taraskevizer 5.3.2 → 6.0.0

package/README.md

@@ -53,7 +53,7 @@ taraskevizer.convertToHtml('энергія планеты');
 // "en<tarF>erg</tarF>ija p<tarF>lan</tarF>ety"
 
 // properties can be rewritten after creating an object
-taraskevizer.abc = ALPHABET.ARABIC;
+taraskevizer.general.abc = ALPHABET.ARABIC;
 taraskevizer.html.g = true;
 
 const latinizerWithJi = new Taraskevizer({
@@ -108,13 +108,13 @@ If set to false, may cause unwanted changes in acronyms.
 
 Is always `true` in `convertAlphabetOnly`.
 
-### OVERRIDE_taraskevize
+### taraskevize
 
 Type: `(text: string) => string`
 
 Default value: internal function `taraskevize`
 
-Can be overridden in order to make additional changes to the text.
+Overriddes internal function in order to change behaviour of taraskevization.
 This function usually uses private api via `__tarask__`
 
 ## html

package/dist/index.cjs

@@ -2004,7 +2004,6 @@ for (const word of gwords)
   wordlist.push([RegExp(word.replace(/ґ/g, "г"), "g"), word]);
 
 // src/tarask.ts
-var import_diff = require("diff");
 var isUpperCase = (str) => str === str.toUpperCase();
 var getLastLetter = (word, i) => {
   for (let i2 = word.length - 1; i2 >= 0; i2--)
@@ -2097,13 +2096,62 @@ var restoreCase = (text, orig) => {
   }
   return text;
 };
-var highlightChanges = (text, orig, highlight) => (0, import_diff.diffChars)(orig, text).reduce(
-  (acc, diff) => acc + (diff.removed ? "�" : diff.added ? highlight(diff.value) : diff.value),
-  ""
-).replace(
-  /([^<>\x1b])�([^<>\x1b])/g,
-  ($0, $1, $2) => highlight($1 + $2)
-).replace(/�/g, "");
+var highlightDiff = (text, orig, isCyrillic, highlight) => {
+  for (let i = 0; i < text.length; i++) {
+    const word = text[i];
+    const oWord = orig[i];
+    if (oWord === word)
+      continue;
+    const wordH = isCyrillic ? replaceG(word, ($0) => gobj[$0]) : word;
+    if (oWord === wordH)
+      continue;
+    if (!/\(/.test(word)) {
+      if (word.length === oWord.length) {
+        const wordLetters = word.split("");
+        for (let j = 0; j < wordLetters.length; j++) {
+          if (wordH[j] !== oWord[j])
+            wordLetters[j] = highlight(wordLetters[j]);
+        }
+        text[i] = wordLetters.join("");
+        continue;
+      }
+      if (isCyrillic) {
+        const word1 = word.replace(/ь/g, "");
+        switch (oWord) {
+          case word1:
+            text[i] = word.replace(/ь/g, highlight("ь"));
+            continue;
+          case word1 + "ь":
+            text[i] = word.slice(0, -1).replace(/ь/g, highlight("ь")) + "ь";
+            continue;
+        }
+      }
+    }
+    const oWordEnd = oWord.length - 1;
+    let fromStart = 0;
+    let fromWordEnd = word.length - 1;
+    let fromOWordEnd = oWordEnd;
