@intlayer/cli 8.1.1 → 8.1.3-canary.0

package/dist/esm/translateDoc/translateFile.mjs

@@ -1,102 +1,3 @@
-import { readAsset } from "../_virtual/_utils_asset.mjs";
-import { sanitizeChunk, validateTranslation } from "./validation.mjs";
-import { chunkInference } from "../utils/chunkInference.mjs";
-import { fixChunkStartEndChars } from "../utils/fixChunkStartEndChars.mjs";
-import { chunkText } from "../utils/calculateChunks.mjs";
-import { formatLocale, formatPath } from "@intlayer/chokidar";
-import { ANSIColors, colon, colorize, colorizeNumber, getAppLogger, retryManager } from "@intlayer/config";
-import { dirname, relative } from "node:path";
-import { mkdirSync, writeFileSync } from "node:fs";
-import { readFile } from "node:fs/promises";
-import { performance } from "node:perf_hooks";
-
-//#region src/translateDoc/translateFile.ts
-const translateFile = async ({ baseFilePath, outputFilePath, locale, baseLocale, configuration, errorState, aiOptions, customInstructions, aiClient, aiConfig, flushStrategy = "incremental", onChunkReceive, limit }) => {
-	if (errorState.shouldStop) return null;
-	const appLogger = getAppLogger(configuration, { config: { prefix: "" } });
-	const fileStartTime = performance.now();
-	try {
-		const chunks = chunkText(await readFile(baseFilePath, "utf-8"));
-		const totalChunks = chunks.length;
-		const filePrefix = `${colon(`${ANSIColors.GREY_DARK}[${formatPath(baseFilePath)}${ANSIColors.GREY_DARK}] `, { colSize: 40 })}${ANSIColors.RESET}`;
-		const prefix = `${colon(`${ANSIColors.GREY_DARK}[${formatPath(baseFilePath)}${ANSIColors.GREY_DARK}][${formatLocale(locale)}${ANSIColors.GREY_DARK}] `, { colSize: 40 })}${ANSIColors.RESET}`;
-		appLogger(`${filePrefix}Split into ${colorizeNumber(totalChunks)} chunks. Queuing...`);
-		const basePrompt = readAsset("./prompts/TRANSLATE_PROMPT.md", "utf-8").replaceAll("{{localeName}}", `${formatLocale(locale, false)}`).replaceAll("{{baseLocaleName}}", `${formatLocale(baseLocale, false)}`).replace("{{applicationContext}}", aiOptions?.applicationContext ?? "-").replace("{{customInstructions}}", customInstructions ?? "-");
-		const translatedParts = new Array(totalChunks).fill("");
-		const runTask = limit ?? ((fn) => fn());
-		const tasks = chunks.map((chunk, i) => runTask(async () => {
-			if (errorState.shouldStop) return null;
-			const chunkLogger = getAppLogger(configuration, { config: { prefix: `${prefix}  ${ANSIColors.GREY_DARK}[${i + 1}/${totalChunks}] ${ANSIColors.RESET}` } });
-			const chunkStartTime = performance.now();
-			const isFirstChunk = i === 0;
-			const fileToTranslateCurrentChunk = chunk.content;
-			const getPrevChunkPrompt = () => `>>> CONTEXT: PREVIOUS SOURCE CONTENT <<<\n\`\`\`\n` + (chunks[i - 1]?.content ?? "") + `\n\`\`\`\n>>> END PREVIOUS CONTEXT <<<`;
-			const getBaseChunkContextPrompt = () => `>>> CONTEXT: NEXT CONTENT <<<\n\`\`\`\n` + (chunks[i + 1]?.content ?? "") + `\n\`\`\`\n>>> END NEXT CONTEXT <<<`;
-			chunkLogger("Process started");
-			const { content: translatedChunk, tokens } = await retryManager(async () => {
-				const result = await chunkInference([
-					{
-						role: "system",
-						content: basePrompt
-					},
-					...chunks[i + 1] ? [{
-						role: "system",
-						content: getBaseChunkContextPrompt()
-					}] : [],
-					...isFirstChunk ? [] : [{
-						role: "system",
-						content: getPrevChunkPrompt()
-					}],
-					{
-						role: "system",
-						content: [`You are translating TARGET CHUNK (${i + 1}/${totalChunks}).`, `Translate ONLY the target chunk. Preserve frontmatter/code exactly.`].join("\n")
-					},
-					{
-						role: "user",
-						content: `>>> TARGET CHUNK START <<<\n${fileToTranslateCurrentChunk}\n>>> TARGET CHUNK END <<<`
-					}
-				], aiOptions, configuration, aiClient, aiConfig);
-				let processedChunk = sanitizeChunk(result?.fileContent, fileToTranslateCurrentChunk);
-				processedChunk = fixChunkStartEndChars(processedChunk, fileToTranslateCurrentChunk);
-				if (!validateTranslation(fileToTranslateCurrentChunk, processedChunk, chunkLogger)) throw new Error(`Validation failed for chunk ${i + 1}/${totalChunks}`);
-				return {
-					content: processedChunk,
-					tokens: result.tokenUsed
-				};
-			})();
-			const chunkDuration = (performance.now() - chunkStartTime).toFixed(0);
-			translatedParts[i] = translatedChunk;
-			if (onChunkReceive) onChunkReceive(translatedChunk, i, totalChunks);
-			if (flushStrategy === "incremental") {
-				if (translatedParts.slice(0, i + 1).every((p) => p && p !== "")) {
-					let endIdx = 0;
-					while (endIdx < totalChunks && translatedParts[endIdx] && translatedParts[endIdx] !== "") endIdx++;
-					const currentContent = translatedParts.slice(0, endIdx).join("");
-					mkdirSync(dirname(outputFilePath), { recursive: true });
-					writeFileSync(outputFilePath, currentContent);
-				}
-			}
-			chunkLogger([`${colorizeNumber(tokens)} tokens used `, `${ANSIColors.GREY_DARK}in ${colorizeNumber(chunkDuration)}ms${ANSIColors.RESET}`].join(""));
-		}));
-		await Promise.all(tasks);
-		const fullContent = translatedParts.join("");
-		if (flushStrategy === "end" || flushStrategy === "incremental") {
-			mkdirSync(dirname(outputFilePath), { recursive: true });
-			writeFileSync(outputFilePath, fullContent);
-		}
-		const totalDuration = ((performance.now() - fileStartTime) / 1e3).toFixed(2);
-		const relativePath = relative(configuration.content.baseDir, outputFilePath);
-		appLogger(`${colorize("✔", ANSIColors.GREEN)} File ${formatPath(relativePath)} completed in ${colorizeNumber(totalDuration)}s.`);
-		return fullContent;
-	} catch (error) {
-		errorState.count++;
-		const errorMessage = error?.message ?? JSON.stringify(error);
-		appLogger(`${colorize("✖", ANSIColors.RED)} Error: ${errorMessage}`);
-		if (errorState.count >= errorState.maxErrors) errorState.shouldStop = true;
-		return null;
-	}
-};
-
-//#endregion
-export { translateFile };
+import{readAsset as e}from"../_virtual/_utils_asset.mjs";import{sanitizeChunk as t,validateTranslation as n}from"./validation.mjs";import{chunkInference as r}from"../utils/chunkInference.mjs";import{fixChunkStartEndChars as i}from"../utils/fixChunkStartEndChars.mjs";import{chunkText as a}from"../utils/calculateChunks.mjs";import{formatLocale as o,formatPath as s}from"@intlayer/chokidar/utils";import{dirname as c,relative as l}from"node:path";import{ANSIColors as u,colon as d,colorize as f,colorizeNumber as p,getAppLogger as m}from"@intlayer/config/logger";import{retryManager as h}from"@intlayer/config/utils";import{mkdirSync as g,writeFileSync as _}from"node:fs";import{readFile as v}from"node:fs/promises";import{performance as y}from"node:perf_hooks";const b=async({baseFilePath:b,outputFilePath:x,locale:S,baseLocale:C,configuration:w,errorState:T,aiOptions:E,customInstructions:D,aiClient:O,aiConfig:k,flushStrategy:A=`incremental`,onChunkReceive:j,limit:M})=>{if(T.shouldStop)return null;let N=m(w,{config:{prefix:``}}),P=y.now();try{let F=a(await v(b,`utf-8`)),I=F.length,L=`${d(`${u.GREY_DARK}[${s(b)}${u.GREY_DARK}] `,{colSize:40})}${u.RESET}`,R=`${d(`${u.GREY_DARK}[${s(b)}${u.GREY_DARK}][${o(S)}${u.GREY_DARK}] `,{colSize:40})}${u.RESET}`;N(`${L}Split into ${p(I)} chunks. Queuing...`);let z=e(`./prompts/TRANSLATE_PROMPT.md`,`utf-8`).replaceAll(`{{localeName}}`,`${o(S,!1)}`).replaceAll(`{{baseLocaleName}}`,`${o(C,!1)}`).replace(`{{applicationContext}}`,E?.applicationContext??`-`).replace(`{{customInstructions}}`,D??`-`),B=Array(I).fill(``),V=M??(e=>e()),H=F.map((e,a)=>V(async()=>{if(T.shouldStop)return null;let o=m(w,{config:{prefix:`${R}  ${u.GREY_DARK}[${a+1}/${I}] ${u.RESET}`}}),s=y.now(),l=a===0,d=e.content,f=()=>">>> CONTEXT: PREVIOUS SOURCE CONTENT <<<\n```\n"+(F[a-1]?.content??``)+"\n```\n>>> END PREVIOUS CONTEXT <<<",v=()=>">>> CONTEXT: NEXT CONTENT <<<\n```\n"+(F[a+1]?.content??``)+"\n```\n>>> END NEXT CONTEXT <<<";o(`Process started`);let{content:b,tokens:S}=await h(async()=>{let e=await r([{role:`system`,content:z},...F[a+1]?[{role:`system`,content:v()}]:[],...l?[]:[{role:`system`,content:f()}],{role:`system`,content:[`You are translating TARGET CHUNK (${a+1}/${I}).`,`Translate ONLY the target chunk. Preserve frontmatter/code exactly.`].join(`
+`)},{role:`user`,content:`>>> TARGET CHUNK START <<<\n${d}\n>>> TARGET CHUNK END <<<`}],E,w,O,k),s=t(e?.fileContent,d);if(s=i(s,d),!n(d,s,o))throw Error(`Validation failed for chunk ${a+1}/${I}`);return{content:s,tokens:e.tokenUsed}})(),C=(y.now()-s).toFixed(0);if(B[a]=b,j&&j(b,a,I),A===`incremental`&&B.slice(0,a+1).every(e=>e&&e!==``)){let e=0;for(;e<I&&B[e]&&B[e]!==``;)e++;let t=B.slice(0,e).join(``);g(c(x),{recursive:!0}),_(x,t)}o([`${p(S)} tokens used `,`${u.GREY_DARK}in ${p(C)}ms${u.RESET}`].join(``))}));await Promise.all(H);let U=B.join(``);(A===`end`||A===`incremental`)&&(g(c(x),{recursive:!0}),_(x,U));let W=((y.now()-P)/1e3).toFixed(2),G=l(w.content.baseDir,x);return N(`${f(`✔`,u.GREEN)} File ${s(G)} completed in ${p(W)}s.`),U}catch(e){T.count++;let t=e?.message??JSON.stringify(e);return N(`${f(`✖`,u.RED)} Error: ${t}`),T.count>=T.maxErrors&&(T.shouldStop=!0),null}};export{b as translateFile};
 //# sourceMappingURL=translateFile.mjs.map

