fix: resolve lint issues (XSS K601, knip unused exports, eslint errors)
- Fix XSS warnings in user.tsx by adding safe attribute to elements - Remove unused exports: ChunkUploadRequest, getArchiveInfo, getFileForDirectDownload, createDirectDownloadHeaders, createConverterArchive - Fix eslint no-unused-vars errors across multiple files - Fix pdfmathtranslate async Promise executor issue - Update tests to use toThrow instead of toMatch for Error objects - Apply prettier formatting
This commit is contained in:
Your Name
parent
db5f47586e
commit
79fc6f7067
21 changed files with 439 additions and 439 deletions
|
|
@ -26,7 +26,10 @@ import { convert as convertVcf, properties as propertiesVcf } from "./vcf";
|
|||
import { convert as convertxelatex, properties as propertiesxelatex } from "./xelatex";
|
||||
import { convert as convertMarkitdown, properties as propertiesMarkitdown } from "./markitdown";
|
||||
import { convert as convertMineru, properties as propertiesMineru } from "./mineru";
|
||||
import { convert as convertPDFMathTranslate, properties as propertiesPDFMathTranslate } from "./pdfmathtranslate";
|
||||
import {
|
||||
convert as convertPDFMathTranslate,
|
||||
properties as propertiesPDFMathTranslate,
|
||||
} from "./pdfmathtranslate";
|
||||
|
||||
// This should probably be reconstructed so that the functions are not imported instead the functions hook into this to make the converters more modular
|
||||
|
||||
|
|
|
|||
|
|
@ -2,7 +2,7 @@ import { execFile as execFileOriginal } from "node:child_process";
|
|||
import { mkdirSync, existsSync, readdirSync, unlinkSync, rmdirSync } from "node:fs";
|
||||
import { join, basename, dirname } from "node:path";
|
||||
import { ExecFileFn } from "./types";
|
||||
import { createConverterArchive, getArchiveFileName } from "../transfer";
|
||||
import { getArchiveFileName } from "../transfer";
|
||||
|
||||
export const properties = {
|
||||
from: {
|
||||
|
|
@ -16,7 +16,7 @@ export const properties = {
|
|||
|
||||
/**
|
||||
* Helper function to create a .tar archive from a directory (no compression)
|
||||
*
|
||||
*
|
||||
* ⚠️ 重要：僅使用 .tar 格式，禁止 .tar.gz / .tgz / .zip
|
||||
*/
|
||||
function createTarArchive(
|
||||
|
|
@ -28,23 +28,19 @@ function createTarArchive(
|
|||
// Use tar command to create archive (without gzip compression)
|
||||
// tar -cf <output.tar> -C <sourceDir> .
|
||||
// 注意：使用 -cf 而非 -czf，避免 gzip 壓縮
|
||||
execFile(
|
||||
"tar",
|
||||
["-cf", outputTar, "-C", sourceDir, "."],
|
||||
(error, stdout, stderr) => {
|
||||
if (error) {
|
||||
reject(`tar error: ${error}`);
|
||||
return;
|
||||
}
|
||||
if (stdout) {
|
||||
console.log(`tar stdout: ${stdout}`);
|
||||
}
|
||||
if (stderr) {
|
||||
console.error(`tar stderr: ${stderr}`);
|
||||
}
|
||||
resolve();
|
||||
},
|
||||
);
|
||||
execFile("tar", ["-cf", outputTar, "-C", sourceDir, "."], (error, stdout, stderr) => {
|
||||
if (error) {
|
||||
reject(`tar error: ${error}`);
|
||||
return;
|
||||
}
|
||||
if (stdout) {
|
||||
console.log(`tar stdout: ${stdout}`);
|
||||
}
|
||||
if (stderr) {
|
||||
console.error(`tar stderr: ${stderr}`);
|
||||
}
|
||||
resolve();
|
||||
});
|
||||
});
|
||||
}
|
||||
|
||||
|
|
@ -75,11 +71,6 @@ export async function convert(
|
|||
execFile: ExecFileFn = execFileOriginal,
|
||||
): Promise<string> {
|
||||
