X-Git-Url: https://git.armaanb.net/?a=blobdiff_plain;f=lightcards%2Fparse.py;h=3a1306bfae98840ba312fc58deff160d667ceeb8;hb=85327074f6aaa4e46ec36cec6fd6f07317e96ea6;hp=f9ca3ee15afc6648ced65bd8d9436e911c19d4cf;hpb=ddc9c9451cc55b4aa8869963439a48a8f1977a3c;p=lightcards.git diff --git a/lightcards/parse.py b/lightcards/parse.py index f9ca3ee..3a1306b 100644 --- a/lightcards/parse.py +++ b/lightcards/parse.py @@ -13,7 +13,9 @@ def md2html(file): try: return markdown.markdown(open(file, "r").read(), extensions=["tables"]) except FileNotFoundError: - sys.exit(f'lightcards: "{file}": No such file or directory') + raise Exception( + f'lightcards: "{file}": No such file or directory' + ) from None def parse_html(html): @@ -22,14 +24,21 @@ 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: + raise Exception("lightcards: No table found") from None + + ths = soup.find_all("th") + if len(ths) != 2: + raise Exception("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):