shopee 0.2.1 → 1.0.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 9f4258fe2aa8adee98fa96ee6fe07c256931bf4d
-  data.tar.gz: a626806c09a7df230329b83a570e9198e0548017
+  metadata.gz: 755028be4d2f5bdc40ddbcd3d607bb4bae24e8ad
+  data.tar.gz: a5428a405378f2370d7756a345ae3a37ab0ac34d
 SHA512:
-  metadata.gz: 628d17e1d2dfacd21e126f10c846d958dc0a08528664d1c8e4e0c2f9c3525c28ee0f412243ad3c174ee2ace8e3f6717d8b3442d4b36dc17fa80614a2588c9467
-  data.tar.gz: 68ef920116e9f218faca5796b5bcf1fd94829885386c2f300887926055ff18d76b7111cbf4432ef25f85061f2ae1274dac842734856b7cb6117668d4ec7f9dcb
+  metadata.gz: c12cb36534a9357af5aacc0a631ebac53137b02f3e9f41af9866901a4e88243ef1741db50d7184bc011233ec2394034b91f676579dcfb8b78a3efac6f4dfe60a
+  data.tar.gz: 0035557bdd8d291913fe8b60fa2298b3966baafca47d79f6762ab6c5011ca54dffd8f23f3c57969d3cdce481fa62357933ad2d713a3b00051f61ceab5a68c022

data/README.md

@@ -21,10 +21,10 @@ $ gem install shopee
 $ shopee allcate
 
 ## 2. List goods by a category
-$ shopee list [category_name]
+$ shopee list [category_name] [page_number]
 
 ## 3. Find top number goods in a category by the keyword
-$ shopee search [category_name] [keyword] [num]
+$ shopee search [category_name] [keyword] [number_of_product]
 
 ```
 
@@ -191,7 +191,8 @@ GPS
 require 'shopee'
 ##Example ：category is '電腦資訊'
 category_name = '電腦資訊'
-shopeecate = ShopeeScrape::ShopeeListGoodsByCate.new(category_name)
+page = 1
+shopeecate = ShopeeScrape::ShopeeListGoodsByCate.new(category_name, page)
 puts shopeecate.goods
 ```
 