package/dist/esm/translateDoc/translateFile.mjs.map

@@ -1 +1 @@
-{"version":3,"file":"translateFile.mjs","names":[],"sources":["../../../src/translateDoc/translateFile.ts"],"sourcesContent":["import { mkdirSync, writeFileSync } from 'node:fs';\nimport { readFile } from 'node:fs/promises';\nimport { dirname, relative } from 'node:path';\nimport { performance } from 'node:perf_hooks';\nimport { readAsset } from 'utils:asset';\nimport { formatLocale, formatPath } from '@intlayer/chokidar';\nimport {\n  ANSIColors,\n  colon,\n  colorize,\n  colorizeNumber,\n  getAppLogger,\n  retryManager,\n} from '@intlayer/config';\nimport { chunkText } from '../utils/calculateChunks';\nimport { chunkInference } from '../utils/chunkInference';\nimport { fixChunkStartEndChars } from '../utils/fixChunkStartEndChars';\nimport type { TranslateFileOptions } from './types';\nimport { sanitizeChunk, validateTranslation } from './validation';\n\nexport const translateFile = async ({\n  baseFilePath,\n  outputFilePath,\n  locale,\n  baseLocale,\n  configuration,\n  errorState,\n  aiOptions,\n  customInstructions,\n  aiClient,\n  aiConfig,\n  flushStrategy = 'incremental',\n  onChunkReceive,\n  limit, // The Global Limiter\n}: TranslateFileOptions): Promise<string | null> => {\n  if (errorState.shouldStop) return null;\n\n  const appLogger = getAppLogger(configuration, { config: { prefix: '' } });\n  const fileStartTime = performance.now();\n\n  try {\n    const fileContent = await readFile(baseFilePath, 'utf-8');\n    const chunks = chunkText(fileContent);\n    const totalChunks = chunks.length;\n\n    const filePrefixText = `${ANSIColors.GREY_DARK}[${formatPath(baseFilePath)}${ANSIColors.GREY_DARK}] `;\n    const filePrefix = `${colon(filePrefixText, { colSize: 40 })}${ANSIColors.RESET}`;\n    const prefixText = `${ANSIColors.GREY_DARK}[${formatPath(baseFilePath)}${ANSIColors.GREY_DARK}][${formatLocale(locale)}${ANSIColors.GREY_DARK}] `;\n    const prefix = `${colon(prefixText, { colSize: 40 })}${ANSIColors.RESET}`;\n\n    appLogger(\n      `${filePrefix}Split into ${colorizeNumber(totalChunks)} chunks. Queuing...`\n    );\n\n    const basePrompt = readAsset('./prompts/TRANSLATE_PROMPT.md', 'utf-8')\n      .replaceAll('{{localeName}}', `${formatLocale(locale, false)}`)\n      .replaceAll('{{baseLocaleName}}', `${formatLocale(baseLocale, false)}`)\n      .replace('{{applicationContext}}', aiOptions?.applicationContext ?? '-')\n      .replace('{{customInstructions}}', customInstructions ?? '-');\n\n    const translatedParts: string[] = new Array(totalChunks).fill('');\n\n    // Fallback if no limiter is provided (runs immediately)\n    const runTask = limit ?? ((fn) => fn());\n\n    // MAP CHUNKS TO GLOBAL TASKS\n    // This pushes ALL chunks for this file into the Global Queue immediately.\n    // They will execute whenever the global concurrency slots open up.\n    const tasks = chunks.map((chunk, i) =>\n      runTask(async () => {\n        if (errorState.shouldStop) return null;\n\n        const chunkLogger = getAppLogger(configuration, {\n          config: {\n            prefix: `${prefix}  ${ANSIColors.GREY_DARK}[${i + 1}/${totalChunks}] ${ANSIColors.RESET}`,\n          },\n        });\n\n        const chunkStartTime = performance.now();\n        const isFirstChunk = i === 0;\n        const fileToTranslateCurrentChunk = chunk.content;\n\n        // Context Preparation\n        const getPrevChunkPrompt = () =>\n          `>>> CONTEXT: PREVIOUS SOURCE CONTENT <<<\\n\\`\\`\\`\\n` +\n          (chunks[i - 1]?.content ?? '') +\n          `\\n\\`\\`\\`\\n>>> END PREVIOUS CONTEXT <<<`;\n\n        const getBaseChunkContextPrompt = () =>\n          `>>> CONTEXT: NEXT CONTENT <<<\\n\\`\\`\\`\\n` +\n          (chunks[i + 1]?.content ?? '') +\n          `\\n\\`\\`\\`\\n>>> END NEXT CONTEXT <<<`;\n\n        chunkLogger('Process started');\n\n        const chunkTranslation = retryManager(async () => {\n          const result = await chunkInference(\n            [\n              { role: 'system', content: basePrompt },\n              ...(chunks[i + 1]\n                ? [\n                    {\n                      role: 'system',\n                      content: getBaseChunkContextPrompt(),\n                    } as const,\n                  ]\n                : []),\n              ...(isFirstChunk\n                ? []\n                : [{ role: 'system', content: getPrevChunkPrompt() } as const]),\n              {\n                role: 'system',\n                content: [\n                  `You are translating TARGET CHUNK (${i + 1}/${totalChunks}).`,\n                  `Translate ONLY the target chunk. Preserve frontmatter/code exactly.`,\n                ].join('\\n'),\n              },\n              {\n                role: 'user',\n                content: `>>> TARGET CHUNK START <<<\\n${fileToTranslateCurrentChunk}\\n>>> TARGET CHUNK END <<<`,\n              },\n            ],\n            aiOptions,\n            configuration,\n            aiClient,\n            aiConfig\n          );\n\n          let processedChunk = sanitizeChunk(\n            result?.fileContent,\n            fileToTranslateCurrentChunk\n          );\n          processedChunk = fixChunkStartEndChars(\n            processedChunk,\n            fileToTranslateCurrentChunk\n          );\n\n          const isValid = validateTranslation(\n            fileToTranslateCurrentChunk,\n            processedChunk,\n            chunkLogger\n          );\n\n          if (!isValid) {\n            // Throwing an error here signals retryManager to try again\n            throw new Error(\n              `Validation failed for chunk ${i + 1}/${totalChunks}`\n            );\n          }\n\n          return { content: processedChunk, tokens: result.tokenUsed };\n        });\n\n        const { content: translatedChunk, tokens } = await chunkTranslation();\n        const chunkEndTime = performance.now();\n        const chunkDuration = (chunkEndTime - chunkStartTime).toFixed(0);\n\n        // Store Result\n        translatedParts[i] = translatedChunk;\n\n        if (onChunkReceive) {\n          onChunkReceive(translatedChunk, i, totalChunks);\n        }\n\n        // Incremental Flush Strategy\n        if (flushStrategy === 'incremental') {\n          const isContiguous = translatedParts\n            .slice(0, i + 1)\n            .every((p) => p && p !== '');\n\n          if (isContiguous) {\n            let endIdx = 0;\n            while (\n              endIdx < totalChunks &&\n              translatedParts[endIdx] &&\n              translatedParts[endIdx] !== ''\n            ) {\n              endIdx++;\n            }\n            const currentContent = translatedParts.slice(0, endIdx).join('');\n            // Write asynchronously/sync is fine here as node handles file locks reasonably well for single process\n            mkdirSync(dirname(outputFilePath), { recursive: true });\n            writeFileSync(outputFilePath, currentContent);\n          }\n        }\n\n        chunkLogger(\n          [\n            `${colorizeNumber(tokens)} tokens used `,\n            `${ANSIColors.GREY_DARK}in ${colorizeNumber(chunkDuration)}ms${ANSIColors.RESET}`,\n          ].join('')\n        );\n      })\n    );\n\n    // Wait for all chunks for this specific file/locale to finish\n    await Promise.all(tasks);\n\n    // Final Flush\n    const fullContent = translatedParts.join('');\n    if (flushStrategy === 'end' || flushStrategy === 'incremental') {\n      mkdirSync(dirname(outputFilePath), { recursive: true });\n      writeFileSync(outputFilePath, fullContent);\n    }\n\n    const fileEndTime = performance.now();\n    const totalDuration = ((fileEndTime - fileStartTime) / 1000).toFixed(2);\n    const relativePath = relative(\n      configuration.content.baseDir,\n      outputFilePath\n    );\n\n    appLogger(\n      `${colorize('✔', ANSIColors.GREEN)} File ${formatPath(relativePath)} completed in ${colorizeNumber(totalDuration)}s.`\n    );\n\n    return fullContent;\n  } catch (error: any) {\n    errorState.count++;\n    const errorMessage = error?.message ?? JSON.stringify(error);\n    appLogger(`${colorize('✖', ANSIColors.RED)} Error: ${errorMessage}`);\n    if (errorState.count >= errorState.maxErrors) errorState.shouldStop = true;\n    return null;\n  }\n};\n"],"mappings":";;;;;;;;;;;;;AAoBA,MAAa,gBAAgB,OAAO,EAClC,cACA,gBACA,QACA,YACA,eACA,YACA,WACA,oBACA,UACA,UACA,gBAAgB,eAChB,gBACA,YACkD;AAClD,KAAI,WAAW,WAAY,QAAO;CAElC,MAAM,YAAY,aAAa,eAAe,EAAE,QAAQ,EAAE,QAAQ,IAAI,EAAE,CAAC;CACzE,MAAM,gBAAgB,YAAY,KAAK;AAEvC,KAAI;EAEF,MAAM,SAAS,UADK,MAAM,SAAS,cAAc,QAAQ,CACpB;EACrC,MAAM,cAAc,OAAO;EAG3B,MAAM,aAAa,GAAG,MADC,GAAG,WAAW,UAAU,GAAG,WAAW,aAAa,GAAG,WAAW,UAAU,KACtD,EAAE,SAAS,IAAI,CAAC,GAAG,WAAW;EAE1E,MAAM,SAAS,GAAG,MADC,GAAG,WAAW,UAAU,GAAG,WAAW,aAAa,GAAG,WAAW,UAAU,IAAI,aAAa,OAAO,GAAG,WAAW,UAAU,KAC1G,EAAE,SAAS,IAAI,CAAC,GAAG,WAAW;AAElE,YACE,GAAG,WAAW,aAAa,eAAe,YAAY,CAAC,qBACxD;EAED,MAAM,aAAa,UAAU,iCAAiC,QAAQ,CACnE,WAAW,kBAAkB,GAAG,aAAa,QAAQ,MAAM,GAAG,CAC9D,WAAW,sBAAsB,GAAG,aAAa,YAAY,MAAM,GAAG,CACtE,QAAQ,0BAA0B,WAAW,sBAAsB,IAAI,CACvE,QAAQ,0BAA0B,sBAAsB,IAAI;EAE/D,MAAM,kBAA4B,IAAI,MAAM,YAAY,CAAC,KAAK,GAAG;EAGjE,MAAM,UAAU,WAAW,OAAO,IAAI;EAKtC,MAAM,QAAQ,OAAO,KAAK,OAAO,MAC/B,QAAQ,YAAY;AAClB,OAAI,WAAW,WAAY,QAAO;GAElC,MAAM,cAAc,aAAa,eAAe,EAC9C,QAAQ,EACN,QAAQ,GAAG,OAAO,IAAI,WAAW,UAAU,GAAG,IAAI,EAAE,GAAG,YAAY,IAAI,WAAW,SACnF,EACF,CAAC;GAEF,MAAM,iBAAiB,YAAY,KAAK;GACxC,MAAM,eAAe,MAAM;GAC3B,MAAM,8BAA8B,MAAM;GAG1C,MAAM,2BACJ,wDACC,OAAO,IAAI,IAAI,WAAW,MAC3B;GAEF,MAAM,kCACJ,6CACC,OAAO,IAAI,IAAI,WAAW,MAC3B;AAEF,eAAY,kBAAkB;GA4D9B,MAAM,EAAE,SAAS,iBAAiB,WAAW,MA1DpB,aAAa,YAAY;IAChD,MAAM,SAAS,MAAM,eACnB;KACE;MAAE,MAAM;MAAU,SAAS;MAAY;KACvC,GAAI,OAAO,IAAI,KACX,CACE;MACE,MAAM;MACN,SAAS,2BAA2B;MACrC,CACF,GACD,EAAE;KACN,GAAI,eACA,EAAE,GACF,CAAC;MAAE,MAAM;MAAU,SAAS,oBAAoB;MAAE,CAAU;KAChE;MACE,MAAM;MACN,SAAS,CACP,qCAAqC,IAAI,EAAE,GAAG,YAAY,KAC1D,sEACD,CAAC,KAAK,KAAK;MACb;KACD;MACE,MAAM;MACN,SAAS,+BAA+B,4BAA4B;MACrE;KACF,EACD,WACA,eACA,UACA,SACD;IAED,IAAI,iBAAiB,cACnB,QAAQ,aACR,4BACD;AACD,qBAAiB,sBACf,gBACA,4BACD;AAQD,QAAI,CANY,oBACd,6BACA,gBACA,YACD,CAIC,OAAM,IAAI,MACR,+BAA+B,IAAI,EAAE,GAAG,cACzC;AAGH,WAAO;KAAE,SAAS;KAAgB,QAAQ,OAAO;KAAW;KAC5D,EAEmE;GAErE,MAAM,iBADe,YAAY,KAAK,GACA,gBAAgB,QAAQ,EAAE;AAGhE,mBAAgB,KAAK;AAErB,OAAI,eACF,gBAAe,iBAAiB,GAAG,YAAY;AAIjD,OAAI,kBAAkB,eAKpB;QAJqB,gBAClB,MAAM,GAAG,IAAI,EAAE,CACf,OAAO,MAAM,KAAK,MAAM,GAAG,EAEZ;KAChB,IAAI,SAAS;AACb,YACE,SAAS,eACT,gBAAgB,WAChB,gBAAgB,YAAY,GAE5B;KAEF,MAAM,iBAAiB,gBAAgB,MAAM,GAAG,OAAO,CAAC,KAAK,GAAG;AAEhE,eAAU,QAAQ,eAAe,EAAE,EAAE,WAAW,MAAM,CAAC;AACvD,mBAAc,gBAAgB,eAAe;;;AAIjD,eACE,CACE,GAAG,eAAe,OAAO,CAAC,gBAC1B,GAAG,WAAW,UAAU,KAAK,eAAe,cAAc,CAAC,IAAI,WAAW,QAC3E,CAAC,KAAK,GAAG,CACX;IACD,CACH;AAGD,QAAM,QAAQ,IAAI,MAAM;EAGxB,MAAM,cAAc,gBAAgB,KAAK,GAAG;AAC5C,MAAI,kBAAkB,SAAS,kBAAkB,eAAe;AAC9D,aAAU,QAAQ,eAAe,EAAE,EAAE,WAAW,MAAM,CAAC;AACvD,iBAAc,gBAAgB,YAAY;;EAI5C,MAAM,kBADc,YAAY,KAAK,GACC,iBAAiB,KAAM,QAAQ,EAAE;EACvE,MAAM,eAAe,SACnB,cAAc,QAAQ,SACtB,eACD;AAED,YACE,GAAG,SAAS,KAAK,WAAW,MAAM,CAAC,QAAQ,WAAW,aAAa,CAAC,gBAAgB,eAAe,cAAc,CAAC,IACnH;AAED,SAAO;UACA,OAAY;AACnB,aAAW;EACX,MAAM,eAAe,OAAO,WAAW,KAAK,UAAU,MAAM;AAC5D,YAAU,GAAG,SAAS,KAAK,WAAW,IAAI,CAAC,UAAU,eAAe;AACpE,MAAI,WAAW,SAAS,WAAW,UAAW,YAAW,aAAa;AACtE,SAAO"}
+{"version":3,"file":"translateFile.mjs","names":[],"sources":["../../../src/translateDoc/translateFile.ts"],"sourcesContent":["import { mkdirSync, writeFileSync } from 'node:fs';\nimport { readFile } from 'node:fs/promises';\nimport { dirname, relative } from 'node:path';\nimport { performance } from 'node:perf_hooks';\nimport { readAsset } from 'utils:asset';\nimport { formatLocale, formatPath } from '@intlayer/chokidar/utils';\nimport {\n  ANSIColors,\n  colon,\n  colorize,\n  colorizeNumber,\n  getAppLogger,\n} from '@intlayer/config/logger';\nimport { retryManager } from '@intlayer/config/utils';\nimport { chunkText } from '../utils/calculateChunks';\nimport { chunkInference } from '../utils/chunkInference';\nimport { fixChunkStartEndChars } from '../utils/fixChunkStartEndChars';\nimport type { TranslateFileOptions } from './types';\nimport { sanitizeChunk, validateTranslation } from './validation';\n\nexport const translateFile = async ({\n  baseFilePath,\n  outputFilePath,\n  locale,\n  baseLocale,\n  configuration,\n  errorState,\n  aiOptions,\n  customInstructions,\n  aiClient,\n  aiConfig,\n  flushStrategy = 'incremental',\n  onChunkReceive,\n  limit, // The Global Limiter\n}: TranslateFileOptions): Promise<string | null> => {\n  if (errorState.shouldStop) return null;\n\n  const appLogger = getAppLogger(configuration, { config: { prefix: '' } });\n  const fileStartTime = performance.now();\n\n  try {\n    const fileContent = await readFile(baseFilePath, 'utf-8');\n    const chunks = chunkText(fileContent);\n    const totalChunks = chunks.length;\n\n    const filePrefixText = `${ANSIColors.GREY_DARK}[${formatPath(baseFilePath)}${ANSIColors.GREY_DARK}] `;\n    const filePrefix = `${colon(filePrefixText, { colSize: 40 })}${ANSIColors.RESET}`;\n    const prefixText = `${ANSIColors.GREY_DARK}[${formatPath(baseFilePath)}${ANSIColors.GREY_DARK}][${formatLocale(locale)}${ANSIColors.GREY_DARK}] `;\n    const prefix = `${colon(prefixText, { colSize: 40 })}${ANSIColors.RESET}`;\n\n    appLogger(\n      `${filePrefix}Split into ${colorizeNumber(totalChunks)} chunks. Queuing...`\n    );\n\n    const basePrompt = readAsset('./prompts/TRANSLATE_PROMPT.md', 'utf-8')\n      .replaceAll('{{localeName}}', `${formatLocale(locale, false)}`)\n      .replaceAll('{{baseLocaleName}}', `${formatLocale(baseLocale, false)}`)\n      .replace('{{applicationContext}}', aiOptions?.applicationContext ?? '-')\n      .replace('{{customInstructions}}', customInstructions ?? '-');\n\n    const translatedParts: string[] = new Array(totalChunks).fill('');\n\n    // Fallback if no limiter is provided (runs immediately)\n    const runTask = limit ?? ((fn) => fn());\n\n    // MAP CHUNKS TO GLOBAL TASKS\n    // This pushes ALL chunks for this file into the Global Queue immediately.\n    // They will execute whenever the global concurrency slots open up.\n    const tasks = chunks.map((chunk, i) =>\n      runTask(async () => {\n        if (errorState.shouldStop) return null;\n\n        const chunkLogger = getAppLogger(configuration, {\n          config: {\n            prefix: `${prefix}  ${ANSIColors.GREY_DARK}[${i + 1}/${totalChunks}] ${ANSIColors.RESET}`,\n          },\n        });\n\n        const chunkStartTime = performance.now();\n        const isFirstChunk = i === 0;\n        const fileToTranslateCurrentChunk = chunk.content;\n\n        // Context Preparation\n        const getPrevChunkPrompt = () =>\n          `>>> CONTEXT: PREVIOUS SOURCE CONTENT <<<\\n\\`\\`\\`\\n` +\n          (chunks[i - 1]?.content ?? '') +\n          `\\n\\`\\`\\`\\n>>> END PREVIOUS CONTEXT <<<`;\n\n        const getBaseChunkContextPrompt = () =>\n          `>>> CONTEXT: NEXT CONTENT <<<\\n\\`\\`\\`\\n` +\n          (chunks[i + 1]?.content ?? '') +\n          `\\n\\`\\`\\`\\n>>> END NEXT CONTEXT <<<`;\n\n        chunkLogger('Process started');\n\n        const chunkTranslation = retryManager(async () => {\n          const result = await chunkInference(\n            [\n              { role: 'system', content: basePrompt },\n              ...(chunks[i + 1]\n                ? [\n                    {\n                      role: 'system',\n                      content: getBaseChunkContextPrompt(),\n                    } as const,\n                  ]\n                : []),\n              ...(isFirstChunk\n                ? []\n                : [{ role: 'system', content: getPrevChunkPrompt() } as const]),\n              {\n                role: 'system',\n                content: [\n                  `You are translating TARGET CHUNK (${i + 1}/${totalChunks}).`,\n                  `Translate ONLY the target chunk. Preserve frontmatter/code exactly.`,\n                ].join('\\n'),\n              },\n              {\n                role: 'user',\n                content: `>>> TARGET CHUNK START <<<\\n${fileToTranslateCurrentChunk}\\n>>> TARGET CHUNK END <<<`,\n              },\n            ],\n            aiOptions,\n            configuration,\n            aiClient,\n            aiConfig\n          );\n\n          let processedChunk = sanitizeChunk(\n            result?.fileContent,\n            fileToTranslateCurrentChunk\n          );\n          processedChunk = fixChunkStartEndChars(\n            processedChunk,\n            fileToTranslateCurrentChunk\n          );\n\n          const isValid = validateTranslation(\n            fileToTranslateCurrentChunk,\n            processedChunk,\n            chunkLogger\n          );\n\n          if (!isValid) {\n            // Throwing an error here signals retryManager to try again\n            throw new Error(\n              `Validation failed for chunk ${i + 1}/${totalChunks}`\n            );\n          }\n\n          return { content: processedChunk, tokens: result.tokenUsed };\n        });\n\n        const { content: translatedChunk, tokens } = await chunkTranslation();\n        const chunkEndTime = performance.now();\n        const chunkDuration = (chunkEndTime - chunkStartTime).toFixed(0);\n\n        // Store Result\n        translatedParts[i] = translatedChunk;\n\n        if (onChunkReceive) {\n          onChunkReceive(translatedChunk, i, totalChunks);\n        }\n\n        // Incremental Flush Strategy\n        if (flushStrategy === 'incremental') {\n          const isContiguous = translatedParts\n            .slice(0, i + 1)\n            .every((p) => p && p !== '');\n\n          if (isContiguous) {\n            let endIdx = 0;\n            while (\n              endIdx < totalChunks &&\n              translatedParts[endIdx] &&\n              translatedParts[endIdx] !== ''\n            ) {\n              endIdx++;\n            }\n            const currentContent = translatedParts.slice(0, endIdx).join('');\n            // Write asynchronously/sync is fine here as node handles file locks reasonably well for single process\n            mkdirSync(dirname(outputFilePath), { recursive: true });\n            writeFileSync(outputFilePath, currentContent);\n          }\n        }\n\n        chunkLogger(\n          [\n            `${colorizeNumber(tokens)} tokens used `,\n            `${ANSIColors.GREY_DARK}in ${colorizeNumber(chunkDuration)}ms${ANSIColors.RESET}`,\n          ].join('')\n        );\n      })\n    );\n\n    // Wait for all chunks for this specific file/locale to finish\n    await Promise.all(tasks);\n\n    // Final Flush\n    const fullContent = translatedParts.join('');\n    if (flushStrategy === 'end' || flushStrategy === 'incremental') {\n      mkdirSync(dirname(outputFilePath), { recursive: true });\n      writeFileSync(outputFilePath, fullContent);\n    }\n\n    const fileEndTime = performance.now();\n    const totalDuration = ((fileEndTime - fileStartTime) / 1000).toFixed(2);\n    const relativePath = relative(\n      configuration.content.baseDir,\n      outputFilePath\n    );\n\n    appLogger(\n      `${colorize('✔', ANSIColors.GREEN)} File ${formatPath(relativePath)} completed in ${colorizeNumber(totalDuration)}s.`\n    );\n\n    return fullContent;\n  } catch (error: any) {\n    errorState.count++;\n    const errorMessage = error?.message ?? JSON.stringify(error);\n    appLogger(`${colorize('✖', ANSIColors.RED)} Error: ${errorMessage}`);\n    if (errorState.count >= errorState.maxErrors) errorState.shouldStop = true;\n    return null;\n  }\n};\n"],"mappings":"yvBAoBA,MAAa,EAAgB,MAAO,CAClC,eACA,iBACA,SACA,aACA,gBACA,aACA,YACA,qBACA,WACA,WACA,gBAAgB,cAChB,iBACA,WACkD,CAClD,GAAI,EAAW,WAAY,OAAO,KAElC,IAAM,EAAY,EAAa,EAAe,CAAE,OAAQ,CAAE,OAAQ,GAAI,CAAE,CAAC,CACnE,EAAgB,EAAY,KAAK,CAEvC,GAAI,CAEF,IAAM,EAAS,EADK,MAAM,EAAS,EAAc,QAAQ,CACpB,CAC/B,EAAc,EAAO,OAGrB,EAAa,GAAG,EADC,GAAG,EAAW,UAAU,GAAG,EAAW,EAAa,GAAG,EAAW,UAAU,IACtD,CAAE,QAAS,GAAI,CAAC,GAAG,EAAW,QAEpE,EAAS,GAAG,EADC,GAAG,EAAW,UAAU,GAAG,EAAW,EAAa,GAAG,EAAW,UAAU,IAAI,EAAa,EAAO,GAAG,EAAW,UAAU,IAC1G,CAAE,QAAS,GAAI,CAAC,GAAG,EAAW,QAElE,EACE,GAAG,EAAW,aAAa,EAAe,EAAY,CAAC,qBACxD,CAED,IAAM,EAAa,EAAU,gCAAiC,QAAQ,CACnE,WAAW,iBAAkB,GAAG,EAAa,EAAQ,GAAM,GAAG,CAC9D,WAAW,qBAAsB,GAAG,EAAa,EAAY,GAAM,GAAG,CACtE,QAAQ,yBAA0B,GAAW,oBAAsB,IAAI,CACvE,QAAQ,yBAA0B,GAAsB,IAAI,CAEzD,EAAgC,MAAM,EAAY,CAAC,KAAK,GAAG,CAG3D,EAAU,IAAW,GAAO,GAAI,EAKhC,EAAQ,EAAO,KAAK,EAAO,IAC/B,EAAQ,SAAY,CAClB,GAAI,EAAW,WAAY,OAAO,KAElC,IAAM,EAAc,EAAa,EAAe,CAC9C,OAAQ,CACN,OAAQ,GAAG,EAAO,IAAI,EAAW,UAAU,GAAG,EAAI,EAAE,GAAG,EAAY,IAAI,EAAW,QACnF,CACF,CAAC,CAEI,EAAiB,EAAY,KAAK,CAClC,EAAe,IAAM,EACrB,EAA8B,EAAM,QAGpC,MACJ,mDACC,EAAO,EAAI,IAAI,SAAW,IAC3B,sCAEI,MACJ,wCACC,EAAO,EAAI,IAAI,SAAW,IAC3B,kCAEF,EAAY,kBAAkB,CA4D9B,GAAM,CAAE,QAAS,EAAiB,UAAW,MA1DpB,EAAa,SAAY,CAChD,IAAM,EAAS,MAAM,EACnB,CACE,CAAE,KAAM,SAAU,QAAS,EAAY,CACvC,GAAI,EAAO,EAAI,GACX,CACE,CACE,KAAM,SACN,QAAS,GAA2B,CACrC,CACF,CACD,EAAE,CACN,GAAI,EACA,EAAE,CACF,CAAC,CAAE,KAAM,SAAU,QAAS,GAAoB,CAAE,CAAU,CAChE,CACE,KAAM,SACN,QAAS,CACP,qCAAqC,EAAI,EAAE,GAAG,EAAY,IAC1D,sEACD,CAAC,KAAK;EAAK,CACb,CACD,CACE,KAAM,OACN,QAAS,+BAA+B,EAA4B,4BACrE,CACF,CACD,EACA,EACA,EACA,EACD,CAEG,EAAiB,EACnB,GAAQ,YACR,EACD,CAYD,GAXA,EAAiB,EACf,EACA,EACD,CAQG,CANY,EACd,EACA,EACA,EACD,CAIC,MAAU,MACR,+BAA+B,EAAI,EAAE,GAAG,IACzC,CAGH,MAAO,CAAE,QAAS,EAAgB,OAAQ,EAAO,UAAW,EAC5D,EAEmE,CAE/D,GADe,EAAY,KAAK,CACA,GAAgB,QAAQ,EAAE,CAUhE,GAPA,EAAgB,GAAK,EAEjB,GACF,EAAe,EAAiB,EAAG,EAAY,CAI7C,IAAkB,eACC,EAClB,MAAM,EAAG,EAAI,EAAE,CACf,MAAO,GAAM,GAAK,IAAM,GAAG,CAEZ,CAChB,IAAI,EAAS,EACb,KACE,EAAS,GACT,EAAgB,IAChB,EAAgB,KAAY,IAE5B,IAEF,IAAM,EAAiB,EAAgB,MAAM,EAAG,EAAO,CAAC,KAAK,GAAG,CAEhE,EAAU,EAAQ,EAAe,CAAE,CAAE,UAAW,GAAM,CAAC,CACvD,EAAc,EAAgB,EAAe,CAIjD,EACE,CACE,GAAG,EAAe,EAAO,CAAC,eAC1B,GAAG,EAAW,UAAU,KAAK,EAAe,EAAc,CAAC,IAAI,EAAW,QAC3E,CAAC,KAAK,GAAG,CACX,EACD,CACH,CAGD,MAAM,QAAQ,IAAI,EAAM,CAGxB,IAAM,EAAc,EAAgB,KAAK,GAAG,EACxC,IAAkB,OAAS,IAAkB,iBAC/C,EAAU,EAAQ,EAAe,CAAE,CAAE,UAAW,GAAM,CAAC,CACvD,EAAc,EAAgB,EAAY,EAI5C,IAAM,IADc,EAAY,KAAK,CACC,GAAiB,KAAM,QAAQ,EAAE,CACjE,EAAe,EACnB,EAAc,QAAQ,QACtB,EACD,CAMD,OAJA,EACE,GAAG,EAAS,IAAK,EAAW,MAAM,CAAC,QAAQ,EAAW,EAAa,CAAC,gBAAgB,EAAe,EAAc,CAAC,IACnH,CAEM,QACA,EAAY,CACnB,EAAW,QACX,IAAM,EAAe,GAAO,SAAW,KAAK,UAAU,EAAM,CAG5D,OAFA,EAAU,GAAG,EAAS,IAAK,EAAW,IAAI,CAAC,UAAU,IAAe,CAChE,EAAW,OAAS,EAAW,YAAW,EAAW,WAAa,IAC/D"}

