yt-transcript-strapi-plugin 0.0.21 → 0.0.25

package/dist/server/index.js

@@ -4,14 +4,10 @@ const types_js = require("@modelcontextprotocol/sdk/types.js");
 const zod = require("zod");
 const node_crypto = require("node:crypto");
 const streamableHttp_js = require("@modelcontextprotocol/sdk/server/streamableHttp.js");
-const textsplitters = require("@langchain/textsplitters");
-const prompts = require("@langchain/core/prompts");
-const openai = require("@langchain/openai");
 const youtubei_js = require("youtubei.js");
 const undici = require("undici");
 const FetchTranscriptSchema = zod.z.object({
-  videoId: zod.z.string().min(1, "Video ID or URL is required"),
-  generateReadable: zod.z.boolean().optional().default(false)
+  videoId: zod.z.string().min(1, "Video ID or URL is required")
 });
 const ListTranscriptsSchema = zod.z.object({
   page: zod.z.number().int().min(1).optional().default(1),
@@ -19,7 +15,18 @@ const ListTranscriptsSchema = zod.z.object({
   sort: zod.z.string().optional().default("createdAt:desc")
 });
 const GetTranscriptSchema = zod.z.object({
-  videoId: zod.z.string().min(1, "Video ID is required")
+  videoId: zod.z.string().min(1, "Video ID is required"),
+  includeFullTranscript: zod.z.boolean().optional().default(false),
+  includeTimecodes: zod.z.boolean().optional().default(false),
+  startTime: zod.z.number().min(0).optional(),
+  endTime: zod.z.number().min(0).optional(),
+  chunkIndex: zod.z.number().int().min(0).optional(),
+  chunkSize: zod.z.number().int().min(30).optional()
+});
+const SearchTranscriptSchema = zod.z.object({
+  videoId: zod.z.string().min(1, "Video ID is required"),
+  query: zod.z.string().min(1, "Search query is required"),
+  maxResults: zod.z.number().int().min(1).max(20).optional().default(5)
 });
 const FindTranscriptsSchema = zod.z.object({
   query: zod.z.string().optional(),
@@ -34,6 +41,7 @@ const ToolSchemas = {
   fetch_transcript: FetchTranscriptSchema,
   list_transcripts: ListTranscriptsSchema,
   get_transcript: GetTranscriptSchema,
+  search_transcript: SearchTranscriptSchema,
   find_transcripts: FindTranscriptsSchema
 };
 function validateToolInput(toolName, input) {
@@ -68,31 +76,64 @@ function extractYouTubeID(urlOrID) {
 }
 const fetchTranscriptTool = {
   name: "fetch_transcript",
-  description: "Fetch a transcript from YouTube for a given video ID or URL. Optionally generates a human-readable version using AI. The transcript is saved to the database for future retrieval.",
+  description: "Fetch a transcript from YouTube for a given video ID or URL. The transcript is saved to the database. Returns metadata and preview only to avoid context overflow. Use get_transcript to retrieve content.",
   inputSchema: {
     type: "object",
     properties: {
       videoId: {
         type: "string",
         description: 'YouTube video ID (e.g., "dQw4w9WgXcQ") or full YouTube URL'
-      },
-      generateReadable: {
-        type: "boolean",
-        description: "If true, uses AI to add punctuation and formatting to make the transcript more readable. Requires OpenAI API key configuration.",
-        default: false
       }
     },
     required: ["videoId"]
   }
 };
-async function handleFetchTranscript(strapi2, args) {
+function getVideoDurationMs$1(timecodes) {
+  if (!timecodes || timecodes.length === 0) return 0;
+  const lastEntry = timecodes[timecodes.length - 1];
+  return lastEntry.end || lastEntry.start + (lastEntry.duration || 0);
+}
+function formatTime$2(ms) {
+  const totalSeconds = Math.floor(ms / 1e3);
+  const hours = Math.floor(totalSeconds / 3600);
+  const minutes = Math.floor(totalSeconds % 3600 / 60);
+  const seconds = totalSeconds % 60;
+  if (hours > 0) {
+    return `${hours}:${minutes.toString().padStart(2, "0")}:${seconds.toString().padStart(2, "0")}`;
+  }
+  return `${minutes}:${seconds.toString().padStart(2, "0")}`;
+}
+function buildMetadataResponse(transcript2, previewLength, cached) {
+  const fullText = transcript2.fullTranscript || "";
+  const timecodes = transcript2.transcriptWithTimeCodes || [];
+  const durationMs = getVideoDurationMs$1(timecodes);
+  const wordCount = fullText.split(/\s+/).length;
+  const preview = fullText.length > previewLength ? fullText.substring(0, previewLength) + "..." : fullText;
+  return {
+    message: cached ? "Transcript already exists in database" : "Transcript fetched and saved successfully",
+    cached,
+    videoId: transcript2.videoId,
+    title: transcript2.title,
+    metadata: {
+      wordCount,
+      characterCount: fullText.length,
+      duration: formatTime$2(durationMs),
+      durationSeconds: Math.floor(durationMs / 1e3)
+    },
+    preview,
+    usage: "Use get_transcript with videoId to retrieve full content, specific time ranges, or paginated chunks."
