Improve filtering criteria for undocumented templates (#70)

master.py

@@ -16,7 +16,7 @@
 # Using {{lang}} and {{if lang}} on non-template pages -> this is apparently somewhat common now to make copy/paste editing easier
 # Pages which link to disambig pages not in hatnote/see also
 # Just... a summary of every single external link. Maybe just 'count per domain' and then list the top 10 pages? I'm finding a LOT of sus links, and it's only the ones that are *broken*.
-# Lang template mis-ordering and lang-template duplicate keys
+# {{lang}} template mis-ordering and lang-template duplicate keys
 # Templates sorted by usage and protect status
 # A 'missing translations' report but for dictionary entries (maybe sorted by usage, too?)
 # A report for "Edits on talkpages (not in the "user talk" namespace) in the past few days", so people can track active discussions?

undocumented_templates.py

@@ -1,6 +1,7 @@
 from re import search, sub
 from utils import pagescraper_queue, plural, time_and_date, whatlinkshere
 from wikitools import wiki
+from wikitools.page import Page
 
 verbose = False
 
@@ -9,12 +10,14 @@ def pagescraper(page, badpages):
   page_visible = sub('<includeonly>.*?</includeonly>', '', page_text)
   if len(page_text) == 0:
     return # Empty templates (usually due to HTTP failures)
-  elif float(len(page_visible)) / len(page_text) > .80:
-    return # Template is self-documenting, as it shows the majority of its contents.
   elif '{{tlx|' in page_visible or '{{tl|' in page_visible:
     return # Page has example usages
   elif search('{{([Dd]oc begin|[Tt]emplate doc|[Dd]ocumentation|[Ww]ikipedia doc|[dD]ictionary/wrapper)}}', page_visible):
     return # Page uses a documentation template
+  elif '{{{' not in page_text:
+    return # Page does not have any arguments 
+  elif not search('{{{[a-zA-Z0-9]+}}}', page_visible):
+    return # All of the arguments have defaults
 
   count = page.get_transclusion_count()
   if count > 0:
@@ -23,13 +26,26 @@ def pagescraper(page, badpages):
     badpages.append([count, page.title])
 
 def main(w):
+  navbox_templates = []
+  navbox = Page(w, 'Template:Navbox')
+  for page in navbox.get_transclusions(namespaces=['Template']):
+    if page.title.lower().startswith('template:navbox'):
+      continue # Exclude alternative navbox templates
+    if page.title.lower().endswith('sandbox'):
+      continue # Sandboxes link to pages but shouldn't be used
+    if 'navbox' not in page.get_wiki_text().lower():
+      continue # Some template pages actually *use* other navboxes, but are not one themselves.
+    navbox_templates.append(page.title)
+
   badpages = []
   with pagescraper_queue(pagescraper, badpages) as page_q:
     for page in w.get_all_templates():
       if '/' in page.title:
         continue # Don't include subpage templates like Template:Dictionary or Template:PatchDiff
       elif page.title[:13] == 'Template:User':
         continue # Don't include userboxes.
+      elif page.title in navbox_templates:
+        continue # Don't include navboxes, they are self-documenting.
       page_q.put(page)
 
   badpages.sort(key=lambda s: (-s[0], s[1]))