return new Promise((resolve, reject) => {
|
||||
// Determine mode based on target format
|
||||
// md-t = table mode (tables as markdown)
|
||||
// md-i = image mode (tables as images)
|
||||
const tableMode = convertTo === "md-t" ? "latex" : "html";
|
||||
|
||||
// Create a temporary output directory for MinerU
|
||||
const outputDir = dirname(targetPath);
|
||||
const inputFileName = basename(filePath, `.${fileType}`);
|
||||
|
|
@ -92,14 +83,7 @@ export async function convert(
|
|||
|
||||
// Build MinerU command arguments
|
||||
// MinerU CLI: magic-pdf -p <input> -o <output_dir> -m auto
|
||||
const args = [
|
||||
"-p",
|
||||
filePath,
|
||||
"-o",
|
||||
mineruOutputDir,
|
||||
"-m",
|
||||
"auto",
|
||||
];
|
||||
const args = ["-p", filePath, "-o", mineruOutputDir, "-m", "auto"];
|
||||
|
||||
// Add table mode option if md-i (render tables as images)
|
||||
if (convertTo === "md-i") {
|
||||
|
|
|
|||
|
|
@ -6,46 +6,47 @@ import { getArchiveFileName } from "../transfer";
|
|||
|
||||
/**
|
||||
* PDFMathTranslate Content Engine
|
||||
*
|
||||
*
|
||||
* 用於翻譯 PDF 文件同時保留數學公式的引擎。
|
||||
*
|
||||
*
|
||||
* 使用 pdf2zh CLI 工具進行翻譯，支援多種目標語言。
|
||||
* 所有輸出一律打包為 .tar 檔案，包含：
|
||||
* - original.pdf（原始 PDF）
|
||||
* - translated-<lang>.pdf（翻譯後的 PDF）
|
||||
*
|
||||
*
|
||||
* 必須在 Docker build 階段預先下載所需模型，
|
||||
* 不允許在 runtime 隱式下載模型。
|
||||
*/
|
||||
|
||||
// 支援的目標語言列表
|
||||
const SUPPORTED_LANGUAGES = [
|
||||
"en", // English
|
||||
"zh", // Chinese (Simplified)
|
||||
"en", // English
|
||||
"zh", // Chinese (Simplified)
|
||||
"zh-TW", // Chinese (Traditional)
|
||||
"ja", // Japanese
|
||||
"ko", // Korean
|
||||
"de", // German
|
||||
"fr", // French
|
||||
"es", // Spanish
|
||||
"it", // Italian
|
||||
"pt", // Portuguese
|
||||
"ru", // Russian
|
||||
"ar", // Arabic
|
||||
"hi", // Hindi
|
||||
"vi", // Vietnamese
|
||||
"th", // Thai
|
||||
"ja", // Japanese
|
||||
"ko", // Korean
|
||||
"de", // German
|
||||
"fr", // French
|
||||
"es", // Spanish
|
||||
"it", // Italian
|
||||
"pt", // Portuguese
|
||||
"ru", // Russian
|
||||
"ar", // Arabic
|
||||
"hi", // Hindi
|
||||
"vi", // Vietnamese
|
||||
"th", // Thai
|
||||
] as const;
|
||||
|
||||
type SupportedLanguage = typeof SUPPORTED_LANGUAGES[number];
|
||||
|
||||
// 模型路徑（Docker 環境中）
|
||||
const MODELS_PATH = process.env.PDFMATHTRANSLATE_MODELS_PATH || "/models/pdfmathtranslate";
|
||||
|
||||
// 生成 from/to 格式映射
|
||||
function generateLanguageMappings(): { from: Record<string, string[]>; to: Record<string, string[]> } {
|
||||
const outputFormats = SUPPORTED_LANGUAGES.map(lang => `pdf-${lang}`);
|
||||
|
||||
function generateLanguageMappings(): {
|
||||
from: Record<string, string[]>;
|
||||
to: Record<string, string[]>;
|
||||
} {
|
||||
const outputFormats = SUPPORTED_LANGUAGES.map((lang) => `pdf-${lang}`);
|
||||
|
||||
return {
|
||||
from: {
|
||||
document: ["pdf"],
|
||||
|
|
@ -91,7 +92,7 @@ function checkModelsExist(): boolean {
|
|||
|
||||
/**
|
||||
* Helper function to create a .tar archive from a directory (no compression)
|
||||
*
|
||||
*
|
||||
* ⚠️ 重要：僅使用 .tar 格式，禁止 .tar.gz / .tgz / .zip
|
||||
*/
|
||||
function createTarArchive(
|
||||
|
|
@ -103,23 +104,19 @@ function createTarArchive(
|
|||
// Use tar command to create archive (without gzip compression)
|
||||
// tar -cf <output.tar> -C <sourceDir> .
