@sisu-ai/tool-web-fetch 8.0.1 → 8.0.2

package/dist/index.d.ts

@@ -1,5 +1,5 @@
-import type { Tool } from '@sisu-ai/core';
-export type WebFetchFormat = 'text' | 'html' | 'json';
+import type { Tool } from "@sisu-ai/core";
+export type WebFetchFormat = "text" | "html" | "json";
 export interface WebFetchArgs {
     url: string;
     format?: WebFetchFormat;

package/dist/index.js

@@ -1,32 +1,35 @@
-import { firstConfigValue } from '@sisu-ai/core';
-import { z } from 'zod';
+import { firstConfigValue } from "@sisu-ai/core";
+import { z } from "zod";
 export const webFetch = {
-    name: 'webFetch',
-    description: 'Fetch a web page by URL and return text, HTML, or JSON. Defaults to text extraction for HTML.',
+    name: "webFetch",
+    description: "Fetch a web page by URL and return text, HTML, or JSON. Defaults to text extraction for HTML.",
     schema: z.object({
         url: z.string().url(),
-        format: z.enum(['text', 'html', 'json']).optional(),
+        format: z.enum(["text", "html", "json"]).optional(),
         maxBytes: z.number().int().positive().max(5_000_000).optional(),
         respectRobots: z.boolean().optional(),
     }),
-    handler: async ({ url, format = 'text', maxBytes, respectRobots }, ctx) => {
-        const ua = firstConfigValue(['WEB_FETCH_USER_AGENT', 'HTTP_USER_AGENT'])
-            || 'SisuWebFetch/0.1 (+https://github.com/finger-gun/sisu)';
-        const capEnv = firstConfigValue(['WEB_FETCH_MAX_BYTES']);
+    handler: async ({ url, format = "text", maxBytes, respectRobots }, ctx) => {
+        const ua = firstConfigValue(["WEB_FETCH_USER_AGENT", "HTTP_USER_AGENT"]) ||
+            "SisuWebFetch/0.1 (+https://github.com/finger-gun/sisu)";
+        const capEnv = firstConfigValue(["WEB_FETCH_MAX_BYTES"]);
         const cap = Number(maxBytes ?? (capEnv !== undefined ? Number(capEnv) : 500_000));
         // robots.txt compliance (default on; disable with arg or env WEB_FETCH_RESPECT_ROBOTS=0)
         const respect = (() => {
-            if (typeof respectRobots === 'boolean')
+            if (typeof respectRobots === "boolean")
                 return respectRobots;
-            const env = firstConfigValue(['WEB_FETCH_RESPECT_ROBOTS', 'RESPECT_ROBOTS']);
+            const env = firstConfigValue([
+                "WEB_FETCH_RESPECT_ROBOTS",
+                "RESPECT_ROBOTS",
+            ]);
             if (env === undefined)
                 return true; // default on
-            return !(env === '0' || /^false$/i.test(env));
+            return !(env === "0" || /^false$/i.test(env));
         })();
         if (respect) {
             const decision = await robotsDecision(url, ua).catch(() => ({ allowed: true }));
             if (!decision.allowed) {
-                ctx?.log?.info?.('[webFetch] blocked by robots.txt', {
+                ctx?.log?.info?.("[webFetch] blocked by robots.txt", {
                     url,
                     userAgent: ua,
                     matchedAgent: decision.matchedAgent,
@@ -36,48 +39,76 @@ export const webFetch = {
                 return {
                     url,
                     status: 403,
-                    contentType: 'text/plain',
-                    text: `Blocked by robots.txt (agent: ${decision.matchedAgent ?? 'unknown'}, rule: ${decision.ruleType ?? 'disallow'} ${decision.rulePattern ?? ''})`.