+    while (wordH[fromStart] === oWord[fromStart])
+      ++fromStart;
+    while (wordH[fromWordEnd] === oWord[fromOWordEnd]) {
+      --fromWordEnd;
+      --fromOWordEnd;
+    }
+    if (oWord.length < word.length) {
+      if (fromOWordEnd === oWordEnd) {
+        text[i] = highlight(word);
+        continue;
+      }
+      if (fromWordEnd < 0)
+        fromWordEnd = 0;
+    }
+    if (fromStart === fromWordEnd + 1) {
+      --fromStart;
+      ++fromWordEnd;
+    }
+    text[i] = word.slice(0, fromStart) + highlight(word.slice(fromStart, fromWordEnd + 1)) + word.slice(fromWordEnd + 1);
+  }
+};
 var replaceWithDict = (text, dict = []) => {
   for (const [pattern, result] of dict)
     text = text.replace(
@@ -2126,9 +2174,11 @@ var __tarask__ = {
 };
 var convertAlphabet = (text, abc) => replaceWithDict(replaceWithDict(text, letters[abc]), lettersUpperCase[abc]);
 var Taraskevizer = class {
-  abc = ALPHABET.CYRILLIC;
-  j = REPLACE_J.NEVER;
-  doEscapeCapitalized = true;
+  general = {
+    abc: ALPHABET.CYRILLIC,
+    j: REPLACE_J.NEVER,
+    doEscapeCapitalized: true
+  };
   html = {
     g: false
   };
@@ -2140,34 +2190,21 @@ var Taraskevizer = class {
   constructor(options) {
     if (!options)
       return;
-    const general = options.general;
-    if (general) {
-      for (const prop of [
-        "abc",
-        "j",
-        "doEscapeCapitalized"
-      ])
-        if (prop in general)
-          this[prop] = general[prop];
-    }
-    if (options.OVERRIDE_taraskevize)
-      this.taraskevize = options.OVERRIDE_taraskevize;
+    Object.assign(this.general, options.general);
     Object.assign(this.html, options.html);
     Object.assign(this.nonHtml, options.nonHtml);
   }
   convert(text) {
     const wrapInColorOf = wrappers.ansiColors;
+    const isCyrillic = this.general.abc === ALPHABET.CYRILLIC;
     const noFixArr = [];
-    text = this.prepare(text, noFixArr, "<");
-    const origInTargetAlphabet = convertAlphabet(text, this.abc);
-    text = this.process(text, origInTargetAlphabet);
+    const { splitted, splittedOrig } = this.process(
+      this.prepare(text, noFixArr, "<")
+    );
     if (this.nonHtml.ansiColors)
-      text = highlightChanges(
-        text,
-        afterJoin(origInTargetAlphabet),
-        wrapInColorOf.fix
-      );
-    if (this.abc === ALPHABET.CYRILLIC && (this.nonHtml.h || this.nonHtml.ansiColors))
+      highlightDiff(splitted, splittedOrig, isCyrillic, wrapInColorOf.fix);
+    text = join(splitted);
+    if (isCyrillic && (this.nonHtml.h || this.nonHtml.ansiColors))
       text = replaceG(
         text,
         this.nonHtml.ansiColors ? this.nonHtml.h ? ($0) => wrapInColorOf.variable(gobj[$0]) : wrapInColorOf.variable("$&") : ($0) => gobj[$0]
