From d54bb032b718e95b9b095b612092fba7cceb90a0 Mon Sep 17 00:00:00 2001 From: "duncan.parkes" Date: Tue, 29 Jan 2008 01:08:58 +0000 Subject: [PATCH] fix: New Forest DC Bolton --- trunk/python_scrapers/AcolnetParser.py | 24 ++++++++++++++++------- trunk/python_scrapers/SitesToGenerate.csv | 4 ++-- 2 files changed, 19 insertions(+), 9 deletions(-) diff --git a/trunk/python_scrapers/AcolnetParser.py b/trunk/python_scrapers/AcolnetParser.py index bea581e..962bbba 100644 --- a/trunk/python_scrapers/AcolnetParser.py +++ b/trunk/python_scrapers/AcolnetParser.py @@ -49,6 +49,7 @@ class AcolnetParser: return soup.findAll("table", {"class": "results-table"}) def _getCouncilReference(self, app_table): +# return app_table.findAll("a")[1].string.strip() return app_table.a.string.strip() def _getDateReceived(self, app_table): @@ -297,19 +298,28 @@ class SurreyHeathParser(AcolnetParser): # Wychavon is rather different, and will need some thought. There is no # advanced search page + +class NewForestDCParser(AcolnetParser): + def _getCouncilReference(self, app_table): + return app_table.findAll("a")[1].string.strip() + +class BoltonParser(AcolnetParser): + def _getCouncilReference(self, app_table): + return app_table.findAll("a")[1].string.strip() + if __name__ == '__main__': - day = 30 - month = 11 - year = 2007 + day = 22 + month = 1 + year = 2008 #parser = AcolnetParser("Babergh", "Babergh", "http://planning.babergh.gov.uk/dcdatav2//acolnetcgi.gov?ACTION=UNWRAP&RIPNAME=Root.pgesearch") #parser = AcolnetParser("Basingstoke", "Basingstoke", "http://planning.basingstoke.gov.uk/DCOnline2/acolnetcgi.exe?ACTION=UNWRAP&RIPNAME=Root.pgesearch") #parser = BassetlawParser("Bassetlaw", "Bassetlaw", "http://www.bassetlaw.gov.uk/planning/acolnetcgi.gov?ACTION=UNWRAP&RIPNAME=Root.pgesearch") - #parser = AcolnetParser("Bolton", "Bolton", "http://www.planning.bolton.gov.uk/PlanningSearch/acolnetcgi.gov?ACTION=UNWRAP&RIPNAME=Root.pgesearch") -# parser = BridgnorthParser("Bridgnorth", "Bridgnorth", "http://www2.bridgnorth-dc.gov.uk/planning/acolnetcgi.gov?ACTION=UNWRAP&RIPNAME=Root.PgeSearch") + #parser = BoltonParser("Bolton", "Bolton", "http://www.planning.bolton.gov.uk/PlanningSearch/acolnetcgi.gov?ACTION=UNWRAP&RIPNAME=Root.pgesearch") + #parser = BridgnorthParser("Bridgnorth", "Bridgnorth", "http://www2.bridgnorth-dc.gov.uk/planning/acolnetcgi.gov?ACTION=UNWRAP&RIPNAME=Root.PgeSearch") #parser = AcolnetParser("Bury", "Bury", "http://e-planning.bury.gov.uk/DCWebPages/acolnetcgi.gov?ACTION=UNWRAP&RIPNAME=Root.pgesearch") - #parser = CanterburyParser("Canterbury", "Canterbury", "http://planning.canterbury.gov.uk/scripts/acolnetcgi.exe?ACTION=UNWRAP&RIPNAME=Root.pgesearch") + parser = CanterburyParser("Canterbury", "Canterbury", "http://planning.canterbury.gov.uk/scripts/acolnetcgi.exe?ACTION=UNWRAP&RIPNAME=Root.pgesearch") #parser = AcolnetParser("Carlisle", "Carlisle", "http://planning.carlisle.gov.uk/acolnet/acolnetcgi.gov?ACTION=UNWRAP&RIPNAME=Root.pgesearch") #parser = AcolnetParser("Croydon", "Croydon", "http://planning.croydon.gov.uk/DCWebPages/acolnetcgi.gov?ACTION=UNWRAP&RIPNAME=Root.pgesearch") #parser = AcolnetParser("Derby", "Derby", "http://eplanning.derby.gov.uk/acolnet/planningpages02/acolnetcgi.exe?ACTION=UNWRAP&RIPNAME=Root.pgesearch") @@ -322,7 +332,7 @@ if __name__ == '__main__': #parser = AcolnetParser("Hertsmere", "Hertsmere", "http://www2.hertsmere.gov.uk/ACOLNET/DCOnline//acolnetcgi.gov?ACTION=UNWRAP&RIPNAME=Root.pgesearch") #parser = AcolnetParser("Lewisham", "Lewisham", "http://acolnet.lewisham.gov.uk/lewis-xslpagesdc/acolnetcgi.exe?ACTION=UNWRAP&RIPNAME=Root.PgeSearch") #parser = AcolnetParser("Mid Suffolk", "Mid