-                        trim(),
+                    contentType: "text/plain",
+                    text: `Blocked by robots.txt (agent: ${decision.matchedAgent ?? "unknown"}, rule: ${decision.ruleType ?? "disallow"} ${decision.rulePattern ?? ""})`.trim(),
                     robotsBlocked: true,
-                    robotsAgent: ua
+                    robotsAgent: ua,
                 };
             }
         }
         const res = await fetch(url, {
-            redirect: 'follow',
-            headers: { 'User-Agent': ua, 'Accept': '*/*' },
+            redirect: "follow",
+            headers: { "User-Agent": ua, Accept: "*/*" },
         });
-        const contentType = res.headers?.get?.('content-type') || '';
+        const contentType = res.headers?.get?.("content-type") || "";
         // Stream read with cap to avoid massive bodies
         const buf = await readWithCap(res, cap);
         const finalUrl = res.url || undefined;
         if (!res.ok) {
-            return { url, finalUrl, status: res.status, contentType, text: truncateText(buf.toString('utf8')) };
+            return {
+                url,
+                finalUrl,
+                status: res.status,
+                contentType,
+                text: truncateText(buf.toString("utf8")),
+            };
         }
         // Handle by requested format and content-type
         const ctLower = contentType.toLowerCase();
-        if (format === 'json' || ctLower.includes('application/json')) {
+        if (format === "json" || ctLower.includes("application/json")) {
             try {
-                const json = JSON.parse(buf.toString('utf8'));
+                const json = JSON.parse(buf.toString("utf8"));
                 return { url, finalUrl, status: res.status, contentType, json };
             }
             catch {
                 // Fall through to text
             }
         }
-        if (format === 'html' || ctLower.includes('text/html') || ctLower.includes('application/xhtml')) {
-            const html = buf.toString('utf8');
-            if (format === 'html') {
-                return { url, finalUrl, status: res.status, contentType, html, title: extractTitle(html) };
+        if (format === "html" ||
+            ctLower.includes("text/html") ||
+            ctLower.includes("application/xhtml")) {
+            const html = buf.toString("utf8");
+            if (format === "html") {
+                return {
+                    url,
+                    finalUrl,
+                    status: res.status,
+                    contentType,
+                    html,
+                    title: extractTitle(html),
+                };
             }
             // format === 'text'
             const text = htmlToText(html);
-            return { url, finalUrl, status: res.status, contentType, text, title: extractTitle(html), html: undefined };
+            return {
+                url,
+                finalUrl,
+                status: res.status,
+                contentType,
+                text,
+                title: extractTitle(html),
+                html: undefined,
+            };
         }
         // Fallback: treat as text/*
-        const text = buf.toString('utf8');
-        return { url, finalUrl, status: res.status, contentType, text: truncateText(text) };
+        const text = buf.toString("utf8");
+        return {
+            url,
+            finalUrl,
+            status: res.status,
+            contentType,
+            text: truncateText(text),
+        };
     },
 };
 export default webFetch;
@@ -87,10 +118,13 @@ async function robotsDecision(targetUrl, userAgent) {
     const origin = `${u.protocol}//${u.host}`;
     const cache = robotsCache.get(origin);
     const now = Date.now();
-    if (!cache || (now - cache.ts) > 60 * 60 * 1000) { // 1h TTL
+    if (!cache || now - cache.ts > 60 * 60 * 1000) {
+        // 1h TTL
         const robotsUrl = `${origin}/robots.txt`;
         try {
-            const res = await fetch(robotsUrl, { headers: { 'User-Agent': userAgent, 'Accept': 'text/plain' } });
+            const res = await fetch(robotsUrl, {
+                headers: { "User-Agent": userAgent, Accept: "text/plain" },
+            });
