@ashx-j/lunr 0.1.0

package/dist/builtin-extensions/pi-web-access/rsc-extract.js

@@ -0,0 +1,330 @@
+// @ts-nocheck
+/**
+ * RSC Content Extractor
+ *
+ * Extracts readable content from Next.js React Server Components (RSC) flight payloads.
+ * RSC pages embed content as JSON in <script>self.__next_f.push([...])</script> tags.
+ */
+export function extractRSCContent(html) {
+    if (!html.includes("self.__next_f.push")) {
+        return null;
+    }
+    // Parse all RSC chunks into a map
+    const chunkMap = new Map();
+    const scriptRegex = /<script>self\.__next_f\.push\(\[1,"([\s\S]*?)"\]\)<\/script>/g;
+    for (const match of html.matchAll(scriptRegex)) {
+        let content;
+        try {
+            content = JSON.parse('"' + match[1] + '"');
+        }
+        catch {
+            continue;
+        }
+        // Parse each line as "id:payload"
+        // Lines are separated by \n, each line is one chunk
+        // Chunk IDs are hex strings, typically 1-4 chars (supports up to 65535 chunks)
+        for (const line of content.split("\n")) {
+            if (!line.trim())
+                continue;
+            const colonIdx = line.indexOf(":");
+            if (colonIdx <= 0 || colonIdx > 4)
+                continue;
+            const id = line.slice(0, colonIdx);
+            if (!/^[0-9a-f]+$/i.test(id))
+                continue;
+            const payload = line.slice(colonIdx + 1);
+            if (!payload)
+                continue;
+            const existing = chunkMap.get(id);
+            if (!existing || payload.length > existing.length) {
+                chunkMap.set(id, payload);
+            }
+        }
+    }
+    if (chunkMap.size === 0)
+        return null;
+    // Extract title
+    const titleMatch = html.match(/<title[^>]*>([^<]+)<\/title>/);
+    const title = titleMatch?.[1]?.split("|")[0]?.trim() || "";
+    // Parse and cache parsed chunks
+    const parsedCache = new Map();
+    function getParsedChunk(id) {
+        if (parsedCache.has(id))
+            return parsedCache.get(id);
+        const chunk = chunkMap.get(id);
+        if (!chunk || !chunk.startsWith("[")) {
+            parsedCache.set(id, null);
+            return null;
+        }
+        try {
+            const parsed = JSON.parse(chunk);
+            parsedCache.set(id, parsed);
+            return parsed;
+        }
+        catch {
+            parsedCache.set(id, null);
+            return null;
+        }
+    }
+    const visitedRefs = new Set();
+    function extractNode(node, ctx = { inTable: false, inCode: false }) {
+        if (node === null || node === undefined)
+            return "";
+        if (typeof node === "string") {
+            // Check if it's a reference like "$L30"
+            const refMatch = node.match(/^\$L([0-9a-f]+)$/i);
+            if (refMatch) {
+                const refId = refMatch[1];
+                if (visitedRefs.has(refId))
+                    return ""; // Prevent cycles
+                visitedRefs.add(refId);
+                const refNode = getParsedChunk(refId);
+                const result = refNode ? extractNode(refNode, ctx) : "";
+                visitedRefs.delete(refId);
+                return result;
+            }
+            // Filter out RSC-specific artifacts, but preserve content inside code blocks
+            if (!ctx.inCode && (node === "$undefined" || node === "$" || /^\$[A-Z]/.test(node)))
+                return "";
+            return node.trim() ? node : "";
+        }
+        if (typeof node === "number")
+            return String(node);
+        if (typeof node === "boolean")
+            return "";
+        if (!Array.isArray(node))
+            return "";
+        // RSC element: ["$", "tag", key, props]
+        if (node[0] === "$" && typeof node[1] === "string") {
+            const tag = node[1];
+            const props = (node[3] || {});
+            // Skip non-content
+            const skipTags = ["script", "style", "svg", "path", "circle", "link", "meta",
+                "template", "button", "input", "nav", "footer", "aside"];
+            if (skipTags.includes(tag))
+                return "";
+            // Component ref like $L25
+            if (tag.startsWith("$L")) {
+                const refId = tag.slice(2);
+                if (visitedRefs.has(refId))
+                    return "";
+                // Check for heading components with baseId
+                if (props.baseId && props.children) {
+                    return `## ${String(props.children)}\n\n`;
+                }
+                visitedRefs.add(refId);
+                const refNode = getParsedChunk(refId);
+                let result = "";
+                if (refNode) {
+                    result = extractNode(refNode, ctx);
+                }
+                else if (props.children) {
+                    result = extractNode(props.children, ctx);
+                }
+                visitedRefs.delete(refId);
+                return result;
+            }
+            const children = props.children;
+            const content = children ? extractNode(children, ctx) : "";
+            switch (tag) {
+                case "h1": return `# ${content.trim()}\n\n`;
+                case "h2": return `## ${content.trim()}\n\n`;
+                case "h3": return `### ${content.trim()}\n\n`;
+                case "h4": return `#### ${content.trim()}\n\n`;
+                case "h5": return `##### ${content.trim()}\n\n`;
+                case "h6": return `###### ${content.trim()}\n\n`;
+                case "p": return ctx.inTable ? content : `${content.trim()}\n\n`;
+                case "code": {
+                    const codeContent = children ? extractNode(children, { ...ctx, inCode: true }) : "";
+                    return ctx.inCode ? codeContent : `\`${codeContent}\``;
+                }
+                case "pre": {
+                    const preContent = children ? extractNode(children, { ...ctx, inCode: true }) : "";
+                    return "```\n" + preContent + "\n```\n\n";
+                }
+                case "strong":
+                case "b": return `**${content}**`;
+                case "em":
+                case "i": return `*${content}*`;
+                case "li": return `- ${content.trim()}\n`;
+                case "ul":
+                case "ol": return content + "\n";
+                case "blockquote": return `> ${content.trim()}\n\n`;
+                case "table": return extractTable(node) + "\n";
+                case "thead":
+                case "tbody":
+                case "tr":
+                case "th":
+                case "td":
