X-Git-Url: https://git.phdru.name/?a=blobdiff_plain;f=reindex_blog.py;h=0f1a8f5f9b185fe34ccecbdf8f2b297f3bd0b3d0;hb=ee424085d4a516ac2e599e66f023f05dd3a7b7b9;hp=c2e16ce32c35029bb31d05f382f46e4d65e55617;hpb=f85b54d41b553cd6d9cdc43a2b16c2675bb99216;p=phdru.name%2Fphdru.name.git

diff --git a/reindex_blog.py b/reindex_blog.py
index c2e16ce..0f1a8f5 100755
--- a/reindex_blog.py
+++ b/reindex_blog.py
@@ -2,7 +2,7 @@
 # -*- coding: koi8-r -*-
 
 __author__ = "Oleg Broytman <phd@phdru.name>"
-__copyright__ = "Copyright (C) 2006-2017 PhiloSoft Design"
+__copyright__ = "Copyright (C) 2006-2021 PhiloSoft Design"
 
 import sys, os
 from Cheetah.Template import Template
@@ -482,6 +482,8 @@ class AbsURLHTMLParser(_HTMLParser):
       self.accumulator += '<a'
       for attrname, value in attrs:
          value = cgi.escape(value, True)
+         if isinstance(value, unicode):
+            value = value.encode('koi8-r')
          if attrname == 'href':
             self.accumulator += ' href="%s"' % urljoin(self.base, value)
          else:
@@ -525,9 +527,11 @@ from rss_20 import rss_20
 from news import NewsItem
 
 if blog_root:
-   baseURL = "http://phdru.name/%s/" % blog_root
+   blog_root_url = blog_root[
+        blog_root.find('/htdocs/phdru.name/') + len('/htdocs/phdru.name/'):]
+   baseURL = "https://phdru.name/%s/" % blog_root_url
 else:
-   baseURL = "http://phdru.name/"
+   baseURL = "https://phdru.name/"
 
 items = []
 for item in tuple(reversed(all_titles_tags))[:10]:
@@ -544,8 +548,17 @@ for item in tuple(reversed(all_titles_tags))[:10]:
    item.categoryList = tags
    body = bodies[(year, month, day, file)]
    body = absolute_urls(body, baseURL + url_path)
+   try:
+       body.decode('utf-8')
+   except UnicodeDecodeError:
+       body = body.decode('koi8-r').encode('utf-8')
    item.body = body
-   item.excerpt = get_first_p(body)
+   excerpt = get_first_p(body)
+   try:
+       excerpt.decode('utf-8')
+   except UnicodeDecodeError:
+       excerpt = excerpt.decode('koi8-r').encode('utf-8')
+   item.excerpt = excerpt
 
 namespace = {
    "title": "Oleg Broytman's blog",