result construction optimization

api/escher_api/escher.py

@@ -1,5 +1,6 @@
 
 from collections import namedtuple
+from unicodedata import name
 import flask as flask
 from datetime import datetime
 from common import database
@@ -362,7 +363,6 @@ def get_chat(db_conn, ts_query, start_time="-infinity", end_time="infinity"):
 
     return results
 
-
 def load_results_data(db_conn, results):
     """
     Replace chat and transcript with all entries in result's timeframe.
@@ -402,32 +402,35 @@ def load_results_data(db_conn, results):
         """
         --sql
         WITH timespans (id, start_time, end_time) AS (VALUES %s)
-        SELECT 
-            timespans.id,
-            buscribe_transcriptions.start_time,
-            buscribe_transcriptions.end_time,
-            names,
-            buscribe_transcriptions.transcription_line
-        FROM timespans JOIN
-            buscribe_transcriptions ON (
-                buscribe_transcriptions.start_time >= timespans.start_time AND buscribe_transcriptions.start_time <= timespans.end_time AND
-                buscribe_transcriptions.end_time >= timespans.start_time AND buscribe_transcriptions.end_time <= timespans.end_time
-                )
-            LEFT OUTER JOIN (
-                SELECT line,
-                    ARRAY(
-                        SELECT speaker_name
-                        FROM buscribe_line_inferred_speakers AS inner_speakers
-                        WHERE inner_speakers.line = buscribe_line_inferred_speakers.line
-                    ) AS names
-                FROM buscribe_line_inferred_speakers
-            ) AS inferred_speakers ON buscribe_transcriptions.id = inferred_speakers.line;
+        SELECT timespans.id,
+            ARRAY(
+                SELECT row_to_json(transcriptions)
+                FROM (
+                        SELECT buscribe_transcriptions.start_time,
+                            buscribe_transcriptions.end_time,
+                            ARRAY(
+                                SELECT speaker_name
+                                FROM buscribe_line_inferred_speakers
+                                WHERE buscribe_line_inferred_speakers.line = buscribe_transcriptions.id
+                            ) as names,
+                            buscribe_transcriptions.transcription_line
+                        FROM buscribe_transcriptions
+                        WHERE buscribe_transcriptions.start_time >= timespans.start_time
+                            AND buscribe_transcriptions.start_time <= timespans.end_time
+                            AND buscribe_transcriptions.end_time >= timespans.start_time
+                            AND buscribe_transcriptions.end_time <= timespans.end_time
+                    ) AS transcriptions
+            ) AS transcriptions
+        FROM timespans;
         """,
         result_timespans
     )
 
-    for transcript_line in cur:
-        results[transcript_line.id].transcript.append(transcript_line)
+    TranscriptRecord = namedtuple("TranscriptRecord", cur.fetchone().transcriptions[0].keys())
+    cur.scroll(-1)
+
+    for result in cur:
+        results[result.id].transcript.extend([TranscriptRecord(**transcription) for transcription in result.transcriptions])
 
     return results