+  };
+}
+async function handleFetchTranscript(strapi, args) {
   const validatedArgs = validateToolInput("fetch_transcript", args);
-  const { videoId: videoIdOrUrl, generateReadable } = validatedArgs;
+  const { videoId: videoIdOrUrl } = validatedArgs;
+  const pluginConfig = await strapi.config.get("plugin::yt-transcript-strapi-plugin");
+  const previewLength = pluginConfig?.previewLength || 500;
   const videoId = extractYouTubeID(videoIdOrUrl);
   if (!videoId) {
     throw new Error(`Invalid YouTube video ID or URL: "${videoIdOrUrl}". Please provide a valid 11-character video ID or YouTube URL.`);
   }
-  const service2 = strapi2.plugin("yt-transcript-strapi-plugin").service("service");
+  const service2 = strapi.plugin("yt-transcript-strapi-plugin").service("service");
   const existingTranscript = await service2.findTranscript(videoId);
   if (existingTranscript) {
     return {
@@ -100,11 +141,7 @@ async function handleFetchTranscript(strapi2, args) {
         {
           type: "text",
           text: JSON.stringify(
-            {
-              message: "Transcript already exists in database",
-              data: existingTranscript,
-              cached: true
-            },
+            buildMetadataResponse(existingTranscript, previewLength, true),
             null,
             2
           )
@@ -122,25 +159,13 @@ async function handleFetchTranscript(strapi2, args) {
     fullTranscript: transcriptData.fullTranscript,
     transcriptWithTimeCodes: transcriptData.transcriptWithTimeCodes
   };
-  if (generateReadable && transcriptData.fullTranscript) {
-    try {
-      const readableTranscript = await service2.generateHumanReadableTranscript(transcriptData.fullTranscript);
-      payload.readableTranscript = readableTranscript;
-    } catch (error) {
-      strapi2.log.warn("[yt-transcript-mcp] Failed to generate readable transcript:", error);
-    }
-  }
   const savedTranscript = await service2.saveTranscript(payload);
   return {
     content: [
       {
         type: "text",
         text: JSON.stringify(
-          {
-            message: "Transcript fetched and saved successfully",
-            data: savedTranscript,
-            cached: false
-          },
+          buildMetadataResponse(savedTranscript, previewLength, false),
           null,
           2
         )
@@ -173,17 +198,17 @@ const listTranscriptsTool = {
     required: []
   }
 };
-async function handleListTranscripts(strapi2, args) {
+async function handleListTranscripts(strapi, args) {
   const validatedArgs = validateToolInput("list_transcripts", args);
   const { page, pageSize, sort } = validatedArgs;
   const start = (page - 1) * pageSize;
-  const transcripts = await strapi2.documents("plugin::yt-transcript-strapi-plugin.transcript").findMany({
+  const transcripts = await strapi.documents("plugin::yt-transcript-strapi-plugin.transcript").findMany({
     sort,
     limit: pageSize,
     start,
     fields: ["id", "documentId", "title", "videoId", "createdAt", "updatedAt"]
   });
-  const allTranscripts = await strapi2.documents("plugin::yt-transcript-strapi-plugin.transcript").findMany({});
+  const allTranscripts = await strapi.documents("plugin::yt-transcript-strapi-plugin.transcript").findMany({});
   const total = allTranscripts.length;
   return {
     content: [
@@ -208,26 +233,305 @@ async function handleListTranscripts(strapi2, args) {
 }
 const getTranscriptTool = {
   name: "get_transcript",
-  description: "Get a specific saved transcript by YouTube video ID. Returns the full transcript data including any readable version if available.",
+  description: "Get a saved transcript by YouTube video ID. Returns metadata and preview by default. Use parameters to get full content or specific time ranges to avoid context overflow.",
   inputSchema: {
     type: "object",
     properties: {
       videoId: {
         type: "string",
         description: 'YouTube video ID (e.g., "dQw4w9WgXcQ") or full YouTube URL'
+      },
+      includeFullTranscript: {
+        type: "boolean",
+        description: "Include the complete transcript text. Warning: may cause context overflow for long videos. Default: false",
+        default: false
+      },
+      includeTimecodes: {
+        type: "boolean",
+        description: "Include the transcript with timecodes array. Warning: significantly increases response size. Default: false",
+        default: false
+      },
+      startTime: {
+        type: "number",
+        description: "Start time in seconds for fetching a specific portion of the transcript"
+      },
+      endTime: {
+        type: "number",
+        description: "End time in seconds for fetching a specific portion of the transcript"
+      },
+      chunkIndex: {
+        type: "number",
+        description: "Chunk index (0-based) when paginating through transcript. Use with chunkSize to paginate through long videos."
+      },
+      chunkSize: {
+        type: "number",
+        description: "Chunk size in seconds. Overrides config default. Use with chunkIndex for pagination."