package/dist/esm/translateDoc/validation.mjs

@@ -1,47 +1,6 @@
-//#region src/translateDoc/validation.ts
-/**
-* Validates that the translated content matches the structure of the original.
-* Throws an error if a mismatch is found, triggering a retry.
-*/
-const validateTranslation = (original, translated, logger) => {
-	const errors = [];
-	if (original.trimStart().startsWith("---")) {
-		if (!translated.trimStart().startsWith("---")) errors.push("YAML Frontmatter missing: Input starts with \"---\", output does not.");
-		const originalDashes = (original.match(/^---$/gm) || []).length;
-		const translatedDashes = (translated.match(/^---$/gm) || []).length;
-		if (originalDashes >= 2 && translatedDashes < 2) errors.push("YAML Frontmatter unclosed: Input has closing \"---\", output is missing it.");
-	}
-	const fenceRegex = /^\s*```/gm;
-	const originalFences = (original.match(fenceRegex) || []).length;
-	const translatedFences = (translated.match(fenceRegex) || []).length;
-	if (originalFences !== translatedFences) errors.push(`Code fence mismatch: Input has ${originalFences}, output has ${translatedFences}`);
-	const ratio = translated.length / (original.length || 1);
-	const isTooLong = ratio > 2.5;
-	const isSignificantLength = original.length > 50;
-	if (isTooLong && isSignificantLength) errors.push(`Length deviation: Output is ${translated.length} chars vs Input ${original.length} (${ratio.toFixed(1)}x). Likely included context.`);
-	const originalLines = original.split("\n").length;
-	const translatedLines = translated.split("\n").length;
-	if (originalLines > 5) {
-		if (translatedLines < originalLines * .4) errors.push(`Line count deviation: Output has ${translatedLines} lines, Input has ${originalLines}. Likely content deletion.`);
-	}
-	if (errors.length > 0) {
-		logger(`Validation Failed: ${errors.join(", ")}`);
-		return false;
-	}
-	return true;
-};
-/**
-* Clean common AI artifacts
-*/
-const sanitizeChunk = (translated, original) => {
-	let cleaned = translated;
-	const match = cleaned.match(/^```(?:markdown|md|txt)?\n([\s\S]*?)\n```$/i);
-	if (match) cleaned = match[1];
-	if (!original.startsWith("\n") && cleaned.startsWith("\n")) cleaned = cleaned.replace(/^\n+/, "");
-	if (!original.startsWith(" ") && cleaned.startsWith(" ")) cleaned = cleaned.trimStart();
-	return cleaned;
-};
-
-//#endregion
-export { sanitizeChunk, validateTranslation };
+const e=(e,t,n)=>{let r=[];if(e.trimStart().startsWith(`---`)){t.trimStart().startsWith(`---`)||r.push(`YAML Frontmatter missing: Input starts with "---", output does not.`);let n=(e.match(/^---$/gm)||[]).length,i=(t.match(/^---$/gm)||[]).length;n>=2&&i<2&&r.push(`YAML Frontmatter unclosed: Input has closing "---", output is missing it.`)}let i=/^\s*```/gm,a=(e.match(i)||[]).length,o=(t.match(i)||[]).length;a!==o&&r.push(`Code fence mismatch: Input has ${a}, output has ${o}`);let s=t.length/(e.length||1),c=s>2.5,l=e.length>50;c&&l&&r.push(`Length deviation: Output is ${t.length} chars vs Input ${e.length} (${s.toFixed(1)}x). Likely included context.`);let u=e.split(`
+`).length,d=t.split(`
+`).length;return u>5&&d<u*.4&&r.push(`Line count deviation: Output has ${d} lines, Input has ${u}. Likely content deletion.`),r.length>0?(n(`Validation Failed: ${r.join(`, `)}`),!1):!0},t=(e,t)=>{let n=e,r=n.match(/^```(?:markdown|md|txt)?\n([\s\S]*?)\n```$/i);return r&&(n=r[1]),!t.startsWith(`
+`)&&n.startsWith(`
+`)&&(n=n.replace(/^\n+/,``)),!t.startsWith(` `)&&n.startsWith(` `)&&(n=n.trimStart()),n};export{t as sanitizeChunk,e as validateTranslation};
 //# sourceMappingURL=validation.mjs.map