@@ -2184,15 +2221,14 @@ var Taraskevizer = class {
   }
   convertToHtml(text) {
     const wrapInTag = wrappers.html;
+    const isCyrillic = this.general.abc === ALPHABET.CYRILLIC;
     const noFixArr = [];
-    text = this.prepare(text, noFixArr, "&lt;");
-    const origInTargetAlphabet = convertAlphabet(text, this.abc);
-    text = highlightChanges(
-      this.process(text, origInTargetAlphabet),
-      afterJoin(origInTargetAlphabet),
-      wrapInTag.fix
+    const { splitted, splittedOrig } = this.process(
+      this.prepare(text, noFixArr, "&lt;")
     );
-    if (this.abc === ALPHABET.CYRILLIC)
+    highlightDiff(splitted, splittedOrig, isCyrillic, wrapInTag.fix);
+    text = join(splitted);
+    if (isCyrillic)
       text = replaceG(
         text,
         this.html.g ? wrapInTag.letterH("$&") : ($0) => wrapInTag.letterH(gobj[$0])
@@ -2206,7 +2242,7 @@ var Taraskevizer = class {
       "<br>"
     );
   }
-  prepare(text, noFixArr, LEFT_ANGLE_BRACKET, doEscapeCapitalized = this.doEscapeCapitalized) {
+  prepare(text, noFixArr, LEFT_ANGLE_BRACKET, doEscapeCapitalized = this.general.doEscapeCapitalized) {
     text = ` ${text.trim()} `.replace(//g, "");
     if (doEscapeCapitalized)
       text = text.replace(
@@ -2218,7 +2254,7 @@ var Taraskevizer = class {
         return LEFT_ANGLE_BRACKET + $3 + ">";
       if ($1)
         $3 = restoreCase(
-          [replaceWithDict($3.toLowerCase(), letters[this.abc])],
+          [replaceWithDict($3.toLowerCase(), letters[this.general.abc])],
           [$3]
         );
       noFixArr.push($2 === "." ? $3 : LEFT_ANGLE_BRACKET + $3 + ">");
@@ -2230,14 +2266,17 @@ var Taraskevizer = class {
     return finalize(
       applyNoFix(
         noFixArr,
-        convertAlphabet(this.prepare(text, noFixArr, "<", false), this.abc)
+        convertAlphabet(
+          this.prepare(text, noFixArr, "<", false),
+          this.general.abc
+        )
       ).replace(/&nbsp;/g, " ").replace(new RegExp(" (\\p{P}|\\p{S}|\\d|&#40) ", "gu"), "$1"),
       "\n"
     );
   }
-  process(text, textInTargetAlphabet) {
-    const { abc, j } = this;
-    const splittedOrig = textInTargetAlphabet.split(" ");
+  process(text) {
+    const { abc, j } = this.general;
+    const splittedOrig = convertAlphabet(text, abc).split(" ");
     text = this.taraskevize(text.toLowerCase());
     if (j && abc !== ALPHABET.LATIN_JI)
       text = replaceIbyJ(text, j === REPLACE_J.ALWAYS);
@@ -2245,7 +2284,7 @@ var Taraskevizer = class {
     let splitted = text.split(" ");
     if (abc !== ALPHABET.ARABIC)
       splitted = restoreCase(splitted, splittedOrig);
-    return join(splitted);
+    return { splittedOrig, splitted };
   }
   taraskevize(text) {
     text = replaceWithDict(text, wordlist);

package/dist/index.d.ts

@@ -47,9 +47,11 @@ declare const __tarask__: {
     readonly afterTarask: ExtendedDict;
 };
 declare class Taraskevizer {
-    abc: Alphabet;
-    j: OptionJ;
-    doEscapeCapitalized: boolean;
+    general: {
+        abc: Alphabet;
+        j: OptionJ;
+        doEscapeCapitalized: boolean;
+    };
     html: {
         g: boolean;
     };

package/dist/index.js

@@ -1973,7 +1973,6 @@ for (const word of gwords)
   wordlist.push([RegExp(word.replace(/ґ/g, "г"), "g"), word]);
 
 // src/tarask.ts
-import { diffChars } from "diff";
 var isUpperCase = (str) => str === str.toUpperCase();
 var getLastLetter = (word, i) => {
   for (let i2 = word.length - 1; i2 >= 0; i2--)
@@ -2066,13 +2065,62 @@ var restoreCase = (text, orig) => {
   }
   return text;
 };
-var highlightChanges = (text, orig, highlight) => diffChars(orig, text).reduce(
-  (acc, diff) => acc + (diff.removed ? "�" : diff.added ? highlight(diff.value) : diff.value),
-  ""
-).replace(
-  /([^<>\x1b])�([^<>\x1b])/g,
-  ($0, $1, $2) => highlight($1 + $2)
-).replace(/�/g, "");
+var highlightDiff = (text, orig, isCyrillic, highlight) => {
+  for (let i = 0; i < text.length; i++) {
+    const word = text[i];
+    const oWord = orig[i];
+    if (oWord === word)
+      continue;
+    const wordH = isCyrillic ? replaceG(word, ($0) => gobj[$0]) : word;
+    if (oWord === wordH)
+      continue;
+    if (!/\(/.test(word)) {
+      if (word.length === oWord.length) {
+        const wordLetters = word.split("");
+        for (let j = 0; j < wordLetters.length; j++) {
+          if (wordH[j] !== oWord[j])
+            wordLetters[j] = highlight(wordLetters[j]);
+        }
+        text[i] = wordLetters.join("");
+        continue;
+      }
+      if (isCyrillic) {
+        const word1 = word.replace(/ь/g, "");
+        switch (oWord) {
+          case word1:
+            text[i] = word.replace(/ь/g, highlight("ь"));
+            continue;
+          case word1 + "ь":
+            text[i] = word.slice(0, -1).replace(/ь/g, highlight("ь")) + "ь";
+            continue;
+        }
+      }
+    }
+    const oWordEnd = oWord.length - 1;
+    let fromStart = 0;
+    let fromWordEnd = word.length - 1;
+    let fromOWordEnd = oWordEnd;
+    while (wordH[fromStart] === oWord[fromStart])
+      ++fromStart;
+    while (wordH[fromWordEnd] === oWord[fromOWordEnd]) {
+      --fromWordEnd;
+      --fromOWordEnd;
+    }
+    if (oWord.length < word.length) {
+      if (fromOWordEnd === oWordEnd) {
+        text[i] = highlight(word);
+        continue;
+      }
+      if (fromWordEnd < 0)
+        fromWordEnd = 0;
+    }
+    if (fromStart === fromWordEnd + 1) {
+      --fromStart;
+      ++fromWordEnd;
+    }
+    text[i] = word.slice(0, fromStart) + highlight(word.slice(fromStart, fromWordEnd + 1)) + word.slice(fromWordEnd + 1);
+  }
+};
 var replaceWithDict = (text, dict = []) => {
   for (const [pattern, result] of dict)
     text = text.replace(
@@ -2095,9 +2143,11 @@ var __tarask__ = {
 };
 var convertAlphabet = (text, abc) => replaceWithDict(replaceWithDict(text, letters[abc]), lettersUpperCase[abc]);
 var Taraskevizer = class {
-  abc = ALPHABET.CYRILLIC;
-  j = REPLACE_J.NEVER;
-  doEscapeCapitalized = true;
+  general = {
+    abc: ALPHABET.CYRILLIC,
+    j: REPLACE_J.NEVER,
+    doEscapeCapitalized: true
+  };
   html = {
     g: false
   };
@@ -2109,34 +2159,21 @@ var Taraskevizer = class {
   constructor(options) {
     if (!options)
       return;
-    const general = options.general;
-    if (general) {
-      for (const prop of [
-        "abc",
-        "j",
-        "doEscapeCapitalized"
-      ])
-        if (prop in general)
-          this[prop] = general[prop];
-    }
-    if (options.OVERRIDE_taraskevize)
-      this.taraskevize = options.OVERRIDE_taraskevize;
+    Object.assign(this.general, options.general);
     Object.assign(this.html, options.html);
     Object.assign(this.nonHtml, options.nonHtml);
   }
   convert(text) {
     const wrapInColorOf = wrappers.ansiColors;
+    const isCyrillic = this.general.abc === ALPHABET.CYRILLIC;