+                    return content;
+                case "div":
+                    if (props.role === "alert" || props["data-slot"] === "alert") {
+                        return `> ${content.trim()}\n\n`;
+                    }
+                    return content;
+                case "a": {
+                    const href = props.href;
+                    return href && !href.startsWith("#") ? `[${content}](${href})` : content;
+                }
+                default: return content;
+            }
+        }
+        // Array of child nodes
+        return node.map(n => extractNode(n, ctx)).join("");
+    }
+    function extractTable(tableNode) {
+        const props = (tableNode[3] || {});
+        const rows = [];
+        let headerRowCount = 0;
+        function walkTable(node, isHeader = false) {
+            if (node === null || node === undefined)
+                return;
+            // Handle string refs
+            if (typeof node === "string") {
+                const refMatch = node.match(/^\$L([0-9a-f]+)$/i);
+                if (refMatch && !visitedRefs.has(refMatch[1])) {
+                    visitedRefs.add(refMatch[1]);
+                    const refNode = getParsedChunk(refMatch[1]);
+                    if (refNode)
+                        walkTable(refNode, isHeader);
+                    visitedRefs.delete(refMatch[1]);
+                }
+                return;
+            }
+            if (!Array.isArray(node))
+                return;
+            if (node[0] === "$") {
+                const tag = node[1];
+                const nodeProps = (node[3] || {});
+                // Handle component refs
+                if (tag.startsWith("$L")) {
+                    const refId = tag.slice(2);
+                    if (!visitedRefs.has(refId)) {
+                        visitedRefs.add(refId);
+                        const refNode = getParsedChunk(refId);
+                        if (refNode)
+                            walkTable(refNode, isHeader);
+                        visitedRefs.delete(refId);
+                    }
+                    return;
+                }
+                if (tag === "thead")
+                    walkTable(nodeProps.children, true);
+                else if (tag === "tbody")
+                    walkTable(nodeProps.children, false);
+                else if (tag === "tr") {
+                    const cells = [];
+                    walkCells(nodeProps.children, cells);
+                    if (cells.length > 0) {
+                        rows.push(cells);
+                        if (isHeader)
+                            headerRowCount++;
+                    }
+                }
+                else
+                    walkTable(nodeProps.children, isHeader);
+            }
+            else {
+                for (const child of node)
+                    walkTable(child, isHeader);
+            }
+        }
+        function walkCells(node, cells) {
+            if (node === null || node === undefined)
+                return;
+            // Handle string refs
+            if (typeof node === "string") {
+                const refMatch = node.match(/^\$L([0-9a-f]+)$/i);
+                if (refMatch && !visitedRefs.has(refMatch[1])) {
+                    visitedRefs.add(refMatch[1]);
+                    const refNode = getParsedChunk(refMatch[1]);
+                    if (refNode)
+                        walkCells(refNode, cells);
+                    visitedRefs.delete(refMatch[1]);
+                }
+                return;
+            }
+            if (!Array.isArray(node))
+                return;
+            if (node[0] === "$" && (node[1] === "td" || node[1] === "th")) {
+                const cellProps = (node[3] || {});
+                const text = extractNode(cellProps.children, { inTable: true, inCode: false })
+                    .trim()
+                    .replace(/\n/g, " ")
+                    .replace(/\\/g, "\\\\") // Escape backslashes first
+                    .replace(/\|/g, "\\|"); // Then escape pipes
+                cells.push(text);
+            }
+            else if (node[0] === "$" && typeof node[1] === "string" && node[1].startsWith("$L")) {
+                // Component ref for a cell
+                const refId = node[1].slice(2);
+                if (!visitedRefs.has(refId)) {
+                    visitedRefs.add(refId);
+                    const refNode = getParsedChunk(refId);
+                    if (refNode)
+                        walkCells(refNode, cells);
+                    visitedRefs.delete(refId);
+                }
+            }
+            else {
+                for (const child of node)
+                    walkCells(child, cells);
+            }
+        }
+        walkTable(props.children);
+        if (rows.length === 0)
+            return "";
+        const colCount = Math.max(...rows.map(r => r.length));
+        let md = "";
+        for (let i = 0; i < rows.length; i++) {
+            const row = rows[i].concat(Array(colCount - rows[i].length).fill(""));
+            md += "| " + row.join(" | ") + " |\n";
+            if (i === headerRowCount - 1 || (headerRowCount === 0 && i === 0)) {
+                md += "| " + Array(colCount).fill("---").join(" | ") + " |\n";
+            }
+        }
+        return md;
+    }
+    // Process main content chunk (usually "23")
+    const mainChunk = getParsedChunk("23");
+    if (mainChunk) {
+        const content = extractNode(mainChunk);
+        if (content.trim().length > 100) {
+            const cleaned = content
+                .replace(/\n{3,}/g, "\n\n")
+                .trim();
+            return { title, content: cleaned };
+        }
+    }
+    // Fallback: try other chunks
+    const contentParts = [];
+    for (const [id] of chunkMap) {
+        if (id === "23")
+            continue;
+        const parsed = getParsedChunk(id);
+        if (!parsed)
+            continue;
+        visitedRefs.clear();
+        const text = extractNode(parsed);
+        if (text.trim().length > 50 &&
+            !text.includes("page was not found") &&
+            !text.includes("404")) {
+            contentParts.push({ order: parseInt(id, 16), text: text.trim() });
+        }
+    }
+    if (contentParts.length === 0)
+        return null;
+    contentParts.sort((a, b) => a.order - b.order);
+    const seen = new Set();
+    const uniqueParts = [];
+    for (const part of contentParts) {
+        const key = part.text.slice(0, 150);
+        if (!seen.has(key)) {
+            seen.add(key);
+            uniqueParts.push(part.text);
+        }
+    }
