When an article is updated, save the update.
authorNeal H. Walfield <neal@walfield.org>
Sat, 5 Nov 2011 11:03:15 +0000 (12:03 +0100)
committerNeal H. Walfield <neal@walfield.org>
Sat, 5 Nov 2011 11:03:15 +0000 (12:03 +0100)
src/rss_sqlite.py

index 4b4d276..ed6ac3b 100644 (file)
@@ -456,8 +456,6 @@ class Feed(BaseObject):
                #reversedEntries = self.getEntries()
                #reversedEntries.reverse()
     
-               ids = self.getIds()
-    
                tmp["entries"].reverse()
                for entry in tmp["entries"]:
                    # Yield so as to make the main thread a bit more
@@ -493,9 +491,9 @@ class Feed(BaseObject):
                                 "date":date, "link":entry["link"], "author":entry["author"], "id":entry["id"]}
                    id = self.generateUniqueId(tmpEntry)
                    
-                   current_version \
-                       = self.db.execute('select date from feed where id=?',
-                                         (id,)).fetchone()
+                   current_version = self.db.execute(
+                       'select date, ROWID from feed where id=?',
+                       (id,)).fetchone()
                    if (current_version is not None
                        and current_version[0] == date):
                        logger.debug("ALREADY DOWNLOADED %s (%s)"
@@ -506,7 +504,6 @@ class Feed(BaseObject):
                        # The version was updated.  Mark it as unread.
                        logger.debug("UPDATED: %s (%s)"
                                     % (entry["title"], entry["link"]))
-                       self.setEntryUnread(id)
                        updated_objects += 1
                    else:
                        logger.debug("NEW: %s (%s)"
@@ -517,7 +514,6 @@ class Feed(BaseObject):
                    soup = BeautifulSoup(self.getArticle(tmpEntry)) #tmpEntry["content"])
                    images = soup('img')
                    baseurl = tmpEntry["link"]
-                   #if not id in ids:
                    if imageCache and len(images) > 0:
                        self.serial_execution_lock.release ()
                        have_serial_execution_lock = False
@@ -547,29 +543,27 @@ class Feed(BaseObject):
                    file = open(tmpEntry["contentLink"], "w")
                    file.write(soup.prettify())
                    file.close()
-                   if id in ids:
-                       self.db.execute("UPDATE feed SET updated=? WHERE id=?;", (currentTime, id) )
-                       self.db.commit()
-                   else:
-                       values = (id, tmpEntry["title"], tmpEntry["contentLink"], tmpEntry["date"], currentTime, tmpEntry["link"], 0)
-                       self.db.execute("INSERT INTO feed (id, title, contentLink, date, updated, link, read) VALUES (?, ?, ?, ?, ?, ?, ?);", values)
-                       self.db.commit()
-#                   else:
-#                       try:
-#                           self.db.execute("UPDATE feed SET updated=? WHERE id=?;", (currentTime, id) )
-#                           self.db.commit()
-#                           filename = configdir+self.key+".d/"+id+".html"
-#                           file = open(filename,"a")
-#                           utime(filename, None)
-#                           file.close()
-#                           images = self.db.execute("SELECT imagePath FROM images where id=?;", (id, )).fetchall()
-#                           for image in images:
-#                                file = open(image[0],"a")
-#                                utime(image[0], None)
-#                                file.close()
-#                       except:
-#                           pass
-    
+
+                   values = {'id': id,
+                             'title': tmpEntry["title"],
+                             'contentLink': tmpEntry["contentLink"],
+                             'date': tmpEntry["date"],
+                             'updated': currentTime,
+                             'link': tmpEntry["link"],
+                             'read': 0}
+
+                   if current_version is not None:
+                       # This is an update.  Ensure that the existing
+                       # entry is replaced.
+                       values['ROWID'] = current_version[1]
+
+                   cols, values = zip(*values.items())
+                   self.db.execute(
+                       "INSERT OR REPLACE INTO feed (%s) VALUES (%s);"
+                       % (','.join(cols), ','.join(('?',) * len(values))),
+                       values)
+                   self.db.commit()
+
                    # Register the object with Woodchuck and mark it as
                    # downloaded.
                    def register_object_transferred(