search
Giò Diani
parent
f6b73be826
commit
44df18fcfb
|
|
@ -1 +0,0 @@
|
|||
*,*:before,*:after{box-sizing:border-box}:root{--c1: #F2F2F2;--c2: #D9D9D9;--c3: #737373;--c4: #404040;--c5: #0D0D0D}html{-moz-text-size-adjust:none;-webkit-text-size-adjust:none;-webkit-font-smoothing:antialiased;text-size-adjust:none}*{margin:0}ul[role=list],ol[role=list]{list-style:none}body{min-height:100vh;line-height:1.5;padding:1.5em;font-family:sans-serif;background:var(--c4);color:var(--c1)}img,picture,video,canvas,svg{display:block;max-width:100%}input,button,textarea,select{font:inherit}input{margin-top:.2em;border-radius:.2em;border:1px solid #fff;background:var(--c3);color:#fff;padding:.2em .5em;display:block}h1,h2,h3,h4,button,input,label{line-height:1.1}fieldset{display:block;border:none;padding:0;margin:0}h1,h2,h3,h4{text-wrap:balance}a:not([class]){text-decoration-skip-ink:auto;color:currentColor}input,button,textarea,select{font-family:inherit;font-size:inherit}textarea:not([rows]){min-height:10em}:target{scroll-margin-block:5ex}dl{display:grid;grid-template-areas:"title desc";gap:.5em 1em}dt{font-weight:600}.visually-hidden{position:absolute;left:-10000px;top:auto;width:1px;height:1px;overflow:hidden}.tablist{height:100%;display:grid;grid-template:"tablist" "panel";grid-template-rows:2.5em 1fr}.controls{background:var(--c5)}.control{display:inline-block}.control label{display:inline-grid;justify-items:center;align-items:center;padding:.5em 1em;height:2.5em;color:var(--c1)}input[type=radio]:checked+label{background:var(--c3)}a:focus,input:focus,input[type=radio]:focus+label{outline:2px dotted;outline-offset:2px}input[type=radio]+label:hover{cursor:pointer;text-decoration:underline}.panel{background:var(--c3);padding:1em;height:calc(50vh - 4em);overflow-y:scroll}#tablist_panel_transcript_panel font{display:block;color:#fff!important}#transcript{padding:0;list-style:none}#transcript a{cursor:pointer}#transcript li{display:flex}#transcript li>div{margin-right:1em}.vjs-title-bar{display:none!important}header{position:fixed;top:0;left:0;width:100%;background:#000;z-index:99;height:2.5em;display:flex;align-items:center;padding:0 2em}header a{text-decoration:none}header a:before{content:"←";padding:0 .5em 0 0}main{width:100%;margin-top:1.5em;height:calc(100vh - 6.5em);display:grid;grid-template-areas:"video tabs" "timeline timeline";grid-template-columns:repeat(2,1fr);grid-template-rows:1fr 50%;gap:1em}ol{padding-left:1em}#video{grid-area:video}#tabs{grid-area:tabs}#timeline{grid-area:timeline;position:relative;background:var(--c3)}#timeline-bar{height:2em;margin-left:20em;background:#aaa;display:flex;justify-content:space-between}#timeline-bar>span{display:inline-flex;align-items:center}#timeline-bar-ctrl{display:inline-block;background:#000;width:.1875em;height:100%;position:absolute;z-index:999}#timeline-bar-ctrl:before{content:"";width:2em;height:2em;border-radius:50%;transform:translate3d(calc(-50% + .09375em),0,0);position:absolute;background:#000}#tracks{height:calc(100% - 2em);overflow-y:scroll}.track{width:100%;min-height:15em;display:grid;grid-template:100% / 20em 1fr}.track+.track{border-top:1px solid var(--c1)}.track-ctrl{padding:1em;background:var(--c3)}.track-ctrl h2{font-size:1.2em}.track-ctrl p{margin-top:1em;font-size:1em;line-height:1.5}.track-ctrl label{margin-top:1em;display:inline-block}.track-viz{background:var(--c3)}ul.segments{padding:0;grid-template:1fr / repeat(4,1fr);list-style:none;display:grid;grid-template:;height:100%}ul.segments li{background:#ccc;color:#000;border-left:.0625em solid #fff;padding:1em}.track-viz svg{width:100%!important}#sentiments-list{list-style:none;padding:0}#sentiments-list li{margin-top:.5em}#sentiments-list span{display:inline-block;background:#d01c8b;width:1.5em;height:1.5em;border-radius:50%;text-align:center}#sentiments-list span.pos{background:#4dac26}#sentiments-list span.neut{background:#f7f7f7;color:#000}
