Adds a functional TAMES scraper using a similar API to the old system scraper.

data/texappscraper/courts.yml

@@ -1,4 +1,8 @@
 ---
+1:
+  name: First Court of Appeals
+  city: Houston
+  scraper: TAMESScraper
 2:
   name: Second Court of Appeals
   city: Fort Worth
@@ -9,6 +13,18 @@
   city: Austin
   scraper: OldSystemScraper
   site: http://www.3rdcoa.courts.state.tx.us
+4:
+  name: Fourth Court of Appeals
+  city: San Antonio
+  scraper: TAMESScraper
+5:
+  name: Fifth Court of Appeals
+  city: Dallas
+  scraper: TAMESScraper
+6:
+  name: Sixth Court of Appeals
+  city: Texarkana
+  scraper: TAMESScraper
 7:
   name: Seventh Court of Appeals
   city: Amarillo
@@ -19,6 +35,10 @@
   city: El Paso
   scraper: OldSystemScraper
   site: http://www.8thcoa.courts.state.tx.us
+9:
+  name: Ninth Court of Appeals
+  city: Texarkana
+  scraper: TAMESScraper
 10:
   name: Tenth Court of Appeals
   city: Waco
@@ -29,3 +49,15 @@
   city: Eastland
   scraper: OldSystemScraper
   site: http://www.11thcoa.courts.state.tx.us
+12:
+  name: Twelfth Court of Appeals
+  city: Tyler
+  scraper: TAMESScraper
+13:
+  name: Thirteenth Court of Appeals
+  city: Corpus Christi/Edunburg
+  scraper: TAMESScraper
+14:
+  name: Fourteenth Court of Appeals
+  city: Houston
+  scraper: TAMESScraper

lib/texappscraper.rb

@@ -1,5 +1,6 @@
 require "texappscraper/version"
 require "texappscraper/old_system_scraper"
+require "texappscraper/tames_scraper"
 require "texappscraper/courts"
 require "texappscraper/cacher"
 

lib/texappscraper/tames_scraper.rb

@@ -0,0 +1,111 @@
+require 'texappscraper/throttled_agent'
+require 'texappscraper/courts'
+
+module TexAppScraper
+  class TAMESScraper
+    THROTTLE_DELAY = 3
+
+    BASE = 'http://www.search.txcourts.gov'
+
+    def initialize court, delay=THROTTLE_DELAY
+      delay = 0 if delay == :no_throttling
+      @court = court
+      @agent = ThrottledAgent.new delay
+      @court_number = court
+      @court = TexAppScraper::COURTS[court]
+    end
+
+    def scrape date
+      cases_with_opinions_on_day(date).each do |docket_number|
+        opinions_for_case(docket_number) do |o|
+          yield o
+        end
+      end
+    end
+
+    DAY_KEY = 'FullDate'
+    CASE_LINK_RE = %r{Case\.aspx\?cn=(\d\d-\d\d-\d\d\d\d\d-..)}
+    def cases_with_opinions_on_day day
+      url = "#{BASE}/Docket.aspx"
+      params = {
+        :coa => format("coa%02d", @court_number),
+        :FullDate => day.strftime('%m/%d/%Y'),
+        :p => 1
+      }
+      page = @agent.get url, params
+      page.links_with(:href => CASE_LINK_RE).map do |link|
+        CASE_LINK_RE.match(link.href)[1]
+      end.uniq
+    end
+
+    CASE_URL = "#{BASE}/Case.aspx"
+    def opinions_for_case docket_number
+      page = @agent.get CASE_URL, { :cn => docket_number, :p => 1 }
+      the_case = case_metadata page
+      yield_opinions page do |o|
+        yield ({
+          :case => the_case,
+          :date => o[:date],
+          :type => o[:type],
+          :url => o[:url]
+        })
+      end
+    end
+
+    OPINION_TYPES = {
+      'Opinion issued' => :opinion,
+      'Memorandum opinion issued' => :memorandum
+    }
+    def yield_opinions page
+      page.search('.//tr[@class="rgRow" or @class="rgAltRow"]').each do |row|
+        tds = row.search('./td').to_a
+        type = tds[1].text
+        next unless tds.count == 5
+        if OPINION_TYPES.keys.include? type
+          links = tds[4]
+          date = Date.strptime tds[0].text.strip, '%m/%d/%Y'
+          links.search('.//tr').each do |linkrow|
+            label = linkrow.css('td').first.text
+            if /opinion/i =~ label
+              yield ({
+                :type => OPINION_TYPES[type],
+                :date => date,
+                :url => BASE + '/' + linkrow.css('a').attr('href')
+              })
+            end
+          end
+        end
+      end
+    end
+
+    META = '//*[@id="ctl00_ContentPlaceHolder1_tblContent"]//tr[2]/td/table//tr/td/table//tr[3]/td/table//tr/td/table//tr'
+    META_KEYS = {
+      'Case Number:' => :docket_number,
+      'Style:' => :style,
+      'v.:' => :versus
+    }
+    META_FORMAT = {
+      :filed => lambda { |x| Date.strptime(x, '%m/%d/%Y')}
+    }
+    def case_metadata page
+      meta = {}
+      page.search(META).to_a.each do |tr|
+        key = tr.at_css('td.BreadCrumbs').text.strip
+        next unless META_KEYS.keys.include? key
+        value = tr.at_css('td.TextNormal').text.gsub("\u00A0"," ").strip
+        meta_key = META_KEYS[key]
+        format = META_FORMAT[meta_key]
+        meta[meta_key] = format.nil? ? value : format.call(value)
+      end
+      if meta[:versus] && meta[:versus].length > 0
+        meta[:style] = "#{meta[:style]} v. #{meta[:versus]}"
+      end
+      return({
+        :court => @court_number,
+        :docket_number => meta[:docket_number],
+        :style => meta[:style]
+      })
+    end
+
+  end
+end

spec/fixtures/day-1-20130110.yml

@@ -0,0 +1,17 @@
+---
+- 01-11-00488-CV
+- 01-11-00827-CV
+- 01-11-01033-CV
+- 01-12-00211-CV
+- 01-12-00491-CV
+- 01-12-00607-CV
+- 01-12-00951-CV
+- 01-12-01021-CV
+- 01-12-01056-CV
+- 01-12-01078-CV
+- 01-11-00729-CR
+- 01-11-00977-CR
+- 01-11-00978-CR
+- 01-11-00979-CR
+- 01-12-00429-CR
+- 01-12-00689-CR