|
||||
// 注意：使用 -cf 而非 -czf，避免 gzip 壓縮
|
||||
execFile(
|
||||
"tar",
|
||||
["-cf", outputTar, "-C", sourceDir, "."],
|
||||
(error, stdout, stderr) => {
|
||||
if (error) {
|
||||
reject(`tar error: ${error}`);
|
||||
return;
|
||||
}
|
||||
if (stdout) {
|
||||
console.log(`tar stdout: ${stdout}`);
|
||||
}
|
||||
if (stderr) {
|
||||
console.error(`tar stderr: ${stderr}`);
|
||||
}
|
||||
resolve();
|
||||
},
|
||||
);
|
||||
execFile("tar", ["-cf", outputTar, "-C", sourceDir, "."], (error, stdout, stderr) => {
|
||||
if (error) {
|
||||
reject(`tar error: ${error}`);
|
||||
return;
|
||||
}
|
||||
if (stdout) {
|
||||
console.log(`tar stdout: ${stdout}`);
|
||||
}
|
||||
if (stderr) {
|
||||
console.error(`tar stderr: ${stderr}`);
|
||||
}
|
||||
resolve();
|
||||
});
|
||||
});
|
||||
}
|
||||
|
||||
|
|
@ -143,7 +140,7 @@ function removeDir(dirPath: string): void {
|
|||
|
||||
/**
|
||||
* 執行 pdf2zh 命令進行 PDF 翻譯
|
||||
*
|
||||
*
|
||||
* @param inputPath 輸入 PDF 路徑
|
||||
* @param outputDir 輸出目錄
|
||||
* @param targetLang 目標語言
|
||||
|
|
@ -160,16 +157,19 @@ function runPdf2zh(
|
|||
// -lo <lang>: 目標語言
|
||||
// -o <dir>: 輸出目錄
|
||||
// -s google: 使用 Google 翻譯（預設，免費）
|
||||
//
|
||||
//
|
||||
// 輸出檔案：
|
||||
// - <filename>-mono.pdf: 純翻譯版本
|
||||
// - <filename>-dual.pdf: 雙語對照版本
|
||||
|
||||
|
||||
const args = [
|
||||
inputPath,
|
||||
"-lo", targetLang,
|
||||
"-o", outputDir,
|
||||
"-s", process.env.PDFMATHTRANSLATE_SERVICE || "google",
|
||||
"-lo",
|
||||
targetLang,
|
||||
"-o",
|
||||
outputDir,
|
||||
"-s",
|
||||
process.env.PDFMATHTRANSLATE_SERVICE || "google",
|
||||
];
|
||||
|
||||
// 如果設定了自訂模型路徑，使用 --onnx 參數
|
||||
|
|
@ -205,9 +205,9 @@ function runPdf2zh(
|
|||
if (!existsSync(monoPath) && !existsSync(dualPath)) {
|
||||
// 嘗試在 outputDir 下尋找任何 PDF 檔案
|
||||
const files = readdirSync(outputDir);
|
||||
const pdfFiles = files.filter(f => f.endsWith(".pdf"));
|
||||
const pdfFiles = files.filter((f) => f.endsWith(".pdf"));
|
||||
console.log(`[PDFMathTranslate] Found PDF files: ${pdfFiles.join(", ")}`);
|
||||
|
||||
|
||||
if (pdfFiles.length === 0) {
|
||||
reject(`No output PDF files found in ${outputDir}`);
|
||||
return;
|
||||
|
|
@ -221,7 +221,7 @@ function runPdf2zh(
|
|||
|
||||
/**
|
||||
* 主要轉換函數
|
||||
*
|
||||
*
|
||||
* @param filePath 輸入 PDF 檔案路徑
|
||||
* @param fileType 檔案類型（應為 "pdf"）
|
||||
* @param convertTo 目標格式（如 "pdf-zh"）
|
||||
|
|
@ -234,93 +234,91 @@ export async function convert(
|
|||
fileType: string,
|
||||
convertTo: string,
|
||||
targetPath: string,
|
||||
options?: unknown,
|
||||
_options?: unknown,
|
||||
execFile: ExecFileFn = execFileOriginal,
|
||||
): Promise<string> {
|
||||
return new Promise(async (resolve, reject) => {
|
||||
try {
|
||||
