Merge pull request #14 from JuLi0n21

minor refactoring

build.py

@@ -1,105 +1,60 @@
 import datetime
 
-def generate_blocklist(header_file, sitelist_file, output_file, format, plus_version: bool, sitelist_file_plus='', filterlist_file_ublock_only=''):
+def generate_blocklist(output_file, format, is_sitelist_plus: bool = False, filterlist_file_ublock_only:str = ''):
   """
   Combines and modifies content of two files into a new file, 
   replacing placeholders, counting entries and adding text characters for proper format support.
   """
+  header_file = "header.txt"
+  sitelist_file = "sitelist.txt"
+  sitelist_file_plus = "sitelist-plus.txt"
   # Get current time in desired format with timezone-aware object
   current_time = datetime.datetime.now(datetime.timezone.utc).strftime("%d %b %Y %H:%M UTC")
 
   # Count entries in filtered file, ignoring lines starting with "!"
   num_entries = 0
   with open(sitelist_file, 'r') as f_filtered:
-    for line in f_filtered:
-      if not line.startswith('!'):
-        num_entries += 1
-  if plus_version == True:
-    with open(sitelist_file_plus, 'r') as f_filtered_plus:
-      for line in f_filtered_plus:
-        if not line.startswith('!'):
-          num_entries += 1
+        num_entries += countUncommentedLines(sitelist_file)
+  if (is_sitelist_plus):
+          num_entries += countUncommentedLines(sitelist_file_plus)
   if format == 'ublock':
-    with open(filterlist_file_ublock_only, 'r') as f_filtered_ublock:
-      for line in f_filtered_ublock:
-        if not line.startswith('!'):
-          num_entries += 1
+          num_entries += countUncommentedLines(filterlist_file_ublock_only)
 
   # Open header and combined files
-  with open(header_file, 'r') as f_header, open(sitelist_file, 'r') as f_domains, open(output_file, 'w') as f_combined:
+  with open(header_file, 'r') as f_header, open(output_file, 'w') as f_combined:
     header_lines = f_header.readlines()
 
     # Replace placeholders in header lines
     for i, line in enumerate(header_lines):
       if line.startswith('Title:'):
-        if plus_version == True:
-          header_lines[i] = line.replace('${filterlist_edition}', 'Plus')
-        elif plus_version == False:
-          header_lines[i] = line.replace('${filterlist_edition}', 'Basic')
-      if line.startswith('Description:'):
-        if plus_version == True:
-          header_lines[i] = line.replace('${sites_blocked}', 'malicious and not recommended')
-        elif plus_version == False:
-          header_lines[i] = line.replace('${sites_blocked}', 'malicious')
-      if line.startswith('Last modified:'):
+          header_lines[i] = line.replace('${filterlist_edition}', 'Plus' if is_sitelist_plus else 'Basic')
+      elif line.startswith('Description:'):
+          header_lines[i] = line.replace('${sites_blocked}', 'malicious and not recommended' if is_sitelist_plus else 'malicious')
+      elif line.startswith('Last modified:'):
         header_lines[i] = line.replace('${current_time}', current_time)
       elif line.startswith('Entries:'):
         header_lines[i] = line.replace('${num_entries}', str(num_entries))
       elif line.startswith('Format:'):
         header_lines[i] = line.replace('${format}', format)
+
       # Add comment symbols before every header line
       if format == 'ublock' or format == 'abp':
         header_lines[i] = ('! ' + header_lines[i])
       if format == 'hosts' or format == 'domains' or format == 'wildcard_domains' or format == 'wildcard_urls':
         header_lines[i] = ('# ' + header_lines[i])
 
-    # Add proper filterlist formatting to the domains
-    domains_lines = f_domains.readlines()
-    for i, line in enumerate(domains_lines):
-      if not line.startswith('!'):
-        if format == 'ublock' or format == 'abp':
-          domains_lines[i] = line.replace("\n", '')
-          domains_lines[i] = ('||' + domains_lines[i] + '^' + "\n")
-        elif format == 'hosts':
-          domains_lines[i] = ('0.0.0.0 ' + domains_lines[i])
-        elif format == 'wildcard_domains':
-          domains_lines[i] = ('*.' + domains_lines[i])
-        elif format == 'wildcard_urls':
-          domains_lines[i] = line.replace("\n", '')
-          domains_lines[i] = ('*://*.' + domains_lines[i] + '/*' + "\n")
-      if line.startswith('!'):
-        if format == 'hosts' or format == 'domains' or format == 'wildcard_domains' or format == 'wildcard_urls':
-          domains_lines[i] = line.replace('!', '#')
-      line.format()
-
-    # Add proper filterlist formatting to the plus version extra domains
-    if plus_version == True:
-      with open(sitelist_file_plus, 'r') as f_domains_plus:
-        domains_lines_plus = f_domains_plus.readlines()
-        for i, line in enumerate(domains_lines_plus):
-          if not line.startswith('!'):
-            if format == 'ublock' or format == 'abp':
-              domains_lines_plus[i] = line.replace("\n", '')
-              domains_lines_plus[i] = ('||' + domains_lines_plus[i] + '^' + "\n")
-            elif format == 'hosts':
-              domains_lines_plus[i] = ('0.0.0.0 ' + domains_lines_plus[i])
-            elif format == 'wildcard_domains':
-              domains_lines_plus[i] = ('*.' + domains_lines_plus[i])
-            elif format == 'wildcard_urls':
-              domains_lines_plus[i] = line.replace("\n", '')
-              domains_lines_plus[i] = ('*://*.' + domains_lines_plus[i] + '/*' + "\n")
-          if line.startswith('!'):
-            if format == 'hosts' or format == 'domains' or format == 'wildcard_domains' or format == 'wildcard_urls':
-             domains_lines_plus[i] = line.replace('!', '#')
-          line.format()
+    # Add proper filterlist formatting
+    domains_lines = formatFile(sitelist_file, format)
+    if (is_sitelist_plus):
+     domains_lines_plus = formatFile(sitelist_file_plus, format)
 
