[{"_1":2,"_172":-5,"_173":-5},"loaderData",{"_3":4,"_91":92},"root",{"_5":6,"_34":35,"_64":65},"ctx",{"_7":8,"_23":-5,"_24":25,"_26":27,"_28":20,"_29":20,"_30":20,"_31":32,"_33":20},"auth",{"_9":-5,"_10":-5,"_11":-5,"_12":-5,"_13":-5,"_14":-5,"_15":-5,"_16":-5,"_17":-5,"_18":-5,"_19":20,"_21":22},"sessionClaims","sessionId","sessionStatus","userId","actor","orgId","orgRole","orgSlug","orgPermissions","factorVerificationAge","isAuthenticated",false,"tokenType","session_token","user","preferencesKeybindings","default","preferencesTheme","light","preferencesFixers","preferencesCopilot","preferencesSelectionAi","notificationCount",0,"flagDenols","ENV",{"_36":37,"_38":39,"_40":41,"_42":43,"_44":45,"_46":47,"_48":49,"_50":51,"_52":53,"_54":55,"_56":57,"_58":59,"_60":61,"_62":63},"REACT_APP_SENTRY_DSN","https://586020d86f05489db63a554690acf1e9@o1398359.ingest.sentry.io/6724499","VAL_RUNTIME_NAME","prod","RENDER_GIT_COMMIT","e3ca0a77ab07ff2ce3097208ffd6aa0e378d3c47","RENDER_GIT_BRANCH","main","CLERK_PUBLISHABLE_KEY","pk_live_Y2xlcmsudmFsLnRvd24k","POSTHOG_API_TOKEN","phc_nmoyuAkStEZuUp69rYlg1DZgrIt19Gx0Y4Hwf8tGgJP","POSTHOG_API_HOST","https://wart.val.town","WEB_API_TEMPLATE","https://{val}.web.val.run","APP_URL","https://www.val.town","MODULE_URL","https://esm.town","VANITY_WEB_API_TEMPLATE","https://{label}.val.run/","DENOLS_URL","https://lsp.val.town","NODE_ENV","production","API_URL","https://api.val.town","clerkState",{"_66":67},"__internal_clerk_state",{"_68":69,"_70":45,"_71":72,"_73":72,"_74":20,"_75":72,"_76":72,"_77":72,"_78":72,"_79":80,"_89":20,"_90":20},"__clerk_ssr_state",{"_21":22,"_9":-5,"_10":-5,"_11":-5,"_12":-5,"_13":-5,"_14":-5,"_15":-5,"_16":-5,"_17":-5,"_18":-5,"_19":20},"__publishableKey","__proxyUrl","","__domain","__isSatellite","__signInUrl","__signUpUrl","__afterSignInUrl","__afterSignUpUrl","__clerk_debug",{"_81":20,"_19":20,"_82":72,"_83":84,"_85":72,"_86":45,"_87":20,"_88":72},"isSignedIn","proxyUrl","reason","session-token-and-uat-missing","message","publishableKey","isSatellite","domain","__telemetryDisabled","__telemetryDebug","routes/embed.x.$handle.$name.$",{"_93":94,"_95":96,"_97":98,"_99":100,"_147":148,"_169":170,"_171":-5},"handle","jpd3v","name","offx","path","backend/twitter-summarization.ts","project",{"_101":102,"_95":96,"_103":104,"_105":-5,"_106":20,"_107":108,"_109":72,"_110":-5,"_12":111,"_112":-5,"_113":20,"_114":-5,"_115":116,"_14":117,"_118":119,"_120":119,"_121":122,"_123":122,"_124":122,"_125":119,"_126":119,"_23":127,"_130":131,"_133":134,"_146":20},"id","a93616c0-55f2-11f0-8cdc-f69ea79377d9","createdAt",["D",1751316115780],"deletedAt","legacyEnvVars","privacy","public","description","imageUrl","da7efaa0-3a43-11f0-b284-7173a8a6f515","pinnedFileId","pinPreview","screenshotVersion","likeCount",1,"f1d76965-949f-4699-906d-07e921194179","userPerm","r","valPerm","blobPerm","rw","emailPerm","sqlitePerm","projectPerm","telemetryPerm",{"_93":94,"_128":129},"profileImageUrl","https://img.clerk.com/eyJ0eXBlIjoicHJveHkiLCJzcmMiOiJodHRwczovL2ltYWdlcy5jbGVyay5kZXYvb2F1dGhfZ2l0aHViL2ltZ18yeGRlTk9ERTV0VzdNb2FqR0NxOGRuM0NlaTEifQ","org",{"_93":94,"_132":129},"avatar","branch",{"_135":136,"_95":43,"_101":137,"_138":102,"_139":-5,"_140":-5,"_141":-5,"_142":-5,"_143":-5,"_144":96,"_145":94},"version",98,"a93670de-55f2-11f0-8cdc-f69ea79377d9","projectId","forkedBranchId","forkedBranchVersion","forkedProjectPrivacy","forkedProjectId","forkedProjectOrgId","projectName","userHandle","isOwned","file",{"_101":149,"_95":150,"_151":152,"_153":154,"_155":137,"_156":157,"_158":159,"_103":160,"_135":116,"_161":162,"_163":164,"_165":20,"_166":-5,"_97":167},"a5decd36-8a7d-11f0-aa86-0224a6c84d84","twitter-summarization.ts","fileId","845e360c-8a7c-11f0-9d13-0224a6c84d84","parentId","266ad892-55f3-11f0-9719-f69ea79377d9","branchId","branchVersion",76,"latest",true,["D",1757093270676],"fileType","script","content","import { OpenAI } from \"https://esm.town/v/std/openai\";\nimport {\n createSummarization,\n getAllTopics,\n getPostsForTopic,\n getSummarizationsForTopic,\n} from \"./database/queries.ts\";\nimport type { Post, Topic } from \"./database/queries.ts\";\nimport { TwitterSentimentAnalyzer } from \"./twitter-sentiment.ts\";\n\n/**\n * OpenAI-powered summarization service for grouped tweets\n * Extends the existing sentiment analysis to provide topic-based summaries\n */\n\nexport interface TopicSummary {\n topicId: number;\n topicTitle: string;\n summary: string;\n keyPoints: string[];\n tweetCount: number;\n timeWindow: string;\n overallSentiment: {\n sentiment: \"positive\" | \"negative\" | \"neutral\";\n confidence: number;\n score: number;\n };\n trendingKeywords: string[];\n influentialAuthors: string[];\n}\n\nexport interface SummarizationRequest {\n topicId?: number;\n timeWindowHours?: number;\n maxTweets?: number;\n includeMedia?: boolean;\n includeSentiment?: boolean;\n}\n\nexport interface SummarizationResult {\n success: boolean;\n generated: number;\n stored: number;\n errors: string[];\n summaries: TopicSummary[];\n}\n\n/**\n * Service class for generating AI-powered summaries of tweets grouped by topics\n */\nexport class TwitterSummarizationService {\n private sentimentAnalyzer: TwitterSentimentAnalyzer;\n\n constructor() {\n // Initialize sentiment analyzer with default config\n this.sentimentAnalyzer = new TwitterSentimentAnalyzer();\n }\n\n /**\n * Generate a comprehensive summary for a specific topic\n */\n async generateTopicSummary(\n topicId: number,\n options: Omit = {},\n ): Promise {\n try {\n const {\n timeWindowHours = 24,\n maxTweets = 50,\n includeMedia = false,\n includeSentiment = true,\n } = options;\n\n // Get the topic details\n const topics = await getAllTopics();\n const topic = topics.find((t) => t.id === topicId);\n if (!topic) {\n throw new Error(`Topic with ID ${topicId} not found`);\n }\n\n // Get recent posts for this topic\n const posts = await getPostsForTopic(topicId, maxTweets);\n\n if (posts.length === 0) {\n console.log(`No posts found for topic ${topicId}`);\n return null;\n }\n\n // Filter posts by time window\n const cutoffTime = new Date(\n Date.now() - timeWindowHours * 60 * 60 * 1000,\n );\n const recentPosts = posts.filter((post) =>\n new Date(post.created_at) > cutoffTime\n );\n\n if (recentPosts.length === 0) {\n console.log(\n `No recent posts found for topic ${topicId} in the last ${timeWindowHours} hours`,\n );\n return null;\n }\n\n // Analyze sentiment if requested\n let sentimentData: {\n sentiment: \"positive\" | \"negative\" | \"neutral\";\n confidence: number;\n score: number;\n } | null = null;\n if (includeSentiment) {\n try {\n const sentimentResults = await this.sentimentAnalyzer.analyzeBatch(\n recentPosts,\n );\n sentimentData = this.aggregateSentiment(sentimentResults);\n } catch (error: unknown) {\n console.warn(\n \"Failed to analyze sentiment, continuing without it:\",\n error,\n );\n sentimentData = {\n sentiment: \"neutral\" as const,\n confidence: 0.5,\n score: 0,\n };\n }\n }\n\n // Group and format posts for summarization\n const groupedData = this.groupPostsForSummarization(recentPosts);\n\n // Generate summary using OpenAI\n const summary = await this.generateSummaryWithAI(\n topic,\n groupedData,\n sentimentData,\n );\n\n // Create the final topic summary\n const topicSummary: TopicSummary = {\n topicId,\n topicTitle: topic.title,\n summary: summary.mainSummary,\n keyPoints: summary.keyPoints,\n tweetCount: recentPosts.length,\n timeWindow: `${timeWindowHours} hours`,\n overallSentiment: sentimentData || {\n sentiment: \"neutral\",\n confidence: 0.5,\n score: 0,\n },\n trendingKeywords: summary.trendingKeywords,\n influentialAuthors: summary.influentialAuthors,\n };\n\n return topicSummary;\n } catch (error: unknown) {\n console.error(`Error generating summary for topic ${topicId}:`, error);\n throw error;\n }\n }\n\n /**\n * Generate summaries for all topics with recent activity\n */\n async generateAllTopicSummaries(\n options: Omit = {},\n ): Promise {\n try {\n const topics = await getAllTopics();\n const summaries: TopicSummary[] = [];\n\n console.log(`Generating summaries for ${topics.length} topics...`);\n\n for (const topic of topics) {\n try {\n console.log(`Processing topic: ${topic.title} (ID: ${topic.id})`);\n const summary = await this.generateTopicSummary(topic.id!, options);\n if (summary) {\n summaries.push(summary);\n console.log(`✓ Generated summary for topic: ${topic.title}`);\n } else {\n console.log(`- No recent activity for topic: ${topic.title}`);\n }\n } catch (error: unknown) {\n console.error(\n `Failed to generate summary for topic ${topic.id} (${topic.title}):`,\n error,\n );\n // Continue with other topics\n }\n }\n\n console.log(`Generated ${summaries.length} summaries successfully`);\n return summaries;\n } catch (error: unknown) {\n console.error(\"Error generating all topic summaries:\", error);\n throw error;\n }\n }\n\n /**\n * Store a topic summary in the database\n */\n async storeTopicSummary(topicSummary: TopicSummary): Promise {\n try {\n const summaryContent = JSON.stringify({\n summary: topicSummary.summary,\n keyPoints: topicSummary.keyPoints,\n sentiment: topicSummary.overallSentiment,\n tweetCount: topicSummary.tweetCount,\n timeWindow: topicSummary.timeWindow,\n trendingKeywords: topicSummary.trendingKeywords,\n influentialAuthors: topicSummary.influentialAuthors,\n });\n\n // Parse the time window to extract start and end dates\n const [startDate, endDate] = topicSummary.timeWindow.split(\" to \");\n\n const summarizationId = await createSummarization({\n topic_id: topicSummary.topicId,\n content: summaryContent,\n prompt: \"Topic summary generated automatically\",\n post_count: topicSummary.tweetCount,\n time_period_start: startDate.trim(),\n time_period_end: endDate.trim(),\n });\n\n return summarizationId;\n } catch (error: unknown) {\n console.error(\"Error storing topic summary:\", error);\n return null;\n }\n }\n\n /**\n * Generate and store summaries for all topics\n */\n async generateAndStoreSummaries(\n options: Omit = {},\n ): Promise {\n const errors: string[] = [];\n let generated = 0;\n let stored = 0;\n const summaries: TopicSummary[] = [];\n\n try {\n const generatedSummaries = await this.generateAllTopicSummaries(options);\n generated = generatedSummaries.length;\n summaries.push(...generatedSummaries);\n\n for (const summary of generatedSummaries) {\n try {\n const summaryId = await this.storeTopicSummary(summary);\n if (summaryId) {\n stored++;\n console.log(\n `✓ Stored summary for topic: ${summary.topicTitle} (ID: ${summaryId})`,\n );\n } else {\n const error =\n `Failed to store summary for topic ${summary.topicId} (${summary.topicTitle})`;\n errors.push(error);\n console.error(error);\n }\n } catch (error: unknown) {\n const errorMessage =\n `Error storing summary for topic ${summary.topicId} (${summary.topicTitle}): ${error}`;\n errors.push(errorMessage);\n console.error(errorMessage);\n }\n }\n\n return {\n success: errors.length === 0,\n generated,\n stored,\n errors,\n summaries,\n };\n } catch (error: unknown) {\n const errorMessage = `Error in generateAndStoreSummaries: ${error}`;\n