Visualization Testing Corpus Analysis

app/static/js/CorpusAnalysis/CQiClient.js

@@ -100,7 +100,69 @@ class CQiCorpus {
 
   getCorpusData() {
     return new Promise((resolve, reject) => {
-      const dummyData = {};
+      const dummyData = {
+          "num_tokens": 2000,    // number of tokens in the corpus
+          "num_unique_words": 500,    // number of unique words in the corpus
+          "num_unique_lemmas": 200,    // number of unique lemmas in the corpus
+          "num_sentences": 90,    // number of sentences in the corpus
+          "average_sentence_length": 11,   // average number of tokens per sentence in the corpus
+          "num_ent_types": 30,    // number of entities in the corpus
+          "num_unique_ent_types":10,
+          "ent_type_freqs": {
+            "str": 10,    // number of ent_types with ent_type "str"
+            // ...
+        },
+          "texts": [
+              {
+                  "num_tokens": 11,    // number of tokens in the text
+                  "num_unique_words": 12,    // number of unique words in the text
+                  "word_freqs": {    // frequency of unique words in the text (sorted by frequency)
+                    "str": "int",    // number of tokens with word "str"
+                    // ...
+                  },
+                  "num_unique_lemmas": 15,    // number of unique lemmas in the text
+                  "lemma_freqs": {    // frequency of unique lemmas in the text (sorted by frequency)
+                    "str": "int",    // number of tokens with lemma "str"
+                    // ...
+                  },
+                  "num_sentences": 4,    // number of sentences in the text
+                  "average_sentence_length": 3,   // average number of tokens per sentence in the text
+                  "num_ent_types": 12,    // number of ent_types in the text
+                  "num_unique_ent_types": 28,    // number of unique ent_types in the text
+                  "num_entities_by_id": {
+                      "1": "int",    // number of entities with id 1
+                      // ...
+                  },            
+                  "author": "Author Name",
+                  "title": "Titel",
+                  "publishing_year": 1950
+              },
+              {
+                "num_tokens": 800,    // number of tokens in the text
+                "num_unique_words": 60,    // number of unique words in the text
+                "word_freqs": {    // frequency of unique words in the text (sorted by frequency)
+                  "testwort": 50,    // number of tokens with word "str"
+                  "testwort2": 1
+                },
+                "num_unique_lemmas": 15,    // number of unique lemmas in the text
+                "lemma_freqs": {    // frequency of unique lemmas in the text (sorted by frequency)
+                  "testlemma": 11,    // number of tokens with lemma "str"
+                  "testlemma2": 1
+                },
+                "num_sentences": 90,    // number of sentences in the text
+                "average_sentence_length": 7,   // average number of tokens per sentence in the text
+                "num_ent_types": 19,
+                "num_unique_ent_types": 5,    // number of unique ent_types in the text
+                "num_entities_by_id": {
+                    "1": "int",    // number of entities with id 1
+                    // ...
+                },            
+                "author": "Author Name 2",
+                "title": "Titel 2",
+                "publishing_year": 1951
+              }
+          ]
+      };
 
       resolve(dummyData);
       /*

app/static/js/CorpusAnalysis/CorpusAnalysisApp.js

@@ -25,7 +25,6 @@ class CorpusAnalysisApp {
   init() {
     this.disableActionElements();
     this.elements.m.initModal.open();
-  
     // Init data
     this.data.cQiClient = new CQiClient(this.settings.corpusId);
     this.data.cQiClient.connect()
@@ -35,7 +34,13 @@ class CorpusAnalysisApp {
       .then(
         cQiCorpus => {
           this.data.corpus = {o: cQiCorpus};
-          // TODO: Don't do this here
+          this.data.corpus.o.getCorpusData()
+            .then(corpusData => {
+              this.renderGeneralCorpusInfo(corpusData);
+              this.renderTextInfoList(corpusData);
+              this.renderTextProportionsGraphic(corpusData);
+            });
+          // TODO: Don't do this hgere
           cQiCorpus.updateDb();
           this.enableActionElements();
           for (let extension of Object.values(this.extensions)) {extension.init();}
@@ -52,6 +57,7 @@ class CorpusAnalysisApp {
           }
         }
       );
+
   
