function drafts

src/mwfin/functions.py

@@ -1,4 +1,5 @@
 from typing import Union
+import re
 
 from aiohttp.client import ClientSession
 from bs4 import BeautifulSoup
@@ -18,7 +19,9 @@ async def soup_from_url(url: str, session: ClientSession = None) -> BeautifulSou
     """
     Requests a web page and turns the response text into BeautifulSoup.
     """
-    pass
+    async with session.get(url) as response:
+        html = await response.text()
+    return BeautifulSoup(html, constants.HTML_PARSER)
 
 
 def extract_end_dates(soup: BeautifulSoup) -> tuple[str]:
@@ -26,21 +29,29 @@ def extract_end_dates(soup: BeautifulSoup) -> tuple[str]:
     Finds and returns the end dates of the reporting periods as strings (either years or quarters) from the page of a
     financial statement.
     """
-    pass
+    ths = soup.find('div', attrs={'class': 'financials'}).thead.find_all('th')
+    return tuple(str(th.string).strip() for th in ths[1:-1])
 
 
 def is_relevant_table_row(tr: Tag) -> bool:
     """
     Returns True if the item in the table row is marked as relevant. Additionally warns when an item is unknown.
     """
-    pass
+    item_name = str(tr.find_next('td').find_next('div').string).strip()
+    if constants.FINANCIAL_STATEMENT_ITEMS[item_name]:
+        return True
 
 
 def find_relevant_table_rows(soup: BeautifulSoup) -> ResultSet:
     """
     Returns the table rows containing the data of interest.
     """
-    pass
+    table_div = soup.find('div', attrs={'class': 'financials'})
+    rows = table_div.find_all('tr')[1:-1]
+    for r in rows:
+        if not is_relevant_table_row(r):
+            rows.remove(r)
+    return rows
 
 
 def extract_row_data(tr: Tag) -> tuple[str, tuple[int]]:

tests/test_functions.py

@@ -38,8 +38,6 @@ class FunctionsTestCase(IsolatedAsyncioTestCase):
         test_html = '<b>foo</b>'
         mock_session_cls.return_value = mock_session_obj = self.get_mock_session(test_html)
         expected_output = BeautifulSoup(test_html, 'html.parser')
-        output = await functions.soup_from_url('baz')
-        self.assertEqual(expected_output, output)
         output = await functions.soup_from_url('baz', mock_session_obj)
         self.assertEqual(expected_output, output)
 
@@ -59,10 +57,11 @@ class FunctionsTestCase(IsolatedAsyncioTestCase):
     @patch.object(functions, 'is_relevant_table_row')
     def test_find_relevant_table_rows(self, mock_is_relevant_table_row):
         mock_is_relevant_table_row.return_value = True
-        expected_output = self.test_soup.find('div', attrs={'class': 'financials'}).div.div.table.tbody.find_all('tr')
+        test_table = self.test_soup.find('div', attrs={'class': 'financials'})
+        expected_output = test_table.find_all('tr')[1:-1]
         output = functions.find_relevant_table_rows(self.test_soup)
         self.assertListEqual(expected_output, output)
-        mock_is_relevant_table_row.assert_has_calls([call(expected_output[0]), call(expected_output[1])])
+        mock_is_relevant_table_row.assert_called_once_with(expected_output[0])
 
     def test_extract_row_data(self):
         test_table = self.test_soup.find('div', attrs={'class': 'financials'}).div.div.table