whale-code 6.5.5 → 6.5.7

package/dist/cli/services/tools/web-tools.js

@@ -3,393 +3,440 @@
  *
  * Extracted from local-tools.ts for single-responsibility.
  */
+
 import { lookup } from "node:dns/promises";
 import { getValidToken, createAuthenticatedClient } from "../auth-service.js";
 import { resolveConfig } from "../config-store.js";
 // ============================================================================
 // SSRF PROTECTION
 // ============================================================================
+
 export function isBlockedUrl(urlStr) {
-    try {
-        const u = new URL(urlStr);
-        const host = u.hostname.toLowerCase();
-        // Block localhost
-        if (host === "localhost" || host === "127.0.0.1" || host === "::1" || host === "[::1]" || host === "0.0.0.0")
-            return true;
-        // Block private IPv4 ranges
-        if (/^10\./.test(host))
-            return true;
-        if (/^172\.(1[6-9]|2\d|3[01])\./.test(host))
-            return true;
-        if (/^192\.168\./.test(host))
-            return true;
-        // Block IPv6 private
-        if (/^fe80:/i.test(host) || /^\[fe80:/i.test(host))
-            return true;
-        if (/^fc00:/i.test(host) || /^\[fc00:/i.test(host))
-            return true;
-        if (/^fd/i.test(host) || /^\[fd/i.test(host))
-            return true;
-        // Block cloud metadata
-        if (host === "169.254.169.254" || /^169\.254\./.test(host))
-            return true;
-        // Block IPv6-mapped IPv4 (e.g. ::ffff:127.0.0.1, ::ffff:169.254.169.254)
-        if (host.includes("::ffff:"))
-            return true;
-        // Block decimal/hex IP representations (e.g. 2130706433 = 127.0.0.1, 0x7f000001)
-        if (/^\d+$/.test(host) || /^0x[0-9a-f]+$/i.test(host))
-            return true;
-        // Block octal IPs (e.g., 0177.0.0.1 = 127.0.0.1)
-        if (/^0\d+\./.test(host))
-            return true;
-        // Block short zero address forms (e.g. 0, 0.0, 0.0.0)
-        if (/^0(\.0)*$/.test(host))
-            return true;
-        // Block internal TLDs
-        if (host.endsWith(".internal") || host.endsWith(".local"))
-            return true;
-        // Block non-HTTP(S)
-        if (u.protocol !== "http:" && u.protocol !== "https:")
-            return true;
-        return false;
-    }
-    catch {
-        return true;
-    }
+  try {
+    const u = new URL(urlStr);
+    const host = u.hostname.toLowerCase();
+
+    // Block localhost
+    if (host === "localhost" || host === "127.0.0.1" || host === "::1" || host === "[::1]" || host === "0.0.0.0") return true;
+
+    // Block private IPv4 ranges
+    if (/^10\./.test(host)) return true;
+    if (/^172\.(1[6-9]|2\d|3[01])\./.test(host)) return true;
+    if (/^192\.168\./.test(host)) return true;
+
+    // Block IPv6 private
+    if (/^fe80:/i.test(host) || /^\[fe80:/i.test(host)) return true;
+    if (/^fc00:/i.test(host) || /^\[fc00:/i.test(host)) return true;
+    if (/^fd/i.test(host) || /^\[fd/i.test(host)) return true;
+
+    // Block cloud metadata
+    if (host === "169.254.169.254" || /^169\.254\./.test(host)) return true;
+
+    // Block IPv6-mapped IPv4 (e.g. ::ffff:127.0.0.1, ::ffff:169.254.169.254)
+    if (host.includes("::ffff:")) return true;
+
+    // Block decimal/hex IP representations (e.g. 2130706433 = 127.0.0.1, 0x7f000001)
+    if (/^\d+$/.test(host) || /^0x[0-9a-f]+$/i.test(host)) return true;
+
+    // Block octal IPs (e.g., 0177.0.0.1 = 127.0.0.1)
+    if (/^0\d+\./.test(host)) return true;
+
+    // Block short zero address forms (e.g. 0, 0.0, 0.0.0)
+    if (/^0(\.0)*$/.test(host)) return true;
+
+    // Block internal TLDs
+    if (host.endsWith(".internal") || host.endsWith(".local")) return true;
+
+    // Block non-HTTP(S)