Suffolk", "http://planning.midsuffolk.gov.uk/planning/acolnetcgi.gov?ACTION=UNWRAP&RIPNAME=Root.pgesearch") - #parser = AcolnetParser("New Forest District Council", "New Forest DC", "http://web3.newforest.gov.uk/planningonline/acolnetcgi.gov?ACTION=UNWRAP&RIPNAME=Root.pgesearch") + #parser = NewForestDCParser("New Forest District Council", "New Forest DC", "http://web3.newforest.gov.uk/planningonline/acolnetcgi.gov?ACTION=UNWRAP&RIPNAME=Root.pgesearch") #parser = AcolnetParser("New Forest National Park Authority", "New Forest NPA", "http://web01.newforestnpa.gov.uk/planningpages/acolnetcgi.gov?ACTION=UNWRAP&RIPNAME=Root.pgesearch") #parser = AcolnetParser("North Hertfordshire", "North Herts", "http://www.north-herts.gov.uk/dcdataonline/Pages/acolnetcgi.gov?ACTION=UNWRAP&RIPNAME=Root.PgeSearch") #parser = AcolnetParser("North Wiltshire", "North Wilts", "http://planning.northwilts.gov.uk/DCOnline/acolnetcgi.gov?ACTION=UNWRAP&RIPNAME=Root.pgesearch") diff --git a/trunk/python_scrapers/SitesToGenerate.csv b/trunk/python_scrapers/SitesToGenerate.csv index 783e349..37bf8b1 100644 --- a/trunk/python_scrapers/SitesToGenerate.csv +++ b/trunk/python_scrapers/SitesToGenerate.csv @@ -111,7 +111,7 @@ "South Bedfordshire District Council", "South Bedfordshire", "http://planning.southbeds.gov.uk/plantech/DCWebPages/acolnetcgi.exe?ACTION=UNWRAP&RIPNAME=Root.PgeSearch", "AcolnetParser", "AcolnetParser" "Suffolk Coastal District Council", "Suffolk Coastal", "http://apps3.suffolkcoastal.gov.uk/DCDataV2/acolnetcgi.gov?ACTION=UNWRAP&RIPNAME=Root.pgesearch", "AcolnetParser", "AcolnetParser" "Surrey Heath Borough Council", "Surrey Heath", "https://www.public.surreyheath-online.gov.uk/whalecom60b1ef305f59f921/whalecom0/Scripts/PlanningPagesOnline/acolnetcgi.gov?ACTION=UNWRAP&RIPNAME=Root.pgesearch", "AcolnetParser", "SurreyHeathParser" -"New Forest District Council", "New Forest DC", "http://web3.newforest.gov.uk/planningonline/acolnetcgi.gov?ACTION=UNWRAP&RIPNAME=Root.pgesearch", "AcolnetParser", "AcolnetParser" +"New Forest District Council", "New Forest DC", "http://web3.newforest.gov.uk/planningonline/acolnetcgi.gov?ACTION=UNWRAP&RIPNAME=Root.pgesearch", "AcolnetParser", "NewForestDCParser" "Craven District Council", "Craven", "http://www.planning.cravendc.gov.uk/fastweb/", "FastWeb", "FastWeb" "Eastleigh Borough Council", "Eastleigh", "http://www.eastleigh.gov.uk/FastWEB/", "FastWeb", "FastWeb" "Eden District Council", "Eden", "http://eforms.eden.gov.uk/fastweb/", "FastWeb", "FastWeb" @@ -145,7 +145,7 @@ "Cumbria County Council", "Cumbria", "http://217.114.50.149:7778/ePlanningOPS/loadResults.do", "AtriumePlanning", "AtriumePlanningParser" "Lincolnshire County Council", "Lincolnshire", "http://apps1.lincolnshire.gov.uk/ePlanning/loadResults.do", "AtriumePlanning", "AtriumePlanningParser" "West Sussex County Council", "West Sussex", "http://eplanning.westsussex.gov.uk/ePlanningOPS/loadResults.do", "AtriumePlanning", "AtriumePlanningParser" -"Bolton Metropolitan Borough Council", "Bolton", "http://www.planning.bolton.gov.uk/PlanningSearch/acolnetcgi.gov?ACTION=UNWRAP&RIPNAME=Root.pgesearch", "AcolnetParser", "AcolnetParser" +"Bolton Metropolitan Borough Council", "Bolton", "http://www.planning.bolton.gov.uk/PlanningSearch/acolnetcgi.gov?ACTION=UNWRAP&RIPNAME=Root.pgesearch", "AcolnetParser", "BoltonParser" "Exeter City Council", "Exeter", "http://pub.exeter.gov.uk/scripts/Acolnet/dataonlineplanning/acolnetcgi.gov?ACTION=UNWRAP&RIPNAME=Root.pgesearch", "AcolnetParser", "AcolnetParser" "Suffolk County Council", "Suffolk", "http://atrium.suffolkcc.gov.uk/ePlanning/loadResults.do", "AtriumePlanning", "AtriumePlanningParser" "Blackburn with Darwen Borough Council", "Blackburn", "http://195.8.175.6/", "PlanningExplorer", "BlackburnParser"