Started to make the program language-independent. Fixed a problem in month names.

[phdru.name/phdru.name.git] / reindex_blog.py

diff --git a/reindex_blog.py b/reindex_blog.py
index 74717438d9e99d5629ae04266b10db48be40a595..a4667b01c56edd9901d9fda68f383c80b3b0d647 100755 (executable)
--- a/reindex_blog.py
+++ b/reindex_blog.py
@@ -10,8 +10,9 @@ __copyright__ = "Copyright (C) 2006 PhiloSoft Design"
 
 import sys, os
 
-blog_filename = sys.argv[1]
+blog_data_root = sys.argv[1]
 blog_root = sys.argv[2]
+blog_filename = os.path.join(blog_data_root, "blog_dict.pickle")
 
 try:
    import cPickle as pickle
@@ -44,8 +45,7 @@ for dirpath, dirs, files in os.walk(blog_root):
    if not d.startswith("20") and not d.isdigit():
       continue
    for file in files:
-      # Ignore index.tmpl and *.html files; supose all other files are *.tmpl
-      if file == "index.tmpl" or file.endswith(".html"):
+      if not file.endswith(".tmpl"):
          continue
       fullpath = os.path.join(dirpath, file)
       template = Template(file=fullpath)
@@ -53,7 +53,7 @@ for dirpath, dirs, files in os.walk(blog_root):
       title = ' '.join(title_parts[6:])
       lead = getattr(template, "Lead", None)
 
-      tags = getattr(template, "Tag", None)
+      tags = template.Tag
       if isinstance(tags, basestring):
          tags = (tags,)
 
@@ -127,6 +127,9 @@ def write_template(level, year, month, day, titles, tags=None):
 #attr $Title = "Oleg BroytMann's blog"
 #attr $Description = "BroytMann Russian Blog Index Document"
 #attr $Copyright = %(cyear)s
+#attr $alternates = (("application/atom+xml", "News [Atom 1.0]", "atom_10.xml"),
+                     ("application/rss+xml",  "News [RSS 2.0]",  "rss_20.xml")
+)
 ##
 #def body_html
 <H1>Журнал</H1>
@@ -177,11 +180,11 @@ def write_template(level, year, month, day, titles, tags=None):
 #attr $Copyright = %(cyear)s
 ##
 #def body_html
-<H1>Журнал: %(day)d %(month_name_ru0)s %(year)s</H1>
+<H1>Журнал: %(day)d %(month_name_ru)s %(year)s</H1>
 """ % {
       "year": year, "cyear": year or 2005,
       "month_abbr_en": months_abbrs_en[imonth], "month_name_en": months_names_en[imonth],
-      "month_name_ru0": months_names_ru0[imonth],
+      "month_name_ru": months_names_ru[imonth],
       "day": iday
    })
 
@@ -217,13 +220,20 @@ def write_template(level, year, month, day, titles, tags=None):
 ''' % (lead, href, title))
 
    if level == 0:
+      new_text.append("""
+<hr>
+
+<p class="head">Новостевая лента в форматах
+<A HREF="atom_10.xml">Atom 1.0 <img src="../../Graphics/atom_10.jpg" border=0></A>
+и <A HREF="rss_20.xml">RSS 2.0 <img src="../../Graphics/rss_20.jpg" border=0></A>.
+</p>
+""")
+
       years = {}
       for year, month, day, file, title, lead in save_titles:
          years[year] = True
       new_text.append('''
-<hr>
-
-<p class="noindent"><a href="tags/">Теги</a>:
+<p class="head"><a href="tags/">Теги</a>:
 ''')
       first_tag = True
       for count, tag, links in all_tags:
@@ -237,7 +247,7 @@ def write_template(level, year, month, day, titles, tags=None):
 ''')
 
       new_text.append('''
-<p class="noindent">По годам:
+<p class="head">По годам:
 ''')
       first_year = True
       for year in sorted(years.keys()):
@@ -250,6 +260,11 @@ def write_template(level, year, month, day, titles, tags=None):
 </p>
 ''')
 
+      new_text.append("""
+<hr>
+<p class="head"><a href="http://phd.livejournal.com/">ЖЖ</a>
+""")
+
    new_text.append("""\
 #end def
 $phd_pp_ru.respond(self)
@@ -289,8 +304,17 @@ for year in sorted(years.keys()):
       write_template(2, year, month, day, month_titles)
    write_template(1, year, month, day, year_titles)
 
+def by_count_rev_tag_link(t1, t2):
+   """Sort all_tags by count in descending order,
+   and by tags and links in ascending order
+   """
+   r = cmp(t1[0], t2[0])
+   if r:
+      return -r
+   return cmp((t1[1], t1[2]), (t2[1], t2[2]))
+
 all_tags = [(len(links), tag, links) for (tag, links) in all_tags.items()]
-all_tags.sort()
+all_tags.sort(by_count_rev_tag_link)
 
 write_template(0, year, month, day, all_titles[-20:], all_tags)
 
@@ -365,7 +389,7 @@ from atom_10 import atom_10
 from rss_20 import rss_20
 from news import NewsItem
 
-baseURL = "http://phd.pp.ru/Russian/blog/"
+baseURL = "http://phd.pp.ru/" + blog_root
 
 items = []
 for item in tuple(reversed(all_titles_tags))[:10]:
@@ -391,9 +415,13 @@ namespace = {
    "lang": "ru",
    "author": "Oleg Broytmann",
    "email": "phd@phd.pp.ru",
+   "generator": os.path.basename(sys.argv[0]),
    "posts": items,
 }
 
+# For english dates
+locale.setlocale(locale.LC_TIME, 'C')
+
 atom_tmpl = str(atom_10(searchList=[namespace]))
 write_if_changed(os.path.join(blog_root, "atom_10.xml"), atom_tmpl)
 rss_tmpl = str(rss_20(searchList=[namespace]))