errors.push(errorMessage);\n console.error(errorMessage);\n return {\n success: false,\n generated,\n stored,\n errors,\n summaries,\n };\n }\n }\n\n /**\n * Group posts for effective summarization\n */\n private groupPostsForSummarization(posts: Post[]): {\n authorCounts: Record;\n contentSample: string[];\n mediaCount: number;\n timeSpread: { earliest: string; latest: string };\n } {\n const authorCounts: Record = {};\n const contentSample: string[] = [];\n let mediaCount = 0;\n\n // Count authors and collect content\n for (const post of posts) {\n authorCounts[post.author] = (authorCounts[post.author] || 0) + 1;\n contentSample.push(`[@${post.author}]: ${post.content}`);\n\n // Count media (this would need to be enhanced based on actual media data structure)\n if (\n post.content.includes(\"http\") ||\n post.content.includes(\"pic.twitter.com\")\n ) {\n mediaCount++;\n }\n }\n\n // Calculate time spread\n const timestamps = posts.map((p) => new Date(p.created_at).getTime())\n .sort();\n const timeSpread = {\n earliest: new Date(timestamps[0]).toISOString(),\n latest: new Date(timestamps[timestamps.length - 1]).toISOString(),\n };\n\n return {\n authorCounts,\n contentSample: contentSample.slice(0, 20), // Limit for token efficiency\n mediaCount,\n timeSpread,\n };\n }\n\n /**\n * Generate summary using OpenAI\n */\n private async generateSummaryWithAI(\n topic: Topic,\n groupedData: ReturnType,\n sentimentData: any,\n ): Promise<{\n mainSummary: string;\n keyPoints: string[];\n trendingKeywords: string[];\n influentialAuthors: string[];\n }> {\n const prompt = this.buildSummarizationPrompt(\n topic,\n groupedData,\n sentimentData,\n );\n\n try {\n const openai = new OpenAI();\n\n const response = await openai.chat.completions.create({\n model: \"gpt-3.5-turbo\",\n messages: [\n {\n role: \"system\",\n content:\n `You are an expert social media analyst specializing in Twitter content summarization. \n Provide concise, insightful summaries that capture the key themes, sentiment, and notable patterns in the conversation.\n Always respond with valid JSON in the exact format requested.`,\n },\n {\n role: \"user\",\n content: prompt,\n },\n ],\n max_tokens: 800,\n temperature: 0.7,\n response_format: { type: \"json_object\" },\n });\n\n const result = response.choices[0].message.content?.trim();\n if (!result) {\n throw new Error(\"No response from OpenAI\");\n }\n\n const summaryData = JSON.parse(result);\n\n // Validate response structure\n if (!summaryData.mainSummary || !Array.isArray(summaryData.keyPoints)) {\n throw new Error(\"Invalid response format from OpenAI\");\n }\n\n return {\n mainSummary: summaryData.mainSummary,\n keyPoints: summaryData.keyPoints || [],\n trendingKeywords: summaryData.trendingKeywords || [],\n influentialAuthors: summaryData.influentialAuthors || [],\n };\n } catch (error: unknown) {\n console.error(\"Error generating summary with AI:\", error);\n\n // Fallback to simple summary\n return this.generateFallbackSummary(topic, groupedData);\n }\n }\n\n /**\n * Build the summarization prompt for OpenAI\n */\n private buildSummarizationPrompt(\n topic: Topic,\n groupedData: ReturnType,\n sentimentData: any,\n ): string {\n const { authorCounts, contentSample, mediaCount, timeSpread } = groupedData;\n\n const topAuthors = Object.entries(authorCounts)\n .sort(([, a], [, b]) => b - a)\n .slice(0, 5)\n .map(([author, count]) => `${author} (${count} posts)`);\n\n const sentimentInfo = sentimentData\n ? `Overall sentiment: ${sentimentData.sentiment} (confidence: ${\n sentimentData.confidence.toFixed(2)\n }, score: ${sentimentData.score.toFixed(2)})`\n : \"Sentiment analysis not available\";\n\n return `\nAnalyze and summarize the following Twitter