Added HTML_Table_to_List

nitish-iiitd · nitish-iiitd · commit 80c46d29d2e8 · 2018-10-15T00:18:56.000+05:30
diff --git a/HTML_Table_to_List/HTMLTableToList.py b/HTML_Table_to_List/HTMLTableToList.py
@@ -0,0 +1,22 @@
+from bs4 import BeautifulSoup
+
+class HTMLTableToList():
+
+	def __init__(self, table_html):
+		self.table_html = table_html
+
+	def get_list(self):
+		list_of_list = []
+		soup = BeautifulSoup(self.table_html,"lxml")
+		table = soup.find('table')
+		all_tr = table.findAll('tr')
+		for tr in all_tr:
+			current_row = []
+			all_th  = tr.findAll('th')
+			all_td  = tr.findAll('td')
+			for th in all_th:
+				current_row.append(th.text)
+			for td in all_td:
+				current_row.append(td.text)
+			list_of_list.append(current_row)
+		return list_of_list
diff --git a/HTML_Table_to_List/README.md b/HTML_Table_to_List/README.md
@@ -0,0 +1,9 @@
+# HTML Table to Python List of List Converter
+A simple tool which takes a HTML table as string, and converts it to python list of list data structure and returns the same.
+
+## Libraries Required
+1. Beautiful Soap
+`$pip install bs4`
+
+## Usage
+A sample script `html_table_to_list_usage.py` has been provided to show the usage of the HTMLTableToList. It takes a string of html table, and prints the corresponding list of list. 
diff --git a/HTML_Table_to_List/__init__.py b/HTML_Table_to_List/__init__.py
diff --git a/HTML_Table_to_List/html_table_to_list_usage.py b/HTML_Table_to_List/html_table_to_list_usage.py
@@ -0,0 +1,35 @@
+from HTMLTableToList import HTMLTableToList
+from pprint import pprint
+
+html_table_string = """<table class="table table-condensed">
+                        <tr>
+                            <th>RGB</th>
+                            <td>53</td><td>72</td><td>35</td>
+                        </tr>
+                        <tr>
+                            <th>HSL</th><td>0.25</td><td>0.35</td><td>0.21</td>
+                        </tr>
+                        <tr>
+                            <th>HSV</th><td>91&deg;</td><td>51&deg;</td><td>28&deg;</td>
+                        </tr>
+                        <tr>
+                            <th>CMYK</th>
+                            <td>0.26</td><td>0.00</td><td>0.51 &nbsp; 0.72</td>
+                        </tr>
+                        <tr>
+                            <th>XYZ</th><td>4.0889</td><td>5.5130</td><td>2.4387</td>
+                        </tr>
+                        <tr>
+                            <th>Yxy</th><td>5.5130</td><td>0.3396</td><td>0.4579</td>
+                        </tr>
+                        <tr>
+                            <th>Hunter Lab</th><td>23.4798</td><td>-10.0046</td><td>10.2778</td>
+                        </tr>
+                        <tr>
+                            <th>CIE-Lab</th><td>28.1490</td><td>-15.1006</td><td>19.7427</td>
+                        </tr>
+                    </table>"""
+
+htmltabletolist = HTMLTableToList(html_table_string) ## args : HTML table as string
+list_of_list = htmltabletolist.get_list()
+pprint(list_of_list)
