@mz1999/defuddle 0.14.1

package/dist/extractors/_conversation.js

@@ -0,0 +1,78 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.ConversationExtractor = void 0;
+const _base_1 = require("./_base");
+const defuddle_1 = require("../defuddle");
+const dom_1 = require("../utils/dom");
+class ConversationExtractor extends _base_1.BaseExtractor {
+    getFootnotes() {
+        return [];
+    }
+    extract() {
+        const messages = this.extractMessages();
+        const metadata = this.getMetadata();
+        const footnotes = this.getFootnotes();
+        const rawContentHtml = this.createContentHtml(messages, footnotes);
+        // Create a temporary document to run Defuddle on our content
+        const tempDoc = this.document.implementation.createHTMLDocument();
+        const container = tempDoc.createElement('article');
+        container.appendChild((0, dom_1.parseHTML)(tempDoc, rawContentHtml));
+        tempDoc.body.appendChild(container);
+        // Run Defuddle on our formatted content
+        const defuddled = new defuddle_1.Defuddle(tempDoc).parse();
+        const contentHtml = defuddled.content;
+        return {
+            content: contentHtml,
+            contentHtml: contentHtml,
+            extractedContent: {
+                messageCount: messages.length.toString(),
+            },
+            variables: {
+                title: metadata.title || 'Conversation',
+                site: metadata.site,
+                description: metadata.description || `${metadata.site} conversation with ${messages.length} messages`,
+                wordCount: defuddled.wordCount?.toString() || '',
+            }
+        };
+    }
+    createContentHtml(messages, footnotes) {
+        const messagesHtml = messages.map((message, index) => {
+            const timestampHtml = message.timestamp ?
+                `<div class="message-timestamp">${message.timestamp}</div>` : '';
+            // Check if content already has paragraph tags
+            const hasParagraphs = /<p[^>]*>[\s\S]*?<\/p>/i.test(message.content);
+            const contentHtml = hasParagraphs ? message.content : `<p>${message.content}</p>`;
+            // Add metadata to data attributes
+            const dataAttributes = message.metadata ?
+                Object.entries(message.metadata)
+                    .map(([key, value]) => `data-${key}="${value}"`)
+                    .join(' ') : '';
+            return `
+			<div class="message message-${message.author.toLowerCase()}" ${dataAttributes}>
+				<div class="message-header">
+					<p class="message-author"><strong>${message.author}</strong></p>
+					${timestampHtml}
+				</div>
+				<div class="message-content">
+					${contentHtml}
+				</div>
+			</div>${index < messages.length - 1 ? '\n<hr>' : ''}`;
+        }).join('\n').trim();
+        // Add footnotes section if we have any
+        const footnotesHtml = footnotes.length > 0 ? `
+			<div id="footnotes">
+				<ol>
+					${footnotes.map((footnote, index) => `
+						<li class="footnote" id="fn:${index + 1}">
+							<p>
+								<a href="${footnote.url}" target="_blank">${footnote.text}</a>&nbsp;<a href="#fnref:${index + 1}" class="footnote-backref">↩</a>
+							</p>
+						</li>
+					`).join('')}
+				</ol>
+			</div>` : '';
+        return `${messagesHtml}\n${footnotesHtml}`.trim();
+    }
+}
+exports.ConversationExtractor = ConversationExtractor;
+//# sourceMappingURL=_conversation.js.map

package/dist/extractors/_conversation.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"_conversation.js","sourceRoot":"","sources":["../../src/extractors/_conversation.ts"],"names":[],"mappings":";;;AAAA,mCAAwC;AAExC,0CAAuC;AACvC,sCAAyC;AAEzC,MAAsB,qBAAsB,SAAQ,qBAAa;IAGtD,YAAY;QACrB,OAAO,EAAE,CAAC;IACX,CAAC;IAED,OAAO;QACN,MAAM,QAAQ,GAAG,IAAI,CAAC,eAAe,EAAE,CAAC;QACxC,MAAM,QAAQ,GAAG,IAAI,CAAC,WAAW,EAAE,CAAC;QACpC,MAAM,SAAS,GAAG,IAAI,CAAC,YAAY,EAAE,CAAC;QACtC,MAAM,cAAc,GAAG,IAAI,CAAC,iBAAiB,CAAC,QAAQ,EAAE,SAAS,CAAC,CAAC;QAEnE,6DAA6D;QAC7D,MAAM,OAAO,GAAG,IAAI,CAAC,QAAQ,CAAC,cAAc,CAAC,kBAAkB,EAAE,CAAC;QAClE,MAAM,SAAS,GAAG,OAAO,CAAC,aAAa,CAAC,SAAS,CAAC,CAAC;QACnD,SAAS,CAAC,WAAW,CAAC,IAAA,eAAS,EAAC,OAAO,EAAE,cAAc,CAAC,CAAC,CAAC;QAC1D,OAAO,CAAC,IAAI,CAAC,WAAW,CAAC,SAAS,CAAC,CAAC;QAEpC,wCAAwC;QACxC,MAAM,SAAS,GAAG,IAAI,mBAAQ,CAAC,OAAO,CAAC,CAAC,KAAK,EAAE,CAAC;QAChD,MAAM,WAAW,GAAG,SAAS,CAAC,OAAO,CAAC;QAEtC,OAAO;YACN,OAAO,EAAE,WAAW;YACpB,WAAW,EAAE,WAAW;YACxB,gBAAgB,EAAE;gBACjB,YAAY,EAAE,QAAQ,CAAC,MAAM,CAAC,QAAQ,EAAE;aACxC;YACD,SAAS,EAAE;gBACV,KAAK,EAAE,QAAQ,CAAC,KAAK,IAAI,cAAc;gBACvC,IAAI,EAAE,QAAQ,CAAC,IAAI;gBACnB,WAAW,EAAE,QAAQ,CAAC,WAAW,IAAI,GAAG,QAAQ,CAAC,IAAI,sBAAsB,QAAQ,CAAC,MAAM,WAAW;gBACrG,SAAS,EAAE,SAAS,CAAC,SAAS,EAAE,QAAQ,EAAE,IAAI,EAAE;aAChD;SACD,CAAC;IACH,CAAC;IAES,iBAAiB,CAAC,QAA+B,EAAE,SAAqB;QACjF,MAAM,YAAY,GAAG,QAAQ,CAAC,GAAG,CAAC,CAAC,OAAO,EAAE,KAAK,EAAE,EAAE;YACpD,MAAM,aAAa,GAAG,OAAO,CAAC,SAAS,CAAC,CAAC;gBACxC,kCAAkC,OAAO,CAAC,SAAS,QAAQ,CAAC,CAAC,CAAC,EAAE,CAAC;YAElE,8CAA8C;YAC9C,MAAM,aAAa,GAAG,wBAAwB,CAAC,IAAI,CAAC,OAAO,CAAC,OAAO,CAAC,CAAC;YACrE,MAAM,WAAW,GAAG,aAAa,CAAC,CAAC,CAAC,OAAO,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,OAAO,CAAC,OAAO,MAAM,CAAC;YAElF,kCAAkC;YAClC,MAAM,cAAc,GAAG,OAAO,CAAC,QAAQ,CAAC,CAAC;gBACxC,MAAM,CAAC,OAAO,CAAC,OAAO,CAAC,QAAQ,CAAC;qBAC9B,GAAG,CAAC,CAAC,CAAC,GAAG,EAAE,KAAK,CAAC,EAAE,EAAE,CAAC,QAAQ,GAAG,KAAK,KAAK,GAAG,CAAC;qBAC/C,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;YAElB,OAAO;iCACuB,OAAO,CAAC,MAAM,CAAC,WAAW,EAAE,KAAK,cAAc;;yCAEvC,OAAO,CAAC,MAAM;OAChD,aAAa;;;OAGb,WAAW;;WAEP,KAAK,GAAG,QAAQ,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC;QACvD,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,CAAC;QAErB,uCAAuC;QACvC,MAAM,aAAa,GAAG,SAAS,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC;;;OAGxC,SAAS,CAAC,GAAG,CAAC,CAAC,QAAQ,EAAE,KAAK,EAAE,EAAE,CAAC;oCACN,KAAK,GAAG,CAAC;;mBAE1B,QAAQ,CAAC,GAAG,qBAAqB,QAAQ,CAAC,IAAI,6BAA6B,KAAK,GAAG,CAAC;;;MAGjG,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC;;UAEN,CAAC,CAAC,CAAC,EAAE,CAAC;QAEd,OAAO,GAAG,YAAY,KAAK,aAAa,EAAE,CAAC,IAAI,EAAE,CAAC;IACnD,CAAC;CACD;AAjFD,sDAiFC"}

