archilyzer

commit 32b81b9082dd5e1cee5695b54cd547679582f9d8
parent 20466b29e57eab7e474707981b5cb0e68493ef5f
Author: I Mean I'm Just Saying <imeanimjustsaying@kiwifarms.st>
Date:   Fri,  2 Oct 2026 00:47:21 -0400

common: the live-chat archive reads no raw replay while the media is away

A build re-normalizes a stale cues file from the tiered raw replay only while
the channel's media is ok or in place; otherwise it stages the cues on disk as
they are and logs how many are older than their raw replay.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

Diffstat:
M
common/controller/archiveLiveChat.ts
 | 
48
++++++++++++++++++++++++++++++++++++++++--------

1 file changed, 40 insertions(+), 8 deletions(-)
diff --git a/common/controller/archiveLiveChat.ts b/common/controller/archiveLiveChat.ts
@@ -26,7 +26,8 @@ import {
 } from "node:fs/promises";
 import pLimit from "p-limit";
 import { listChannelStatsFromDisk, type ChannelStat } from "./channels";
-import { normalizeLiveChat } from "./normalizeLiveChat";
+import { isLiveChatCuesFresh, normalizeLiveChat } from "./normalizeLiveChat";
+import { inspectChannelMedia } from "../lib/channelMedia";
 import {
   archiveExtension,
   archiveSidecarPath,
@@ -139,9 +140,22 @@ async function stageChannel(
 
   log(`Archive live chat ${ch.slug}: scanning ${videoIds.length} videos`);
 
+  // THE RAW REPLAY IS MEDIA (release 17): `transcript.live_chat.json` is a
+  // tiered link onto the channel's media drive. A build re-normalizes a stale
+  // cues file from it only while that drive is reachable; otherwise the cues
+  // the corpus disk holds are staged as they are (the freshness check `lstat`s
+  // the link, so asking it costs the drive nothing), and a video with no cues
+  // yet is left out of this build. Never a read of a drive that is unmounted,
+  // stalled or mid-move.
+  const media = await inspectChannelMedia(paths, ch.slug, ch.config, {
+    fresh: true,
+  });
+  const rawReadable = media.status === "ok" || media.status === "in-place";
+
   let normalizedCount = 0;
   let linkedCount = 0;
   let normalizeFailed = 0;
+  let keptStale = 0;
 
   const requiresRewrite = !build.includeMetadata || build.prettyPrint;
 
@@ -157,13 +171,24 @@ async function stageChannel(
         }
         let outcome;
         try {
-          outcome = await normalizeLiveChat({
-            videoDir,
-            channelSlug: ch.slug,
-            configName: ch.config.name,
-            // A build moves no media (release 17): the jobs tier, this reads.
-            tier: false,
-          });
+          if (rawReadable) {
+            outcome = await normalizeLiveChat({
+              videoDir,
+              channelSlug: ch.slug,
+              configName: ch.config.name,
+              // A build moves no media (release 17): the jobs tier, this reads.
+              tier: false,
+            });
+          } else {
+            const { fresh, cuesPath } = await isLiveChatCuesFresh(videoDir);
+            const have = await stat(cuesPath).then(
+              (st) => st.isFile(),
+              () => false,
+            );
+            if (!have) return;
+            if (!fresh) keptStale++;
+            outcome = { status: "fresh" as const, cuesPath };
+          }
         } catch (err) {
           normalizeFailed++;
           log(`  ! normalize ${ch.slug}/${id}: ${(err as Error).message}`);
@@ -189,6 +214,13 @@ async function stageChannel(
   if (normalizeFailed > 0) {
     log(`  ${ch.slug}: ${normalizeFailed} videos failed to normalize`);
   }
+  if (!rawReadable) {
+    log(
+      `  ${ch.slug}: its media is not reachable (${media.detail ?? media.status}) — ` +
+        `staged the live-chat cues on disk as they are` +
+        (keptStale > 0 ? ` (${keptStale} older than their raw replay)` : ""),
+    );
+  }
 
   if (linkedCount > 0) {
     await writeFile(