Add searching feature

2025-06-08 21:53:45 +09:00
parent cd0e4065fc
commit b7f3ca6a27
16 changed files with 564 additions and 194 deletions
--- a/services/batch-scheduler.ts
+++ b/services/batch-scheduler.ts
@ -103,9 +103,12 @@ class BatchScheduler {
          }

          // Episode category migration
-          const { migrateEpisodesWithCategories, getEpisodeCategoryMigrationStatus } =
-            await import("./database.js");
-          const episodeMigrationStatus = await getEpisodeCategoryMigrationStatus();
+          const {
+            migrateEpisodesWithCategories,
+            getEpisodeCategoryMigrationStatus,
+          } = await import("./database.js");
+          const episodeMigrationStatus =
+            await getEpisodeCategoryMigrationStatus();

          if (!episodeMigrationStatus.migrationComplete) {
            console.log("🔄 Running episode category migration...");
@ -117,10 +120,7 @@ class BatchScheduler {

          this.migrationCompleted = true;
        } catch (migrationError) {
-          console.error(
-            "❌ Error during category migrations:",
-            migrationError,
-          );
+          console.error("❌ Error during category migrations:", migrationError);
          // Don't fail the entire batch process due to migration error
          this.migrationCompleted = true; // Mark as completed to avoid retrying every batch
        }
--- a/services/content-extractor.ts
+++ b/services/content-extractor.ts
@ -67,7 +67,7 @@ export async function extractArticleContent(
    }

    // Wait for potential dynamic content
-    await new Promise(resolve => setTimeout(resolve, 2000));
+    await new Promise((resolve) => setTimeout(resolve, 2000));

    // Extract content using page.evaluate
    const extractedData = await page.evaluate(() => {
@ -112,7 +112,9 @@ export async function extractArticleContent(
        "";

      // Extract description
-      const descriptionMeta = document.querySelector('meta[name="description"]');
+      const descriptionMeta = document.querySelector(
+        'meta[name="description"]',
+      );
      const ogDescriptionMeta = document.querySelector(
        'meta[property="og:description"]',
      );
--- a/services/database.ts
+++ b/services/database.ts
@ -218,7 +218,9 @@ function initializeDatabase(): Database {

  // Ensure the category column exists in episodes
  const episodeInfos = db.prepare("PRAGMA table_info(episodes);").all();
-  const hasEpisodeCategory = episodeInfos.some((col: any) => col.name === "category");
+  const hasEpisodeCategory = episodeInfos.some(
+    (col: any) => col.name === "category",
+  );

  if (!hasEpisodeCategory) {
    db.exec("ALTER TABLE episodes ADD COLUMN category TEXT DEFAULT NULL;");
@ -581,6 +583,88 @@ export async function fetchEpisodeWithSourceInfo(
  }
 }