package/dist/extractors/chatgpt.d.ts

@@ -0,0 +1,14 @@
+import { ConversationExtractor } from './_conversation';
+import { ConversationMessage, ConversationMetadata, Footnote } from '../types/extractors';
+export declare class ChatGPTExtractor extends ConversationExtractor {
+    private articles;
+    private footnotes;
+    private footnoteCounter;
+    private cachedMessages;
+    constructor(document: Document, url: string);
+    canExtract(): boolean;
+    protected extractMessages(): ConversationMessage[];
+    protected getFootnotes(): Footnote[];
+    protected getMetadata(): ConversationMetadata;
+    private getTitle;
+}

package/dist/extractors/chatgpt.js

@@ -0,0 +1,138 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.ChatGPTExtractor = void 0;
+const _conversation_1 = require("./_conversation");
+const dom_1 = require("../utils/dom");
+class ChatGPTExtractor extends _conversation_1.ConversationExtractor {
+    constructor(document, url) {
+        super(document, url);
+        this.cachedMessages = null;
+        this.articles = document.querySelectorAll('article[data-testid^="conversation-turn-"]');
+        this.footnotes = [];
+        this.footnoteCounter = 0;
+    }
+    canExtract() {
+        return !!this.articles && this.articles.length > 0;
+    }
+    extractMessages() {
+        if (this.cachedMessages)
+            return this.cachedMessages;
+        const messages = [];
+        this.footnotes = [];
+        this.footnoteCounter = 0;
+        if (!this.articles)
+            return messages;
+        this.articles.forEach((article) => {
+            // Get the localized author text from the sr-only heading and clean it
+            const authorElement = article.querySelector('h5.sr-only, h6.sr-only');
+            const authorText = authorElement?.textContent
+                ?.trim()
+                ?.replace(/:\s*$/, '') // Remove colon and any trailing whitespace
+                || '';
+            let currentAuthorRole = '';
+            const authorRole = article.getAttribute('data-message-author-role');
+            if (authorRole) {
+                currentAuthorRole = authorRole;
+            }
+            let messageContent = (0, dom_1.serializeHTML)(article);
+            messageContent = messageContent.replace(/\u200B/g, '');
+            // Remove specific elements from the message content
+            const tempDiv = this.document.createElement('div');
+            tempDiv.appendChild((0, dom_1.parseHTML)(this.document, messageContent));
+            tempDiv.querySelectorAll('h5.sr-only, h6.sr-only, span[data-state="closed"]').forEach(el => el.remove());
+            messageContent = (0, dom_1.serializeHTML)(tempDiv);
+            // Process inline references using regex to find the containers
+            // Look for spans containing citation links (a[target=_blank][rel=noopener]), replacing entire structure
+            // Also capture optional preceding ZeroWidthSpace
+            const citationPattern = /(&ZeroWidthSpace;)?(<span[^>]*?>\s*<a(?=[^>]*?href="([^"]+)")(?=[^>]*?target="_blank")(?=[^>]*?rel="noopener")[^>]*?>[\s\S]*?<\/a>\s*<\/span>)/gi;
+            messageContent = messageContent.replace(citationPattern, (match, zws, spanStructure, url) => {
+                // url is captured group 3
+                let domain = '';
+                let fragmentText = '';
+                try {
+                    // Extract domain without www.
+                    domain = new URL(url).hostname.replace(/^www\./, '');
+                    // Extract and decode the fragment text if it exists
+                    const hashParts = url.split('#:~:text=');
+                    if (hashParts.length > 1) {
+                        fragmentText = decodeURIComponent(hashParts[1]);
+                        fragmentText = fragmentText.replace(/%2C/g, ',');
+                        const parts = fragmentText.split(',');
+                        if (parts.length > 1 && parts[0].trim()) {
+                            fragmentText = ` — ${parts[0].trim()}...`;
+                        }
+                        else if (parts[0].trim()) {
+                            fragmentText = ` — ${fragmentText.trim()}`;
+                        }
+                        else {
+                            fragmentText = '';
+                        }
+                    }
+                }
+                catch (e) {
+                    console.error(`Failed to parse URL: ${url}`, e);
+                    domain = url;
+                }
+                // Check if this URL already exists in our footnotes
+                let footnoteIndex = this.footnotes.findIndex(fn => fn.url === url);
+                let footnoteNumber;
+                if (footnoteIndex === -1) {
+                    this.footnoteCounter++;
+                    footnoteNumber = this.footnoteCounter;
+                    this.footnotes.push({
+                        url,
+                        text: `<a href="${url}">${domain}</a>${fragmentText}`
+                    });
+                }
+                else {
+                    footnoteNumber = footnoteIndex + 1;
+                }
+                // Return just the footnote reference, replacing the ZWS (if captured) and the entire span structure
+                return `<sup id="fnref:${footnoteNumber}"><a href="#fn:${footnoteNumber}">${footnoteNumber}</a></sup>`;
+            });
+            // Clean up any stray empty paragraph tags
+            messageContent = messageContent
+                .replace(/<p[^>]*>\s*<\/p>/g, '');
+            messages.push({
+                author: authorText,
+                content: messageContent.trim(),
+                metadata: {
+                    role: currentAuthorRole || 'unknown'
+                }
+            });
+        });
+        this.cachedMessages = messages;
+        return messages;
+    }
+    getFootnotes() {
+        return this.footnotes;
+    }
+    getMetadata() {
+        const title = this.getTitle();
+        const messages = this.extractMessages();
+        return {
+            title,
+            site: 'ChatGPT',
+            url: this.url,
+            messageCount: messages.length,
+            description: `ChatGPT conversation with ${messages.length} messages`
+        };
+    }
+    getTitle() {
+        // Try to get the page title first
+        const pageTitle = this.document.title?.trim();
+        if (pageTitle && pageTitle !== 'ChatGPT') {
+            return pageTitle;
+        }
+        // Fall back to first user message
+        const firstUserTurn = this.articles?.item(0)?.querySelector('.text-message');
+        if (firstUserTurn) {
+            const text = firstUserTurn.textContent || '';
+            // Truncate to first 50 characters if longer
+            return text.length > 50 ? text.slice(0, 50) + '...' : text;
+        }
+        return 'ChatGPT Conversation';
+    }
+}
+exports.ChatGPTExtractor = ChatGPTExtractor;
+//# sourceMappingURL=chatgpt.js.map