+    if (u.protocol !== "http:" && u.protocol !== "https:") return true;
+    return false;
+  } catch {
+    return true;
+  }
 }
 async function resolveAndCheckUrl(url) {
-    if (isBlockedUrl(url))
-        return true;
-    try {
-        const { hostname } = new URL(url);
-        // Skip IP-based hostnames (already checked by isBlockedUrl)
-        if (/^[\d.]+$/.test(hostname) || hostname.includes(":"))
-            return false;
-        const { address } = await lookup(hostname);
-        // IPv6 private/loopback detection
-        if (address.includes(":")) {
-            const lower = address.toLowerCase();
-            if (lower === "::1" || lower === "::" || lower.startsWith("fe80") ||
-                lower.startsWith("fc") || lower.startsWith("fd") ||
-                lower.startsWith("::ffff:"))
-                return true;
-            return false;
-        }
-        // IPv4 private/internal ranges
-        const parts = address.split(".").map(Number);
-        if (parts[0] === 10)
-            return true; // 10.0.0.0/8
-        if (parts[0] === 172 && parts[1] >= 16 && parts[1] <= 31)
-            return true; // 172.16.0.0/12
-        if (parts[0] === 192 && parts[1] === 168)
-            return true; // 192.168.0.0/16
-        if (parts[0] === 127)
-            return true; // 127.0.0.0/8
-        if (parts[0] === 169 && parts[1] === 254)
-            return true; // link-local
-        if (address === "0.0.0.0")
-            return true;
-        return false;
-    }
-    catch {
-        return false; // DNS resolution failed, allow (will fail at fetch anyway)
+  if (isBlockedUrl(url)) return true;
+  try {
+    const {
+      hostname
+    } = new URL(url);
+    // Skip IP-based hostnames (already checked by isBlockedUrl)
+    if (/^[\d.]+$/.test(hostname) || hostname.includes(":")) return false;
+    const {
+      address
+    } = await lookup(hostname);
+    // IPv6 private/loopback detection
+    if (address.includes(":")) {
+      const lower = address.toLowerCase();
+      if (lower === "::1" || lower === "::" || lower.startsWith("fe80") || lower.startsWith("fc") || lower.startsWith("fd") || lower.startsWith("::ffff:")) return true;
+      return false;
     }
+    // IPv4 private/internal ranges
+    const parts = address.split(".").map(Number);
+    if (parts[0] === 10) return true; // 10.0.0.0/8
+    if (parts[0] === 172 && parts[1] >= 16 && parts[1] <= 31) return true; // 172.16.0.0/12
+    if (parts[0] === 192 && parts[1] === 168) return true; // 192.168.0.0/16
+    if (parts[0] === 127) return true; // 127.0.0.0/8
+    if (parts[0] === 169 && parts[1] === 254) return true; // link-local
+    if (address === "0.0.0.0") return true;
+    return false;
+  } catch {
+    return false; // DNS resolution failed, allow (will fail at fetch anyway)
+  }
 }
+
 // ============================================================================
 // WEB FETCH
 // ============================================================================
+
 export async function webFetch(input) {
-    const url = input.url;
-    if (!url)
-        return { success: false, output: "url is required" };
-    if (await resolveAndCheckUrl(url)) {
-        return { success: false, output: "URL blocked: cannot fetch localhost, private IPs, or internal addresses" };
+  const url = input.url;
+  if (!url) return {
+    success: false,
+    output: "url is required"
+  };
+  if (await resolveAndCheckUrl(url)) {
+    return {
+      success: false,
+      output: "URL blocked: cannot fetch localhost, private IPs, or internal addresses"
+    };
+  }
+  try {
+    const controller = new AbortController();
+    const fetchTimer = setTimeout(() => controller.abort(), 30000); // 30s timeout
+    let response = await fetch(url, {
+      headers: {
