From 98954c038581597a15bc3416ed17a21d7c41e083 Mon Sep 17 00:00:00 2001 From: Adrian Short Date: Fri, 30 Jun 2017 09:26:55 +0100 Subject: [PATCH] Add LA name and GSS code --- scraper.rb | 12 +++++++----- 1 file changed, 7 insertions(+), 5 deletions(-) diff --git a/scraper.rb b/scraper.rb index 47e3820..6300c7d 100644 --- a/scraper.rb +++ b/scraper.rb @@ -9,17 +9,19 @@ require 'date' # Use the column names from planningalerts.org.au: # https://www.planningalerts.org.au/how_to_write_a_scraper +LA_NAME = "Kingston upon Thames" +LA_GSS = "E09000021" # https://mapit.mysociety.org/area/2480.html BASEURL = "https://maps.kingston.gov.uk/propertyServices/planning/" # Parse and save a single planning application def parse(app) record = {} - record['title'] = app.at("h4").inner_text - matches = record['title'].match(/(\d+\/\d+\/\w+)\s+-\s+(.+)/) - record['council_reference'] = matches[1] - record['type'] = matches[2] - + record['la_name'] = LA_NAME + record['la_gss'] = LA_GSS + + record['council_reference'], record['type'] = app.at("h4").inner_text.split(' - ') + app.search("a").each do |link| record['info_url'] = BASEURL + link['href'].strip if link['href'].match(/Details/) record['map_url'] = link['href'].strip if link['href'].match(/\?map=/)