1
0
mirror of https://github.com/cookiengineer/audacity synced 2025-06-15 15:49:36 +02:00

Bug 1370 - mw2html: Filipino pages present.

Now excludes links containing /xx/, /xxx/ and  /xx_XX/.  Case sensitive, otherwise we exclude /FAQ/ and /SC4/
This commit is contained in:
James Crook 2016-06-22 19:18:30 +01:00
parent de9579dbe9
commit b4d6f5f03a

View File

@ -343,11 +343,13 @@ def html_remove_image_history(doc):
def html_remove_translation_links(doc):
"""
Remove translation links (the international flags).
We identify them by the pattern for a 2 letter language code, /[\s\S][\s\S][/"]
We identify them by the pattern for a 2 or 3 letter language code, /[a-z]{2,3}[/"]
in the URL.
The second version deals with links like /pt_PT and /zh_CN
We are case sensitive, so as not to treat FAQ as a language code.
"""
doc = re.sub(r'<a href="[^"]+/[\s\S][\s\S][/"][\s\S]+?</a>', r'<!--Removed Translation Flag-->', doc)
doc = re.sub(r'<a href="[^"]+/[\s\S][\s\S]_[\s\S][\s\S][/"][\s\S]+?</a>', r'<!--Removed Translation Flag2-->', doc)
doc = re.sub(r'<a href="[^"]+/[a-z]{2,3}[/"][\s\S]+?</a>', r'<!--Removed Translation Flag-->', doc)
doc = re.sub(r'<a href="[^"]+/[a-z]{2}_[A-Z]{2}[/"][\s\S]+?</a>', r'<!--Removed Translation Flag2-->', doc)
return doc
def monobook_hack_skin_html(doc):