fixing the setup-pages-for-supported-languages script

* lang detection now uses the actual codes from config.yaml so pt-br and zh-hans get recognized properly instead of trying to re-localize them on every run
* removed the menu/aliases regex stripping since nothing in content uses those keys anymore, which also kills a latent bug where _index.md copies were getting stripped wrongly
* dropped the unused imports and the rootfiles dict that was never read

Author: bennyvasquez

setup-pages-for-supported-languages.py

@@ -1,46 +1,39 @@
 #!/usr/bin/env python3
 # original source from https://github.com/guardianproject/info/blob/master/setup-pages-for-supported-languages.py
+"""Generate per-language placeholder copies of every content/*.md file.
 
-import glob
+For each Hugo content page that doesn't yet have a localized sibling, copy
+the source file to content/<page>.<lang>.md for every language defined in
+config.yaml. This ensures Hugo emits localized URLs for every configured
+language, even when Weblate hasn't supplied a translation yet.
+
+Run from the repo root. Used by the publish workflows; safe to run locally.
+"""
 import os
-import re
-import shutil
-import sys
+
 import yaml
-from yaml.loader import SafeLoader
 
 with open('config.yaml') as fp:
-    #config = yaml.load(fp)
-    config = yaml.load(fp, Loader=SafeLoader)
+    config = yaml.safe_load(fp)
 
-rootfiles = dict()
-languages = config['languages'].keys()
-localized_pat = re.compile(r'.+\.([a-z][a-z](|-[A-Z][A-Z]|-Han[st]))$')
+# Sort longest-first so e.g. "pt-br" wins over "pt" when checking suffixes.
+languages = sorted(config['languages'].keys(), key=len, reverse=True)
 
-for root, dirs, files in os.walk('content'):
-    #if root.startswith('content/blog'):
-    #    continue  # skip since it is not translated
+for root, _dirs, files in os.walk('content'):
     for f in files:
         fileroot, extension = os.path.splitext(f)
-        if extension == '.md':
-            m = localized_pat.match(fileroot)
-            if m:
-                lang = m.group(1)
-                if lang in languages:
-                    #print(f, '\t', m.group(1))
-                    filename = os.path.join(root, f.replace('.' + lang + '.md', '.md'))
-                    if not filename in rootfiles:
-                        rootfiles[filename] = []
-                    rootfiles[filename].append(lang)
-                    continue
-            for lang in languages:
-                lang_file = os.path.join(root, fileroot + '.' + lang + extension)
-                if not os.path.exists(lang_file):
-                    with open(os.path.join(root, f)) as fp:
-                        contents = fp.read()
-                    # for some reason, index.*.md copies need the menu entries
-                    if f != 'index.md':
-                        contents = re.sub(r'\naliases:.*?\n([^ ])', r'\n\1', contents, flags=re.DOTALL)
-                        contents = re.sub(r'\nmenu:.*?\n([^ ])', r'\n\1', contents, flags=re.DOTALL)
-                    with open(lang_file, 'w') as fp:
-                        fp.write(contents)
+        if extension != '.md':
+            continue
+
+        # Skip files that are already localized variants (e.g. foo.de.md).
+        if any(fileroot.endswith('.' + lang) for lang in languages):
+            continue
+
+        for lang in languages:
+            lang_file = os.path.join(root, fileroot + '.' + lang + extension)
+            if os.path.exists(lang_file):
+                continue
+            with open(os.path.join(root, f)) as src:
+                contents = src.read()
+            with open(lang_file, 'w') as dst:
+                dst.write(contents)