From 3c51c3be85bb0d1bdb87ea0d6632f1c256912f27 Mon Sep 17 00:00:00 2001 From: Dimitri Staessens Date: Sun, 6 Oct 2019 21:37:45 +0200 Subject: build: Add some required modules for node --- node_modules/ret/lib/index.js | 282 ++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 282 insertions(+) create mode 100644 node_modules/ret/lib/index.js (limited to 'node_modules/ret/lib/index.js') diff --git a/node_modules/ret/lib/index.js b/node_modules/ret/lib/index.js new file mode 100644 index 0000000..0e151c3 --- /dev/null +++ b/node_modules/ret/lib/index.js @@ -0,0 +1,282 @@ +var util = require('./util'); +var types = require('./types'); +var sets = require('./sets'); +var positions = require('./positions'); + + +module.exports = function(regexpStr) { + var i = 0, l, c, + start = { type: types.ROOT, stack: []}, + + // Keep track of last clause/group and stack. + lastGroup = start, + last = start.stack, + groupStack = []; + + + var repeatErr = function(i) { + util.error(regexpStr, 'Nothing to repeat at column ' + (i - 1)); + }; + + // Decode a few escaped characters. + var str = util.strToChars(regexpStr); + l = str.length; + + // Iterate through each character in string. + while (i < l) { + c = str[i++]; + + switch (c) { + // Handle escaped characters, inclues a few sets. + case '\\': + c = str[i++]; + + switch (c) { + case 'b': + last.push(positions.wordBoundary()); + break; + + case 'B': + last.push(positions.nonWordBoundary()); + break; + + case 'w': + last.push(sets.words()); + break; + + case 'W': + last.push(sets.notWords()); + break; + + case 'd': + last.push(sets.ints()); + break; + + case 'D': + last.push(sets.notInts()); + break; + + case 's': + last.push(sets.whitespace()); + break; + + case 'S': + last.push(sets.notWhitespace()); + break; + + default: + // Check if c is integer. + // In which case it's a reference. + if (/\d/.test(c)) { + last.push({ type: types.REFERENCE, value: parseInt(c, 10) }); + + // Escaped character. + } else { + last.push({ type: types.CHAR, value: c.charCodeAt(0) }); + } + } + + break; + + + // Positionals. + case '^': + last.push(positions.begin()); + break; + + case '$': + last.push(positions.end()); + break; + + + // Handle custom sets. + case '[': + // Check if this class is 'anti' i.e. [^abc]. + var not; + if (str[i] === '^') { + not = true; + i++; + } else { + not = false; + } + + // Get all the characters in class. + var classTokens = util.tokenizeClass(str.slice(i), regexpStr); + + // Increase index by length of class. + i += classTokens[1]; + last.push({ + type: types.SET, + set: classTokens[0], + not: not, + }); + + break; + + + // Class of any character except \n. + case '.': + last.push(sets.anyChar()); + break; + + + // Push group onto stack. + case '(': + // Create group. + var group = { + type: types.GROUP, + stack: [], + remember: true, + }; + + c = str[i]; + + // If if this is a special kind of group. + if (c === '?') { + c = str[i + 1]; + i += 2; + + // Match if followed by. + if (c === '=') { + group.followedBy = true; + + // Match if not followed by. + } else if (c === '!') { + group.notFollowedBy = true; + + } else if (c !== ':') { + util.error(regexpStr, + 'Invalid group, character \'' + c + + '\' after \'?\' at column ' + (i - 1)); + } + + group.remember = false; + } + + // Insert subgroup into current group stack. + last.push(group); + + // Remember the current group for when the group closes. + groupStack.push(lastGroup); + + // Make this new group the current group. + lastGroup = group; + last = group.stack; + break; + + + // Pop group out of stack. + case ')': + if (groupStack.length === 0) { + util.error(regexpStr, 'Unmatched ) at column ' + (i - 1)); + } + lastGroup = groupStack.pop(); + + // Check if this group has a PIPE. + // To get back the correct last stack. + last = lastGroup.options ? + lastGroup.options[lastGroup.options.length - 1] : lastGroup.stack; + break; + + + // Use pipe character to give more choices. + case '|': + // Create array where options are if this is the first PIPE + // in this clause. + if (!lastGroup.options) { + lastGroup.options = [lastGroup.stack]; + delete lastGroup.stack; + } + + // Create a new stack and add to options for rest of clause. + var stack = []; + lastGroup.options.push(stack); + last = stack; + break; + + + // Repetition. + // For every repetition, remove last element from last stack + // then insert back a RANGE object. + // This design is chosen because there could be more than + // one repetition symbols in a regex i.e. `a?+{2,3}`. + case '{': + var rs = /^(\d+)(,(\d+)?)?\}/.exec(str.slice(i)), min, max; + if (rs !== null) { + if (last.length === 0) { + repeatErr(i); + } + min = parseInt(rs[1], 10); + max = rs[2] ? rs[3] ? parseInt(rs[3], 10) : Infinity : min; + i += rs[0].length; + + last.push({ + type: types.REPETITION, + min: min, + max: max, + value: last.pop(), + }); + } else { + last.push({ + type: types.CHAR, + value: 123, + }); + } + break; + + case '?': + if (last.length === 0) { + repeatErr(i); + } + last.push({ + type: types.REPETITION, + min: 0, + max: 1, + value: last.pop(), + }); + break; + + case '+': + if (last.length === 0) { + repeatErr(i); + } + last.push({ + type: types.REPETITION, + min: 1, + max: Infinity, + value: last.pop(), + }); + break; + + case '*': + if (last.length === 0) { + repeatErr(i); + } + last.push({ + type: types.REPETITION, + min: 0, + max: Infinity, + value: last.pop(), + }); + break; + + + // Default is a character that is not `\[](){}?+*^$`. + default: + last.push({ + type: types.CHAR, + value: c.charCodeAt(0), + }); + } + + } + + // Check if any groups have not been closed. + if (groupStack.length !== 0) { + util.error(regexpStr, 'Unterminated group'); + } + + return start; +}; + +module.exports.types = types; -- cgit v1.2.3