@xbrowser/cli 1.6.0 → 1.6.2

package/dist/cli.js

@@ -2178,10 +2178,45 @@ function createTurndown() {
     },
     replacement: () => ""
   });
+  turndown.addRule("complexTables", {
+    filter: (node) => {
+      if (typeof node !== "object" || node === null) return false;
+      const n = node;
+      return n.nodeName?.toLowerCase() === "table";
+    },
+    replacement: (_content, node) => {
+      const html = node.outerHTML || "";
+      if (!html) return "";
+      const rows = [];
+      const rowRegex = /<tr[^>]*>([\s\S]*?)<\/tr>/gi;
+      let rowMatch;
+      while ((rowMatch = rowRegex.exec(html)) !== null) {
+        const cells = [];
+        const cellRegex = /<t[dh][^>]*>([\s\S]*?)<\/t[dh]>/gi;
+        let cellMatch;
+        while ((cellMatch = cellRegex.exec(rowMatch[1])) !== null) {
+          cells.push(cellMatch[1].replace(/<[^>]+>/g, "").trim().replace(/\n/g, " "));
+        }
+        if (cells.length > 0) rows.push(cells);
+      }
+      if (rows.length === 0) return "";
+      const mdRows = rows.map((cells) => {
+        const escaped = cells.map((c) => c.replace(/\|/g, "\\|") || "");
+        return `| ${escaped.join(" | ")} |`;
+      });
+      const hasHeader = /<th/i.test(html);
+      if (hasHeader && mdRows.length > 0) {
+        const colCount = rows[0].length;
+        const sep = `| ${Array(colCount).fill("---").join(" | ")} |`;
+        mdRows.splice(1, 0, sep);
+      }
+      return "\n\n" + mdRows.join("\n") + "\n\n";
+    }
+  });
   return turndown;
 }
 function postClean(md) {
-  md = md.replace(/<(?:table|div|tbody|thead|tr|td|th|span|colgroup|col)\b[^>]*(?:>[\s\S]{200,}?<\/(?:table|div|tbody|thead|tr|td|th|span|colgroup|col)>)/g, "\n[\u26A0\uFE0F HTML block removed \u2014 complex table/layout not converted to Markdown]\n");
+  md = md.replace(/<[^>]+>/g, "");
   md = md.replace(/\n{3,}/g, "\n\n");
   md = md.replace(/!\[[^\]]*\]\(\s*\)/g, "");
   md = md.replace(/\[([^\]]*)\]\(\s*\)/g, "$1");
@@ -12523,7 +12558,7 @@ async function handleEvalMode(argv) {
 }
 async function handleChainInput(input, argv) {
   const cdpEndpoint = argv ? extractCdpFromArgv(argv) : void 0;
-  const jsonMode = argv ? argv.includes("--json") || argv.includes("-j") : false;
+  const jsonMode = argv ? argv.some((a) => a === "--json" || a.startsWith("--json=") || a.includes(" --json") || a.startsWith("--json")) || argv.includes("-j") : false;
   const chainResult = await executeChain(input, { cdpEndpoint });
   if (jsonMode) {
     const output = {

package/dist/daemon-main.js

@@ -2139,10 +2139,45 @@ function createTurndown() {
     },
     replacement: () => ""
   });
+  turndown.addRule("complexTables", {
+    filter: (node) => {
+      if (typeof node !== "object" || node === null) return false;
+      const n = node;
+      return n.nodeName?.toLowerCase() === "table";
+    },
+    replacement: (_content, node) => {
+      const html = node.outerHTML || "";
+      if (!html) return "";
+      const rows = [];
+      const rowRegex = /<tr[^>]*>([\s\S]*?)<\/tr>/gi;
+      let rowMatch;
+      while ((rowMatch = rowRegex.exec(html)) !== null) {
+        const cells = [];
+        const cellRegex = /<t[dh][^>]*>([\s\S]*?)<\/t[dh]>/gi;
+        let cellMatch;
+        while ((cellMatch = cellRegex.exec(rowMatch[1])) !== null) {
+          cells.push(cellMatch[1].replace(/<[^>]+>/g, "").trim().replace(/\n/g, " "));
+        }
+        if (cells.length > 0) rows.push(cells);
+      }
+      if (rows.length === 0) return "";
+      const mdRows = rows.map((cells) => {
+        const escaped = cells.map((c) => c.replace(/\|/g, "\\|") || "");
+        return `| ${escaped.join(" | ")} |`;
+      });
+      const hasHeader = /<th/i.test(html);
+      if (hasHeader && mdRows.length > 0) {
+        const colCount = rows[0].length;
+        const sep = `| ${Array(colCount).fill("---").join(" | ")} |`;
+        mdRows.splice(1, 0, sep);
+      }
+      return "\n\n" + mdRows.join("\n") + "\n\n";
+    }
+  });
   return turndown;
 }
 function postClean(md) {
-  md = md.replace(/<(?:table|div|tbody|thead|tr|td|th|span|colgroup|col)\b[^>]*(?:>[\s\S]{200,}?<\/(?:table|div|tbody|thead|tr|td|th|span|colgroup|col)>)/g, "\n[\u26A0\uFE0F HTML block removed \u2014 complex table/layout not converted to Markdown]\n");
+  md = md.replace(/<[^>]+>/g, "");
   md = md.replace(/\n{3,}/g, "\n\n");
   md = md.replace(/!\[[^\]]*\]\(\s*\)/g, "");
   md = md.replace(/\[([^\]]*)\]\(\s*\)/g, "$1");

