From fba8464daef023e52bcfb978a838f35666507d53 Mon Sep 17 00:00:00 2001 From: logique Date: Wed, 14 Jan 2026 15:32:27 +0330 Subject: [PATCH] Add indexer for almas movie series --- indexer.py | 42 ++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 42 insertions(+) create mode 100644 indexer.py diff --git a/indexer.py b/indexer.py new file mode 100644 index 0000000..fc0b10a --- /dev/null +++ b/indexer.py @@ -0,0 +1,42 @@ +from glob import glob +from urllib.parse import urlparse + + +links_dict = {} + +def parse_saymyname(link: str): + parsed = urlparse(link) + splitted_path = parsed.path.split("/") + first_letter = splitted_path[2] + + first_letter_list = links_dict.get(first_letter) + if not first_letter_list: + links_dict[first_letter] = list() + + joined_path = '/'.join(splitted_path[:4]) + url = f"{parsed.scheme}://{parsed.hostname}{joined_path}" + + if not url in links_dict[first_letter]: + links_dict[first_letter].append(url) + +# Iterating series +for file in glob("data/*.saymyname.website-series.txt"): + with open(file, "r") as fp: + links = fp.readlines() + + for link in links: + if "saymyname" in link: + parse_saymyname(link) + + +text = "# Almas Movie Archive (Only Series)" +for k, v in links_dict.items(): + print(f"{k}: {len(v)}") + + for link in v: + parsed_link = urlparse(link) + name = parsed_link.path.split("/")[-1] + text += f"{k}: [{name}]({link.replace(' ', '%20')})\n\n" + +with open("series.md", "w") as fp: + fp.write(text)