     const noFixArr = [];
-    text = this.prepare(text, noFixArr, "<");
-    const origInTargetAlphabet = convertAlphabet(text, this.abc);
-    text = this.process(text, origInTargetAlphabet);
+    const { splitted, splittedOrig } = this.process(
+      this.prepare(text, noFixArr, "<")
+    );
     if (this.nonHtml.ansiColors)
-      text = highlightChanges(
-        text,
-        afterJoin(origInTargetAlphabet),
-        wrapInColorOf.fix
-      );
-    if (this.abc === ALPHABET.CYRILLIC && (this.nonHtml.h || this.nonHtml.ansiColors))
+      highlightDiff(splitted, splittedOrig, isCyrillic, wrapInColorOf.fix);
+    text = join(splitted);
+    if (isCyrillic && (this.nonHtml.h || this.nonHtml.ansiColors))
       text = replaceG(
         text,
         this.nonHtml.ansiColors ? this.nonHtml.h ? ($0) => wrapInColorOf.variable(gobj[$0]) : wrapInColorOf.variable("$&") : ($0) => gobj[$0]
@@ -2153,15 +2190,14 @@ var Taraskevizer = class {
   }
   convertToHtml(text) {
     const wrapInTag = wrappers.html;
+    const isCyrillic = this.general.abc === ALPHABET.CYRILLIC;
     const noFixArr = [];
-    text = this.prepare(text, noFixArr, "&lt;");
-    const origInTargetAlphabet = convertAlphabet(text, this.abc);
-    text = highlightChanges(
-      this.process(text, origInTargetAlphabet),
-      afterJoin(origInTargetAlphabet),
-      wrapInTag.fix
+    const { splitted, splittedOrig } = this.process(
+      this.prepare(text, noFixArr, "&lt;")
     );
-    if (this.abc === ALPHABET.CYRILLIC)
+    highlightDiff(splitted, splittedOrig, isCyrillic, wrapInTag.fix);
+    text = join(splitted);
+    if (isCyrillic)
       text = replaceG(
         text,
         this.html.g ? wrapInTag.letterH("$&") : ($0) => wrapInTag.letterH(gobj[$0])
@@ -2175,7 +2211,7 @@ var Taraskevizer = class {
       "<br>"
     );
   }
-  prepare(text, noFixArr, LEFT_ANGLE_BRACKET, doEscapeCapitalized = this.doEscapeCapitalized) {
+  prepare(text, noFixArr, LEFT_ANGLE_BRACKET, doEscapeCapitalized = this.general.doEscapeCapitalized) {
     text = ` ${text.trim()} `.replace(//g, "");
     if (doEscapeCapitalized)
       text = text.replace(
@@ -2187,7 +2223,7 @@ var Taraskevizer = class {
         return LEFT_ANGLE_BRACKET + $3 + ">";
       if ($1)
         $3 = restoreCase(
-          [replaceWithDict($3.toLowerCase(), letters[this.abc])],
+          [replaceWithDict($3.toLowerCase(), letters[this.general.abc])],
           [$3]
         );
       noFixArr.push($2 === "." ? $3 : LEFT_ANGLE_BRACKET + $3 + ">");
@@ -2199,14 +2235,17 @@ var Taraskevizer = class {
     return finalize(
       applyNoFix(
         noFixArr,
-        convertAlphabet(this.prepare(text, noFixArr, "<", false), this.abc)
+        convertAlphabet(
+          this.prepare(text, noFixArr, "<", false),
+          this.general.abc
+        )
       ).replace(/&nbsp;/g, " ").replace(new RegExp(" (\\p{P}|\\p{S}|\\d|&#40) ", "gu"), "$1"),
       "\n"
     );
   }
-  process(text, textInTargetAlphabet) {
-    const { abc, j } = this;
-    const splittedOrig = textInTargetAlphabet.split(" ");
+  process(text) {
+    const { abc, j } = this.general;
+    const splittedOrig = convertAlphabet(text, abc).split(" ");
     text = this.taraskevize(text.toLowerCase());
     if (j && abc !== ALPHABET.LATIN_JI)
       text = replaceIbyJ(text, j === REPLACE_J.ALWAYS);
@@ -2214,7 +2253,7 @@ var Taraskevizer = class {
     let splitted = text.split(" ");
     if (abc !== ALPHABET.ARABIC)
       splitted = restoreCase(splitted, splittedOrig);
-    return join(splitted);
+    return { splittedOrig, splitted };
   }
   taraskevize(text) {
     text = replaceWithDict(text, wordlist);

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "taraskevizer",
-	"version": "5.3.2",
+	"version": "6.0.0",
 	"author": "GooseOb",
 	"repository": {
 		"type": "git",
@@ -8,9 +8,6 @@
 	},
 	"main": "dist/index.js",
 	"module": "dist/index.js",
-	"dependencies": {
-		"diff": "^5.2.0"
-	},
 	"devDependencies": {
 		"@digitak/esrun": "^3.2.26",
 		"@types/diff": "^5.2.0",