+    const content = uniqueParts.join("\n\n").replace(/\n{3,}/g, "\n\n").trim();
+    return content.length > 100 ? { title, content } : null;
+}
+//# sourceMappingURL=rsc-extract.js.map

package/dist/builtin-extensions/pi-web-access/rsc-extract.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"rsc-extract.js","sourceRoot":"","sources":["../../../src/builtin-extensions/pi-web-access/rsc-extract.ts"],"names":[],"mappings":"AAAA,cAAc;AACd;;;;;GAKG;AAOH,MAAM,UAAU,iBAAiB,CAAC,IAAY,EAA2B;IACvE,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,oBAAoB,CAAC,EAAE,CAAC;QACzC,OAAO,IAAI,CAAC;IACd,CAAC;IAED,kCAAkC;IAClC,MAAM,QAAQ,GAAG,IAAI,GAAG,EAAkB,CAAC;IAC3C,MAAM,WAAW,GAAG,+DAA+D,CAAC;IAEpF,KAAK,MAAM,KAAK,IAAI,IAAI,CAAC,QAAQ,CAAC,WAAW,CAAC,EAAE,CAAC;QAC/C,IAAI,OAAe,CAAC;QACpB,IAAI,CAAC;YACH,OAAO,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,GAAG,KAAK,CAAC,CAAC,CAAC,GAAG,GAAG,CAAC,CAAC;QAC7C,CAAC;QAAC,MAAM,CAAC;YACP,SAAS;QACX,CAAC;QAED,kCAAkC;QAClC,oDAAoD;QACpD,+EAA+E;QAC/E,KAAK,MAAM,IAAI,IAAI,OAAO,CAAC,KAAK,CAAC,IAAI,CAAC,EAAE,CAAC;YACvC,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE;gBAAE,SAAS;YAE3B,MAAM,QAAQ,GAAG,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC;YACnC,IAAI,QAAQ,IAAI,CAAC,IAAI,QAAQ,GAAG,CAAC;gBAAE,SAAS;YAE5C,MAAM,EAAE,GAAG,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,QAAQ,CAAC,CAAC;YACnC,IAAI,CAAC,cAAc,CAAC,IAAI,CAAC,EAAE,CAAC;gBAAE,SAAS;YAEvC,MAAM,OAAO,GAAG,IAAI,CAAC,KAAK,CAAC,QAAQ,GAAG,CAAC,CAAC,CAAC;YACzC,IAAI,CAAC,OAAO;gBAAE,SAAS;YAEvB,MAAM,QAAQ,GAAG,QAAQ,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;YAClC,IAAI,CAAC,QAAQ,IAAI,OAAO,CAAC,MAAM,GAAG,QAAQ,CAAC,MAAM,EAAE,CAAC;gBAClD,QAAQ,CAAC,GAAG,CAAC,EAAE,EAAE,OAAO,CAAC,CAAC;YAC5B,CAAC;QACH,CAAC;IACH,CAAC;IAED,IAAI,QAAQ,CAAC,IAAI,KAAK,CAAC;QAAE,OAAO,IAAI,CAAC;IAErC,gBAAgB;IAChB,MAAM,UAAU,GAAG,IAAI,CAAC,KAAK,CAAC,8BAA8B,CAAC,CAAC;IAC9D,MAAM,KAAK,GAAG,UAAU,EAAE,CAAC,CAAC,CAAC,EAAE,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,IAAI,EAAE,CAAC;IAE3D,gCAAgC;IAChC,MAAM,WAAW,GAAG,IAAI,GAAG,EAAmB,CAAC;IAE/C,SAAS,cAAc,CAAC,EAAU,EAAkB;QAClD,IAAI,WAAW,CAAC,GAAG,CAAC,EAAE,CAAC;YAAE,OAAO,WAAW,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;QAEpD,MAAM,KAAK,GAAG,QAAQ,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;QAC/B,IAAI,CAAC,KAAK,IAAI,CAAC,KAAK,CAAC,UAAU,CAAC,GAAG,CAAC,EAAE,CAAC;YACrC,WAAW,CAAC,GAAG,CAAC,EAAE,EAAE,IAAI,CAAC,CAAC;YAC1B,OAAO,IAAI,CAAC;QACd,CAAC;QAED,IAAI,CAAC;YACH,MAAM,MAAM,GAAG,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC;YACjC,WAAW,CAAC,GAAG,CAAC,EAAE,EAAE,MAAM,CAAC,CAAC;YAC5B,OAAO,MAAM,CAAC;QAChB,CAAC;QAAC,MAAM,CAAC;YACP,WAAW,CAAC,GAAG,CAAC,EAAE,EAAE,IAAI,CAAC,CAAC;YAC1B,OAAO,IAAI,CAAC;QACd,CAAC;IAAA,CACF;IAID,MAAM,WAAW,GAAG,IAAI,GAAG,EAAU,CAAC;IAEtC,SAAS,WAAW,CAAC,IAAU,EAAE,GAAG,GAAG,EAAE,OAAO,EAAE,KAAK,EAAE,MAAM,EAAE,KAAK,EAAE,EAAU;QAChF,IAAI,IAAI,KAAK,IAAI,IAAI,IAAI,KAAK,SAAS;YAAE,OAAO,EAAE,CAAC;QAEnD,IAAI,OAAO,IAAI,KAAK,QAAQ,EAAE,CAAC;YAC7B,wCAAwC;YACxC,MAAM,QAAQ,GAAG,IAAI,CAAC,KAAK,CAAC,mBAAmB,CAAC,CAAC;YACjD,IAAI,QAAQ,EAAE,CAAC;gBACb,MAAM,KAAK,GAAG,QAAQ,CAAC,CAAC,CAAC,CAAC;gBAC1B,IAAI,WAAW,CAAC,GAAG,CAAC,KAAK,CAAC;oBAAE,OAAO,EAAE,CAAC,CAAC,iBAAiB;gBACxD,WAAW,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC;gBACvB,MAAM,OAAO,GAAG,cAAc,CAAC,KAAK,CAAC,CAAC;gBACtC,MAAM,MAAM,GAAG,OAAO,CAAC,CAAC,CAAC,WAAW,CAAC,OAAO,EAAE,GAAG,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;gBACxD,WAAW,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC;gBAC1B,OAAO,MAAM,CAAC;YAChB,CAAC;YACD,6EAA6E;YAC7E,IAAI,CAAC,GAAG,CAAC,MAAM,IAAI,CAAC,IAAI,KAAK,YAAY,IAAI,IAAI,KAAK,GAAG,IAAI,UAAU,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;gBAAE,OAAO,EAAE,CAAC;YAC/F,OAAO,IAAI,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC;QACjC,CAAC;QAED,IAAI,OAAO,IAAI,KAAK,QAAQ;YAAE,OAAO,MAAM,CAAC,IAAI,CAAC,CAAC;QAClD,IAAI,OAAO,IAAI,KAAK,SAAS;YAAE,OAAO,EAAE,CAAC;QACzC,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,IAAI,CAAC;YAAE,OAAO,EAAE,CAAC;QAEpC,wCAAwC;QACxC,IAAI,IAAI,CAAC,CAAC,CAAC,KAAK,GAAG,IAAI,OAAO,IAAI,CAAC,CAAC,CAAC,KAAK,QAAQ,EAAE,CAAC;YACnD,MAAM,GAAG,GAAG,IAAI,CAAC,CAAC,CAAW,CAAC;YAC9B,MAAM,KAAK,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,EAAE,CAA4B,CAAC;YAEzD,mBAAmB;YACnB,MAAM,QAAQ,GAAG,CAAC,QAAQ,EAAE,OAAO,EAAE,KAAK,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,EAAE,MAAM;gBAC1D,UAAU,EAAE,QAAQ,EAAE,OAAO,EAAE,KAAK,EAAE,QAAQ,EAAE,OAAO,CAAC,CAAC;YAC3E,IAAI,QAAQ,CAAC,QAAQ,CAAC,GAAG,CAAC;gBAAE,OAAO,EAAE,CAAC;YAEtC,0BAA0B;YAC1B,IAAI,GAAG,CAAC,UAAU,CAAC,IAAI,CAAC,EAAE,CAAC;gBACzB,MAAM,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;gBAC3B,IAAI,WAAW,CAAC,GAAG,CAAC,KAAK,CAAC;oBAAE,OAAO,EAAE,CAAC;gBAEtC,2CAA2C;gBAC3C,IAAI,KAAK,CAAC,MAAM,IAAI,KAAK,CAAC,QAAQ,EAAE,CAAC;oBACnC,OAAO,MAAM,MAAM,CAAC,KAAK,CAAC,QAAQ,CAAC,MAAM,CAAC;gBAC5C,CAAC;gBAED,WAAW,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC;gBACvB,MAAM,OAAO,GAAG,cAAc,CAAC,KAAK,CAAC,CAAC;gBACtC,IAAI,MAAM,GAAG,EAAE,CAAC;gBAChB,IAAI,OAAO,EAAE,CAAC;oBACZ,MAAM,GAAG,WAAW,CAAC,OAAO,EAAE,GAAG,CAAC,CAAC;gBACrC,CAAC;qBAAM,IAAI,KAAK,CAAC,QAAQ,EAAE,CAAC;oBAC1B,MAAM,GAAG,WAAW,CAAC,KAAK,CAAC,QAAgB,EAAE,GAAG,CAAC,CAAC;gBACpD,CAAC;gBACD,WAAW,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC;gBAC1B,OAAO,MAAM,CAAC;YAChB,CAAC;YAED,MAAM,QAAQ,GAAG,KAAK,CAAC,QAAQ,CAAC;YAChC,MAAM,OAAO,GAAG,QAAQ,CAAC,CAAC,CAAC,WAAW,CAAC,QAAgB,EAAE,GAAG,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;YAEnE,QAAQ,GAAG,EAAE,CAAC;gBACZ,KAAK,IAAI,EAAE,OAAO,KAAK,OAAO,CAAC,IAAI,EAAE,MAAM,CAAC;gBAC5C,KAAK,IAAI,EAAE,OAAO,MAAM,OAAO,CAAC,IAAI,EAAE,MAAM,CAAC;gBAC7C,KAAK,IAAI,EAAE,OAAO,OAAO,OAAO,CAAC,IAAI,EAAE,MAAM,CAAC;gBAC9C,KAAK,IAAI,EAAE,OAAO,QAAQ,OAAO,CAAC,IAAI,EAAE,MAAM,CAAC;gBAC/C,KAAK,IAAI,EAAE,OAAO,SAAS,OAAO,CAAC,IAAI,EAAE,MAAM,CAAC;gBAChD,KAAK,IAAI,EAAE,OAAO,UAAU,OAAO,CAAC,IAAI,EAAE,MAAM,CAAC;gBACjD,KAAK,GAAG,EAAE,OAAO,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,GAAG,OAAO,CAAC,IAAI,EAAE,MAAM,CAAC;gBACjE,KAAK,MAAM,EAAE,CAAC;oBACZ,MAAM,WAAW,GAAG,QAAQ,CAAC,CAAC,CAAC,WAAW,CAAC,QAAgB,EAAE,EAAE,GAAG,GAAG,EAAE,MAAM,EAAE,IAAI,EAAE,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;oBAC5F,OAAO,GAAG,CAAC,MAAM,CAAC,CAAC,CAAC,WAAW,CAAC,CAAC,CAAC,KAAK,WAAW,IAAI,CAAC;gBACzD,CAAC;gBACD,KAAK,KAAK,EAAE,CAAC;oBACX,MAAM,UAAU,GAAG,QAAQ,CAAC,CAAC,CAAC,WAAW,CAAC,QAAgB,EAAE,EAAE,GAAG,GAAG,EAAE,MAAM,EAAE,IAAI,EAAE,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;oBAC3F,OAAO,OAAO,GAAG,UAAU,GAAG,WAAW,CAAC;gBAC5C,CAAC;gBACD,KAAK,QAAQ,CAAC;gBAAC,KAAK,GAAG,EAAE,OAAO,KAAK,OAAO,IAAI,CAAC;gBACjD,KAAK,IAAI,CAAC;gBAAC,KAAK,GAAG,EAAE,OAAO,IAAI,OAAO,GAAG,CAAC;gBAC3C,KAAK,IAAI,EAAE,OAAO,KAAK,OAAO,CAAC,IAAI,EAAE,IAAI,CAAC;gBAC1C,KAAK,IAAI,CAAC;gBAAC,KAAK,IAAI,EAAE,OAAO,OAAO,GAAG,IAAI,CAAC;gBAC5C,KAAK,YAAY,