X-Git-Url: https://git.armaanb.net/?a=blobdiff_plain;f=lightcards%2Fparse.py;h=6fc2a759cf2ac152f636b9a376efa9ddf2d9db29;hb=572d28cd80f72a865840afdb4aa5bf98b8d63b9b;hp=d70c4e977935752853167a775dcc7f8aacb96dc6;hpb=23bedf0b57ac06fbee12af000f42c992d3227dd1;p=lightcards.git diff --git a/lightcards/parse.py b/lightcards/parse.py index d70c4e9..6fc2a75 100644 --- a/lightcards/parse.py +++ b/lightcards/parse.py @@ -11,25 +11,32 @@ from .deck import Card def md2html(file): """Use the markdown module to convert input to HTML""" try: - return markdown.markdown(open(file, "r").read(), extensions=['tables']) + return markdown.markdown(open(file, "r").read(), extensions=["tables"]) except FileNotFoundError: - print(f"lightcards: \"{file}\": No such file or directory") - exit(1) + sys.exit(f'lightcards: "{file}": No such file or directory') def parse_html(html): """Use BeautifulSoup to parse the 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):