tas_councils/scrapers/kentish.rb

# Kentish Council — Advertised / Planning Applications (site page, not PlanBuild)

require "nokogiri"
require "uri"
require "cgi"

require_relative "../lib/enrich"
require_relative "../lib/log"
TABLE = ENV.fetch("TABLE_NAME")              # run_all.sh -> da_kentish
# Set this to the exact page you use for Kentish (from your original file)
URL   = "https://www.kentish.tas.gov.au/services/building-and-planning-services/planningapp"

DB.ensure_table!(TABLE)

def abs_url(base, href)
  return "" if href.to_s.strip.empty?
  URI.join(base, href).to_s rescue href.to_s
end

# Reference formats like:
#   DA 2025/00123
#   DA2025/00123
#   Application No. DA 2025/123
REF_RX1 = %r{\bDA\s*(20\d{2})\s*/\s*([A-Za-z0-9\-\._]+)}i      # DA 2025/0123
REF_RX2 = %r{\bDA(20\d{2})\s*[-\/]?\s*([0-9]{3,})\b}i          # DA2025-0123 or DA2025/0123
REF_RX3 = %r{\bDA\s*([0-9]{1,4})\s*-\s*(20\d{2})\b}i           # DA 114-2025

def extract_ref(str)
  s = CGI.unescape(str.to_s)
  if (m = s.match(REF_RX1))
    return "DA #{m[1]} / #{m[2]}"
  end
  if (m = s.match(REF_RX2))
    return "DA #{m[1]} / #{m[2]}"
  end
  if (m = s.match(REF_RX3))
    return "DA #{m[2]} / #{m[1]}"
  end
  nil
end

DATE_RX = /
  (\b\d{1,2}\/\d{1,2}\/\d{2,4}\b|
   \b\d{1,2}\s+[A-Za-z]{3,}\s+\d{4}\b|
   \b[A-Za-z]{3,}\s+\d{1,2},?\s+\d{4}\b)
/x

def extract_on_notice_raw(text)
  s = text.to_s.gsub(/\s+/, " ")

  if (m = s.match(/\bon\s*notice\s*(until|to)\s*[:\-]?\s*([A-Za-z0-9\/ ,]+)/i))
    if (d = m[2].match(DATE_RX))
      return d[1]
    end
  end

  if (m = s.match(/clos(?:e|ing|es)\s*(on)?\s*[:\-]?\s*([A-Za-z0-9\/ ,]+)/i))
    if (d = m[2].match(DATE_RX))
      return d[1]
    end
  end

  if (d = s.match(DATE_RX))
    return d[1]
  end

  ""
end

def first_meaningful_text(node)
  return "" unless node
  t = node.text.to_s.strip.gsub(/\s+/, " ")
  t
end

def nearest_context_text(a)
  host = a.ancestors("li, p, div, tr").first || a.parent
  first_meaningful_text(host)
end

def parse_document_list(doc, base_url)
  # Look for clear “items”: pdf links, or list/table rows containing one
  anchors = doc.css("a").select { |a|
    href = a["href"].to_s
    a.text.to_s.strip.match?(/application|permit|advertis/i) || href.downcase.end_with?(".pdf")
  }

  rows = []
  anchors.each do |a|
    href = a["href"].to_s
    pdf  = abs_url(base_url, href)
    ctx  = nearest_context_text(a)
    link_text = a.text.to_s.strip

    text_for_parse = [link_text, ctx].uniq.join(" — ")

    # Try to pull fields
    ref  = extract_ref(text_for_parse)
    addr = if link_text.length > 6
      link_text
    else
      ctx[0, 140]
    end

    on_raw = extract_on_notice_raw(text_for_parse)
    on_dt  = Util.parse_aus_date(on_raw)

    desc = if text_for_parse =~ /proposal\s*[:\-]\s*([^—\-]+)\b/i
      $1.strip
    else
      "Development Application"
    end

    next if ref.nil? || addr.to_s.strip.empty?

    rows << {
      council_reference: ref,
      address: addr.to_s.strip,
      description: desc,
      date_received: on_dt,
      date_received_raw: on_raw,
      document_url: pdf
    }
  end

  rows
end

begin
  html = Http.get(URL)
rescue StandardError => e
  Log.warn "scraper", "Failed to fetch #{URL}: #{e.class} #{e.message}"
  exit 1
end

doc = Nokogiri::HTML(html)
items = parse_document_list(doc, URL)

puts "Found #{items.length} item(s) for #{TABLE}"

items.each do |r|
  DB.upsert(TABLE, {
    description:       r[:description],
    date_received:     r[:date_received],
    date_received_raw: r[:date_received_raw],
    on_notice_to:      r[:date_received],
    on_notice_to_raw:  r[:date_received_raw],
    address:           r[:address],
    council_reference: r[:council_reference],
    document_url:      r[:document_url],
    applicant:         "",
    owner:             ""
  })

  enrich_after_upsert!(
    table:             TABLE,
    council_reference: r[:council_reference],
    address:           r[:address]
  )

  puts "Upserted #{r[:council_reference]} -> #{r[:address]}"
end

puts "Done #{TABLE}."