     // Add event listeners
     for (let extensionSelectorElement of this.elements.overview.querySelectorAll('.extension-selector')) {
@@ -95,4 +101,40 @@ class CorpusAnalysisApp {
       }
     }
   }
+
+  renderGeneralCorpusInfo(corpusData) {
+    let corpusGeneralInfoListElement = document.querySelector('.corpus-general-info-list');
+    corpusGeneralInfoListElement.querySelector('.corpus-num-tokens').innerHTML = `<b>Number of tokens:</b> ${this.data.corpus.o.size}`;
+    corpusGeneralInfoListElement.querySelector('.corpus-text-count').innerHTML = `<b>Corpus text count:</b> ${corpusData.texts.length}`;
+    corpusGeneralInfoListElement.querySelector('.corpus-num-unique-words').innerHTML = `<b>Corpus unique word count:</b> ${corpusData.num_unique_words}`;
+    corpusGeneralInfoListElement.querySelector('.corpus-num-unique-lemmas').innerHTML = `<b>Corpus unique lemma count:</b> ${corpusData.num_unique_lemmas}`;
+    // corpusGeneralInfoListElement.querySelector('.corpus-most-frequent-words').innerHTML = `<b>Corpus most frequent words:</b> ${corpusData.most_frequent_words.join(', ');
+    corpusGeneralInfoListElement.querySelector('.corpus-num-sentences').innerHTML = `<b>Corpus sentence count:</b> ${corpusData.num_sentences}`;
+    corpusGeneralInfoListElement.querySelector('.corpus-average-sentence-length').innerHTML = `<b>Corpus average sentence length:</b> ${corpusData.average_sentence_length}`;
+    corpusGeneralInfoListElement.querySelector('.corpus-num-ent-types').innerHTML = `<b>Corpus entity count:</b> ${corpusData.num_ent_types}`;
+    corpusGeneralInfoListElement.querySelector('.corpus-num-unique-ent-types').innerHTML = `<b>Corpus unique entity count:</b> ${corpusData.num_unique_ent_types}`;
+  }
+
+  renderTextInfoList(corpusData) {
+    let corpusTextInfoListElement = document.querySelector('.corpus-text-info-list');
+    let corpusTextInfoList = new CorpusTextInfoList(corpusTextInfoListElement);
+    corpusTextInfoList.add(corpusData.texts);
+    
+  }
+
+  renderTextProportionsGraphic(corpusData) {
+    let textProportionsGraphicElement = document.querySelector('#text-proportions-graphic');
+    let graphData = [
+      {
+        values: corpusData.texts.map(text => text.num_tokens),
+        labels: corpusData.texts.map(text => `${text.title} (${text.publishing_year})`),
+        type: 'pie'
+      }
+    ];
+    let graphLayout = {
+      height: 400,
+      width: 500
+    };
+    Plotly.newPlot(textProportionsGraphicElement, graphData, graphLayout);
+  }
 }

app/static/js/ResourceLists/CorpusTextInfoList.js

@@ -0,0 +1,97 @@
+class CorpusTextInfoList extends ResourceList {
+
+  static autoInit() {
+    for (let corpusTextInfoListElement of document.querySelectorAll('.corpus-text-info-list:not(.no-autoinit)')) {
+      new CorpusTextInfoList(corpusTextInfoListElement);
+    }
+  }
+
+  static defaultOptions = {
+    page: 4
+  };
+
+  constructor(listContainerElement, options = {}) {
+    let _options = Utils.mergeObjectsDeep(
+      CorpusTextInfoList.defaultOptions,
+      options
+    );
+    super(listContainerElement, _options);
+    this.isInitialized = false;
+  }
+
+  get item() {
+    return (values) => {
+      return `
+        <tr class="list-item clickable hoverable">
+          <td><span class="title"></span> (<span class="publishing_year"></span>)</td>
+          <td><span class="num_tokens"></span></td>
+          <td><span class="num_unique_words"></span></td>
+          <td><span class="num_unique_lemmas"></span></td>
+          <td><span class="num_sentences"></span></td>
+          <td><span class="average_sentence_length"></span></td>
+          <td><span class="num_ent_types"></span></td>
+          <td><span class="num_unique_ent_types"></span></td>
+        </tr>
+      `.trim();
+    }
+  }
+
+  get valueNames() {
+    return [
+      'title',
+      'publishing_year',
+      'num_tokens',
+      'num_unique_words',
+      'num_unique_lemmas',
+      'num_sentences',
+      'average_sentence_length',
+      'num_ent_types',
+      'num_unique_ent_types'
+    ];
+  }
+
+  initListContainerElement() {
+    if (!this.listContainerElement.hasAttribute('id')) {
+      this.listContainerElement.id = Utils.generateElementId('corpus-file-list-');
+    }
+    let listSearchElementId = Utils.generateElementId(`${this.listContainerElement.id}-search-`);
+    this.listContainerElement.innerHTML = `
+      <div class="input-field">
+        <i class="material-icons prefix">search</i>
+        <input id="${listSearchElementId}" class="search" type="text"></input>
+        <label for="${listSearchElementId}">Search corpus file</label>
+      </div>
+      <table>
+        <thead>
+          <tr>
+            <th>Text</th>
+            <th>Number of tokens</th>
+            <th>Number of unique words</th>
+            <th>Number of unique lemmas</th>
+            <th>Number of sentences</th>
+            <th>Average sentence length</th>
+            <th>Number of entity types</th>
+            <th>Number of unique entity types</th>
+          </tr>
+        </thead>
+        <tbody class="list"></tbody>
+      </table>
+      <ul class="pagination"></ul>
+    `.trim();
+  }
+
+  mapResourceToValue(corpusTextData) {
+    console.log(corpusTextData);
+    return {
+      title: corpusTextData.title,
+      publishing_year: corpusTextData.publishing_year,
+      num_tokens: corpusTextData.num_tokens,
+      num_unique_words: corpusTextData.num_unique_words,
+      num_unique_lemmas: corpusTextData.num_unique_lemmas,
+      num_sentences: corpusTextData.num_sentences,
+      average_sentence_length: corpusTextData.average_sentence_length,
+      num_ent_types: corpusTextData.num_ent_types,
+      num_unique_ent_types: corpusTextData.num_unique_ent_types
+    };
+  }
+}

