peam 0.1.1

package/dist/index.js

@@ -0,0 +1,250 @@
+'use strict';
+
+var parser = require('@peam-ai/parser');
+var search = require('@peam-ai/search');
+var chalk = require('chalk');
+var commander = require('commander');
+var fg = require('fast-glob');
+var fs = require('fs');
+var path = require('path');
+
+function _interopDefault (e) { return e && e.__esModule ? e : { default: e }; }
+
+var chalk__default = /*#__PURE__*/_interopDefault(chalk);
+var fg__default = /*#__PURE__*/_interopDefault(fg);
+
+var __require = /* @__PURE__ */ ((x) => typeof require !== "undefined" ? require : typeof Proxy !== "undefined" ? new Proxy(x, {
+  get: (a, b) => (typeof require !== "undefined" ? require : a)[b]
+}) : x)(function(x) {
+  if (typeof require !== "undefined") return require.apply(this, arguments);
+  throw Error('Dynamic require of "' + x + '" is not supported');
+});
+var packageJson = JSON.parse(fs.readFileSync(path.join(__dirname, "../package.json"), "utf-8"));
+function parseExcludePatterns(value) {
+  return value.split(",").map((p) => p.trim()).filter(Boolean);
+}
+var log = {
+  success: (message) => console.log(chalk__default.default.green("\u2713 ") + message),
+  error: (message) => console.error(chalk__default.default.red("\u2717 ") + message),
+  warn: (message) => console.warn(chalk__default.default.yellow("\u26A0 ") + message),
+  info: (message) => console.log(chalk__default.default.blue("\u2139 ") + message),
+  text: (message) => console.log(message),
+  cyan: (text) => chalk__default.default.cyan(text),
+  gray: (text) => chalk__default.default.gray(text),
+  bold: (text) => chalk__default.default.bold(text),
+  yellow: (text) => chalk__default.default.yellow(text),
+  red: (text) => chalk__default.default.red(text),
+  green: (text) => chalk__default.default.green(text)
+};
+async function discoverHtmlFiles(sourceDir, globPattern, projectDir) {
+  const pages = [];
+  try {
+    const sourceFullPath = path.join(projectDir, sourceDir);
+    const htmlFiles = await fg__default.default(globPattern, {
+      cwd: sourceFullPath,
+      absolute: true,
+      onlyFiles: true,
+      dot: false,
+      ignore: ["**/_next/**", "**/_astro/**", "**/404.{html,htm}", "**/500.{html,htm}"]
+    });
+    for (const htmlFilePath of htmlFiles) {
+      const relativePath = path.relative(sourceFullPath, htmlFilePath);
+      const pathname = parser.filePathToPathname(relativePath);
+      pages.push({
+        pathname,
+        htmlFilePath,
+        relativeHtmlPath: relativePath
+      });
+    }
+  } catch (error) {
+    log.warn(`Error discovering HTML files: ${error}`);
+  }
+  return pages;
+}
+async function indexPages(config) {
+  var _a;
+  log.text("\n" + log.bold(log.cyan("Peam Static Site Indexer")) + "\n");
+  log.text(log.bold("Configuration:"));
+  log.text(`  Project Directory: ${log.gray(config.projectDir)}`);
+  log.text(`  Source Directory: ${log.gray(config.source)}`);
+  log.text(`  Glob Pattern: ${log.gray(config.glob)}`);
+  log.text(`  Output Directory: ${log.gray(config.outputDir)}`);
+  log.text(`  Index Filename: ${log.gray(config.indexFilename)}`);
+  log.text(`  Respect robots.txt: ${log.gray(config.respectRobotsTxt.toString())}`);
+  if (config.exclude.length > 0) {
+    log.text(`  Exclude Patterns: ${log.gray(config.exclude.join(", "))}`);
+  }
+  log.text("");
+  const sourcePath = path.join(config.projectDir, config.source);
+  if (!fs.existsSync(sourcePath)) {
+    log.error(`Source directory not found: ${sourcePath}`);
+    log.text(log.yellow("   Please build your site first or specify the correct --source directory"));
+    process.exit(1);
+  }
+  const searchPaths = [path.join(config.source, "robots.txt"), "public/robots.txt", "robots.txt"];
+  const robotsResult = config.respectRobotsTxt ? parser.loadRobotsTxt(config.projectDir, searchPaths, config.robotsTxtPath) : null;
+  if (robotsResult && config.respectRobotsTxt) {
+    log.info(`robots.txt loaded from ${log.cyan(path.relative(config.projectDir, robotsResult.path))}`);
+  } else if (config.respectRobotsTxt) {
+    log.info("No robots.txt found, all paths will be indexed");
+  }
+  log.text(log.bold("Discovering HTML files..."));
+  log.text("");
+  log.text(`  Scanning: ${log.gray(config.source)}`);
+  log.text(`  Pattern: ${log.gray(config.glob)}`);
+  log.text("");
+  const discoveredPages = await discoverHtmlFiles(config.source, config.glob, config.projectDir);
+  if (discoveredPages.length === 0) {
+    log.text("");
+    log.warn("No HTML files found");
+    log.text(log.yellow("   Check that your source directory contains HTML files"));
+    log.text(log.yellow(`   Source: ${sourcePath}`));
+    log.text(log.yellow(`   Pattern: ${config.glob}`));
+    process.exit(1);
+  }
+  const uniquePages = /* @__PURE__ */ new Map();
+  for (const page of discoveredPages) {
+    if (!uniquePages.has(page.pathname)) {
+      uniquePages.set(page.pathname, page);
+    }
+  }
+  log.text("");
+  log.success(`Found ${log.bold(uniquePages.size.toString())} unique pages`);
+  log.text("");
+  log.text(log.bold("Processing pages..."));
+  log.text("");
+  const processedPages = [];
+  for (const [pathname, page] of uniquePages) {
+    const result = parser.shouldIncludePath(pathname, (_a = robotsResult == null ? void 0 : robotsResult.parser) != null ? _a : null, config.exclude, config.respectRobotsTxt);
+    if (!result.included) {
+      if (result.reason === "robots-txt") {
+        log.error(`Excluded by robots.txt: ${pathname}`);
+      } else if (result.reason === "exclude-pattern") {
+        log.error(`Excluded by pattern: ${pathname}`);
+      }
+      continue;
+    }
+    try {
+      const html = fs.readFileSync(page.htmlFilePath, "utf-8");
+      const structuredPage = parser.parseHTML(html);
+      if (!structuredPage) {
+        log.warn(`No content extracted from ${log.gray(pathname)}`);
+        continue;
+      }
+      log.success(`${log.cyan(pathname)}`);
+      processedPages.push({
+        path: pathname,
+        htmlFile: page.relativeHtmlPath,
+        structuredPage
+      });
+    } catch (error) {
+      log.error(`Error processing ${log.gray(pathname)}: ${error}`);
+    }
+  }
+  log.text("");
+  log.success(`Successfully processed ${log.bold(processedPages.length.toString())} pages`);
+  log.text("");
+  log.text(log.bold("Creating search index..."));
+  log.text("");
+  const searchIndexData = await search.buildSearchIndex(processedPages);
+  log.success(`Added ${log.bold(processedPages.length.toString())} pages to search index`);
+  log.text("");
+  log.text(log.bold("Saving index..."));
+  log.text("");
+  const outputPath = path.join(config.projectDir, config.outputDir);
+  fs.mkdirSync(outputPath, { recursive: true });
+  const generatedPath = path.join(outputPath, "generated");
+  fs.mkdirSync(generatedPath, { recursive: true });
+  const searchIndexFile = path.join(generatedPath, config.indexFilename);
+  fs.writeFileSync(searchIndexFile, JSON.stringify(searchIndexData));
+  const indexJsContent = `// Auto-generated by Peam - DO NOT EDIT THIS FILE
+import index from "./generated/${config.indexFilename}";
+export default index;
+`;
+  const indexJsFile = path.join(outputPath, "index.js");
+  fs.writeFileSync(indexJsFile, indexJsContent);
+  log.success(`Index saved to: ${log.cyan(path.relative(config.projectDir, searchIndexFile))}`);
+  log.text(`  Total pages indexed: ${log.bold(processedPages.length.toString())}`);
+  log.text(
+    `  Index size: ${log.bold((Buffer.byteLength(JSON.stringify(searchIndexData), "utf8") / 1024).toFixed(2))} KB`
+  );
+  log.text("");
+  log.success(log.bold("Indexing complete!"));
+  log.text("");
+}
+function probeSourceDirectory(projectDir) {
+  const commonDirs = [".next", ".build", ".out"];
+  for (const dir of commonDirs) {
+    const fullPath = path.join(projectDir, dir);
+    if (fs.existsSync(fullPath)) {
+      return dir;
+    }
+  }
+  return null;
+}
+async function main() {
+  const program = new commander.Command();
+  program.name("peam").description("Peam static site indexer (Next.js, Hugo, etc.)").version(packageJson.version).option("--source <path>", "Source directory containing HTML files (auto-detected if not provided)").option("--glob <pattern>", "Glob pattern for HTML files", "**/*.{html,htm}").option("--output-dir <path>", "Output directory for index", ".peam").option("--index-filename <name>", "Name of index file", "index.json").option("--ignore-robots-txt", "Disable robots.txt checking").option("--robots-path <path>", "Custom path to robots.txt file").option("--exclude <patterns>", "Comma-separated exclude patterns", parseExcludePatterns, []).option("--project-dir <path>", "Project root directory", process.cwd()).addHelpText(
+    "after",
+    `
+Examples:
+  # Auto-detect build directory
+  $ peam
+
+  # Hugo
+  $ peam --source public
+
+  # Next.js
+  $ peam --source .next
+
+  # Custom output directory
+  $ peam --source dist --glob "**/*.html"
+
+  # Exclude patterns
+  $ peam --exclude "/admin/**,/api/*,/private-*"
+
+For Next.js 15+, the @peam-ai/next integration is recommended for production use.
+
+More information: https://peam.ai
+    `
+  ).parse();
+  const options = program.opts();
+  try {
+    let sourceDir = options.source;
+    if (!sourceDir) {
+      const probedDir = probeSourceDirectory(options.projectDir);
+      if (probedDir) {
+        sourceDir = probedDir;
+        log.info(`Auto-detected source directory: ${log.cyan(probedDir)}`);
+        log.text("");
+      } else {
+        log.error("No build output directory found");
+        log.text(log.yellow("   Searched for: .next, .build, .out"));
+        log.text(log.yellow("   Please build your site first or specify --source <directory>"));
+        process.exit(1);
+      }
+    }
+    const config = {
+      source: sourceDir,
+      glob: options.glob,
+      outputDir: options.outputDir,
+      indexFilename: options.indexFilename,
+      respectRobotsTxt: !options.ignoreRobotsTxt,
+      robotsTxtPath: options.robotsPath,
+      exclude: options.exclude,
+      projectDir: options.projectDir
+    };
+    await indexPages(config);
+  } catch (error) {
+    log.text("");
+    log.error(`Fatal error: ${error}`);
+    process.exit(1);
+  }
+}
+if (__require.main === module) {
+  main();
+}
+
+exports.indexPages = indexPages;
+//# sourceMappingURL=index.js.map
+//# sourceMappingURL=index.js.map