|
||||
File diff suppressed because one or more lines are too long
File diff suppressed because one or more lines are too long
|
|
@ -0,0 +1 @@
|
|||
*,*:before,*:after{box-sizing:border-box}:root{--c1: #F2F2F2;--c2: #D9D9D9;--c3: #737373;--c4: #404040;--c5: #0D0D0D}html{-moz-text-size-adjust:none;-webkit-text-size-adjust:none;-webkit-font-smoothing:antialiased;text-size-adjust:none}*{margin:0}ul[role=list],ol[role=list]{list-style:none}body{min-height:100vh;line-height:1.5;padding:1.5em;font-family:sans-serif;background:var(--c4);color:var(--c1)}img,picture,video,canvas,svg{display:block;max-width:100%}input,button,textarea,select{font:inherit}button,input{margin-top:.2em;border-radius:.2em;border:1px solid #fff;background:var(--c3);color:#fff;padding:.2em .5em;display:block}.searchform button{border-radius:0 .2em .2em 0}button:hover{background:#fff;color:#000}.searchform input{border-radius:.2em 0 0 .2em;border-right:0}h1,h2,h3,h4,button,input,label{line-height:1.1}fieldset{display:block;border:none;padding:0;margin:0}h1,h2,h3,h4{text-wrap:balance}a:not([class]){text-decoration-skip-ink:auto;color:currentColor}input,button,textarea,select{font-family:inherit;font-size:inherit}textarea:not([rows]){min-height:10em}:target{scroll-margin-block:5ex}dl{display:grid;grid-template-areas:"title desc";gap:.5em 1em}dt{font-weight:600}.visually-hidden{position:absolute;left:-10000px;top:auto;width:1px;height:1px;overflow:hidden}.tablist{height:100%;display:grid;grid-template:"tablist" "panel";grid-template-rows:2.5em 1fr}.controls{background:var(--c5)}.control{display:inline-block}.control label{display:inline-grid;justify-items:center;align-items:center;padding:.5em 1em;height:2.5em;color:var(--c1)}input[type=radio]:checked+label{background:var(--c3)}a:focus,input:focus,input[type=radio]:focus+label{outline:2px dotted;outline-offset:2px}input[type=radio]+label:hover{cursor:pointer;text-decoration:underline}.panel{background:var(--c3);padding:1em;height:calc(50vh - 4em);overflow-y:scroll}#tablist_panel_transcript_panel font,#tablist_panel_search_panel font{display:block;color:#fff!important}.searchform{display:flex}#searchresults{margin-top:1em}#searchresults,#transcript{padding:0;list-style:none}#searchresults a,#transcript a{cursor:pointer}#searchresults li,#transcript li{display:flex}#searchresults li>div,#transcript li>div{margin-right:1em}.vjs-title-bar{display:none!important}#topicslist a{background:#fff;color:#000;cursor:pointer;padding:.2em;border-radius:.3em;display:inline-block;margin-top:.3em}header{position:fixed;top:0;left:0;width:100%;background:#000;z-index:99;height:2.5em;display:flex;align-items:center;padding:0 2em}header a{text-decoration:none}header a:before{content:"←";padding:0 .5em 0 0}main{width:100%;margin-top:1.5em;height:calc(100vh - 6.5em);display:grid;grid-template-areas:"video tabs" "timeline timeline";grid-template-columns:repeat(2,1fr);grid-template-rows:1fr 