2022-05-31 22:34:14 +02:00
|
|
|
-- mod-version:3
|
2019-12-28 12:16:32 +01:00
|
|
|
local syntax = require "core.syntax"
|
|
|
|
|
2023-11-29 17:00:09 +01:00
|
|
|
-- Regex pattern explanation:
|
|
|
|
-- This will match / and will look ahead for something that looks like a regex.
|
|
|
|
--
|
|
|
|
-- (?!/) Don't match empty regexes.
|
|
|
|
--
|
|
|
|
-- (?>...) this is using an atomic group to minimize backtracking, as that'd
|
|
|
|
-- cause "Catastrophic Backtracking" in some cases.
|
|
|
|
--
|
|
|
|
-- [^\\[\/]++ will match anything that's isn't an escape, a start of character
|
|
|
|
-- class or an end of pattern, without backtracking (the second +).
|
|
|
|
--
|
|
|
|
-- \\. will match anything that's escaped.
|
|
|
|
--
|
|
|
|
-- \[(?:[^\\\]++]|\\.)*+\] will match character classes.
|
|
|
|
--
|
|
|
|
-- /[gmiyuvsd]*\s*[\n,;\)\]\}\.]) will match the end of pattern delimiter, optionally
|
|
|
|
-- followed by pattern options, and anything that can
|
|
|
|
-- be after a pattern.
|
|
|
|
--
|
2024-03-25 00:20:32 +01:00
|
|
|
-- Demo with some unit tests (click on the Unit Tests entry): https://regex101.com/r/Vx5L5V/1
|
2023-11-29 17:00:09 +01:00
|
|
|
-- Note that it has a couple of changes to make it work on that platform.
|
|
|
|
local regex_pattern = {
|
2024-03-25 00:20:32 +01:00
|
|
|
[=[\/(?=(?!\/)(?:(?>[^\\[\/]++|\\.|\[(?:[^\\\]]++|\\.)*+\])*+)++\/[gmiyuvsd]*\s*(?:[\n,;\)\]\}\.]|\/[\/*]))()]=],
|
2023-11-29 17:00:09 +01:00
|
|
|
"/()[gmiyuvsd]*", "\\"
|
|
|
|
}
|
|
|
|
|
|
|
|
-- For the moment let's not actually differentiate the insides of the regex,
|
|
|
|
-- as this will need new token types...
|
|
|
|
local inner_regex_syntax = {
|
|
|
|
patterns = {
|
|
|
|
{ pattern = "%(()%?[:!=><]", type = { "string", "string" } },
|
|
|
|
{ pattern = "[.?+*%(%)|]", type = "string" },
|
|
|
|
{ pattern = "{%d*,?%d*}", type = "string" },
|
|
|
|
{ regex = { [=[\[()\^?]=], [=[(?:\]|(?=\n))()]=], "\\" },
|
|
|
|
type = { "string", "string" },
|
|
|
|
syntax = { -- Inside character class
|
|
|
|
patterns = {
|
|
|
|
{ pattern = "\\\\", type = "string" },
|
|
|
|
{ pattern = "\\%]", type = "string" },
|
|
|
|
{ pattern = "[^%]\n]", type = "string" }
|
|
|
|
},
|
|
|
|
symbols = {}
|
|
|
|
}
|
|
|
|
},
|
|
|
|
{ regex = "\\/", type = "string" },
|
|
|
|
{ regex = "[^/\n]", type = "string" },
|
|
|
|
},
|
|
|
|
symbols = {}
|
|
|
|
}
|
|
|
|
|
2019-12-28 12:16:32 +01:00
|
|
|
syntax.add {
|
2021-11-21 03:46:43 +01:00
|
|
|
name = "JavaScript",
|
2022-06-20 16:12:25 +02:00
|
|
|
files = { "%.js$", "%.json$", "%.cson$", "%.mjs$", "%.cjs$" },
|
2019-12-28 12:16:32 +01:00
|
|
|
comment = "//",
|
2021-12-26 08:05:27 +01:00
|
|
|
block_comment = { "/*", "*/" },
|
2019-12-28 12:16:32 +01:00
|
|
|
patterns = {
|
2024-03-25 00:20:32 +01:00
|
|
|
{ pattern = "//.*", type = "comment" },
|
|
|
|