package/dist/index.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/cli.ts"],"names":["readFileSync","join","chalk","fg","relative","filePathToPathname","existsSync","loadRobotsTxt","shouldIncludePath","parseHTML","buildSearchIndex","mkdirSync","writeFileSync","Command"],"mappings":";;;;;;;;;;;;;;;;;;;;;AAeA,IAAM,WAAA,GAAc,KAAK,KAAA,CAAMA,eAAA,CAAaC,UAAK,SAAA,EAAW,iBAAiB,CAAA,EAAG,OAAO,CAAC,CAAA;AAmBxF,SAAS,qBAAqB,KAAA,EAAyB;AACrD,EAAA,OAAO,KAAA,CACJ,KAAA,CAAM,GAAG,CAAA,CACT,GAAA,CAAI,CAAC,CAAA,KAAM,CAAA,CAAE,IAAA,EAAM,CAAA,CACnB,MAAA,CAAO,OAAO,CAAA;AACnB;AAEA,IAAM,GAAA,GAAM;AAAA,EACV,OAAA,EAAS,CAAC,OAAA,KAAoB,OAAA,CAAQ,IAAIC,sBAAA,CAAM,KAAA,CAAM,SAAI,CAAA,GAAI,OAAO,CAAA;AAAA,EACrE,KAAA,EAAO,CAAC,OAAA,KAAoB,OAAA,CAAQ,MAAMA,sBAAA,CAAM,GAAA,CAAI,SAAI,CAAA,GAAI,OAAO,CAAA;AAAA,EACnE,IAAA,EAAM,CAAC,OAAA,KAAoB,OAAA,CAAQ,KAAKA,sBAAA,CAAM,MAAA,CAAO,SAAI,CAAA,GAAI,OAAO,CAAA;AAAA,EACpE,IAAA,EAAM,CAAC,OAAA,KAAoB,OAAA,CAAQ,IAAIA,sBAAA,CAAM,IAAA,CAAK,SAAI,CAAA,GAAI,OAAO,CAAA;AAAA,EACjE,IAAA,EAAM,CAAC,OAAA,KAAoB,OAAA,CAAQ,IAAI,OAAO,CAAA;AAAA,EAE9C,IAAA,EAAM,CAAC,IAAA,KAAiBA,sBAAA,CAAM,KAAK,IAAI,CAAA;AAAA,EACvC,IAAA,EAAM,CAAC,IAAA,KAAiBA,sBAAA,CAAM,KAAK,IAAI,CAAA;AAAA,EACvC,IAAA,EAAM,CAAC,IAAA,KAAiBA,sBAAA,CAAM,KAAK,IAAI,CAAA;AAAA,EACvC,MAAA,EAAQ,CAAC,IAAA,KAAiBA,sBAAA,CAAM,OAAO,IAAI,CAAA;AAAA,EAC3C,GAAA,EAAK,CAAC,IAAA,KAAiBA,sBAAA,CAAM,IAAI,IAAI,CAAA;AAAA,EACrC,KAAA,EAAO,CAAC,IAAA,KAAiBA,sBAAA,CAAM,MAAM,IAAI;AAC3C,CAAA;AAEA,eAAe,iBAAA,CACb,SAAA,EACA,WAAA,EACA,UAAA,EAC2B;AAC3B,EAAA,MAAM,QAA0B,EAAC;AAEjC,EAAA,IAAI;AACF,IAAA,MAAM,cAAA,GAAiBD,SAAA,CAAK,UAAA,EAAY,SAAS,CAAA;AAEjD,IAAA,MAAM,SAAA,GAAY,MAAME,mBAAA,CAAG,WAAA,EAAa;AAAA,MACtC,GAAA,EAAK,cAAA;AAAA,MACL,QAAA,EAAU,IAAA;AAAA,MACV,SAAA,EAAW,IAAA;AAAA,MACX,GAAA,EAAK,KAAA;AAAA,MACL,MAAA,EAAQ,CAAC,aAAA,EAAe,cAAA,EAAgB,qBAAqB,mBAAmB;AAAA,KACjF,CAAA;AAED,IAAA,KAAA,MAAW,gBAAgB,SAAA,EAAW;AACpC,MAAA,MAAM,YAAA,GAAeC,aAAA,CAAS,cAAA,EAAgB,YAAY,CAAA;AAC1D,MAAA,MAAM,QAAA,GAAWC,0BAAmB,YAAY,CAAA;AAEhD,MAAA,KAAA,CAAM,IAAA,CAAK;AAAA,QACT,QAAA;AAAA,QACA,YAAA;AAAA,QACA,gBAAA,EAAkB;AAAA,OACnB,CAAA;AAAA,IACH;AAAA,EACF,SAAS,KAAA,EAAO;AACd,IAAA,GAAA,CAAI,IAAA,CAAK,CAAA,8BAAA,EAAiC,KAAK,CAAA,CAAE,CAAA;AAAA,EACnD;AAEA,EAAA,OAAO,KAAA;AACT;AAEA,eAAe,WAAW,MAAA,EAAsC;AA3FhE,EAAA,IAAA,EAAA;AA4FE,EAAA,GAAA,CAAI,IAAA,CAAK,OAAO,GAAA,CAAI,IAAA,CAAK,IAAI,IAAA,CAAK,0BAA0B,CAAC,CAAA,GAAI,IAAI,CAAA;AACrE,EAAA,GAAA,CAAI,IAAA,CAAK,GAAA,CAAI,IAAA,CAAK,gBAAgB,CAAC,CAAA;AACnC,EAAA,GAAA,CAAI,KAAK,CAAA,qBAAA,EAAwB,GAAA,CAAI,KAAK,MAAA,CAAO,UAAU,CAAC,CAAA,CAAE,CAAA;AAC9D,EAAA,GAAA,CAAI,KAAK,CAAA,oBAAA,EAAuB,GAAA,CAAI,KAAK,MAAA,CAAO,MAAM,CAAC,CAAA,CAAE,CAAA;AACzD,EAAA,GAAA,CAAI,KAAK,CAAA,gBAAA,EAAmB,GAAA,CAAI,KAAK,MAAA,CAAO,IAAI,CAAC,CAAA,CAAE,CAAA;AACnD,EAAA,GAAA,CAAI,KAAK,CAAA,oBAAA,EAAuB,GAAA,CAAI,KAAK,MAAA,CAAO,SAAS,CAAC,CAAA,CAAE,CAAA;AAC5D,EAAA,GAAA,CAAI,KAAK,CAAA,kBAAA,EAAqB,GAAA,CAAI,KAAK,MAAA,CAAO,aAAa,CAAC,CAAA,CAAE,CAAA;AAC9D,EAAA,GAAA,CAAI,IAAA,CAAK,yBAAyB,GAAA,CAAI,IAAA,CAAK,OAAO,gBAAA,CAAiB,QAAA,EAAU,CAAC,CAAA,CAAE,CAAA;AAChF,EAAA,IAAI,MAAA,CAAO,OAAA,CAAQ,MAAA,GAAS,CAAA,EAAG;AAC7B,IAAA,GAAA,CAAI,IAAA,CAAK,CAAA,oBAAA,EAAuB,GAAA,CAAI,IAAA,CAAK,MAAA,CAAO,QAAQ,IAAA,CAAK,IAAI,CAAC,CAAC,CAAA,CAAE,CAAA;AAAA,EACvE;AACA,EAAA,GAAA,CAAI,KAAK,EAAE,CAAA;AAEX,EAAA,MAAM,UAAA,GAAaJ,SAAA,CAAK,MAAA,CAAO,UAAA,EAAY,OAAO,MAAM,CAAA;AACxD,EAAA,IAAI,CAACK,aAAA,CAAW,UAAU,CAAA,EAAG;AAC3B,IAAA,GAAA,CAAI,KAAA,CAAM,CAAA,4BAAA,EAA+B,UAAU,CAAA,CAAE,CAAA;AACrD,IAAA,GAAA,CAAI,IAAA,CAAK,GAAA,CAAI,MAAA,CAAO,2EAA2E,CAAC,CAAA;AAChG,IAAA,OAAA,CAAQ,KAAK,CAAC,CAAA;AAAA,EAChB;AAEA,EAAA,MAAM,WAAA,GAAc,CAACL,SAAA,CAAK,MAAA,CAAO,QAAQ,YAAY,CAAA,EAAG,qBAAqB,YAAY,CAAA;AAEzF,EAAA,MAAM,YAAA,GAAe,OAAO,gBAAA,GACxBM,oBAAA,CAAc,OAAO,UAAA,EAAY,WAAA,EAAa,MAAA,CAAO,aAAa,CAAA,GAClE,IAAA;AAEJ,EAAA,IAAI,YAAA,IAAgB,OAAO,gBAAA,EAAkB;AAC3C,IAAA,GAAA,CAAI,IAAA,CAAK,CAAA,uBAAA,EAA0B,GAAA,CAAI,IAAA,CAAKH,aAAA,CAAS,MAAA,CAAO,UAAA,EAAY,YAAA,CAAa,IAAI,CAAC,CAAC,CAAA,CAAE,CAAA;AAAA,EAC/F,CAAA,MAAA,IAAW,OAAO,gBAAA,EAAkB;AAClC,IAAA,GAAA,CAAI,KAAK,gDAAgD,CAAA;AAAA,EAC3D;AAEA,EAAA,GAAA,CAAI,IAAA,CAAK,GAAA,CAAI,IAAA,CAAK,2BAA2B,CAAC,CAAA;AAC9C,EAAA,GAAA,CAAI,KAAK,EAAE,CAAA;AACX,EAAA,GAAA,CAAI,KAAK,CAAA,YAAA,EAAe,GAAA,CAAI,KAAK,MAAA,CAAO,MAAM,CAAC,CAAA,CAAE,CAAA;AACjD,EAAA,GAAA,CAAI,KAAK,CAAA,WAAA,EAAc,GAAA,CAAI,KAAK,MAAA,CAAO,IAAI,CAAC,CAAA,CAAE,CAAA;AAC9C,EAAA,GAAA,CAAI,KAAK,EAAE,CAAA;AAEX,EAAA,MAAM,eAAA,GAAkB,MAAM,iBAAA,CAAkB,MAAA,CAAO,QAAQ,MAAA,CAAO,IAAA,EAAM,OAAO,UAAU,CAAA;AAE7F,EAAA,IAAI,eAAA,CAAgB,WAAW,CAAA,EAAG;AAChC,IAAA,GAAA,CAAI,KAAK,EAAE,CAAA;AACX,IAAA,GAAA,CAAI,KAAK,qBAAqB,CAAA;AAC9B,IAAA,GAAA,CAAI,IAAA,CAAK,GAAA,CAAI,MAAA,CAAO,yDAAyD,CAAC,CAAA;AAC9E,IAAA,GAAA,CAAI,KAAK,GAAA,CAAI,MAAA,CAAO,CAAA,WAAA,EAAc,UAAU,EAAE,CAAC,CAAA;AAC/C,IAAA,GAAA,CAAI,KAAK,GAAA,CAAI,MAAA,CAAO,eAAe,MAAA,CAAO,IAAI,EAAE,CAAC,CAAA;AACjD,IAAA,OAAA,CAAQ,KAAK,CAAC,CAAA;AAAA,EAChB;AAEA,EAAA,MAAM,WAAA,uBAAkB,GAAA,EAA4B;AACpD,EAAA,KAAA,MAAW,QAAQ,eAAA,EAAiB;AAClC,IAAA,IAAI,CAAC,WAAA,CAAY,GAAA,CAAI,IAAA,CAAK,QAAQ,CAAA,EAAG;AACnC,MAAA,WAAA,CAAY,GAAA,CAAI,IAAA,CAAK,QAAA,EAAU,IAAI,CAAA;AAAA,IACrC;AAAA,EACF;AAEA,EAAA,GAAA,CAAI,KAAK,EAAE,CAAA;AACX,EAAA,GAAA,CAAI,OAAA,CAAQ,SAAS,GAAA,CAAI,IAAA,CAAK,YAAY,IAAA,CAAK,QAAA,EAAU,CAAC,CAAA,aAAA,CAAe,CAAA;AACzE,EAAA,GAAA,CAAI,KAAK,EAAE,CAAA;AACX,EAAA,GAAA,CAAI,IAAA,CAAK,GAAA,CAAI,IAAA,CAAK,qBAAqB,CAAC,CAAA;AACxC,EAAA,GAAA,CAAI,KAAK,EAAE,CAAA;AAEX,EAAA,MAAM,iBAID,EAAC;AAEN,EAAA,KAAA,MAAW,CAAC,QAAA,EAAU,IAAI,CAAA,IAAK,WAAA,EAAa;AAC1C,IAAA,MAAM,MAAA,GAASI,wBAAA,CAAkB,QAAA,EAAA,CAAU,EAAA,GAAA,YAAA,IAAA,IAAA,GAAA,MAAA,GAAA,YAAA,CAAc,MAAA,KAAd,YAAwB,IAAA,EAAM,MAAA,CAAO,OAAA,EAAS,MAAA,CAAO,gBAAgB,CAAA;AAEhH,IAAA,IAAI,CAAC,OAAO,QAAA,EAAU;AAEpB,MAAA,IAAI,MAAA,CAAO,WAAW,YAAA,EAAc;AAClC,QAAA,GAAA,CAAI,KAAA,CAAM,CAAA,wBAAA,EAA2B,QAAQ,CAAA,CAAE,CAAA;AAAA,MACjD,CAAA,MAAA,IAAW,MAAA,CAAO,MAAA,KAAW,iBAAA,EAAmB;AAC9C,QAAA,GAAA,CAAI,KAAA,CAAM,CAAA,qBAAA,EAAwB,QAAQ,CAAA,CAAE,CAAA;AAAA,MAC9C;AACA,MAAA;AAAA,IACF;AAEA,IAAA,IAAI;AACF,MAAA,MAAM,IAAA,GAAOR,eAAA,CAAa,IAAA,CAAK,YAAA,EAAc,OAAO,CAAA;AAEpD,MAAA,MAAM,cAAA,GAAiBS,iBAAU,IAAI,CAAA;AAErC,MAAA,IAAI,CAAC,cAAA,EAAgB;AACnB,QAAA,GAAA,CAAI,KAAK,CAAA,0BAAA,EAA6B,GAAA,CAAI,IAAA,CAAK,QAAQ,CAAC,CAAA,CAAE,CAAA;AAC1D,QAAA;AAAA,MACF;AAEA,MAAA,GAAA,CAAI,QAAQ,CAAA,EAAG,GAAA,CAAI,IAAA,CAAK,QAAQ,CAAC,CAAA,CAAE,CAAA;AAEnC,MAAA,cAAA,CAAe,IAAA,CAAK;AAAA,QAClB,IAAA,EAAM,QAAA;AAAA,QACN,UAAU,IAAA,CAAK,gBAAA;AAAA,QACf;AAAA,OACD,CAAA;AAAA,IACH,SAAS,KAAA,EAAO;AACd,MAAA,GAAA,CAAI,KAAA,CAAM,oBAAoB,GAAA,CAAI,IAAA,CAAK,QAAQ,CAAC,CAAA,EAAA,EAAK,KAAK,CAAA,CAAE,CAAA;AAAA,IAC9D;AAAA,EACF;AAEA,EAAA,GAAA,CAAI,KAAK,EAAE,CAAA;AACX,EAAA,GAAA,CAAI,OAAA,CAAQ,0BAA0B,GAAA,CAAI,IAAA,CAAK,eAAe,MAAA,CAAO,QAAA,EAAU,CAAC,CAAA,MAAA,CAAQ,CAAA;AACxF,EAAA,GAAA,CAAI,KAAK,EAAE,CAAA;AACX,EAAA,GAAA,CAAI,IAAA,CAAK,GAAA,CAAI,IAAA,CAAK,0BAA0B,CAAC,CAAA;AAC7C,EAAA,GAAA,CAAI,KAAK,EAAE,CAAA;AAEX,EAAA,MAAM,eAAA,GAAkB,MAAMC,uBAAA,CAAiB,cAAc,CAAA;AAE7D,EAAA,GAAA,CAAI,OAAA,CAAQ,SAAS,GAAA,CAAI,IAAA,CAAK,eAAe,MAAA,CAAO,QAAA,EAAU,CAAC,CAAA,sBAAA,CAAwB,CAAA;AACvF,EAAA,GAAA,CAAI,KAAK,EAAE,CAAA;AACX,EAAA,GAAA,CAAI,IAAA,CAAK,GAAA,CAAI,IAAA,CAAK,iBAAiB,CAAC,CAAA;AACpC,EAAA,GAAA,CAAI,KAAK,EAAE,CAAA;AAEX,EAAA,MAAM,UAAA,GAAaT,SAAA,CAAK,MAAA,CAAO,UAAA,EAAY,OAAO,SAAS,CAAA;AAC3D,EAAAU,YAAA,CAAU,UAAA,EAAY,EAAE,SAAA,EAAW,IAAA,EAAM,CAAA;AAEzC,EAAA,MAAM,aAAA,GAAgBV,SAAA,CAAK,UAAA,EAAY,WAAW,CAAA;AAClD,EAAAU,YAAA,CAAU,aAAA,EAAe,EAAE,SAAA,EAAW,IAAA,EAAM,CAAA;AAE5C,EAAA,MAAM,eAAA,GAAkBV,SAAA,CAAK,aAAA,EAAe,MAAA,CAAO,aAAa,CAAA;AAChE,EAAAW,gBAAA,CAAc,eAAA,EAAiB,IAAA,CAAK,SAAA,CAAU,eAAe,CAAC,CAAA;AAE9D,EAAA,MAAM,cAAA,GAAiB,CAAA;AAAA,+BAAA,EACQ,OAAO,aAAa,CAAA;AAAA;AAAA,CAAA;AAGnD,EAAA,MAAM,WAAA,GAAcX,SAAA,CAAK,UAAA,EAAY,UAAU,CAAA;AAC/C,EAAAW,gBAAA,CAAc,aAAa,cAAc,CAAA;AAEzC,EAAA,GAAA,CAAI,OAAA,CAAQ,CAAA,gBAAA,EAAmB,GAAA,CAAI,IAAA,CAAKR,aAAA,CAAS,OAAO,UAAA,EAAY,eAAe,CAAC,CAAC,CAAA,CAAE,CAAA;AACvF,EAAA,GAAA,CAAI,IAAA,CAAK,0BAA0B,GAAA,CAAI,IAAA,CAAK,eAAe,MAAA,CAAO,QAAA,EAAU,CAAC,CAAA,CAAE,CAAA;AAC/E,EAAA,GAAA,CAAI,IAAA;AAAA,IACF,CAAA,cAAA,EAAiB,GAAA,CAAI,IAAA,CAAA,CAAM,MAAA,CAAO,WAAW,IAAA,CAAK,SAAA,CAAU,eAAe,CAAA,EAAG,MAAM,CAAA,GAAI,IAAA,EAAM,OAAA,CAAQ,CAAC,CAAC,CAAC,CAAA,GAAA;AAAA,GAC3G;AACA,EAAA,GAAA,CAAI,KAAK,EAAE,CAAA;AACX,EAAA,GAAA,CAAI,OAAA,CAAQ,GAAA,CAAI,IAAA,CAAK,oBAAoB,CAAC,CAAA;AAC1C,EAAA,GAAA,CAAI,KAAK,EAAE,CAAA;AACb;AAEA,SAAS,qBAAqB,UAAA,EAAmC;AAC/D,EAAA,MAAM,UAAA,GAAa,CAAC,OAAA,EAAS,QAAA,EAAU,MAAM,CAAA;AAE7C,EAAA,KAAA,MAAW,OAAO,UAAA,EAAY;AAC5B,IAAA,MAAM,QAAA,GAAWH,SAAA,CAAK,UAAA,EAAY,GAAG,CAAA;AACrC,IAAA,IAAIK,aAAA,CAAW,QAAQ,CAAA,EAAG;AACxB,MAAA,OAAO,GAAA;AAAA,IACT;AAAA,EACF;AAEA,EAAA,OAAO,IAAA;AACT;AAEA,eAAe,IAAA,GAAO;AACpB,EAAA,MAAM,OAAA,GAAU,IAAIO,iBAAA,EAAQ;AAE5B,EAAA,OAAA,CACG,IAAA,CAAK,MAAM,CAAA,CACX,WAAA,CAAY,gDAAgD,CAAA,CAC5D,OAAA,CAAQ,WAAA,CAAY,OAAO,CAAA,CAC3B,MAAA,CAAO,mBAAmB,wEAAwE,CAAA,CAClG,MAAA,CAAO,kBAAA,EAAoB,6BAAA,EAA+B,iBAAiB,EAC3E,MAAA,CAAO,qBAAA,EAAuB,4BAAA,EAA8B,OAAO,CAAA,CACnE,MAAA,CAAO,2BAA2B,oBAAA,EAAsB,YAAY,EACpE,MAAA,CAAO,qBAAA,EAAuB,6BAA6B,CAAA,CAC3D,MAAA,CAAO,sBAAA,EAAwB,gCAAgC,CAAA,CAC/D,MAAA,CAAO,wBAAwB,kCAAA,EAAoC,oBAAA,EAAsB,EAAE,CAAA,CAC3F,MAAA,CAAO,wBAAwB,wBAAA,EAA0B,OAAA,CAAQ,GAAA,EAAK,CAAA,CACtE,WAAA;AAAA,IACC,OAAA;AAAA,IACA;AAAA;AAAA;AAAA;;AAAA;AAAA;;AAAA;AAAA;;AAAA;AAAA;;AAAA;AAAA;;AAAA;;AAAA;AAAA,IAAA;AAAA,IAsBD,KAAA,EAAM;AAET,EAAA,MAAM,OAAA,GAAU,QAAQ,IAAA,EAAK;AAE7B,EAAA,IAAI;AACF,IAAA,IAAI,YAAY,OAAA,CAAQ,MAAA;AACxB,IAAA,IAAI,CAAC,SAAA,EAAW;AACd,MAAA,MAAM,SAAA,GAAY,oBAAA,CAAqB,OAAA,CAAQ,UAAU,CAAA;AACzD,MAAA,IAAI,SAAA,EAAW;AACb,QAAA,SAAA,GAAY,SAAA;AACZ,QAAA,GAAA,CAAI,KAAK,CAAA,gCAAA,EAAmC,GAAA,CAAI,IAAA,CAAK,SAAS,CAAC,CAAA,CAAE,CAAA;AACjE,QAAA,GAAA,CAAI,KAAK,EAAE,CAAA;AAAA,MACb,CAAA,MAAO;AACL,QAAA,GAAA,CAAI,MAAM,iCAAiC,CAAA;AAC3C,QAAA,GAAA,CAAI,IAAA,CAAK,GAAA,CAAI,MAAA,CAAO,sCAAsC,CAAC,CAAA;AAC3D,QAAA,GAAA,CAAI,IAAA,CAAK,GAAA,CAAI,MAAA,CAAO,iEAAiE,CAAC,CAAA;AACtF,QAAA,OAAA,CAAQ,KAAK,CAAC,CAAA;AAAA,MAChB;AAAA,IACF;AAEA,IAAA,MAAM,MAAA,GAAwB;AAAA,MAC5B,MAAA,EAAQ,SAAA;AAAA,MACR,MAAM,OAAA,CAAQ,IAAA;AAAA,MACd,WAAW,OAAA,CAAQ,SAAA;AAAA,MACnB,eAAe,OAAA,CAAQ,aAAA;AAAA,MACvB,gBAAA,EAAkB,CAAC,OAAA,CAAQ,eAAA;AAAA,MAC3B,eAAe,OAAA,CAAQ,UAAA;AAAA,MACvB,SAAS,OAAA,CAAQ,OAAA;AAAA,MACjB,YAAY,OAAA,CAAQ;AAAA,KACtB;AAEA,IAAA,MAAM,WAAW,MAAM,CAAA;AAAA,EACzB,SAAS,KAAA,EAAO;AACd,IAAA,GAAA,CAAI,KAAK,EAAE,CAAA;AACX,IAAA,GAAA,CAAI,KAAA,CAAM,CAAA,aAAA,EAAgB,KAAK,CAAA,CAAE,CAAA;AACjC,IAAA,OAAA,CAAQ,KAAK,CAAC,CAAA;AAAA,EAChB;AACF;AAEA,IAAI,SAAA,CAAQ,SAAS,MAAA,EAAQ;AAC3B,EAAA,IAAA,EAAK;AACP","file":"index.js","sourcesContent":["/**\n * Peam Static Site Indexer\n *\n * Scans a static site output directory, discovers HTML files,\n * parses them into structured pages, and creates a searchable index.\n */\n\nimport { filePathToPathname, loadRobotsTxt, parseHTML, shouldIncludePath, type StructuredPage } from '@peam-ai/parser';\nimport { buildSearchIndex } from '@peam-ai/search';\nimport chalk from 'chalk';\nimport { Command } from 'commander';\nimport fg from 'fast-glob';\nimport { existsSync, mkdirSync, readFileSync, writeFileSync } from 'fs';\nimport { join, relative } from 'path';\n\nconst packageJson = JSON.parse(readFileSync(join(__dirname, '../package.json'), 'utf-8'));\n\ninterface IndexerConfig {\n  source: string;\n  outputDir: string;\n  indexFilename: string;\n  respectRobotsTxt: boolean;\n  robotsTxtPath?: string;\n  exclude: string[];\n  glob: string;\n  projectDir: string;\n}\n\ninterface DiscoveredPage {\n  pathname: string;\n  htmlFilePath: string;\n  relativeHtmlPath: string;\n}\n\nfunction parseExcludePatterns(value: string): string[] {\n  return value\n    .split(',')\n    .map((p) => p.trim())\n    .filter(Boolean);\n}\n\nconst log = {\n  success: (message: string) => console.log(chalk.green('✓ ') + message),\n  error: (message: string) => console.error(chalk.red('✗ ') + message),\n  warn: (message: string) => console.warn(chalk.yellow('⚠ ') + message),\n  info: (message: string) => console.log(chalk.blue('ℹ ') + message),\n  text: (message: string) => console.log(message),\n\n  cyan: (text: string) => chalk.cyan(text),\n  gray: (text: string) => chalk.gray(text),\n  bold: (text: string) => chalk.bold(text),\n  yellow: (text: string) => chalk.yellow(text),\n  red: (text: string) => chalk.red(text),\n  green: (text: string) => chalk.green(text),\n};\n\nasync function discoverHtmlFiles(\n  sourceDir: string,\n  globPattern: string,\n  projectDir: string\n): Promise<DiscoveredPage[]> {\n  const pages: DiscoveredPage[] = [];\n\n  try {\n    const sourceFullPath = join(projectDir, sourceDir);\n\n    const htmlFiles = await fg(globPattern, {\n      cwd: sourceFullPath,\n      absolute: true,\n      onlyFiles: true,\n      dot: false,\n      ignore: ['**/_next/**', '**/_astro/**', '**/404.{html,htm}', '**/500.{html,htm}'],\n    });\n\n    for (const htmlFilePath of htmlFiles) {\n      const relativePath = relative(sourceFullPath, htmlFilePath);\n      const pathname = filePathToPathname(relativePath);\n\n      pages.push({\n        pathname,\n        htmlFilePath,\n        relativeHtmlPath: relativePath,\n      });\n    }\n  } catch (error) {\n    log.warn(`Error discovering HTML files: ${error}`);\n  }\n\n  return pages;\n}\n\nasync function indexPages(config: IndexerConfig): Promise<void> {\n  log.text('\\n' + log.bold(log.cyan('Peam Static Site Indexer')) + '\\n');\n  log.text(log.bold('Configuration:'));\n  log.text(`  Project Directory: ${log.gray(config.projectDir)}`);\n  log.text(`  Source Directory: ${log.gray(config.source)}`);\n  log.text(`  Glob Pattern: ${log.gray(config.glob)}`);\n  log.text(`  Output Directory: ${log.gray(config.outputDir)}`);\n  log.text(`  Index Filename: ${log.gray(config.indexFilename)}`);\n  log.text(`  Respect robots.txt: ${log.gray(config.respectRobotsTxt.toString())}`);\n  if (config.exclude.length > 0) {\n    log.text(`  Exclude Patterns: ${log.gray(config.exclude.join(', '))}`);\n  }\n  log.text('');\n\n  const sourcePath = join(config.projectDir, config.source);\n  if (!existsSync(sourcePath)) {\n    log.error(`Source directory not found: ${sourcePath}`);\n    log.text(log.yellow('   Please build your site first or specify the correct --source