Pair char preprocessor added (#5)

hmhard · Aug 12, 2024 · 1741cd2 · 1741cd2
1 parent fe9b844
commit 1741cd2
Show file tree

Hide file tree

Showing 22 changed files with 1,192,158 additions and 1,106,098 deletions.
diff --git a/clear-non-alpha.py b/clear-non-alpha.py
@@ -1,6 +1,7 @@
 import json
 import re
 
+# file_path = 'outputs/word_pair_dictionary.json'
 file_path = 'outputs/word_dictionary.json'
 
 valid_keys_file = 'outputs/valid_keys.json'
@@ -17,11 +18,11 @@
 invalid_keys = {}
 
 def is_only_emoji(s):
-    return all(emoji_pattern.match(c) for c in s)
+    return all(emoji_pattern.search(c) for c in s)
 
 # Separate keys based on whether they are only symbols, only numbers, or only emojis
 for k, v in data.items():
-    if symbols_pattern.match(k) or numbers_pattern.match(k) or is_only_emoji(k):
+    if symbols_pattern.search(k) or numbers_pattern.search(k) or is_only_emoji(k):
         invalid_keys[k] = v
     else:
         valid_keys[k] = v

diff --git a/index.html b/index.html
@@ -4,32 +4,42 @@
 <head>
     <meta charset="UTF-8">
     <meta name="viewport" content="width=device-width, initial-scale=1.0">
-    <title>Tikvah Telegram Channel Too 500 words</title>
+    <title>Tikvah Telegram Channel Top used words</title>
     <script src="https://cdn.tailwindcss.com"></script>
 
     <style>
         #chartdiv {
             width: 100%;
-            height: 550px;
+            height: 580px;
+        }
+
+        #chartdiv-pair {
+            width: 100%;
+            height: 580px;
         }
     </style>
 </head>
 
 <body>
 
-    <div class=" text-2xl text-center text-pink-500 underline py-3"><a href="https://t.me/s/tikvahethiopia">Tikvah Telegram Channel</a> Most Used 500 Words Until 2024-08-07</div>
+    <div class=" text-2xl text-center text-pink-500 underline py-3"><a href="https://t.me/s/tikvahethiopia">Tikvah Telegram Channel</a> Most Used 300 Words Until 2024-08-07</div>
 
 
     <section class="section">
 
         <div id="chartdiv"></div>
+        <div class=" text-2xl text-center text-pink-500 underline py-7"> Most Used 250 Pair Words Until 2024-08-07</div>
+
+        <div id="chartdiv-pair"></div>
+
+
 
-        <div class="has-text-centered"></div>
     </section>
     <div class="text-center underline">
         <a href="https://github.com/hmhard/tikvah-tg-channel-analysis">View on Github</a>
     </div>
-    <script src="./top-500.js" type="text/javascript"></script>
+    <script src="./top-300.js" type="text/javascript"></script>
+    <script src="./top-pair-250.js" type="text/javascript"></script>
     <!-- Load d3.js -->
     <script src="https://code.jquery.com/jquery-3.3.1.slim.min.js" integrity="sha256-3edrmyuQ0w65f8gfBsqowzjJe2iM6n0nKciPUp8y+7E=" crossorigin="anonymous"></script>
 
@@ -43,39 +53,40 @@
             tag: item,
             value: data[item]
         }));
+        var pair_datas = Object.keys(pair_data).map(item => ({
+            tag: item,
+            value: pair_data[item]
+        }));
         var myWords = Object.keys(data);
+        var pairWords = Object.keys(pair_data);
         $(window).on('load', function() {
             loadText();
-
-            $('#redraw').on('click', function() {
-                loadText();
-            });
         });
 
         function loadText() {
+            drawWorldCloud(pairWords.join(' '), pair_datas, "chartdiv-pair");
 
-
-            drawWorldCloud(myWords.join(' '));
+            drawWorldCloud(myWords.join(' '), datas, "chartdiv");
         }
 
 
-        function drawWorldCloud(sentence) {
+        function drawWorldCloud(sentence, datav, div) {
 
             am4core.useTheme(am4themes_animated);
 
-            var chart = am4core.create("chartdiv", am4plugins_wordCloud.WordCloud);
+            var chart = am4core.create(div, am4plugins_wordCloud.WordCloud);
             var series = chart.series.push(new am4plugins_wordCloud.WordCloudSeries());
 
             series.accuracy = 4;
             series.step = 15;
             series.rotationThreshold = 0.7;
             series.maxCount = 5000;
-            series.minWordLength = 2; // 最少頻度
+            series.minWordLength = 2;
             series.labels.template.tooltipText = "{word}: {value}";
             series.fontFamily = "'M PLUS 1p', sans-serif";
             series.maxFontSize = am4core.percent(50);
 
-            series.data = datas
+            series.data = datav
             series.dataFields.word = "tag";
             series.dataFields.value = "value";