taraskevizer 5.2.6 → 5.3.0

package/dist/index.cjs

@@ -2003,6 +2003,7 @@ 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--)
@@ -2094,58 +2095,13 @@ var restoreCase = (text, orig) => {
   }
   return text;
 };
-var highlightChanges = (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;
-    }
-    text[i] = word.slice(0, fromStart) + highlight(word.slice(fromStart, fromWordEnd + 1)) + word.slice(fromWordEnd + 1);
-  }
-};
+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 replaceWithDict = (text, dict = []) => {
   for (const [pattern, result] of dict)
     text = text.replace(
@@ -2199,15 +2155,13 @@ var Taraskevizer = class {
   }
   convert(text) {
     const wrapInColorOf = wrappers.ansiColors;
-    const isCyrillic = this.abc === ALPHABET.CYRILLIC;
     const noFixArr = [];
-    const { splitted, splittedOrig } = this.process(
-      this.prepare(text, noFixArr, "<")
-    );
+    text = this.prepare(text, noFixArr, "<");
+    const origInTargetAlphabet = convertAlphabet(text, this.abc);
+    text = this.process(text, origInTargetAlphabet);
     if (this.nonHtml.ansiColors)
-      highlightChanges(splitted, splittedOrig, isCyrillic, wrapInColorOf.fix);
-    text = join(splitted);
-    if (isCyrillic && (this.nonHtml.h || this.nonHtml.ansiColors))
+      text = highlightChanges(text, origInTargetAlphabet, wrapInColorOf.fix);
+    if (this.abc === ALPHABET.CYRILLIC && (this.nonHtml.h || this.nonHtml.ansiColors))
       text = replaceG(
         text,
         this.nonHtml.ansiColors ? this.nonHtml.h ? ($0) => wrapInColorOf.variable(gobj[$0]) : wrapInColorOf.variable("$&") : ($0) => gobj[$0]
@@ -2224,14 +2178,15 @@ var Taraskevizer = class {
   }
   convertToHtml(text) {
     const wrapInTag = wrappers.html;
-    const isCyrillic = this.abc === ALPHABET.CYRILLIC;
     const noFixArr = [];
-    const { splitted, splittedOrig } = this.process(
-      this.prepare(text, noFixArr, "&lt;")
+    text = this.prepare(text, noFixArr, "&lt;");
+    const origInTargetAlphabet = convertAlphabet(text, this.abc);
+    text = highlightChanges(
+      this.process(text, origInTargetAlphabet),
+      origInTargetAlphabet,
+      wrapInTag.fix
     );
-    highlightChanges(splitted, splittedOrig, isCyrillic, wrapInTag.fix);
-    text = join(splitted);
-    if (isCyrillic)
+    if (this.abc === ALPHABET.CYRILLIC)
       text = replaceG(
         text,
         this.html.g ? wrapInTag.letterH("$&") : ($0) => wrapInTag.letterH(gobj[$0])
@@ -2274,9 +2229,9 @@ var Taraskevizer = class {
       "\n"
     );
   }
-  process(text) {
+  process(text, textInTargetAlphabet) {
     const { abc, j } = this;
-    const splittedOrig = convertAlphabet(text, abc).split(" ");
+    const splittedOrig = textInTargetAlphabet.split(" ");
     text = this.taraskevize(text.toLowerCase());
     if (j && abc !== ALPHABET.LATIN_JI)
       text = replaceIbyJ(text, j === REPLACE_J.ALWAYS);
@@ -2284,7 +2239,7 @@ var Taraskevizer = class {
     let splitted = text.split(" ");
     if (abc !== ALPHABET.ARABIC)
       splitted = restoreCase(splitted, splittedOrig);
-    return { splittedOrig, splitted };
+    return join(splitted);
   }
   taraskevize(text) {
     text = replaceWithDict(text, wordlist);

