From: root Date: Mon, 03 Dec 2012 02:56:21 +0000 Subject: Merge branch 'master' of ssh://apples.lambdacomplex.org/git/disclosr X-Git-Url: https://maxious.lambdacomplex.org/git/?p=disclosr.git&a=commitdiff&h=e2b91982ab62f9f22d63338fea8e9f462d4004ca --- Merge branch 'master' of ssh://apples.lambdacomplex.org/git/disclosr Former-commit-id: fb652de2c04eb0dcdcc435c47141e3a0034e996b --- --- /dev/null +++ b/documents/about.php @@ -1,1 +1,11 @@ + +

About

+ + --- a/documents/genericScrapers.py +++ b/documents/genericScrapers.py @@ -137,7 +137,12 @@ self.getDescription(description,row, doc) if notes != None: doc.update({ 'notes': (''.join(notes.stripped_strings))}) - foidocsdb.save(doc) + badtitles = ['-','Summary of FOI Request','FOI request(in summary form)','Summary of FOI request received by the ASC', +'Summary of FOI request received by agency/minister','Description of Documents Requested','FOI request','Description of FOI Request','Summary of request','Description','Summary', +'Summary of FOIrequest received by agency/minister','Summary of FOI request received','Description of FOI Request',"FOI request",'Results 1 to 67 of 67'] + if doc['title'] not in badtitles and doc['description'] != '': + print "saving" + foidocsdb.save(doc) else: print "already saved "+hash --- a/documents/rss.xml.php +++ b/documents/rss.xml.php @@ -28,7 +28,7 @@ $newItem = $TestFeed->createNewItem(); //Add elements to the feed item $newItem->setTitle($row->value->title); - $newItem->setLink("view.php?id=".$row->value->docID); + $newItem->setLink("view.php?id=".$row->value->_id); $newItem->setDate(date("c", strtotime($row->value->date))); $newItem->setDescription(displayLogEntry($row,$idtoname)); $newItem->addElement('guid', $row->value->_id,array('isPermaLink'=>'true')); --- a/documents/scrapers/227cb6eb7d2c9f8a6e846df7447d6caa.py +++ b/documents/scrapers/227cb6eb7d2c9f8a6e846df7447d6caa.py @@ -21,9 +21,10 @@ for row in soup.find(class_ = "ms-rteTable-GreyAlternating").find_all('tr'): if row != None: rowtitle = row.find('th').string - description = description + "\n" + rowtitle + ": " + if rowtitle != None: + description = description + "\n" + rowtitle + ": " for text in row.find('td').stripped_strings: - description = description + text + description = description + text for atag in row.find_all("a"): if atag.has_key('href'): links.append(scrape.fullurl(link,atag['href'])) --- a/documents/template.inc.php +++ b/documents/template.inc.php @@ -89,7 +89,7 @@