directory'));\n    process.exit(1);\n  }\n\n  const searchPaths = [join(config.source, 'robots.txt'), 'public/robots.txt', 'robots.txt'];\n\n  const robotsResult = config.respectRobotsTxt\n    ? loadRobotsTxt(config.projectDir, searchPaths, config.robotsTxtPath)\n    : null;\n\n  if (robotsResult && config.respectRobotsTxt) {\n    log.info(`robots.txt loaded from ${log.cyan(relative(config.projectDir, robotsResult.path))}`);\n  } else if (config.respectRobotsTxt) {\n    log.info('No robots.txt found, all paths will be indexed');\n  }\n\n  log.text(log.bold('Discovering HTML files...'));\n  log.text('');\n  log.text(`  Scanning: ${log.gray(config.source)}`);\n  log.text(`  Pattern: ${log.gray(config.glob)}`);\n  log.text('');\n\n  const discoveredPages = await discoverHtmlFiles(config.source, config.glob, config.projectDir);\n\n  if (discoveredPages.length === 0) {\n    log.text('');\n    log.warn('No HTML files found');\n    log.text(log.yellow('   Check that your source directory contains HTML files'));\n    log.text(log.yellow(`   Source: ${sourcePath}`));\n    log.text(log.yellow(`   Pattern: ${config.glob}`));\n    process.exit(1);\n  }\n\n  const uniquePages = new Map<string, DiscoveredPage>();\n  for (const page of discoveredPages) {\n    if (!uniquePages.has(page.pathname)) {\n      uniquePages.set(page.pathname, page);\n    }\n  }\n\n  log.text('');\n  log.success(`Found ${log.bold(uniquePages.size.toString())} unique pages`);\n  log.text('');\n  log.text(log.bold('Processing pages...'));\n  log.text('');\n\n  const processedPages: Array<{\n    path: string;\n    htmlFile: string;\n    structuredPage: StructuredPage;\n  }> = [];\n\n  for (const [pathname, page] of uniquePages) {\n    const result = shouldIncludePath(pathname, robotsResult?.parser ?? null, config.exclude, config.respectRobotsTxt);\n\n    if (!result.included) {\n      // Log the specific reason for exclusion\n      if (result.reason === 'robots-txt') {\n        log.error(`Excluded by robots.txt: ${pathname}`);\n      } else if (result.reason === 'exclude-pattern') {\n        log.error(`Excluded by pattern: ${pathname}`);\n      }\n      continue;\n    }\n\n    try {\n      const html = readFileSync(page.htmlFilePath, 'utf-8');\n\n      const structuredPage = parseHTML(html);\n\n      if (!structuredPage) {\n        log.warn(`No content extracted from ${log.gray(pathname)}`);\n        continue;\n      }\n\n      log.success(`${log.cyan(pathname)}`);\n\n      processedPages.push({\n        path: pathname,\n        htmlFile: page.relativeHtmlPath,\n        structuredPage,\n      });\n    } catch (error) {\n      log.error(`Error processing ${log.gray(pathname)}: ${error}`);\n    }\n  }\n\n  log.text('');\n  log.success(`Successfully processed ${log.bold(processedPages.length.toString())} pages`);\n  log.text('');\n  log.text(log.bold('Creating search index...'));\n  log.text('');\n\n  const searchIndexData = await buildSearchIndex(processedPages);\n\n  log.success(`Added ${log.bold(processedPages.length.toString())} pages to search index`);\n  log.text('');\n  log.text(log.bold('Saving index...'));\n  log.text('');\n\n  const outputPath = join(config.projectDir, config.outputDir);\n  mkdirSync(outputPath, { recursive: true });\n\n  const generatedPath = join(outputPath, 'generated');\n  mkdirSync(generatedPath, { recursive: true });\n\n  const searchIndexFile = join(generatedPath, config.indexFilename);\n  writeFileSync(searchIndexFile, JSON.stringify(searchIndexData));\n\n  const indexJsContent = `// Auto-generated by Peam - DO NOT EDIT THIS FILE\nimport index from \"./generated/${config.indexFilename}\";\nexport default index;\n`;\n  const indexJsFile = join(outputPath, 'index.js');\n  writeFileSync(indexJsFile, indexJsContent);\n\n  log.success(`Index saved to: ${log.cyan(relative(config.projectDir, searchIndexFile))}`);\n  log.text(`  Total pages indexed: ${log.bold(processedPages.length.toString())}`);\n  log.text(\n    `  Index size: ${log.bold((Buffer.byteLength(JSON.stringify(searchIndexData), 'utf8') / 1024).toFixed(2))} KB`\n  );\n  log.text('');\n  log.success(log.bold('Indexing complete!'));\n  log.text('');\n}\n\nfunction probeSourceDirectory(projectDir: string): string | null {\n  const commonDirs = ['.next', '.build', '.out'];\n\n  for (const dir of commonDirs) {\n    const fullPath = join(projectDir, dir);\n    if (existsSync(fullPath)) {\n      return dir;\n    }\n  }\n\n  return null;\n}\n\nasync function main() {\n  const program = new Command();\n\n  program\n    .name('peam')\n    .description('Peam static site indexer (Next.js, Hugo, etc.)')\n    .version(packageJson.version)\n    .option('--source <path>', 'Source directory containing HTML files (auto-detected if not provided)')\n    .option('--glob <pattern>', 'Glob pattern for HTML files', '**/*.{html,htm}')\n    .option('--output-dir <path>', 'Output directory for index', '.peam')\n    .option('--index-filename <name>', 'Name of index file', 'index.json')\n    .option('--ignore-robots-txt', 'Disable robots.txt checking')\n    .option('--robots-path <path>', 'Custom path to robots.txt file')\n    .option('--exclude <patterns>', 'Comma-separated exclude patterns', parseExcludePatterns, [])\n    .option('--project-dir <path>', 'Project root directory', process.cwd())\n    .addHelpText(\n      'after',\n      `\nExamples:\n  # Auto-detect build directory\n  $ peam\n\n  # Hugo\n  $ peam --source public\n\n  # Next.js\n  $ peam --source .next\n\n  # Custom output directory\n  $ peam --source dist --glob \"**/*.html\"\n\n  # Exclude patterns\n  $ peam --exclude \"/admin/**,/api/*,/private-*\"\n\nFor Next.js 15+, the @peam-ai/next integration is recommended for production use.\n\nMore information: https://peam.ai\n    `\n    )\n    .parse();\n\n  const options = program.opts();\n\n  try {\n    let sourceDir = options.source;\n    if (!sourceDir) {\n      const probedDir = probeSourceDirectory(options.projectDir);\n      if (probedDir) {\n        sourceDir = probedDir;\n        log.info(`Auto-detected source directory: ${log.cyan(probedDir)}`);\n        log.text('');\n      } else {\n        log.error('No build output directory found');\n        log.text(log.yellow('   Searched for: .next, .build, .out'));\n        log.text(log.yellow('   Please build your site first or specify --source <directory>'));\n        process.exit(1);\n      }\n    }\n\n    const config: IndexerConfig = {\n      source: sourceDir,\n      glob: options.glob,\n      outputDir: options.outputDir,\n      indexFilename: options.indexFilename,\n      respectRobotsTxt: !options.ignoreRobotsTxt,\n      robotsTxtPath: options.robotsPath,\n      exclude: options.exclude,\n      projectDir: options.projectDir,\n    };\n\n    await indexPages(config);\n  } catch (error) {\n    log.text('');\n    log.error(`Fatal error: ${error}`);\n    process.exit(1);\n  }\n}\n\nif (require.main === module) {\n  main();\n}\n\nexport { indexPages, type IndexerConfig };\n"]}

