fixes; another function implemented

src/mwfin/functions.py

@@ -1,5 +1,5 @@
+import logging
 from typing import Union
-import re
 
 from aiohttp.client import ClientSession
 from bs4 import BeautifulSoup
@@ -8,6 +8,8 @@ from bs4.element import ResultSet, Tag
 from . import constants
 
 
+log = logging.getLogger(__name__)
+
 # The resulting dictionary's keys correspond to the name of the item (row) in the financial statement,
 # while its values will always be tuples with a length corresponding to the number of periods (columns)
 # and elements being the actual numbers, with the exception of the first key-value-pair, which will represent
@@ -37,21 +39,21 @@ 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.
     """
-    item_name = str(tr.find_next('td').find_next('div').string).strip()
-    if constants.FINANCIAL_STATEMENT_ITEMS[item_name]:
-        return True
+    if tr.name != 'tr':
+        return False
+    item_name = str(tr.td.div.string).strip()
+    try:
+        return constants.FINANCIAL_STATEMENT_ITEMS[item_name]
+    except KeyError:
+        log.warning(f"Unknown item name '{item_name}' found in financial statement.")
+        return False
 
 
 def find_relevant_table_rows(soup: BeautifulSoup) -> ResultSet:
     """
     Returns the table rows containing the data of interest.
     """
-    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
+    return soup.find('div', attrs={'class': 'financials'}).tbody.find_all(is_relevant_table_row)
 
 
 def extract_row_data(tr: Tag) -> tuple[str, tuple[int]]: