fix(eda): fixed bugs come with random generated datasets

sfu-db · Apr 20, 2021 · 53ecf76 · 53ecf76
1 parent 56896d0
commit 53ecf76
Show file tree

Hide file tree

Showing 13 changed files with 225 additions and 287 deletions.
diff --git a/dataprep/clean/clean_country.py b/dataprep/clean/clean_country.py
@@ -4,7 +4,6 @@
 from functools import lru_cache
 from operator import itemgetter
 from os import path
-from re import error
 from typing import Any, Union
 
 import dask
@@ -289,14 +288,11 @@ def _check_country(country: str, input_format: str, strict: bool, clean: bool) -
 
     if strict and input_format == "regex":
         for form in ("name", "official"):
-            try:
-                ind = DATA[
-                    DATA[form].str.contains(f"^{country}$", flags=re.IGNORECASE, na=False)
-                ].index
-                if np.size(ind) > 0:
-                    return (ind[0], "success") if clean else True
-            except error:
-                return (None, "unknown") if clean else False
+            ind = DATA[
+                DATA[form].str.contains(f"^{re.escape(country)}$", flags=re.IGNORECASE, na=False)
+            ].index
+            if np.size(ind) > 0:
+                return (ind[0], "success") if clean else True
 
     elif not strict and input_format in ("regex", "name", "official"):
         for index, country_regex in enumerate(REGEXES):
@@ -305,7 +301,9 @@ def _check_country(country: str, input_format: str, strict: bool, clean: bool) -
 
     else:
         ind = DATA[
-            DATA[input_format].str.contains(f"^{country}$", flags=re.IGNORECASE, na=False)
+            DATA[input_format].str.contains(
+                f"^{re.escape(country)}$", flags=re.IGNORECASE, na=False
+            )
         ].index
         if np.size(ind) > 0:
             return (ind[0], "success") if clean else True

diff --git a/dataprep/eda/create_report/formatter.py b/dataprep/eda/create_report/formatter.py
@@ -217,7 +217,6 @@ def format_basic(df: dd.DataFrame, cfg: Config) -> Dict[str, Any]:
             for tab in rndrd.tabs:
                 fig = tab.child
                 fig.sizing_mode = "stretch_width"
-                # fig.title = Title(text=tab.title, align="center", text_font_size="20px")
                 figs_corr.append(fig)
                 res["correlation_names"].append(tab.title)
             res["correlations"] = components(figs_corr)

diff --git a/dataprep/eda/create_report/templates/base.html b/dataprep/eda/create_report/templates/base.html
@@ -19,7 +19,7 @@
       <a href="#variables">Variables ≡</a>
       <div class="subcontent">
         {% for var in context.components.variables.keys() %}
-        <a href="#{{ var }}">{{ var }}</a>
+        <a href="#{{ var|e }}">{{ var|e }}</a>
         {% endfor %}
       </div>
       {% endif %}
@@ -92,11 +92,11 @@ <h1 class="tb-title">Missing Values</h1>
   {% if context.components.has_interaction and context.components.interactions[0] != 0%}
   {{ context.components.interactions[0] }}
   {% endif %}
-  
+
   <footer class="footer">
     <p>Report generated with <a href="https://dataprep.ai/">DataPrep</a></p>
   </footer>
 
 </body>
 
-</html>
+</html>
diff --git a/dataprep/eda/create_report/templates/correlation.html b/dataprep/eda/create_report/templates/correlation.html
@@ -1,20 +1,27 @@
 <div class="correlation-container">
   <div class="corr-plot">
-    <div class="report-tab">
-      <button class="corr-tablinks active" onclick="openTab(event, '{{context.components.correlation_names[0]}}')" >{{context.components.correlation_names[0]}}</button>
-      <button class="corr-tablinks" onclick="openTab(event, '{{context.components.correlation_names[1]}}')">{{context.components.correlation_names[1]}}</button>
-      <button class="corr-tablinks" onclick="openTab(event, '{{context.components.correlation_names[2]}}')">{{context.components.correlation_names[2]}}</button>
+    <div class="vp-switch">
+      {% for corr_name in context.components.correlation_names %}
+      {% if loop.index == 1 %}
+      <input type="radio" name="correlation-names" id="corr-name-{{ corr_name }}" checked />
+      <label for="corr-name-{{ corr_name }}" onclick="openTab(this)" >{{ corr_name }}</label>
+      {% else %}
+      <input type="radio" name="correlation-names" id="corr-name-{{ corr_name }}" />
+      <label for="corr-name-{{ corr_name }}" onclick="openTab(this)" >{{ corr_name }}</label>
+      {% endif %}
+      {% endfor %}
     </div>
 
