Merge ~rodrigo-zaiden/ubuntu-security-tools:backup-wiki into ubuntu-security-tools:master

[Rodrigo Figueiredo Zaiden]

backupwiki: Add Include Macro pattern to search for included pages

When a wiki page uses Include macro, it includes a page inside
another. We would like to verify if the included page should be
added to the backup list

Adding the pattern '<<Include\((\w+(\/\w+){0,}).*\)>>' that will search for the Include macro and select a page candidate to be backed up (example: SecurityTeam/Page1/Page2/.../PageN)

Unmerged commits

f101b55... by Rodrigo Figueiredo Zaiden on 2022-01-04

backupwiki: Add Include Macro pattern to search for included pages

When a wiki page uses Include Macro, it includes a page inside
another. We would like to verify if the included page should be
added to the backup list

diff --git a/utilities/backup_ubuntu_wiki.py b/utilities/backup_ubuntu_wiki.py
index 558c173..c5c8403 100755
--- a/utilities/backup_ubuntu_wiki.py
+++ b/utilities/backup_ubuntu_wiki.py
@@ -17,7 +17,7 @@ from urllib.parse import urlparse, urljoin
 import requests
 
 host = 'wiki.ubuntu.com'
-backup_dir = "/home/steve/backups/" + host
+backup_dir = os.getenv("HOME") + "/backups/" + host
 
 subtrees = [
     'Security', 'SecurityTeam', 'AppArmor', 'DebuggingSecurity',
@@ -208,8 +208,13 @@ def main():
 
         urls_saved.add(next_page)
 
-        for m in re.finditer('\[\[\s*([\w/.:]+)\s*(\|\s*[^\]]*\s*)?\]\]', contents):
+        for m in re.finditer('\[\[\s*([\w/.:]+)\s*(\|\s*[^\]]*\s*)?\]\]|<<Include\((\w+(\/\w+){0,}).*\)>>', contents):
             link = m.group(1)
+            if link == None:
+                link = m.group(3)
+                if config.debug:
+                    print(f'Found Include macro with {link} (referrer: {next_page})', file=sys.stderr)
+
             # Ugh, mediawiki converts everything to have the first
             # letter be uppercase; need to convert references
             #if link[0] in string.lowercase: