From 40c43572e97b499e40774152d5662c7c005d05b1 Mon Sep 17 00:00:00 2001 From: samsonsin Date: Thu, 28 Aug 2025 19:04:16 +0200 Subject: [PATCH 1/2] Updated default TTS Multi Voice .js file --- .../main/assets/defaultData/speech_rule.js | 108 +++++++++++------- 1 file changed, 68 insertions(+), 40 deletions(-) diff --git a/app/src/main/assets/defaultData/speech_rule.js b/app/src/main/assets/defaultData/speech_rule.js index 3fd088e27..2e6f0aeb2 100644 --- a/app/src/main/assets/defaultData/speech_rule.js +++ b/app/src/main/assets/defaultData/speech_rule.js @@ -1,52 +1,80 @@ let SpeechRuleJS = { - name: "旁白/对话", + name: "MultiVoice", id: "ttsrv.multi_voice", - author: "TTS Server", - version: 4, - tags: {narration: "旁白", dialogue: "对话"}, + author: "samsonsin", + version: 1, + tags: { + narration: "Narration", + dialogue: "Dialogue", + squareBrackets: "Square Brackets", + curlyBrackets: "Curly Brackets", + }, handleText(text) { - const list = []; - let tmpStr = ""; - let endTag = "narration"; - - text.split("").forEach((char, index) => { - tmpStr += char; - - if (char === '“') { - endTag = "dialogue"; - list.push({text: tmpStr, tag: "narration"}); - tmpStr = ""; - } else if (char === '”') { - endTag = "narration"; - tmpStr = tmpStr.slice(0, -1) - list.push({text: tmpStr, tag: "dialogue"}); - tmpStr = ""; - } else if (index === text.length - 1) { - list.push({text: tmpStr, tag: endTag}); + //Define regex patterns for different text types. The order matters, as the first match takes precedence. + let regexExpressions = [ + { regex: /.+/g, tag: "narration" }, + { regex: /["“][^"“”]*["”]?/g, tag: "dialogue" }, + { regex: /\\[([^\\]]*?\\s+[^\\]]*?\\s+[^\\]]*?)\\]/g, tag: "squareBrackets" }, + { regex: /{[^}]*?}/g, tag: "curlyBrackets" }, + + //Final rule to match and ignore specific characters by assigning them a null tag. + { regex: /[•*{}"“”]/g, tag: null} + ]; + + //Initialize an array to hold the regex matches. + let regexMatches = new Array(text.length + 1).fill(null); + + //Iterate through each regex pattern and assign the corresponding tag to the matched char in the regexMatches array + regexExpressions.forEach((rule) => { + while ((queryResult = rule.regex.exec(text)) !== null) { + //If the match is empty, increment the last index of the regex and continue to the next iteration, skipping the char + if (queryResult[0].length === 0) { + rule.regex.lastIndex++; + continue; + } + regexMatches.fill( + rule.tag, + queryResult.index, + rule.regex.lastIndex + ); } }); - return list; - }, + //two moving indices to track the current position in the text and the end of last segment. When discovering a new segment, + //push current segment to output array and update the previous tag and index. regexMatches final index is null to ensure the last segment is captured. + let previousTag = regexMatches[0]; + let previousIndex = 0; + let textSegments = []; + for (let i = 1; i < regexMatches.length; i++) { + var isInMiddleOfTextSegment = previousTag === regexMatches[i]; + var isCurrentSelectionEmptyOrContainsNull = + text.slice(previousIndex, i).trim().length === 0 || + previousTag === null; - splitText(text) { - let separatorStr = "。？?！!;；" - - let list = [] - let tmpStr = "" - text.split("").forEach((char, index) => { - tmpStr += char - - if (separatorStr.includes(char)) { - list.push(tmpStr) - tmpStr = "" - } else if (index === text.length - 1) { - list.push(tmpStr); + if (isInMiddleOfTextSegment) continue; + if (isCurrentSelectionEmptyOrContainsNull) { + previousTag = regexMatches[i]; + previousIndex = i; + continue; } - }) - return list.filter(item => item.replace(/[“”]/g, '').trim().length > 0); - } + textSegments.push({ + text: text.slice(previousIndex, i).trim(), + tag: previousTag, + }); + previousTag = regexMatches[i]; + previousIndex = i; + } + return textSegments; + }, + splitText(text) { + //Regex matches immediately after specific characters, splitting the text into segments. + //Then asserts that each segment, after removing some characters, is non-empty. + //This implementation leverages regex operations rather than custom made logic, and is equivalent to the original in my testing + return (text + .split(/[。？?！!;；]\\K/g) + .filter((item) => item.replace(/[“”]/g, "").trim().length > 0)); + }, }; From c8ea4a4f21c7c0e731098f211eae96b2f35a45ae Mon Sep 17 00:00:00 2001 From: samsonsin Date: Tue, 9 Sep 2025 03:53:12 +0200 Subject: [PATCH 2/2] Fixed regex brackets rule. Now it actually works, and supports 3 deep nesting as well as multiple brackets in a row. --- app/src/main/assets/defaultData/speech_rule.js | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/app/src/main/assets/defaultData/speech_rule.js b/app/src/main/assets/defaultData/speech_rule.js index 2e6f0aeb2..6cac9bf64 100644 --- a/app/src/main/assets/defaultData/speech_rule.js +++ b/app/src/main/assets/defaultData/speech_rule.js @@ -15,7 +15,7 @@ let SpeechRuleJS = { let regexExpressions = [ { regex: /.+/g, tag: "narration" }, { regex: /["“][^"“”]*["”]?/g, tag: "dialogue" }, - { regex: /\\[([^\\]]*?\\s+[^\\]]*?\\s+[^\\]]*?)\\]/g, tag: "squareBrackets" }, + { regex: /\[[^\]\[]*(\[[^\]\[]*(\[[^\]\[]*\])*[^\]\[]*\])*[^\]\[]*(\[[^\]\[]*(\[[^\]\[]*\])*[^\]\[]*\])*[^\]\[]*(\[[^\]\[]*(\[[^\]\[]*\])*[^\]\[]*\])*[^\]\[]*\]/g, tag: "squareBrackets" }, { regex: /{[^}]*?}/g, tag: "curlyBrackets" }, //Final rule to match and ignore specific characters by assigning them a null tag.