+        "User-Agent": "WhaleCode/3.0 (CLI Agent)",
+        "Accept": "text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8"
+      },
+      signal: controller.signal,
+      redirect: "manual"
+    });
+    clearTimeout(fetchTimer);
+
+    // Handle redirects manually to prevent SSRF via redirect to internal addresses
+    if (response.status >= 300 && response.status < 400) {
+      const location = response.headers.get("location");
+      if (!location) {
+        return {
+          success: false,
+          output: `Redirect ${response.status} with no Location header`
+        };
+      }
+      // Resolve relative redirects against the original URL
+      const resolvedLocation = new URL(location, url).toString();
+      if (await resolveAndCheckUrl(resolvedLocation)) {
+        return {
+          success: false,
+          output: "Redirect target is a blocked address"
+        };
+      }
+      const redirectController = new AbortController();
+      const redirectTimer = setTimeout(() => redirectController.abort(), 30000);
+      response = await fetch(resolvedLocation, {
+        headers: {
+          "User-Agent": "WhaleCode/3.0 (CLI Agent)",
+          "Accept": "text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8"
+        },
+        signal: redirectController.signal,
+        redirect: "manual"
+      });
+      clearTimeout(redirectTimer);
+      // If still redirecting, give up rather than following an unbounded chain
+      if (response.status >= 300 && response.status < 400) {
+        return {
+          success: false,
+          output: "Too many redirects"
+        };
+      }
     }
-    try {
-        const controller = new AbortController();
-        const fetchTimer = setTimeout(() => controller.abort(), 30000); // 30s timeout
-        let response = await fetch(url, {
-            headers: {
-                "User-Agent": "WhaleCode/3.0 (CLI Agent)",
-                "Accept": "text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8",
-            },
-            signal: controller.signal,
-            redirect: "manual",
-        });
-        clearTimeout(fetchTimer);
-        // Handle redirects manually to prevent SSRF via redirect to internal addresses
-        if (response.status >= 300 && response.status < 400) {
-            const location = response.headers.get("location");
-            if (!location) {
-                return { success: false, output: `Redirect ${response.status} with no Location header` };
-            }
-            // Resolve relative redirects against the original URL
-            const resolvedLocation = new URL(location, url).toString();
-            if (await resolveAndCheckUrl(resolvedLocation)) {
-                return { success: false, output: "Redirect target is a blocked address" };
-            }
-            const redirectController = new AbortController();
-            const redirectTimer = setTimeout(() => redirectController.abort(), 30000);
-            response = await fetch(resolvedLocation, {
-                headers: {
-                    "User-Agent": "WhaleCode/3.0 (CLI Agent)",
-                    "Accept": "text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8",
-                },
-                signal: redirectController.signal,
-                redirect: "manual",
-            });
-            clearTimeout(redirectTimer);
-            // If still redirecting, give up rather than following an unbounded chain
-            if (response.status >= 300 && response.status < 400) {
-                return { success: false, output: "Too many redirects" };
-            }
-        }
-        if (!response.ok) {
-            return { success: false, output: `HTTP ${response.status}: ${response.statusText}` };
-        }