package/dist/index.js

@@ -1972,6 +1972,7 @@ 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--)
@@ -2063,58 +2064,13 @@ var restoreCase = (text, orig) => {
   }
   return text;
 };
-var highlightChanges = (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;
-    }
-    text[i] = word.slice(0, fromStart) + highlight(word.slice(fromStart, fromWordEnd + 1)) + word.slice(fromWordEnd + 1);
-  }
-};
+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 replaceWithDict = (text, dict = []) => {
   for (const [pattern, result] of dict)
     text = text.replace(
@@ -2168,15 +2124,13 @@ var Taraskevizer = class {
   }
   convert(text) {
     const wrapInColorOf = wrappers.ansiColors;
-    const isCyrillic = this.abc === ALPHABET.CYRILLIC;
     const noFixArr = [];
-    const { splitted, splittedOrig } = this.process(
-      this.prepare(text, noFixArr, "<")
-    );
+    text = this.prepare(text, noFixArr, "<");
+    const origInTargetAlphabet = convertAlphabet(text, this.abc);
+    text = this.process(text, origInTargetAlphabet);
     if (this.nonHtml.ansiColors)
-      highlightChanges(splitted, splittedOrig, isCyrillic, wrapInColorOf.fix);
-    text = join(splitted);
-    if (isCyrillic && (this.nonHtml.h || this.nonHtml.ansiColors))
+      text = highlightChanges(text, origInTargetAlphabet, wrapInColorOf.fix);
+    if (this.abc === ALPHABET.CYRILLIC && (this.nonHtml.h || this.nonHtml.ansiColors))
       text = replaceG(
         text,
         this.nonHtml.ansiColors ? this.nonHtml.h ? ($0) => wrapInColorOf.variable(gobj[$0]) : wrapInColorOf.variable("$&") : ($0) => gobj[$0]
@@ -2193,14 +2147,15 @@ var Taraskevizer = class {
   }
   convertToHtml(text) {
     const wrapInTag = wrappers.html;
-    const isCyrillic = this.abc === ALPHABET.CYRILLIC;
     const noFixArr = [];
-    const { splitted, splittedOrig } = this.process(
-      this.prepare(text, noFixArr, "&lt;")
+    text = this.prepare(text, noFixArr, "&lt;");
+    const origInTargetAlphabet = convertAlphabet(text, this.abc);
+    text = highlightChanges(
+      this.process(text, origInTargetAlphabet),
+      origInTargetAlphabet,
+      wrapInTag.fix
     );
-    highlightChanges(splitted, splittedOrig, isCyrillic, wrapInTag.fix);
-    text = join(splitted);
-    if (isCyrillic)
+    if (this.abc === ALPHABET.CYRILLIC)
       text = replaceG(
         text,
         this.html.g ? wrapInTag.letterH("$&") : ($0) => wrapInTag.letterH(gobj[$0])
@@ -2243,9 +2198,9 @@ var Taraskevizer = class {
       "\n"
     );
   }
-  process(text) {
+  process(text, textInTargetAlphabet) {
     const { abc, j } = this;
-    const splittedOrig = convertAlphabet(text, abc).split(" ");
+    const splittedOrig = textInTargetAlphabet.split(" ");
     text = this.taraskevize(text.toLowerCase());
     if (j && abc !== ALPHABET.LATIN_JI)
       text = replaceIbyJ(text, j === REPLACE_J.ALWAYS);
@@ -2253,7 +2208,7 @@ var Taraskevizer = class {
     let splitted = text.split(" ");
     if (abc !== ALPHABET.ARABIC)
       splitted = restoreCase(splitted, splittedOrig);
-    return { splittedOrig, splitted };
+    return join(splitted);
   }
   taraskevize(text) {
     text = replaceWithDict(text, wordlist);

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "taraskevizer",
-	"version": "5.2.6",
+	"version": "5.3.0",
 	"author": "GooseOb",
 	"repository": {
 		"type": "git",
@@ -8,14 +8,18 @@
 	},
 	"main": "dist/index.js",
 	"module": "dist/index.js",
+	"dependencies": {
+		"diff": "^5.2.0"
+	},
 	"devDependencies": {
-		"@digitak/esrun": "^3.2.25",
-		"@types/node": "^20.10.4",
-		"bun-types": "^1.0.17",
-		"husky": "^8.0.3",
-		"prettier": "^3.1.1",
-		"tsup": "^7.2.0",
-		"typescript": "^5.3.3"
+		"@digitak/esrun": "^3.2.26",
+		"@types/diff": "^5.2.0",
+		"@types/node": "^20.12.7",
+		"bun-types": "^1.1.4",
+		"husky": "^9.0.11",
+		"prettier": "^3.2.5",
+		"tsup": "^8.0.2",
+		"typescript": "^5.4.5"
 	},
 	"bin": {
 		"tarask": "dist/bin.js"
@@ -46,7 +50,7 @@
 		"dev-bun": "bun test --watch",
 		"test": "esrun --send-code-mode=temporaryFile test",
 		"test-cli": "bun run build && esrun --send-code-mode=temporaryFile test",
-		"prepare": "husky install",
+		"prepare": "husky",
 		"typecheck": "tsc --project src/tsconfig.json"
 	},
 	"type": "module",