             const txt = await res.text();
             const rules = res.ok ? parseRobots(txt) : null;
             robotsCache.set(origin, { ts: now, rules });
@@ -102,7 +136,7 @@ async function robotsDecision(targetUrl, userAgent) {
     const rules = robotsCache.get(origin)?.rules;
     if (!rules)
         return { allowed: true };
-    return evaluateRobotsDetailed(rules, userAgent, u.pathname + (u.search || ''));
+    return evaluateRobotsDetailed(rules, userAgent, u.pathname + (u.search || ""));
 }
 function parseRobots(text) {
     const lines = text.split(/\r?\n/);
@@ -110,31 +144,31 @@ function parseRobots(text) {
     let current = null;
     for (const raw of lines) {
         const line = raw.trim();
-        if (!line || line.startsWith('#'))
+        if (!line || line.startsWith("#"))
             continue;
         const m = line.match(/^(user-agent|allow|disallow)\s*:\s*(.*)$/i);
         if (!m)
             continue;
         const key = m[1].toLowerCase();
         const val = m[2].trim();
-        if (key === 'user-agent') {
+        if (key === "user-agent") {
             // Start a new group if we already had one and it contains rules
-            if (!current || (current.allows.length + current.disallows.length) > 0) {
+            if (!current || current.allows.length + current.disallows.length > 0) {
                 current = { agents: [], allows: [], disallows: [] };
                 groups.push(current);
             }
             current.agents.push(val.toLowerCase());
         }
-        else if (key === 'allow') {
+        else if (key === "allow") {
             if (!current) {
-                current = { agents: ['*'], allows: [], disallows: [] };
+                current = { agents: ["*"], allows: [], disallows: [] };
                 groups.push(current);
             }
             current.allows.push(val);
         }
-        else if (key === 'disallow') {
+        else if (key === "disallow") {
             if (!current) {
-                current = { agents: ['*'], allows: [], disallows: [] };
+                current = { agents: ["*"], allows: [], disallows: [] };
                 groups.push(current);
             }
             current.disallows.push(val);
@@ -145,18 +179,20 @@ function parseRobots(text) {
 function evaluateRobotsDetailed(rules, userAgent, pathWithQuery) {
     // Match exact agent token (product) ignoring case, or '*'.
     // Example: 'SisuWebFetch/0.1 (+...)' -> baseAgent 'sisuwebfetch'
-    const baseAgent = (userAgent.split(/[\/\s]/)[0] || '').toLowerCase();
+    const baseAgent = (userAgent.split(/[\s/]/)[0] || "").toLowerCase();
     const agentMatches = (agent) => {
-        if (agent === '*')
+        if (agent === "*")
             return true;
         return agent.toLowerCase() === baseAgent;
     };
     const matching = rules.groups
-        .map(g => ({ g, matchedAgent: g.agents.find(agentMatches) }))
-        .filter(x => !!x.matchedAgent);
+        .map((g) => ({ g, matchedAgent: g.agents.find(agentMatches) }))
+        .filter((x) => !!x.matchedAgent);
     const selected = matching.length
         ? matching
-        : rules.groups.filter(g => g.agents.includes('*')).map(g => ({ g, matchedAgent: '*' }));
+        : rules.groups
+            .filter((g) => g.agents.includes("*"))
+            .map((g) => ({ g, matchedAgent: "*" }));
     if (!selected.length)
         return { allowed: true };
     // longest match wins between allow and disallow
@@ -172,7 +208,7 @@ function evaluateRobotsDetailed(rules, userAgent, pathWithQuery) {
                 const L = pat.length;
                 if (L > bestLen) {
                     bestLen = L;
-                    bestType = 'allow';
+                    bestType = "allow";
                     bestPat = pat;
                     bestAgent = matchedAgent;
                 }
@@ -185,28 +221,38 @@ function evaluateRobotsDetailed(rules, userAgent, pathWithQuery) {
                 const L = pat.length;
                 if (L > bestLen) {
                     bestLen = L;
-                    bestType = 'disallow';
+                    bestType = "disallow";
                     bestPat = pat;
                     bestAgent = matchedAgent;
                 }
             }