     # Write modified header and filtered content
     if format == 'abp':
       f_combined.writelines("[Adblock Plus]\n")
     f_combined.writelines(header_lines)
     f_combined.writelines("\n")
     f_combined.writelines(domains_lines)
-    if plus_version == True:
+    if (is_sitelist_plus):
       if format == 'domains' or format == 'hosts' or format == 'wildcard_domains':
         f_combined.writelines("\n")
       f_combined.writelines(domains_lines_plus)
@@ -108,20 +63,50 @@ def generate_blocklist(header_file, sitelist_file, output_file, format, plus_ver
 
   print("Generated filterlist with " + format + " format as " + output_file + "!")
 
+# Add proper filterlist formatting to the domains
+def formatFile(filename: str, format: str):
+  with open(filename, 'r') as f:
+    file = f.readlines()
+    for i, line in enumerate(file):
+      if not line.startswith('!'):
+        if format == 'ublock' or format == 'abp':
+          file[i] = line.replace("\n", '')
+          file[i] = ('||' + file[i] + '^' + "\n")
+        elif format == 'hosts':
+          file[i] = ('0.0.0.0 ' + file[i])
+        elif format == 'wildcard_domains':
+          file[i] = ('*.' + file[i])
+        elif format == 'wildcard_urls':
+          file[i] = line.replace("\n", '')
+          file[i] = ('*://*.' + file[i] + '/*' + "\n")
+      else:
+        if format == 'hosts' or format == 'domains' or format == 'wildcard_domains' or format == 'wildcard_urls':
+          file[i] = line.replace('!', '#')
+      line.format()
+  return file
+
+def countUncommentedLines(filename: str) -> int:
+  count = 0
+  with open(filename, 'r') as file:
+   for line in file:
+      if not line.startswith('!'):
+        count += 1
+  return count  
+
 # Generate Header to make filterlist
 print("Generating filterlist.")
 # Generate Plus version filterlist files
-generate_blocklist("header.txt", "sitelist.txt", "filterlist.txt", "ublock", True, "sitelist-plus.txt", "filters-ublock-only.txt")
-generate_blocklist("header.txt", "sitelist.txt", "filterlist-abp.txt", "abp", True, "sitelist-plus.txt")
-generate_blocklist("header.txt", "sitelist.txt", "filterlist-domains.txt", "domains", True, "sitelist-plus.txt")
-generate_blocklist("header.txt", "sitelist.txt", "filterlist-wildcard-domains.txt", "wildcard_domains", True, "sitelist-plus.txt")
-generate_blocklist("header.txt", "sitelist.txt", "filterlist-wildcard-urls.txt", "wildcard_urls", True, "sitelist-plus.txt")
-generate_blocklist("header.txt", "sitelist.txt", "filterlist-hosts.txt", "hosts", True, "sitelist-plus.txt")
+generate_blocklist("filterlist.txt", "ublock", True, "filters-ublock-only.txt")
+generate_blocklist("filterlist-abp.txt", "abp", True)
+generate_blocklist("filterlist-domains.txt", "domains", True)
+generate_blocklist("filterlist-wildcard-domains.txt", "wildcard_domains", True)
+generate_blocklist("filterlist-wildcard-urls.txt", "wildcard_urls", True)
+generate_blocklist("filterlist-hosts.txt", "hosts", True)
 # Generate Basic version filterlist files
-generate_blocklist("header.txt", "sitelist.txt", "filterlist-basic.txt", "ublock", False, filterlist_file_ublock_only="filters-ublock-only.txt")
-generate_blocklist("header.txt", "sitelist.txt", "filterlist-basic-abp.txt", "abp", False)
-generate_blocklist("header.txt", "sitelist.txt", "filterlist-basic-domains.txt", "domains", False)
-generate_blocklist("header.txt", "sitelist.txt", "filterlist-basic-wildcard-domains.txt", "wildcard_domains", False)
-generate_blocklist("header.txt", "sitelist.txt", "filterlist-basic-wildcard-urls.txt", "wildcard_urls", False)
-generate_blocklist("header.txt", "sitelist.txt", "filterlist-basic-hosts.txt", "hosts", False)
-print("Build Finished")
+generate_blocklist("filterlist-basic.txt", "ublock", False, "filters-ublock-only.txt")
+generate_blocklist("filterlist-basic-abp.txt", "abp")
+generate_blocklist("filterlist-basic-domains.txt", "domains")
+generate_blocklist("filterlist-basic-wildcard-domains.txt", "wildcard_domains")
+generate_blocklist("filterlist-basic-wildcard-urls.txt", "wildcard_urls")
+generate_blocklist("filterlist-basic-hosts.txt", "hosts")
+print("Build Finished")