50%;gap:1em}ol{padding-left:1em}#video{grid-area:video}#tabs{grid-area:tabs}#timeline{grid-area:timeline;position:relative;background:var(--c3)}#timeline-bar{height:2em;margin-left:20em;background:#aaa;display:flex;justify-content:space-between}#timeline-bar>span{display:inline-flex;align-items:center}#timeline-bar-ctrl{display:inline-block;background:#000;width:.1875em;height:100%;position:absolute;z-index:999}#timeline-bar-ctrl:before{content:"";width:2em;height:2em;border-radius:50%;transform:translate3d(calc(-50% + .09375em),0,0);position:absolute;background:#000}#tracks{height:calc(100% - 2em);overflow-y:scroll}.track{width:100%;min-height:15em;display:grid;grid-template:100% / 20em 1fr}.track+.track{border-top:1px solid var(--c1)}.track-ctrl{padding:1em;background:var(--c3)}.track-ctrl h2{font-size:1.2em}.track-ctrl p{margin-top:1em;font-size:1em;line-height:1.5}.track-ctrl label{margin-top:1em;display:inline-block}.track-viz{background:var(--c3)}ul.segments{padding:0;grid-template:1fr / repeat(4,1fr);list-style:none;display:grid;grid-template:;height:100%}ul.segments li{background:#ccc;color:#000;border-left:.0625em solid #fff;padding:1em}.track-viz svg{width:100%!important}#sentiments-list{list-style:none;padding:0}#sentiments-list li{margin-top:.5em}#sentiments-list span{display:inline-block;background:#d01c8b;width:1.5em;height:1.5em;border-radius:50%;text-align:center}#sentiments-list span.pos{background:#4dac26}#sentiments-list span.neut{background:#f7f7f7;color:#000}
|
||||
|
|
@ -5,12 +5,12 @@
|
|||
"isEntry": true
|
||||
},
|
||||
"resources/css/app.css": {
|
||||
"file": "assets/app-BDv93WNO.css",
|
||||
"file": "assets/app-DQLXVmOM.css",
|
||||
"src": "resources/css/app.css",
|
||||
"isEntry": true
|
||||
},
|
||||
"resources/js/app.js": {
|
||||
"file": "assets/app-CrWEnB8o.js",
|
||||
"file": "assets/app-DGnSPD5n.js",
|
||||
"name": "app",
|
||||
"src": "resources/js/app.js",
|
||||
"isEntry": true
|
||||
|
|
|
|||
|
|
@ -46,7 +46,7 @@ img, picture, video, canvas, svg {
|
|||
input, button, textarea, select {
|
||||
font: inherit;
|
||||
}
|
||||
|
||||
button,
|
||||
input{
|
||||
margin-top: .2em;
|
||||
border-radius: .2em;
|
||||
|
|
@ -57,6 +57,21 @@ input{
|
|||
display: block;
|
||||
}
|
||||
|
||||
.searchform button{
|
||||
border-radius: 0 .2em .2em 0;
|
||||
}
|
||||
|
||||
button:hover{
|
||||
background: #fff;
|
||||
color: #000;
|
||||
}
|
||||
|
||||
.searchform input{
|
||||
border-radius: .2em 0 0 .2em;
|
||||
border-right: 0;
|
||||
}
|
||||
|
||||
|
||||
|
||||
h1,
|
||||
h2,
|
||||
|
|
@ -175,24 +190,37 @@ input[type="radio"] + label:hover {
|
|||
overflow-y: scroll;
|
||||
}
|
||||
|
||||
#tablist_panel_transcript_panel font{
|
||||
#tablist_panel_transcript_panel font,
|
||||
#tablist_panel_search_panel font{
|
||||
display: block;
|
||||
color: #fff !important
|
||||
}
|
||||
|
||||
.searchform{
|
||||
display: flex;
|
||||
}
|
||||
|
||||
#searchresults{
|
||||
margin-top: 1em;
|
||||
}
|
||||
|
||||
#searchresults,
|
||||
#transcript{
|
||||
padding: 0;
|
||||
list-style: none;
|
||||
}
|
||||
|
||||
#searchresults a,
|
||||
#transcript a{
|
||||
cursor: pointer;
|
||||
}
|
||||
|
||||
#searchresults li,
|
||||
#transcript li{
|
||||
display: flex;
|
||||
}
|
||||
|
||||
#searchresults li>div,
|
||||
#transcript li>div{
|
||||
margin-right: 1em;
|
||||
}
|
||||
|
|
@ -201,6 +229,15 @@ input[type="radio"] + label:hover {
|
|||
display: none !important;
|
||||
}
|
||||
|
||||
#topicslist a{
|
||||
background: #fff;
|
||||
color: #000;
|
||||
cursor: pointer;
|
||||
padding: .2em;
|
||||
border-radius: .3em;
|
||||
display: inline-block;