package/dist/index.js

@@ -2218,10 +2218,45 @@ function createTurndown() {
     },
     replacement: () => ""
   });
+  turndown.addRule("complexTables", {
+    filter: (node) => {
+      if (typeof node !== "object" || node === null) return false;
+      const n = node;
+      return n.nodeName?.toLowerCase() === "table";
+    },
+    replacement: (_content, node) => {
+      const html = node.outerHTML || "";
+      if (!html) return "";
+      const rows = [];
+      const rowRegex = /<tr[^>]*>([\s\S]*?)<\/tr>/gi;
+      let rowMatch;
+      while ((rowMatch = rowRegex.exec(html)) !== null) {
+        const cells = [];
+        const cellRegex = /<t[dh][^>]*>([\s\S]*?)<\/t[dh]>/gi;
+        let cellMatch;
+        while ((cellMatch = cellRegex.exec(rowMatch[1])) !== null) {
+          cells.push(cellMatch[1].replace(/<[^>]+>/g, "").trim().replace(/\n/g, " "));
+        }
+        if (cells.length > 0) rows.push(cells);
+      }
+      if (rows.length === 0) return "";
+      const mdRows = rows.map((cells) => {
+        const escaped = cells.map((c) => c.replace(/\|/g, "\\|") || "");
+        return `| ${escaped.join(" | ")} |`;
+      });
+      const hasHeader = /<th/i.test(html);
+      if (hasHeader && mdRows.length > 0) {
+        const colCount = rows[0].length;
+        const sep = `| ${Array(colCount).fill("---").join(" | ")} |`;
+        mdRows.splice(1, 0, sep);
+      }
+      return "\n\n" + mdRows.join("\n") + "\n\n";
+    }
+  });
   return turndown;
 }
 function postClean(md) {
-  md = md.replace(/<(?:table|div|tbody|thead|tr|td|th|span|colgroup|col)\b[^>]*(?:>[\s\S]{200,}?<\/(?:table|div|tbody|thead|tr|td|th|span|colgroup|col)>)/g, "\n[\u26A0\uFE0F HTML block removed \u2014 complex table/layout not converted to Markdown]\n");
+  md = md.replace(/<[^>]+>/g, "");
   md = md.replace(/\n{3,}/g, "\n\n");
   md = md.replace(/!\[[^\]]*\]\(\s*\)/g, "");
   md = md.replace(/\[([^\]]*)\]\(\s*\)/g, "$1");
@@ -12846,7 +12881,7 @@ async function handleEvalMode(argv) {
 }
 async function handleChainInput(input, argv) {
   const cdpEndpoint = argv ? extractCdpFromArgv(argv) : void 0;
-  const jsonMode = argv ? argv.includes("--json") || argv.includes("-j") : false;
+  const jsonMode = argv ? argv.some((a) => a === "--json" || a.startsWith("--json=") || a.includes(" --json") || a.startsWith("--json")) || argv.includes("-j") : false;
   const chainResult = await executeChain(input, { cdpEndpoint });
   if (jsonMode) {
     const output = {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@xbrowser/cli",
-  "version": "1.6.0",
+  "version": "1.6.2",
   "description": "Browser automation CLI for web scraping, headless browsing, SEO analysis, and AI agent workflows. A command-line alternative to Playwright, Puppeteer, and Selenium.",
   "type": "module",
   "bin": {