From 4a9496c60f55ab1e294d8c10346f44762d0e54b6 Mon Sep 17 00:00:00 2001
From: HeNine <>
Date: Fri, 22 Oct 2021 15:47:15 +0200
Subject: [PATCH] Add highlighting

---
 buscribe-api/buscribeapi/buscribeapi.py | 13 +++++++++----
 buscribe-web/line.less                  |  4 ++++
 buscribe_data.sql                       | 12 +++++++++++-
 3 files changed, 24 insertions(+), 5 deletions(-)

diff --git a/buscribe-api/buscribeapi/buscribeapi.py b/buscribe-api/buscribeapi/buscribeapi.py
index 14cd43a..5245287 100644
--- a/buscribe-api/buscribeapi/buscribeapi.py
+++ b/buscribe-api/buscribeapi/buscribeapi.py
@@ -95,16 +95,19 @@ def get_json():
                      "end_bus_time": round_bus_time(row.start_time - app.bustime_start),
                      "verifier": row.verifier,
                      "speakers": row.names,
-                     "text": row.transcription_line if row.transcription_line is not None else ""} for row in results])
+                     "text": row.highlighted_text if row.highlighted_text is not None else ""} for row in results])
 
 
 def fetch_lines(db_conn, start_time, end_time, ts_query=None, limit=None, offset=None):
-    query = "SELECT * FROM buscribe_all_transcriptions WHERE start_time > %(start_time)s AND end_time < %(end_time)s "
+    query = "SELECT *" + \
+            (
+                ",ts_headline(transcription_line, convert_query(%(text_query)s), 'StartSel=''<span class=\"highlight\">'', StopSel=</span>') AS highlighted_text" if ts_query is not None else ",transcription_line AS highlighted_text") + \
+            " FROM buscribe_all_transcriptions WHERE start_time > %(start_time)s AND end_time < %(end_time)s "
 
     if ts_query is not None:
         query += "AND (coalesce(transcription_line_ts, ''::tsvector) || coalesce(names_ts, ''::tsvector)) @@ " \
-                 "(CASE WHEN websearch_to_tsquery(%(text_query)s)::text != '' THEN websearch_to_tsquery(%(text_query)s)::text || ':*' ELSE '' END)::tsquery " \
-                 "ORDER BY ts_rank_cd(coalesce(transcription_line_ts, ''::tsvector) || coalesce(names_ts, ''::tsvector), (CASE WHEN websearch_to_tsquery(%(text_query)s)::text != '' THEN websearch_to_tsquery(%(text_query)s)::text || ':*' ELSE '' END)::tsquery) DESC, " \
+                 "convert_query(%(text_query)s) " \
+                 "ORDER BY ts_rank_cd(coalesce(transcription_line_ts, ''::tsvector) || coalesce(names_ts, ''::tsvector), convert_query(%(text_query)s)) DESC, " \
                  "start_time "
     else:
         query += "ORDER BY start_time "
@@ -117,6 +120,8 @@ def fetch_lines(db_conn, start_time, end_time, ts_query=None, limit=None, offset
 
     query += ";"
 
+    print(query)
+
     return database.query(db_conn, query,
                           start_time=start_time if start_time is not None else '-infinity',
                           end_time=end_time if end_time is not None else 'infinity',
diff --git a/buscribe-web/line.less b/buscribe-web/line.less
index e5f0c8b..c86646d 100644
--- a/buscribe-web/line.less
+++ b/buscribe-web/line.less
@@ -44,6 +44,10 @@
 
     grid-column: text;
     //grid-row: span 2;
+    span.highlight {
+      font-weight: bold;
+      color: powderblue;
+    }
   }
 
   .line_speakers {
diff --git a/buscribe_data.sql b/buscribe_data.sql
index 497d273..b06607f 100644
--- a/buscribe_data.sql
+++ b/buscribe_data.sql
@@ -111,4 +111,14 @@ SELECT id,
        null                                       AS names_ts
 FROM buscribe_transcriptions;
 
-ROLLBACK;
\ No newline at end of file
+ROLLBACK;
+
+-- Convert last lexeme in a query to prefix query.
+CREATE FUNCTION convert_query(query_text text) RETURNS tsquery AS
+$$
+DECLARE
+    ws_query text := websearch_to_tsquery(query_text)::text;
+BEGIN
+    RETURN (CASE WHEN ws_query != '' THEN ws_query || ':*' ELSE '' END)::tsquery;
+END;
+$$ LANGUAGE plpgsql;
\ No newline at end of file