       }
     },
     required: ["videoId"]
   }
 };
-async function handleGetTranscript(strapi2, args) {
+function getTranscriptForTimeRange(timecodes, startTimeMs, endTimeMs) {
+  const entries = timecodes.filter(
+    (entry) => entry.start >= startTimeMs && entry.start < endTimeMs
+  );
+  const text = entries.map((e) => e.text).join(" ");
+  return { text, entries };
+}
+function getVideoDurationMs(timecodes) {
+  if (!timecodes || timecodes.length === 0) return 0;
+  const lastEntry = timecodes[timecodes.length - 1];
+  return lastEntry.end || lastEntry.start + (lastEntry.duration || 0);
+}
+function formatTime$1(ms) {
+  const totalSeconds = Math.floor(ms / 1e3);
+  const hours = Math.floor(totalSeconds / 3600);
+  const minutes = Math.floor(totalSeconds % 3600 / 60);
+  const seconds = totalSeconds % 60;
+  if (hours > 0) {
+    return `${hours}:${minutes.toString().padStart(2, "0")}:${seconds.toString().padStart(2, "0")}`;
+  }
+  return `${minutes}:${seconds.toString().padStart(2, "0")}`;
+}
+async function handleGetTranscript(strapi, args) {
   const validatedArgs = validateToolInput("get_transcript", args);
-  const { videoId: videoIdOrUrl } = validatedArgs;
+  const {
+    videoId: videoIdOrUrl,
+    includeFullTranscript,
+    includeTimecodes,
+    startTime,
+    endTime,
+    chunkIndex,
+    chunkSize: chunkSizeOverride
+  } = validatedArgs;
+  const pluginConfig = await strapi.config.get("plugin::yt-transcript-strapi-plugin");
+  const defaultChunkSize = pluginConfig?.chunkSizeSeconds || 300;
+  const previewLength = pluginConfig?.previewLength || 500;
+  const maxFullTranscriptLength = pluginConfig?.maxFullTranscriptLength || 5e4;
+  const chunkSizeSeconds = chunkSizeOverride || defaultChunkSize;
+  const videoId = extractYouTubeID(videoIdOrUrl);
+  if (!videoId) {
+    throw new Error(`Invalid YouTube video ID or URL: "${videoIdOrUrl}". Please provide a valid 11-character video ID or YouTube URL.`);
+  }
+  const service2 = strapi.plugin("yt-transcript-strapi-plugin").service("service");
+  const transcript2 = await service2.findTranscript(videoId);
+  if (!transcript2) {
+    return {
+      content: [
+        {
+          type: "text",
+          text: JSON.stringify(
+            {
+              error: true,
+              message: `No transcript found for video ID: ${videoId}. Use fetch_transcript to fetch it from YouTube first.`,
+              videoId
+            },
+            null,
+            2
+          )
+        }
+      ]
+    };
+  }
+  const timecodes = transcript2.transcriptWithTimeCodes || [];
+  const fullText = transcript2.fullTranscript || "";
+  const durationMs = getVideoDurationMs(timecodes);
+  const totalChunks = Math.ceil(durationMs / (chunkSizeSeconds * 1e3));
+  const wordCount = fullText.split(/\s+/).length;
+  const response = {
+    videoId: transcript2.videoId,
+    title: transcript2.title,
+    metadata: {
+      wordCount,
+      characterCount: fullText.length,
+      duration: formatTime$1(durationMs),
+      durationSeconds: Math.floor(durationMs / 1e3),
+      totalChunks,
+      chunkSizeSeconds
+    }
+  };
+  if (startTime !== void 0 || endTime !== void 0) {
+    const startMs = (startTime || 0) * 1e3;
+    const endMs = endTime !== void 0 ? endTime * 1e3 : durationMs;
+    const { text, entries } = getTranscriptForTimeRange(timecodes, startMs, endMs);
+    response.timeRange = {
+      startTime: startTime || 0,
+      endTime: endTime || Math.floor(durationMs / 1e3),
+      startFormatted: formatTime$1(startMs),
+      endFormatted: formatTime$1(endMs)
+    };
+    response.transcript = text;
+    if (includeTimecodes) {
+      response.transcriptWithTimeCodes = entries;
+    }
+  } else if (chunkIndex !== void 0) {
+    const chunkStartMs = chunkIndex * chunkSizeSeconds * 1e3;
+    const chunkEndMs = Math.min((chunkIndex + 1) * chunkSizeSeconds * 1e3, durationMs);
+    if (chunkStartMs >= durationMs) {
+      response.error = `Chunk index ${chunkIndex} is out of range. Total chunks: ${totalChunks} (0-${totalChunks - 1})`;
+    } else {
+      const { text, entries } = getTranscriptForTimeRange(timecodes, chunkStartMs, chunkEndMs);
+      response.chunk = {
+        index: chunkIndex,
+        totalChunks,
+        startTime: Math.floor(chunkStartMs / 1e3),
+        endTime: Math.floor(chunkEndMs / 1e3),
+        startFormatted: formatTime$1(chunkStartMs),
+        endFormatted: formatTime$1(chunkEndMs)
+      };
+      response.transcript = text;
+      if (includeTimecodes) {
+        response.transcriptWithTimeCodes = entries;
+      }
