30}c", "\\\\k", "\\kf", re.ICASE)
{\kf10}a{\kf15}b{\kf30}c
Example

Add one to each \k duration:

function add_one(str)
    return tostring(tonumber(str) + 1)
end
>>> re.sub("{\\k10}a{\\k15}b{\\k30}c", "\\\\k(\[[:digit:]]+)", add_one)
{\k11}a{\k16}b{\k31}c

{::template name=“automation_navbox” /}