// 1. 檢查模型（警告但不阻止，因為 pdf2zh 可能會自動下載）
|
||||
checkModelsExist();
|
||||
try {
|
||||
// 1. 檢查模型（警告但不阻止，因為 pdf2zh 可能會自動下載）
|
||||
checkModelsExist();
|
||||
|
||||
// 2. 提取目標語言
|
||||
const targetLang = extractTargetLanguage(convertTo);
|
||||
console.log(`[PDFMathTranslate] Translating to: ${targetLang}`);
|
||||
// 2. 提取目標語言
|
||||
const targetLang = extractTargetLanguage(convertTo);
|
||||
console.log(`[PDFMathTranslate] Translating to: ${targetLang}`);
|
||||
|
||||
// 3. 建立臨時輸出目錄
|
||||
const outputDir = dirname(targetPath);
|
||||
const inputFileName = basename(filePath, `.${fileType}`);
|
||||
const tempDir = join(outputDir, `${inputFileName}_pdfmathtranslate_${Date.now()}`);
|
||||
// 3. 建立臨時輸出目錄
|
||||
const outputDir = dirname(targetPath);
|
||||
const inputFileName = basename(filePath, `.${fileType}`);
|
||||
const tempDir = join(outputDir, `${inputFileName}_pdfmathtranslate_${Date.now()}`);
|
||||
|
||||
if (!existsSync(tempDir)) {
|
||||
mkdirSync(tempDir, { recursive: true });
|
||||
}
|
||||
|
||||
// 4. 建立封裝用目錄
|
||||
const archiveDir = join(tempDir, "archive");
|
||||
mkdirSync(archiveDir, { recursive: true });
|
||||
|
||||
// 5. 執行 pdf2zh 翻譯
|
||||
const { monoPath, dualPath } = await runPdf2zh(filePath, tempDir, targetLang, execFile);
|
||||
|
||||
// 6. 複製原始檔案到封裝目錄
|
||||
const originalDest = join(archiveDir, "original.pdf");
|
||||
copyFileSync(filePath, originalDest);
|
||||
|
||||
// 7. 複製翻譯後的檔案（優先使用 mono，因為它是純翻譯版本）
|
||||
const translatedDest = join(archiveDir, `translated-${targetLang}.pdf`);
|
||||
|
||||
// 檢查各種可能的輸出檔案名稱
|
||||
const possibleOutputs = [
|
||||
monoPath,
|
||||
dualPath,
|
||||
join(tempDir, `${inputFileName}-mono.pdf`),
|
||||
join(tempDir, `${inputFileName}-dual.pdf`),
|
||||
];
|
||||
|
||||
let foundOutput = false;
|
||||
for (const outputPath of possibleOutputs) {
|
||||
if (existsSync(outputPath)) {
|
||||
copyFileSync(outputPath, translatedDest);
|
||||
foundOutput = true;
|
||||
console.log(`[PDFMathTranslate] Using output: ${outputPath}`);
|
||||
break;
|
||||
}
|
||||
}
|
||||
|
||||
// 如果找不到預期的輸出檔案，嘗試找任何 PDF
|
||||
if (!foundOutput) {
|
||||
const allFiles = readdirSync(tempDir);
|
||||
const pdfFiles = allFiles.filter(f => f.endsWith(".pdf") && f !== basename(filePath));
|
||||
const firstPdfName = pdfFiles[0];
|
||||
|
||||
if (firstPdfName) {
|
||||
const firstPdf = join(tempDir, firstPdfName);
|
||||
copyFileSync(firstPdf, translatedDest);
|
||||
foundOutput = true;
|
||||
console.log(`[PDFMathTranslate] Using fallback output: ${firstPdf}`);
|
||||
}
|
||||
}
|
||||
|
||||
if (!foundOutput) {
|
||||
throw new Error("No translated PDF output found");
|
||||
}
|
||||
|
||||
// 8. 建立 .tar 封裝
|
||||
const tarPath = getArchiveFileName(targetPath);
|
||||