package/dist/esm/translateDoc/validation.mjs.map

@@ -1 +1 @@
-{"version":3,"file":"validation.mjs","names":[],"sources":["../../../src/translateDoc/validation.ts"],"sourcesContent":["import type { Logger } from '@intlayer/config';\n\n/**\n * Validates that the translated content matches the structure of the original.\n * Throws an error if a mismatch is found, triggering a retry.\n */\nexport const validateTranslation = (\n  original: string,\n  translated: string,\n  logger: Logger\n): boolean => {\n  const errors: string[] = [];\n\n  // YAML Frontmatter Integrity (CRITICAL)\n  if (original.trimStart().startsWith('---')) {\n    if (!translated.trimStart().startsWith('---')) {\n      errors.push(\n        'YAML Frontmatter missing: Input starts with \"---\", output does not.'\n      );\n    }\n    const originalDashes = (original.match(/^---$/gm) || []).length;\n    const translatedDashes = (translated.match(/^---$/gm) || []).length;\n    if (originalDashes >= 2 && translatedDashes < 2) {\n      errors.push(\n        'YAML Frontmatter unclosed: Input has closing \"---\", output is missing it.'\n      );\n    }\n  }\n\n  // Code Fence Check\n  const fenceRegex = /^\\s*```/gm;\n  const originalFences = (original.match(fenceRegex) || []).length;\n  const translatedFences = (translated.match(fenceRegex) || []).length;\n\n  if (originalFences !== translatedFences) {\n    errors.push(\n      `Code fence mismatch: Input has ${originalFences}, output has ${translatedFences}`\n    );\n  }\n\n  // Length/Duplication Check\n  const ratio = translated.length / (original.length || 1);\n  const isTooLong = ratio > 2.5;\n  const isSignificantLength = original.length > 50;\n\n  if (isTooLong && isSignificantLength) {\n    errors.push(\n      `Length deviation: Output is ${translated.length} chars vs Input ${original.length} (${ratio.toFixed(1)}x). Likely included context.`\n    );\n  }\n\n  // Line Count Heuristic\n  const originalLines = original.split('\\n').length;\n  const translatedLines = translated.split('\\n').length;\n\n  if (originalLines > 5) {\n    if (translatedLines < originalLines * 0.4) {\n      errors.push(\n        `Line count deviation: Output has ${translatedLines} lines, Input has ${originalLines}. Likely content deletion.`\n      );\n    }\n  }\n\n  if (errors.length > 0) {\n    logger(`Validation Failed: ${errors.join(', ')}`);\n    return false;\n  }\n\n  return true;\n};\n\n/**\n * Clean common AI artifacts\n */\nexport const sanitizeChunk = (translated: string, original: string): string => {\n  let cleaned = translated;\n  const wrapRegex = /^```(?:markdown|md|txt)?\\n([\\s\\S]*?)\\n```$/i;\n  const match = cleaned.match(wrapRegex);\n  if (match) cleaned = match[1];\n\n  if (!original.startsWith('\\n') && cleaned.startsWith('\\n')) {\n    cleaned = cleaned.replace(/^\\n+/, '');\n  }\n  if (!original.startsWith(' ') && cleaned.startsWith(' ')) {\n    cleaned = cleaned.trimStart();\n  }\n  return cleaned;\n};\n"],"mappings":";;;;;AAMA,MAAa,uBACX,UACA,YACA,WACY;CACZ,MAAM,SAAmB,EAAE;AAG3B,KAAI,SAAS,WAAW,CAAC,WAAW,MAAM,EAAE;AAC1C,MAAI,CAAC,WAAW,WAAW,CAAC,WAAW,MAAM,CAC3C,QAAO,KACL,wEACD;EAEH,MAAM,kBAAkB,SAAS,MAAM,UAAU,IAAI,EAAE,EAAE;EACzD,MAAM,oBAAoB,WAAW,MAAM,UAAU,IAAI,EAAE,EAAE;AAC7D,MAAI,kBAAkB,KAAK,mBAAmB,EAC5C,QAAO,KACL,8EACD;;CAKL,MAAM,aAAa;CACnB,MAAM,kBAAkB,SAAS,MAAM,WAAW,IAAI,EAAE,EAAE;CAC1D,MAAM,oBAAoB,WAAW,MAAM,WAAW,IAAI,EAAE,EAAE;AAE9D,KAAI,mBAAmB,iBACrB,QAAO,KACL,kCAAkC,eAAe,eAAe,mBACjE;CAIH,MAAM,QAAQ,WAAW,UAAU,SAAS,UAAU;CACtD,MAAM,YAAY,QAAQ;CAC1B,MAAM,sBAAsB,SAAS,SAAS;AAE9C,KAAI,aAAa,oBACf,QAAO,KACL,+BAA+B,WAAW,OAAO,kBAAkB,SAAS,OAAO,IAAI,MAAM,QAAQ,EAAE,CAAC,8BACzG;CAIH,MAAM,gBAAgB,SAAS,MAAM,KAAK,CAAC;CAC3C,MAAM,kBAAkB,WAAW,MAAM,KAAK,CAAC;AAE/C,KAAI,gBAAgB,GAClB;MAAI,kBAAkB,gBAAgB,GACpC,QAAO,KACL,oCAAoC,gBAAgB,oBAAoB,cAAc,4BACvF;;AAIL,KAAI,OAAO,SAAS,GAAG;AACrB,SAAO,sBAAsB,OAAO,KAAK,KAAK,GAAG;AACjD,SAAO;;AAGT,QAAO;;;;;AAMT,MAAa,iBAAiB,YAAoB,aAA6B;CAC7E,IAAI,UAAU;CAEd,MAAM,QAAQ,QAAQ,MADJ,8CACoB;AACtC,KAAI,MAAO,WAAU,MAAM;AAE3B,KAAI,CAAC,SAAS,WAAW,KAAK,IAAI,QAAQ,WAAW,KAAK,CACxD,WAAU,QAAQ,QAAQ,QAAQ,GAAG;AAEvC,KAAI,CAAC,SAAS,WAAW,IAAI,IAAI,QAAQ,WAAW,IAAI,CACtD,WAAU,QAAQ,WAAW;AAE/B,QAAO"}
+{"version":3,"file":"validation.mjs","names":[],"sources":["../../../src/translateDoc/validation.ts"],"sourcesContent":["import type { Logger } from '@intlayer/config/logger';\n\n/**\n * Validates that the translated content matches the structure of the original.\n * Throws an error if a mismatch is found, triggering a retry.\n */\nexport const validateTranslation = (\n  original: string,\n  translated: string,\n  logger: Logger\n): boolean => {\n  const errors: string[] = [];\n\n  // YAML Frontmatter Integrity (CRITICAL)\n  if (original.trimStart().startsWith('---')) {\n    if (!translated.trimStart().startsWith('---')) {\n      errors.push(\n        'YAML Frontmatter missing: Input starts with \"---\", output does not.'\n      );\n    }\n    const originalDashes = (original.match(/^---$/gm) || []).length;\n    const translatedDashes = (translated.match(/^---$/gm) || []).length;\n    if (originalDashes >= 2 && translatedDashes < 2) {\n      errors.push(\n        'YAML Frontmatter unclosed: Input has closing \"---\", output is missing it.'\n      );\n    }\n  }\n\n  // Code Fence Check\n  const fenceRegex = /^\\s*```/gm;\n  const originalFences = (original.match(fenceRegex) || []).length;\n  const translatedFences = (translated.match(fenceRegex) || []).length;\n\n  if (originalFences !== translatedFences) {\n    errors.push(\n      `Code fence mismatch: Input has ${originalFences}, output has ${translatedFences}`\n    );\n  }\n\n  // Length/Duplication Check\n  const ratio = translated.length / (original.length || 1);\n  const isTooLong = ratio > 2.5;\n  const isSignificantLength = original.length > 50;\n\n  if (isTooLong && isSignificantLength) {\n    errors.push(\n      `Length deviation: Output is ${translated.length} chars vs Input ${original.length} (${ratio.toFixed(1)}x). Likely included context.`\n    );\n  }\n\n  // Line Count Heuristic\n  const originalLines = original.split('\\n').length;\n  const translatedLines = translated.split('\\n').length;\n\n  if (originalLines > 5) {\n    if (translatedLines < originalLines * 0.4) {\n      errors.push(\n        `Line count deviation: Output has ${translatedLines} lines, Input has ${originalLines}. Likely content deletion.`\n      );\n    }\n  }\n\n  if (errors.length > 0) {\n    logger(`Validation Failed: ${errors.join(', ')}`);\n    return false;\n  }\n\n  return true;\n};\n\n/**\n * Clean common AI artifacts\n */\nexport const sanitizeChunk = (translated: string, original: string): string => {\n  let cleaned = translated;\n  const wrapRegex = /^```(?:markdown|md|txt)?\\n([\\s\\S]*?)\\n```$/i;\n  const match = cleaned.match(wrapRegex);\n  if (match) cleaned = match[1];\n\n  if (!original.startsWith('\\n') && cleaned.startsWith('\\n')) {\n    cleaned = cleaned.replace(/^\\n+/, '');\n  }\n  if (!original.startsWith(' ') && cleaned.startsWith(' ')) {\n    cleaned = cleaned.trimStart();\n  }\n  return cleaned;\n};\n"],"mappings":"AAMA,MAAa,GACX,EACA,EACA,IACY,CACZ,IAAM,EAAmB,EAAE,CAG3B,GAAI,EAAS,WAAW,CAAC,WAAW,MAAM,CAAE,CACrC,EAAW,WAAW,CAAC,WAAW,MAAM,EAC3C,EAAO,KACL,sEACD,CAEH,IAAM,GAAkB,EAAS,MAAM,UAAU,EAAI,EAAE,EAAE,OACnD,GAAoB,EAAW,MAAM,UAAU,EAAI,EAAE,EAAE,OACzD,GAAkB,GAAK,EAAmB,GAC5C,EAAO,KACL,4EACD,CAKL,IAAM,EAAa,YACb,GAAkB,EAAS,MAAM,EAAW,EAAI,EAAE,EAAE,OACpD,GAAoB,EAAW,MAAM,EAAW,EAAI,EAAE,EAAE,OAE1D,IAAmB,GACrB,EAAO,KACL,kCAAkC,EAAe,eAAe,IACjE,CAIH,IAAM,EAAQ,EAAW,QAAU,EAAS,QAAU,GAChD,EAAY,EAAQ,IACpB,EAAsB,EAAS,OAAS,GAE1C,GAAa,GACf,EAAO,KACL,+BAA+B,EAAW,OAAO,kBAAkB,EAAS,OAAO,IAAI,EAAM,QAAQ,EAAE,CAAC,8BACzG,CAIH,IAAM,EAAgB,EAAS,MAAM;EAAK,CAAC,OACrC,EAAkB,EAAW,MAAM;EAAK,CAAC,OAe/C,OAbI,EAAgB,GACd,EAAkB,EAAgB,IACpC,EAAO,KACL,oCAAoC,EAAgB,oBAAoB,EAAc,4BACvF,CAID,EAAO,OAAS,GAClB,EAAO,sBAAsB,EAAO,KAAK,KAAK,GAAG,CAC1C,IAGF,IAMI,GAAiB,EAAoB,IAA6B,CAC7E,IAAI,EAAU,EAER,EAAQ,EAAQ,MADJ,8CACoB,CAStC,OARI,IAAO,EAAU,EAAM,IAEvB,CAAC,EAAS,WAAW;EAAK,EAAI,EAAQ,WAAW;EAAK,GACxD,EAAU,EAAQ,QAAQ,OAAQ,GAAG,EAEnC,CAAC,EAAS,WAAW,IAAI,EAAI,EAAQ,WAAW,IAAI,GACtD,EAAU,EAAQ,WAAW,EAExB"}

