From 752d156cab9b2fcbb825f0cd8e872153fcfe67e6 Mon Sep 17 00:00:00 2001 From: Cassard Mathieu Date: Sat, 28 Dec 2013 15:22:36 +0100 Subject: [PATCH 1/4] Fix date parsing in RegionCentre.py --- jobboards/RegionCentre.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/jobboards/RegionCentre.py b/jobboards/RegionCentre.py index 90a5785..c1061ef 100755 --- a/jobboards/RegionCentre.py +++ b/jobboards/RegionCentre.py @@ -37,7 +37,7 @@ def fetch_url(self, url): xmldoc = minidom.parse(os.path.join(self.processingDir, filename)) MainPubDate = xmldoc.getElementsByTagName('pubDate')[0].firstChild.data - epochPubDate = datetime.datetime.strptime(MainPubDate, "%a, %d %b %Y %H:%M:%S +0200").strftime('%s') + epochPubDate = datetime.datetime.strptime(MainPubDate, "%a, %d %b %Y %H:%M:%S +0100").strftime('%s') print "main date " + MainPubDate # if (epochPubDate <= self.lastFetchDate): From 81e684cbd7ae799875335eced05a1c176888afed Mon Sep 17 00:00:00 2001 From: Cassard Mathieu Date: Sat, 28 Dec 2013 16:11:09 +0100 Subject: [PATCH 2/4] Dirty encoding fix on RegionSudOuest --- jobboards/RegionSudOuest.py | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/jobboards/RegionSudOuest.py b/jobboards/RegionSudOuest.py index 260f328..4ff197f 100755 --- a/jobboards/RegionSudOuest.py +++ b/jobboards/RegionSudOuest.py @@ -34,7 +34,10 @@ def fetch_url(self, url): filename = url.split('/')[-1] utilities.download_file(url, self.processingDir) - xmldoc = minidom.parse(os.path.join(self.processingDir, filename)) + fileXML = open(os.path.join(self.processingDir, filename), 'r+') + fileXML = fileXML.read() + fileXML = fileXML.replace('US-ASCII','iso-8859-1') + xmldoc = minidom.parseString(fileXML) MainPubDate = xmldoc.getElementsByTagName('pubDate')[0].firstChild.data epochPubDate = datetime.datetime.strptime(MainPubDate, "%a, %d %b %Y %H:%M:%S +0200").strftime('%s') From efa6430cd2df56245ff74d97b7f573d8ec68ae17 Mon Sep 17 00:00:00 2001 From: Cassard Mathieu Date: Sat, 28 Dec 2013 16:11:38 +0100 Subject: [PATCH 3/4] Fix date parsing in RegionSudOuest.py --- jobboards/RegionSudOuest.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/jobboards/RegionSudOuest.py b/jobboards/RegionSudOuest.py index 4ff197f..2be8ccd 100755 --- a/jobboards/RegionSudOuest.py +++ b/jobboards/RegionSudOuest.py @@ -40,7 +40,7 @@ def fetch_url(self, url): xmldoc = minidom.parseString(fileXML) MainPubDate = xmldoc.getElementsByTagName('pubDate')[0].firstChild.data - epochPubDate = datetime.datetime.strptime(MainPubDate, "%a, %d %b %Y %H:%M:%S +0200").strftime('%s') + epochPubDate = datetime.datetime.strptime(MainPubDate, "%a, %d %b %Y %H:%M:%S +0100").strftime('%s') print "main date " + MainPubDate # if (epochPubDate <= self.lastFetchDate): From 8720f9f64c9cacd0470549bfef779cc446dade63 Mon Sep 17 00:00:00 2001 From: Cassard Mathieu Date: Sat, 28 Dec 2013 16:11:56 +0100 Subject: [PATCH 4/4] Fix date parsing in RegionOuest.py --- jobboards/RegionOuest.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/jobboards/RegionOuest.py b/jobboards/RegionOuest.py index 0db74b8..877079f 100755 --- a/jobboards/RegionOuest.py +++ b/jobboards/RegionOuest.py @@ -37,7 +37,7 @@ def fetch_url(self, url): xmldoc = minidom.parse(os.path.join(self.processingDir, filename)) MainPubDate = xmldoc.getElementsByTagName('pubDate')[0].firstChild.data - epochPubDate = datetime.datetime.strptime(MainPubDate, "%a, %d %b %Y %H:%M:%S +0200").strftime('%s') + epochPubDate = datetime.datetime.strptime(MainPubDate, "%a, %d %b %Y %H:%M:%S +0100").strftime('%s') print "main date " + MainPubDate # if (epochPubDate <= self.lastFetchDate):