| 1 |
"use strict"; |
| 2 |
var wp; |
| 3 |
(wp ||= {}).wordcount = (() => { |
| 4 |
var __defProp = Object.defineProperty; |
| 5 |
var __getOwnPropDesc = Object.getOwnPropertyDescriptor; |
| 6 |
var __getOwnPropNames = Object.getOwnPropertyNames; |
| 7 |
var __hasOwnProp = Object.prototype.hasOwnProperty; |
| 8 |
var __export = (target, all) => { |
| 9 |
for (var name in all) |
| 10 |
__defProp(target, name, { get: all[name], enumerable: true }); |
| 11 |
}; |
| 12 |
var __copyProps = (to, from, except, desc) => { |
| 13 |
if (from && typeof from === "object" || typeof from === "function") { |
| 14 |
for (let key of __getOwnPropNames(from)) |
| 15 |
if (!__hasOwnProp.call(to, key) && key !== except) |
| 16 |
__defProp(to, key, { get: () => from[key], enumerable: !(desc = __getOwnPropDesc(from, key)) || desc.enumerable }); |
| 17 |
} |
| 18 |
return to; |
| 19 |
}; |
| 20 |
var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: true }), mod); |
| 21 |
|
| 22 |
// packages/wordcount/build-module/index.mjs |
| 23 |
var index_exports = {}; |
| 24 |
__export(index_exports, { |
| 25 |
count: () => count |
| 26 |
}); |
| 27 |
|
| 28 |
// packages/wordcount/build-module/defaultSettings.mjs |
| 29 |
var defaultSettings = { |
| 30 |
HTMLRegExp: /<\/?[a-z][^>]*?>/gi, |
| 31 |
HTMLcommentRegExp: /<!--[\s\S]*?-->/g, |
| 32 |
spaceRegExp: / | /gi, |
| 33 |
HTMLEntityRegExp: /&\S+?;/g, |
| 34 |
// \u2014 = em-dash. |
| 35 |
connectorRegExp: /--|\u2014/g, |
| 36 |
// Characters to be removed from input text. |
| 37 |
removeRegExp: new RegExp( |
| 38 |
[ |
| 39 |
"[", |
| 40 |
// Basic Latin (extract) |
| 41 |
"!-/:-@[-`{-~", |
| 42 |
// Latin-1 Supplement (extract) |
| 43 |
"\x80-\xBF\xD7\xF7", |
| 44 |
/* |
| 45 |
* The following range consists of: |
| 46 |
* General Punctuation |
| 47 |
* Superscripts and Subscripts |
| 48 |
* Currency Symbols |
| 49 |
* Combining Diacritical Marks for Symbols |
| 50 |
* Letterlike Symbols |
| 51 |
* Number Forms |
| 52 |
* Arrows |
| 53 |
* Mathematical Operators |
| 54 |
* Miscellaneous Technical |
| 55 |
* Control Pictures |
| 56 |
* Optical Character Recognition |
| 57 |
* Enclosed Alphanumerics |
| 58 |
* Box Drawing |
| 59 |
* Block Elements |
| 60 |
* Geometric Shapes |
| 61 |
* Miscellaneous Symbols |
| 62 |
* Dingbats |
| 63 |
* Miscellaneous Mathematical Symbols-A |
| 64 |
* Supplemental Arrows-A |
| 65 |
* Braille Patterns |
| 66 |
* Supplemental Arrows-B |
| 67 |
* Miscellaneous Mathematical Symbols-B |
| 68 |
* Supplemental Mathematical Operators |
| 69 |
* Miscellaneous Symbols and Arrows |
| 70 |
*/ |
| 71 |
"\u2000-\u2BFF", |
| 72 |
// Supplemental Punctuation. |
| 73 |
"\u2E00-\u2E7F", |
| 74 |
"]" |
| 75 |
].join(""), |
| 76 |
"g" |
| 77 |
), |
| 78 |
// Remove UTF-16 surrogate points, see https://en.wikipedia.org/wiki/UTF-16#U.2BD800_to_U.2BDFFF |
| 79 |
astralRegExp: /[\uD800-\uDBFF][\uDC00-\uDFFF]/g, |
| 80 |
wordsRegExp: /\S\s+/g, |
| 81 |
characters_excluding_spacesRegExp: /\S/g, |
| 82 |
/* |
| 83 |
* Match anything that is not a formatting character, excluding: |
| 84 |
* \f = form feed |
| 85 |
* \n = new line |
| 86 |
* \r = carriage return |
| 87 |
* \t = tab |
| 88 |
* \v = vertical tab |
| 89 |
* \u00AD = soft hyphen |
| 90 |
* \u2028 = line separator |
| 91 |
* \u2029 = paragraph separator |
| 92 |
*/ |
| 93 |
characters_including_spacesRegExp: /[^\f\n\r\t\v\u00AD\u2028\u2029]/g, |
| 94 |
l10n: { |
| 95 |
type: "words" |
| 96 |
} |
| 97 |
}; |
| 98 |
|
| 99 |
// packages/wordcount/build-module/stripTags.mjs |
| 100 |
function stripTags(settings, text) { |
| 101 |
return text.replace(settings.HTMLRegExp, "\n"); |
| 102 |
} |
| 103 |
|
| 104 |
// packages/wordcount/build-module/transposeAstralsToCountableChar.mjs |
| 105 |
function transposeAstralsToCountableChar(settings, text) { |
| 106 |
return text.replace(settings.astralRegExp, "a"); |
| 107 |
} |
| 108 |
|
| 109 |
// packages/wordcount/build-module/stripHTMLEntities.mjs |
| 110 |
function stripHTMLEntities(settings, text) { |
| 111 |
return text.replace(settings.HTMLEntityRegExp, ""); |
| 112 |