+      if (chunkIndex < totalChunks - 1) {
+        response.nextChunk = `Use chunkIndex: ${chunkIndex + 1} to get the next portion`;
+      }
+      if (chunkIndex > 0) {
+        response.previousChunk = `Use chunkIndex: ${chunkIndex - 1} to get the previous portion`;
+      }
+    }
+  } else if (includeFullTranscript || fullText.length <= maxFullTranscriptLength) {
+    response.transcript = fullText;
+    if (includeTimecodes) {
+      response.transcriptWithTimeCodes = timecodes;
+    }
+    if (includeFullTranscript && fullText.length > maxFullTranscriptLength) {
+      response.warning = "Full transcript included. For long videos, consider using chunkIndex, startTime/endTime, or search_transcript to reduce response size.";
+    } else if (fullText.length <= maxFullTranscriptLength) {
+      response.note = "Full transcript auto-loaded (fits within context limit).";
+    }
+  } else {
+    const preview = fullText.length > previewLength ? fullText.substring(0, previewLength) + "..." : fullText;
+    response.preview = preview;
+    response.isLargeTranscript = true;
+    response.usage = {
+      fullTranscript: "Set includeFullTranscript: true to get complete text (warning: may exceed context)",
+      search: "Use search_transcript to find relevant portions by keyword (recommended for large transcripts)",
+      timeRange: "Use startTime and endTime (in seconds) to get a specific portion",
+      pagination: `Use chunkIndex (0-${totalChunks - 1}) to paginate through ${chunkSizeSeconds}s chunks`
+    };
+  }
+  return {
+    content: [
+      {
+        type: "text",
+        text: JSON.stringify(response, null, 2)
+      }
+    ]
+  };
+}
+const searchTranscriptTool = {
+  name: "search_transcript",
+  description: "Search within a saved transcript using BM25 scoring. Returns the most relevant segments matching your query with timestamps. Use this to find specific content in long videos without loading the entire transcript.",
+  inputSchema: {
+    type: "object",
+    properties: {
+      videoId: {
+        type: "string",
+        description: 'YouTube video ID (e.g., "dQw4w9WgXcQ") or full YouTube URL'
+      },
+      query: {
+        type: "string",
+        description: "Search query - keywords or phrases to find in the transcript"
+      },
+      maxResults: {
+        type: "number",
+        description: "Maximum number of results to return (default: 5, max: 20)",
+        default: 5
+      }
+    },
+    required: ["videoId", "query"]
+  }
+};
+function tokenize(text) {
+  return text.toLowerCase().replace(/[^\w\s]/g, " ").split(/\s+/).filter((word) => word.length > 1);
+}
+function calculateIDF(segments, vocabulary) {
+  const idf = /* @__PURE__ */ new Map();
+  const N = segments.length;
+  for (const term of vocabulary) {
+    const docsWithTerm = segments.filter(
+      (seg) => tokenize(seg.text).includes(term)
+    ).length;
+    idf.set(term, Math.log((N - docsWithTerm + 0.5) / (docsWithTerm + 0.5) + 1));
+  }
+  return idf;
+}
+function bm25Score(segmentTokens, queryTokens, idf, avgDocLength, k1 = 1.5, b = 0.75) {
+  const docLength = segmentTokens.length;
+  let score = 0;
+  const tf = /* @__PURE__ */ new Map();
+  for (const token of segmentTokens) {
+    tf.set(token, (tf.get(token) || 0) + 1);
+  }
+  for (const term of queryTokens) {
+    const termFreq = tf.get(term) || 0;
+    const termIdf = idf.get(term) || 0;
+    if (termFreq > 0) {
+      const numerator = termFreq * (k1 + 1);
+      const denominator = termFreq + k1 * (1 - b + b * (docLength / avgDocLength));
+      score += termIdf * (numerator / denominator);
+    }
+  }
+  return score;
+}
+function formatTime(ms) {
+  const totalSeconds = Math.floor(ms / 1e3);
+  const hours = Math.floor(totalSeconds / 3600);
+  const minutes = Math.floor(totalSeconds % 3600 / 60);
+  const seconds = totalSeconds % 60;
+  if (hours > 0) {
+    return `${hours}:${minutes.toString().padStart(2, "0")}:${seconds.toString().padStart(2, "0")}`;
+  }
+  return `${minutes}:${seconds.toString().padStart(2, "0")}`;
+}
+function createSegments(timecodes, segmentDurationMs) {
+  if (!timecodes || timecodes.length === 0) return [];
+  const segments = [];
+  let currentSegment = [];
+  let segmentStartTime = timecodes[0].start;
+  for (const entry of timecodes) {
+    const segmentEndTime = segmentStartTime + segmentDurationMs;
+    if (entry.start < segmentEndTime) {
+      currentSegment.push(entry);
+    } else {
+      if (currentSegment.length > 0) {
+        const endTime = currentSegment[currentSegment.length - 1].end || currentSegment[currentSegment.length - 1].start + (currentSegment[currentSegment.length - 1].duration || 0);