+// Search episodes with feed information
+export async function searchEpisodesWithFeedInfo(
+  query: string,
+  category?: string,
+): Promise<EpisodeWithFeedInfo[]> {
+  try {
+    let whereClause = `
+      WHERE f.active = 1 
+      AND (
+        e.title LIKE ? 
+        OR e.description LIKE ? 
+        OR a.title LIKE ? 
+        OR a.description LIKE ? 
+        OR a.content LIKE ?
+      )
+    `;
+
+    const searchPattern = `%${query}%`;
+    const params = [
+      searchPattern,
+      searchPattern,
+      searchPattern,
+      searchPattern,
+      searchPattern,
+    ];
+
+    if (category) {
+      whereClause += " AND f.category = ?";
+      params.push(category);
+    }
+
+    const stmt = db.prepare(`
+      SELECT 
+        e.id,
+        e.title,
+        e.description,
+        e.audio_path as audioPath,
+        e.duration,
+        e.file_size as fileSize,
+        e.category,
+        e.created_at as createdAt,
+        e.article_id as articleId,
+        a.title as articleTitle,
+        a.link as articleLink,
+        a.pub_date as articlePubDate,
+        f.id as feedId,
+        f.title as feedTitle,
+        f.url as feedUrl,
+        f.category as feedCategory
+      FROM episodes e
+      JOIN articles a ON e.article_id = a.id
+      JOIN feeds f ON a.feed_id = f.id
+      ${whereClause}
+      ORDER BY e.created_at DESC
+    `);
+
+    const rows = stmt.all(...params) as any[];
+
+    return rows.map((row) => ({
+      id: row.id,
+      title: row.title,
+      description: row.description,
+      audioPath: row.audioPath,
+      duration: row.duration,
+      fileSize: row.fileSize,
+      category: row.category,
+      createdAt: row.createdAt,
+      articleId: row.articleId,
+      articleTitle: row.articleTitle,
+      articleLink: row.articleLink,
+      articlePubDate: row.articlePubDate,
+      feedId: row.feedId,
+      feedTitle: row.feedTitle,
+      feedUrl: row.feedUrl,
+      feedCategory: row.feedCategory,
+    }));
+  } catch (error) {
+    console.error("Error searching episodes with feed info:", error);
+    throw error;
+  }
+}
+
 export async function getAllFeedsIncludingInactive(): Promise<Feed[]> {
  try {
    const stmt = db.prepare("SELECT * FROM feeds ORDER BY created_at DESC");
@ -1309,7 +1393,9 @@ export async function deleteEpisode(episodeId: string): Promise<boolean> {
 }

 // Episode category management functions
-export async function getEpisodesByCategory(category?: string): Promise<EpisodeWithFeedInfo[]> {
+export async function getEpisodesByCategory(
+  category?: string,
+): Promise<EpisodeWithFeedInfo[]> {
  try {
    let stmt;
    let rows;
@ -1458,7 +1544,10 @@ export async function getEpisodeCategoryStats(): Promise<{
  }
 }

-export async function updateEpisodeCategory(episodeId: string, category: string): Promise<boolean> {
+export async function updateEpisodeCategory(
+  episodeId: string,
+  category: string,
+): Promise<boolean> {
  try {
    const stmt = db.prepare("UPDATE episodes SET category = ? WHERE id = ?");
    const result = stmt.run(category, episodeId);
@ -1519,10 +1608,7 @@ export async function migrateEpisodesWithCategories(): Promise<void> {
        // Add a small delay to avoid rate limiting
        await new Promise((resolve) => setTimeout(resolve, 1000));
      } catch (error) {
-        console.error(
-          `❌ Failed to classify episode ${episode.title}:`,
-          error,
-        );
+        console.error(`❌ Failed to classify episode ${episode.title}:`, error);
        errorCount++;

        // Set a default category for failed classifications
--- a/services/llm.ts
+++ b/services/llm.ts
@ -155,16 +155,17 @@ export async function openAI_ClassifyEpisode(

  // Build the text for classification based on available data
  let textForClassification = `タイトル: ${title}`;
-  
+
  if (description && description.trim()) {
    textForClassification += `\n説明: ${description}`;
  }
-  
+
  if (content && content.trim()) {
    const maxContentLength = 1500;
-    const truncatedContent = content.length > maxContentLength 
-      ? content.substring(0, maxContentLength) + "..." 
-      : content;
+    const truncatedContent =
+      content.length > maxContentLength
+        ? content.substring(0, maxContentLength) + "..."
+        : content;
    textForClassification += `\n内容: ${truncatedContent}`;
  }

--- a/services/podcast.ts
+++ b/services/podcast.ts
@ -3,10 +3,10 @@ import fsSync from "node:fs";
 import path from "node:path";
 import { dirname } from "path";
 import { config } from "./config.js";
-import { 
-  fetchEpisodesWithFeedInfo, 
-  getEpisodesByCategory, 
-  fetchEpisodesByFeedId 
+import {
+  fetchEpisodesByFeedId,
+  fetchEpisodesWithFeedInfo,
+  getEpisodesByCategory,
 } from "./database.js";

 function escapeXml(text: string): string {
@ -86,14 +86,14 @@ function filterValidEpisodes(episodes: any[]): any[] {