-    <div id={{context.components.correlation_names[0]}} class="corr-tabcontent">
-      {{ context.components.correlations[1][0]}}
+    {% for plot in context.components.correlations[1] %}
+    {% if loop.index == 1 %}
+    <div class="corr-tabcontent" style="display: block;">
+      {{ plot }}
     </div>
-    <div id={{context.components.correlation_names[1]}} style="display:none" class="corr-tabcontent">
-      {{ context.components.correlations[1][1]}}
+    {% else %}
+    <div class="corr-tabcontent">
+      {{ plot }}
     </div>
-    <div id={{context.components.correlation_names[2]}} style="display:none" class="corr-tabcontent">
-      {{ context.components.correlations[1][2]}}
-    </div>
-
+    {% endif %}
+    {% endfor %}
   </div>
-</div>
+</div>
diff --git a/dataprep/eda/create_report/templates/missing.html b/dataprep/eda/create_report/templates/missing.html
@@ -1,24 +1,28 @@
 <div class="missing-container">
   <div class="miss-plot">
-    <div class="report-tab">
-      <button class="missing-tablinks active" onclick="openTab(event, '{{context.components.missing_tabs[0]}}')" >{{context.components.missing_tabs[0]}}</button>
-      <button class="missing-tablinks" onclick="openTab(event, '{{context.components.missing_tabs[1]}}')">{{context.components.missing_tabs[1]}}</button>
-      <button class="missing-tablinks" onclick="openTab(event, '{{context.components.missing_tabs[2]}}')">{{context.components.missing_tabs[2]}}</button>
-      <button class="missing-tablinks" onclick="openTab(event, '{{context.components.missing_tabs[3]}}')">{{context.components.missing_tabs[3]}}</button>
+    <div class="vp-switch">
+      {% for miss_name in context.components.missing_tabs %}
+      {% if loop.index == 1 %}
+      <input type="radio" name="miss-names" id="miss-name-{{ miss_name }}" checked />
+      <label for="miss-name-{{ miss_name }}" onclick="openTab(this)" >{{ miss_name }}</label>
+      {% else %}
+      <input type="radio" name="miss-names" id="miss-name-{{ miss_name }}" />
+      <label for="miss-name-{{ miss_name }}" onclick="openTab(this)" >{{ miss_name }}</label>
+      {% endif %}
+      {% endfor %}
     </div>
 
-    <div id="{{context.components.missing_tabs[0]}}" class="missing-tabcontent">
-      {{ context.components.missing[1][0]}}
+    {% for plot in context.components.missing[1] %}
+    {% if loop.index == 1 %}
+    <div class="missing-tabcontent" style="display: block;">
+      {{ plot }}
     </div>
-    <div id="{{context.components.missing_tabs[1]}}" style="display:none" class="missing-tabcontent">
-      {{ context.components.missing[1][1]}}
-    </div>
-    <div id="{{context.components.missing_tabs[2]}}" style="display:none" class="missing-tabcontent">
-      {{ context.components.missing[1][2]}}
-    </div>
-    <div id="{{context.components.missing_tabs[3]}}" style="display:none" class="missing-tabcontent">
-      {{ context.components.missing[1][3]}}
+    {% else %}
+    <div class="missing-tabcontent">
+      {{ plot }}
     </div>
+    {% endif %}
+    {% endfor %}
 