EAAE,OAAO,KAAK,OAAO,CAAC,IAAI,EAAE,MAAM,CAAC;gBACpD,KAAK,OAAO,EAAE,OAAO,YAAY,CAAC,IAAiB,CAAC,GAAG,IAAI,CAAC;gBAC5D,KAAK,OAAO,CAAC;gBAAC,KAAK,OAAO,CAAC;gBAAC,KAAK,IAAI,CAAC;gBAAC,KAAK,IAAI,CAAC;gBAAC,KAAK,IAAI;oBACzD,OAAO,OAAO,CAAC;gBACjB,KAAK,KAAK;oBACR,IAAI,KAAK,CAAC,IAAI,KAAK,OAAO,IAAI,KAAK,CAAC,WAAW,CAAC,KAAK,OAAO,EAAE,CAAC;wBAC7D,OAAO,KAAK,OAAO,CAAC,IAAI,EAAE,MAAM,CAAC;oBACnC,CAAC;oBACD,OAAO,OAAO,CAAC;gBACjB,KAAK,GAAG,EAAE,CAAC;oBACT,MAAM,IAAI,GAAG,KAAK,CAAC,IAA0B,CAAC;oBAC9C,OAAO,IAAI,IAAI,CAAC,IAAI,CAAC,UAAU,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,IAAI,OAAO,KAAK,IAAI,GAAG,CAAC,CAAC,CAAC,OAAO,CAAC;gBAC3E,CAAC;gBACD,SAAS,OAAO,OAAO,CAAC;YAC1B,CAAC;QACH,CAAC;QAED,uBAAuB;QACvB,OAAQ,IAAe,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,WAAW,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;IAAA,CAChE;IAED,SAAS,YAAY,CAAC,SAAoB,EAAU;QAClD,MAAM,KAAK,GAAG,CAAC,SAAS,CAAC,CAAC,CAAC,IAAI,EAAE,CAA4B,CAAC;QAC9D,MAAM,IAAI,GAAe,EAAE,CAAC;QAC5B,IAAI,cAAc,GAAG,CAAC,CAAC;QAEvB,SAAS,SAAS,CAAC,IAAa,EAAE,QAAQ,GAAG,KAAK,EAAQ;YACxD,IAAI,IAAI,KAAK,IAAI,IAAI,IAAI,KAAK,SAAS;gBAAE,OAAO;YAEhD,qBAAqB;YACrB,IAAI,OAAO,IAAI,KAAK,QAAQ,EAAE,CAAC;gBAC7B,MAAM,QAAQ,GAAG,IAAI,CAAC,KAAK,CAAC,mBAAmB,CAAC,CAAC;gBACjD,IAAI,QAAQ,IAAI,CAAC,WAAW,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;oBAC9C,WAAW,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC;oBAC7B,MAAM,OAAO,GAAG,cAAc,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC;oBAC5C,IAAI,OAAO;wBAAE,SAAS,CAAC,OAAO,EAAE,QAAQ,CAAC,CAAC;oBAC1C,WAAW,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC;gBAClC,CAAC;gBACD,OAAO;YACT,CAAC;YAED,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,IAAI,CAAC;gBAAE,OAAO;YAEjC,IAAI,IAAI,CAAC,CAAC,CAAC,KAAK,GAAG,EAAE,CAAC;gBACpB,MAAM,GAAG,GAAG,IAAI,CAAC,CAAC,CAAW,CAAC;gBAC9B,MAAM,SAAS,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,EAAE,CAA4B,CAAC;gBAE7D,wBAAwB;gBACxB,IAAI,GAAG,CAAC,UAAU,CAAC,IAAI,CAAC,EAAE,CAAC;oBACzB,MAAM,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;oBAC3B,IAAI,CAAC,WAAW,CAAC,GAAG,CAAC,KAAK,CAAC,EAAE,CAAC;wBAC5B,WAAW,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC;wBACvB,MAAM,OAAO,GAAG,cAAc,CAAC,KAAK,CAAC,CAAC;wBACtC,IAAI,OAAO;4BAAE,SAAS,CAAC,OAAO,EAAE,QAAQ,CAAC,CAAC;wBAC1C,WAAW,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC;oBAC5B,CAAC;oBACD,OAAO;gBACT,CAAC;gBAED,IAAI,GAAG,KAAK,OAAO;oBAAE,SAAS,CAAC,SAAS,CAAC,QAAQ,EAAE,IAAI,CAAC,CAAC;qBACpD,IAAI,GAAG,KAAK,OAAO;oBAAE,SAAS,CAAC,SAAS,CAAC,QAAQ,EAAE,KAAK,CAAC,CAAC;qBAC1D,IAAI,GAAG,KAAK,IAAI,EAAE,CAAC;oBACtB,MAAM,KAAK,GAAa,EAAE,CAAC;oBAC3B,SAAS,CAAC,SAAS,CAAC,QAAQ,EAAE,KAAK,CAAC,CAAC;oBACrC,IAAI,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;wBACrB,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;wBACjB,IAAI,QAAQ;4BAAE,cAAc,EAAE,CAAC;oBACjC,CAAC;gBACH,CAAC;;oBAAM,SAAS,CAAC,SAAS,CAAC,QAAQ,EAAE,QAAQ,CAAC,CAAC;YACjD,CAAC;iBAAM,CAAC;gBACN,KAAK,MAAM,KAAK,IAAI,IAAI;oBAAE,SAAS,CAAC,KAAK,EAAE,QAAQ,CAAC,CAAC;YACvD,CAAC;QAAA,CACF;QAED,SAAS,SAAS,CAAC,IAAa,EAAE,KAAe,EAAQ;YACvD,IAAI,IAAI,KAAK,IAAI,IAAI,IAAI,KAAK,SAAS;gBAAE,OAAO;YAEhD,qBAAqB;YACrB,IAAI,OAAO,IAAI,KAAK,QAAQ,EAAE,CAAC;gBAC7B,MAAM,QAAQ,GAAG,IAAI,CAAC,KAAK,CAAC,mBAAmB,CAAC,CAAC;gBACjD,IAAI,QAAQ,IAAI,CAAC,WAAW,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;oBAC9C,WAAW,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC;oBAC7B,MAAM,OAAO,GAAG,cAAc,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC;oBAC5C,IAAI,OAAO;wBAAE,SAAS,CAAC,OAAO,EAAE,KAAK,CAAC,CAAC;oBACvC,WAAW,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC;gBAClC,CAAC;gBACD,OAAO;YACT,CAAC;YAED,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,IAAI,CAAC;gBAAE,OAAO;YAEjC,IAAI,IAAI,CAAC,CAAC,CAAC,KAAK,GAAG,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC,KAAK,IAAI,IAAI,IAAI,CAAC,CAAC,CAAC,KAAK,IAAI,CAAC,EAAE,CAAC;gBAC9D,MAAM,SAAS,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,EAAE,CAA4B,CAAC;gBAC7D,MAAM,IAAI,GAAG,WAAW,CAAC,SAAS,CAAC,QAAQ,EAAE,EAAE,OAAO,EAAE,IAAI,EAAE,MAAM,EAAE,KAAK,EAAE,CAAC;qBAC3E,IAAI,EAAE;qBACN,OAAO,CAAC,KAAK,EAAE,GAAG,CAAC;qBACnB,OAAO,CAAC,KAAK,EAAE,MAAM,CAAC,CAAE,2BAA2B;qBACnD,OAAO,CAAC,KAAK,EAAE,KAAK,CAAC,CAAC,CAAE,oBAAoB;gBAC/C,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;YACnB,CAAC;iBAAM,IAAI,IAAI,CAAC,CAAC,CAAC,KAAK,GAAG,IAAI,OAAO,IAAI,CAAC,CAAC,CAAC,KAAK,QAAQ,IAAK,IAAI,CAAC,CAAC,CAAY,CAAC,UAAU,CAAC,IAAI,CAAC,EAAE,CAAC;gBAClG,2BAA2B;gBAC3B,MAAM,KAAK,GAAI,IAAI,CAAC,CAAC,CAAY,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;gBAC3C,IAAI,CAAC,WAAW,CAAC,GAAG,CAAC,KAAK,CAAC,EAAE,CAAC;oBAC5B,WAAW,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC;oBACvB,MAAM,OAAO,GAAG,cAAc,CAAC,KAAK,CAAC,CAAC;oBACtC,IAAI,OAAO;wBAAE,SAAS,CAAC,OAAO,EAAE,KAAK,CAAC,CAAC;oBACvC,WAAW,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC;gBAC5B,CAAC;YACH,CAAC;iBAAM,CAAC;gBACN,KAAK,MAAM,KAAK,IAAI,IAAI;oBAAE,SAAS,CAAC,KAAK,EAAE,KAAK,CAAC,CAAC;YACpD,CAAC;QAAA,CACF;QAED,SAAS,CAAC,KAAK,CAAC,QAAQ,CAAC,CAAC;QAC1B,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC;YAAE,OAAO,EAAE,CAAC;QAEjC,MAAM,QAAQ,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC;QACtD,IAAI,EAAE,GAAG,EAAE,CAAC;QACZ,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;YACrC,MAAM,GAAG,GAAG,IAAI,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,QAAQ,GAAG,IAAI,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC,CAAC;YACtE,EAAE,IAAI,IAAI,GAAG,GAAG,CAAC,IAAI,CAAC,KAAK,CAAC,GAAG,MAAM,CAAC;YACtC,IAAI,CAAC,KAAK,cAAc,GAAG,CAAC,IAAI,CAAC,cAAc,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC;gBAClE,EAAE,IAAI,IAAI,GAAG,KAAK,CAAC,QAAQ,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,GAAG,MAAM,CAAC;YAChE,CAAC;QACH,CAAC;QACD,OAAO,EAAE,CAAC;IAAA,CACX;IAED,4CAA4C;IAC5C,MAAM,SAAS,GAAG,cAAc,CAAC,IAAI,CAAC,CAAC;IAEvC,IAAI,SAAS,EAAE,CAAC;QACd,MAAM,OAAO,GAAG,WAAW,CAAC,SAAS,CAAC,CAAC;QACvC,IAAI,OAAO,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,GAAG,EAAE,CAAC;YAChC,MAAM,OAAO,GAAG,OAAO;iBACpB,OAAO,CAAC,SAAS,EAAE,MAAM,CAAC;iBAC1B,IAAI,EAAE,CAAC;YACV,OAAO,EAAE,KAAK,EAAE,OAAO,EAAE,OAAO,EAAE,CAAC;QACrC,CAAC;IACH,CAAC;IAED,6BAA6B;IAC7B,MAAM,YAAY,GAAsC,EAAE,CAAC;IAE3D,KAAK,MAAM,CAAC,EAAE,CAAC,IAAI,QAAQ,EAAE,CAAC;QAC5B,IAAI,EAAE,KAAK,IAAI;YAAE,SAAS;QAC1B,MAAM,MAAM,GAAG,cAAc,CAAC,EAAE,CAAC,CAAC;QAClC,IAAI,CAAC,MAAM;YAAE,SAAS;QAEtB,WAAW,CAAC,KAAK,EAAE,CAAC;QACpB,MAAM,IAAI,GAAG,WAAW,CAAC,MAAM,CAAC,CAAC;QAEjC,IAAI,IAAI,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,EAAE;YACvB,CAAC,IAAI,CAAC,QAAQ,CAAC,oBAAoB,CAAC;YACpC,CAAC,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,CAAC;YAC1B,YAAY,CAAC,IAAI,CAAC,EAAE,KAAK,EAAE,QAAQ,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,IAAI,EAAE,IAAI,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC;QACpE,CAAC;IACH,CAAC;IAED,IAAI,YAAY,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,IAAI,CAAC;IAE3C,YAAY,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,KAAK,GAAG,CAAC,CAAC,KAAK,CAAC,CAAC;IAE/C,MAAM,IAAI,GAAG,IAAI,GAAG,EAAU,CAAC;IAC/B,MAAM,WAAW,GAAa,EAAE,CAAC;IACjC,KAAK,MAAM,IAAI,IAAI,YAAY,EAAE,CAAC;QAChC,MAAM,GAAG,GAAG,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC;QACpC,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,CAAC;YACnB,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;YACd,WAAW,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QAC9B,CAAC;IACH,CAAC;IAED,MAAM,OAAO,GAAG,WAAW,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,OAAO,CAAC,SAAS,EAAE,MAAM,CAAC,CAAC,IAAI,EAAE,CAAC;IAC3E,OAAO,OAAO,CAAC,MAAM,GAAG,GAAG,CAAC,CAAC,CAAC,EAAE,KAAK,EAAE,OAAO,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC;AAAA,CACzD","sourcesContent":["// @ts-nocheck\n/**\n * RSC Content Extractor\n * \n * Extracts readable content from Next.js React Server Components (RSC) flight payloads.\n * RSC pages embed content as JSON in <script>self.__next_f.push([...])</script> tags.\n */\n\nexport interface RSCExtractResult {\n  title: string;\n  content: string;\n}\n\nexport function extractRSCContent(html: string): RSCExtractResult | null {\n  if (!html.includes(\"self.__next_f.push\")) {\n    return null;\n  }\n\n  // Parse all RSC chunks into a map\n  const chunkMap = new Map<string, string>();\n  const scriptRegex = /<script>self\\.__next_f\\.push\\(\\[1,\"([\\s\\S]*?)\"\\]\\)<\\/script>/g;\n\n  for (const match of html.matchAll(scriptRegex)) {\n    let content: string;\n    try {\n      content = JSON.parse('\"' + match[1] + '\"');\n    } catch {\n      continue;\n    }\n\n    // Parse each line as \"id:payload\"\n    // Lines are separated by \\n, each line is one chunk\n    // Chunk IDs are hex strings, typically 1-4 chars (supports up to 65535 chunks)\n    for (const line of content.split(\"\\n\")) {\n      if (!line.trim()) continue;\n      \n      const colonIdx = line.indexOf(\":\");\n      if (colonIdx <= 0 || colonIdx > 4) continue;\n      \n      const id = line.slice(0, colonIdx);\n      if (!/^[0-9a-f]+$/i.test(id)) continue;\n      \n      const payload = line.slice(colonIdx + 1);\n      if (!payload) continue;\n      \n      const existing = chunkMap.get(id);\n      if (!existing || payload.length > existing.length) {\n        chunkMap.set(id, payload);\n      }\n    }\n  }\n\n  if (chunkMap.size === 0) return null;\n\n  // Extract title\n  const titleMatch = html.match(/<title[^>]*>([^<]+)<\\/title>/);\n  const title = titleMatch?.[1]?.split(\"|\")[0]?.trim() || \"\";\n\n  // Parse and cache parsed chunks\n  const parsedCache = new Map<string, unknown>();\n  \n  function getParsedChunk(id: string): unknown | null {\n    if (parsedCache.has(id)) return parsedCache.get(id);\n    \n    const chunk = chunkMap.get(id);\n    if (!chunk || !chunk.startsWith(\"[\")) {\n      parsedCache.set(id, null);\n      return null;\n    }\n    \n    try {\n      const parsed = JSON.parse(chunk);\n      parsedCache.set(id, parsed);\n      return parsed;\n    } catch {\n      parsedCache.set(id, null);\n      return null;\n    }\n  }\n\n  // Extract markdown from nodes, resolving refs on the fly\n  type Node = unknown;\n  const visitedRefs = new Set<string>();\n\n  function extractNode(node: Node, ctx = { inTable: false, inCode: false }): string {\n    if (node === null || node === undefined) return \"\";\n    \n    if (typeof node === \"string\") {\n      // Check if it's a reference like \"$L30\"\n      const refMatch = node.match(/^\\$L([0-9a-f]+)$/i);\n      if (refMatch) {\n        const refId = refMatch[1];\n        if (visitedRefs.has(refId)) return \"\"; // Prevent cycles\n        visitedRefs.add(refId);\n        const refNode = getParsedChunk(refId);\n        const result = refNode ? extractNode(refNode, ctx) : \"\";\n        visitedRefs.delete(refId);\n        return result;\n      }\n      // Filter out RSC-specific artifacts, but preserve content inside code blocks\n      if (!ctx.inCode && (node === \"$undefined\" || node === \"$\" || /^\\$[A-Z]/.test(node))) return \"\";\n      return node.trim() ? node : \"\";\n    }\n    \n    if (typeof node === \"number\") return String(node);\n    if (typeof node === \"boolean\") return \"\";\n    if (!Array.isArray(node)) return \"\";\n\n    // RSC element: [\"$\", \"tag\", key, props]\n    if (node[0] === \"$\" && typeof node[1] === \"string\") {\n      const tag = node[1] as string;\n      const props = (node[3] || {}) as Record<string, unknown>;\n\n      // Skip non-content\n      const skipTags = [\"script\", \"style\", \"svg\", \"path\", \"circle\", \"link\", \"meta\", \n                        \"template\", \"button\", \"input\", \"nav\", \"footer\", \"aside\"];\n      if (skipTags.includes(tag)) return \"\";\n\n      // Component ref like $L25\n      if (tag.startsWith(\"$L\")) {\n        const refId = tag.slice(2);\n        if (visitedRefs.has(refId)) return \"\";\n        \n        // Check for heading components with baseId\n        if (props.baseId && props.children) {\n          return `## ${String(props.children)}\\n\\n`;\n        }\n        \n        visitedRefs.add(refId);\n        const refNode = getParsedChunk(refId);\n        let result = \"\";\n        if (refNode) {\n          result = extractNode(refNode, ctx);\n        } else if (props.children) {\n          result = extractNode(props.children as Node, ctx);\n        }\n        visitedRefs.delete(refId);\n        return result;\n      }\n\n      const children = props.children;\n      const content = children ? extractNode(children as Node, ctx) : \"\";\n\n      switch (tag) {\n        case \"h1\": return `# ${content.trim()}\\n\\n`;\n        case \"h2\": return `## ${content.trim()}\\n\\n`;\n        case \"h3\": return `### ${content.trim()}\\n\\n`;\n        case \"h4\": return `#### ${content.trim()}\\n\\n`;\n        case \"h5\": return `##### ${content.trim()}\\n\\n`;\n        case \"h6\": return `###### ${content.trim()}\\n\\n`;\n        case \"p\": return ctx.inTable ? content : `${content.trim()}\\n\\n`;\n        case \"code\": {\n          const codeContent = children ? extractNode(children as Node, { ...ctx, inCode: true }) : \"\";\n          return ctx.inCode ? codeContent : `\\`${codeContent}\\``;\n        }\n        case \"pre\": {\n          const preContent = children ? extractNode(children as Node, { ...ctx, inCode: true }) : \"\";\n          return \"```\\n\" + preContent + \"\\n```\\n\\n\";\n        }\n        case \"strong\": case \"b\": return `**${content}**`;\n        case \"em\": case \"i\": return `*${content}*`;\n        case \"li\": return `- ${content.trim()}\\n`;\n        case \"ul\": case \"ol\": return content + \"\\n\";\n        case \"blockquote\": return `> ${content.trim()}\\n\\n`;\n        case \"table\": return extractTable(node as unknown[]) + \"\\n\";\n        case \"thead\": case \"tbody\": case \"tr\": case \"th\": case \"td\":\n          return content;\n        case \"div\":\n          if (props.role === \"alert\" || props[\"data-slot\"] === \"alert\") {\n            return `> ${content.trim()}\\n\\n`;\n          }\n          return content;\n        case \"a\": {\n          const href = props.href as string | undefined;\n          return href && !href.startsWith(\"#\") ? `[${content}](${href})` : content;\n        }\n        default: return content;\n      }\n    }\n\n    // Array of child nodes\n    return (node as Node[]).map(n => extractNode(n, ctx)).join(\"\");\n  }\n\n  function extractTable(tableNode: unknown[]): string {\n    const props = (tableNode[3] || {}) as Record<string, unknown>;\n    const rows: string[][] = [];\n    let headerRowCount = 0;\n\n    function walkTable(node: unknown, isHeader = false): void {\n      if (node === null || node === undefined) return;\n      \n      // Handle string refs\n      if (typeof node === \"string\") {\n        const refMatch = node.match(/^\\$L([0-9a-f]+)$/i);\n        if (refMatch && !visitedRefs.has(refMatch[1])) {\n          visitedRefs.add(refMatch[1]);\n          const refNode = getParsedChunk(refMatch[1]);\n          if (refNode) walkTable(refNode, isHeader);\n          visitedRefs.delete(refMatch[1]);\n        }\n        return;\n      }\n      \n      if (!Array.isArray(node)) return;\n      \n      if (node[0] === \"$\") {\n        const tag = node[1] as string;\n        const nodeProps = (node[3] || {}) as Record<string, unknown>;\n        \n        // Handle component refs\n        if (tag.startsWith(\"$L\")) {\n          const refId = tag.slice(2);\n          if (!visitedRefs.has(refId)) {\n            visitedRefs.add(refId);\n            const refNode = getParsedChunk(refId);\n            if (refNode) walkTable(refNode, isHeader);\n            visitedRefs.delete(refId);\n          }\n          return;\n        }\n        \n        if (tag === \"thead\") walkTable(nodeProps.children, true);\n        else if (tag === \"tbody\") walkTable(nodeProps.children, false);\n        else if (tag === \"tr\") {\n          const cells: string[] = [];\n          walkCells(nodeProps.children, cells);\n          if (cells.length > 0) {\n            rows.push(cells);\n            if (isHeader) headerRowCount++;\n          }\n        } else walkTable(nodeProps.children, isHeader);\n      } else {\n        for (const child of node) walkTable(child, isHeader);\n      }\n    }\n\n    function walkCells(node: unknown, cells: string[]): void {\n      if (node === null || node === undefined) return;\n      \n      // Handle string refs\n      if (typeof node === \"string\") {\n        const refMatch = node.match(/^\\$L([0-9a-f]+)$/i);\n        if (refMatch && !visitedRefs.has(refMatch[1])) {\n          visitedRefs.add(refMatch[1]);\n          const refNode = getParsedChunk(refMatch[1]);\n          if (refNode) walkCells(refNode, cells);\n          visitedRefs.delete(refMatch[1]);\n        }\n        return;\n      }\n      \n      if (!Array.isArray(node)) return;\n      \n      if (node[0] === \"$\" && (node[1] === \"td\" || node[1] === \"th\")) {\n        const cellProps = (node[3] || {}) as Record<string, unknown>;\n        const text = extractNode(cellProps.children, { inTable: true, inCode: false })\n          .trim()\n          .replace(/\\n/g, \" \")\n          .replace(/\\\\/g, \"\\\\\\\\\")  // Escape backslashes first\n          .replace(/\\|/g, \"\\\\|\");  // Then escape pipes\n        cells.push(text);\n      } else if (node[0] === \"$\" && typeof node[1] === \"string\" && (node[1] as string).startsWith(\"$L\")) {\n        // Component ref for a cell\n        const refId = (node[1] as string).slice(2);\n        if (!visitedRefs.has(refId)) {\n          visitedRefs.add(refId);\n          const refNode = getParsedChunk(refId);\n          if (refNode) walkCells(refNode, cells);\n          visitedRefs.delete(refId);\n        }\n      } else {\n        for (const child of node) walkCells(child, cells);\n      }\n    }\n\n    walkTable(props.children);\n    if (rows.length === 0) return \"\";\n\n    const colCount = Math.max(...rows.map(r => r.length));\n    let md = \"\";\n    for (let i = 0; i < rows.length; i++) {\n      const row = rows[i].concat(Array(colCount - rows[i].length).fill(\"\"));\n      md += \"| \" + row.join(\" | \") + \" |\\n\";\n      if (i === headerRowCount - 1 || (headerRowCount === 0 && i === 0)) {\n        md += \"| \" + Array(colCount).fill(\"---\").join(\" | \") + \" |\\n\";\n      }\n    }\n    return md;\n  }\n\n  // Process main content chunk (usually \"23\")\n  const mainChunk = getParsedChunk(\"23\");\n  \n  if (mainChunk) {\n    const content = extractNode(mainChunk);\n    if (content.trim().length > 100) {\n      const cleaned = content\n        .replace(/\\n{3,}/g, \"\\n\\n\")\n        .trim();\n      return { title, content: cleaned };\n    }\n  }\n\n  // Fallback: try other chunks\n  const contentParts: { order: number; text: string }[] = [];\n\n  for (const [id] of chunkMap) {\n    if (id === \"23\") continue;\n    const parsed = getParsedChunk(id);\n    if (!parsed) continue;\n\n    visitedRefs.clear();\n    const text = extractNode(parsed);\n\n    if (text.trim().length > 50 && \n        !text.includes(\"page was not found\") && \n        !text.includes(\"404\")) {\n      contentParts.push({ order: parseInt(id, 16), text: text.trim() });\n    }\n  }\n\n  if (contentParts.length === 0) return null;\n\n  contentParts.sort((a, b) => a.order - b.order);\n  \n  const seen = new Set<string>();\n  const uniqueParts: string[] = [];\n  for (const part of contentParts) {\n    const key = part.text.slice(0, 150);\n    if (!seen.has(key)) {\n      seen.add(key);\n      uniqueParts.push(part.text);\n    }\n  }\n\n  const content = uniqueParts.join(\"\\n\\n\").replace(/\\n{3,}/g, \"\\n\\n\").trim();\n  return content.length > 100 ? { title, content } : null;\n}\n"]}

package/dist/builtin-extensions/pi-web-access/ssrf-protection.d.ts

@@ -0,0 +1,24 @@
+export type LookupAddress = {
+    address: string;
+    family: number;
+};
+export type Lookup = (hostname: string) => Promise<LookupAddress[]>;
+type Fetch = typeof fetch;
+interface ValidationOptions {
+    lookup?: Lookup;
+    /**
+     * CIDR ranges (e.g. "198.18.0.0/15") to exempt from the SSRF guard.