package/dist/extractors/chatgpt.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"chatgpt.js","sourceRoot":"","sources":["../../src/extractors/chatgpt.ts"],"names":[],"mappings":";;;AAAA,mDAAwD;AAExD,sCAAwD;AAExD,MAAa,gBAAiB,SAAQ,qCAAqB;IAM1D,YAAY,QAAkB,EAAE,GAAW;QAC1C,KAAK,CAAC,QAAQ,EAAE,GAAG,CAAC,CAAC;QAHd,mBAAc,GAAiC,IAAI,CAAC;QAI3D,IAAI,CAAC,QAAQ,GAAG,QAAQ,CAAC,gBAAgB,CAAC,4CAA4C,CAAC,CAAC;QACxF,IAAI,CAAC,SAAS,GAAG,EAAE,CAAC;QACpB,IAAI,CAAC,eAAe,GAAG,CAAC,CAAC;IAC1B,CAAC;IAED,UAAU;QACT,OAAO,CAAC,CAAC,IAAI,CAAC,QAAQ,IAAI,IAAI,CAAC,QAAQ,CAAC,MAAM,GAAG,CAAC,CAAC;IACpD,CAAC;IAES,eAAe;QACxB,IAAI,IAAI,CAAC,cAAc;YAAE,OAAO,IAAI,CAAC,cAAc,CAAC;QAEpD,MAAM,QAAQ,GAA0B,EAAE,CAAC;QAC3C,IAAI,CAAC,SAAS,GAAG,EAAE,CAAC;QACpB,IAAI,CAAC,eAAe,GAAG,CAAC,CAAC;QAEzB,IAAI,CAAC,IAAI,CAAC,QAAQ;YAAE,OAAO,QAAQ,CAAC;QAEpC,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,CAAC,OAAO,EAAE,EAAE;YACjC,sEAAsE;YACtE,MAAM,aAAa,GAAG,OAAO,CAAC,aAAa,CAAC,wBAAwB,CAAC,CAAC;YACtE,MAAM,UAAU,GAAG,aAAa,EAAE,WAAW;gBAC5C,EAAE,IAAI,EAAE;gBACR,EAAE,OAAO,CAAC,OAAO,EAAE,EAAE,CAAC,CAAC,2CAA2C;mBAC/D,EAAE,CAAC;YAEP,IAAI,iBAAiB,GAAG,EAAE,CAAC;YAE3B,MAAM,UAAU,GAAG,OAAO,CAAC,YAAY,CAAC,0BAA0B,CAAC,CAAC;YACpE,IAAI,UAAU,EAAE,CAAC;gBAChB,iBAAiB,GAAG,UAAU,CAAC;YAChC,CAAC;YAED,IAAI,cAAc,GAAG,IAAA,mBAAa,EAAC,OAAO,CAAC,CAAC;YAC5C,cAAc,GAAG,cAAc,CAAC,OAAO,CAAC,SAAS,EAAE,EAAE,CAAC,CAAC;YAEvD,oDAAoD;YACpD,MAAM,OAAO,GAAG,IAAI,CAAC,QAAQ,CAAC,aAAa,CAAC,KAAK,CAAC,CAAC;YACnD,OAAO,CAAC,WAAW,CAAC,IAAA,eAAS,EAAC,IAAI,CAAC,QAAQ,EAAE,cAAc,CAAC,CAAC,CAAC;YAC9D,OAAO,CAAC,gBAAgB,CAAC,mDAAmD,CAAC,CAAC,OAAO,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,MAAM,EAAE,CAAC,CAAC;YACzG,cAAc,GAAG,IAAA,mBAAa,EAAC,OAAO,CAAC,CAAC;YAExC,+DAA+D;YAC/D,wGAAwG;YACxG,iDAAiD;YACjD,MAAM,eAAe,GAAG,kJAAkJ,CAAC;YAE3K,cAAc,GAAG,cAAc,CAAC,OAAO,CAAC,eAAe,EAAE,CAAC,KAAK,EAAE,GAAG,EAAE,aAAa,EAAE,GAAG,EAAE,EAAE;gBAC3F,0BAA0B;gBAC1B,IAAI,MAAM,GAAG,EAAE,CAAC;gBAChB,IAAI,YAAY,GAAG,EAAE,CAAC;gBAEtB,IAAI,CAAC;oBACJ,8BAA8B;oBAC9B,MAAM,GAAG,IAAI,GAAG,CAAC,GAAG,CAAC,CAAC,QAAQ,CAAC,OAAO,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAC;oBAErD,oDAAoD;oBACpD,MAAM,SAAS,GAAG,GAAG,CAAC,KAAK,CAAC,WAAW,CAAC,CAAC;oBACzC,IAAI,SAAS,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;wBAC1B,YAAY,GAAG,kBAAkB,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC;wBAChD,YAAY,GAAG,YAAY,CAAC,OAAO,CAAC,MAAM,EAAE,GAAG,CAAC,CAAC;wBAEjD,MAAM,KAAK,GAAG,YAAY,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;wBACtC,IAAI,KAAK,CAAC,MAAM,GAAG,CAAC,IAAI,KAAK,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC;4BACzC,YAAY,GAAG,MAAM,KAAK,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,KAAK,CAAC;wBAC3C,CAAC;6BAAM,IAAI,KAAK,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC;4BAC5B,YAAY,GAAG,MAAM,YAAY,CAAC,IAAI,EAAE,EAAE,CAAC;wBAC5C,CAAC;6BAAM,CAAC;4BACP,YAAY,GAAG,EAAE,CAAC;wBACnB,CAAC;oBACF,CAAC;gBACF,CAAC;gBAAC,OAAO,CAAC,EAAE,CAAC;oBACZ,OAAO,CAAC,KAAK,CAAC,wBAAwB,GAAG,EAAE,EAAE,CAAC,CAAC,CAAC;oBAChD,MAAM,GAAG,GAAG,CAAC;gBACd,CAAC;gBAED,oDAAoD;gBACpD,IAAI,aAAa,GAAG,IAAI,CAAC,SAAS,CAAC,SAAS,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,GAAG,KAAK,GAAG,CAAC,CAAC;gBACnE,IAAI,cAAsB,CAAC;gBAE3B,IAAI,aAAa,KAAK,CAAC,CAAC,EAAE,CAAC;oBAC1B,IAAI,CAAC,eAAe,EAAE,CAAC;oBACvB,cAAc,GAAG,IAAI,CAAC,eAAe,CAAC;oBACtC,IAAI,CAAC,SAAS,CAAC,IAAI,CAAC;wBACnB,GAAG;wBACH,IAAI,EAAE,YAAY,GAAG,KAAK,MAAM,OAAO,YAAY,EAAE;qBACrD,CAAC,CAAC;gBACJ,CAAC;qBAAM,CAAC;oBACP,cAAc,GAAG,aAAa,GAAG,CAAC,CAAC;gBACpC,CAAC;gBAED,oGAAoG;gBACpG,OAAO,kBAAkB,cAAc,kBAAkB,cAAc,KAAK,cAAc,YAAY,CAAC;YACxG,CAAC,CAAC,CAAC;YAEH,0CAA0C;YAC1C,cAAc,GAAG,cAAc;iBAC7B,OAAO,CAAC,mBAAmB,EAAE,EAAE,CAAC,CAAC;YAEnC,QAAQ,CAAC,IAAI,CAAC;gBACb,MAAM,EAAE,UAAU;gBAClB,OAAO,EAAE,cAAc,CAAC,IAAI,EAAE;gBAC9B,QAAQ,EAAE;oBACT,IAAI,EAAE,iBAAiB,IAAI,SAAS;iBACpC;aACD,CAAC,CAAC;QAEJ,CAAC,CAAC,CAAC;QAEH,IAAI,CAAC,cAAc,GAAG,QAAQ,CAAC;QAC/B,OAAO,QAAQ,CAAC;IACjB,CAAC;IAES,YAAY;QACrB,OAAO,IAAI,CAAC,SAAS,CAAC;IACvB,CAAC;IAES,WAAW;QACpB,MAAM,KAAK,GAAG,IAAI,CAAC,QAAQ,EAAE,CAAC;QAC9B,MAAM,QAAQ,GAAG,IAAI,CAAC,eAAe,EAAE,CAAC;QAExC,OAAO;YACN,KAAK;YACL,IAAI,EAAE,SAAS;YACf,GAAG,EAAE,IAAI,CAAC,GAAG;YACb,YAAY,EAAE,QAAQ,CAAC,MAAM;YAC7B,WAAW,EAAE,6BAA6B,QAAQ,CAAC,MAAM,WAAW;SACpE,CAAC;IACH,CAAC;IAEO,QAAQ;QACf,kCAAkC;QAClC,MAAM,SAAS,GAAG,IAAI,CAAC,QAAQ,CAAC,KAAK,EAAE,IAAI,EAAE,CAAC;QAC9C,IAAI,SAAS,IAAI,SAAS,KAAK,SAAS,EAAE,CAAC;YAC1C,OAAO,SAAS,CAAC;QAClB,CAAC;QAED,kCAAkC;QAClC,MAAM,aAAa,GAAG,IAAI,CAAC,QAAQ,EAAE,IAAI,CAAC,CAAC,CAAC,EAAE,aAAa,CAAC,eAAe,CAAC,CAAC;QAC7E,IAAI,aAAa,EAAE,CAAC;YACnB,MAAM,IAAI,GAAG,aAAa,CAAC,WAAW,IAAI,EAAE,CAAC;YAC7C,4CAA4C;YAC5C,OAAO,IAAI,CAAC,MAAM,GAAG,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC,GAAG,KAAK,CAAC,CAAC,CAAC,IAAI,CAAC;QAC5D,CAAC;QAED,OAAO,sBAAsB,CAAC;IAC/B,CAAC;CACD;AA3JD,4CA2JC"}

