Rough cut compatible captions

buscribe-api/buscribeapi/buscribeapi.py

@@ -2,6 +2,7 @@ import json
 from datetime import timedelta
 
 import flask as flask
+import common
 from common import dateutil, database
 from dateutil.parser import ParserError
 from flask import request, jsonify, Response, render_template
@@ -11,9 +12,16 @@ app = flask.Flask('buscribe')
 
 @app.template_filter()
 def convert_vtt_timedelta(delta: timedelta):
+    """Converts a timedelta to a VTT compatible format."""
     return f'{delta.days * 24 + delta.seconds // 3600:02}:{(delta.seconds % 3600) // 60:02}:{delta.seconds % 60:02}.{delta.microseconds // 1000:03}'
 
 
+@app.template_filter()
+def create_seconds_timedelta(seconds):
+    """Converts a float of seconds to a timedelta."""
+    return timedelta(seconds=seconds)
+
+
 def round_bus_time(delta: timedelta):
     """Round bus time down to the second."""
     return f'{delta.days * 24 + delta.seconds // 3600:02}:{(delta.seconds % 3600) // 60:02}:{delta.seconds % 60:02}'
@@ -44,10 +52,15 @@ def get_vtt():
 
     db_conn = app.db_manager.get_conn()
 
+    segments = common.get_best_segments(app.segments_dir,
+                                        start_time,
+                                        end_time)
+    segments_start_time = segments[0].start
+
     results = fetch_lines(db_conn, start_time, end_time)
 
     return Response(
-        render_template("busubs.jinja", results=results, bustime_start=app.bustime_start,
+        render_template("busubs.jinja", results=results, start_time=segments_start_time,
                         duration_extend=timedelta(seconds=0.3)),
         mimetype="text/vtt"
     )
@@ -102,7 +115,7 @@ def fetch_lines(db_conn, start_time, end_time, ts_query=None, limit=None, offset
     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 "
+            " 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)) @@ " \

buscribe-api/buscribeapi/main.py

@@ -47,7 +47,9 @@ def servelet(server):
                'postgresql://USER:PASSWORD@HOST/DBNAME?KEY=VALUE')
 @argh.arg('--bustime-start',
           help='The start time in UTC for the event, for UTC-Bustime conversion')
-def main(database="", host='0.0.0.0', port=8010, bustime_start=None):
+@argh.arg('--base-dir',
+          help='Directory from which segments will be grabbed. Default is current working directory.')
+def main(database="", host='0.0.0.0', port=8010, bustime_start=None, base_dir=None):
     if bustime_start is None:
         logging.error("Missing --bustime-start!")
         exit(1)
@@ -60,6 +62,8 @@ def main(database="", host='0.0.0.0', port=8010, bustime_start=None):
         logging.error("Invalid --bustime-start!")
         exit(1)
 
+    app.segments_dir = base_dir
+
     app.db_manager = DBManager(dsn=database)
 
     stopping = gevent.event.Event()

buscribe-api/templates/busub.jinja

@@ -1,2 +1,10 @@
-{{ (row.start_time - bustime_start - duration_extend)|convert_vtt_timedelta }} --> {{ (row.end_time - bustime_start + duration_extend)|convert_vtt_timedelta }}
-- {{ row.transcription_line }}
+{{ (row.start_time - start_time)|convert_vtt_timedelta }} --> {{ (row.end_time - start_time + duration_extend)|convert_vtt_timedelta }}
+<c.{%- if row.verifier is not none -%}verified_cc{%- else -%}machine_cc{%- endif -%}>
+    {%- if row.transcription_json is none -%}
+        {{ row.transcription_line }}
+    {%- else -%}
+        {%- set line_start_time = row.transcription_json["result"][0]["start"] -%}
+        {%- for word in row.transcription_json["result"] -%}
+             <{{ ((row.start_time - start_time) + (word["start"] - line_start_time)|create_seconds_timedelta)|convert_vtt_timedelta }}>{{ word["word"] + " " }}
+        {%- endfor -%}
+    {%- endif -%}</c>