|
||||
margin-top: .3em;
|
||||
}
|
||||
|
||||
/* Layout */
|
||||
header{
|
||||
|
|
@ -374,3 +411,5 @@ ul.segments li{
|
|||
background: #f7f7f7;
|
||||
color: #000;
|
||||
}
|
||||
|
||||
|
||||
|
|
|
|||
|
|
@ -6,10 +6,12 @@ import { gsap } from "gsap";
|
|||
import { Draggable } from "gsap/Draggable";
|
||||
import * as echarts from 'echarts';
|
||||
|
||||
window.echarts = echarts;
|
||||
|
||||
// Tabs
|
||||
function changeTab(el){
|
||||
|
||||
let tabId = el.target.getAttribute('id');
|
||||
let tabId = el.getAttribute('id');
|
||||
let panels = document.querySelectorAll('.panel');
|
||||
|
||||
panels.forEach(p => {
|
||||
|
|
@ -24,7 +26,7 @@ function changeTab(el){
|
|||
const tabs = document.querySelectorAll('input[type="radio"]');
|
||||
tabs.forEach(tab => {
|
||||
tab.addEventListener('click', e => {
|
||||
changeTab(e);
|
||||
changeTab(e.target);
|
||||
})
|
||||
})
|
||||
|
||||
|
|
@ -47,8 +49,6 @@ function setWordCountChart(){
|
|||
// let max = totalWordCount / count;
|
||||
let data = [];
|
||||
|
||||
|
||||
|
||||
let wordWordsLen = wordWords.length;
|
||||
let chunks = Math.floor(wordWordsLen / count);
|
||||
|
||||
|
|
@ -119,14 +119,26 @@ function setWordCountChart(){
|
|||
}
|
||||
|
||||
const wordCountChartOptions = {
|
||||
tooltip: {
|
||||
trigger: 'axis',
|
||||
valueFormatter: (value) => 'Gesprochene Worte ' + new Intl.NumberFormat('de-CH').format(value)
|
||||
},
|
||||
grid: grid,
|
||||
xAxis: xAxis,
|
||||
yAxis: yAxis,
|
||||
series: series
|
||||
series: series,
|
||||
tooltip: {
|
||||
trigger: 'axis',
|
||||
valueFormatter: (value) => 'Gesprochene Worte ' + new Intl.NumberFormat('de-CH').format(value),
|
||||
formatter: (v) => {
|
||||
|
||||
let index = v[0].componentIndex;
|
||||
//console.log(wordCountChartOptions.series[0]);
|
||||
let count_tooltip = 0;
|
||||
for(let i = 0; i < index; i++){
|
||||
count_tooltip = count_tooltip + parseInt(wordCountChartOptions.series[i].data.slice(-1));
|
||||
}
|
||||
count_tooltip = parseInt(v[0].axisValue) + count_tooltip;
|
||||
return `Segment: ${index + 1}<br>Wort: ${count_tooltip}`;
|
||||
|
||||
}
|
||||
},
|
||||
}
|
||||
|
||||
|
||||
|
|
@ -152,7 +164,12 @@ const chartSentimentsOptions = {
|
|||
tooltip: {
|
||||
trigger: 'axis',
|
||||
position: "top",
|
||||
formatter: 'Sekunde: {b0}<br>Sentiment: {c0}'
|
||||
formatter: (e) => {
|
||||
let sek = e[0].axisValue;
|
||||
let sent = e[0].data;
|
||||
sent = sent == 0 ? 'Neutral' : sent == 1 ? 'Positiv' : 'Negativ';
|
||||
return `Sekunde: ${sek}<br>Sentiment: ${sent}`
|
||||
}
|
||||
},
|
||||
grid: {
|
||||
show: false,
|
||||
|
|
@ -191,6 +208,7 @@ const chartSentimentsOptions = {
|
|||
|
||||
chartSentiments.setOption(chartSentimentsOptions);
|
||||
|
||||
|
||||
// VIDEO
|
||||
|
||||
|
||||
|
|
@ -201,7 +219,7 @@ gsap.registerPlugin(Draggable);
|
|||
|
||||
const player = new Pillarbox('my-player', {
|
||||
controls: false,
|
||||
muted: true,
|
||||
muted: false,
|
||||
srgOptions: {
|
||||
liveui: false
|
||||
}
|
||||
|
|
@ -288,3 +306,62 @@ document.addEventListener("DOMContentLoaded", (event) => {
|
|||
},
|
||||
});
|
||||
});
|
||||
|
||||
chartSentiments.on('click', 'series', (e) => {
|
||||
player.currentTime(chartSentimentsOptions.xAxis.data[e.dataIndex]);
|
||||
player.play();
|
||||
});
|
||||
|
||||
// Suche
|
||||
const searchInput = document.querySelector('#search')
|
||||
const searchInputBtn = document.querySelector('#searchbutton')
|
||||
const searchresults = document.querySelector('#searchresults')