package/dist/index.mjs

@@ -0,0 +1,238 @@
+import { __require } from './chunk-BJTO5JO5.mjs';
+import { loadRobotsTxt, shouldIncludePath, parseHTML, filePathToPathname } from '@peam-ai/parser';
+import { buildSearchIndex } from '@peam-ai/search';
+import chalk from 'chalk';
+import { Command } from 'commander';
+import fg from 'fast-glob';
+import { readFileSync, existsSync, mkdirSync, writeFileSync } from 'fs';
+import { join, relative } from 'path';
+
+var packageJson = JSON.parse(readFileSync(join(__dirname, "../package.json"), "utf-8"));
+function parseExcludePatterns(value) {
+  return value.split(",").map((p) => p.trim()).filter(Boolean);
+}
+var log = {
+  success: (message) => console.log(chalk.green("\u2713 ") + message),
+  error: (message) => console.error(chalk.red("\u2717 ") + message),
+  warn: (message) => console.warn(chalk.yellow("\u26A0 ") + message),
+  info: (message) => console.log(chalk.blue("\u2139 ") + message),
+  text: (message) => console.log(message),
+  cyan: (text) => chalk.cyan(text),
+  gray: (text) => chalk.gray(text),
+  bold: (text) => chalk.bold(text),
+  yellow: (text) => chalk.yellow(text),
+  red: (text) => chalk.red(text),
+  green: (text) => chalk.green(text)
+};
+async function discoverHtmlFiles(sourceDir, globPattern, projectDir) {
+  const pages = [];
+  try {
+    const sourceFullPath = join(projectDir, sourceDir);
+    const htmlFiles = await fg(globPattern, {
+      cwd: sourceFullPath,
+      absolute: true,
+      onlyFiles: true,
+      dot: false,
+      ignore: ["**/_next/**", "**/_astro/**", "**/404.{html,htm}", "**/500.{html,htm}"]
+    });
+    for (const htmlFilePath of htmlFiles) {
+      const relativePath = relative(sourceFullPath, htmlFilePath);
+      const pathname = filePathToPathname(relativePath);
+      pages.push({
+        pathname,
+        htmlFilePath,
+        relativeHtmlPath: relativePath
+      });
+    }
+  } catch (error) {
+    log.warn(`Error discovering HTML files: ${error}`);
+  }
+  return pages;
+}
+async function indexPages(config) {
+  var _a;
+  log.text("\n" + log.bold(log.cyan("Peam Static Site Indexer")) + "\n");
+  log.text(log.bold("Configuration:"));
+  log.text(`  Project Directory: ${log.gray(config.projectDir)}`);
+  log.text(`  Source Directory: ${log.gray(config.source)}`);
+  log.text(`  Glob Pattern: ${log.gray(config.glob)}`);
+  log.text(`  Output Directory: ${log.gray(config.outputDir)}`);
+  log.text(`  Index Filename: ${log.gray(config.indexFilename)}`);
+  log.text(`  Respect robots.txt: ${log.gray(config.respectRobotsTxt.toString())}`);
+  if (config.exclude.length > 0) {
+    log.text(`  Exclude Patterns: ${log.gray(config.exclude.join(", "))}`);
+  }
+  log.text("");
+  const sourcePath = join(config.projectDir, config.source);
+  if (!existsSync(sourcePath)) {
+    log.error(`Source directory not found: ${sourcePath}`);
+    log.text(log.yellow("   Please build your site first or specify the correct --source directory"));
+    process.exit(1);
+  }
+  const searchPaths = [join(config.source, "robots.txt"), "public/robots.txt", "robots.txt"];
+  const robotsResult = config.respectRobotsTxt ? loadRobotsTxt(config.projectDir, searchPaths, config.robotsTxtPath) : null;
+  if (robotsResult && config.respectRobotsTxt) {
+    log.info(`robots.txt loaded from ${log.cyan(relative(config.projectDir, robotsResult.path))}`);
+  } else if (config.respectRobotsTxt) {
+    log.info("No robots.txt found, all paths will be indexed");
+  }
+  log.text(log.bold("Discovering HTML files..."));
+  log.text("");
+  log.text(`  Scanning: ${log.gray(config.source)}`);
+  log.text(`  Pattern: ${log.gray(config.glob)}`);
+  log.text("");
+  const discoveredPages = await discoverHtmlFiles(config.source, config.glob, config.projectDir);
+  if (discoveredPages.length === 0) {
+    log.text("");
+    log.warn("No HTML files found");
+    log.text(log.yellow("   Check that your source directory contains HTML files"));
+    log.text(log.yellow(`   Source: ${sourcePath}`));
+    log.text(log.yellow(`   Pattern: ${config.glob}`));
+    process.exit(1);
+  }
+  const uniquePages = /* @__PURE__ */ new Map();
+  for (const page of discoveredPages) {
+    if (!uniquePages.has(page.pathname)) {
+      uniquePages.set(page.pathname, page);
+    }
+  }
+  log.text("");
+  log.success(`Found ${log.bold(uniquePages.size.toString())} unique pages`);
+  log.text("");
+  log.text(log.bold("Processing pages..."));
+  log.text("");
+  const processedPages = [];
+  for (const [pathname, page] of uniquePages) {
+    const result = shouldIncludePath(pathname, (_a = robotsResult == null ? void 0 : robotsResult.parser) != null ? _a : null, config.exclude, config.respectRobotsTxt);
+    if (!result.included) {
+      if (result.reason === "robots-txt") {
+        log.error(`Excluded by robots.txt: ${pathname}`);
+      } else if (result.reason === "exclude-pattern") {
+        log.error(`Excluded by pattern: ${pathname}`);
+      }
+      continue;
+    }
+    try {
+      const html = readFileSync(page.htmlFilePath, "utf-8");
+      const structuredPage = parseHTML(html);
+      if (!structuredPage) {
+        log.warn(`No content extracted from ${log.gray(pathname)}`);
+        continue;
+      }
+      log.success(`${log.cyan(pathname)}`);
+      processedPages.push({
+        path: pathname,
+        htmlFile: page.relativeHtmlPath,
+        structuredPage
+      });
+    } catch (error) {
+      log.error(`Error processing ${log.gray(pathname)}: ${error}`);
+    }
+  }
+  log.text("");
+  log.success(`Successfully processed ${log.bold(processedPages.length.toString())} pages`);
+  log.text("");
+  log.text(log.bold("Creating search index..."));
+  log.text("");
+  const searchIndexData = await buildSearchIndex(processedPages);
+  log.success(`Added ${log.bold(processedPages.length.toString())} pages to search index`);
+  log.text("");
+  log.text(log.bold("Saving index..."));
+  log.text("");
+  const outputPath = join(config.projectDir, config.outputDir);
+  mkdirSync(outputPath, { recursive: true });
+  const generatedPath = join(outputPath, "generated");
+  mkdirSync(generatedPath, { recursive: true });
+  const searchIndexFile = join(generatedPath, config.indexFilename);
+  writeFileSync(searchIndexFile, JSON.stringify(searchIndexData));
+  const indexJsContent = `// Auto-generated by Peam - DO NOT EDIT THIS FILE
+import index from "./generated/${config.indexFilename}";
+export default index;
+`;
+  const indexJsFile = join(outputPath, "index.js");
+  writeFileSync(indexJsFile, indexJsContent);
+  log.success(`Index saved to: ${log.cyan(relative(config.projectDir, searchIndexFile))}`);
+  log.text(`  Total pages indexed: ${log.bold(processedPages.length.toString())}`);
+  log.text(
+    `  Index size: ${log.bold((Buffer.byteLength(JSON.stringify(searchIndexData), "utf8") / 1024).toFixed(2))} KB`
+  );
+  log.text("");
+  log.success(log.bold("Indexing complete!"));
+  log.text("");
+}
+function probeSourceDirectory(projectDir) {
+  const commonDirs = [".next", ".build", ".out"];
+  for (const dir of commonDirs) {
+    const fullPath = join(projectDir, dir);
+    if (existsSync(fullPath)) {
+      return dir;
+    }
+  }
+  return null;
+}
+async function main() {
+  const program = new Command();
+  program.name("peam").description("Peam static site indexer (Next.js, Hugo, etc.)").version(packageJson.version).option("--source <path>", "Source directory containing HTML files (auto-detected if not provided)").option("--glob <pattern>", "Glob pattern for HTML files", "**/*.{html,htm}").option("--output-dir <path>", "Output directory for index", ".peam").option("--index-filename <name>", "Name of index file", "index.json").option("--ignore-robots-txt", "Disable robots.txt checking").option("--robots-path <path>", "Custom path to robots.txt file").option("--exclude <patterns>", "Comma-separated exclude patterns", parseExcludePatterns, []).option("--project-dir <path>", "Project root directory", process.cwd()).addHelpText(
+    "after",
+    `
+Examples:
+  # Auto-detect build directory
+  $ peam
+
+  # Hugo
+  $ peam --source public
+
+  # Next.js
+  $ peam --source .next
+
+  # Custom output directory
+  $ peam --source dist --glob "**/*.html"
+
+  # Exclude patterns
+  $ peam --exclude "/admin/**,/api/*,/private-*"
+
+For Next.js 15+, the @peam-ai/next integration is recommended for production use.
+
+More information: https://peam.ai
+    `
+  ).parse();
+  const options = program.opts();
+  try {
+    let sourceDir = options.source;
+    if (!sourceDir) {
+      const probedDir = probeSourceDirectory(options.projectDir);
+      if (probedDir) {
+        sourceDir = probedDir;
+        log.info(`Auto-detected source directory: ${log.cyan(probedDir)}`);
+        log.text("");
+      } else {
+        log.error("No build output directory found");
+        log.text(log.yellow("   Searched for: .next, .build, .out"));
+        log.text(log.yellow("   Please build your site first or specify --source <directory>"));
+        process.exit(1);
+      }
+    }
+    const config = {
+      source: sourceDir,
+      glob: options.glob,
+      outputDir: options.outputDir,
+      indexFilename: options.indexFilename,
+      respectRobotsTxt: !options.ignoreRobotsTxt,
+      robotsTxtPath: options.robotsPath,
+      exclude: options.exclude,
+      projectDir: options.projectDir
+    };
+    await indexPages(config);
+  } catch (error) {
+    log.text("");
+    log.error(`Fatal error: ${error}`);
+    process.exit(1);
+  }
+}
+if (__require.main === module) {
+  main();
+}
+
+export { indexPages };
+//# sourceMappingURL=index.mjs.map
+//# sourceMappingURL=index.mjs.map