-        const contentType = response.headers.get("content-type") || "";
-        const body = await response.text();
-        // JSON — return as-is (pretty-printed)
-        if (contentType.includes("application/json")) {
-            try {
-                const parsed = JSON.parse(body);
-                const pretty = JSON.stringify(parsed, null, 2);
-                return { success: true, output: pretty.slice(0, 80000) };
-            }
-            catch {
-                return { success: true, output: body.slice(0, 80000) };
-            }
-        }
-        // Plain text — return as-is
-        if (contentType.includes("text/plain")) {
-            return { success: true, output: body.slice(0, 80000) };
-        }
-        // HTML — convert to readable text
-        const text = htmlToText(body);
-        const truncated = text.length > 500_000 ? text.slice(0, 500_000) + "\n\n... (safety truncated)" : text;
-        return { success: true, output: `# ${url}\n\n${truncated}` };
+    if (!response.ok) {
+      return {
+        success: false,
+        output: `HTTP ${response.status}: ${response.statusText}`
+      };
     }
-    catch (err) {
-        return { success: false, output: `Fetch error: ${err.message || err}` };
+    const contentType = response.headers.get("content-type") || "";
+    const body = await response.text();
+
+    // JSON — return as-is (pretty-printed)
+    if (contentType.includes("application/json")) {
+      try {
+        const parsed = JSON.parse(body);
+        const pretty = JSON.stringify(parsed, null, 2);
+        return {
+          success: true,
+          output: pretty.slice(0, 80000)
+        };
+      } catch {
+        return {
+          success: true,
+          output: body.slice(0, 80000)
+        };
+      }
     }
+
+    // Plain text — return as-is
+    if (contentType.includes("text/plain")) {
+      return {
+        success: true,
+        output: body.slice(0, 80000)
+      };
+    }
+
+    // HTML — convert to readable text
+    const text = htmlToText(body);
+    const truncated = text.length > 500_000 ? text.slice(0, 500_000) + "\n\n... (safety truncated)" : text;
+    return {
+      success: true,
+      output: `# ${url}\n\n${truncated}`
+    };
+  } catch (err) {
+    return {
+      success: false,
+      output: `Fetch error: ${err.message || err}`
+    };
+  }
 }
+
 // ============================================================================
 // HTML TO TEXT
 // ============================================================================
+
 /** Remove all instances of a tag and its content, handling nesting (innermost first) */
 function removeNestedTag(html, tag) {
-    const pattern = new RegExp(`<${tag}[^>]*>(?:(?!<${tag}[\\s>/])[\\s\\S])*?<\\/${tag}>`, "gi");
-    let result = html;
-    let prev = "";
-    let safety = 0;
-    while (result !== prev && safety++ < 50) {
-        prev = result;
-        result = result.replace(pattern, "");
-    }
-    result = result.replace(new RegExp(`<${tag}[^>]*>`, "gi"), "");
-    return result;
+  const pattern = new RegExp(`<${tag}[^>]*>(?:(?!<${tag}[\\s>/])[\\s\\S])*?<\\/${tag}>`, "gi");
+  let result = html;
+  let prev = "";
+  let safety = 0;
+  while (result !== prev && safety++ < 50) {
+    prev = result;
+    result = result.replace(pattern, "");
+  }
+  result = result.replace(new RegExp(`<${tag}[^>]*>`, "gi"), "");
+  return result;
 }
 function stripTags(html) {
-    return html.replace(/<[^>]+>/g, "");
+  return html.replace(/<[^>]+>/g, "");
 }
 function decodeEntities(text) {
-    return text
-        .replace(/&nbsp;/g, " ")
-        .replace(/&amp;/g, "&")
-        .replace(/&lt;/g, "<")
-        .replace(/&gt;/g, ">")
-        .replace(/&quot;/g, '"')
-        .replace(/&#39;/g, "'")
-        .replace(/&#x27;/g, "'")