package/dist/extractors/claude.d.ts

@@ -0,0 +1,10 @@
+import { ConversationExtractor } from './_conversation';
+import { ConversationMessage, ConversationMetadata } from '../types/extractors';
+export declare class ClaudeExtractor extends ConversationExtractor {
+    private articles;
+    constructor(document: Document, url: string);
+    canExtract(): boolean;
+    protected extractMessages(): ConversationMessage[];
+    protected getMetadata(): ConversationMetadata;
+    private getTitle;
+}

package/dist/extractors/claude.js

@@ -0,0 +1,91 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.ClaudeExtractor = void 0;
+const _conversation_1 = require("./_conversation");
+const dom_1 = require("../utils/dom");
+class ClaudeExtractor extends _conversation_1.ConversationExtractor {
+    constructor(document, url) {
+        super(document, url);
+        // Find all message blocks - both user and assistant messages
+        this.articles = document.querySelectorAll('div[data-testid="user-message"], div[data-testid="assistant-message"], div.font-claude-response');
+    }
+    canExtract() {
+        return !!this.articles && this.articles.length > 0;
+    }
+    extractMessages() {
+        const messages = [];
+        if (!this.articles)
+            return messages;
+        this.articles.forEach((article) => {
+            let role;
+            let content;
+            if (article.hasAttribute('data-testid')) {
+                // Handle user messages
+                if (article.getAttribute('data-testid') === 'user-message') {
+                    role = 'you';
+                    content = (0, dom_1.serializeHTML)(article);
+                }
+                // Skip non-message elements
+                else {
+                    return;
+                }
+            }
+            else if (article.classList.contains('font-claude-response')) {
+                // Handle Claude messages
+                role = 'assistant';
+                const assistantBody = article.querySelector('.standard-markdown') || article;
+                content = (0, dom_1.serializeHTML)(assistantBody);
+            }
+            else {
+                // Skip unknown elements
+                return;
+            }
+            if (content) {
+                // Normalize content similar to ChatGPT extractor
+                content = content.replace(/\u200B/g, '').replace(/<p[^>]*>\s*<\/p>/g, '');
+                messages.push({
+                    author: role === 'you' ? 'You' : 'Claude',
+                    content: content.trim(),
+                    metadata: {
+                        role: role
+                    }
+                });
+            }
+        });
+        return messages;
+    }
+    getMetadata() {
+        const title = this.getTitle();
+        const messages = this.extractMessages();
+        return {
+            title,
+            site: 'Claude',
+            url: this.url,
+            messageCount: messages.length,
+            description: `Claude conversation with ${messages.length} messages`
+        };
+    }
+    getTitle() {
+        // Try to get the page title first
+        const pageTitle = this.document.title?.trim();
+        if (pageTitle && pageTitle !== 'Claude') {
+            // Remove ' - Claude' suffix if present
+            return pageTitle.replace(/ - Claude$/, '');
+        }
+        // Try to get title from header
+        const headerTitle = this.document.querySelector('header .font-tiempos')?.textContent?.trim();
+        if (headerTitle) {
+            return headerTitle;
+        }
+        // Fall back to first user message
+        const firstUserMessage = this.articles?.item(0)?.querySelector('[data-testid="user-message"]');
+        if (firstUserMessage) {
+            const text = firstUserMessage.textContent || '';
+            // Truncate to first 50 characters if longer
+            return text.length > 50 ? text.slice(0, 50) + '...' : text;
+        }
+        return 'Claude Conversation';
+    }
+}
+exports.ClaudeExtractor = ClaudeExtractor;
+//# sourceMappingURL=claude.js.map