} |
| 113 |
|
| 114 |
// packages/wordcount/build-module/stripConnectors.mjs |
| 115 |
function stripConnectors(settings, text) { |
| 116 |
return text.replace(settings.connectorRegExp, " "); |
| 117 |
} |
| 118 |
|
| 119 |
// packages/wordcount/build-module/stripRemovables.mjs |
| 120 |
function stripRemovables(settings, text) { |
| 121 |
return text.replace(settings.removeRegExp, ""); |
| 122 |
} |
| 123 |
|
| 124 |
// packages/wordcount/build-module/stripHTMLComments.mjs |
| 125 |
function stripHTMLComments(settings, text) { |
| 126 |
return text.replace(settings.HTMLcommentRegExp, ""); |
| 127 |
} |
| 128 |
|
| 129 |
// packages/wordcount/build-module/stripShortcodes.mjs |
| 130 |
function stripShortcodes(settings, text) { |
| 131 |
if (settings.shortcodesRegExp) { |
| 132 |
return text.replace(settings.shortcodesRegExp, "\n"); |
| 133 |
} |
| 134 |
return text; |
| 135 |
} |
| 136 |
|
| 137 |
// packages/wordcount/build-module/stripSpaces.mjs |
| 138 |
function stripSpaces(settings, text) { |
| 139 |
return text.replace(settings.spaceRegExp, " "); |
| 140 |
} |
| 141 |
|
| 142 |
// packages/wordcount/build-module/transposeHTMLEntitiesToCountableChars.mjs |
| 143 |
function transposeHTMLEntitiesToCountableChars(settings, text) { |
| 144 |
return text.replace(settings.HTMLEntityRegExp, "a"); |
| 145 |
} |
| 146 |
|
| 147 |
// packages/wordcount/build-module/index.mjs |
| 148 |
function loadSettings(type = "words", userSettings = {}) { |
| 149 |
const mergedSettings = { ...defaultSettings, ...userSettings }; |
| 150 |
const settings = { |
| 151 |
...mergedSettings, |
| 152 |
type, |
| 153 |
shortcodes: [] |
| 154 |
}; |
| 155 |
settings.shortcodes = settings.l10n?.shortcodes ?? []; |
| 156 |
if (settings.shortcodes && settings.shortcodes.length) { |
| 157 |
settings.shortcodesRegExp = new RegExp( |
| 158 |
"\\[\\/?(?:" + settings.shortcodes.join("|") + ")[^\\]]*?\\]", |
| 159 |
"g" |
| 160 |
); |
| 161 |
} |
| 162 |
if (settings.type !== "characters_excluding_spaces" && settings.type !== "characters_including_spaces") { |
| 163 |
settings.type = "words"; |
| 164 |
} |
| 165 |
return settings; |
| 166 |
} |
| 167 |
function countWords(text, regex, settings) { |
| 168 |
text = [ |
| 169 |
stripTags.bind(null, settings), |
| 170 |
stripHTMLComments.bind(null, settings), |
| 171 |
stripShortcodes.bind(null, settings), |
| 172 |
stripSpaces.bind(null, settings), |
| 173 |
stripHTMLEntities.bind(null, settings), |
| 174 |
stripConnectors.bind(null, settings), |
| 175 |
stripRemovables.bind(null, settings) |
| 176 |
].reduce((result, fn) => fn(result), text); |
| 177 |
text = text + "\n"; |
| 178 |
return text.match(regex)?.length ?? 0; |
| 179 |
} |
| 180 |
function countCharacters(text, regex, settings) { |
| 181 |
text = [ |
| 182 |
stripTags.bind(null, settings), |
| 183 |
stripHTMLComments.bind(null, settings), |
| 184 |
stripShortcodes.bind(null, settings), |
| 185 |
transposeAstralsToCountableChar.bind(null, settings), |
| 186 |
stripSpaces.bind(null, settings), |
| 187 |
transposeHTMLEntitiesToCountableChars.bind(null, settings) |
| 188 |
].reduce((result, fn) => fn(result), text); |
| 189 |
text = text + "\n"; |
| 190 |
return text.match(regex)?.length ?? 0; |
| 191 |
} |
| 192 |
function count(text, type, userSettings) { |
| 193 |
const settings = loadSettings(type, userSettings); |
| 194 |
let matchRegExp; |
| 195 |
switch (settings.type) { |
| 196 |
case "words": |
| 197 |
matchRegExp = settings.wordsRegExp; |
| 198 |
return countWords(text, matchRegExp, settings); |
| 199 |
case "characters_including_spaces": |
| 200 |
matchRegExp = settings.characters_including_spacesRegExp; |
| 201 |
return countCharacters(text, matchRegExp, settings); |
| 202 |
case "characters_excluding_spaces": |
| 203 |
matchRegExp = settings.characters_excluding_spacesRegExp; |
| 204 |
return countCharacters(text, matchRegExp, settings); |
| 205 |
default: |
| 206 |
return 0; |
| 207 |
} |
| 208 |
} |
| 209 |
return __toCommonJS(index_exports); |
| 210 |
})(); |
| 211 |
//# sourceMappingURL=index.js.map |
| 212 |
|