|
||||
|
||||
function search(term){
|
||||
|
||||
if(term.length > 1){
|
||||
|
||||
searchresults.innerHTML = '';
|
||||
|
||||
transcript.querySelectorAll('li').forEach((el)=>{
|
||||
let text = el.querySelector('font').innerText.toLowerCase();
|
||||
term = term.toLowerCase();
|
||||
|
||||
if(text.includes(term)){
|
||||
let clone = el.cloneNode(true);
|
||||
searchresults.appendChild(clone);
|
||||
}
|
||||
})
|
||||
|
||||
if(!searchresults.querySelector('li')){
|
||||
searchresults.innerHTML = `<li>Keine Ergebnisse für «${term}»</li>`;
|
||||
}else{
|
||||
searchresults.querySelectorAll('[data-start]').forEach(c => {
|
||||
c.addEventListener('click', (e) => {
|
||||
player.currentTime(e.target.getAttribute('data-start'));
|
||||
player.play();
|
||||
});
|
||||
});
|
||||
}
|
||||
}else{
|
||||
searchresults.innerHTML = `<li>Bitte gib mind. 2 Buchstaben ein.</li>`;
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
searchInputBtn.addEventListener('click', () => {
|
||||
search(searchInput.value)
|
||||
})
|
||||
|
||||
const topics = document.querySelectorAll('#topicslist a');
|
||||
|
||||
topics.forEach(el => {
|
||||
el.addEventListener('click', e => {
|
||||
search(e.target.innerText);
|
||||
searchInput.value = e.target.innerText;
|
||||
let tab = document.getElementById('tablist_panel_search');
|
||||
tab.checked = true
|
||||
changeTab(document.getElementById('tablist_panel_search'));
|
||||
})
|
||||
})
|
||||
|
||||
|
|
|
|||
|
|
@ -33,6 +33,11 @@
|
|||
for="tablist_panel_terms"><span class="visually-hidden">Zeige Tabinhalt
|
||||
</span>Topics</label>
|
||||
</div>
|
||||
<div class="control">
|
||||
<input class="visually-hidden" id="tablist_panel_search" name="tablist" type="radio" /><label
|
||||
for="tablist_panel_search"><span class="visually-hidden">Zeige Tabinhalt
|
||||
</span>Suche</label>
|
||||
</div>
|
||||
</fieldset>
|
||||
<div class="panel" id="tablist_panel_episode_panel">
|
||||
<h2 class="visually-hidden">
|
||||
|
|
@ -71,15 +76,27 @@
|
|||
<h2 class="visually-hidden">
|
||||
Topics (Tabinhalt)
|
||||
</h2>
|
||||
<ol>
|
||||
@foreach($topics as $topic)
|
||||
<ol id="topicslist">
|
||||
@foreach($topics['topics'] as $topic)
|
||||
<li>
|
||||
{{ implode(', ', $topic) }}
|
||||
@foreach($topic as $t)
|
||||
<a title="Suche nach {{ $t }}">{{ $t }}</a>
|
||||
@endforeach
|
||||
</li>
|
||||
@endforeach
|
||||
</ol>
|
||||
|
||||
</div>
|
||||
<div class="panel" id="tablist_panel_search_panel" hidden>
|
||||
<h2 class="visually-hidden">
|
||||
Suche (Tabinhalt)
|
||||
</h2>
|
||||
<div class="searchform">
|
||||
<input type="text" id="search">
|
||||
<button id="searchbutton">Suche</button>
|
||||
</div>
|
||||
<ol id="searchresults"></ol>
|
||||
</div>
|
||||
</div>
|
||||
|
||||
</div>
|
||||
|
|
@ -120,9 +137,13 @@
|
|||
<li><span class="neut">0</span> = Neutral ({{ round($senti_subdata['verteilung_perc']['neutral']) }}%)</li >
|
||||
<li><span class="neg">-1</span> = Negativ ({{ round($senti_subdata['verteilung_perc']['negative']) }}%)</li></p>
|
||||
</div>
|
||||
<div class="track-viz" id="sentiment-track" data-time="{{ implode(',', $senti_subdata['time']) }}" data-sentiments="{{ $senti_subdata['data'] }}" data-weights="{{ $senti_subdata['weights'] }}"></div>
|
||||
<div
|
||||
class="track-viz"
|
||||
id="sentiment-track"
|
||||
data-time="{{ implode(',', $senti_subdata['time']) }}" data-sentiments="{{ $senti_subdata['data'] }}" data-weights="{{ $senti_subdata['weights'] }}"></div>
