Explorar el Código

Fix the Lewisham scraper.

It looks like there has been an update to most of the Acolnet sites.
I'll need to go through and look at the rest...
import/raw
duncan.parkes hace 17 años
padre
commit
bd05c50ae6
Se han modificado 2 ficheros con 6 adiciones y 3 borrados
  1. +5
    -2
      trunk/python_scrapers/AcolnetParser.py
  2. +1
    -1
      trunk/python_scrapers/SitesToGenerate.csv

+ 5
- 2
trunk/python_scrapers/AcolnetParser.py Ver fichero

@@ -311,6 +311,9 @@ class BoltonParser(AcolnetParser):
def _getCouncilReference(self, app_table):
return app_table.findAll("a")[1].string.strip()
class LewishamParser(AcolnetParser):
def _getCouncilReference(self, app_table):
return app_table.findAll("a")[1].string.strip()
if __name__ == '__main__':
day = 22
@@ -330,11 +333,11 @@ if __name__ == '__main__':
#parser = AcolnetParser("East Lindsey", "East Lindsey", "http://www.e-lindsey.gov.uk/planning/AcolnetCGI.exe?ACTION=UNWRAP&RIPNAME=Root.pgesearch", "AcolnetParser")
#parser = AcolnetParser("Exeter City Council", "Exeter", "http://pub.exeter.gov.uk/scripts/Acolnet/dataonlineplanning/acolnetcgi.gov?ACTION=UNWRAP&RIPNAME=Root.pgesearch")
#parser = AcolnetParser("Fylde", "Fylde", "http://www2.fylde.gov.uk/planning/acolnetcgi.gov?ACTION=UNWRAP&RIPNAME=Root.pgesearch")
parser = AcolnetParser("Guildford", "Guildford", "http://www.guildford.gov.uk/DLDC_Version_2/acolnetcgi.exe?ACTION=UNWRAP&RIPNAME=Root.pgesearch")
#parser = AcolnetParser("Guildford", "Guildford", "http://www.guildford.gov.uk/DLDC_Version_2/acolnetcgi.exe?ACTION=UNWRAP&RIPNAME=Root.pgesearch")
#parser = AcolnetParser("Harlow", "Harlow", "http://planning.harlow.gov.uk/PlanningSearch/acolnetcgi.exe?ACTION=UNWRAP&RIPNAME=Root.pgesearch")
#parser = AcolnetParser("Havant", "Havant", "http://www3.havant.gov.uk/scripts/planningpages/acolnetcgi.exe?ACTION=UNWRAP&RIPNAME=Root.pgesearch")
#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 = LewishamParser("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 = NewForestDCParser("New Forest District Council", "New Forest DC", "http://web3.newforest.gov.uk/planningonline/acolnetcgi.gov?ACTION=UNWRAP&RIPNAME=Root.pgesearch")
#parser = NewForestNPAParser("New Forest National Park Authority", "New Forest NPA", "http://web01.newforestnpa.gov.uk/planningpages/acolnetcgi.gov?ACTION=UNWRAP&RIPNAME=Root.pgesearch")


+ 1
- 1
trunk/python_scrapers/SitesToGenerate.csv Ver fichero

@@ -98,7 +98,7 @@
"Guildford Borough Council", "Guildford", "http://www.guildford.gov.uk/DLDC_Version_2/acolnetcgi.exe?ACTION=UNWRAP&RIPNAME=Root.pgesearch", "AcolnetParser", "AcolnetParser"
"Harlow Council", "Harlow", "http://planning.harlow.gov.uk/PlanningSearch/acolnetcgi.exe?ACTION=UNWRAP&RIPNAME=Root.pgesearch", "AcolnetParser", "AcolnetParser"
"Hertsmere Borough Council", "Hertsmere", "http://www2.hertsmere.gov.uk/ACOLNET/DCOnline//acolnetcgi.gov?ACTION=UNWRAP&RIPNAME=Root.pgesearch", "AcolnetParser", "AcolnetParser"
"London Borough of Lewisham", "Lewisham", "http://acolnet.lewisham.gov.uk/lewis-xslpagesdc/acolnetcgi.exe?ACTION=UNWRAP&RIPNAME=Root.PgeSearch", "AcolnetParser", "AcolnetParser"
"London Borough of Lewisham", "Lewisham", "http://acolnet.lewisham.gov.uk/lewis-xslpagesdc/acolnetcgi.exe?ACTION=UNWRAP&RIPNAME=Root.PgeSearch", "AcolnetParser", "LewishamParser"
"North Hertfordshire District Council", "North Hertfordshire", "http://www.north-herts.gov.uk/dcdataonline/Pages/acolnetcgi.gov?ACTION=UNWRAP&RIPNAME=Root.PgeSearch", "AcolnetParser", "AcolnetParser"
"New Forest National Park", "New Forest NP", "http://web01.newforestnpa.gov.uk/planningpages/acolnetcgi.gov?ACTION=UNWRAP&RIPNAME=Root.pgesearch", "AcolnetParser", "NewForestNPAParser"
"Bridgnorth District Council", "Bridgnorth", "http://www2.bridgnorth-dc.gov.uk/planning/acolnetcgi.gov?ACTION=UNWRAP&RIPNAME=Root.PgeSearch", "AcolnetParser", "BridgnorthParser"


Cargando…
Cancelar
Guardar