botrun-horse 2.29.2 → 2.30.0

package/botrun-c/.claude/skills/PDF/346/212/200/350/203/275/scripts/pdf-to-text.mjs

@@ -0,0 +1,336 @@
+#!/usr/bin/env node
+/**
+ * PDF 轉文字腳本 - 使用 pdftotext 將 PDF 轉換為純文字
+ *
+ * Usage: node pdf-to-text.mjs "file.pdf" [--output <path>]
+ *        node pdf-to-text.mjs "pages_dir/" [--batch]
+ *
+ * 模式：
+ * - 單一檔案：轉換單一 PDF 為 TXT
+ * - 批次模式：轉換目錄下所有 PDF 為 TXT（保留檔名）
+ *
+ * 特性：
+ * - 使用 pdftotext -layout 保留排版
+ * - 支援批次處理（適用於切割後的頁面）
+ * - UTF-8 編碼輸出
+ */
+
+import { execSync } from 'child_process';
+import { existsSync, statSync, readdirSync, mkdirSync } from 'fs';
+import { basename, dirname, resolve, isAbsolute, join, extname } from 'path';
+
+// ============================================================================
+// 配置
+// ============================================================================
+
+const WORKING_DIR = process.env.WORKING_DIR || process.cwd();
+
+// ============================================================================
+// 日誌
+// ============================================================================
+
+function log(...args) {
+  console.error('[PDF-ToText]', ...args);
+}
+
+// ============================================================================
+// 執行命令
+// ============================================================================
+
+function execCommand(command, timeout = 60000) {
+  try {
+    const result = execSync(command, {
+      encoding: 'utf-8',
+      timeout,
+      stdio: ['pipe', 'pipe', 'pipe']
+    });
+    return { success: true, output: result.trim() };
+  } catch (error) {
+    return { success: false, error: error.message, output: '' };
+  }
+}
+
+// ============================================================================
+// 路徑驗證
+// ============================================================================
+
+function validatePath(filePath) {
+  const absolutePath = isAbsolute(filePath) ? filePath : resolve(WORKING_DIR, filePath);
+
+  if (!existsSync(absolutePath)) {
+    return { valid: false, error: `路徑不存在: ${filePath}` };
+  }
+
+  const stats = statSync(absolutePath);
+
+  return {
+    valid: true,
+    absolutePath,
+    isFile: stats.isFile(),
+    isDirectory: stats.isDirectory(),
+    sizeBytes: stats.size
+  };
+}
+
+// ============================================================================
+// 單檔轉換
+// ============================================================================
+
+function convertSinglePdf(pdfPath, outputPath) {
+  // pdftotext 選項：
+  // -layout: 保留排版
+  // -enc UTF-8: UTF-8 編碼
+  // -nopgbrk: 不加分頁符號
+  const command = `pdftotext -layout -enc UTF-8 -nopgbrk "${pdfPath}" "${outputPath}"`;
+
+  const result = execCommand(command);
+
+  if (!result.success) {
+    return {
+      success: false,
+      error: `pdftotext 執行失敗: ${result.error}`
+    };
+  }
+
+  // 檢查輸出檔案
+  if (!existsSync(outputPath)) {
+    return {
+      success: false,
+      error: '轉換後檔案不存在'
+    };
+  }
+
+  const outputStats = statSync(outputPath);
+
+  return {
+    success: true,
+    outputPath,
+    outputSize: outputStats.size
+  };
+}
+
+// ============================================================================
+// 批次轉換
+// ============================================================================
+
+async function batchConvert(inputDir, options = {}) {
+  const { outputDir: customOutputDir } = options;
+
+  log(`批次轉換目錄: ${inputDir}`);
+
+  // 驗證路徑
+  const validation = validatePath(inputDir);
+  if (!validation.valid) {
+    return { success: false, error: validation.error };
+  }
+
+  if (!validation.isDirectory) {
+    return { success: false, error: `不是目錄: ${inputDir}` };
+  }
+
+  const absoluteInputDir = validation.absolutePath;
+
+  // 找出所有 PDF 檔案
+  const pdfFiles = readdirSync(absoluteInputDir)
+    .filter(f => f.toLowerCase().endsWith('.pdf'))
+    .sort();
+
+  if (pdfFiles.length === 0) {
+    return {
+      success: false,
+      error: '目錄中沒有 PDF 檔案'
+    };
+  }
+
+  log(`找到 ${pdfFiles.length} 個 PDF 檔案`);
+
+  // 決定輸出目錄（預設同目錄）
+  const outputDir = customOutputDir || absoluteInputDir;
+  if (!existsSync(outputDir)) {
+    mkdirSync(outputDir, { recursive: true });
+  }
+
+  // 批次轉換
+  const results = [];
+  let successCount = 0;
+  let failCount = 0;
+  const startTime = Date.now();
+
+  for (let i = 0; i < pdfFiles.length; i++) {
+    const pdfFile = pdfFiles[i];
+    const pdfPath = join(absoluteInputDir, pdfFile);
+    const txtFile = pdfFile.replace(/\.pdf$/i, '.txt');
+    const txtPath = join(outputDir, txtFile);
+
+    const result = convertSinglePdf(pdfPath, txtPath);