   </div>
-</div>
+</div>
diff --git a/dataprep/eda/create_report/templates/overview.html b/dataprep/eda/create_report/templates/overview.html
@@ -2,7 +2,7 @@
     <div class="ov-stats-container">
         <div class="tb-container">
             <h3 class="tb-title">Dataset Statistics</h3>
-            <table class="rp-table-overview">
+            <table class="rp-table">
                 {% for h, d in context.components.overview[0].items() %}
                 <tr>
                     <th>{{ h }}</th>
@@ -12,7 +12,7 @@ <h3 class="tb-title">Dataset Statistics</h3>
                 <tr>
                     <th style="vertical-align: top;">Variable Types</th>
                     <td>
-                        <ul>
+                        <ul style="list-style: none;padding-left: 0;">
                             {% for tp, num in context.components.overview[1].items() %}
                             <li>{{ tp }}: {{ num }}</li>
                             {% endfor %}
@@ -24,15 +24,15 @@ <h3 class="tb-title">Dataset Statistics</h3>
     </div>
     {% if context.components.overview_insights %}
         <div class="insights">
+            <h3>Dataset Insights</h3>
             {% for page, content in context.components.overview_insights.items() %}
             <div class="page-box" id="page-{{ loop.index }}">
-                <h3>Dataset Insights</h3>
                 <table>
                     {% for entry in content %}
                     {% for ins_type, insight in entry.items() %}
                     <tr>
                         <th id="des">
-                            {{ insight.replace('/*', '<span class="col-name' + context.rnd | string() + '">').replace('*/', '</span>') }}
+                            {{ insight|escape|replace('/*start*/', '<span class="col-name">')|replace('/*end*/', '</span>') }}
                         </th>
                         <td><span class="ins-type">{{ ins_type }}</span></td>
                     </tr>
@@ -41,18 +41,17 @@ <h3>Dataset Insights</h3>
                 </table>
             </div>
             {% endfor %}
+            <div class="page-switch-box">
+                <ul class="page-switch">
+                    {% for _ in context.components.overview_insights %}
+                    {% if loop.length > 1 %}
+                    <li id="switch-{{ loop.index }}" onclick="switchPage(this)">{{ loop.index }}</li>
+                    {% endif %}
+                    {% endfor %}
+                </ul>
+            </div>
         </div>
     {% endif %}
     </div>
-    {% if context.components.overview_insights %}
-    <div class="page-switch-box">
-        <ul class="page-switch">
-            {% for _ in context.components.overview_insights %}
-            {% if loop.length > 1 %}
-            <li id="switch-{{ loop.index }}" onclick="switchPage(this)">{{ loop.index }}</li>
-            {% endif %}
-            {% endfor %}
-        </ul>
-    </div>
-    {% endif %}
-</div>
+
+</div>
diff --git a/dataprep/eda/create_report/templates/scripts.html b/dataprep/eda/create_report/templates/scripts.html
@@ -17,19 +17,10 @@
 {% endif %}
 
 <script>
-    setTimeout(init, 100);
-    function init() {
-        const tableRows_num = document.querySelectorAll('.tb-numerical tr');
-        highlightTableValue(tableRows_num, '#f00');
-        scientificNotationStrip(tableRows_num);
-
-        const tableRows_cat = document.querySelectorAll('.tb-categorical tr');
-        highlightTableValue(tableRows_cat, '#f00');
-        scientificNotationStrip(tableRows_cat);
-    }
+    setTimeout(init, 50);
 
-    window.onload = () => {
-        let tableRows = document.getElementsByTagName('tr');
+    function init() {
+        const tableRows = document.querySelectorAll('.rp-table tr');
         highlightTableValue(tableRows, '#f00');
         scientificNotationStrip(tableRows);
     }
@@ -102,25 +93,13 @@
         }
     }
 
-    function openTab(evt, tabName) {
-      var i, tabcontent, tablinks;
-
-      tabcontent = document.getElementsByClassName(evt.currentTarget.parentElement.nextElementSibling.className);
-      for (i = 0; i < tabcontent.length; i++) {
-        tabcontent[i].style.display = "none";
-      }
-
-      tablinks = document.getElementsByClassName(evt.currentTarget.className);
-      for (i = 0; i < tablinks.length; i++) {
-        tablinks[i].style.background = "transparent";
-        tablinks[i].className = tablinks[i].className.replace(" active", "");
-      }
-      document.getElementById(tabName).style.display = "flex";
-      for (i = 0; i < tablinks.length; i++){
-      }
-      evt.currentTarget.className += " active";
-
-      evt.currentTarget.style.background = "#b5d6ea";
+    function openTab(e) {
+        const btnIndex = [...e.parentElement.children].indexOf(e);
+        const allContentEle = e.parentElement.parentElement.children;
+        for (let i of [...allContentEle].slice(1)) {
+            i.style.display = 'none';
+        }
+        allContentEle[Math.round(btnIndex/2)].style.display = (btnIndex == 1) ? 'flex' : 'block';
     }
 
     function switchPage(e) {
@@ -137,4 +116,4 @@
         e.style.backgroundColor = '#5DADE2';
     }
 
-</script>
+</script>