atp_scraper 0.5.1 → 0.6.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: f7fe84730985d940e8b413c65f1b33cc375c30ec
-  data.tar.gz: ef3f2b93d9fa239d98254e0c272ef4cd3cd6d7e4
+  metadata.gz: f7d4557a6bb017dd9f6607a7a55a7ffd8d53b219
+  data.tar.gz: 1bbae10d70dd2deac57b51c0f86c7b80fda6fc55
 SHA512:
-  metadata.gz: b614765df55a42cfdababaf9c798970d868676fff8be8e7d4b8538d87081d9a0800721dd03231aeec982ff3dd2e187dee53951bca0350cc6ca32c4013cb7148e
-  data.tar.gz: bd0e5db0f6f37e89321c0436b734a59878db052c9ab7b39ef114a4de9ecabc18bf38d2b59737724ff0af651ff684cdef57f7ef9bf7f07fb6eee875a19d4f41a4
+  metadata.gz: b71241b3908338ba1d07c73d708bfd2ea9ffe5c7f3226a17243a296a78c9d5a49a01f3d54499b23f26303f5adbeabf4a3b4985578a231e3d486825e8dd2fa920
+  data.tar.gz: 8e4854126992fab5cd987e38365c150a4976edfd157da2e3dd516fbbfda32d9f013de39678aa55952661782cf4959e404e7afc7caa07f54b1cb03b2ea5b6cf5b

data/README.md

@@ -1,3 +1,4 @@
+[![Gem Version](https://badge.fury.io/rb/atp_scraper.svg)](https://badge.fury.io/rb/atp_scraper)
 [![Build Status](https://travis-ci.org/mosuke5/atp_scraper.svg?branch=master)](https://travis-ci.org/mosuke5/atp_scraper)
 # AtpScraper
 AtpScraper is a tool scraping tennis data from atpworldtour.com
@@ -49,6 +50,7 @@ AtpScraper::Get.player_activity("n409", 2016)
 #   score: "673 64 63 674 26",
 #   win_loss: "L",
 #   tournament_name: "Australian Open",
+#   tournament_category: "grandslam",
 #   tournament_location: "Melbourne, Australia",
 #   tournament_start_date: "2016.01.18",
 #   tournament_end_date: "2016.01.31",

data/lib/atp_scraper/activities/record.rb

@@ -1,7 +1,17 @@
 module Activities
   # Activity Record Class
   class Record
-    def self.pickup_record(record_doc)
+    def initialize(doc)
+      @record = doc
+    end
+
+    def info
+      pickup_record(@record)
+    end
+
+    private
+
+    def pickup_record(record_doc)
       result = {}
       record_doc.css("td").each_with_index do |td, n|
         record_content = td.content.strip

data/lib/atp_scraper/activities/tournament.rb

@@ -1,5 +1,75 @@
 module Activities
   # Activity Tournamnet Class
   class Tournament
+    def initialize(doc)
+      @tournament = doc
+    end
+
+    def info
+      pickup_info(@tournament)
+    end
+
+    def records
+      search_records_doc
+    end
+
+    private
+
+    def pickup_info(tournament_doc)
+      tournament_date = pickup_text(".tourney-dates")
+      surface = pickup_surface
+      caption = pickup_text(".activity-tournament-caption")
+      {
+        name: pickup_text(".tourney-title"),
+        category: pickup_category,
+        location: pickup_text(".tourney-location"),
+        date: divide_tournament_date(tournament_date),
+        year: tournament_date[0, 4],
+        surface: surface[:surface],
+        surface_inout: surface[:inout],
+        ranking: pickup_player_rank(caption)
+      }
+    end
+
+    # Before: String "2011.01.03 - 2011.01.08"
+    # After:  Hash { start: 2011.01.03, end: 2011.01.08 }
+    def divide_tournament_date(date)
+      date = date.split('-').map(&:strip)
+      { start: date[0], end: date[1] }
+    end
+
+    def pickup_text(selector)
+      @tournament.css(selector).first.content.strip
+    end
+
+    def pickup_category
+      # ex) /~/media/images/tourtypes/categorystamps_itf_118x64.png?xxxxx
+      badge_url = @tournament.css(".tourney-badge-wrapper img").attr("src").value
+      badge_url.match(/categorystamps_(.*)_[0-9]*x[0-9]*.png/)[1]
+    end
+
+    def pickup_surface
+      surface = @tournament
+                .css(".tourney-details")[1]
+                .css(".item-details")
+                .first.content.gsub(/\t|\s/, "")
+      divide_surface(surface)
+    end
+    
+    # "OutdoorHard" => { surface: "Hard", inout: "Outdoor" } 
+    def divide_surface(surface)
+      inout = surface.match(/^(Outdoor|Indoor)/)
+      return { surface: surface, inout: nil } if inout.nil?
+      { surface: surface.gsub(/#{inout[0]}/, ''), inout: inout[0] }
+    end
+
+    def pickup_player_rank(tournament_caption)
+      rank = tournament_caption.match(/ATP Ranking:(.+), Prize/)
+      rank[1].strip
+    end
+
+    def search_records_doc
+      @tournament.css(".mega-table tbody tr")
+    end
   end
 end

