v3: add backlinks between dumps (via either dump url or dump target url mentioned through dump description and dump comments)

api/services/backlink-service.ts

@@ -0,0 +1,116 @@
+import { type Dump } from "../model/interfaces.ts";
+import { db, dumpRowToApi, isDumpRow } from "../model/db.ts";
+import { UUID_RE } from "../lib/slugify.ts";
+
+// Mirrors dump-service SELECT_COLS — kept local to avoid a circular import
+// (dump-service.ts must import relinkBacklinks from this file).
+const SELECT_COLS =
+  "id, kind, title, slug, comment, user_id, created_at, updated_at, url, rich_content, file_name, file_mime, file_size, vote_count, is_private, custom_thumbnail_mime," +
+  " (SELECT COUNT(*) FROM comments WHERE dump_id = dumps.id AND deleted = 0) as comment_count";
+
+// Matches http(s) URLs in free text — stops at whitespace or markdown-link
+// delimiters (so `[text](url)` and `<url>` don't swallow the closing char).
+const URL_RE = /https?:\/\/[^\s)\]<>"']+/g;
+
+// A link to another dump's page on this same site (e.g. pasted from the
+// address bar) — host-agnostic, since the configured public URL can differ
+// between dev/prod/reverse-proxy setups. Matches `/dumps/{slug-or-id}`,
+// ignoring any trailing path/query/hash (so `/edit` or `#comment-x` links
+// still resolve to the same target dump).
+const DUMP_PATH_RE = /^\/dumps\/([^/?#]+)/;
+
+function canonicalize(url: string): string | null {
+  try {
+    return new URL(url).toString();
+  } catch {
+    return null;
+  }
+}
+
+/** Resolves a same-site `/dumps/{slug-or-id}` path segment to a dump id. */
+function resolveDumpPathId(idOrSlug: string): string | null {
+  const row = UUID_RE.test(idOrSlug)
+    ? db.prepare(`SELECT id FROM dumps WHERE id = ?;`).get(idOrSlug)
+    : db.prepare(`SELECT id FROM dumps WHERE slug = ?;`).get(idOrSlug);
+  return (row as { id: string } | undefined)?.id ?? null;
+}
+
+/**
+ * Re-derive the backlinks contributed by one text source (a dump's own
+ * `comment` field, or a single comment's `body`) and persist them. Always
+ * deletes the source's previous contribution first, so editing/clearing the
+ * text correctly removes stale links — unlike `linkAttachments`/`notifyMentions`,
+ * which are purely additive.
+ */
+export function relinkBacklinks(
+  sourceType: "dump_comment" | "comment",
+  sourceId: string,
+  fromDumpId: string,
+  text: string,
+): void {
+  const targetIds = new Set<string>();
+  const externalUrls = new Set<string>();
+
+  for (const match of text.matchAll(URL_RE)) {
+    const canonical = canonicalize(match[0]);
+    if (!canonical) continue;
+    let parsed: URL;
+    try {
+      parsed = new URL(canonical);
+    } catch {
+      continue;
+    }
+    const pathMatch = DUMP_PATH_RE.exec(parsed.pathname);
+    if (pathMatch) {
+      const resolved = resolveDumpPathId(decodeURIComponent(pathMatch[1]));
+      if (resolved) targetIds.add(resolved);
+    } else {
+      externalUrls.add(canonical);
+    }
+  }
+
+  if (externalUrls.size > 0) {
+    const placeholders = [...externalUrls].map(() => "?").join(", ");
+    const rows = db.prepare(
+      `SELECT id FROM dumps WHERE url IN (${placeholders});`,
+    ).all(...externalUrls) as { id: string }[];
+    for (const row of rows) targetIds.add(row.id);
+  }
+
+  targetIds.delete(fromDumpId); // no self-links
+
+  db.exec("BEGIN;");
+  try {
+    db.prepare(
+      `DELETE FROM dump_backlinks WHERE source_type = ? AND source_id = ?;`,
+    ).run(sourceType, sourceId);
+    const createdAt = new Date().toISOString();
+    for (const toDumpId of targetIds) {
+      db.prepare(
+        `INSERT INTO dump_backlinks (source_type, source_id, from_dump_id, to_dump_id, created_at)
+         VALUES (?, ?, ?, ?, ?);`,
+      ).run(sourceType, sourceId, fromDumpId, toDumpId, createdAt);
+    }
+    db.exec("COMMIT;");
+  } catch (err) {
+    db.exec("ROLLBACK;");
+    throw err;
+  }
+}
+
+/** Bidirectional, privacy-filtered, capped at 20 — backlinks are a small, bounded list. */
+export function getRelatedDumps(
+  dumpId: string,
+  requestingUserId?: string,
+): Dump[] {
+  const rows = db.prepare(
+    `SELECT ${SELECT_COLS} FROM dumps WHERE id IN (
+       SELECT to_dump_id FROM dump_backlinks WHERE from_dump_id = ?
+       UNION
+       SELECT from_dump_id FROM dump_backlinks WHERE to_dump_id = ?
+     ) AND (is_private = 0 OR user_id = ?)
+     ORDER BY created_at DESC LIMIT 20;`,
+  ).all(dumpId, dumpId, requestingUserId ?? "");
+
+  return rows.filter(isDumpRow).map(dumpRowToApi);
+}