]> git.armaanb.net Git - lightcards.git/blobdiff - lightcards/parse.py
Fix pickle implementation
[lightcards.git] / lightcards / parse.py
index f9ca3ee15afc6648ced65bd8d9436e911c19d4cf..6fc2a759cf2ac152f636b9a376efa9ddf2d9db29 100644 (file)
@@ -22,14 +22,21 @@ def parse_html(html):
     def clean_text(inp):
         return inp.get_text().rstrip()
 
-    soup = BeautifulSoup(html, "html.parser")
+    soup = BeautifulSoup(html, "html.parser").find("table")
     outp = []
 
-    for x in soup.find_all("tr"):
-        outp.append(Card([clean_text(y) for y in x.find_all("td")[:2]]))
+    try:
+        for x in soup.find_all("tr"):
+            outp.append(Card(tuple([clean_text(y) for y in x.find_all("td")])))
+    except AttributeError:
+        sys.exit("lightcards: No table found")
+
+    ths = soup.find_all("th")
+    if len(ths) != 2:
+        sys.exit("lightcards: Headings malformed")
 
     # Return a tuple of nested lists
-    return ([clean_text(x) for x in soup.find_all("th")][:2], outp[1:])
+    return ([clean_text(x) for x in ths], outp[1:])
 
 
 def main(file):