bibclean: an abstract that arrived as an index

parity/oracle/abstracts.py

@@ -0,0 +1,44 @@
+"""Turn an inverted abstract back into prose the way the policy pipeline did.
+
+The function below is the original, copied out of the journals module verbatim
+— importing it opens a database connection and a catalogue client, and neither
+decides anything.
+
+    python3 oracle/abstracts.py < cases.ndjson
+
+One case per line: {name, index}. Writes one JSON answer per line.
+"""
+
+from __future__ import annotations
+
+import json
+import sys
+
+
+def _decode_abstract(idx: dict | None) -> str | None:
+    if not isinstance(idx, dict) or not idx:
+        return None
+    positions: list[tuple[int, str]] = []
+    for word, locs in idx.items():
+        if not isinstance(locs, list):
+            continue
+        for p in locs:
+            if isinstance(p, int):
+                positions.append((p, str(word)))
+    if not positions:
+        return None
+    positions.sort(key=lambda t: t[0])
+    return " ".join(w for _, w in positions)
+
+
+for line in sys.stdin:
+    if not line.strip():
+        continue
+    case = json.loads(line)
+    sys.stdout.write(
+        json.dumps(
+            {"name": case["name"], "abstract": _decode_abstract(case["index"])},
+            sort_keys=True,
+        )
+        + "\n"
+    )