Add season field to episode scripts and improve global search; v2.0.50

Tag scripts with a season number (with an RSS-feed backfill for
existing uploads), surface it in search results, and fix word-boundary
matching plus a title/script-match dedup bug in the search merge logic.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

server/routes/episode-scripts.js

@@ -3,28 +3,63 @@ import { requireAdminAuth } from '../auth.js'
 import { state } from '../state.js'
 import { queueEpisodeScriptsWrite } from '../data.js'
 import { MAX_EPISODE_SCRIPT_LENGTH } from '../config.js'
+import { fetchAllEpisodes } from './episodes.js'
+
+function escapeRegExp(s) {
+  return s.replace(/[.*+?^${}()|[\]\\]/g, '\\$&')
+}
+
+// Word-boundary containment so "sin" doesn't match inside "using"/"resin".
+function containsWord(haystackLower, word) {
+  return new RegExp(`\\b${escapeRegExp(word)}\\b`, 'i').test(haystackLower)
+}
 
 /**
- * Simple whitespace-normalised search over episode script text.
- * Returns { episodeNumber, title, filename, snippet, score } for matches.
+ * Whitespace-normalised search over episode script text, with word-boundary
+ * matching, an exact-phrase score boost, and a "season N" browse mode.
+ * Returns { episodeNumber, title, filename, season, snippet, score } for matches.
  */
 function searchScripts(query) {
   if (!query || query.trim().length < 2) return []
-  const q = query.trim().toLowerCase()
-  const words = q.split(/\s+/).filter(w => w.length >= 2)
+  const q = query.trim()
+  const qLower = q.toLowerCase()
+
+  // "season N" browses every script tagged with that season, no text match needed.
+  const seasonQuery = /^season\s+(\d+)$/i.exec(q)
+  if (seasonQuery) {
+    const season = seasonQuery[1]
+    return Object.entries(state.episodeScripts)
+      .filter(([, entry]) => String(entry.season ?? '') === season)
+      .map(([episodeNumber, entry]) => ({
+        episodeNumber,
+        title: entry.title ?? `Episode ${episodeNumber}`,
+        filename: entry.filename ?? '',
+        season: entry.season ?? null,
+        snippet: '',
+        score: 1,
+      }))
+      .sort((a, b) => Number(a.episodeNumber) - Number(b.episodeNumber))
+      .slice(0, 8)
+  }
+
+  const words = qLower.split(/\s+/).filter(w => w.length >= 2)
+  if (words.length === 0) return []
   const results = []
 
   for (const [episodeNumber, entry] of Object.entries(state.episodeScripts)) {
     if (!entry?.text) continue
     const text = entry.text.toLowerCase()
 
-    // Count how many words appear in the text
-    const matchCount = words.filter(w => text.includes(w)).length
-    if (matchCount === 0) continue
+    const matchedWords = words.filter(w => containsWord(text, w))
+    if (matchedWords.length === 0) continue
+
+    // Exact multi-word phrase matches rank above partial word-overlap matches.
+    const isExactPhrase = words.length > 1 && text.includes(qLower)
+    const score = isExactPhrase ? 1 : matchedWords.length / words.length
 
     // Find a snippet around the first matching word
-    const firstWord = words.find(w => text.includes(w))
-    const idx = firstWord ? text.indexOf(firstWord) : 0
+    const firstWord = matchedWords[0]
+    const idx = text.indexOf(firstWord)
     const start = Math.max(0, idx - 80)
     const end = Math.min(entry.text.length, idx + 180)
     let snippet = entry.text.slice(start, end).replace(/\s+/g, ' ').trim()
@@ -35,8 +70,9 @@ function searchScripts(query) {
       episodeNumber,
       title: entry.title ?? `Episode ${episodeNumber}`,
       filename: entry.filename ?? '',
+      season: entry.season ?? null,
       snippet,
-      score: matchCount / words.length, // 0–1: fraction of query words found
+      score,
     })
   }
 
@@ -57,10 +93,11 @@ export function register(app) {
       episodeNumber,
       title: entry.title ?? `Episode ${episodeNumber}`,
       filename: entry.filename ?? '',
+      season: entry.season ?? null,
       uploadedAt: entry.uploadedAt ?? null,
       wordCount: entry.text ? entry.text.split(/\s+/).filter(Boolean).length : 0,
     }))
-    list.sort((a, b) => Number(a.episodeNumber) - Number(b.episodeNumber))
+    list.sort((a, b) => (Number(a.season) || 0) - (Number(b.season) || 0) || Number(a.episodeNumber) - Number(b.episodeNumber))
     res.json({ scripts: list })
   })
 
@@ -74,6 +111,8 @@ export function register(app) {
     const dataUrl = typeof req.body?.dataUrl === 'string' ? req.body.dataUrl : ''
     const filename = typeof req.body?.filename === 'string' ? req.body.filename.trim() : `episode-${episodeNumber}.docx`
     const title = typeof req.body?.title === 'string' ? req.body.title.trim() : `Episode ${episodeNumber}`
+    const seasonRaw = typeof req.body?.season === 'string' ? req.body.season.trim() : (typeof req.body?.season === 'number' ? String(req.body.season) : '')
+    const season = /^\d+$/.test(seasonRaw) ? seasonRaw : null
 
     if (!dataUrl.startsWith('data:')) {
       res.status(400).json({ message: 'dataUrl must be a valid data URL.' }); return
@@ -108,6 +147,7 @@ export function register(app) {
     state.episodeScripts[episodeNumber] = {
       title,
       filename,
+      season,
       text,
       uploadedAt: new Date().toISOString(),
       wordCount: text.split(/\s+/).filter(Boolean).length,
@@ -117,10 +157,46 @@ export function register(app) {
     res.json({
       ok: true,
       episodeNumber,
+      season,
       wordCount: state.episodeScripts[episodeNumber].wordCount,
     })
   })
 
+  // Admin: fill in season numbers for existing scripts from the podcast RSS feed
+  // (matches by episode number; never overwrites a season that's already set)
+  app.post('/api/admin-episode-scripts/backfill-seasons', requireAdminAuth, async (_req, res) => {
+    let episodes
+    try {
+      episodes = await fetchAllEpisodes()
+    } catch (err) {
+      console.error('[episode-scripts] backfill: RSS fetch failed:', err.message)
+      res.status(502).json({ message: 'Could not fetch the podcast RSS feed to look up seasons.' })
+      return
+    }
+
+    const seasonByEpisodeNumber = new Map()
+    for (const ep of episodes) {
+      if (ep.episode && ep.season) seasonByEpisodeNumber.set(ep.episode, ep.season)
+    }
+
+    let updated = 0
+    const stillMissing = []
+    for (const [episodeNumber, entry] of Object.entries(state.episodeScripts)) {
+      if (entry.season) continue
+      const found = seasonByEpisodeNumber.get(episodeNumber)
+      if (found) {
+        entry.season = found
+        updated += 1
+      } else {
+        stillMissing.push(episodeNumber)
+      }
+    }
+
+    if (updated > 0) queueEpisodeScriptsWrite()
+
+    res.json({ ok: true, updated, stillMissing })
+  })
+
   // Admin: delete a script
   app.delete('/api/admin-episode-scripts/:episodeNumber', requireAdminAuth, (req, res) => {
     const episodeNumber = req.params.episodeNumber?.trim()