package/dist/esm/translation-alignment/alignBlocks.mjs

@@ -1,67 +1,2 @@
-import { computeJaccardSimilarity } from "./computeSimilarity.mjs";
-
-//#region src/translation-alignment/alignBlocks.ts
-const alignEnglishAndFrenchBlocks = (defaultBlocks, secondaryBlocks) => {
-	const defaultLength = defaultBlocks.length;
-	const secondaryLength = secondaryBlocks.length;
-	const scoreMatrix = Array.from({ length: defaultLength + 1 }, () => Array.from({ length: secondaryLength + 1 }, () => 0));
-	const traceMatrix = Array.from({ length: defaultLength + 1 }, () => Array.from({ length: secondaryLength + 1 }, () => "diagonal"));
-	const gapPenalty = -2;
-	const computeMatchScore = (defaultIndex, secondaryIndex) => {
-		const defaultBlock = defaultBlocks[defaultIndex];
-		const secondaryBlock = secondaryBlocks[secondaryIndex];
-		const typeBonus = defaultBlock.type === secondaryBlock.type ? 2 : 0;
-		const anchorSimilarity = computeJaccardSimilarity(defaultBlock.anchorText, secondaryBlock.anchorText, 3);
-		return typeBonus + (Math.min(defaultBlock.content.length, secondaryBlock.content.length) / Math.max(defaultBlock.content.length, secondaryBlock.content.length) > .75 ? 1 : 0) + anchorSimilarity * 8;
-	};
-	for (let i = 1; i <= defaultLength; i += 1) {
-		scoreMatrix[i][0] = scoreMatrix[i - 1][0] + gapPenalty;
-		traceMatrix[i][0] = "up";
-	}
-	for (let j = 1; j <= secondaryLength; j += 1) {
-		scoreMatrix[0][j] = scoreMatrix[0][j - 1] + gapPenalty;
-		traceMatrix[0][j] = "left";
-	}
-	for (let i = 1; i <= defaultLength; i += 1) for (let j = 1; j <= secondaryLength; j += 1) {
-		const match = scoreMatrix[i - 1][j - 1] + computeMatchScore(i - 1, j - 1);
-		const deleteGap = scoreMatrix[i - 1][j] + gapPenalty;
-		const insertGap = scoreMatrix[i][j - 1] + gapPenalty;
-		const best = Math.max(match, deleteGap, insertGap);
-		scoreMatrix[i][j] = best;
-		traceMatrix[i][j] = best === match ? "diagonal" : best === deleteGap ? "up" : "left";
-	}
-	const result = [];
-	let i = defaultLength;
-	let j = secondaryLength;
-	while (i > 0 || j > 0) if (i > 0 && j > 0 && traceMatrix[i][j] === "diagonal") {
-		const englishIndex = i - 1;
-		const frenchIndex = j - 1;
-		const similarityScore = computeJaccardSimilarity(defaultBlocks[englishIndex].anchorText, secondaryBlocks[frenchIndex].anchorText, 3);
-		result.unshift({
-			englishIndex,
-			frenchIndex,
-			similarityScore
-		});
-		i -= 1;
-		j -= 1;
-	} else if (i > 0 && (j === 0 || traceMatrix[i][j] === "up")) {
-		result.unshift({
-			englishIndex: i - 1,
-			frenchIndex: null,
-			similarityScore: 0
-		});
-		i -= 1;
-	} else if (j > 0 && (i === 0 || traceMatrix[i][j] === "left")) {
-		result.unshift({
-			englishIndex: -1,
-			frenchIndex: j - 1,
-			similarityScore: 0
-		});
-		j -= 1;
-	}
-	return result;
-};
-
-//#endregion
-export { alignEnglishAndFrenchBlocks };
+import{computeJaccardSimilarity as e}from"./computeSimilarity.mjs";const t=(t,n)=>{let r=t.length,i=n.length,a=Array.from({length:r+1},()=>Array.from({length:i+1},()=>0)),o=Array.from({length:r+1},()=>Array.from({length:i+1},()=>`diagonal`)),s=(r,i)=>{let a=t[r],o=n[i],s=a.type===o.type?2:0,c=e(a.anchorText,o.anchorText,3);return s+(Math.min(a.content.length,o.content.length)/Math.max(a.content.length,o.content.length)>.75?1:0)+c*8};for(let e=1;e<=r;e+=1)a[e][0]=a[e-1][0]+-2,o[e][0]=`up`;for(let e=1;e<=i;e+=1)a[0][e]=a[0][e-1]+-2,o[0][e]=`left`;for(let e=1;e<=r;e+=1)for(let t=1;t<=i;t+=1){let n=a[e-1][t-1]+s(e-1,t-1),r=a[e-1][t]+-2,i=a[e][t-1]+-2,c=Math.max(n,r,i);a[e][t]=c,o[e][t]=c===n?`diagonal`:c===r?`up`:`left`}let c=[],l=r,u=i;for(;l>0||u>0;)if(l>0&&u>0&&o[l][u]===`diagonal`){let r=l-1,i=u-1,a=e(t[r].anchorText,n[i].anchorText,3);c.unshift({englishIndex:r,frenchIndex:i,similarityScore:a}),--l,--u}else l>0&&(u===0||o[l][u]===`up`)?(c.unshift({englishIndex:l-1,frenchIndex:null,similarityScore:0}),--l):u>0&&(l===0||o[l][u]===`left`)&&(c.unshift({englishIndex:-1,frenchIndex:u-1,similarityScore:0}),--u);return c};export{t as alignEnglishAndFrenchBlocks};
 //# sourceMappingURL=alignBlocks.mjs.map

package/dist/esm/translation-alignment/alignBlocks.mjs.map

@@ -1 +1 @@
-{"version":3,"file":"alignBlocks.mjs","names":[],"sources":["../../../src/translation-alignment/alignBlocks.ts"],"sourcesContent":["import { computeJaccardSimilarity } from './computeSimilarity';\nimport type { AlignmentPair, FingerprintedBlock } from './types';\n\nexport const alignEnglishAndFrenchBlocks = (\n  defaultBlocks: FingerprintedBlock[],\n  secondaryBlocks: FingerprintedBlock[]\n): AlignmentPair[] => {\n  // Needleman–Wunsch style global alignment using anchor similarity and type equality\n  const defaultLength = defaultBlocks.length;\n  const secondaryLength = secondaryBlocks.length;\n\n  const scoreMatrix: number[][] = Array.from(\n    { length: defaultLength + 1 },\n    () => Array.from({ length: secondaryLength + 1 }, () => 0)\n  );\n  const traceMatrix: ('diagonal' | 'up' | 'left')[][] = Array.from(\n    { length: defaultLength + 1 },\n    () => Array.from({ length: secondaryLength + 1 }, () => 'diagonal')\n  );\n\n  const gapPenalty = -2;\n\n  const computeMatchScore = (\n    defaultIndex: number,\n    secondaryIndex: number\n  ): number => {\n    const defaultBlock = defaultBlocks[defaultIndex];\n    const secondaryBlock = secondaryBlocks[secondaryIndex];\n    const typeBonus = defaultBlock.type === secondaryBlock.type ? 2 : 0;\n    const anchorSimilarity = computeJaccardSimilarity(\n      defaultBlock.anchorText,\n      secondaryBlock.anchorText,\n      3\n    );\n    const lengthRatio =\n      Math.min(defaultBlock.content.length, secondaryBlock.content.length) /\n      Math.max(defaultBlock.content.length, secondaryBlock.content.length);\n    const lengthBonus = lengthRatio > 0.75 ? 1 : 0;\n    return typeBonus + lengthBonus + anchorSimilarity * 8; // weighted toward anchor similarity\n  };\n\n  // initialize first row and column\n  for (let i = 1; i <= defaultLength; i += 1) {\n    scoreMatrix[i][0] = scoreMatrix[i - 1][0] + gapPenalty;\n    traceMatrix[i][0] = 'up';\n  }\n  for (let j = 1; j <= secondaryLength; j += 1) {\n    scoreMatrix[0][j] = scoreMatrix[0][j - 1] + gapPenalty;\n    traceMatrix[0][j] = 'left';\n  }\n\n  // fill\n  for (let i = 1; i <= defaultLength; i += 1) {\n    for (let j = 1; j <= secondaryLength; j += 1) {\n      const match = scoreMatrix[i - 1][j - 1] + computeMatchScore(i - 1, j - 1);\n      const deleteGap = scoreMatrix[i - 1][j] + gapPenalty;\n      const insertGap = scoreMatrix[i][j - 1] + gapPenalty;\n\n      const best = Math.max(match, deleteGap, insertGap);\n      scoreMatrix[i][j] = best;\n      traceMatrix[i][j] =\n        best === match ? 'diagonal' : best === deleteGap ? 'up' : 'left';\n    }\n  }\n\n  // traceback\n  const result: AlignmentPair[] = [];\n  let i = defaultLength;\n  let j = secondaryLength;\n  while (i > 0 || j > 0) {\n    if (i > 0 && j > 0 && traceMatrix[i][j] === 'diagonal') {\n      const englishIndex = i - 1;\n      const frenchIndex = j - 1;\n      const similarityScore = computeJaccardSimilarity(\n        defaultBlocks[englishIndex].anchorText,\n        secondaryBlocks[frenchIndex].anchorText,\n        3\n      );\n      result.unshift({ englishIndex, frenchIndex, similarityScore });\n      i -= 1;\n      j -= 1;\n    } else if (i > 0 && (j === 0 || traceMatrix[i][j] === 'up')) {\n      result.unshift({\n        englishIndex: i - 1,\n        frenchIndex: null,\n        similarityScore: 0,\n      });\n      i -= 1;\n    } else if (j > 0 && (i === 0 || traceMatrix[i][j] === 'left')) {\n      // french block has no corresponding english block (deleted)\n      result.unshift({\n        englishIndex: -1,\n        frenchIndex: j - 1,\n        similarityScore: 0,\n      });\n      j -= 1;\n    }\n  }\n  return result;\n};\n"],"mappings":";;;AAGA,MAAa,+BACX,eACA,oBACoB;CAEpB,MAAM,gBAAgB,cAAc;CACpC,MAAM,kBAAkB,gBAAgB;CAExC,MAAM,cAA0B,MAAM,KACpC,EAAE,QAAQ,gBAAgB,GAAG,QACvB,MAAM,KAAK,EAAE,QAAQ,kBAAkB,GAAG,QAAQ,EAAE,CAC3D;CACD,MAAM,cAAgD,MAAM,KAC1D,EAAE,QAAQ,gBAAgB,GAAG,QACvB,MAAM,KAAK,EAAE,QAAQ,kBAAkB,GAAG,QAAQ,WAAW,CACpE;CAED,MAAM,aAAa;CAEnB,MAAM,qBACJ,cACA,mBACW;EACX,MAAM,eAAe,cAAc;EACnC,MAAM,iBAAiB,gBAAgB;EACvC,MAAM,YAAY,aAAa,SAAS,eAAe,OAAO,IAAI;EAClE,MAAM,mBAAmB,yBACvB,aAAa,YACb,eAAe,YACf,EACD;AAKD,SAAO,aAHL,KAAK,IAAI,aAAa,QAAQ,QAAQ,eAAe,QAAQ,OAAO,GACpE,KAAK,IAAI,aAAa,QAAQ,QAAQ,eAAe,QAAQ,OAAO,GACpC,MAAO,IAAI,KACZ,mBAAmB;;AAItD,MAAK,IAAI,IAAI,GAAG,KAAK,eAAe,KAAK,GAAG;AAC1C,cAAY,GAAG,KAAK,YAAY,IAAI,GAAG,KAAK;AAC5C,cAAY,GAAG,KAAK;;AAEtB,MAAK,IAAI,IAAI,GAAG,KAAK,iBAAiB,KAAK,GAAG;AAC5C,cAAY,GAAG,KAAK,YAAY,GAAG,IAAI,KAAK;AAC5C,cAAY,GAAG,KAAK;;AAItB,MAAK,IAAI,IAAI,GAAG,KAAK,eAAe,KAAK,EACvC,MAAK,IAAI,IAAI,GAAG,KAAK,iBAAiB,KAAK,GAAG;EAC5C,MAAM,QAAQ,YAAY,IAAI,GAAG,IAAI,KAAK,kBAAkB,IAAI,GAAG,IAAI,EAAE;EACzE,MAAM,YAAY,YAAY,IAAI,GAAG,KAAK;EAC1C,MAAM,YAAY,YAAY,GAAG,IAAI,KAAK;EAE1C,MAAM,OAAO,KAAK,IAAI,OAAO,WAAW,UAAU;AAClD,cAAY,GAAG,KAAK;AACpB,cAAY,GAAG,KACb,SAAS,QAAQ,aAAa,SAAS,YAAY,OAAO;;CAKhE,MAAM,SAA0B,EAAE;CAClC,IAAI,IAAI;CACR,IAAI,IAAI;AACR,QAAO,IAAI,KAAK,IAAI,EAClB,KAAI,IAAI,KAAK,IAAI,KAAK,YAAY,GAAG,OAAO,YAAY;EACtD,MAAM,eAAe,IAAI;EACzB,MAAM,cAAc,IAAI;EACxB,MAAM,kBAAkB,yBACtB,cAAc,cAAc,YAC5B,gBAAgB,aAAa,YAC7B,EACD;AACD,SAAO,QAAQ;GAAE;GAAc;GAAa;GAAiB,CAAC;AAC9D,OAAK;AACL,OAAK;YACI,IAAI,MAAM,MAAM,KAAK,YAAY,GAAG,OAAO,OAAO;AAC3D,SAAO,QAAQ;GACb,cAAc,IAAI;GAClB,aAAa;GACb,iBAAiB;GAClB,CAAC;AACF,OAAK;YACI,IAAI,MAAM,MAAM,KAAK,YAAY,GAAG,OAAO,SAAS;AAE7D,SAAO,QAAQ;GACb,cAAc;GACd,aAAa,IAAI;GACjB,iBAAiB;GAClB,CAAC;AACF,OAAK;;AAGT,QAAO"}
+{"version":3,"file":"alignBlocks.mjs","names":[],"sources":["../../../src/translation-alignment/alignBlocks.ts"],"sourcesContent":["import { computeJaccardSimilarity } from './computeSimilarity';\nimport type { AlignmentPair, FingerprintedBlock } from './types';\n\nexport const alignEnglishAndFrenchBlocks = (\n  defaultBlocks: FingerprintedBlock[],\n  secondaryBlocks: FingerprintedBlock[]\n): AlignmentPair[] => {\n  // Needleman–Wunsch style global alignment using anchor similarity and type equality\n  const defaultLength = defaultBlocks.length;\n  const secondaryLength = secondaryBlocks.length;\n\n  const scoreMatrix: number[][] = Array.from(\n    { length: defaultLength + 1 },\n    () => Array.from({ length: secondaryLength + 1 }, () => 0)\n  );\n  const traceMatrix: ('diagonal' | 'up' | 'left')[][] = Array.from(\n    { length: defaultLength + 1 },\n    () => Array.from({ length: secondaryLength + 1 }, () => 'diagonal')\n  );\n\n  const gapPenalty = -2;\n\n  const computeMatchScore = (\n    defaultIndex: number,\n    secondaryIndex: number\n  ): number => {\n    const defaultBlock = defaultBlocks[defaultIndex];\n    const secondaryBlock = secondaryBlocks[secondaryIndex];\n    const typeBonus = defaultBlock.type === secondaryBlock.type ? 2 : 0;\n    const anchorSimilarity = computeJaccardSimilarity(\n      defaultBlock.anchorText,\n      secondaryBlock.anchorText,\n      3\n    );\n    const lengthRatio =\n      Math.min(defaultBlock.content.length, secondaryBlock.content.length) /\n      Math.max(defaultBlock.content.length, secondaryBlock.content.length);\n    const lengthBonus = lengthRatio > 0.75 ? 1 : 0;\n    return typeBonus + lengthBonus + anchorSimilarity * 8; // weighted toward anchor similarity\n  };\n\n  // initialize first row and column\n  for (let i = 1; i <= defaultLength; i += 1) {\n    scoreMatrix[i][0] = scoreMatrix[i - 1][0] + gapPenalty;\n    traceMatrix[i][0] = 'up';\n  }\n  for (let j = 1; j <= secondaryLength; j += 1) {\n    scoreMatrix[0][j] = scoreMatrix[0][j - 1] + gapPenalty;\n    traceMatrix[0][j] = 'left';\n  }\n\n  // fill\n  for (let i = 1; i <= defaultLength; i += 1) {\n    for (let j = 1; j <= secondaryLength; j += 1) {\n      const match = scoreMatrix[i - 1][j - 1] + computeMatchScore(i - 1, j - 1);\n      const deleteGap = scoreMatrix[i - 1][j] + gapPenalty;\n      const insertGap = scoreMatrix[i][j - 1] + gapPenalty;\n\n      const best = Math.max(match, deleteGap, insertGap);\n      scoreMatrix[i][j] = best;\n      traceMatrix[i][j] =\n        best === match ? 'diagonal' : best === deleteGap ? 'up' : 'left';\n    }\n  }\n\n  // traceback\n  const result: AlignmentPair[] = [];\n  let i = defaultLength;\n  let j = secondaryLength;\n  while (i > 0 || j > 0) {\n    if (i > 0 && j > 0 && traceMatrix[i][j] === 'diagonal') {\n      const englishIndex = i - 1;\n      const frenchIndex = j - 1;\n      const similarityScore = computeJaccardSimilarity(\n        defaultBlocks[englishIndex].anchorText,\n        secondaryBlocks[frenchIndex].anchorText,\n        3\n      );\n      result.unshift({ englishIndex, frenchIndex, similarityScore });\n      i -= 1;\n      j -= 1;\n    } else if (i > 0 && (j === 0 || traceMatrix[i][j] === 'up')) {\n      result.unshift({\n        englishIndex: i - 1,\n        frenchIndex: null,\n        similarityScore: 0,\n      });\n      i -= 1;\n    } else if (j > 0 && (i === 0 || traceMatrix[i][j] === 'left')) {\n      // french block has no corresponding english block (deleted)\n      result.unshift({\n        englishIndex: -1,\n        frenchIndex: j - 1,\n        similarityScore: 0,\n      });\n      j -= 1;\n    }\n  }\n  return result;\n};\n"],"mappings":"mEAGA,MAAa,GACX,EACA,IACoB,CAEpB,IAAM,EAAgB,EAAc,OAC9B,EAAkB,EAAgB,OAElC,EAA0B,MAAM,KACpC,CAAE,OAAQ,EAAgB,EAAG,KACvB,MAAM,KAAK,CAAE,OAAQ,EAAkB,EAAG,KAAQ,EAAE,CAC3D,CACK,EAAgD,MAAM,KAC1D,CAAE,OAAQ,EAAgB,EAAG,KACvB,MAAM,KAAK,CAAE,OAAQ,EAAkB,EAAG,KAAQ,WAAW,CACpE,CAIK,GACJ,EACA,IACW,CACX,IAAM,EAAe,EAAc,GAC7B,EAAiB,EAAgB,GACjC,EAAY,EAAa,OAAS,EAAe,KAAO,EAAI,EAC5D,EAAmB,EACvB,EAAa,WACb,EAAe,WACf,EACD,CAKD,OAAO,GAHL,KAAK,IAAI,EAAa,QAAQ,OAAQ,EAAe,QAAQ,OAAO,CACpE,KAAK,IAAI,EAAa,QAAQ,OAAQ,EAAe,QAAQ,OAAO,CACpC,IAAO,EAAI,GACZ,EAAmB,GAItD,IAAK,IAAI,EAAI,EAAG,GAAK,EAAe,GAAK,EACvC,EAAY,GAAG,GAAK,EAAY,EAAI,GAAG,GAAK,GAC5C,EAAY,GAAG,GAAK,KAEtB,IAAK,IAAI,EAAI,EAAG,GAAK,EAAiB,GAAK,EACzC,EAAY,GAAG,GAAK,EAAY,GAAG,EAAI,GAAK,GAC5C,EAAY,GAAG,GAAK,OAItB,IAAK,IAAI,EAAI,EAAG,GAAK,EAAe,GAAK,EACvC,IAAK,IAAI,EAAI,EAAG,GAAK,EAAiB,GAAK,EAAG,CAC5C,IAAM,EAAQ,EAAY,EAAI,GAAG,EAAI,GAAK,EAAkB,EAAI,EAAG,EAAI,EAAE,CACnE,EAAY,EAAY,EAAI,GAAG,GAAK,GACpC,EAAY,EAAY,GAAG,EAAI,GAAK,GAEpC,EAAO,KAAK,IAAI,EAAO,EAAW,EAAU,CAClD,EAAY,GAAG,GAAK,EACpB,EAAY,GAAG,GACb,IAAS,EAAQ,WAAa,IAAS,EAAY,KAAO,OAKhE,IAAM,EAA0B,EAAE,CAC9B,EAAI,EACJ,EAAI,EACR,KAAO,EAAI,GAAK,EAAI,GAClB,GAAI,EAAI,GAAK,EAAI,GAAK,EAAY,GAAG,KAAO,WAAY,CACtD,IAAM,EAAe,EAAI,EACnB,EAAc,EAAI,EAClB,EAAkB,EACtB,EAAc,GAAc,WAC5B,EAAgB,GAAa,WAC7B,EACD,CACD,EAAO,QAAQ,CAAE,eAAc,cAAa,kBAAiB,CAAC,CAC9D,IACA,SACS,EAAI,IAAM,IAAM,GAAK,EAAY,GAAG,KAAO,OACpD,EAAO,QAAQ,CACb,aAAc,EAAI,EAClB,YAAa,KACb,gBAAiB,EAClB,CAAC,CACF,KACS,EAAI,IAAM,IAAM,GAAK,EAAY,GAAG,KAAO,UAEpD,EAAO,QAAQ,CACb,aAAc,GACd,YAAa,EAAI,EACjB,gBAAiB,EAClB,CAAC,CACF,KAGJ,OAAO"}

