fix BtB downlaoding when name dont match exactly

also fix how some levels seem to share the same zip file...

Author: noisecode3

database/scrape_trle_download.py

@@ -195,18 +195,29 @@ def _get_trle_btb_download_info(trle_btb_url, lid):
     parted = trle_btb_url.split("BtB")
     btb_soup = None
     for case in ["web", "Web"]:
-        if parted[1] == f"/{case}/index.html" and len(parted) > 1:
+        if parted[1] == f"/{case}/index.html" and len(parted) == 2:
             btb_soup = scrape_common.get_soup(parted[0] + "BtB" + f"/{case}/downloads.html")
+            break
 
     if not btb_soup:
-        print("Fucked up")
+        print("No BtB download page soup")
         sys.exit(1)
+
     trle_info = _get_trle_info(lid)
     trle_title = trle_info[0]
     trle_zip_size = trle_info[2]
+
+    btb_title = scrape_common.normalize_level_name(" - ".join(trle_title.split(" - ")[1:]))
+
     link_tag = None
-    if isinstance(trle_info[0], str):
-        link_tag = btb_soup.find("a", string=trle_title.split(" - ")[1])
+    for a in btb_soup.find_all("a"):
+        if scrape_common.normalize_level_name(a.get_text()) == btb_title:
+            link_tag = a
+            break
+
+    if not link_tag:
+        print("No BtB link tag found")
+        sys.exit(1)
 
     download_url = None
     if isinstance(link_tag, Tag):

database/tombll_create.py

@@ -165,6 +165,9 @@ def database_zip_file(zip_file, level_id, con):
     Returns:
         int: The ID of the newly inserted zip file.
     """
+    # SQL queries for selecting, inserting, and linking tags
+    query_select_id = "SELECT ZipID FROM Zip WHERE name = ?"
+
     # SQL queries for inserting ZIP file data and linking ZIP files to a level
     query_insert = '''
         INSERT INTO Zip (name, size, md5sum, url, version, release)
@@ -182,10 +185,12 @@ def database_zip_file(zip_file, level_id, con):
     )
 
     # Insert the ZIP file and get its ID
-    zip_id = tombll_common.query_return_id(query_insert, insert_args, con)
+    zip_id = tombll_common.query_return_id(query_select_id, (zip_file.get('name'), ), con)
+    if zip_id is None:
+        zip_id = tombll_common.query_return_id(query_insert, insert_args, con)
 
     # Link the ZIP file to the level in ZipList table
-    query_insert_middle = "INSERT INTO ZipList (zipID, levelID) VALUES (?, ?)"
+    query_insert_middle = "INSERT OR IGNORE INTO ZipList (zipID, levelID) VALUES (?, ?)"
     middle_args = (zip_id, level_id)
     tombll_common.query_run(query_insert_middle, middle_args, con)
     return zip_id