buscribe-api/templates/busubs.jinja

@@ -1,5 +1,7 @@
 WEBVTT
 
 {% for row in results %}
-{% include "busub.jinja" %}
+    {%- if row.transcription_line is not none %}
+        {%- include "busub.jinja" %}
+    {%- endif %}
 {% endfor %}

buscribe_data.sql

@@ -87,7 +87,8 @@ SELECT buscribe_transcriptions.id,
        names,
        verified_line                                                 AS transcription_line,
        setweight(to_tsvector('english', verified_line), 'C')         AS transcription_line_ts,
-       setweight(to_tsvector(array_to_string(names, ' ')), 'C')      AS names_ts
+       setweight(to_tsvector(array_to_string(names, ' ')), 'C')      AS names_ts,
+       null                                                          AS transcription_json
 FROM buscribe_transcriptions
          LEFT OUTER JOIN buscribe_verified_lines ON buscribe_transcriptions.id = buscribe_verified_lines.line
          LEFT OUTER JOIN (
@@ -108,7 +109,8 @@ SELECT id,
        null                                       AS names,
        transcription_line,
        to_tsvector('english', transcription_line) AS transcription_line_ts,
-       null                                       AS names_ts
+       null                                       AS names_ts,
+       transcription_json
 FROM buscribe_transcriptions;
 
 ROLLBACK;

professor/script.js

@@ -8,7 +8,7 @@ function pageReady() {
         controls: true,
         autoplay: false,
         width: 900,
-        height: 420,
+        height: 900 / 16 * 9,
         playbackRates: [0.5, 1, 1.25, 1.5, 2],
         inactivityTimeout: 0,
         controlBar: {
@@ -16,24 +16,46 @@ function pageReady() {
             volumePanel: {
                 inline: false,
             },
-        },
-        sources: [{src: `//localhost:8005/professor/line/${line_id}/playlist.m3u8`}]
+        }
     });
 
+    // this changes the background color to red
+    const bgColorSelector = document.querySelector('.vjs-bg-color > select');
+    bgColorSelector.value = "#000";
+
+    // this changes the background opacity to 0.5
+    const bgOpacitySelector = document.querySelector('.vjs-bg-opacity > select');
+    bgOpacitySelector.value = "0.5"
+
     fetch(`//localhost:8005/professor/line/${line_id}`)
         .then(response => response.json())
-        .then(fillLineInfo);
+        .then(fillLineInfo)
+        .then(initializePlayer);
 
 }
 
 function fillLineInfo(line_json) {
     // document.getElementById("original_transcription").innerText = line_json.line_data.text;
+    line = line_json
     document.getElementById("original_transcription").innerHTML = line_json.line_data.result
         .map(word => `<span style="opacity: ${word.conf}">${word.word}</span>`).join(" ");
     document.getElementById("new_transcription")
         .attributes.getNamedItem("placeholder").value = line_json.line_data.text;
 }
 
+function initializePlayer() {
+    videojs.getPlayer("player").src([
+        {src: `//localhost:8005/professor/line/${line_id}/playlist.m3u8`}
+    ]);
+    videojs.getPlayer("player").addRemoteTextTrack({
+        kind: "captions",
+        src: `//localhost:8010/buscribe/vtt?start_time=${line.start_time}&end_time=${line.end_time}`,
+        srclang: "en",
+        label: "English",
+        default: true
+    }, false);
+}
+
 async function submit() {
     document.getElementById("update_indicator").innerText = "⭯"
 

professor/style.less

@@ -56,4 +56,8 @@ button {
   height: 1.3em;
   margin-left: 1em;
   vertical-align: middle;
-}
+}
+
+span.verified_cc {
+  color: #c1ffc1;
+}