package/dist/index.mjs.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/cli.ts"],"names":[],"mappings":";;;;;;;;;AAeA,IAAM,WAAA,GAAc,KAAK,KAAA,CAAM,YAAA,CAAa,KAAK,SAAA,EAAW,iBAAiB,CAAA,EAAG,OAAO,CAAC,CAAA;AAmBxF,SAAS,qBAAqB,KAAA,EAAyB;AACrD,EAAA,OAAO,KAAA,CACJ,KAAA,CAAM,GAAG,CAAA,CACT,GAAA,CAAI,CAAC,CAAA,KAAM,CAAA,CAAE,IAAA,EAAM,CAAA,CACnB,MAAA,CAAO,OAAO,CAAA;AACnB;AAEA,IAAM,GAAA,GAAM;AAAA,EACV,OAAA,EAAS,CAAC,OAAA,KAAoB,OAAA,CAAQ,IAAI,KAAA,CAAM,KAAA,CAAM,SAAI,CAAA,GAAI,OAAO,CAAA;AAAA,EACrE,KAAA,EAAO,CAAC,OAAA,KAAoB,OAAA,CAAQ,MAAM,KAAA,CAAM,GAAA,CAAI,SAAI,CAAA,GAAI,OAAO,CAAA;AAAA,EACnE,IAAA,EAAM,CAAC,OAAA,KAAoB,OAAA,CAAQ,KAAK,KAAA,CAAM,MAAA,CAAO,SAAI,CAAA,GAAI,OAAO,CAAA;AAAA,EACpE,IAAA,EAAM,CAAC,OAAA,KAAoB,OAAA,CAAQ,IAAI,KAAA,CAAM,IAAA,CAAK,SAAI,CAAA,GAAI,OAAO,CAAA;AAAA,EACjE,IAAA,EAAM,CAAC,OAAA,KAAoB,OAAA,CAAQ,IAAI,OAAO,CAAA;AAAA,EAE9C,IAAA,EAAM,CAAC,IAAA,KAAiB,KAAA,CAAM,KAAK,IAAI,CAAA;AAAA,EACvC,IAAA,EAAM,CAAC,IAAA,KAAiB,KAAA,CAAM,KAAK,IAAI,CAAA;AAAA,EACvC,IAAA,EAAM,CAAC,IAAA,KAAiB,KAAA,CAAM,KAAK,IAAI,CAAA;AAAA,EACvC,MAAA,EAAQ,CAAC,IAAA,KAAiB,KAAA,CAAM,OAAO,IAAI,CAAA;AAAA,EAC3C,GAAA,EAAK,CAAC,IAAA,KAAiB,KAAA,CAAM,IAAI,IAAI,CAAA;AAAA,EACrC,KAAA,EAAO,CAAC,IAAA,KAAiB,KAAA,CAAM,MAAM,IAAI;AAC3C,CAAA;AAEA,eAAe,iBAAA,CACb,SAAA,EACA,WAAA,EACA,UAAA,EAC2B;AAC3B,EAAA,MAAM,QAA0B,EAAC;AAEjC,EAAA,IAAI;AACF,IAAA,MAAM,cAAA,GAAiB,IAAA,CAAK,UAAA,EAAY,SAAS,CAAA;AAEjD,IAAA,MAAM,SAAA,GAAY,MAAM,EAAA,CAAG,WAAA,EAAa;AAAA,MACtC,GAAA,EAAK,cAAA;AAAA,MACL,QAAA,EAAU,IAAA;AAAA,MACV,SAAA,EAAW,IAAA;AAAA,MACX,GAAA,EAAK,KAAA;AAAA,MACL,MAAA,EAAQ,CAAC,aAAA,EAAe,cAAA,EAAgB,qBAAqB,mBAAmB;AAAA,KACjF,CAAA;AAED,IAAA,KAAA,MAAW,gBAAgB,SAAA,EAAW;AACpC,MAAA,MAAM,YAAA,GAAe,QAAA,CAAS,cAAA,EAAgB,YAAY,CAAA;AAC1D,MAAA,MAAM,QAAA,GAAW,mBAAmB,YAAY,CAAA;AAEhD,MAAA,KAAA,CAAM,IAAA,CAAK;AAAA,QACT,QAAA;AAAA,QACA,YAAA;AAAA,QACA,gBAAA,EAAkB;AAAA,OACnB,CAAA;AAAA,IACH;AAAA,EACF,SAAS,KAAA,EAAO;AACd,IAAA,GAAA,CAAI,IAAA,CAAK,CAAA,8BAAA,EAAiC,KAAK,CAAA,CAAE,CAAA;AAAA,EACnD;AAEA,EAAA,OAAO,KAAA;AACT;AAEA,eAAe,WAAW,MAAA,EAAsC;AA3FhE,EAAA,IAAA,EAAA;AA4FE,EAAA,GAAA,CAAI,IAAA,CAAK,OAAO,GAAA,CAAI,IAAA,CAAK,IAAI,IAAA,CAAK,0BAA0B,CAAC,CAAA,GAAI,IAAI,CAAA;AACrE,EAAA,GAAA,CAAI,IAAA,CAAK,GAAA,CAAI,IAAA,CAAK,gBAAgB,CAAC,CAAA;AACnC,EAAA,GAAA,CAAI,KAAK,CAAA,qBAAA,EAAwB,GAAA,CAAI,KAAK,MAAA,CAAO,UAAU,CAAC,CAAA,CAAE,CAAA;AAC9D,EAAA,GAAA,CAAI,KAAK,CAAA,oBAAA,EAAuB,GAAA,CAAI,KAAK,MAAA,CAAO,MAAM,CAAC,CAAA,CAAE,CAAA;AACzD,EAAA,GAAA,CAAI,KAAK,CAAA,gBAAA,EAAmB,GAAA,CAAI,KAAK,MAAA,CAAO,IAAI,CAAC,CAAA,CAAE,CAAA;AACnD,EAAA,GAAA,CAAI,KAAK,CAAA,oBAAA,EAAuB,GAAA,CAAI,KAAK,MAAA,CAAO,SAAS,CAAC,CAAA,CAAE,CAAA;AAC5D,EAAA,GAAA,CAAI,KAAK,CAAA,kBAAA,EAAqB,GAAA,CAAI,KAAK,MAAA,CAAO,aAAa,CAAC,CAAA,CAAE,CAAA;AAC9D,EAAA,GAAA,CAAI,IAAA,CAAK,yBAAyB,GAAA,CAAI,IAAA,CAAK,OAAO,gBAAA,CAAiB,QAAA,EAAU,CAAC,CAAA,CAAE,CAAA;AAChF,EAAA,IAAI,MAAA,CAAO,OAAA,CAAQ,MAAA,GAAS,CAAA,EAAG;AAC7B,IAAA,GAAA,CAAI,IAAA,CAAK,CAAA,oBAAA,EAAuB,GAAA,CAAI,IAAA,CAAK,MAAA,CAAO,QAAQ,IAAA,CAAK,IAAI,CAAC,CAAC,CAAA,CAAE,CAAA;AAAA,EACvE;AACA,EAAA,GAAA,CAAI,KAAK,EAAE,CAAA;AAEX,EAAA,MAAM,UAAA,GAAa,IAAA,CAAK,MAAA,CAAO,UAAA,EAAY,OAAO,MAAM,CAAA;AACxD,EAAA,IAAI,CAAC,UAAA,CAAW,UAAU,CAAA,EAAG;AAC3B,IAAA,GAAA,CAAI,KAAA,CAAM,CAAA,4BAAA,EAA+B,UAAU,CAAA,CAAE,CAAA;AACrD,IAAA,GAAA,CAAI,IAAA,CAAK,GAAA,CAAI,MAAA,CAAO,2EAA2E,CAAC,CAAA;AAChG,IAAA,OAAA,CAAQ,KAAK,CAAC,CAAA;AAAA,EAChB;AAEA,EAAA,MAAM,WAAA,GAAc,CAAC,IAAA,CAAK,MAAA,CAAO,QAAQ,YAAY,CAAA,EAAG,qBAAqB,YAAY,CAAA;AAEzF,EAAA,MAAM,YAAA,GAAe,OAAO,gBAAA,GACxB,aAAA,CAAc,OAAO,UAAA,EAAY,WAAA,EAAa,MAAA,CAAO,aAAa,CAAA,GAClE,IAAA;AAEJ,EAAA,IAAI,YAAA,IAAgB,OAAO,gBAAA,EAAkB;AAC3C,IAAA,GAAA,CAAI,IAAA,CAAK,CAAA,uBAAA,EAA0B,GAAA,CAAI,IAAA,CAAK,QAAA,CAAS,MAAA,CAAO,UAAA,EAAY,YAAA,CAAa,IAAI,CAAC,CAAC,CAAA,CAAE,CAAA;AAAA,EAC/F,CAAA,MAAA,IAAW,OAAO,gBAAA,EAAkB;AAClC,IAAA,GAAA,CAAI,KAAK,gDAAgD,CAAA;AAAA,EAC3D;AAEA,EAAA,GAAA,CAAI,IAAA,CAAK,GAAA,CAAI,IAAA,CAAK,2BAA2B,CAAC,CAAA;AAC9C,EAAA,GAAA,CAAI,KAAK,EAAE,CAAA;AACX,EAAA,GAAA,CAAI,KAAK,CAAA,YAAA,EAAe,GAAA,CAAI,KAAK,MAAA,CAAO,MAAM,CAAC,CAAA,CAAE,CAAA;AACjD,EAAA,GAAA,CAAI,KAAK,CAAA,WAAA,EAAc,GAAA,CAAI,KAAK,MAAA,CAAO,IAAI,CAAC,CAAA,CAAE,CAAA;AAC9C,EAAA,GAAA,CAAI,KAAK,EAAE,CAAA;AAEX,EAAA,MAAM,eAAA,GAAkB,MAAM,iBAAA,CAAkB,MAAA,CAAO,QAAQ,MAAA,CAAO,IAAA,EAAM,OAAO,UAAU,CAAA;AAE7F,EAAA,IAAI,eAAA,CAAgB,WAAW,CAAA,EAAG;AAChC,IAAA,GAAA,CAAI,KAAK,EAAE,CAAA;AACX,IAAA,GAAA,CAAI,KAAK,qBAAqB,CAAA;AAC9B,IAAA,GAAA,CAAI,IAAA,CAAK,GAAA,CAAI,MAAA,CAAO,yDAAyD,CAAC,CAAA;AAC9E,IAAA,GAAA,CAAI,KAAK,GAAA,CAAI,MAAA,CAAO,CAAA,WAAA,EAAc,UAAU,EAAE,CAAC,CAAA;AAC/C,IAAA,GAAA,CAAI,KAAK,GAAA,CAAI,MAAA,CAAO,eAAe,MAAA,CAAO,IAAI,EAAE,CAAC,CAAA;AACjD,IAAA,OAAA,CAAQ,KAAK,CAAC,CAAA;AAAA,EAChB;AAEA,EAAA,MAAM,WAAA,uBAAkB,GAAA,EAA4B;AACpD,EAAA,KAAA,MAAW,QAAQ,eAAA,EAAiB;AAClC,IAAA,IAAI,CAAC,WAAA,CAAY,GAAA,CAAI,IAAA,CAAK,QAAQ,CAAA,EAAG;AACnC,MAAA,WAAA,CAAY,GAAA,CAAI,IAAA,CAAK,QAAA,EAAU,IAAI,CAAA;AAAA,IACrC;AAAA,EACF;AAEA,EAAA,GAAA,CAAI,KAAK,EAAE,CAAA;AACX,EAAA,GAAA,CAAI,OAAA,CAAQ,SAAS,GAAA,CAAI,IAAA,CAAK,YAAY,IAAA,CAAK,QAAA,EAAU,CAAC,CAAA,aAAA,CAAe,CAAA;AACzE,EAAA,GAAA,CAAI,KAAK,EAAE,CAAA;AACX,EAAA,GAAA,CAAI,IAAA,CAAK,GAAA,CAAI,IAAA,CAAK,qBAAqB,CAAC,CAAA;AACxC,EAAA,GAAA,CAAI,KAAK,EAAE,CAAA;AAEX,EAAA,MAAM,iBAID,EAAC;AAEN,EAAA,KAAA,MAAW,CAAC,QAAA,EAAU,IAAI,CAAA,IAAK,WAAA,EAAa;AAC1C,IAAA,MAAM,MAAA,GAAS,iBAAA,CAAkB,QAAA,EAAA,CAAU,EAAA,GAAA,YAAA,IAAA,IAAA,GAAA,MAAA,GAAA,YAAA,CAAc,MAAA,KAAd,YAAwB,IAAA,EAAM,MAAA,CAAO,OAAA,EAAS,MAAA,CAAO,gBAAgB,CAAA;AAEhH,IAAA,IAAI,CAAC,OAAO,QAAA,EAAU;AAEpB,MAAA,IAAI,MAAA,CAAO,WAAW,YAAA,EAAc;AAClC,QAAA,GAAA,CAAI,KAAA,CAAM,CAAA,wBAAA,EAA2B,QAAQ,CAAA,CAAE,CAAA;AAAA,MACjD,CAAA,MAAA,IAAW,MAAA,CAAO,MAAA,KAAW,iBAAA,EAAmB;AAC9C,QAAA,GAAA,CAAI,KAAA,CAAM,CAAA,qBAAA,EAAwB,QAAQ,CAAA,CAAE,CAAA;AAAA,MAC9C;AACA,MAAA;AAAA,IACF;AAEA,IAAA,IAAI;AACF,MAAA,MAAM,IAAA,GAAO,YAAA,CAAa,IAAA,CAAK,YAAA,EAAc,OAAO,CAAA;AAEpD,MAAA,MAAM,cAAA,GAAiB,UAAU,IAAI,CAAA;AAErC,MAAA,IAAI,CAAC,cAAA,EAAgB;AACnB,QAAA,GAAA,CAAI,KAAK,CAAA,0BAAA,EAA6B,GAAA,CAAI,IAAA,CAAK,QAAQ,CAAC,CAAA,CAAE,CAAA;AAC1D,QAAA;AAAA,MACF;AAEA,MAAA,GAAA,CAAI,QAAQ,CAAA,EAAG,GAAA,CAAI,IAAA,CAAK,QAAQ,CAAC,CAAA,CAAE,CAAA;AAEnC,MAAA,cAAA,CAAe,IAAA,CAAK;AAAA,QAClB,IAAA,EAAM,QAAA;AAAA,QACN,UAAU,IAAA,CAAK,gBAAA;AAAA,QACf;AAAA,OACD,CAAA;AAAA,IACH,SAAS,KAAA,EAAO;AACd,MAAA,GAAA,CAAI,KAAA,CAAM,oBAAoB,GAAA,CAAI,IAAA,CAAK,QAAQ,CAAC,CAAA,EAAA,EAAK,KAAK,CAAA,CAAE,CAAA;AAAA,IAC9D;AAAA,EACF;AAEA,EAAA,GAAA,CAAI,KAAK,EAAE,CAAA;AACX,EAAA,GAAA,CAAI,OAAA,CAAQ,0BAA0B,GAAA,CAAI,IAAA,CAAK,eAAe,MAAA,CAAO,QAAA,EAAU,CAAC,CAAA,MAAA,CAAQ,CAAA;AACxF,EAAA,GAAA,CAAI,KAAK,EAAE,CAAA;AACX,EAAA,GAAA,CAAI,IAAA,CAAK,GAAA,CAAI,IAAA,CAAK,0BAA0B,CAAC,CAAA;AAC7C,EAAA,GAAA,CAAI,KAAK,EAAE,CAAA;AAEX,EAAA,MAAM,eAAA,GAAkB,MAAM,gBAAA,CAAiB,cAAc,CAAA;AAE7D,EAAA,GAAA,CAAI,OAAA,CAAQ,SAAS,GAAA,CAAI,IAAA,CAAK,eAAe,MAAA,CAAO,QAAA,EAAU,CAAC,CAAA,sBAAA,CAAwB,CAAA;AACvF,EAAA,GAAA,CAAI,KAAK,EAAE,CAAA;AACX,EAAA,GAAA,CAAI,IAAA,CAAK,GAAA,CAAI,IAAA,CAAK,iBAAiB,CAAC,CAAA;AACpC,EAAA,GAAA,CAAI,KAAK,EAAE,CAAA;AAEX,EAAA,MAAM,UAAA,GAAa,IAAA,CAAK,MAAA,CAAO,UAAA,EAAY,OAAO,SAAS,CAAA;AAC3D,EAAA,SAAA,CAAU,UAAA,EAAY,EAAE,SAAA,EAAW,IAAA,EAAM,CAAA;AAEzC,EAAA,MAAM,aAAA,GAAgB,IAAA,CAAK,UAAA,EAAY,WAAW,CAAA;AAClD,EAAA,SAAA,CAAU,aAAA,EAAe,EAAE,SAAA,EAAW,IAAA,EAAM,CAAA;AAE5C,EAAA,MAAM,eAAA,GAAkB,IAAA,CAAK,aAAA,EAAe,MAAA,CAAO,aAAa,CAAA;AAChE,EAAA,aAAA,CAAc,eAAA,EAAiB,IAAA,CAAK,SAAA,CAAU,eAAe,CAAC,CAAA;AAE9D,EAAA,MAAM,cAAA,GAAiB,CAAA;AAAA,+BAAA,EACQ,OAAO,aAAa,CAAA;AAAA;AAAA,CAAA;AAGnD,EAAA,MAAM,WAAA,GAAc,IAAA,CAAK,UAAA,EAAY,UAAU,CAAA;AAC/C,EAAA,aAAA,CAAc,aAAa,cAAc,CAAA;AAEzC,EAAA,GAAA,CAAI,OAAA,CAAQ,CAAA,gBAAA,EAAmB,GAAA,CAAI,IAAA,CAAK,QAAA,CAAS,OAAO,UAAA,EAAY,eAAe,CAAC,CAAC,CAAA,CAAE,CAAA;AACvF,EAAA,GAAA,CAAI,IAAA,CAAK,0BAA0B,GAAA,CAAI,IAAA,CAAK,eAAe,MAAA,CAAO,QAAA,EAAU,CAAC,CAAA,CAAE,CAAA;AAC/E,EAAA,GAAA,CAAI,IAAA;AAAA,IACF,CAAA,cAAA,EAAiB,GAAA,CAAI,IAAA,CAAA,CAAM,MAAA,CAAO,WAAW,IAAA,CAAK,SAAA,CAAU,eAAe,CAAA,EAAG,MAAM,CAAA,GAAI,IAAA,EAAM,OAAA,CAAQ,CAAC,CAAC,CAAC,CAAA,GAAA;AAAA,GAC3G;AACA,EAAA,GAAA,CAAI,KAAK,EAAE,CAAA;AACX,EAAA,GAAA,CAAI,OAAA,CAAQ,GAAA,CAAI,IAAA,CAAK,oBAAoB,CAAC,CAAA;AAC1C,EAAA,GAAA,CAAI,KAAK,EAAE,CAAA;AACb;AAEA,SAAS,qBAAqB,UAAA,EAAmC;AAC/D,EAAA,MAAM,UAAA,GAAa,CAAC,OAAA,EAAS,QAAA,EAAU,MAAM,CAAA;AAE7C,EAAA,KAAA,MAAW,OAAO,UAAA,EAAY;AAC5B,IAAA,MAAM,QAAA,GAAW,IAAA,CAAK,UAAA,EAAY,GAAG,CAAA;AACrC,IAAA,IAAI,UAAA,CAAW,QAAQ,CAAA,EAAG;AACxB,MAAA,OAAO,GAAA;AAAA,IACT;AAAA,EACF;AAEA,EAAA,OAAO,IAAA;AACT;AAEA,eAAe,IAAA,GAAO;AACpB,EAAA,MAAM,OAAA,GAAU,IAAI,OAAA,EAAQ;AAE5B,EAAA,OAAA,CACG,IAAA,CAAK,MAAM,CAAA,CACX,WAAA,CAAY,gDAAgD,CAAA,CAC5D,OAAA,CAAQ,WAAA,CAAY,OAAO,CAAA,CAC3B,MAAA,CAAO,mBAAmB,wEAAwE,CAAA,CAClG,MAAA,CAAO,kBAAA,EAAoB,6BAAA,EAA+B,iBAAiB,EAC3E,MAAA,CAAO,qBAAA,EAAuB,4BAAA,EAA8B,OAAO,CAAA,CACnE,MAAA,CAAO,2BAA2B,oBAAA,EAAsB,YAAY,EACpE,MAAA,CAAO,qBAAA,EAAuB,6BAA6B,CAAA,CAC3D,MAAA,CAAO,sBAAA,EAAwB,gCAAgC,CAAA,CAC/D,MAAA,CAAO,wBAAwB,kCAAA,EAAoC,oBAAA,EAAsB,EAAE,CAAA,CAC3F,MAAA,CAAO,wBAAwB,wBAAA,EAA0B,OAAA,CAAQ,GAAA,EAAK,CAAA,CACtE,WAAA;AAAA,IACC,OAAA;AAAA,IACA;AAAA;AAAA;AAAA;;AAAA;AAAA;;AAAA;AAAA;;AAAA;AAAA;;AAAA;AAAA;;AAAA;;AAAA;AAAA,IAAA;AAAA,IAsBD,KAAA,EAAM;AAET,EAAA,MAAM,OAAA,GAAU,QAAQ,IAAA,EAAK;AAE7B,EAAA,IAAI;AACF,IAAA,IAAI,YAAY,OAAA,CAAQ,MAAA;AACxB,IAAA,IAAI,CAAC,SAAA,EAAW;AACd,MAAA,MAAM,SAAA,GAAY,oBAAA,CAAqB,OAAA,CAAQ,UAAU,CAAA;AACzD,MAAA,IAAI,SAAA,EAAW;AACb,QAAA,SAAA,GAAY,SAAA;AACZ,QAAA,GAAA,CAAI,KAAK,CAAA,gCAAA,EAAmC,GAAA,CAAI,IAAA,CAAK,SAAS,CAAC,CAAA,CAAE,CAAA;AACjE,QAAA,GAAA,CAAI,KAAK,EAAE,CAAA;AAAA,MACb,CAAA,MAAO;AACL,QAAA,GAAA,CAAI,MAAM,iCAAiC,CAAA;AAC3C,QAAA,GAAA,CAAI,IAAA,CAAK,GAAA,CAAI,MAAA,CAAO,sCAAsC,CAAC,CAAA;AAC3D,QAAA,GAAA,CAAI,IAAA,CAAK,GAAA,CAAI,MAAA,CAAO,iEAAiE,CAAC,CAAA;AACtF,QAAA,OAAA,CAAQ,KAAK,CAAC,CAAA;AAAA,MAChB;AAAA,IACF;AAEA,IAAA,MAAM,MAAA,GAAwB;AAAA,MAC5B,MAAA,EAAQ,SAAA;AAAA,MACR,MAAM,OAAA,CAAQ,IAAA;AAAA,MACd,WAAW,OAAA,CAAQ,SAAA;AAAA,MACnB,eAAe,OAAA,CAAQ,aAAA;AAAA,MACvB,gBAAA,EAAkB,CAAC,OAAA,CAAQ,eAAA;AAAA,MAC3B,eAAe,OAAA,CAAQ,UAAA;AAAA,MACvB,SAAS,OAAA,CAAQ,OAAA;AAAA,MACjB,YAAY,OAAA,CAAQ;AAAA,KACtB;AAEA,IAAA,MAAM,WAAW,MAAM,CAAA;AAAA,EACzB,SAAS,KAAA,EAAO;AACd,IAAA,GAAA,CAAI,KAAK,EAAE,CAAA;AACX,IAAA,GAAA,CAAI,KAAA,CAAM,CAAA,aAAA,EAAgB,KAAK,CAAA,CAAE,CAAA;AACjC,IAAA,OAAA,CAAQ,KAAK,CAAC,CAAA;AAAA,EAChB;AACF;AAEA,IAAI,SAAA,CAAQ,SAAS,MAAA,EAAQ;AAC3B,EAAA,IAAA,EAAK;AACP","file":"index.mjs","sourcesContent":["/**\n * Peam Static Site Indexer\n *\n * Scans a static site output directory, discovers HTML files,\n * parses them into structured pages, and creates a searchable index.