         }
     }
-    if (bestType === 'disallow')
-        return { allowed: false, matchedAgent: bestAgent, ruleType: 'disallow', rulePattern: bestPat };
-    return { allowed: true, matchedAgent: bestAgent, ruleType: bestType, rulePattern: bestPat };
+    if (bestType === "disallow")
+        return {
+            allowed: false,
+            matchedAgent: bestAgent,
+            ruleType: "disallow",
+            rulePattern: bestPat,
+        };
+    return {
+        allowed: true,
+        matchedAgent: bestAgent,
+        ruleType: bestType,
+        rulePattern: bestPat,
+    };
 }
 function patternMatches(pat, path) {
     // Support '*' wildcard and '$' end anchor; treat path as starting with '/'
     const p = pat.trim();
-    if (p === '')
+    if (p === "")
         return false;
     // Empty disallow means allow all; already handled by return false above
     // Convert to regex
-    const escaped = p.replace(/[.+?^${}()|\[\]\\]/g, r => '\\' + r);
-    let reStr = '^' + escaped.replace(/\*/g, '.*');
-    if (reStr.endsWith('\$')) {
-        reStr = reStr.slice(0, -2) + '$';
+    const escaped = p.replace(/[.+?^${}()|[\]\\]/g, (r) => "\\" + r);
+    let reStr = "^" + escaped.replace(/\*/g, ".*");
+    if (reStr.endsWith("$")) {
+        reStr = reStr.slice(0, -1) + "$";
     }
     const re = new RegExp(reStr);
     return re.test(path);
@@ -214,9 +260,9 @@ function patternMatches(pat, path) {
 async function readWithCap(res, cap) {
     // If body is not a stream (older fetch mocks), try res.text()
     const anyRes = res;
-    if (!anyRes.body || typeof anyRes.body.getReader !== 'function') {
-        const t = typeof anyRes.text === 'function' ? await anyRes.text() : '';
-        return Buffer.from(String(t), 'utf8');
+    if (!anyRes.body || typeof anyRes.body.getReader !== "function") {
+        const t = typeof anyRes.text === "function" ? await anyRes.text() : "";
+        return Buffer.from(String(t), "utf8");
     }
     const reader = anyRes.body.getReader();
     const chunks = [];
@@ -249,34 +295,34 @@ function extractTitle(html) {
 function htmlToText(html) {
     // Remove script/style robustly: allow attributes and sloppy closing tags like </script foo="bar"> or </script >
     let s = html
-        .replace(/<script\b[^>]*>[\s\S]*?<\/script\b[^>]*>/gi, ' ')
-        .replace(/<style\b[^>]*>[\s\S]*?<\/style\b[^>]*>/gi, ' ');
+        .replace(/<script\b[^>]*>[\s\S]*?<\/script\b[^>]*>/gi, " ")
+        .replace(/<style\b[^>]*>[\s\S]*?<\/style\b[^>]*>/gi, " ");
     // Remove HTML comments, including non-standard end '--!>' browsers tolerate
-    s = s.replace(/<!--[\s\S]*?--!?>(\n)?/g, ' ');
+    s = s.replace(/<!--[\s\S]*?--!?>(\n)?/g, " ");
     // Replace <br> and block tags with newlines
-    s = s.replace(/<(br|BR)\s*\/?>(\n)?/g, '\n');
-    s = s.replace(/<\/(p|div|section|article|h[1-6]|li|ul|ol|header|footer|main)>/gi, '\n');
+    s = s.replace(/<(br|BR)\s*\/?>(\n)?/g, "\n");
+    s = s.replace(/<\/(p|div|section|article|h[1-6]|li|ul|ol|header|footer|main)>/gi, "\n");
     // Strip remaining tags
-    s = s.replace(/<[^>]+>/g, ' ');
+    s = s.replace(/<[^>]+>/g, " ");
     // Decode entities
     s = decodeHTMLEntities(s);
     // Collapse whitespace
-    s = s.replace(/\s+/g, ' ').trim();
+    s = s.replace(/\s+/g, " ").trim();
     return truncateText(s);
 }
 function truncateText(text, max = 200_000) {
-    return text.length > max ? text.slice(0, max) + '…' : text;
+    return text.length > max ? text.slice(0, max) + "…" : text;
 }
 // Minimal HTML entity decoder for common entities
 function decodeHTMLEntities(s) {
     const map = {
-        '&amp;': '&',
-        '&lt;': '<',
-        '&gt;': '>',
-        '&quot;': '"',
-        '&#39;': "'",
-        '&apos;': "'",
-        '&nbsp;': ' ',
+        "&amp;": "&",
+        "&lt;": "<",
+        "&gt;": ">",
+        "&quot;": '"',
+        "&#39;": "'",
+        "&apos;": "'",
+        "&nbsp;": " ",
     };
     return s.replace(/&(amp|lt|gt|quot|#39|apos|nbsp);/g, (m) => map[m] || m);
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@sisu-ai/tool-web-fetch",
-  "version": "8.0.1",
+  "version": "8.0.2",
   "license": "Apache-2.0",
   "type": "module",
   "main": "dist/index.js",