|
||||
</div>
|
||||
</div>
|
||||
</div>
|
||||
</main>
|
||||
|
||||
@endsection
|
||||
|
|
|
|||
|
|
@ -24,8 +24,7 @@ Route::get('/detail/{id}', function(int $id) {
|
|||
$subtitles = $ep->subtitles;
|
||||
$mediacomposition = json_decode($ep->mediacomposition, 1);
|
||||
$durationSteps = $mediacomposition['chapterList'][0]['duration'] / 1000 / 10;
|
||||
$topics = json_decode($ep->topics);
|
||||
|
||||
$topics = json_decode($ep->topics, 1);
|
||||
|
||||
$subdata = json_decode($ep->subtitle_data, 1);
|
||||
$subdata_senti = json_decode($ep->sentiments_from_sub, 1);
|
||||
|
|
@ -34,7 +33,6 @@ Route::get('/detail/{id}', function(int $id) {
|
|||
$subdata_senti['weights'] = "";
|
||||
$subdata_senti['verteilung'] = ['neutral' => 0, 'negative' => 0, 'positive' => 0];
|
||||
|
||||
|
||||
foreach ($subdata_senti['sentiments'] as $value) {
|
||||
$subdata_senti['data'] .= $value[0].",";
|
||||
$subdata_senti['weights'] .= (50 * $value[1]).",";
|
||||
|
|
@ -57,7 +55,6 @@ Route::get('/detail/{id}', function(int $id) {
|
|||
$parser = new Podlove\Webvtt\Parser();
|
||||
$subtitles = $parser->parse($subtitles);
|
||||
|
||||
|
||||
return view('detail', ['eps' => $eps, 'title' => $title, 'subtitles' => $subtitles, 'mediacomposition' => $mediacomposition, 'durationSteps' => $durationSteps, 'dom_color' => $ep->viz_data, 'subdata' => $subdata, 'senti_subdata' => $subdata_senti, 'wpm' => $wordsPerMinute, 'topics' => $topics]);
|
||||
|
||||
});
|
||||
|
|
|
|||
Binary file not shown.
|
|
@ -1,6 +1,8 @@
|
|||
import sqlite3
|
||||
from pathlib import Path
|
||||
|
||||
con = sqlite3.connect("/home/gio/Code/VANA/database.sqlite")
|
||||
db_path = str(Path(__file__).parents[4]) + '/database.sqlite'
|
||||
con = sqlite3.connect(db_path)
|
||||
cur = con.cursor()
|
||||
|
||||
def get_subtitle(id):
|
||||
|
|
@ -40,5 +42,3 @@ def save_topics(id, data):
|
|||
cur.execute("UPDATE episodes SET topics = ? WHERE id = ?", [data, id])
|
||||
con.commit()
|
||||
con.close()
|
||||
|
||||
|
||||
|
|
|
|||
|
|
@ -6,7 +6,7 @@ from pathlib import Path
|
|||
|
||||
import polars as pl
|
||||
from database import queries
|
||||
from germansentiment import SentimentModel
|
||||
from import SentimentModel
|
||||
|
||||
|
||||
def get_sent(ep):
|
||||
|
|
|
|||
|
|
@ -2,6 +2,7 @@
|
|||
|
||||
import argparse
|
||||
import json
|
||||
import re
|
||||
from pathlib import Path
|
||||
|
||||
import polars as pl
|
||||
|
|
@ -46,8 +47,24 @@ def get_topics(ep):
|
|||
top.append(word[0])
|
||||
topics_list.append(top)
|
||||
|
||||
return json.dumps(topics_list)
|
||||
output_list = []
|
||||
|
||||
for patterns in topics_list:
|
||||
patterns = [re.sub(r'\.', '\.', p) for p in patterns]
|
||||
patterns = [re.sub(r'\*', '\*', p) for p in patterns]
|
||||
patterns = '|'.join(patterns)
|
||||
|
||||
df = df.with_columns(pl.col("sentences").str.extract_all(patterns).alias("matches"))
|
||||
output = {"timecode" : [], "matches" : []}
|
||||
|
||||
for row in df.rows(named=True):
|
||||
if len(row['matches']) > 0:
|
||||
output["timecode"].append(row["start"])
|
||||
output["matches"].append(row["matches"])
|
||||
|
||||
output_list.append(output)
|
||||
|
||||
return json.dumps({"topics" : topics_list, "verteilung" : output_list})
|
||||
|
||||
# CLI
|
||||
parser = argparse.ArgumentParser(
|
||||
|
|
|
|||
File diff suppressed because it is too large
Load Diff
BIN
database.sqlite
BIN
database.sqlite
Binary file not shown.
Loading…
Reference in New Issue