\n */\n\nimport { filePathToPathname, loadRobotsTxt, parseHTML, shouldIncludePath, type StructuredPage } from '@peam-ai/parser';\nimport { buildSearchIndex } from '@peam-ai/search';\nimport chalk from 'chalk';\nimport { Command } from 'commander';\nimport fg from 'fast-glob';\nimport { existsSync, mkdirSync, readFileSync, writeFileSync } from 'fs';\nimport { join, relative } from 'path';\n\nconst packageJson = JSON.parse(readFileSync(join(__dirname, '../package.json'), 'utf-8'));\n\ninterface IndexerConfig {\n  source: string;\n  outputDir: string;\n  indexFilename: string;\n  respectRobotsTxt: boolean;\n  robotsTxtPath?: string;\n  exclude: string[];\n  glob: string;\n  projectDir: string;\n}\n\ninterface DiscoveredPage {\n  pathname: string;\n  htmlFilePath: string;\n  relativeHtmlPath: string;\n}\n\nfunction parseExcludePatterns(value: string): string[] {\n  return value\n    .split(',')\n    .map((p) => p.trim())\n    .filter(Boolean);\n}\n\nconst log = {\n  success: (message: string) => console.log(chalk.green('✓ ') + message),\n  error: (message: string) => console.error(chalk.red('✗ ') + message),\n  warn: (message: string) => console.warn(chalk.yellow('⚠ ') + message),\n  info: (message: string) => console.log(chalk.blue('ℹ ') + message),\n  text: (message: string) => console.log(message),\n\n  cyan: (text: string) => chalk.cyan(text),\n  gray: (text: string) => chalk.gray(text),\n  bold: (text: string) => chalk.bold(text),\n  yellow: (text: string) => chalk.yellow(text),\n  red: (text: string) => chalk.red(text),\n  green: (text: string) => chalk.green(text),\n};\n\nasync function discoverHtmlFiles(\n  sourceDir: string,\n  globPattern: string,\n  projectDir: string\n): Promise<DiscoveredPage[]> {\n  const pages: DiscoveredPage[] = [];\n\n  try {\n    const sourceFullPath = join(projectDir, sourceDir);\n\n    const htmlFiles = await fg(globPattern, {\n      cwd: sourceFullPath,\n      absolute: true,\n      onlyFiles: true,\n      dot: false,\n      ignore: ['**/_next/**', '**/_astro/**', '**/404.{html,htm}', '**/500.{html,htm}'],\n    });\n\n    for (const htmlFilePath of htmlFiles) {\n      const relativePath = relative(sourceFullPath, htmlFilePath);\n      const pathname = filePathToPathname(relativePath);\n\n      pages.push({\n        pathname,\n        htmlFilePath,\n        relativeHtmlPath: relativePath,\n      });\n    }\n  } catch (error) {\n    log.warn(`Error discovering HTML files: ${error}`);\n  }\n\n  return pages;\n}\n\nasync function indexPages(config: IndexerConfig): Promise<void> {\n  log.text('\\n' + log.bold(log.cyan('Peam Static Site Indexer')) + '\\n');\n  log.text(log.bold('Configuration:'));\n  log.text(`  Project Directory: ${log.gray(config.projectDir)}`);\n  log.text(`  Source Directory: ${log.gray(config.source)}`);\n  log.text(`  Glob Pattern: ${log.gray(config.glob)}`);\n  log.text(`  Output Directory: ${log.gray(config.outputDir)}`);\n  log.text(`  Index Filename: ${log.gray(config.indexFilename)}`);\n  log.text(`  Respect robots.txt: ${log.gray(config.respectRobotsTxt.toString())}`);\n  if (config.exclude.length > 0) {\n    log.text(`  Exclude Patterns: ${log.gray(config.exclude.join(', '))}`);\n  }\n  log.text('');\n\n  const sourcePath = join(config.projectDir, config.source);\n  if (!existsSync(sourcePath)) {\n    log.error(`Source directory not found: ${sourcePath}`);\n    log.text(log.yellow('   Please build your site first or specify the correct --source directory'));\n    process.exit(1);\n  }\n\n  const searchPaths = [join(config.source, 'robots.txt'), 'public/robots.txt', 'robots.txt'];\n\n  const robotsResult = config.respectRobotsTxt\n    ? loadRobotsTxt(config.projectDir, searchPaths, config.robotsTxtPath)\n    : null;\n\n  if (robotsResult && config.respectRobotsTxt) {\n    log.info(`robots.txt loaded from ${log.cyan(relative(config.projectDir, robotsResult.path))}`);\n  } else if (config.respectRobotsTxt) {\n    log.info('No robots.txt found, all paths will be indexed');\n  }\n\n  log.text(log.bold('Discovering HTML files...'));\n  log.text('');\n  log.text(`  Scanning: ${log.gray(config.source)}`);\n  log.text(`  Pattern: ${log.gray(config.glob)}`);\n  log.text('');\n\n  const discoveredPages = await discoverHtmlFiles(config.source, config.glob, config.projectDir);\n\n  if (discoveredPages.length === 0) {\n    log.text('');\n    log.warn('No HTML files found');\n    log.text(log.yellow('   Check that your source directory contains HTML files'));\n    log.text(log.yellow(`   Source: ${sourcePath}`));\n    log.text(log.yellow(`   Pattern: ${config.glob}`));\n    process.exit(1);\n  }\n\n  const uniquePages = new Map<string, DiscoveredPage>();\n  for (const page of discoveredPages) {\n    if (!uniquePages.has(page.pathname)) {\n      uniquePages.set(page.pathname, page);\n    }\n  }\n\n  log.text('');\n  log.success(`Found ${log.bold(uniquePages.size.toString())} unique pages`);\n  log.text('');\n  log.text(log.bold('Processing pages...'));\n  log.text('');\n\n  const processedPages: Array<{\n    path: string;\n    htmlFile: string;\n    structuredPage: StructuredPage;\n  }> = [];\n\n  for (const [pathname, page] of uniquePages) {\n    const result = shouldIncludePath(pathname, robotsResult?.parser ?? null, config.exclude, config.respectRobotsTxt);\n\n    if (!result.included) {\n      // Log the specific reason for exclusion\n      if (result.reason === 'robots-txt') {\n        log.error(`Excluded by robots.txt: ${pathname}`);\n      } else if (result.reason === 'exclude-pattern') {\n        log.error(`Excluded by pattern: ${pathname}`);\n      }\n      continue;\n    }\n\n    try {\n      const html = readFileSync(page.htmlFilePath, 'utf-8');\n\n      const structuredPage = parseHTML(html);\n\n      if (!structuredPage) {\n        log.warn(`No content extracted from ${log.gray(pathname)}`);\n        continue;\n      }\n\n      log.success(`${log.cyan(pathname)}`);\n\n      processedPages.push({\n        path: pathname,\n        htmlFile: page.relativeHtmlPath,\n        structuredPage,\n      });\n    } catch (error) {\n      log.error(`Error processing ${log.gray(pathname)}: ${error}`);\n    }\n  }\n\n  log.text('');\n  log.success(`Successfully processed ${log.bold(processedPages.length.toString())} pages`);\n  log.text('');\n  log.text(log.bold('Creating search index...'));\n  log.text('');\n\n  const searchIndexData = await buildSearchIndex(processedPages);\n\n  log.success(`Added ${log.bold(processedPages.length.toString())} pages to search index`);\n  log.text('');\n  log.text(log.bold('Saving index...'));\n  log.text('');\n\n  const outputPath = join(config.projectDir, config.outputDir);\n  mkdirSync(outputPath, { recursive: true });\n\n  const generatedPath = join(outputPath, 'generated');\n  mkdirSync(generatedPath, { recursive: true });\n\n  const searchIndexFile = join(generatedPath, config.indexFilename);\n  writeFileSync(searchIndexFile, JSON.stringify(searchIndexData));\n\n  const indexJsContent = `// Auto-generated by Peam - DO NOT EDIT THIS FILE\nimport index from \"./generated/${config.indexFilename}\";\nexport default index;\n`;\n  const indexJsFile = join(outputPath, 'index.js');\n  writeFileSync(indexJsFile, indexJsContent);\n\n  log.success(`Index saved to: ${log.cyan(relative(config.projectDir, searchIndexFile))}`);\n  log.text(`  Total pages indexed: ${log.bold(processedPages.length.toString())}`);\n  log.text(\n    `  Index size: ${log.bold((Buffer.byteLength(JSON.stringify(searchIndexData), 'utf8') / 1024).toFixed(2))} KB`\n  );\n  log.text('');\n  log.success(log.bold('Indexing complete!'));\n  log.text('');\n}\n\nfunction probeSourceDirectory(projectDir: string): string | null {\n  const commonDirs = ['.next', '.build', '.out'];\n\n  for (const dir of commonDirs) {\n    const fullPath = join(projectDir, dir);\n    if (existsSync(fullPath)) {\n      return dir;\n    }\n  }\n\n  return null;\n}\n\nasync function main() {\n  const program = new Command();\n\n  program\n    .name('peam')\n    .description('Peam static site indexer (Next.js, Hugo, etc.)')\n    .version(packageJson.version)\n    .option('--source <path>', 'Source directory containing HTML files (auto-detected if not provided)')\n    .option('--glob <pattern>', 'Glob pattern for HTML files', '**/*.{html,htm}')\n    .option('--output-dir <path>', 'Output directory for index', '.peam')\n    .option('--index-filename <name>', 'Name of index file', 'index.json')\n    .option('--ignore-robots-txt', 'Disable robots.txt checking')\n    .option('--robots-path <path>', 'Custom path to robots.txt file')\n    .option('--exclude <patterns>', 'Comma-separated exclude patterns', parseExcludePatterns, [])\n    .option('--project-dir <path>', 'Project root directory', process.cwd())\n    .addHelpText(\n      'after',\n      `\nExamples:\n  # Auto-detect build directory\n  $ peam\n\n  # Hugo\n  $ peam --source public\n\n  # Next.js\n  $ peam --source .next\n\n  # Custom output directory\n  $ peam --source dist --glob \"**/*.html\"\n\n  # Exclude patterns\n  $ peam --exclude \"/admin/**,/api/*,/private-*\"\n\nFor Next.js 15+, the @peam-ai/next integration is recommended for production use.\n\nMore information: https://peam.ai\n    `\n    )\n    .parse();\n\n  const options = program.opts();\n\n  try {\n    let sourceDir = options.source;\n    if (!sourceDir) {\n      const probedDir = probeSourceDirectory(options.projectDir);\n      if (probedDir) {\n        sourceDir = probedDir;\n        log.info(`Auto-detected source directory: ${log.cyan(probedDir)}`);\n        log.text('');\n      } else {\n        log.error('No build output directory found');\n        log.text(log.yellow('   Searched for: .next, .build, .out'));\n        log.text(log.yellow('   Please build your site first or specify --source <directory>'));\n        process.exit(1);\n      }\n    }\n\n    const config: IndexerConfig = {\n      source: sourceDir,\n      glob: options.glob,\n      outputDir: options.outputDir,\n      indexFilename: options.indexFilename,\n      respectRobotsTxt: !options.ignoreRobotsTxt,\n      robotsTxtPath: options.robotsPath,\n      exclude: options.exclude,\n      projectDir: options.projectDir,\n    };\n\n    await indexPages(config);\n  } catch (error) {\n    log.text('');\n    log.error(`Fatal error: ${error}`);\n    process.exit(1);\n  }\n}\n\nif (require.main === module) {\n  main();\n}\n\nexport { indexPages, type IndexerConfig };\n"]}