-        .replace(/&#x([0-9a-fA-F]+);/g, (_, hex) => String.fromCharCode(parseInt(hex, 16)))
-        .replace(/&#(\d+);/g, (_, code) => String.fromCharCode(parseInt(code)));
+  return text.replace(/&nbsp;/g, " ").replace(/&amp;/g, "&").replace(/&lt;/g, "<").replace(/&gt;/g, ">").replace(/&quot;/g, '"').replace(/&#39;/g, "'").replace(/&#x27;/g, "'").replace(/&#x([0-9a-fA-F]+);/g, (_, hex) => String.fromCharCode(parseInt(hex, 16))).replace(/&#(\d+);/g, (_, code) => String.fromCharCode(parseInt(code)));
 }
+
 /** Enhanced HTML -> readable text/markdown converter */
 export function htmlToText(html) {
-    let c = html;
-    // 1. Extract main content area (skips nav, sidebar, footer automatically)
-    const mainMatch = c.match(/<main[^>]*>([\s\S]*)<\/main>/i)
-        || c.match(/<article[^>]*>([\s\S]*)<\/article>/i);
-    if (mainMatch) {
-        c = mainMatch[1];
-    }
-    else {
-        const bodyMatch = c.match(/<body[^>]*>([\s\S]*)<\/body>/i);
-        if (bodyMatch)
-            c = bodyMatch[1];
-    }
-    // 2. Remove non-content elements (nesting-aware)
-    for (const tag of [
-        "script", "style", "nav", "footer", "aside", "header",
-        "form", "svg", "iframe", "select", "button", "noscript",
-    ]) {
-        c = removeNestedTag(c, tag);
-    }
-    // 3. Remove HTML comments
-    c = c.replace(/<!--[\s\S]*?-->/g, "");
-    // 4. Convert semantic elements -> markdown
-    // Code blocks first (preserve contents)
-    c = c.replace(/<pre[^>]*>([\s\S]*?)<\/pre>/gi, (_, inner) => "\n```\n" + stripTags(inner).trim() + "\n```\n");
-    c = c.replace(/<code[^>]*>([\s\S]*?)<\/code>/gi, "`$1`");
-    // Headings
-    c = c.replace(/<h([1-6])[^>]*>([\s\S]*?)<\/h\1>/gi, (_, level, text) => "\n" + "#".repeat(parseInt(level)) + " " + stripTags(text).trim() + "\n\n");
-    // Links — skip empty, anchor-only, and javascript: hrefs
-    c = c.replace(/<a[^>]*href="([^"]*)"[^>]*>([\s\S]*?)<\/a>/gi, (_, href, text) => {
-        const linkText = stripTags(text).trim();
-        if (!linkText)
-            return "";
-        if (href.startsWith("#") || href.startsWith("javascript:"))
-            return linkText;
-        return `[${linkText}](${href})`;
-    });
-    // Tables -> pipe-delimited markdown
-    c = c.replace(/<table[^>]*>([\s\S]*?)<\/table>/gi, (_, tableContent) => {
-        const rows = [];
-        const rowRegex = /<tr[^>]*>([\s\S]*?)<\/tr>/gi;
-        let rowMatch;
-        let isFirstRow = true;
-        while ((rowMatch = rowRegex.exec(tableContent)) !== null) {
-            const cells = [];
-            const cellRegex = /<(?:td|th)[^>]*>([\s\S]*?)<\/(?:td|th)>/gi;
-            let cellMatch;
-            while ((cellMatch = cellRegex.exec(rowMatch[1])) !== null) {
-                cells.push(stripTags(cellMatch[1]).trim());
-            }
-            if (cells.length > 0) {
-                rows.push("| " + cells.join(" | ") + " |");
-                if (isFirstRow) {
-                    rows.push("| " + cells.map(() => "---").join(" | ") + " |");
-                    isFirstRow = false;
-                }
-            }
+  let c = html;
+
+  // 1. Extract main content area (skips nav, sidebar, footer automatically)
+  const mainMatch = c.match(/<main[^>]*>([\s\S]*)<\/main>/i) || c.match(/<article[^>]*>([\s\S]*)<\/article>/i);
+  if (mainMatch) {
+    c = mainMatch[1];
+  } else {
+    const bodyMatch = c.match(/<body[^>]*>([\s\S]*)<\/body>/i);
+    if (bodyMatch) c = bodyMatch[1];
+  }
+
+  // 2. Remove non-content elements (nesting-aware)
+  for (const tag of ["script", "style", "nav", "footer", "aside", "header", "form", "svg", "iframe", "select", "button", "noscript"]) {
+    c = removeNestedTag(c, tag);
+  }
+
+  // 3. Remove HTML comments
+  c = c.replace(/<!--[\s\S]*?-->/g, "");
+