package/dist/extractors/claude.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"claude.js","sourceRoot":"","sources":["../../src/extractors/claude.ts"],"names":[],"mappings":";;;AAAA,mDAAwD;AAExD,sCAA6C;AAE7C,MAAa,eAAgB,SAAQ,qCAAqB;IAGzD,YAAY,QAAkB,EAAE,GAAW;QAC1C,KAAK,CAAC,QAAQ,EAAE,GAAG,CAAC,CAAC;QACrB,6DAA6D;QAC7D,IAAI,CAAC,QAAQ,GAAG,QAAQ,CAAC,gBAAgB,CAAC,iGAAiG,CAAC,CAAC;IAC9I,CAAC;IAED,UAAU;QACT,OAAO,CAAC,CAAC,IAAI,CAAC,QAAQ,IAAI,IAAI,CAAC,QAAQ,CAAC,MAAM,GAAG,CAAC,CAAC;IACpD,CAAC;IAES,eAAe;QACxB,MAAM,QAAQ,GAA0B,EAAE,CAAC;QAE3C,IAAI,CAAC,IAAI,CAAC,QAAQ;YAAE,OAAO,QAAQ,CAAC;QAEpC,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,CAAC,OAAO,EAAE,EAAE;YACjC,IAAI,IAAY,CAAC;YACjB,IAAI,OAAe,CAAC;YAEpB,IAAI,OAAO,CAAC,YAAY,CAAC,aAAa,CAAC,EAAE,CAAC;gBACzC,uBAAuB;gBACvB,IAAI,OAAO,CAAC,YAAY,CAAC,aAAa,CAAC,KAAK,cAAc,EAAE,CAAC;oBAC5D,IAAI,GAAG,KAAK,CAAC;oBACb,OAAO,GAAG,IAAA,mBAAa,EAAC,OAAO,CAAC,CAAC;gBAClC,CAAC;gBACD,4BAA4B;qBACvB,CAAC;oBACL,OAAO;gBACR,CAAC;YACF,CAAC;iBAAM,IAAI,OAAO,CAAC,SAAS,CAAC,QAAQ,CAAC,sBAAsB,CAAC,EAAE,CAAC;gBAC/D,yBAAyB;gBACzB,IAAI,GAAG,WAAW,CAAC;gBACnB,MAAM,aAAa,GAAI,OAAO,CAAC,aAAa,CAAC,oBAAoB,CAAiB,IAAK,OAAuB,CAAC;gBAC/G,OAAO,GAAG,IAAA,mBAAa,EAAC,aAAa,CAAC,CAAC;YACxC,CAAC;iBAAM,CAAC;gBACP,wBAAwB;gBACxB,OAAO;YACR,CAAC;YAED,IAAI,OAAO,EAAE,CAAC;gBACb,iDAAiD;gBACjD,OAAO,GAAG,OAAO,CAAC,OAAO,CAAC,SAAS,EAAE,EAAE,CAAC,CAAC,OAAO,CAAC,mBAAmB,EAAE,EAAE,CAAC,CAAC;gBAC1E,QAAQ,CAAC,IAAI,CAAC;oBACb,MAAM,EAAE,IAAI,KAAK,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,QAAQ;oBACzC,OAAO,EAAE,OAAO,CAAC,IAAI,EAAE;oBACvB,QAAQ,EAAE;wBACT,IAAI,EAAE,IAAI;qBACV;iBACD,CAAC,CAAC;YACJ,CAAC;QACF,CAAC,CAAC,CAAC;QAEH,OAAO,QAAQ,CAAC;IACjB,CAAC;IAES,WAAW;QACpB,MAAM,KAAK,GAAG,IAAI,CAAC,QAAQ,EAAE,CAAC;QAC9B,MAAM,QAAQ,GAAG,IAAI,CAAC,eAAe,EAAE,CAAC;QAExC,OAAO;YACN,KAAK;YACL,IAAI,EAAE,QAAQ;YACd,GAAG,EAAE,IAAI,CAAC,GAAG;YACb,YAAY,EAAE,QAAQ,CAAC,MAAM;YAC7B,WAAW,EAAE,4BAA4B,QAAQ,CAAC,MAAM,WAAW;SACnE,CAAC;IACH,CAAC;IAEO,QAAQ;QACf,kCAAkC;QAClC,MAAM,SAAS,GAAG,IAAI,CAAC,QAAQ,CAAC,KAAK,EAAE,IAAI,EAAE,CAAC;QAC9C,IAAI,SAAS,IAAI,SAAS,KAAK,QAAQ,EAAE,CAAC;YACzC,uCAAuC;YACvC,OAAO,SAAS,CAAC,OAAO,CAAC,YAAY,EAAE,EAAE,CAAC,CAAC;QAC5C,CAAC;QAED,+BAA+B;QAC/B,MAAM,WAAW,GAAG,IAAI,CAAC,QAAQ,CAAC,aAAa,CAAC,sBAAsB,CAAC,EAAE,WAAW,EAAE,IAAI,EAAE,CAAC;QAC7F,IAAI,WAAW,EAAE,CAAC;YACjB,OAAO,WAAW,CAAC;QACpB,CAAC;QAED,kCAAkC;QAClC,MAAM,gBAAgB,GAAG,IAAI,CAAC,QAAQ,EAAE,IAAI,CAAC,CAAC,CAAC,EAAE,aAAa,CAAC,8BAA8B,CAAC,CAAC;QAC/F,IAAI,gBAAgB,EAAE,CAAC;YACtB,MAAM,IAAI,GAAG,gBAAgB,CAAC,WAAW,IAAI,EAAE,CAAC;YAChD,4CAA4C;YAC5C,OAAO,IAAI,CAAC,MAAM,GAAG,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC,GAAG,KAAK,CAAC,CAAC,CAAC,IAAI,CAAC;QAC5D,CAAC;QAED,OAAO,qBAAqB,CAAC;IAC9B,CAAC;CACD;AA/FD,0CA+FC"}

package/dist/extractors/gemini.d.ts

@@ -0,0 +1,14 @@
+import { ConversationExtractor } from './_conversation';
+import { ConversationMessage, ConversationMetadata, Footnote } from '../types/extractors';
+export declare class GeminiExtractor extends ConversationExtractor {
+    private conversationContainers;
+    private footnotes;
+    private messageCount;
+    constructor(document: Document, url: string);
+    canExtract(): boolean;
+    protected extractMessages(): ConversationMessage[];
+    private extractSources;
+    protected getFootnotes(): Footnote[];
+    protected getMetadata(): ConversationMetadata;
+    private getTitle;
+}