package/dist/logger.d.mts

@@ -0,0 +1 @@
+export { loggers } from '@peam-ai/logger';

package/dist/logger.d.ts

@@ -0,0 +1 @@
+export { loggers } from '@peam-ai/logger';

package/dist/logger.js

@@ -0,0 +1,12 @@
+'use strict';
+
+var logger = require('@peam-ai/logger');
+
+
+
+Object.defineProperty(exports, "loggers", {
+  enumerable: true,
+  get: function () { return logger.loggers; }
+});
+//# sourceMappingURL=logger.js.map
+//# sourceMappingURL=logger.js.map

package/dist/logger.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":[],"names":[],"mappings":"","file":"logger.js","sourcesContent":[]}

package/dist/logger.mjs

@@ -0,0 +1,4 @@
+import './chunk-BJTO5JO5.mjs';
+export { loggers } from '@peam-ai/logger';
+//# sourceMappingURL=logger.mjs.map
+//# sourceMappingURL=logger.mjs.map

package/dist/logger.mjs.map

@@ -0,0 +1 @@
+{"version":3,"sources":[],"names":[],"mappings":"","file":"logger.mjs","sourcesContent":[]}

package/dist/parser.d.mts

@@ -0,0 +1 @@
+export { RobotsTxtResult, StructuredPage, createRobotsParser, loadRobotsTxt, parseHTML, shouldIncludePath } from '@peam-ai/parser';

package/dist/parser.d.ts

@@ -0,0 +1 @@
+export { RobotsTxtResult, StructuredPage, createRobotsParser, loadRobotsTxt, parseHTML, shouldIncludePath } from '@peam-ai/parser';

package/dist/parser.js

@@ -0,0 +1,24 @@
+'use strict';
+
+var parser = require('@peam-ai/parser');
+
+
+
+Object.defineProperty(exports, "createRobotsParser", {
+  enumerable: true,
+  get: function () { return parser.createRobotsParser; }
+});
+Object.defineProperty(exports, "loadRobotsTxt", {
+  enumerable: true,
+  get: function () { return parser.loadRobotsTxt; }
+});
+Object.defineProperty(exports, "parseHTML", {
+  enumerable: true,
+  get: function () { return parser.parseHTML; }
+});
+Object.defineProperty(exports, "shouldIncludePath", {
+  enumerable: true,
+  get: function () { return parser.shouldIncludePath; }
+});
+//# sourceMappingURL=parser.js.map
+//# sourceMappingURL=parser.js.map

package/dist/parser.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":[],"names":[],"mappings":"","file":"parser.js","sourcesContent":[]}