archilyzer

commit a9386a445802c085015c5582b86877bd97ccdce2
parent 79446981bf0273369ca05877c001d171ada3bcb5
Author: I Mean I'm Just Saying <imeanimjustsaying@kiwifarms.st>
Date:   Thu, 16 Apr 2026 14:01:47 -0400

Bump MAX_HITS and use larger search window

Diffstat:
M
app/TranscriptSearch.tsx
 | 
78
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++---------

1 file changed, 69 insertions(+), 9 deletions(-)
diff --git a/app/TranscriptSearch.tsx b/app/TranscriptSearch.tsx
@@ -29,14 +29,19 @@ type HitGroup = {
   hits: Hit[];
 };
 
-const MAX_HITS = 500;
+const MAX_HITS = 1000;
 
 export default function TranscriptSearch({
   transcripts,
 }: {
   transcripts: Summary[];
 }) {
-  const { q: query, re: useRegex, v: activeVideo, t: activeTime } = useUrlParams();
+  const {
+    q: query,
+    re: useRegex,
+    v: activeVideo,
+    t: activeTime,
+  } = useUrlParams();
   const [input, setInput] = useState(query);
   useEffect(() => {
     setInput(query);
@@ -116,13 +121,33 @@ export default function TranscriptSearch({
     for (const t of transcripts) {
       const full = fetched[t.slug];
       if (!full?.cues) continue;
+      const cues = full.cues;
       const hits: Hit[] = [];
-      for (const c of full.cues) {
-        if (matches(c.text)) {
-          hits.push({ start: Math.round(c.start), text: c.text });
-          total++;
-          if (total >= MAX_HITS) break;
-        }
+      for (let i = 0; i < cues.length; i++) {
+        const cur = cues[i];
+        const prevText = i > 0 ? cues[i - 1].text : "";
+        const nextText = i < cues.length - 1 ? cues[i + 1].text : "";
+        const sep1 = prevText ? " " : "";
+        const sep2 = nextText ? " " : "";
+        const windowText = prevText + sep1 + cur.text + sep2 + nextText;
+        const curStart = prevText.length + sep1.length;
+        const curEnd = curStart + cur.text.length;
+        const m = findFirstMatchInRange(
+          windowText,
+          curStart,
+          curEnd,
+          trimmed,
+          useRegex,
+          regex.re,
+        );
+        if (!m) continue;
+        const crosses = m.idx < curStart || m.idx + m.length > curEnd;
+        hits.push({
+          start: Math.round(cur.start),
+          text: crosses ? windowText : cur.text,
+        });
+        total++;
+        if (total >= MAX_HITS) break;
       }
       if (hits.length > 0) {
         groups.push({
@@ -138,7 +163,6 @@ export default function TranscriptSearch({
       }
     }
     return { groups, totalHits: total, capped: false };
-    // eslint-disable-next-line react-hooks/exhaustive-deps
   }, [trimmed, transcripts, fetched, summaryBySlug, useRegex, regex.re]);
 
   return (
@@ -336,6 +360,42 @@ function TranscriptGrid({
   );
 }
 
+function findFirstMatchInRange(
+  haystack: string,
+  rangeStart: number,
+  rangeEnd: number,
+  query: string,
+  useRegex: boolean,
+  regex: RegExp | null,
+): { idx: number; length: number } | null {
+  if (useRegex) {
+    if (!regex) return null;
+    const flags = regex.flags.includes("g") ? regex.flags : regex.flags + "g";
+    const re = new RegExp(regex.source, flags);
+    let m: RegExpExecArray | null;
+    while ((m = re.exec(haystack)) !== null) {
+      if (m.index >= rangeStart && m.index < rangeEnd) {
+        return { idx: m.index, length: m[0].length };
+      }
+      if (m.index >= rangeEnd) return null;
+      if (m[0].length === 0) re.lastIndex++;
+    }
+    return null;
+  }
+  if (!query) return null;
+  const lower = haystack.toLowerCase();
+  const ql = query.toLowerCase();
+  let from = 0;
+  while (from <= haystack.length) {
+    const idx = lower.indexOf(ql, from);
+    if (idx === -1) return null;
+    if (idx >= rangeStart && idx < rangeEnd) return { idx, length: ql.length };
+    if (idx >= rangeEnd) return null;
+    from = idx + 1;
+  }
+  return null;
+}
+
 function formatSeconds(s: number): string {
   const h = Math.floor(s / 3600);
   const m = Math.floor((s % 3600) / 60);