+  // 4. Convert semantic elements -> markdown
+
+  // Code blocks first (preserve contents)
+  c = c.replace(/<pre[^>]*>([\s\S]*?)<\/pre>/gi, (_, inner) => "\n```\n" + stripTags(inner).trim() + "\n```\n");
+  c = c.replace(/<code[^>]*>([\s\S]*?)<\/code>/gi, "`$1`");
+
+  // Headings
+  c = c.replace(/<h([1-6])[^>]*>([\s\S]*?)<\/h\1>/gi, (_, level, text) => "\n" + "#".repeat(parseInt(level)) + " " + stripTags(text).trim() + "\n\n");
+
+  // Links — skip empty, anchor-only, and javascript: hrefs
+  c = c.replace(/<a[^>]*href="([^"]*)"[^>]*>([\s\S]*?)<\/a>/gi, (_, href, text) => {
+    const linkText = stripTags(text).trim();
+    if (!linkText) return "";
+    if (href.startsWith("#") || href.startsWith("javascript:")) return linkText;
+    return `[${linkText}](${href})`;
+  });
+
+  // Tables -> pipe-delimited markdown
+  c = c.replace(/<table[^>]*>([\s\S]*?)<\/table>/gi, (_, tableContent) => {
+    const rows = [];
+    const rowRegex = /<tr[^>]*>([\s\S]*?)<\/tr>/gi;
+    let rowMatch;
+    let isFirstRow = true;
+    while ((rowMatch = rowRegex.exec(tableContent)) !== null) {
+      const cells = [];
+      const cellRegex = /<(?:td|th)[^>]*>([\s\S]*?)<\/(?:td|th)>/gi;
+      let cellMatch;
+      while ((cellMatch = cellRegex.exec(rowMatch[1])) !== null) {
+        cells.push(stripTags(cellMatch[1]).trim());
+      }
+      if (cells.length > 0) {
+        rows.push("| " + cells.join(" | ") + " |");
+        if (isFirstRow) {
+          rows.push("| " + cells.map(() => "---").join(" | ") + " |");
+          isFirstRow = false;
         }
-        return "\n" + rows.join("\n") + "\n";
-    });
-    // List items
-    c = c.replace(/<li[^>]*>([\s\S]*?)<\/li>/gi, (_, text) => "- " + stripTags(text).trim() + "\n");
-    // Bold, italic
-    c = c.replace(/<(strong|b)[^>]*>([\s\S]*?)<\/\1>/gi, "**$2**");
-    c = c.replace(/<(em|i)[^>]*>([\s\S]*?)<\/\1>/gi, "*$2*");
-    // Images -> alt text
-    c = c.replace(/<img[^>]*alt="([^"]*)"[^>]*>/gi, "[$1]");
-    c = c.replace(/<img[^>]*>/gi, "");
-    // Horizontal rules
-    c = c.replace(/<hr\s*\/?>/gi, "\n---\n");
-    // 5. Block elements -> newlines (replace tags independently, NOT as pairs)
-    c = c.replace(/<\/(?:p|div|section|article|main|blockquote|dd|dt|figcaption|figure|details|summary)>/gi, "\n\n");
-    c = c.replace(/<(?:p|div|section|article|main|blockquote|dd|dt|figcaption|figure|details|summary)[^>]*>/gi, "");
-    c = c.replace(/<br\s*\/?>/gi, "\n");
-    c = c.replace(/<\/(?:li|tr|thead|tbody|tfoot|ul|ol|dl)>/gi, "\n");
-    // 6. Strip all remaining tags
-    c = c.replace(/<[^>]+>/g, "");
-    // 7. Decode HTML entities
-    c = decodeEntities(c);
-    // 8. Clean whitespace
-    c = c
-        .replace(/[ \t]+/g, " ")
-        .replace(/ *\n */g, "\n")
-        .replace(/\n{3,}/g, "\n\n")
-        .trim();
-    return c;
+      }
+    }
+    return "\n" + rows.join("\n") + "\n";
+  });
+
+  // List items
+  c = c.replace(/<li[^>]*>([\s\S]*?)<\/li>/gi, (_, text) => "- " + stripTags(text).trim() + "\n");
+
+  // Bold, italic
+  c = c.replace(/<(strong|b)[^>]*>([\s\S]*?)<\/\1>/gi, "**$2**");
+  c = c.replace(/<(em|i)[^>]*>([\s\S]*?)<\/\1>/gi, "*$2*");
+
+  // Images -> alt text
+  c = c.replace(/<img[^>]*alt="([^"]*)"[^>]*>/gi, "[$1]");
+  c = c.replace(/<img[^>]*>/gi, "");
+
+  // Horizontal rules
+  c = c.replace(/<hr\s*\/?>/gi, "\n---\n");
+
+  // 5. Block elements -> newlines (replace tags independently, NOT as pairs)
+  c = c.replace(/<\/(?:p|div|section|article|main|blockquote|dd|dt|figcaption|figure|details|summary)>/gi, "\n\n");
+  c = c.replace(/<(?:p|div|section|article|main|blockquote|dd|dt|figcaption|figure|details|summary)[^>]*>/gi, "");
+  c = c.replace(/<br\s*\/?>/gi, "\n");
+  c = c.replace(/<\/(?:li|tr|thead|tbody|tfoot|ul|ol|dl)>/gi, "\n");
+
+  // 6. Strip all remaining tags