data/lib/atp_scraper/activity.rb

@@ -12,6 +12,7 @@
 ###################
 
 require 'atp_scraper/activities/record'
+require 'atp_scraper/activities/tournament'
 module AtpScraper
   # Scrape activity data
   class Activity
@@ -23,14 +24,12 @@ module AtpScraper
 
     def pickup_activity_data
       result = []
-      player = {}
 
       search_tournaments_doc(@activity_doc).each do |tournament_doc|
-        tournament = pickup_tournament_info(tournament_doc)
-        player[:rank] = pickup_player_rank(tournament[:caption])
-        search_records_doc(tournament_doc).each do |record_doc|
-          record = Record.pickup_record(record_doc)
-          record_hash = create_record(record, player, tournament)
+        tournament = Tournament.new(tournament_doc)
+        tournament.records.each do |record_doc|
+          record = Record.new(record_doc)
+          record_hash = create_record(record.info, tournament.info)
           result.push(record_hash)
         end
       end
@@ -43,21 +42,18 @@ module AtpScraper
       activity_doc.css(".activity-tournament-table")
     end
 
-    def search_records_doc(tournament_doc)
-      tournament_doc.css(".mega-table tbody tr")
-    end
-
-    def create_record(record, player, tournament)
+    def create_record(record, tournament)
       {
         year: tournament[:year],
         player_name: @player_name,
-        player_rank: player[:rank],
+        player_rank: tournament[:ranking],
         opponent_name: record[:opponent_name],
         opponent_rank: record[:opponent_rank],
         round: record[:round],
         score: record[:score],
         win_loss: record[:win_loss],
         tournament_name: tournament[:name],
+        tournament_category: tournament[:category],
         tournament_location: tournament[:location],
         tournament_start_date: tournament[:date][:start],
         tournament_end_date: tournament[:date][:end],
@@ -71,49 +67,5 @@ module AtpScraper
         .css("meta[property=\"pageTransitionTitle\"]")
         .attr("content").value
     end
-
-    def pickup_tournament_info(tournament_doc)
-      tournament_date = pickup_text(tournament_doc, ".tourney-dates")
-      surface = pickup_surface(tournament_doc)
-      {
-        name: pickup_text(tournament_doc, ".tourney-title"),
-        location: pickup_text(tournament_doc, ".tourney-location"),
-        date: divide_tournament_date(tournament_date),
-        year: tournament_date[0, 4],
-        caption: pickup_text(tournament_doc, ".activity-tournament-caption"),
-        surface: surface[:surface],
-        surface_inout: surface[:inout]
-      }
-    end
-
-    def pickup_player_rank(tournament_caption)
-      rank = tournament_caption.match(/ATP Ranking:(.+), Prize/)
-      rank[1].strip
-    end
-
-    # Before: String "2011.01.03 - 2011.01.08"
-    # After:  Hash { start: 2011.01.03, end: 2011.01.08 }
-    def divide_tournament_date(date)
-      date = date.split('-').map(&:strip)
-      { start: date[0], end: date[1] }
-    end
-
-    def pickup_text(doc, selector)
-      doc.css(selector).first.content.strip
-    end
-
-    def pickup_surface(tournament_doc)
-      surface = tournament_doc
-                .css(".tourney-details")[1]
-                .css(".item-details")
-                .first.content.gsub(/\t|\s/, "")
-      divide_surface(surface)
-    end
-
-    def divide_surface(surface)
-      inout = surface.match(/^(Outdoor|Indoor)/)
-      return { surface: surface, inout: nil } if inout.nil?
-      { surface: surface.gsub(/#{inout[0]}/, ''), inout: inout[0] }
-    end
   end
 end

data/lib/atp_scraper/version.rb

@@ -1,3 +1,3 @@
 module AtpScraper
-  VERSION = "0.5.1"
+  VERSION = "0.6.0"
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: atp_scraper
 version: !ruby/object:Gem::Version
-  version: 0.5.1
+  version: 0.6.0
 platform: ruby
 authors:
 - mosuke5
 autorequire: 
 bindir: exe
 cert_chain: []
-date: 2016-02-28 00:00:00.000000000 Z
+date: 2016-02-29 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: nokogiri