+
+    if (result.success) {
+      successCount++;
+      results.push({
+        input: pdfFile,
+        output: txtFile,
+        success: true,
+        size: result.outputSize
+      });
+    } else {
+      failCount++;
+      results.push({
+        input: pdfFile,
+        output: txtFile,
+        success: false,
+        error: result.error
+      });
+    }
+
+    // 進度報告（每 10 個或最後一個）
+    if ((i + 1) % 10 === 0 || i === pdfFiles.length - 1) {
+      log(`進度: ${i + 1}/${pdfFiles.length}`);
+    }
+  }
+
+  const duration = Date.now() - startTime;
+
+  // 計算總輸出大小
+  const totalOutputSize = results
+    .filter(r => r.success)
+    .reduce((sum, r) => sum + r.size, 0);
+
+  log(`批次轉換完成: ${successCount} 成功, ${failCount} 失敗`);
+
+  return {
+    success: true,
+    inputDir: absoluteInputDir,
+    outputDir,
+    totalFiles: pdfFiles.length,
+    successCount,
+    failCount,
+    totalOutputSize,
+    totalOutputSizeMB: Math.round(totalOutputSize / 1024 / 1024 * 100) / 100,
+    duration,
+    files: results
+  };
+}
+
+// ============================================================================
+// 單檔模式
+// ============================================================================
+
+async function convertPdf(pdfPath, options = {}) {
+  const { outputPath: customOutputPath } = options;
+
+  log(`轉換中: ${pdfPath}`);
+
+  // 驗證路徑
+  const validation = validatePath(pdfPath);
+  if (!validation.valid) {
+    return { success: false, error: validation.error };
+  }
+
+  // 如果是目錄，走批次模式
+  if (validation.isDirectory) {
+    return batchConvert(pdfPath, options);
+  }
+
+  const { absolutePath, sizeBytes } = validation;
+
+  // 檢查是否為 PDF
+  if (!absolutePath.toLowerCase().endsWith('.pdf')) {
+    return { success: false, error: `不是 PDF 檔案: ${pdfPath}` };
+  }
+
+  // 檢查 pdftotext 是否可用
+  const pdftotextCheck = execCommand('which pdftotext 2>/dev/null');
+  if (!pdftotextCheck.success || !pdftotextCheck.output) {
+    return {
+      success: false,
+      error: '未安裝 pdftotext，請先安裝：brew install poppler'
+    };
+  }
+
+  // 決定輸出路徑
+  const fileName = basename(absolutePath, '.pdf');
+  const fileDir = dirname(absolutePath);
+  const outputPath = customOutputPath || join(fileDir, `${fileName}.txt`);
+
+  // 執行轉換
+  const startTime = Date.now();
+  const result = convertSinglePdf(absolutePath, outputPath);
+  const duration = Date.now() - startTime;
+
+  if (!result.success) {
+    return result;
+  }
+
+  log(`轉換完成: ${outputPath}`);
+
+  return {
+    success: true,
+    inputFile: absolutePath,
+    inputSize: sizeBytes,
+    inputSizeMB: Math.round(sizeBytes / 1024 / 1024 * 100) / 100,
+    outputFile: outputPath,
+    outputSize: result.outputSize,
+    outputSizeKB: Math.round(result.outputSize / 1024 * 100) / 100,
+    duration
+  };
+}
+
+// ============================================================================
+// Main
+// ============================================================================
+
+async function main() {
+  const args = process.argv.slice(2);
+
+  if (args.length === 0) {
+    console.error('Usage: node pdf-to-text.mjs "file.pdf" [--output <path>]');
+    console.error('       node pdf-to-text.mjs "pages_dir/" [--output-dir <dir>]');
+    console.error('');
+    console.error('將 PDF 轉換為純文字檔案');
+    console.error('');
+    console.error('單檔模式：');
+    console.error('  node pdf-to-text.mjs document.pdf');
+    console.error('  → 輸出：document.txt');
+    console.error('');
+    console.error('批次模式：');
+    console.error('  node pdf-to-text.mjs document_pages/');
+    console.error('  → 輸出：page-001.txt, page-002.txt, ...');
+    console.error('');
+    console.error('選項：');
+    console.error('  --output <path>      單檔輸出路徑');
+    console.error('  --output-dir <dir>   批次輸出目錄');
+    process.exit(1);
+  }
+
+  // 解析參數
+  let inputPath = null;
+  let outputPath = null;
+  let outputDir = null;
+
+  for (let i = 0; i < args.length; i++) {
+    if (args[i] === '--output' && args[i + 1]) {
+      outputPath = args[++i];
+    } else if (args[i] === '--output-dir' && args[i + 1]) {
+      outputDir = args[++i];
+    } else if (!args[i].startsWith('--')) {
+      inputPath = args[i];
+    }
+  }
+
+  if (!inputPath) {
+    console.error('錯誤：請提供 PDF 檔案或目錄路徑');
+    process.exit(1);
+  }
+
+  const options = outputPath ? { outputPath } : outputDir ? { outputDir } : {};
+  const result = await convertPdf(inputPath, options);
+
+  // 輸出 JSON 結果（stdout）
+  console.log(JSON.stringify(result, null, 2));
+
+  if (!result.success) {
+    process.exit(1);
+  }
+}
+
+main();