+        segments.push({
+          text: currentSegment.map((e) => e.text).join(" "),
+          startTime: Math.floor(segmentStartTime / 1e3),
+          endTime: Math.floor(endTime / 1e3),
+          startFormatted: formatTime(segmentStartTime),
+          endFormatted: formatTime(endTime)
+        });
+      }
+      segmentStartTime = entry.start;
+      currentSegment = [entry];
+    }
+  }
+  if (currentSegment.length > 0) {
+    const endTime = currentSegment[currentSegment.length - 1].end || currentSegment[currentSegment.length - 1].start + (currentSegment[currentSegment.length - 1].duration || 0);
+    segments.push({
+      text: currentSegment.map((e) => e.text).join(" "),
+      startTime: Math.floor(segmentStartTime / 1e3),
+      endTime: Math.floor(endTime / 1e3),
+      startFormatted: formatTime(segmentStartTime),
+      endFormatted: formatTime(endTime)
+    });
+  }
+  return segments;
+}
+async function handleSearchTranscript(strapi, args) {
+  const validatedArgs = validateToolInput("search_transcript", args);
+  const { videoId: videoIdOrUrl, query, maxResults: maxResultsInput } = validatedArgs;
+  const pluginConfig = await strapi.config.get("plugin::yt-transcript-strapi-plugin");
+  const segmentSeconds = pluginConfig?.searchSegmentSeconds || 30;
+  const maxResults = Math.min(Math.max(maxResultsInput || 5, 1), 20);
   const videoId = extractYouTubeID(videoIdOrUrl);
   if (!videoId) {
     throw new Error(`Invalid YouTube video ID or URL: "${videoIdOrUrl}". Please provide a valid 11-character video ID or YouTube URL.`);
   }
-  const service2 = strapi2.plugin("yt-transcript-strapi-plugin").service("service");
+  const service2 = strapi.plugin("yt-transcript-strapi-plugin").service("service");
   const transcript2 = await service2.findTranscript(videoId);
   if (!transcript2) {
     return {
@@ -247,13 +551,90 @@ async function handleGetTranscript(strapi2, args) {
       ]
     };
   }
+  const timecodes = transcript2.transcriptWithTimeCodes || [];
+  if (timecodes.length === 0) {
+    return {
+      content: [
+        {
+          type: "text",
+          text: JSON.stringify(
+            {
+              error: true,
+              message: "Transcript has no timecode data for searching.",
+              videoId
+            },
+            null,
+            2
+          )
+        }
+      ]
+    };
+  }
+  const segments = createSegments(timecodes, segmentSeconds * 1e3);
+  if (segments.length === 0) {
+    return {
+      content: [
+        {
+          type: "text",
+          text: JSON.stringify(
+            {
+              error: true,
+              message: "Could not create searchable segments from transcript.",
+              videoId
+            },
+            null,
+            2
+          )
+        }
+      ]
+    };
+  }
+  const queryTokens = tokenize(query);
+  if (queryTokens.length === 0) {
+    return {
+      content: [
+        {
+          type: "text",
+          text: JSON.stringify(
+            {
+              error: true,
+              message: "Query is empty or contains only stop words.",
+              query
+            },
+            null,
+            2
+          )
+        }
+      ]
+    };
+  }
+  const vocabulary = new Set(queryTokens);
+  const idf = calculateIDF(segments, vocabulary);
+  const avgDocLength = segments.reduce((sum, seg) => sum + tokenize(seg.text).length, 0) / segments.length;
+  const scoredSegments = segments.map((segment) => ({
+    ...segment,
+    score: bm25Score(tokenize(segment.text), queryTokens, idf, avgDocLength)
+  }));
+  const results = scoredSegments.filter((seg) => seg.score > 0).sort((a, b) => b.score - a.score).slice(0, maxResults);
   return {
     content: [
       {
         type: "text",
         text: JSON.stringify(
           {
-            data: transcript2
+            videoId: transcript2.videoId,
+            title: transcript2.title,
+            query,
+            totalSegments: segments.length,
+            matchingResults: results.length,
+            results: results.map((r) => ({
+              text: r.text,
+              startTime: r.startTime,
+              endTime: r.endTime,
+              timeRange: `${r.startFormatted} - ${r.endFormatted}`,
+              score: Math.round(r.score * 100) / 100
+            })),
+            usage: results.length > 0 ? `Use get_transcript with startTime: ${results[0].startTime} and endTime: ${results[0].endTime} to get full context for the top result.` : "No matches found. Try different keywords."