package/dist/esm/translation-alignment/computeSimilarity.mjs

@@ -1,23 +1,2 @@
-//#region src/translation-alignment/computeSimilarity.ts
-const generateCharacterShingles = (text, shingleLength) => {
-	const normalized = text.replace(/\s+/g, " ").trim();
-	const set = /* @__PURE__ */ new Set();
-	if (normalized.length < shingleLength) {
-		if (normalized.length > 0) set.add(normalized);
-		return set;
-	}
-	for (let index = 0; index <= normalized.length - shingleLength; index += 1) set.add(normalized.slice(index, index + shingleLength));
-	return set;
-};
-const computeJaccardSimilarity = (a, b, shingleLength = 3) => {
-	const setA = generateCharacterShingles(a, shingleLength);
-	const setB = generateCharacterShingles(b, shingleLength);
-	if (setA.size === 0 && setB.size === 0) return 1;
-	const intersectionSize = Array.from(setA).filter((token) => setB.has(token)).length;
-	const unionSize = new Set([...Array.from(setA), ...Array.from(setB)]).size;
-	return unionSize === 0 ? 0 : intersectionSize / unionSize;
-};
-
-//#endregion
-export { computeJaccardSimilarity, generateCharacterShingles };
+const e=(e,t)=>{let n=e.replace(/\s+/g,` `).trim(),r=new Set;if(n.length<t)return n.length>0&&r.add(n),r;for(let e=0;e<=n.length-t;e+=1)r.add(n.slice(e,e+t));return r},t=(t,n,r=3)=>{let i=e(t,r),a=e(n,r);if(i.size===0&&a.size===0)return 1;let o=Array.from(i).filter(e=>a.has(e)).length,s=new Set([...Array.from(i),...Array.from(a)]).size;return s===0?0:o/s};export{t as computeJaccardSimilarity,e as generateCharacterShingles};
 //# sourceMappingURL=computeSimilarity.mjs.map

package/dist/esm/translation-alignment/computeSimilarity.mjs.map

@@ -1 +1 @@
-{"version":3,"file":"computeSimilarity.mjs","names":[],"sources":["../../../src/translation-alignment/computeSimilarity.ts"],"sourcesContent":["// Character shingle Jaccard similarity (language agnostic)\nexport const generateCharacterShingles = (\n  text: string,\n  shingleLength: number\n): Set<string> => {\n  const normalized = text.replace(/\\s+/g, ' ').trim();\n  const set = new Set<string>();\n  if (normalized.length < shingleLength) {\n    if (normalized.length > 0) {\n      set.add(normalized);\n    }\n    return set;\n  }\n  for (let index = 0; index <= normalized.length - shingleLength; index += 1) {\n    set.add(normalized.slice(index, index + shingleLength));\n  }\n  return set;\n};\n\nexport const computeJaccardSimilarity = (\n  a: string,\n  b: string,\n  shingleLength: number = 3\n): number => {\n  const setA = generateCharacterShingles(a, shingleLength);\n  const setB = generateCharacterShingles(b, shingleLength);\n  if (setA.size === 0 && setB.size === 0) return 1;\n  const intersectionSize = Array.from(setA).filter((token) =>\n    setB.has(token)\n  ).length;\n  const unionSize = new Set([...Array.from(setA), ...Array.from(setB)]).size;\n  return unionSize === 0 ? 0 : intersectionSize / unionSize;\n};\n"],"mappings":";AACA,MAAa,6BACX,MACA,kBACgB;CAChB,MAAM,aAAa,KAAK,QAAQ,QAAQ,IAAI,CAAC,MAAM;CACnD,MAAM,sBAAM,IAAI,KAAa;AAC7B,KAAI,WAAW,SAAS,eAAe;AACrC,MAAI,WAAW,SAAS,EACtB,KAAI,IAAI,WAAW;AAErB,SAAO;;AAET,MAAK,IAAI,QAAQ,GAAG,SAAS,WAAW,SAAS,eAAe,SAAS,EACvE,KAAI,IAAI,WAAW,MAAM,OAAO,QAAQ,cAAc,CAAC;AAEzD,QAAO;;AAGT,MAAa,4BACX,GACA,GACA,gBAAwB,MACb;CACX,MAAM,OAAO,0BAA0B,GAAG,cAAc;CACxD,MAAM,OAAO,0BAA0B,GAAG,cAAc;AACxD,KAAI,KAAK,SAAS,KAAK,KAAK,SAAS,EAAG,QAAO;CAC/C,MAAM,mBAAmB,MAAM,KAAK,KAAK,CAAC,QAAQ,UAChD,KAAK,IAAI,MAAM,CAChB,CAAC;CACF,MAAM,YAAY,IAAI,IAAI,CAAC,GAAG,MAAM,KAAK,KAAK,EAAE,GAAG,MAAM,KAAK,KAAK,CAAC,CAAC,CAAC;AACtE,QAAO,cAAc,IAAI,IAAI,mBAAmB"}
+{"version":3,"file":"computeSimilarity.mjs","names":[],"sources":["../../../src/translation-alignment/computeSimilarity.ts"],"sourcesContent":["// Character shingle Jaccard similarity (language agnostic)\nexport const generateCharacterShingles = (\n  text: string,\n  shingleLength: number\n): Set<string> => {\n  const normalized = text.replace(/\\s+/g, ' ').trim();\n  const set = new Set<string>();\n  if (normalized.length < shingleLength) {\n    if (normalized.length > 0) {\n      set.add(normalized);\n    }\n    return set;\n  }\n  for (let index = 0; index <= normalized.length - shingleLength; index += 1) {\n    set.add(normalized.slice(index, index + shingleLength));\n  }\n  return set;\n};\n\nexport const computeJaccardSimilarity = (\n  a: string,\n  b: string,\n  shingleLength: number = 3\n): number => {\n  const setA = generateCharacterShingles(a, shingleLength);\n  const setB = generateCharacterShingles(b, shingleLength);\n  if (setA.size === 0 && setB.size === 0) return 1;\n  const intersectionSize = Array.from(setA).filter((token) =>\n    setB.has(token)\n  ).length;\n  const unionSize = new Set([...Array.from(setA), ...Array.from(setB)]).size;\n  return unionSize === 0 ? 0 : intersectionSize / unionSize;\n};\n"],"mappings":"AACA,MAAa,GACX,EACA,IACgB,CAChB,IAAM,EAAa,EAAK,QAAQ,OAAQ,IAAI,CAAC,MAAM,CAC7C,EAAM,IAAI,IAChB,GAAI,EAAW,OAAS,EAItB,OAHI,EAAW,OAAS,GACtB,EAAI,IAAI,EAAW,CAEd,EAET,IAAK,IAAI,EAAQ,EAAG,GAAS,EAAW,OAAS,EAAe,GAAS,EACvE,EAAI,IAAI,EAAW,MAAM,EAAO,EAAQ,EAAc,CAAC,CAEzD,OAAO,GAGI,GACX,EACA,EACA,EAAwB,IACb,CACX,IAAM,EAAO,EAA0B,EAAG,EAAc,CAClD,EAAO,EAA0B,EAAG,EAAc,CACxD,GAAI,EAAK,OAAS,GAAK,EAAK,OAAS,EAAG,MAAO,GAC/C,IAAM,EAAmB,MAAM,KAAK,EAAK,CAAC,OAAQ,GAChD,EAAK,IAAI,EAAM,CAChB,CAAC,OACI,EAAY,IAAI,IAAI,CAAC,GAAG,MAAM,KAAK,EAAK,CAAE,GAAG,MAAM,KAAK,EAAK,CAAC,CAAC,CAAC,KACtE,OAAO,IAAc,EAAI,EAAI,EAAmB"}

package/dist/esm/translation-alignment/fingerprintBlock.mjs

@@ -1,21 +1,2 @@
-import crypto from "node:crypto";
-
-//#region src/translation-alignment/fingerprintBlock.ts
-const computeStringDigest = (text) => crypto.createHash("sha256").update(text).digest("hex");
-const fingerprintBlock = (block, previousBlock, nextBlock) => {
-	const semanticDigest = computeStringDigest(block.semanticText);
-	const anchorDigest = computeStringDigest(block.anchorText);
-	const compositeKey = `${semanticDigest}:${anchorDigest}`;
-	const contextKey = computeStringDigest(`${computeStringDigest(previousBlock?.semanticText ?? "")}:${computeStringDigest(nextBlock?.semanticText ?? "")}`);
-	return {
-		...block,
-		semanticDigest,
-		anchorDigest,
-		compositeKey,
-		contextKey
-	};
-};
-
-//#endregion
-export { fingerprintBlock };
+import e from"node:crypto";const t=t=>e.createHash(`sha256`).update(t).digest(`hex`),n=(e,n,r)=>{let i=t(e.semanticText),a=t(e.anchorText),o=`${i}:${a}`,s=t(`${t(n?.semanticText??``)}:${t(r?.semanticText??``)}`);return{...e,semanticDigest:i,anchorDigest:a,compositeKey:o,contextKey:s}};export{n as fingerprintBlock};
 //# sourceMappingURL=fingerprintBlock.mjs.map