+     * Useful when a host runs a TUN/fake-IP proxy (Surge, Clash, Mihomo, ...)
+     * that resolves public domains into a reserved range. Entries are validated
+     * strictly; an invalid entry throws so misconfiguration is not silent.
+     */
+    allowRanges?: string[];
+}
+interface FetchRemoteOptions extends ValidationOptions {
+    fetch?: Fetch;
+    maxRedirects?: number;
+}
+export declare function validateRemoteUrl(rawUrl: string | URL, options?: ValidationOptions): Promise<URL>;
+export declare function fetchRemoteUrl(url: string | URL, init?: RequestInit, options?: FetchRemoteOptions): Promise<Response>;
+export {};
+//# sourceMappingURL=ssrf-protection.d.ts.map

package/dist/builtin-extensions/pi-web-access/ssrf-protection.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"ssrf-protection.d.ts","sourceRoot":"","sources":["../../../src/builtin-extensions/pi-web-access/ssrf-protection.ts"],"names":[],"mappings":"AAOA,MAAM,MAAM,aAAa,GAAG;IAAE,OAAO,EAAE,MAAM,CAAC;IAAC,MAAM,EAAE,MAAM,CAAA;CAAE,CAAC;AAChE,MAAM,MAAM,MAAM,GAAG,CAAC,QAAQ,EAAE,MAAM,KAAK,OAAO,CAAC,aAAa,EAAE,CAAC,CAAC;AACpE,KAAK,KAAK,GAAG,OAAO,KAAK,CAAC;AAE1B,UAAU,iBAAiB;IAC1B,MAAM,CAAC,EAAE,MAAM,CAAC;IAChB;;;;;OAKG;IACH,WAAW,CAAC,EAAE,MAAM,EAAE,CAAC;CACvB;AAQD,UAAU,kBAAmB,SAAQ,iBAAiB;IACrD,KAAK,CAAC,EAAE,KAAK,CAAC;IACd,YAAY,CAAC,EAAE,MAAM,CAAC;CACtB;AAMD,wBAAsB,iBAAiB,CAAC,MAAM,EAAE,MAAM,GAAG,GAAG,EAAE,OAAO,GAAE,iBAAsB,GAAG,OAAO,CAAC,GAAG,CAAC,CAgC3G;AAED,wBAAsB,cAAc,CACnC,GAAG,EAAE,MAAM,GAAG,GAAG,EACjB,IAAI,GAAE,WAAgB,EACtB,OAAO,GAAE,kBAAuB,GAC9B,OAAO,CAAC,QAAQ,CAAC,CAsBnB","sourcesContent":["// @ts-nocheck\nimport { lookup as dnsLookup } from \"node:dns/promises\";\nimport net from \"node:net\";\n\nconst DEFAULT_MAX_REDIRECTS = 5;\nconst REDIRECT_STATUSES = new Set([301, 302, 303, 307, 308]);\n\nexport type LookupAddress = { address: string; family: number };\nexport type Lookup = (hostname: string) => Promise<LookupAddress[]>;\ntype Fetch = typeof fetch;\n\ninterface ValidationOptions {\n\tlookup?: Lookup;\n\t/**\n\t * CIDR ranges (e.g. \"198.18.0.0/15\") to exempt from the SSRF guard.\n\t * Useful when a host runs a TUN/fake-IP proxy (Surge, Clash, Mihomo, ...)\n\t * that resolves public domains into a reserved range. Entries are validated\n\t * strictly; an invalid entry throws so misconfiguration is not silent.\n\t */\n\tallowRanges?: string[];\n}\n\n/** Parsed entry from `allowRanges`: a network address (4 or 16 bytes) + prefix length. */\ninterface ParsedCidr {\n\tbytes: Uint8Array;\n\tprefix: number;\n}\n\ninterface FetchRemoteOptions extends ValidationOptions {\n\tfetch?: Fetch;\n\tmaxRedirects?: number;\n}\n\nasync function defaultLookup(hostname: string): Promise<LookupAddress[]> {\n\treturn dnsLookup(hostname, { all: true, verbatim: true });\n}\n\nexport async function validateRemoteUrl(rawUrl: string | URL, options: ValidationOptions = {}): Promise<URL> {\n\tconst url = rawUrl instanceof URL ? rawUrl : new URL(rawUrl);\n\tif (url.protocol !== \"http:\" && url.protocol !== \"https:\") {\n\t\tthrow new Error(\"Only HTTP and HTTPS URLs can be fetched remotely\");\n\t}\n\n\tconst hostname = normalizeHostname(url.hostname);\n\tif (!hostname) throw new Error(\"URL must include a hostname\");\n\tif (hostname === \"localhost\" || hostname.endsWith(\".localhost\")) {\n\t\tthrow new Error(`Blocked internal hostname: ${hostname}`);\n\t}\n\n\tconst allowRanges = parseAllowRanges(options.allowRanges);\n\n\tif (net.isIP(hostname)) {\n\t\tassertPublicAddress(hostname, hostname, allowRanges);\n\t\treturn url;\n\t}\n\n\tlet addresses: LookupAddress[];\n\ttry {\n\t\taddresses = await (options.lookup ?? defaultLookup)(hostname);\n\t} catch (err) {\n\t\tconst message = err instanceof Error ? err.message : String(err);\n\t\tthrow new Error(`Failed to resolve ${hostname}: ${message}`);\n\t}\n\n\tif (addresses.length === 0) throw new Error(`Failed to resolve ${hostname}: no addresses returned`);\n\tfor (const { address } of addresses) {\n\t\tassertPublicAddress(address, hostname, allowRanges);\n\t}\n\treturn url;\n}\n\nexport async function fetchRemoteUrl(\n\turl: string | URL,\n\tinit: RequestInit = {},\n\toptions: FetchRemoteOptions = {},\n): Promise<Response> {\n\tconst fetchImpl = options.fetch ?? fetch;\n\tconst maxRedirects = options.maxRedirects ?? DEFAULT_MAX_REDIRECTS;\n\tlet current = await validateRemoteUrl(url, options);\n\tlet requestInit = init;\n\n\tfor (let redirects = 0; redirects <= maxRedirects; redirects++) {\n\t\tconst response = await fetchImpl(current, { ...requestInit, redirect: \"manual\" });\n\t\tif (!REDIRECT_STATUSES.has(response.status)) return response;\n\n\t\tconst location = response.headers.get(\"location\");\n\t\tif (!location) return response;\n\t\tif (redirects === maxRedirects) throw new Error(`Too many redirects fetching ${current.toString()}`);\n\n\t\tcurrent = await validateRemoteUrl(new URL(location, current), options);\n\t\tif (response.status === 303 || ((response.status === 301 || response.status === 302) && requestInit.method?.toUpperCase() === \"POST\")) {\n\t\t\tconst { body: _body, ...nextInit } = requestInit;\n\t\t\trequestInit = { ...nextInit, method: \"GET\" };\n\t\t}\n\t}\n\n\tthrow new Error(`Too many redirects fetching ${current.toString()}`);\n}\n\nfunction normalizeHostname(hostname: string): string {\n\treturn hostname.toLowerCase().replace(/^\\[|\\]$/g, \"\").replace(/\\.$/, \"\");\n}\n\nfunction assertPublicAddress(address: string, hostname: string, allowRanges: ParsedCidr[] = []): void {\n\tconst normalized = normalizeHostname(address);\n\tconst ipVersion = net.isIP(normalized);\n\tif (ipVersion === 0) throw new Error(`Resolved non-IP address for ${hostname}: ${address}`);\n\t// Explicitly-allowed ranges bypass the private/reserved checks below. This lets\n\t// users exempt synthetic ranges produced by TUN/fake-IP proxies (e.g. 198.18/15).\n\tif (isInAllowedRange(normalized, ipVersion, allowRanges)) return;\n\tif (ipVersion === 4 && isBlockedIPv4(normalized)) {\n\t\tthrow new Error(`Blocked internal address for ${hostname}: ${normalized}`);\n\t}\n\tif (ipVersion === 6 && isBlockedIPv6(normalized)) {\n\t\tthrow new Error(`Blocked internal address for ${hostname}: ${normalized}`);\n\t}\n}\n\nfunction isBlockedIPv4(address: string): boolean {\n\tconst parts = address.split(\".