+  c = c.replace(/<[^>]+>/g, "");
+
+  // 7. Decode HTML entities
+  c = decodeEntities(c);
+
+  // 8. Clean whitespace
+  c = c.replace(/[ \t]+/g, " ").replace(/ *\n */g, "\n").replace(/\n{3,}/g, "\n\n").trim();
+  return c;
 }
+
 // ============================================================================
 // WEB SEARCH (Exa API)
 // ============================================================================
+
 // Cache the Exa API key within a session
 let cachedExaKey = null;
 async function getExaApiKey() {
-    if (cachedExaKey)
-        return cachedExaKey;
-    try {
-        const config = resolveConfig();
-        // Tier 1: Service role key (MCP server mode)
-        if (config.supabaseUrl && config.supabaseKey) {
-            const { createClient } = await import("@supabase/supabase-js");
-            const client = createClient(config.supabaseUrl, config.supabaseKey, {
-                auth: { persistSession: false, autoRefreshToken: false },
-            });
-            const { data } = await client.from("platform_secrets").select("value").eq("key", "exa_api_key").single();
-            if (data?.value) {
-                cachedExaKey = data.value;
-                return cachedExaKey;
-            }
-        }
-        // Tier 2: User JWT
-        const token = await getValidToken();
-        if (token) {
-            const client = createAuthenticatedClient(token);
-            const { data } = await client.from("platform_secrets").select("value").eq("key", "exa_api_key").single();
-            if (data?.value) {
-                cachedExaKey = data.value;
-                return cachedExaKey;
-            }
+  if (cachedExaKey) return cachedExaKey;
+  try {
+    const config = resolveConfig();
+
+    // Tier 1: Service role key (MCP server mode)
+    if (config.supabaseUrl && config.supabaseKey) {
+      const {
+        createClient
+      } = await import("@supabase/supabase-js");
+      const client = createClient(config.supabaseUrl, config.supabaseKey, {
+        auth: {
+          persistSession: false,
+          autoRefreshToken: false
         }
+      });
+      const {
+        data
+      } = await client.from("platform_secrets").select("value").eq("key", "exa_api_key").single();
+      if (data?.value) {
+        cachedExaKey = data.value;
+        return cachedExaKey;
+      }
     }
-    catch { /* swallow */ }
-    // Tier 3: Environment variable fallback
-    const envKey = process.env["EXA_API_KEY"];
-    if (envKey) {
-        cachedExaKey = envKey;
+
+    // Tier 2: User JWT
+    const token = await getValidToken();
+    if (token) {
+      const client = createAuthenticatedClient(token);
+      const {
+        data
+      } = await client.from("platform_secrets").select("value").eq("key", "exa_api_key").single();
+      if (data?.value) {
+        cachedExaKey = data.value;
         return cachedExaKey;
+      }
     }
-    return null;
+  } catch {/* swallow */}
+
+  // Tier 3: Environment variable fallback
+  const envKey = process.env["EXA_API_KEY"];
+  if (envKey) {
+    cachedExaKey = envKey;
+    return cachedExaKey;
+  }
+  return null;
 }
 export async function webSearch(input) {
-    const query = input.query;
-    if (!query)
-        return { success: false, output: "query is required" };
-    const allowedDomains = input.allowed_domains;
-    const blockedDomains = input.blocked_domains;
-    const apiKey = await getExaApiKey();
-    if (!apiKey) {
-        return { success: false, output: "Exa API key not configured. Add 'exa_api_key' to platform_secrets table." };
-    }
-    try {
-        const searchBody = {
-            query,
-            numResults: 10,
-            type: "auto",
-            contents: { text: { maxCharacters: 1200, includeHtmlTags: false } },
-        };