{ pattern = { "/%*", "%*/" }, type = "comment" },
|
|
|
|
{ regex = regex_pattern, syntax = inner_regex_syntax, type = {"string", "string"} },
|
|
|
|
{ pattern = { '"', '"', '\\' }, type = "string" },
|
|
|
|
{ pattern = { "'", "'", '\\' }, type = "string" },
|
|
|
|
{ pattern = { "`", "`", '\\' }, type = "string" },
|
|
|
|
-- Use (?:\/(?!\/|\*))? to avoid that a regex can start after a number, while also allowing // and /* comments
|
|
|
|
{ regex = [[-?0[xXbBoO][\da-fA-F_]+n?()\s*()(?:\/(?!\/|\*))?]], type = {"number", "normal", "operator"} },
|
|
|
|
{ regex = [[-?\d+[0-9.eE_n]*()\s*()(?:\/(?!\/|\*))?]], type = {"number", "normal", "operator"} },
|
|
|
|
{ regex = [[-?\.?\d+()\s*()(?:\/(?!\/|\*))?]], type = {"number", "normal", "operator"} },
|
|
|
|
{ pattern = "[%+%-=/%*%^%%<>!~|&]", type = "operator" },
|
|
|
|
{ pattern = "[%a_][%w_]*%f[(]", type = "function" },
|
|
|
|
{ pattern = "[%a_][%w_]*", type = "symbol" },
|
2019-12-28 12:16:32 +01:00
|
|
|
},
|
|
|
|
symbols = {
|
2020-05-10 10:51:45 +02:00
|
|
|
["async"] = "keyword",
|
|
|
|
["await"] = "keyword",
|
|
|
|
["break"] = "keyword",
|
|
|
|
["case"] = "keyword",
|
|
|
|
["catch"] = "keyword",
|
|
|
|
["class"] = "keyword",
|
|
|
|
["const"] = "keyword",
|
|
|
|
["continue"] = "keyword",
|
|
|
|
["debugger"] = "keyword",
|
|
|
|
["default"] = "keyword",
|
|
|
|
["delete"] = "keyword",
|
|
|
|
["do"] = "keyword",
|
|
|
|
["else"] = "keyword",
|
|
|
|
["export"] = "keyword",
|
|
|
|
["extends"] = "keyword",
|
|
|
|
["finally"] = "keyword",
|
|
|
|
["for"] = "keyword",
|
|
|
|
["function"] = "keyword",
|
2020-05-10 10:56:28 +02:00
|
|
|
["get"] = "keyword",
|
2020-05-10 10:51:45 +02:00
|
|
|
["if"] = "keyword",
|
|
|
|
["import"] = "keyword",
|
2024-03-24 19:25:44 +01:00
|
|
|
["from"] = "keyword",
|
2020-05-10 10:51:45 +02:00
|
|
|
["in"] = "keyword",
|
2021-05-17 08:56:21 +02:00
|
|
|
["of"] = "keyword",
|
2020-05-10 10:51:45 +02:00
|
|
|
["instanceof"] = "keyword",
|
|
|
|
["let"] = "keyword",
|
|
|
|
["new"] = "keyword",
|
|
|
|
["return"] = "keyword",
|
2020-05-10 10:56:28 +02:00
|
|
|
["set"] = "keyword",
|
2020-06-05 23:37:51 +02:00
|
|
|
["static"] = "keyword",
|
2020-05-10 10:51:45 +02:00
|
|
|
["super"] = "keyword",
|
|
|
|
["switch"] = "keyword",
|
|
|
|
["throw"] = "keyword",
|
|
|
|
["try"] = "keyword",
|
|
|
|
["typeof"] = "keyword",
|
|
|
|
["var"] = "keyword",
|
|
|
|
["void"] = "keyword",
|
|
|
|
["while"] = "keyword",
|
|
|
|
["with"] = "keyword",
|
|
|
|
["yield"] = "keyword",
|
2020-06-05 23:37:51 +02:00
|
|
|
["true"] = "literal",
|
|
|
|
["false"] = "literal",
|
|
|
|
["null"] = "literal",
|
|
|
|
["undefined"] = "literal",
|
|
|
|
["arguments"] = "keyword2",
|
|
|
|
["Infinity"] = "keyword2",
|
|
|
|
["NaN"] = "keyword2",
|
|
|
|
["this"] = "keyword2",
|
2019-12-28 12:16:32 +01:00
|
|
|
},
|
|
|
|
}
|