Changeset 18249

Timestamp:

08/26/08 16:40:23 (5 years ago)

Author:

mattn

Message:

* 画面仕様変更に対応
* 絵文字だけのタイトルだと見えないのでtitleにはaltを使用

Location:

websites/wassr-fav.appspot.com

Files:

• wassr-fav.py (modified) (2 diffs)
• wassr-fav.rss (modified) (1 diff)

websites/wassr-fav.appspot.com/wassr-fav.py

@@ -4 +4 @@
 import logging
 import wsgiref.handlers
-from BeautifulSoup import BeautifulSoup
+from BeautifulSoup import BeautifulSoup,Tag
 from google.appengine.api import urlfetch
 from google.appengine.ext import webapp
 from google.appengine.ext.webapp import template
 
-def to_text(soup):
-  ret = ''.join([e for e in soup.recursiveChildGenerator() if isinstance(e,unicode)]).replace(' ', ' ')
-  soup = BeautifulSoup(ret, convertEntities=BeautifulSoup.HTML_ENTITIES)
-  return soup and soup.contents and soup.contents[0].encode('utf-8', 'replace').replace('\n', '') or ''
+def to_text(soup, use_img_alt=False):
+  ret = ""
+  for e in soup.recursiveChildGenerator():
+    if isinstance(e,Tag) and e.name == 'img' and not "icn-balloon" in e['src']:
+      if use_img_alt:
+        ret += "(%s)" % e['alt']
+      else:
+        e['src'] = "http://wassr.jp%s" % e['src']
+        ret += unicode(e)
+    elif isinstance(e,unicode):
+      ret += e.strip()
+  ret = ret.replace(' ', ' ').replace('  ', '\n')
+  #ret = ' '.join([e.strip() for e in soup.recursiveChildGenerator() if isinstance(e,unicode) or (isinstance(e,Tag) and e.name == 'img')]).replace(' ', ' ').replace('  ', '\n')
+  #soup = BeautifulSoup(ret, convertEntities=BeautifulSoup.HTML_ENTITIES)
+  #return soup and soup.contents and soup.contents[0].encode('utf-8', 'replace').replace('\n', '') or ''
+  return ret
 
 class MainPage(webapp.RequestHandler):
@@ -37 +49 @@
       r2 = re.compile(r'(.*)\([^\)]+\) (.*)')
       for page_fav in page_favs:
-        author = r1.sub(lambda x: x.group(1), page_fav.find('p', { 'class' : 'favorited_user noteTxt' }).find('a')['href'])
+        author = r1.sub(lambda x: x.group(1), page_fav.find('form', { 'class' : 'followbutton favorited_user noteTxt' }).find('a')['href'])
         #name = to_text(page_fav.find('p', { 'class' : 'favorited_user noteTxt' }).find('a'))
         template_values['favs'].append({
-          'title' : to_text(page_fav.find('p', { 'class' : 'message description' })).strip(),
+          'title' : to_text(page_fav.find('p', { 'class' : 'message description' }), True).strip(),
           'link' : 'http://wassr.jp%s' % page_fav.find('a', { 'class' : 'MsgDateTime' })['href'],
           'guid' : 'http://wassr.jp%s#%s' % (page_fav.find('a', { 'class' : 'MsgDateTime' })['href'], author),

websites/wassr-fav.appspot.com/wassr-fav.rss

@@ -6 +6 @@
     <description>「{{ user|escape }}」がもらったイイネ</description>{% for fav in favs %}
     <item>
-        <title>「{{ fav.title|escape }}」のイイネ</title>
+        <title>「{{ fav.title }}」のイイネ</title>
         <link>{{ fav.link|escape }}</link>
         <author>{{ fav.author|escape }}</author>