-        if (allowedDomains?.length)
-            searchBody.includeDomains = allowedDomains;
-        if (blockedDomains?.length)
-            searchBody.excludeDomains = blockedDomains;
-        const response = await fetch("https://api.exa.ai/search", {
-            method: "POST",
-            headers: {
-                "x-api-key": apiKey,
-                "Content-Type": "application/json",
-                "Accept": "application/json",
-            },
-            body: JSON.stringify(searchBody),
-            signal: AbortSignal.timeout(15000),
-        });
-        if (!response.ok) {
-            const errBody = await response.text();
-            return { success: false, output: `Exa API error (${response.status}): ${errBody}` };
+  const query = input.query;
+  if (!query) return {
+    success: false,
+    output: "query is required"
+  };
+  const allowedDomains = input.allowed_domains;
+  const blockedDomains = input.blocked_domains;
+  const apiKey = await getExaApiKey();
+  if (!apiKey) {
+    return {
+      success: false,
+      output: "Exa API key not configured. Add 'exa_api_key' to platform_secrets table."
+    };
+  }
+  try {
+    const searchBody = {
+      query,
+      numResults: 10,
+      type: "auto",
+      contents: {
+        text: {
+          maxCharacters: 1200,
+          includeHtmlTags: false
         }
-        const data = await response.json();
-        const results = (data.results || []).map((r, i) => {
-            const parts = [
-                `${i + 1}. **${r.title || "Untitled"}**`,
-                `   ${r.url}`,
-            ];
-            if (r.publishedDate)
-                parts.push(`   Published: ${r.publishedDate}`);
-            if (r.text)
-                parts.push(`   ${r.text.slice(0, 500)}`);
-            return parts.join("\n");
-        });
-        return {
-            success: true,
-            output: `Found ${results.length} results for "${query}":\n\n${results.join("\n\n")}`,
-        };
+      }
+    };
+    if (allowedDomains?.length) searchBody.includeDomains = allowedDomains;
+    if (blockedDomains?.length) searchBody.excludeDomains = blockedDomains;
+    const response = await fetch("https://api.exa.ai/search", {
+      method: "POST",
+      headers: {
+        "x-api-key": apiKey,
+        "Content-Type": "application/json",
+        "Accept": "application/json"
+      },
+      body: JSON.stringify(searchBody),
+      signal: AbortSignal.timeout(15000)
+    });
+    if (!response.ok) {
+      const errBody = await response.text();
+      return {
+        success: false,
+        output: `Exa API error (${response.status}): ${errBody}`
+      };
     }
-    catch (err) {
-        if (err.name === "TimeoutError" || err.message?.includes("timeout")) {
-            return { success: false, output: "Exa search timed out (15s)" };
-        }
-        return { success: false, output: `Web search error: ${err.message || err}` };
+    const data = await response.json();
+    const results = (data.results || []).map((r, i) => {
+      const parts = [`${i + 1}. **${r.title || "Untitled"}**`, `   ${r.url}`];
+      if (r.publishedDate) parts.push(`   Published: ${r.publishedDate}`);
+      if (r.text) parts.push(`   ${r.text.slice(0, 500)}`);
+      return parts.join("\n");
+    });
+    return {
+      success: true,
+      output: `Found ${results.length} results for "${query}":\n\n${results.join("\n\n")}`
+    };
+  } catch (err) {
+    if (err.name === "TimeoutError" || err.message?.includes("timeout")) {
+      return {
+        success: false,
+        output: "Exa search timed out (15s)"
+      };
     }
+    return {
+      success: false,
+      output: `Web search error: ${err.message || err}`
+    };
+  }
 }
+//# sourceMappingURL=web-tools.js.map