package/dist/extractors/gemini.js

@@ -0,0 +1,111 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.GeminiExtractor = void 0;
+const _conversation_1 = require("./_conversation");
+const dom_1 = require("../utils/dom");
+class GeminiExtractor extends _conversation_1.ConversationExtractor {
+    constructor(document, url) {
+        super(document, url);
+        this.messageCount = null;
+        this.conversationContainers = document.querySelectorAll('div.conversation-container');
+        this.footnotes = [];
+    }
+    canExtract() {
+        return !!this.conversationContainers && this.conversationContainers.length > 0;
+    }
+    extractMessages() {
+        this.messageCount = 0;
+        const messages = [];
+        if (!this.conversationContainers)
+            return messages;
+        this.extractSources();
+        this.conversationContainers.forEach((container) => {
+            const userQuery = container.querySelector('user-query');
+            if (userQuery) {
+                const queryText = userQuery.querySelector('.query-text');
+                if (queryText) {
+                    const content = (0, dom_1.serializeHTML)(queryText);
+                    messages.push({
+                        author: 'You',
+                        content: content.trim(),
+                        metadata: { role: 'user' }
+                    });
+                }
+            }
+            const modelResponse = container.querySelector('model-response');
+            if (modelResponse) {
+                const regularContent = modelResponse.querySelector('.model-response-text .markdown');
+                const extendedContent = modelResponse.querySelector('#extended-response-markdown-content');
+                const contentElement = extendedContent || regularContent;
+                if (contentElement) {
+                    let content = (0, dom_1.serializeHTML)(contentElement);
+                    const tempDiv = this.document.createElement('div');
+                    tempDiv.appendChild((0, dom_1.parseHTML)(this.document, content));
+                    tempDiv.querySelectorAll('.table-content').forEach(el => {
+                        // `table-content` is a PARTIAL selector in defuddle (table of contents, will be removed), but a real table in Gemini (should be kept).
+                        el.classList.remove('table-content');
+                    });
+                    content = (0, dom_1.serializeHTML)(tempDiv);
+                    messages.push({
+                        author: 'Gemini',
+                        content: content.trim(),
+                        metadata: { role: 'assistant' }
+                    });
+                }
+            }
+        });
+        this.messageCount = messages.length;
+        return messages;
+    }
+    extractSources() {
+        const browseItems = this.document.querySelectorAll('browse-item');
+        if (browseItems && browseItems.length > 0) {
+            browseItems.forEach(item => {
+                const link = item.querySelector('a');
+                if (link instanceof HTMLAnchorElement) {
+                    const url = link.href;
+                    const domain = link.querySelector('.domain')?.textContent?.trim() || '';
+                    const title = link.querySelector('.title')?.textContent?.trim() || '';
+                    if (url && (domain || title)) {
+                        this.footnotes.push({
+                            url,
+                            text: title ? `${domain}: ${title}` : domain
+                        });
+                    }
+                }
+            });
+        }
+    }
+    getFootnotes() {
+        return this.footnotes;
+    }
+    getMetadata() {
+        const title = this.getTitle();
+        const messageCount = this.messageCount ?? this.extractMessages().length;
+        return {
+            title,
+            site: 'Gemini',
+            url: this.url,
+            messageCount,
+            description: `Gemini conversation with ${messageCount} messages`
+        };
+    }
+    getTitle() {
+        const pageTitle = this.document.title?.trim();
+        if (pageTitle && pageTitle !== 'Gemini' && !pageTitle.includes('Gemini')) {
+            return pageTitle;
+        }
+        const researchTitle = this.document.querySelector('.title-text')?.textContent?.trim();
+        if (researchTitle) {
+            return researchTitle;
+        }
+        const firstUserQuery = this.conversationContainers?.item(0)?.querySelector('.query-text');
+        if (firstUserQuery) {
+            const text = firstUserQuery.textContent || '';
+            return text.length > 50 ? text.slice(0, 50) + '...' : text;
+        }
+        return 'Gemini Conversation';
+    }
+}
+exports.GeminiExtractor = GeminiExtractor;
+//# sourceMappingURL=gemini.js.map