app/static/js/ResourceLists/ResourceList.js

@@ -15,6 +15,7 @@ class ResourceList {
     UserList.autoInit();
     AdminUserList.autoInit();
     CorpusFollowerList.autoInit();
+    CorpusTextInfoList.autoInit();
   }
 
   static defaultOptions = {

app/templates/_scripts.html.j2

@@ -1,6 +1,8 @@
 <script src="https://cdnjs.cloudflare.com/ajax/libs/fast-json-patch/3.1.1/fast-json-patch.min.js" integrity="sha512-5uDdefwnzyq4N+SkmMBmekZLZNmc6dLixvVxCdlHBfqpyz0N3bzLdrJ55OLm7QrZmgZuhLGgHLDtJwU6RZoFCA==" crossorigin="anonymous" referrerpolicy="no-referrer"></script>
 <script src="https://cdnjs.cloudflare.com/ajax/libs/list.js/2.3.1/list.min.js" integrity="sha512-93wYgwrIFL+b+P3RvYxi/WUFRXXUDSLCT2JQk9zhVGXuS2mHl2axj6d+R6pP+gcU5isMHRj1u0oYE/mWyt/RjA==" crossorigin="anonymous" referrerpolicy="no-referrer"></script>
 <script src="https://cdnjs.cloudflare.com/ajax/libs/socket.io/4.5.4/socket.io.min.js" integrity="sha512-HTENHrkQ/P0NGDFd5nk6ibVtCkcM7jhr2c7GyvXp5O+4X6O5cQO9AhqFzM+MdeBivsX7Hoys2J7pp2wdgMpCvw==" crossorigin="anonymous" referrerpolicy="no-referrer"></script>
+<script src="https://cdn.plot.ly/plotly-2.24.1.min.js" charset="utf-8"></script>
+
 {%- assets
   filters='rjsmin',
   output='gen/app.%(version)s.js',
@@ -49,6 +51,7 @@
   'js/ResourceLists/UserList.js',
   'js/ResourceLists/AdminUserList.js',
   'js/ResourceLists/CorpusFollowerList.js',
+  'js/ResourceLists/CorpusTextInfoList.js',
   'js/ResourceLists/DetailledPublicCorpusList.js'
 %}
 <script src="{{ ASSET_URL }}"></script>

app/templates/corpora/analysis.html.j2

@@ -33,6 +33,58 @@
   {% endfor %}
 </div>
 
+<div class="row">
+  <div class="col s12">
+    <h4><i class="material-icons left">query_stats</i>Visualizations</h4>
+  </div>
+  <div class="col s4" >
+    <div class="card hoverable">
+      <div class="card-content">
+        <span class="card-title">General information about the Corpus</span>
+        <p></p>
+        <br>
+        <ul class="corpus-general-info-list">
+          <li class="corpus-num-tokens"></li>
+          <br>
+          <li class="corpus-text-count"></li>
+          <br>
+          <li class="corpus-num-unique-words"></li>
+          <br>
+          <li class="corpus-num-unique-lemmas"></li>
+          <br>
+          <li class="corpus-num-sentences"></li>
+          <br>
+          <li class="corpus-average-sentence-length"></li>
+          <br>
+          <li class="corpus-num-ent-types"></li>
+          <br>
+          <li class="corpus-num-unique-ent-types"></li>
+          <br>
+        </ul>
+      </div>
+    </div>
+  </div>
+  <div class="col s8">
+    <div class="card hoverable">
+      <div class="card-content">
+        <span class="card-title">Text information</span>
+        <div class="corpus-text-info-list no-autoinit"></div>
+      </div>
+    </div>
+  </div>
+</div>
+<div class="row">
+  <div class="col 12">
+    <div class="card hoverable">
+      <div class="card-content">
+        <span class="card-title">Text proportions within the corpus</span>
+        <div id="text-proportions-graphic"></div>
+      </div>
+    </div>
+  </div>
+</div>
+
+
 {% for extension in extensions %}
 <div id="{{ extension.id_prefix }}-container">
   {{ extension.container_content }}