package/dist/esm/translation-alignment/fingerprintBlock.mjs.map

@@ -1 +1 @@
-{"version":3,"file":"fingerprintBlock.mjs","names":[],"sources":["../../../src/translation-alignment/fingerprintBlock.ts"],"sourcesContent":["import crypto from 'node:crypto';\nimport type { FingerprintedBlock, NormalizedBlock } from './types';\n\nconst computeStringDigest = (text: string): string =>\n  crypto.createHash('sha256').update(text).digest('hex');\n\nexport const fingerprintBlock = (\n  block: NormalizedBlock,\n  previousBlock: NormalizedBlock | null,\n  nextBlock: NormalizedBlock | null\n): FingerprintedBlock => {\n  const semanticDigest = computeStringDigest(block.semanticText);\n  const anchorDigest = computeStringDigest(block.anchorText);\n  const compositeKey = `${semanticDigest}:${anchorDigest}`;\n\n  const previousDigest = computeStringDigest(previousBlock?.semanticText ?? '');\n  const nextDigest = computeStringDigest(nextBlock?.semanticText ?? '');\n  const contextKey = computeStringDigest(`${previousDigest}:${nextDigest}`);\n\n  return {\n    ...block,\n    semanticDigest,\n    anchorDigest,\n    compositeKey,\n    contextKey,\n  };\n};\n"],"mappings":";;;AAGA,MAAM,uBAAuB,SAC3B,OAAO,WAAW,SAAS,CAAC,OAAO,KAAK,CAAC,OAAO,MAAM;AAExD,MAAa,oBACX,OACA,eACA,cACuB;CACvB,MAAM,iBAAiB,oBAAoB,MAAM,aAAa;CAC9D,MAAM,eAAe,oBAAoB,MAAM,WAAW;CAC1D,MAAM,eAAe,GAAG,eAAe,GAAG;CAI1C,MAAM,aAAa,oBAAoB,GAFhB,oBAAoB,eAAe,gBAAgB,GAAG,CAEpB,GADtC,oBAAoB,WAAW,gBAAgB,GAAG,GACI;AAEzE,QAAO;EACL,GAAG;EACH;EACA;EACA;EACA;EACD"}
+{"version":3,"file":"fingerprintBlock.mjs","names":[],"sources":["../../../src/translation-alignment/fingerprintBlock.ts"],"sourcesContent":["import crypto from 'node:crypto';\nimport type { FingerprintedBlock, NormalizedBlock } from './types';\n\nconst computeStringDigest = (text: string): string =>\n  crypto.createHash('sha256').update(text).digest('hex');\n\nexport const fingerprintBlock = (\n  block: NormalizedBlock,\n  previousBlock: NormalizedBlock | null,\n  nextBlock: NormalizedBlock | null\n): FingerprintedBlock => {\n  const semanticDigest = computeStringDigest(block.semanticText);\n  const anchorDigest = computeStringDigest(block.anchorText);\n  const compositeKey = `${semanticDigest}:${anchorDigest}`;\n\n  const previousDigest = computeStringDigest(previousBlock?.semanticText ?? '');\n  const nextDigest = computeStringDigest(nextBlock?.semanticText ?? '');\n  const contextKey = computeStringDigest(`${previousDigest}:${nextDigest}`);\n\n  return {\n    ...block,\n    semanticDigest,\n    anchorDigest,\n    compositeKey,\n    contextKey,\n  };\n};\n"],"mappings":"2BAGA,MAAM,EAAuB,GAC3B,EAAO,WAAW,SAAS,CAAC,OAAO,EAAK,CAAC,OAAO,MAAM,CAE3C,GACX,EACA,EACA,IACuB,CACvB,IAAM,EAAiB,EAAoB,EAAM,aAAa,CACxD,EAAe,EAAoB,EAAM,WAAW,CACpD,EAAe,GAAG,EAAe,GAAG,IAIpC,EAAa,EAAoB,GAFhB,EAAoB,GAAe,cAAgB,GAAG,CAEpB,GADtC,EAAoB,GAAW,cAAgB,GAAG,GACI,CAEzE,MAAO,CACL,GAAG,EACH,iBACA,eACA,eACA,aACD"}

package/dist/esm/translation-alignment/index.mjs

@@ -1,11 +1 @@
-import { computeJaccardSimilarity, generateCharacterShingles } from "./computeSimilarity.mjs";
-import { alignEnglishAndFrenchBlocks } from "./alignBlocks.mjs";
-import { fingerprintBlock } from "./fingerprintBlock.mjs";
-import { mapChangedLinesToBlocks } from "./mapChangedLinesToBlocks.mjs";
-import { normalizeBlock } from "./normalizeBlock.mjs";
-import { planAlignmentActions } from "./planActions.mjs";
-import { identifySegmentsToReview, mergeReviewedSegments } from "./rebuildDocument.mjs";
-import { segmentDocument } from "./segmentDocument.mjs";
-import { buildAlignmentPlan } from "./pipeline.mjs";
-
-export { alignEnglishAndFrenchBlocks, buildAlignmentPlan, computeJaccardSimilarity, fingerprintBlock, generateCharacterShingles, identifySegmentsToReview, mapChangedLinesToBlocks, mergeReviewedSegments, normalizeBlock, planAlignmentActions, segmentDocument };
+import{computeJaccardSimilarity as e,generateCharacterShingles as t}from"./computeSimilarity.mjs";import{alignEnglishAndFrenchBlocks as n}from"./alignBlocks.mjs";import{fingerprintBlock as r}from"./fingerprintBlock.mjs";import{mapChangedLinesToBlocks as i}from"./mapChangedLinesToBlocks.mjs";import{normalizeBlock as a}from"./normalizeBlock.mjs";import{planAlignmentActions as o}from"./planActions.mjs";import{identifySegmentsToReview as s,mergeReviewedSegments as c}from"./rebuildDocument.mjs";import{segmentDocument as l}from"./segmentDocument.mjs";import{buildAlignmentPlan as u}from"./pipeline.mjs";export{n as alignEnglishAndFrenchBlocks,u as buildAlignmentPlan,e as computeJaccardSimilarity,r as fingerprintBlock,t as generateCharacterShingles,s as identifySegmentsToReview,i as mapChangedLinesToBlocks,c as mergeReviewedSegments,a as normalizeBlock,o as planAlignmentActions,l as segmentDocument};

package/dist/esm/translation-alignment/mapChangedLinesToBlocks.mjs

@@ -1,17 +1,2 @@
-//#region src/translation-alignment/mapChangedLinesToBlocks.ts
-const mapChangedLinesToBlocks = (blocks, changedLines) => {
-	const changedSet = /* @__PURE__ */ new Set();
-	if (!changedLines || changedLines.length === 0) return changedSet;
-	const changedLookup = new Set(changedLines);
-	blocks.forEach((block, index) => {
-		for (let line = block.lineStart; line <= block.lineEnd; line += 1) if (changedLookup.has(line)) {
-			changedSet.add(index);
-			break;
-		}
-	});
-	return changedSet;
-};
-
-//#endregion
-export { mapChangedLinesToBlocks };
+const e=(e,t)=>{let n=new Set;if(!t||t.length===0)return n;let r=new Set(t);return e.forEach((e,t)=>{for(let i=e.lineStart;i<=e.lineEnd;i+=1)if(r.has(i)){n.add(t);break}}),n};export{e as mapChangedLinesToBlocks};
 //# sourceMappingURL=mapChangedLinesToBlocks.mjs.map

package/dist/esm/translation-alignment/mapChangedLinesToBlocks.mjs.map

@@ -1 +1 @@
-{"version":3,"file":"mapChangedLinesToBlocks.mjs","names":[],"sources":["../../../src/translation-alignment/mapChangedLinesToBlocks.ts"],"sourcesContent":["import type { Block, LineChange } from './types';\n\nexport const mapChangedLinesToBlocks = (\n  blocks: Block[],\n  changedLines: LineChange[]\n): Set<number> => {\n  const changedSet = new Set<number>();\n  if (!changedLines || changedLines.length === 0) return changedSet;\n\n  const changedLookup = new Set<number>(changedLines);\n\n  blocks.forEach((block, index) => {\n    for (let line = block.lineStart; line <= block.lineEnd; line += 1) {\n      if (changedLookup.has(line)) {\n        changedSet.add(index);\n        break;\n      }\n    }\n  });\n\n  return changedSet;\n};\n"],"mappings":";AAEA,MAAa,2BACX,QACA,iBACgB;CAChB,MAAM,6BAAa,IAAI,KAAa;AACpC,KAAI,CAAC,gBAAgB,aAAa,WAAW,EAAG,QAAO;CAEvD,MAAM,gBAAgB,IAAI,IAAY,aAAa;AAEnD,QAAO,SAAS,OAAO,UAAU;AAC/B,OAAK,IAAI,OAAO,MAAM,WAAW,QAAQ,MAAM,SAAS,QAAQ,EAC9D,KAAI,cAAc,IAAI,KAAK,EAAE;AAC3B,cAAW,IAAI,MAAM;AACrB;;GAGJ;AAEF,QAAO"}
+{"version":3,"file":"mapChangedLinesToBlocks.mjs","names":[],"sources":["../../../src/translation-alignment/mapChangedLinesToBlocks.ts"],"sourcesContent":["import type { Block, LineChange } from './types';\n\nexport const mapChangedLinesToBlocks = (\n  blocks: Block[],\n  changedLines: LineChange[]\n): Set<number> => {\n  const changedSet = new Set<number>();\n  if (!changedLines || changedLines.length === 0) return changedSet;\n\n  const changedLookup = new Set<number>(changedLines);\n\n  blocks.forEach((block, index) => {\n    for (let line = block.lineStart; line <= block.lineEnd; line += 1) {\n      if (changedLookup.has(line)) {\n        changedSet.add(index);\n        break;\n      }\n    }\n  });\n\n  return changedSet;\n};\n"],"mappings":"AAEA,MAAa,GACX,EACA,IACgB,CAChB,IAAM,EAAa,IAAI,IACvB,GAAI,CAAC,GAAgB,EAAa,SAAW,EAAG,OAAO,EAEvD,IAAM,EAAgB,IAAI,IAAY,EAAa,CAWnD,OATA,EAAO,SAAS,EAAO,IAAU,CAC/B,IAAK,IAAI,EAAO,EAAM,UAAW,GAAQ,EAAM,QAAS,GAAQ,EAC9D,GAAI,EAAc,IAAI,EAAK,CAAE,CAC3B,EAAW,IAAI,EAAM,CACrB,QAGJ,CAEK"}

package/dist/esm/translation-alignment/normalizeBlock.mjs

@@ -1,21 +1,2 @@
-//#region src/translation-alignment/normalizeBlock.ts
-const removeMarkdownFormatting = (text) => {
-	return text.replace(/`{1,3}[^`]*`{1,3}/g, " ").replace(/\*\*([^*]+)\*\*/g, "$1").replace(/\*([^*]+)\*/g, "$1").replace(/_([^_]+)_/g, "$1").replace(/~~([^~]+)~~/g, "$1").replace(/!?\[[^\]]*\]\([^)]*\)/g, " ").replace(/^\s*#{1,6}\s+/gm, "").replace(/^\s*>\s?/gm, "").replace(/^\s*[-*+]\s+/gm, "").replace(/^\s*\d+\.\s+/gm, "");
-};
-const collapseWhitespace = (text) => text.replace(/\s+/g, " ").trim();
-const stripLettersKeepDigitsAndSymbols = (text) => {
-	return text.replace(/\p{L}+/gu, "");
-};
-const normalizeBlock = (block) => {
-	const semanticCollapsed = collapseWhitespace(removeMarkdownFormatting(block.content).toLowerCase());
-	const anchorCollapsed = collapseWhitespace(stripLettersKeepDigitsAndSymbols(block.content));
-	return {
-		...block,
-		semanticText: semanticCollapsed,
-		anchorText: anchorCollapsed
-	};
-};
-
-//#endregion
-export { normalizeBlock };
+const e=e=>e.replace(/`{1,3}[^`]*`{1,3}/g,` `).replace(/\*\*([^*]+)\*\*/g,`$1`).replace(/\*([^*]+)\*/g,`$1`).replace(/_([^_]+)_/g,`$1`).replace(/~~([^~]+)~~/g,`$1`).replace(/!?\[[^\]]*\]\([^)]*\)/g,` `).replace(/^\s*#{1,6}\s+/gm,``).replace(/^\s*>\s?/gm,``).replace(/^\s*[-*+]\s+/gm,``).replace(/^\s*\d+\.\s+/gm,``),t=e=>e.replace(/\s+/g,` `).trim(),n=e=>e.replace(/\p{L}+/gu,``),r=r=>{let i=t(e(r.content).toLowerCase()),a=t(n(r.content));return{...r,semanticText:i,anchorText:a}};export{r as normalizeBlock};
 //# sourceMappingURL=normalizeBlock.mjs.map

package/dist/esm/translation-alignment/normalizeBlock.mjs.map

@@ -1 +1 @@
-{"version":3,"file":"normalizeBlock.mjs","names":[],"sources":["../../../src/translation-alignment/normalizeBlock.ts"],"sourcesContent":["import type { Block, NormalizedBlock } from './types';\n\nconst removeMarkdownFormatting = (text: string): string => {\n  return text\n    .replace(/`{1,3}[^`]*`{1,3}/g, ' ')\n    .replace(/\\*\\*([^*]+)\\*\\*/g, '$1')\n    .replace(/\\*([^*]+)\\*/g, '$1')\n    .replace(/_([^_]+)_/g, '$1')\n    .replace(/~~([^~]+)~~/g, '$1')\n    .replace(/!?\\[[^\\]]*\\]\\([^)]*\\)/g, ' ')\n    .replace(/^\\s*#{1,6}\\s+/gm, '')\n    .replace(/^\\s*>\\s?/gm, '')\n    .replace(/^\\s*[-*+]\\s+/gm, '')\n    .replace(/^\\s*\\d+\\.\\s+/gm, '');\n};\n\nconst collapseWhitespace = (text: string): string =>\n  text.replace(/\\s+/g, ' ').trim();\n\nconst stripLettersKeepDigitsAndSymbols = (text: string): string => {\n  // Keep digits and non-letter characters, remove all letters (including accents)\n  return text.replace(/\\p{L}+/gu, '');\n};\n\nexport const normalizeBlock = (block: Block): NormalizedBlock => {\n  const contentWithoutMarkdown = removeMarkdownFormatting(block.content);\n  const semanticLowercased = contentWithoutMarkdown.toLowerCase();\n  const semanticCollapsed = collapseWhitespace(semanticLowercased);\n\n  const anchorOnlySymbols = stripLettersKeepDigitsAndSymbols(block.content);\n  const anchorCollapsed = collapseWhitespace(anchorOnlySymbols);\n\n  return {\n    ...block,\n    semanticText: semanticCollapsed,\n    anchorText: anchorCollapsed,\n  };\n};\n"],"mappings":";AAEA,MAAM,4BAA4B,SAAyB;AACzD,QAAO,KACJ,QAAQ,sBAAsB,IAAI,CAClC,QAAQ,oBAAoB,KAAK,CACjC,QAAQ,gBAAgB,KAAK,CAC7B,QAAQ,cAAc,KAAK,CAC3B,QAAQ,gBAAgB,KAAK,CAC7B,QAAQ,0BAA0B,IAAI,CACtC,QAAQ,mBAAmB,GAAG,CAC9B,QAAQ,cAAc,GAAG,CACzB,QAAQ,kBAAkB,GAAG,CAC7B,QAAQ,kBAAkB,GAAG;;AAGlC,MAAM,sBAAsB,SAC1B,KAAK,QAAQ,QAAQ,IAAI,CAAC,MAAM;AAElC,MAAM,oCAAoC,SAAyB;AAEjE,QAAO,KAAK,QAAQ,YAAY,GAAG;;AAGrC,MAAa,kBAAkB,UAAkC;CAG/D,MAAM,oBAAoB,mBAFK,yBAAyB,MAAM,QAAQ,CACpB,aAAa,CACC;CAGhE,MAAM,kBAAkB,mBADE,iCAAiC,MAAM,QAAQ,CACZ;AAE7D,QAAO;EACL,GAAG;EACH,cAAc;EACd,YAAY;EACb"}
+{"version":3,"file":"normalizeBlock.mjs","names":[],"sources":["../../../src/translation-alignment/normalizeBlock.ts"],"sourcesContent":["import type { Block, NormalizedBlock } from './types';\n\nconst removeMarkdownFormatting = (text: string): string => {\n  return text\n    .replace(/`{1,3}[^`]*`{1,3}/g, ' ')\n    .replace(/\\*\\*([^*]+)\\*\\*/g, '$1')\n    .replace(/\\*([^*]+)\\*/g, '$1')\n    .replace(/_([^_]+)_/g, '$1')\n    .replace(/~~([^~]+)~~/g, '$1')\n    .replace(/!?\\[[^\\]]*\\]\\([^)]*\\)/g, ' ')\n    .replace(/^\\s*#{1,6}\\s+/gm, '')\n    .replace(/^\\s*>\\s?/gm, '')\n    .replace(/^\\s*[-*+]\\s+/gm, '')\n    .replace(/^\\s*\\d+\\.\\s+/gm, '');\n};\n\nconst collapseWhitespace = (text: string): string =>\n  text.replace(/\\s+/g, ' ').trim();\n\nconst stripLettersKeepDigitsAndSymbols = (text: string): string => {\n  // Keep digits and non-letter characters, remove all letters (including accents)\n  return text.replace(/\\p{L}+/gu, '');\n};\n\nexport const normalizeBlock = (block: Block): NormalizedBlock => {\n  const contentWithoutMarkdown = removeMarkdownFormatting(block.content);\n  const semanticLowercased = contentWithoutMarkdown.toLowerCase();\n  const semanticCollapsed = collapseWhitespace(semanticLowercased);\n\n  const anchorOnlySymbols = stripLettersKeepDigitsAndSymbols(block.content);\n  const anchorCollapsed = collapseWhitespace(anchorOnlySymbols);\n\n  return {\n    ...block,\n    semanticText: semanticCollapsed,\n    anchorText: anchorCollapsed,\n  };\n};\n"],"mappings":"AAEA,MAAM,EAA4B,GACzB,EACJ,QAAQ,qBAAsB,IAAI,CAClC,QAAQ,mBAAoB,KAAK,CACjC,QAAQ,eAAgB,KAAK,CAC7B,QAAQ,aAAc,KAAK,CAC3B,QAAQ,eAAgB,KAAK,CAC7B,QAAQ,yBAA0B,IAAI,CACtC,QAAQ,kBAAmB,GAAG,CAC9B,QAAQ,aAAc,GAAG,CACzB,QAAQ,iBAAkB,GAAG,CAC7B,QAAQ,iBAAkB,GAAG,CAG5B,EAAsB,GAC1B,EAAK,QAAQ,OAAQ,IAAI,CAAC,MAAM,CAE5B,EAAoC,GAEjC,EAAK,QAAQ,WAAY,GAAG,CAGxB,EAAkB,GAAkC,CAG/D,IAAM,EAAoB,EAFK,EAAyB,EAAM,QAAQ,CACpB,aAAa,CACC,CAG1D,EAAkB,EADE,EAAiC,EAAM,QAAQ,CACZ,CAE7D,MAAO,CACL,GAAG,EACH,aAAc,EACd,WAAY,EACb"}

