Commit: dfb2a0929cc61bf4d186cfed707e85d67e04e92a - newsfour (git)

news4 - RSS aggrigation system

Commit MetaInfo

• Révision: dfb2a0929cc61bf4d186cfed707e85d67e04e92a (tree)
• l'heure: 2012-09-24 22:43:51
• Auteur: hylom <hylom@hylo...>
• Commiter: hylom

Message de Log

add filter, fix css

Change Summary

• modified: README
• modified: css/gnews.css
• modified: fetcher.py
• add: filters/__init__.py
• add: filters/slashdotjp.py
• modified: gnews.py
• modified: renderer.py
• modified: templates/index.tmpl.html

Modification

--- a/README
+++ b/README
@@ -4,4 +4,7 @@ requires:
 	mako (Makotenplate library for Python)
 
 	feedparser (feed library for Python)
-	http://code.google.com/p/feedparser
\ No newline at end of file
+	http://code.google.com/p/feedparser
+
+        python-dateutil
+        
\ No newline at end of file

--- a/css/gnews.css
+++ b/css/gnews.css
@@ -3,6 +3,28 @@
 .entry-header{}
 .entry-body{}
 .entry-continue {
-	margin-bottom: 1em;
+  margin-bottom: 1em;
 }
-.entry-footer{}
\ No newline at end of file
+.entry-footer{
+  color: gray;
+}
+
+header {
+  border-bottom: 1px solid gray;
+}
+
+.main-contents h3 {
+  font-size: 130%;
+}
+
+.sidebar-left h3 {
+  font-size: 100%;
+  font-weight: normal;
+}
+
+.sidebar-right h3 {
+  font-size: 100%;
+  font-weight: normal;
+}
+
+  
\ No newline at end of file

--- a/fetcher.py
+++ b/fetcher.py
@@ -1,6 +1,10 @@
+#-*- coding: utf-8 -*-
 'fetcher.py - RSS fetcher'
 
+import re
+
 import feedparser
+from config import config as config, target_rss as target_rss
 
 class FeedFetcher(object):
     'Feed fetching and parsing'
@@ -13,6 +17,7 @@ class FeedFetcher(object):
         entries = []
         for e in f['entries']:
             entry = {
+#                'title': e.title.decode('utf8') if isinstance(e.title, str) else e.title,
                 'title': e.title,
                 'link': e.link,
                 'body': e.description,
@@ -22,7 +27,41 @@ class FeedFetcher(object):
             entries.append(entry)
         return entries
 
+    def _embeded_filter(self, entry):
+        # remove PR entry
+        if re.search(u'^PR', entry['title']):
+            print 'delete PR entry - %s' % entry['title']
+            return None
+        return entry
+
     def get_entries(self):
         'get entries'
-        return self._fetch()
+        entries = self._fetch()
+        entries = [self._embeded_filter(x) for x in entries]
+        entries = [x for x in entries if x]
+
+        if 'filter' in self._feed:
+            entry_filter = self._get_filter()
+            entries = [entry_filter(x) for x in entries]
+            # remove entry which is None
+            entries = [x for x in entries if x]
+        return entries
+
+    def _get_filter(self):
+        'load filter by seed settings'
+        filter_name = self._feed.get('filter', None)
+
+        # fallback when filter isn't defined
+        if filter_name is None:
+            return lambda x:x
+        
+        # import module
+        mods = __import__(config['filter_directory'], 
+                          globals(),
+                          locals(),
+                          [filter_name,])
+        mod = mods.__getattribute__(filter_name)
 
+        # return module's entry_filter function
+        return mod.entry_filter
+        

--- /dev/null
+++ b/filters/__init__.py
@@ -0,0 +1 @@
+# __init__.py stub

--- /dev/null
+++ b/filters/slashdotjp.py
@@ -0,0 +1,16 @@
+# filter for slashdot.jp
+# -*- coding: utf-8 -*-
+
+import re
+
+
+def entry_filter(entry):
+    # すべて読む、関連ストーリーを削除
+    body = entry['body']
+    body = re.sub(ur'<p> <a href=.*>\s*すべて読む</a>.*?</p>', '', body)
+    body = re.sub(ur'<p>\s*関連ストーリー：.*?</p>', '', body)
+    entry['body'] = body
+
+    return entry
+
+

--- a/gnews.py
+++ b/gnews.py
@@ -8,14 +8,18 @@ import os.path
 
 def main():
     "TODO: argv check"
+
+    # fetch rss feed
     entries = []
     for feed in target_rss:
         f = fetcher.FeedFetcher(feed)
         e = f.get_entries()
         entries.extend(e)
 
+    # sort by date
     entries.sort(lambda x,y: -cmp(x["date"],y["date"]))
 
+    # render entries
     r = renderer.Renderer()
     for (output_file, tmpl) in config["templates"]:
         output_fullpath = os.path.join(

--- a/renderer.py
+++ b/renderer.py
@@ -1,12 +1,19 @@
 'renderer.py - rendering html and some items'
 
+import datetime
+
 from mako.template import Template
 from mako.lookup import TemplateLookup
 from mako.exceptions import RichTraceback
+import dateutil.parser
 
 from config import config, target_rss
 from propertizer import propertize
 
+def date_format(date):
+    dt = dateutil.parser.parse(date)
+    return dt.strftime('%Y/%m/%d %H:%M')
+
 class Renderer(object):
     def __init__(self):
         self.template_dir = config['template_directory']
@@ -25,7 +32,10 @@ class Renderer(object):
         t = self._get_template(template)
 
         
-        kwargs = { 'entries': entries }
+        kwargs = {
+            'entries': entries,
+            'date_format': date_format,
+            }
         kwargs.update(config["site_parameter"])
         for key in kwargs:
             d = propertize(kwargs[key])

--- a/templates/index.tmpl.html
+++ b/templates/index.tmpl.html
@@ -1,44 +1,87 @@
 <!DOCTYPE html>
 <html lang="ja">
   <head>
-	<meta charset="UTF-8">
+    <meta charset="UTF-8">
     <title>${site_name}</title>
     <!-- Bootstrap -->
     <link href="${css_directory}/bootstrap.min.css" rel="stylesheet">
     <link href="${css_directory}/gnews.css" rel="stylesheet">
   </head>
+
   <body class="wrap">
-	<div class="containter">
-	  <div class="row">
-		<div class="span12">
-		  <h1>${site_name}</h1>
-		</div>
-	  </div>
-	  <div class="row">
-		<div class="span3"></div>
-		<div class="span9">
-		  <script src="http://code.jquery.com/jquery-latest.js"></script>
-		  <script src="${js_directory}/bootstrap.min.js"></script>
-
-		  % for entry in entries:
-		  <div class="entry">
-			<div class="entry-header">
-			  <h3>
-				<a href='${entry.link}'  target="_blank_">${entry.title}</a>
-			  </h3>
-			</div>
-			<div class="entry-body">${entry.body}</div>
-			<div class="entry-footer">
-			<div class="entry-continue">
-			  <a href='${entry.link}'>［続きを読む］</a>
-			</div>
-			<div class="entry-footer">
-			  <span>情報元：<a href='${entry.feed.source_url}'>${entry.feed.name}</a></span>
-			  <span>${entry.date}</span></div>
-			</div>
-		  % endfor
-		</div>
-	  </div>
+    <div class="container">
+
+      <!-- タイトル -->
+      <div class="row">
+	<div class="span12">
+          <header>
+	    <h1>${site_name}</h1>
+          </header>
 	</div>
+      </div>
+
+      <!-- コンテンツ本体 -->
+      <div class="row">
+
+        <!-- 左サイドバー -->
+	<div class="span2 sidebar-left">
+          <div class="keywords">
+            <h3>キーワード：</h3>
+            <ul class="nav nav-pills nav-stacked">
+              <li><a href="#">iPhone</a></li>
+              <li><a href="#">iOS</a></li>
+              <li><a href="#">Internet</a></li>
+              <li><a href="#">Firefox</a></li>
+              <li><a href="#">Facebook</a></li>
+            </ul>
+          </div>
+        </div>
+        
+        <!-- メインコンテンツ -->
+	<div class="span7 main-contents">
+	  <script src="http://code.jquery.com/jquery-latest.js"></script>
+	  <script src="${js_directory}/bootstrap.min.js"></script>
+          
+	  % for entry in entries:
+	  <div class="entry">
+            <!-- ヘッダ -->
+	    <div class="entry-header">
+	      <h3>
+		<a href='${entry.link}'  target="_blank_">${entry.title}</a>
+	      </h3>
+	    </div>
+
+
+            <!-- 本文テキスト -->
+	    <div class="entry-body">${entry.body}</div>
+
+            <!-- フッタ -->
+	    <div class="entry-footer">
+	      <div class="entry-continue">
+		<a href='${entry.link}'>［続きを読む］</a>
+	      </div>
+	      <div class="entry-footer">
+		<span>情報元：<a href='${entry.feed.source_url}'>${entry.feed.name}</a></span>
+		<span>（${date_format(entry.date)}）</span></div>
+	    </div>
+	  </div>
+	  % endfor
+
+	</div><!-- main-contents -->
+        
+        <!-- 右サイドバー -->
+	<div class="span3 sidebar-right">
+          <div class="feed-provider">
+            <h3>情報提供サイト:</h3>
+            <ul class="nav nav-pills nav-stacked">
+              <li><a href="#">Slashdot Japan</a></li>
+              <li><a href="#">SourceForge.JP Magazine</a></li>
+            </ul>
+          </div>
+        </div>
+
+      </div><!-- .row -->
+
+    </div><!-- .container -->
   </body>
 </html>