search for standard sites
pub-search.waow.tech
search
zig
blog
atproto
fix: backfill base_path when publication arrives after document
when a document arrives on the firehose before its publication,
it gets indexed with an empty base_path. this adds a reactive
backfill in insertPublication() that patches any such documents,
bumps indexed_at for sync, and clears embedded_at for re-embedding.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
+16
+16
backend/src/ingest/indexer.zig
+16
backend/src/ingest/indexer.zig
···
244
244
&.{ uri, did, rkey, name, description orelse "", base_path orelse "" },
245
245
);
246
246
247
+
// backfill: if documents arrived before this publication, they have empty base_path
248
+
if (base_path) |bp| {
249
+
if (bp.len > 0) {
250
+
c.exec(
251
+
\\UPDATE documents SET
252
+
\\ base_path = ?,
253
+
\\ indexed_at = strftime('%Y-%m-%dT%H:%M:%S', 'now'),
254
+
\\ embedded_at = NULL
255
+
\\WHERE publication_uri = ?
256
+
\\ AND (base_path IS NULL OR base_path = '')
257
+
, &.{ bp, uri }) catch |err| {
258
+
logfire.warn("indexer: base_path backfill failed for pub {s}: {}", .{ uri, err });
259
+
};
260
+
}
261
+
}
262
+
247
263
// update FTS index (includes base_path for subdomain search)
248
264
c.exec("DELETE FROM publications_fts WHERE uri = ?", &.{uri}) catch {};
249
265
c.exec(