package/dist/esm/translation-alignment/pipeline.mjs

@@ -1,36 +1,2 @@
-import { alignEnglishAndFrenchBlocks } from "./alignBlocks.mjs";
-import { fingerprintBlock } from "./fingerprintBlock.mjs";
-import { mapChangedLinesToBlocks } from "./mapChangedLinesToBlocks.mjs";
-import { normalizeBlock } from "./normalizeBlock.mjs";
-import { planAlignmentActions } from "./planActions.mjs";
-import { identifySegmentsToReview, mergeReviewedSegments } from "./rebuildDocument.mjs";
-import { segmentDocument } from "./segmentDocument.mjs";
-
-//#region src/translation-alignment/pipeline.ts
-const buildAlignmentPlan = ({ englishText, frenchText, changedLines, similarityOptions }) => {
-	const englishBlocksRaw = segmentDocument(englishText);
-	const frenchBlocksRaw = segmentDocument(frenchText);
-	const englishNormalized = englishBlocksRaw.map(normalizeBlock);
-	const frenchNormalized = frenchBlocksRaw.map(normalizeBlock);
-	const englishBlocks = englishNormalized.map((block, index, array) => fingerprintBlock(block, array[index - 1] ?? null, array[index + 1] ?? null));
-	const frenchBlocks = frenchNormalized.map((block, index, array) => fingerprintBlock(block, array[index - 1] ?? null, array[index + 1] ?? null));
-	const plan = planAlignmentActions(alignEnglishAndFrenchBlocks(englishBlocks, frenchBlocks), mapChangedLinesToBlocks(englishBlocks, Array.isArray(changedLines) ? changedLines : []), {
-		minimumMatchForReuse: similarityOptions?.minimumMatchForReuse ?? .9,
-		minimumMatchForNearDuplicate: similarityOptions?.minimumMatchForNearDuplicate ?? .8
-	});
-	const { segmentsToReview } = identifySegmentsToReview({
-		englishBlocks,
-		frenchBlocks,
-		plan
-	});
-	return {
-		englishBlocks,
-		frenchBlocks,
-		plan,
-		segmentsToReview
-	};
-};
-
-//#endregion
-export { buildAlignmentPlan, mergeReviewedSegments };
+import{alignEnglishAndFrenchBlocks as e}from"./alignBlocks.mjs";import{fingerprintBlock as t}from"./fingerprintBlock.mjs";import{mapChangedLinesToBlocks as n}from"./mapChangedLinesToBlocks.mjs";import{normalizeBlock as r}from"./normalizeBlock.mjs";import{planAlignmentActions as i}from"./planActions.mjs";import{identifySegmentsToReview as a,mergeReviewedSegments as o}from"./rebuildDocument.mjs";import{segmentDocument as s}from"./segmentDocument.mjs";const c=({englishText:o,frenchText:c,changedLines:l,similarityOptions:u})=>{let d=s(o),f=s(c),p=d.map(r),m=f.map(r),h=p.map((e,n,r)=>t(e,r[n-1]??null,r[n+1]??null)),g=m.map((e,n,r)=>t(e,r[n-1]??null,r[n+1]??null)),_=i(e(h,g),n(h,Array.isArray(l)?l:[]),{minimumMatchForReuse:u?.minimumMatchForReuse??.9,minimumMatchForNearDuplicate:u?.minimumMatchForNearDuplicate??.8}),{segmentsToReview:v}=a({englishBlocks:h,frenchBlocks:g,plan:_});return{englishBlocks:h,frenchBlocks:g,plan:_,segmentsToReview:v}};export{c as buildAlignmentPlan,o as mergeReviewedSegments};
 //# sourceMappingURL=pipeline.mjs.map

package/dist/esm/translation-alignment/pipeline.mjs.map

@@ -1 +1 @@
-{"version":3,"file":"pipeline.mjs","names":[],"sources":["../../../src/translation-alignment/pipeline.ts"],"sourcesContent":["import { alignEnglishAndFrenchBlocks } from './alignBlocks';\nimport { fingerprintBlock } from './fingerprintBlock';\nimport { mapChangedLinesToBlocks } from './mapChangedLinesToBlocks';\nimport { normalizeBlock } from './normalizeBlock';\nimport { planAlignmentActions } from './planActions';\nimport {\n  identifySegmentsToReview,\n  mergeReviewedSegments,\n  type SegmentToReview,\n} from './rebuildDocument';\nimport { segmentDocument } from './segmentDocument';\nimport type {\n  AlignmentPlan,\n  FingerprintedBlock,\n  SimilarityOptions,\n} from './types';\n\nexport type BuildAlignmentPlanInput = {\n  englishText: string;\n  frenchText: string;\n  changedLines: number[] | undefined;\n  similarityOptions?: Partial<SimilarityOptions>;\n};\n\nexport type BuildAlignmentPlanOutput = {\n  englishBlocks: FingerprintedBlock[];\n  frenchBlocks: FingerprintedBlock[];\n  plan: AlignmentPlan;\n  segmentsToReview: SegmentToReview[];\n};\n\nexport const buildAlignmentPlan = ({\n  englishText,\n  frenchText,\n  changedLines,\n  similarityOptions,\n}: BuildAlignmentPlanInput): BuildAlignmentPlanOutput => {\n  const englishBlocksRaw = segmentDocument(englishText);\n  const frenchBlocksRaw = segmentDocument(frenchText);\n\n  const englishNormalized = englishBlocksRaw.map(normalizeBlock);\n  const frenchNormalized = frenchBlocksRaw.map(normalizeBlock);\n\n  const englishBlocks: FingerprintedBlock[] = englishNormalized.map(\n    (block, index, array) =>\n      fingerprintBlock(\n        block,\n        array[index - 1] ?? null,\n        array[index + 1] ?? null\n      )\n  );\n  const frenchBlocks: FingerprintedBlock[] = frenchNormalized.map(\n    (block, index, array) =>\n      fingerprintBlock(\n        block,\n        array[index - 1] ?? null,\n        array[index + 1] ?? null\n      )\n  );\n\n  const alignment = alignEnglishAndFrenchBlocks(englishBlocks, frenchBlocks);\n\n  const changedIndexes = mapChangedLinesToBlocks(\n    englishBlocks,\n    Array.isArray(changedLines) ? changedLines : []\n  );\n\n  const plan = planAlignmentActions(alignment, changedIndexes, {\n    minimumMatchForReuse: similarityOptions?.minimumMatchForReuse ?? 0.9,\n    minimumMatchForNearDuplicate:\n      similarityOptions?.minimumMatchForNearDuplicate ?? 0.8,\n  });\n\n  const { segmentsToReview } = identifySegmentsToReview({\n    englishBlocks,\n    frenchBlocks,\n    plan,\n  });\n\n  return { englishBlocks, frenchBlocks, plan, segmentsToReview };\n};\n\nexport { mergeReviewedSegments };\nexport type { SegmentToReview };\n"],"mappings":";;;;;;;;;AA+BA,MAAa,sBAAsB,EACjC,aACA,YACA,cACA,wBACuD;CACvD,MAAM,mBAAmB,gBAAgB,YAAY;CACrD,MAAM,kBAAkB,gBAAgB,WAAW;CAEnD,MAAM,oBAAoB,iBAAiB,IAAI,eAAe;CAC9D,MAAM,mBAAmB,gBAAgB,IAAI,eAAe;CAE5D,MAAM,gBAAsC,kBAAkB,KAC3D,OAAO,OAAO,UACb,iBACE,OACA,MAAM,QAAQ,MAAM,MACpB,MAAM,QAAQ,MAAM,KACrB,CACJ;CACD,MAAM,eAAqC,iBAAiB,KACzD,OAAO,OAAO,UACb,iBACE,OACA,MAAM,QAAQ,MAAM,MACpB,MAAM,QAAQ,MAAM,KACrB,CACJ;CASD,MAAM,OAAO,qBAPK,4BAA4B,eAAe,aAAa,EAEnD,wBACrB,eACA,MAAM,QAAQ,aAAa,GAAG,eAAe,EAAE,CAChD,EAE4D;EAC3D,sBAAsB,mBAAmB,wBAAwB;EACjE,8BACE,mBAAmB,gCAAgC;EACtD,CAAC;CAEF,MAAM,EAAE,qBAAqB,yBAAyB;EACpD;EACA;EACA;EACD,CAAC;AAEF,QAAO;EAAE;EAAe;EAAc;EAAM;EAAkB"}
+{"version":3,"file":"pipeline.mjs","names":[],"sources":["../../../src/translation-alignment/pipeline.ts"],"sourcesContent":["import { alignEnglishAndFrenchBlocks } from './alignBlocks';\nimport { fingerprintBlock } from './fingerprintBlock';\nimport { mapChangedLinesToBlocks } from './mapChangedLinesToBlocks';\nimport { normalizeBlock } from './normalizeBlock';\nimport { planAlignmentActions } from './planActions';\nimport {\n  identifySegmentsToReview,\n  mergeReviewedSegments,\n  type SegmentToReview,\n} from './rebuildDocument';\nimport { segmentDocument } from './segmentDocument';\nimport type {\n  AlignmentPlan,\n  FingerprintedBlock,\n  SimilarityOptions,\n} from './types';\n\nexport type BuildAlignmentPlanInput = {\n  englishText: string;\n  frenchText: string;\n  changedLines: number[] | undefined;\n  similarityOptions?: Partial<SimilarityOptions>;\n};\n\nexport type BuildAlignmentPlanOutput = {\n  englishBlocks: FingerprintedBlock[];\n  frenchBlocks: FingerprintedBlock[];\n  plan: AlignmentPlan;\n  segmentsToReview: SegmentToReview[];\n};\n\nexport const buildAlignmentPlan = ({\n  englishText,\n  frenchText,\n  changedLines,\n  similarityOptions,\n}: BuildAlignmentPlanInput): BuildAlignmentPlanOutput => {\n  const englishBlocksRaw = segmentDocument(englishText);\n  const frenchBlocksRaw = segmentDocument(frenchText);\n\n  const englishNormalized = englishBlocksRaw.map(normalizeBlock);\n  const frenchNormalized = frenchBlocksRaw.map(normalizeBlock);\n\n  const englishBlocks: FingerprintedBlock[] = englishNormalized.map(\n    (block, index, array) =>\n      fingerprintBlock(\n        block,\n        array[index - 1] ?? null,\n        array[index + 1] ?? null\n      )\n  );\n  const frenchBlocks: FingerprintedBlock[] = frenchNormalized.map(\n    (block, index, array) =>\n      fingerprintBlock(\n        block,\n        array[index - 1] ?? null,\n        array[index + 1] ?? null\n      )\n  );\n\n  const alignment = alignEnglishAndFrenchBlocks(englishBlocks, frenchBlocks);\n\n  const changedIndexes = mapChangedLinesToBlocks(\n    englishBlocks,\n    Array.isArray(changedLines) ? changedLines : []\n  );\n\n  const plan = planAlignmentActions(alignment, changedIndexes, {\n    minimumMatchForReuse: similarityOptions?.minimumMatchForReuse ?? 0.9,\n    minimumMatchForNearDuplicate:\n      similarityOptions?.minimumMatchForNearDuplicate ?? 0.8,\n  });\n\n  const { segmentsToReview } = identifySegmentsToReview({\n    englishBlocks,\n    frenchBlocks,\n    plan,\n  });\n\n  return { englishBlocks, frenchBlocks, plan, segmentsToReview };\n};\n\nexport { mergeReviewedSegments };\nexport type { SegmentToReview };\n"],"mappings":"qcA+BA,MAAa,GAAsB,CACjC,cACA,aACA,eACA,uBACuD,CACvD,IAAM,EAAmB,EAAgB,EAAY,CAC/C,EAAkB,EAAgB,EAAW,CAE7C,EAAoB,EAAiB,IAAI,EAAe,CACxD,EAAmB,EAAgB,IAAI,EAAe,CAEtD,EAAsC,EAAkB,KAC3D,EAAO,EAAO,IACb,EACE,EACA,EAAM,EAAQ,IAAM,KACpB,EAAM,EAAQ,IAAM,KACrB,CACJ,CACK,EAAqC,EAAiB,KACzD,EAAO,EAAO,IACb,EACE,EACA,EAAM,EAAQ,IAAM,KACpB,EAAM,EAAQ,IAAM,KACrB,CACJ,CASK,EAAO,EAPK,EAA4B,EAAe,EAAa,CAEnD,EACrB,EACA,MAAM,QAAQ,EAAa,CAAG,EAAe,EAAE,CAChD,CAE4D,CAC3D,qBAAsB,GAAmB,sBAAwB,GACjE,6BACE,GAAmB,8BAAgC,GACtD,CAAC,CAEI,CAAE,oBAAqB,EAAyB,CACpD,gBACA,eACA,OACD,CAAC,CAEF,MAAO,CAAE,gBAAe,eAAc,OAAM,mBAAkB"}