           },
           null,
           2
@@ -312,11 +693,10 @@ function truncateText(text, maxLength) {
 function truncateTranscripts(transcripts) {
   return transcripts.map((transcript2) => ({
     ...transcript2,
-    fullTranscript: truncateText(transcript2.fullTranscript, TRANSCRIPT_PREVIEW_LENGTH),
-    readableTranscript: truncateText(transcript2.readableTranscript, TRANSCRIPT_PREVIEW_LENGTH)
+    fullTranscript: truncateText(transcript2.fullTranscript, TRANSCRIPT_PREVIEW_LENGTH)
   }));
 }
-async function handleFindTranscripts(strapi2, args) {
+async function handleFindTranscripts(strapi, args) {
   const validatedArgs = validateToolInput("find_transcripts", args);
   const { query, videoId, title, includeFullContent, page, pageSize, sort } = validatedArgs;
   const start = (page - 1) * pageSize;
@@ -331,17 +711,16 @@ async function handleFindTranscripts(strapi2, args) {
     filters.$or = [
       { title: { $containsi: query } },
       { videoId: { $containsi: query } },
-      { fullTranscript: { $containsi: query } },
-      { readableTranscript: { $containsi: query } }
+      { fullTranscript: { $containsi: query } }
     ];
   }
-  const transcripts = await strapi2.documents("plugin::yt-transcript-strapi-plugin.transcript").findMany({
+  const transcripts = await strapi.documents("plugin::yt-transcript-strapi-plugin.transcript").findMany({
     filters,
     sort,
     limit: pageSize,
     start
   });
-  const allMatching = await strapi2.documents("plugin::yt-transcript-strapi-plugin.transcript").findMany({
+  const allMatching = await strapi.documents("plugin::yt-transcript-strapi-plugin.transcript").findMany({
     filters
   });
   const total = allMatching.length;
@@ -377,15 +756,17 @@ const tools = [
   fetchTranscriptTool,
   listTranscriptsTool,
   getTranscriptTool,
+  searchTranscriptTool,
   findTranscriptsTool
 ];
 const toolHandlers = {
   fetch_transcript: handleFetchTranscript,
   list_transcripts: handleListTranscripts,
   get_transcript: handleGetTranscript,
+  search_transcript: handleSearchTranscript,
   find_transcripts: handleFindTranscripts
 };
-async function handleToolCall(strapi2, request) {
+async function handleToolCall(strapi, request) {
   const { name, arguments: args } = request.params;
   const handler = toolHandlers[name];
   if (!handler) {
@@ -393,13 +774,13 @@ async function handleToolCall(strapi2, request) {
   }
   const startTime = Date.now();
   try {
-    const result = await handler(strapi2, args || {});
+    const result = await handler(strapi, args || {});
     const duration = Date.now() - startTime;
-    strapi2.log.debug(`[yt-transcript-mcp] Tool ${name} executed successfully in ${duration}ms`);
+    strapi.log.debug(`[yt-transcript-mcp] Tool ${name} executed successfully in ${duration}ms`);
     return result;
   } catch (error) {
     const duration = Date.now() - startTime;
-    strapi2.log.error(`[yt-transcript-mcp] Tool ${name} failed after ${duration}ms`, {
+    strapi.log.error(`[yt-transcript-mcp] Tool ${name} failed after ${duration}ms`, {
       error: error instanceof Error ? error.message : String(error)
     });
     return {
@@ -420,7 +801,7 @@ async function handleToolCall(strapi2, request) {
     };
   }
 }
-function createMcpServer(strapi2) {
+function createMcpServer(strapi) {
   const server = new index_js.Server(
     {
       name: "yt-transcript-mcp",
@@ -433,53 +814,57 @@ function createMcpServer(strapi2) {
     }
   );
   server.setRequestHandler(types_js.ListToolsRequestSchema, async () => {
-    strapi2.log.debug("[yt-transcript-mcp] Listing tools");
+    strapi.log.debug("[yt-transcript-mcp] Listing tools");
     return { tools };
   });
   server.setRequestHandler(types_js.CallToolRequestSchema, async (request) => {
-    strapi2.log.debug(`[yt-transcript-mcp] Tool call: ${request.params.name}`);
-    return handleToolCall(strapi2, request);
+    strapi.log.debug(`[yt-transcript-mcp] Tool call: ${request.params.name}`);
+    return handleToolCall(strapi, request);
   });
-  strapi2.log.info("[yt-transcript-mcp] MCP server created with tools:", {
+  strapi.log.info("[yt-transcript-mcp] MCP server created with tools:", {
     tools: tools.map((t) => t.name)
   });
   return server;
 }
-const bootstrap = async ({ strapi: strapi2 }) => {
-  const plugin = strapi2.plugin("yt-transcript-strapi-plugin");
-  plugin.createMcpServer = () => createMcpServer(strapi2);
+const bootstrap = async ({ strapi }) => {
+  const plugin = strapi.plugin("yt-transcript-strapi-plugin");
+  plugin.createMcpServer = () => createMcpServer(strapi);
   plugin.sessions = /* @__PURE__ */ new Map();
-  strapi2.log.info("[yt-transcript-mcp] MCP plugin initialized");
