result finalization

api/escher_api/escher.py

@@ -1,8 +1,13 @@
 
+from collections import namedtuple
 import flask as flask
 from datetime import datetime
 from common import database
 
+from gevent.pool import Pool
+
+from psycopg2.extras import execute_values
+
 app = flask.Flask('escher')
 
 
@@ -219,11 +224,13 @@ def get_transcript(db_conn, ts_query, start_time="-infinity", end_time="infinity
     p = 0.02
     l = b_p + bus_duration
     a = a_p + n_m
+
     # Lomax distribution is posterior predictive for exponential
     message_duration_diff = l * ((1 - p)**-(1/a) - 1)
+
     current_result = Result()
     results = []
-    print(message_duration_diff)
+    # print(message_duration_diff)
     for transcript_line in db_results:
 
         # Current result set is new
@@ -231,8 +238,8 @@ def get_transcript(db_conn, ts_query, start_time="-infinity", end_time="infinity
             current_result.transcript.append(transcript_line)
         # New message is within window
         elif (transcript_line.start_time - current_result.transcript[-1].end_time).total_seconds() <= message_duration_diff:
-            print((transcript_line.start_time -
-                  current_result.transcript[-1].end_time).total_seconds())
+            # print((transcript_line.start_time -
+            #       current_result.transcript[-1].end_time).total_seconds())
             current_result.transcript.append(transcript_line)
         # New message is outside window
         else:
@@ -356,11 +363,76 @@ def get_chat(db_conn, ts_query, start_time="-infinity", end_time="infinity"):
     return results
 
 
-def load_result_data(result):
-    pass
+def load_results_data(db_conn, results):
+    """
+    Replace chat and transcript with all entries in result's timeframe.
+    """
+    
+    # ggroup = Pool(size=30)
+
+    # results = ggroup.map(lambda result: load_result_data(db_manager, result), results)
+
+    result_timespans = [(i, result.start_time, result.end_time) for (i, result) in enumerate(results)]
+    
+    # Clear lists so we can later insert new lines
+    for result in results:
+        result.chat = []
+        result.transcript = []
+
+
+    cur = db_conn.cursor()
+
+    execute_values(cur,
+        """
+        --sql
+        WITH timespans (id, start_time, end_time) AS (VALUES %s)
+        SELECT 
+            timespans.id,
+	        pub_time,
+	        content->'tags'->>'display-name' AS name,
+	        content->'params'->>1 AS content FROM timespans JOIN chat ON (pub_time BETWEEN start_time AND end_time);
+        """,
+        result_timespans
+    )
+
+    for chat_line in cur:
+        results[chat_line.id].chat.append(chat_line)
+
+    execute_values(cur,
+        """
+        --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;
+        """,
+        result_timespans
+    )
+
+    for transcript_line in cur:
+        results[transcript_line.id].transcript.append(transcript_line)
+
+    return results
 
 
-def merge_results(transcript: list[Result], vst: list[Result], chat: list[Result]):
+def merge_results(transcript: list[Result], vst: list[Result], chat: list[Result], limit: int, offset = 0):
     """
     Merge different types of results in order of importance.
 
@@ -383,12 +455,9 @@ def merge_results(transcript: list[Result], vst: list[Result], chat: list[Result
             else:
                 transcript_i += 1
 
-        # print(vst_result)
         while chat_i < len(chat) and chat[chat_i].start_time < vst_result.end_time:
-            # print(vst_result)
             if overlap(vst_result, chat[chat_i]):
                 vst_result.chat.extend(chat.pop(chat_i).chat)
-                # print(vst_result)
             else:
                 chat_i += 1
 
@@ -405,7 +474,7 @@ def merge_results(transcript: list[Result], vst: list[Result], chat: list[Result
     merged = transcript + vst + chat
     merged.sort(key=lambda result: result.start_time)
     merged.sort(key=lambda result: result.weight, reverse=True)
-    return merged
+    return merged[offset:min((offset + limit), len(merged))]
 
 
 def overlap(result_a, result_b):