 // Generate RSS XML from episodes
 function generateRSSXml(
-  episodes: any[], 
-  title: string, 
-  description: string, 
-  link?: string
+  episodes: any[],
+  title: string,
+  description: string,
+  link?: string,
 ): string {
  const lastBuildDate = new Date().toUTCString();
  const itemsXml = episodes.map(createItemXml).join("\n");
-  
+
  return `<?xml version="1.0" encoding="UTF-8"?>
 <rss version="2.0">
  <channel>
@ -121,9 +121,9 @@ export async function updatePodcastRSS(): Promise<void> {

    const outputPath = path.join(config.paths.publicDir, "podcast.xml");
    const rssXml = generateRSSXml(
-      validEpisodes, 
-      config.podcast.title, 
-      config.podcast.description
+      validEpisodes,
+      config.podcast.title,
+      config.podcast.description,
    );

    // Ensure directory exists
@ -151,7 +151,7 @@ export async function generateCategoryRSS(category: string): Promise<string> {

    const title = `${config.podcast.title} - ${category}`;
    const description = `${config.podcast.description} カテゴリ: ${category}`;
-    
+
    return generateRSSXml(validEpisodes, title, description);
  } catch (error) {
    console.error(`Error generating category RSS for "${category}":`, error);
@ -162,9 +162,15 @@ export async function generateCategoryRSS(category: string): Promise<string> {
 export async function saveCategoryRSS(category: string): Promise<void> {
  try {
    const rssXml = await generateCategoryRSS(category);
-    const safeCategory = category.replace(/[^a-zA-Z0-9\u3040-\u309F\u30A0-\u30FF\u4E00-\u9FAF]/g, "_");
-    const outputPath = path.join(config.paths.publicDir, `podcast_category_${safeCategory}.xml`);
-    
+    const safeCategory = category.replace(
+      /[^a-zA-Z0-9\u3040-\u309F\u30A0-\u30FF\u4E00-\u9FAF]/g,
+      "_",
+    );
+    const outputPath = path.join(
+      config.paths.publicDir,
+      `podcast_category_${safeCategory}.xml`,
+    );
+
    // Ensure directory exists
    await fs.mkdir(dirname(outputPath), { recursive: true });
    await fs.writeFile(outputPath, rssXml);
@ -187,10 +193,11 @@ export async function generateFeedRSS(feedId: string): Promise<string> {
    );

    // Use feed info for RSS metadata if available
-    const feedTitle = validEpisodes.length > 0 ? validEpisodes[0].feedTitle : "Unknown Feed";
+    const feedTitle =
+      validEpisodes.length > 0 ? validEpisodes[0].feedTitle : "Unknown Feed";
    const title = `${config.podcast.title} - ${feedTitle}`;
    const description = `${config.podcast.description} フィード: ${feedTitle}`;
-    
+
    return generateRSSXml(validEpisodes, title, description);
  } catch (error) {
    console.error(`Error generating feed RSS for "${feedId}":`, error);
@ -201,8 +208,11 @@ export async function generateFeedRSS(feedId: string): Promise<string> {
 export async function saveFeedRSS(feedId: string): Promise<void> {
  try {
    const rssXml = await generateFeedRSS(feedId);
-    const outputPath = path.join(config.paths.publicDir, `podcast_feed_${feedId}.xml`);
-    
+    const outputPath = path.join(
+      config.paths.publicDir,
+      `podcast_feed_${feedId}.xml`,
+    );
+
    // Ensure directory exists
    await fs.mkdir(dirname(outputPath), { recursive: true });
    await fs.writeFile(outputPath, rssXml);