-  strapi2.log.info("[yt-transcript-mcp] MCP endpoint available at: /api/yt-transcript-strapi-plugin/mcp");
+  strapi.log.info("[yt-transcript-mcp] MCP plugin initialized");
+  strapi.log.info("[yt-transcript-mcp] MCP endpoint available at: /api/yt-transcript-strapi-plugin/mcp");
 };
-const destroy = ({ strapi: strapi2 }) => {
+const destroy = ({ strapi }) => {
 };
-const register = ({ strapi: strapi2 }) => {
+const register = ({ strapi }) => {
 };
 const config = {
   default: {
-    openAIApiKey: "",
-    model: "gpt-4o-mini",
-    temp: 0.7,
-    maxTokens: 4096,
-    proxyUrl: ""
+    proxyUrl: "",
     // Optional: HTTP/HTTPS proxy for YouTube requests (e.g., 'http://user:pass@proxy.example.com:8080')
+    chunkSizeSeconds: 300,
+    // Default chunk size for transcript pagination (5 minutes)
+    previewLength: 500,
+    // Default preview length in characters
+    maxFullTranscriptLength: 5e4,
+    // Auto-load full transcript if under this character count (~12K tokens)
+    searchSegmentSeconds: 30
+    // Segment size for BM25 search scoring
   },
   validator(config2) {
-    if (config2.openAIApiKey && typeof config2.openAIApiKey !== "string") {
-      throw new Error("openAIApiKey must be a string");
+    if (config2.proxyUrl && typeof config2.proxyUrl !== "string") {
+      throw new Error("proxyUrl must be a string");
     }
-    if (config2.model && typeof config2.model !== "string") {
-      throw new Error("model must be a string");
+    if (config2.chunkSizeSeconds !== void 0 && (typeof config2.chunkSizeSeconds !== "number" || config2.chunkSizeSeconds < 30)) {
+      throw new Error("chunkSizeSeconds must be a number >= 30");
     }
-    if (config2.temp !== void 0 && (typeof config2.temp !== "number" || config2.temp < 0 || config2.temp > 2)) {
-      throw new Error("temp must be a number between 0 and 2");
+    if (config2.previewLength !== void 0 && (typeof config2.previewLength !== "number" || config2.previewLength < 100)) {
+      throw new Error("previewLength must be a number >= 100");
     }
-    if (config2.maxTokens !== void 0 && (typeof config2.maxTokens !== "number" || config2.maxTokens < 1)) {
-      throw new Error("maxTokens must be a positive number");
+    if (config2.maxFullTranscriptLength !== void 0 && (typeof config2.maxFullTranscriptLength !== "number" || config2.maxFullTranscriptLength < 1e3)) {
+      throw new Error("maxFullTranscriptLength must be a number >= 1000");
     }
-    if (config2.proxyUrl && typeof config2.proxyUrl !== "string") {
-      throw new Error("proxyUrl must be a string");
+    if (config2.searchSegmentSeconds !== void 0 && (typeof config2.searchSegmentSeconds !== "number" || config2.searchSegmentSeconds < 10)) {
+      throw new Error("searchSegmentSeconds must be a number >= 10");
     }
   }
 };
@@ -513,9 +898,6 @@ const attributes = {
   },
   transcriptWithTimeCodes: {
     type: "json"
-  },
-  readableTranscript: {
-    type: "richtext"
   }
 };
 const schema = {
@@ -532,41 +914,34 @@ const transcript = {
 const contentTypes = {
   transcript
 };
-const controller = ({ strapi: strapi2 }) => ({
+const controller = ({ strapi }) => ({
   async getTranscript(ctx) {
     const videoId = extractYouTubeID(ctx.params.videoId);
     if (!videoId) {
       return ctx.body = { error: "Invalid YouTube URL or ID", data: null };
     }
-    const found = await strapi2.plugin("yt-transcript-strapi-plugin").service("service").findTranscript(videoId);
+    const found = await strapi.plugin("yt-transcript-strapi-plugin").service("service").findTranscript(videoId);
     if (found) {
       return ctx.body = { data: found };
     }
-    const transcriptData = await strapi2.plugin("yt-transcript-strapi-plugin").service("service").getTranscript(videoId);
-    let readableTranscript = null;
-    try {
-      readableTranscript = await strapi2.plugin("yt-transcript-strapi-plugin").service("service").generateHumanReadableTranscript(transcriptData.fullTranscript);
-    } catch (error) {
-      strapi2.log.debug("[yt-transcript] Readable transcript generation skipped");
-    }
+    const transcriptData = await strapi.plugin("yt-transcript-strapi-plugin").service("service").getTranscript(videoId);
     const payload = {
       videoId,
       title: transcriptData?.title || "No title found",
       fullTranscript: transcriptData?.fullTranscript,
-      transcriptWithTimeCodes: transcriptData?.transcriptWithTimeCodes,
-      readableTranscript
+      transcriptWithTimeCodes: transcriptData?.transcriptWithTimeCodes
     };
-    const transcript2 = await strapi2.plugin("yt-transcript-strapi-plugin").service("service").saveTranscript(payload);
+    const transcript2 = await strapi.plugin("yt-transcript-strapi-plugin").service("service").saveTranscript(payload);