package/dist/extractors/gemini.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"gemini.js","sourceRoot":"","sources":["../../src/extractors/gemini.ts"],"names":[],"mappings":";;;AAAA,mDAAwD;AAExD,sCAAwD;AAExD,MAAa,eAAgB,SAAQ,qCAAqB;IAKzD,YAAY,QAAkB,EAAE,GAAW;QAC1C,KAAK,CAAC,QAAQ,EAAE,GAAG,CAAC,CAAC;QAHd,iBAAY,GAAkB,IAAI,CAAC;QAI1C,IAAI,CAAC,sBAAsB,GAAG,QAAQ,CAAC,gBAAgB,CAAC,4BAA4B,CAAC,CAAC;QACtF,IAAI,CAAC,SAAS,GAAG,EAAE,CAAC;IACrB,CAAC;IAED,UAAU;QACT,OAAO,CAAC,CAAC,IAAI,CAAC,sBAAsB,IAAI,IAAI,CAAC,sBAAsB,CAAC,MAAM,GAAG,CAAC,CAAC;IAChF,CAAC;IAES,eAAe;QACxB,IAAI,CAAC,YAAY,GAAG,CAAC,CAAC;QACtB,MAAM,QAAQ,GAA0B,EAAE,CAAC;QAE3C,IAAI,CAAC,IAAI,CAAC,sBAAsB;YAAE,OAAO,QAAQ,CAAC;QAElD,IAAI,CAAC,cAAc,EAAE,CAAC;QAEtB,IAAI,CAAC,sBAAsB,CAAC,OAAO,CAAC,CAAC,SAAS,EAAE,EAAE;YACjD,MAAM,SAAS,GAAG,SAAS,CAAC,aAAa,CAAC,YAAY,CAAC,CAAC;YACxD,IAAI,SAAS,EAAE,CAAC;gBACf,MAAM,SAAS,GAAG,SAAS,CAAC,aAAa,CAAC,aAAa,CAAC,CAAC;gBACzD,IAAI,SAAS,EAAE,CAAC;oBACf,MAAM,OAAO,GAAG,IAAA,mBAAa,EAAC,SAAS,CAAC,CAAC;oBACzC,QAAQ,CAAC,IAAI,CAAC;wBACb,MAAM,EAAE,KAAK;wBACb,OAAO,EAAE,OAAO,CAAC,IAAI,EAAE;wBACvB,QAAQ,EAAE,EAAE,IAAI,EAAE,MAAM,EAAE;qBAC1B,CAAC,CAAC;gBACJ,CAAC;YACF,CAAC;YAED,MAAM,aAAa,GAAG,SAAS,CAAC,aAAa,CAAC,gBAAgB,CAAC,CAAC;YAChE,IAAI,aAAa,EAAE,CAAC;gBACnB,MAAM,cAAc,GAAG,aAAa,CAAC,aAAa,CAAC,gCAAgC,CAAC,CAAC;gBACrF,MAAM,eAAe,GAAG,aAAa,CAAC,aAAa,CAAC,qCAAqC,CAAC,CAAC;gBAC3F,MAAM,cAAc,GAAG,eAAe,IAAI,cAAc,CAAC;gBAEzD,IAAI,cAAc,EAAE,CAAC;oBACpB,IAAI,OAAO,GAAG,IAAA,mBAAa,EAAC,cAAc,CAAC,CAAC;oBAE5C,MAAM,OAAO,GAAG,IAAI,CAAC,QAAQ,CAAC,aAAa,CAAC,KAAK,CAAC,CAAC;oBACnD,OAAO,CAAC,WAAW,CAAC,IAAA,eAAS,EAAC,IAAI,CAAC,QAAQ,EAAE,OAAO,CAAC,CAAC,CAAC;oBAEvD,OAAO,CAAC,gBAAgB,CAAC,gBAAgB,CAAC,CAAC,OAAO,CAAC,EAAE,CAAC,EAAE;wBACvD,uIAAuI;wBACvI,EAAE,CAAC,SAAS,CAAC,MAAM,CAAC,eAAe,CAAC,CAAC;oBACtC,CAAC,CAAC,CAAC;oBAEH,OAAO,GAAG,IAAA,mBAAa,EAAC,OAAO,CAAC,CAAC;oBAEjC,QAAQ,CAAC,IAAI,CAAC;wBACb,MAAM,EAAE,QAAQ;wBAChB,OAAO,EAAE,OAAO,CAAC,IAAI,EAAE;wBACvB,QAAQ,EAAE,EAAE,IAAI,EAAE,WAAW,EAAE;qBAC/B,CAAC,CAAC;gBACJ,CAAC;YACF,CAAC;QACF,CAAC,CAAC,CAAC;QACH,IAAI,CAAC,YAAY,GAAG,QAAQ,CAAC,MAAM,CAAC;QACpC,OAAO,QAAQ,CAAC;IACjB,CAAC;IAEO,cAAc;QACrB,MAAM,WAAW,GAAG,IAAI,CAAC,QAAQ,CAAC,gBAAgB,CAAC,aAAa,CAAC,CAAC;QAElE,IAAI,WAAW,IAAI,WAAW,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YAC3C,WAAW,CAAC,OAAO,CAAC,IAAI,CAAC,EAAE;gBAC1B,MAAM,IAAI,GAAG,IAAI,CAAC,aAAa,CAAC,GAAG,CAAC,CAAC;gBACrC,IAAI,IAAI,YAAY,iBAAiB,EAAE,CAAC;oBACvC,MAAM,GAAG,GAAG,IAAI,CAAC,IAAI,CAAC;oBACtB,MAAM,MAAM,GAAG,IAAI,CAAC,aAAa,CAAC,SAAS,CAAC,EAAE,WAAW,EAAE,IAAI,EAAE,IAAI,EAAE,CAAC;oBACxE,MAAM,KAAK,GAAG,IAAI,CAAC,aAAa,CAAC,QAAQ,CAAC,EAAE,WAAW,EAAE,IAAI,EAAE,IAAI,EAAE,CAAC;oBAEtE,IAAI,GAAG,IAAI,CAAC,MAAM,IAAI,KAAK,CAAC,EAAE,CAAC;wBAC9B,IAAI,CAAC,SAAS,CAAC,IAAI,CAAC;4BACnB,GAAG;4BACH,IAAI,EAAE,KAAK,CAAC,CAAC,CAAC,GAAG,MAAM,KAAK,KAAK,EAAE,CAAC,CAAC,CAAC,MAAM;yBAC5C,CAAC,CAAC;oBACJ,CAAC;gBACF,CAAC;YACF,CAAC,CAAC,CAAC;QACJ,CAAC;IACF,CAAC;IAES,YAAY;QACrB,OAAO,IAAI,CAAC,SAAS,CAAC;IACvB,CAAC;IAES,WAAW;QACpB,MAAM,KAAK,GAAG,IAAI,CAAC,QAAQ,EAAE,CAAC;QAC9B,MAAM,YAAY,GAAG,IAAI,CAAC,YAAY,IAAI,IAAI,CAAC,eAAe,EAAE,CAAC,MAAM,CAAC;QACxE,OAAO;YACN,KAAK;YACL,IAAI,EAAE,QAAQ;YACd,GAAG,EAAE,IAAI,CAAC,GAAG;YACb,YAAY;YACZ,WAAW,EAAE,4BAA4B,YAAY,WAAW;SAChE,CAAC;IACH,CAAC;IAEO,QAAQ;QACf,MAAM,SAAS,GAAG,IAAI,CAAC,QAAQ,CAAC,KAAK,EAAE,IAAI,EAAE,CAAC;QAC9C,IAAI,SAAS,IAAI,SAAS,KAAK,QAAQ,IAAI,CAAC,SAAS,CAAC,QAAQ,CAAC,QAAQ,CAAC,EAAE,CAAC;YAC1E,OAAO,SAAS,CAAC;QAClB,CAAC;QAED,MAAM,aAAa,GAAG,IAAI,CAAC,QAAQ,CAAC,aAAa,CAAC,aAAa,CAAC,EAAE,WAAW,EAAE,IAAI,EAAE,CAAC;QACtF,IAAI,aAAa,EAAE,CAAC;YACnB,OAAO,aAAa,CAAC;QACtB,CAAC;QAED,MAAM,cAAc,GAAG,IAAI,CAAC,sBAAsB,EAAE,IAAI,CAAC,CAAC,CAAC,EAAE,aAAa,CAAC,aAAa,CAAC,CAAC;QAC1F,IAAI,cAAc,EAAE,CAAC;YACpB,MAAM,IAAI,GAAG,cAAc,CAAC,WAAW,IAAI,EAAE,CAAC;YAC9C,OAAO,IAAI,CAAC,MAAM,GAAG,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC,GAAG,KAAK,CAAC,CAAC,CAAC,IAAI,CAAC;QAC5D,CAAC;QAED,OAAO,qBAAqB,CAAC;IAC9B,CAAC;CACD;AA7HD,0CA6HC"}

package/dist/extractors/github.d.ts

@@ -0,0 +1,20 @@
+import { BaseExtractor } from './_base';
+import { ExtractorResult } from '../types/extractors';
+export declare class GitHubExtractor extends BaseExtractor {
+    private isIssue;
+    private isPR;
+    constructor(document: Document, url: string);
+    canExtract(): boolean;
+    extract(): ExtractorResult;
+    private createContentHtml;
+    private getIssueContent;
+    private extractComments;
+    private getPRBody;
+    private getPRContent;
+    private extractPRComments;
+    private extractAuthor;
+    private cleanBodyContent;
+    private extractNumber;
+    private extractRepoInfo;
+    private createDescription;
+}