\").map(part => Number(part));\n\tif (parts.length !== 4 || parts.some(part => !Number.isInteger(part) || part < 0 || part > 255)) return true;\n\tconst [a, b] = parts;\n\treturn a === 0 ||\n\t\ta === 10 ||\n\t\ta === 127 ||\n\t\t(a === 100 && b >= 64 && b <= 127) ||\n\t\t(a === 169 && b === 254) ||\n\t\t(a === 172 && b >= 16 && b <= 31) ||\n\t\t(a === 192 && b === 168) ||\n\t\t(a === 198 && (b === 18 || b === 19)) ||\n\t\ta >= 224;\n}\n\nfunction isBlockedIPv6(address: string): boolean {\n\tconst groups = parseIPv6(address);\n\tif (!groups) return true;\n\n\tconst first = groups[0];\n\tif (groups.every(group => group === 0)) return true;\n\tif (groups.slice(0, 7).every(group => group === 0) && groups[7] === 1) return true;\n\tif ((first & 0xfe00) === 0xfc00) return true;\n\tif ((first & 0xffc0) === 0xfe80) return true;\n\n\tconst isMappedIPv4 = groups.slice(0, 5).every(group => group === 0) && groups[5] === 0xffff;\n\tif (isMappedIPv4) {\n\t\tconst ipv4 = [groups[6] >> 8, groups[6] & 0xff, groups[7] >> 8, groups[7] & 0xff].join(\".\");\n\t\treturn isBlockedIPv4(ipv4);\n\t}\n\n\treturn false;\n}\n\nfunction parseIPv6(address: string): number[] | null {\n\tif (address.includes(\".\")) {\n\t\tconst lastColon = address.lastIndexOf(\":\");\n\t\tconst ipv4 = address.slice(lastColon + 1);\n\t\tif (net.isIP(ipv4) !== 4) return null;\n\t\tconst octets = ipv4.split(\".\").map(part => Number(part));\n\t\taddress = `${address.slice(0, lastColon)}:${((octets[0] << 8) | octets[1]).toString(16)}:${((octets[2] << 8) | octets[3]).toString(16)}`;\n\t}\n\n\tconst pieces = address.split(\"::\");\n\tif (pieces.length > 2) return null;\n\n\tconst left = pieces[0] ? pieces[0].split(\":\") : [];\n\tconst right = pieces.length === 2 && pieces[1] ? pieces[1].split(\":\") : [];\n\tconst missing = 8 - left.length - right.length;\n\tif (pieces.length === 1 && missing !== 0) return null;\n\tif (pieces.length === 2 && missing < 0) return null;\n\n\tconst groups = [...left, ...Array(missing).fill(\"0\"), ...right].map(part => {\n\t\tif (!/^[0-9a-f]{1,4}$/i.test(part)) return -1;\n\t\treturn parseInt(part, 16);\n\t});\n\treturn groups.length === 8 && groups.every(group => group >= 0 && group <= 0xffff) ? groups : null;\n}\n\n/** Parse `allowRanges` config value into validated CIDR rules. Throws on malformed entries. */\nfunction parseAllowRanges(input: unknown): ParsedCidr[] {\n\tif (input === undefined || input === null) return [];\n\tif (!Array.isArray(input)) {\n\t\tthrow new Error(\"ssrf.allowRanges must be an array of CIDR strings\");\n\t}\n\tconst rules: ParsedCidr[] = [];\n\tfor (const entry of input) {\n\t\tif (typeof entry !== \"string\") {\n\t\t\tthrow new Error(`ssrf.allowRanges entries must be strings, got ${typeof entry}`);\n\t\t}\n\t\tconst rule = parseCidr(entry.trim());\n\t\tif (!rule) {\n\t\t\tthrow new Error(`Invalid CIDR notation in ssrf.allowRanges: \"${entry}\"`);\n\t\t}\n\t\trules.push(rule);\n\t}\n\treturn rules;\n}\n\n/** Parse a single CIDR (e.g. \"198.18.0.0/15\", \"fd00::/8\") or bare host (\"1.2.3.4\"). Returns null if invalid. */\nfunction parseCidr(raw: string): ParsedCidr | null {\n\tif (!raw) return null;\n\tconst slash = raw.lastIndexOf(\"/\");\n\tconst addrPart = slash >= 0 ? raw.slice(0, slash) : raw;\n\tconst prefixPart = slash >= 0 ? raw.slice(slash + 1) : null;\n\t// A slash must be followed by digits. Number(\"\")/Number(\" \") are 0, which\n\t// would silently turn \"198.18.0.0/\" into /0 and exempt every address.\n\tif (prefixPart !== null && !/^\\d+$/.test(prefixPart)) return null;\n\tconst version = net.isIP(addrPart);\n\n\tif (version === 4) {\n\t\tconst bytes = ipv4ToBytes(addrPart);\n\t\tif (!bytes) return null;\n\t\tconst prefix = prefixPart === null ? 32 : Number(prefixPart);\n\t\tif (!Number.isInteger(prefix) || prefix < 1 || prefix > 32) return null;\n\t\treturn { bytes, prefix };\n\t}\n\tif (version === 6) {\n\t\tconst groups = parseIPv6(addrPart);\n\t\tif (!groups) return null;\n\t\tconst prefix = prefixPart === null ? 128 : Number(prefixPart);\n\t\tif (!Number.isInteger(prefix) || prefix < 1 || prefix > 128) return null;\n\t\treturn { bytes: ipv6GroupsToBytes(groups), prefix };\n\t}\n\treturn null;\n}\n\nfunction ipv4ToBytes(address: string): Uint8Array | null {\n\tconst parts = address.split(\".\");\n\tif (parts.length !== 4) return null;\n\tconst bytes = new Uint8Array(4);\n\tfor (let i = 0; i < 4; i++) {\n\t\tconst octet = Number(parts[i]);\n\t\tif (!Number.isInteger(octet) || octet < 0 || octet > 255) return null;\n\t\tbytes[i] = octet;\n\t}\n\treturn bytes;\n}\n\nfunction ipv6GroupsToBytes(groups: number[]): Uint8Array {\n\tconst bytes = new Uint8Array(16);\n\tfor (let i = 0; i < 8; i++) {\n\t\tbytes[i * 2] = groups[i] >> 8;\n\t\tbytes[i * 2 + 1] = groups[i] & 0xff;\n\t}\n\treturn bytes;\n}\n\nfunction ipToBytes(address: string, version: number): Uint8Array | null {\n\tif (version === 4) return ipv4ToBytes(address);\n\tif (version === 6) {\n\t\tconst groups = parseIPv6(address);\n\t\treturn groups ? ipv6GroupsToBytes(groups) : null;\n\t}\n\treturn null;\n}\n\n/** True if `address` (already validated as `ipVersion`) falls within any allowed CIDR. */\nfunction isInAllowedRange(address: string, ipVersion: number, allowRanges: ParsedCidr[]): boolean {\n\tif (allowRanges.length === 0) return false;\n\tconst addrBytes = ipToBytes(address, ipVersion);\n\tif (!addrBytes) return false;\n\tfor (const rule of allowRanges) {\n\t\t// Only compare same-family rules (4-byte IPv4 vs 16-byte IPv6).\n\t\tif (rule.bytes.length !== addrBytes.length) continue;\n\t\tif (bytesMatchPrefix(addrBytes, rule.bytes, rule.prefix)) return true;\n\t}\n\treturn false;\n}\n\n/** Compare the leading `prefix` bits of two equal-length byte arrays. */\nfunction bytesMatchPrefix(addr: Uint8Array, network: Uint8Array, prefix: number): boolean {\n\tconst fullBytes = prefix >> 3;\n\tconst remBits = prefix & 7;\n\tfor (let i = 0; i < fullBytes; i++) {\n\t\tif (addr[i] !== network[i]) return false;\n\t}\n\tif (remBits > 0 && fullBytes < addr.length) {\n\t\tconst mask = (0xff << (8 - remBits)) & 0xff;\n\t\tif ((addr[fullBytes] & mask) !== (network[fullBytes] & mask)) return false;\n\t}\n\treturn true;\n}\n"]}