     ctx.body = { data: transcript2 };
   }
 });
-const mcpController = ({ strapi: strapi2 }) => ({
+const mcpController = ({ strapi }) => ({
   /**
    * Handle MCP requests (POST, GET, DELETE)
    * Creates a new server+transport per session for proper isolation
    */
   async handle(ctx) {
-    const plugin = strapi2.plugin("yt-transcript-strapi-plugin");
+    const plugin = strapi.plugin("yt-transcript-strapi-plugin");
     if (!plugin.createMcpServer) {
       ctx.status = 503;
       ctx.body = {
@@ -586,12 +961,12 @@ const mcpController = ({ strapi: strapi2 }) => ({
         await server.connect(transport);
         session = { server, transport, createdAt: Date.now() };
         plugin.sessions.set(sessionId, session);
-        strapi2.log.debug(`[yt-transcript-mcp] New session created: ${sessionId}`);
+        strapi.log.debug(`[yt-transcript-mcp] New session created: ${sessionId}`);
       }
       await session.transport.handleRequest(ctx.req, ctx.res, ctx.request.body);
       ctx.respond = false;
     } catch (error) {
-      strapi2.log.error("[yt-transcript-mcp] Error handling MCP request", {
+      strapi.log.error("[yt-transcript-mcp] Error handling MCP request", {
         error: error instanceof Error ? error.message : String(error),
         method: ctx.method,
         path: ctx.path
@@ -671,18 +1046,6 @@ const routes = {
     routes: [...admin]
   }
 };
-async function initializeModel({
-  openAIApiKey,
-  model,
-  temp
-}) {
-  return new openai.ChatOpenAI({
-    temperature: temp,
-    openAIApiKey,
-    modelName: model,
-    maxTokens: 1e3
-  });
-}
 function isRequestLike(input) {
   return typeof input === "object" && input !== null && "url" in input && typeof input.url === "string" && "method" in input;
 }
@@ -825,49 +1188,14 @@ const fetchTranscript = async (videoId, options2) => {
     );
   }
 };
-async function processTextChunks(chunks, model) {
-  const punctuationPrompt = prompts.PromptTemplate.fromTemplate(
-    "Add proper punctuation and capitalization to the following text chunk:\n\n{chunk}"
-  );
-  const punctuationChain = punctuationPrompt.pipe(model);
-  const processedChunks = await Promise.all(
-    chunks.map(async (chunk) => {
-      const result = await punctuationChain.invoke({ chunk });
-      return result.content;
-    })
-  );
-  return processedChunks.join(" ");
-}
-async function generateModifiedTranscript(rawTranscript) {
-  const pluginSettings = await strapi.config.get(
-    "plugin::yt-transcript-strapi-plugin"
-  );
-  if (!pluginSettings.openAIApiKey || !pluginSettings.model || !pluginSettings.temp || !pluginSettings.maxTokens) {
-    throw new Error("Missing required configuration for YTTranscript");
-  }
-  const chatModel = await initializeModel({
-    openAIApiKey: pluginSettings.openAIApiKey,
-    model: pluginSettings.model,
-    temp: pluginSettings.temp,
-    maxTokens: pluginSettings.maxTokens
-  });
-  const splitter = new textsplitters.TokenTextSplitter({
-    chunkSize: 1e3,
-    chunkOverlap: 200
-  });
-  const transcriptChunks = await splitter.createDocuments([rawTranscript]);
-  const chunkTexts = transcriptChunks.map((chunk) => chunk.pageContent);
-  const modifiedTranscript = await processTextChunks(chunkTexts, chatModel);
-  return modifiedTranscript;
-}
-const service = ({ strapi: strapi2 }) => ({
+const service = ({ strapi }) => ({
   async getTranscript(identifier) {
     const youtubeIdRegex = /^[a-zA-Z0-9_-]{11}$/;
     const isValid = youtubeIdRegex.test(identifier);
     if (!isValid) {
       return { error: "Invalid video ID", data: null };
     }
-    const pluginSettings = await strapi2.config.get(
+    const pluginSettings = await strapi.config.get(
       "plugin::yt-transcript-strapi-plugin"
     );
     const transcriptData = await fetchTranscript(identifier, {
@@ -880,20 +1208,16 @@ const service = ({ strapi: strapi2 }) => ({
     };
   },
   async saveTranscript(payload) {
-    return await strapi2.documents("plugin::yt-transcript-strapi-plugin.transcript").create({
+    return await strapi.documents("plugin::yt-transcript-strapi-plugin.transcript").create({
       data: payload
     });
   },
   async findTranscript(videoId) {
-    const transcriptData = await strapi2.documents("plugin::yt-transcript-strapi-plugin.transcript").findFirst({
+    const transcriptData = await strapi.documents("plugin::yt-transcript-strapi-plugin.transcript").findFirst({
       filters: { videoId }
     });
     if (!transcriptData) return null;
     return transcriptData;
-  },
-  async generateHumanReadableTranscript(transcript2) {
-    const modifiedTranscript = await generateModifiedTranscript(transcript2);
-    return modifiedTranscript;
   }
 });
 const services = {