const tarDir = dirname(tarPath);
|
||||
if (!existsSync(tarDir)) {
|
||||
mkdirSync(tarDir, { recursive: true });
|
||||
}
|
||||
|
||||
await createTarArchive(archiveDir, tarPath, execFile);
|
||||
console.log(`[PDFMathTranslate] Created archive: ${tarPath}`);
|
||||
|
||||
// 9. 清理臨時目錄
|
||||
removeDir(tempDir);
|
||||
|
||||
resolve("Done");
|
||||
} catch (error) {
|
||||
reject(`PDFMathTranslate error: ${error}`);
|
||||
if (!existsSync(tempDir)) {
|
||||
mkdirSync(tempDir, { recursive: true });
|
||||
}
|
||||
});
|
||||
|
||||
// 4. 建立封裝用目錄
|
||||
const archiveDir = join(tempDir, "archive");
|
||||
mkdirSync(archiveDir, { recursive: true });
|
||||
|
||||
// 5. 執行 pdf2zh 翻譯
|
||||
const { monoPath, dualPath } = await runPdf2zh(filePath, tempDir, targetLang, execFile);
|
||||
|
||||
// 6. 複製原始檔案到封裝目錄
|
||||
const originalDest = join(archiveDir, "original.pdf");
|
||||
copyFileSync(filePath, originalDest);
|
||||
|
||||
// 7. 複製翻譯後的檔案（優先使用 mono，因為它是純翻譯版本）
|
||||
const translatedDest = join(archiveDir, `translated-${targetLang}.pdf`);
|
||||
|
||||
// 檢查各種可能的輸出檔案名稱
|
||||
const possibleOutputs = [
|
||||
monoPath,
|
||||
dualPath,
|
||||
join(tempDir, `${inputFileName}-mono.pdf`),
|
||||
join(tempDir, `${inputFileName}-dual.pdf`),
|
||||
];
|
||||
|
||||
let foundOutput = false;
|
||||
for (const outputPath of possibleOutputs) {
|
||||
if (existsSync(outputPath)) {
|
||||
copyFileSync(outputPath, translatedDest);
|
||||
foundOutput = true;
|
||||
console.log(`[PDFMathTranslate] Using output: ${outputPath}`);
|
||||
break;
|
||||
}
|
||||
}
|
||||
|
||||
// 如果找不到預期的輸出檔案，嘗試找任何 PDF
|
||||
if (!foundOutput) {
|
||||
const allFiles = readdirSync(tempDir);
|
||||
const pdfFiles = allFiles.filter((f) => f.endsWith(".pdf") && f !== basename(filePath));
|
||||
const firstPdfName = pdfFiles[0];
|
||||
|
||||
if (firstPdfName) {
|
||||
const firstPdf = join(tempDir, firstPdfName);
|
||||
copyFileSync(firstPdf, translatedDest);
|
||||
foundOutput = true;
|
||||
console.log(`[PDFMathTranslate] Using fallback output: ${firstPdf}`);
|
||||
}
|
||||
}
|
||||
|
||||
if (!foundOutput) {
|
||||
throw new Error("No translated PDF output found");
|
||||
}
|
||||
|
||||
// 8. 建立 .tar 封裝
|
||||
const tarPath = getArchiveFileName(targetPath);
|
||||
const tarDir = dirname(tarPath);
|
||||
if (!existsSync(tarDir)) {
|
||||
mkdirSync(tarDir, { recursive: true });
|
||||
}
|
||||
|
||||
await createTarArchive(archiveDir, tarPath, execFile);
|
||||
console.log(`[PDFMathTranslate] Created archive: ${tarPath}`);
|
||||
|
||||
// 9. 清理臨時目錄
|
||||
removeDir(tempDir);
|
||||
|
||||
return "Done";
|
||||
} catch (error) {
|
||||
throw new Error(`PDFMathTranslate error: ${error}`);
|
||||
}
|
||||
}
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue