tas_councils/scrapers/hobartcity.rb

# Hobart City Council – PlanBuild "Currently Advertised" scraper
# Table name is injected by run_all.sh as TABLE_NAME=da_hobartcity

require "nokogiri"
require "open-uri"
require_relative "../lib/db"
require_relative "../lib/util"
require_relative "../lib/http"
require_relative "../lib/enrich"

TABLE  = ENV.fetch("TABLE_NAME")
URL    = "https://portal.planbuild.tas.gov.au/external/advertisement/search"
# Optional: restrict results to one LGA (substring match)
COUNCIL_FILTER = ENV.fetch("COUNCIL_FILTER", "Hobart City Council").strip

DB.ensure_table!(TABLE)

html = Http.get(URL)

doc  = Nokogiri::HTML(html)

# PlanBuild markup shifts occasionally. We try a few result wrappers.
result_blocks = doc.css(".advertisement-result, .panel.panel-default, .panel.panel-info, .result-row, .row")

found = 0
result_blocks.each do |blk|
  text = blk.text.strip.gsub(/\s+/, " ")

  # Skip blocks that do not look like a single advertised item
  next unless text.match?(/Application/i) || text.match?(/Reference/i) || text.match?(/Council/i)

  # Extract fields using common column patterns first
  address_el = blk.at_css(".col-xs-8, .col-sm-8, .address, [data-field='address']")
  ref_el     = blk.at_css(".col-xs-4, .col-sm-4, .reference, [data-field='reference']")

  address = address_el&.text&.strip.to_s
  council_reference = ref_el&.text&.strip.to_s

  # Fallbacks from label-value pairs (e.g., "Address: …", "Reference: …")
  if address.empty?
    m = text.match(/Address:\s*(.+?)(?:\s{2,}|Reference:|$)/i)
    address = m[1].strip if m
  end
  if council_reference.empty?
    # m = text.match(/(Application|Reference)\s*(No\.?|Number)?:\s*([A-Za-z0-9\-\./_]+)/i)
    REF_RX = %r{(Application|Reference)\s*(No\.?|Number)?:\s*([A-Za-z0-9\-._/]+)}i
    m = text.match(REF_RX)
    council_reference = (m && m[3]) ? m[3].strip : council_reference
  end

  # Try to find the LGA/council name in the block text
  # Common patterns: "Council: Hobart City Council" or a badge/label nearby
  council_name = nil
  if (m = text.match(/Council:\s*([A-Za-z \-]+Council)/i))
    council_name = m[1].strip
  end
  # Light filter: if a filter is set and we can't see Hobart in this block, skip it
  if COUNCIL_FILTER != "" && council_name && !council_name.include?(COUNCIL_FILTER)
    next
  elsif COUNCIL_FILTER != "" && council_name.nil?
    # If no explicit council field, do a substring check across the block text
    next unless text.include?(COUNCIL_FILTER)
  end

  # Optional extras if present in the block
  # Patterns seen across councils vary, so treat all as best-effort
  description = ""
  if (m = text.match(/(Type of Work|Proposal|Description):\s*(.+?)(?:\s{2,}|Address:|Application|Reference|$)/i))
    description = m[2].strip
  end

  date_received_raw = ""
  if (m = text.match(/(Date Lodged|Date Received|Lodged):\s*([0-9]{1,2}\/[0-9]{1,2}\/[0-9]{2,4})/i))
    date_received_raw = m[2].strip
  end
  date_received = Util.parse_aus_date(date_received_raw)

  # If we still don't have key fields, skip
  next if address.empty? || council_reference.empty?

  DB.upsert(TABLE, {
    description: description,
    date_received: date_received,
    date_received_raw: date_received_raw,
    address: address,
    council_reference: council_reference,
    applicant: "",  # PlanBuild usually doesn't expose these in the list
    owner: ""
  })
  
  enrich_after_upsert!(
    table: TABLE,
    council_reference: council_reference,
    address: address
  )

  puts "Upserted #{council_reference} | #{address}"
  found += 1
end

puts "Done #{TABLE}. Found #{found} item(s)."