conversation about \"${topic.title}\":\n\nTOPIC DESCRIPTION: ${topic.description || \"No description provided\"}\n\nCONVERSATION DATA:\n- Total posts: ${contentSample.length}\n- Time range: ${timeSpread.earliest} to ${timeSpread.latest}\n- Posts with media: ${mediaCount}\n- ${sentimentInfo}\n\nTOP CONTRIBUTORS: ${topAuthors.join(\", \")}\n\nSAMPLE POSTS:\n${contentSample.join(\"\\n\\n\")}\n\nPlease provide a comprehensive analysis in the following JSON format:\n{\n \"mainSummary\": \"A 2-3 sentence overview of the main themes and discussion points\",\n \"keyPoints\": [\"Point 1\", \"Point 2\", \"Point 3\", \"Point 4\", \"Point 5\"],\n \"trendingKeywords\": [\"keyword1\", \"keyword2\", \"keyword3\", \"keyword4\", \"keyword5\"],\n \"influentialAuthors\": [\"author1\", \"author2\", \"author3\"]\n}\n\nFocus on:\n1. The main themes and topics being discussed\n2. Any notable trends, events, or developments mentioned\n3. The overall tone and sentiment of the conversation\n4. Key players or influencers in the discussion\n5. Important keywords that capture the essence of the conversation\n\nProvide only valid JSON, no additional text.\n`;\n }\n\n /**\n * Generate a simple fallback summary when AI fails\n */\n private generateFallbackSummary(\n topic: Topic,\n groupedData: ReturnType,\n ): {\n mainSummary: string;\n keyPoints: string[];\n trendingKeywords: string[];\n influentialAuthors: string[];\n } {\n const { authorCounts, contentSample, mediaCount } = groupedData;\n\n const topAuthors = Object.entries(authorCounts)\n .sort(([, a], [, b]) => b - a)\n .slice(0, 3)\n .map(([author]) => author);\n\n return {\n mainSummary:\n `Discussion about ${topic.title} with ${contentSample.length} posts from ${\n Object.keys(authorCounts).length\n } contributors. ${mediaCount} posts included media content.`,\n keyPoints: [\n `${contentSample.length} total posts analyzed`,\n `${Object.keys(authorCounts).length} unique contributors`,\n `${mediaCount} posts with media content`,\n `Most active contributor: ${topAuthors[0] || \"Unknown\"}`,\n \"Detailed analysis unavailable due to processing error\",\n ],\n trendingKeywords: [topic.title.toLowerCase()],\n influentialAuthors: topAuthors,\n };\n }\n\n /**\n * Aggregate sentiment analysis results\n */\n private aggregateSentiment(sentimentResults: any[]): {\n sentiment: \"positive\" | \"negative\" | \"neutral\";\n confidence: number;\n score: number;\n } {\n if (sentimentResults.length === 0) {\n return { sentiment: \"neutral\", confidence: 0.5, score: 0 };\n }\n\n // Calculate average sentiment score and confidence\n const avgScore =\n sentimentResults.reduce((sum, result) => sum + result.score, 0) /\n sentimentResults.length;\n const avgConfidence =\n sentimentResults.reduce((sum, result) => sum + result.confidence, 0) /\n sentimentResults.length;\n\n // Determine overall sentiment based on average score\n let sentiment: \"positive\" | \"negative\" | \"neutral\" = \"neutral\";\n if (avgScore > 0.1) {\n sentiment = \"positive\";\n } else if (avgScore < -0.1) {\n sentiment = \"negative\";\n }\n\n return {\n sentiment,\n confidence: avgConfidence,\n score: avgScore,\n };\n }\n\n /**\n * Get recent summaries for a topic (for checking if summarization is needed)\n */\n async getRecentSummaries(\n topicId: number,\n hoursBack: number = 24,\n ): Promise {\n try {\n return await getSummarizationsForTopic(topicId, 10); // Get up to 10 recent summaries\n } catch (error: unknown) {\n console.error(\n `Error getting recent summaries for topic ${topicId}:`,\n error,\n );\n return [];\n }\n }\n}\n","deleted","label",[168,150],"backend","permalink","https://www.val.town/x/jpd3v/offx/code/backend/twitter-summarization.ts","extension","actionData","errors"]