@@ -235,9 +236,8 @@ puts shopeecate.goods
 ```ruby
 require 'shopee'
 ##Example ：category is '電腦資訊'; keyword is 'samsung'
-shopeecate = ShopeeScrape::ShopeeListGoodsByCate.new(category)
-goods = shopeecate.goods
-puts shopeecate.search_keyword(goods, keyword, num)
+shopeecate = ShopeeScrape::ShopeeListGoodsByCate.new()
+puts shopeecate.search_keyword('電腦資訊', 'Asus', '10')
 ```
 - Expected output
 ```

data/bin/shopee

@@ -12,11 +12,11 @@ Commander.configure do
   program :description, 'Scrape informations from Mobile01'
 
     command :list do |c|
-      c.syntax = 'shopee list [args]'
+      c.syntax = 'shopee list [category] [page]'
       c.description = 'List goods by the category'
 
       c.action do |args, options|
-          shopeecate = ShopeeScrape::ShopeeListGoodsByCate.new(args[0])
+          shopeecate = ShopeeScrape::ShopeeListGoodsByCate.new(args[0], args[1])
           puts shopeecate.goods
       end
     end
@@ -32,13 +32,12 @@ Commander.configure do
     end
 
     command :search do |c|
-      c.syntax = 'shopee search [args]'
+      c.syntax = 'shopee search [category] [keyword] [number of products]'
       c.description = 'search similar good'
 
       c.action do |args, options|
-        shopeecate = ShopeeScrape::ShopeeListGoodsByCate.new(args[0])
-        goods = shopeecate.goods
-        puts shopeecate.search_keyword(goods, args[1], args[2])
+        shopeecate = ShopeeScrape::ShopeeListGoodsByCate.new()
+        puts shopeecate.search_keyword(args[0], args[1], args[2])
       end
     end
 

data/lib/shopee.rb

@@ -4,6 +4,10 @@ require 'json'
 require 'fuzzystringmatch'
 
 module ShopeeScrape
+
+  # =============================
+  # Class For List products by category and search products with keyword
+  # =============================
   class ShopeeListGoodsByCate
 
     require_relative './data/mobile_category'
@@ -13,51 +17,40 @@ module ShopeeScrape
     GOOD_NUM = "//div[contains(@class, 'num')]"
     GOOD_UPTIME = "//div[contains(@class, 'updated')]"
 
-    def initialize(category)
-      parse_html(category)
+    def initialize(category=nil, page=nil)
+      puts category, page
+      if !category.nil? && !page.nil?
+        get_page_html(category, page)
+      end
     end
 
     def goods
       @goods ||= extract_goods
     end
 
-    def search_keyword(good, keyword, list_num)
-      @similar ||= find_similiar_goods(goods, keyword, list_num)
+    def search_keyword(category, keyword, list_num)
+      @similar ||= find_similiar_goods(category, keyword, list_num)
     end
 
     private
 
-    def parse_html(id)
+    def get_page_html(cate, page)
       @document = []
-      url = ALL_LINK[id]
-      @document << Oga.parse_html(open(url))
-
-      page_num = 2
-      if CATEGORY_LIST.include?(id) == true
-        page_num = 5
-      end
+      url = ALL_LINK[cate]
 
-      check = 1
-      i = 1
-
-      while check == 1 && i < page_num
-        i += 1
-        begin
-          url_t = url +'&p='+i.to_s
-          @document << Oga.parse_html(open(url_t))
-        rescue
-          check = 0
+      if CATEGORY_LIST.include?(cate)
+        if page != 1
+          @document << Oga.parse_html(open(url << '&p=' << page.to_s))
+        else
+          @document << Oga.parse_html(open(url))
         end
       end
-
     end
 
     def extract_goods
-
       results = []
 
       @document.each do |doc|
-
         name = []
         price = []
         num = []
@@ -83,7 +76,6 @@ module ShopeeScrape
 
         doc.xpath(GOOD_INFO).map do |good|
           link << good.css('a').attribute('href')[0].to_s
-          # puts good.css('img').attribute('src')[0].to_s
           pic << good.css('img').attribute('src')[0].to_s
         end
 
@@ -93,6 +85,7 @@ module ShopeeScrape
         if number > 32
           number = 32
         end
+
         for i in 2..number-1
           element = {}
           element['name'] = name[i]
@@ -105,17 +98,32 @@ module ShopeeScrape
         end
 
       end
+
       puts results.length
       results
     end
 
-    def find_similiar_goods(goods, keyword, list_num)
+    def find_similiar_goods(category, keyword, list_num)
+      goods = []
+      goodsSet = []
+
+      for i in 1..5
+        shopeecate = ShopeeScrape::ShopeeListGoodsByCate.new(category, i)
+        goodsSet << shopeecate.goods
+      end
+
+      goodsSet.each do |good|
+        good.each do |g|
+          goods << g
+        end
+      end
+
       jarow = FuzzyStringMatch::JaroWinkler.create( :native )
       rank = {}
+
       goods.each do |good|
-        good_name = good['name']
-        value = jarow.getDistance(good_name ,keyword)
-        rank[good_name] = value
+        value = jarow.getDistance(good['name'] ,keyword)
+        rank[good['name']] = value
       end
 
       rank_after_sort = Hash[rank.sort_by{|k, v| v}.reverse]
@@ -136,8 +144,6 @@ module ShopeeScrape
     end
   end
 
-
-
   # =============================
   # List all category of mobile01
   # =============================

data/lib/shopee/version.rb

@@ -1,4 +1,4 @@
 module ShopeeScrape
-  VERSION = '0.2.1'
+  VERSION = '1.0.0'
   DATE = '2016-01-09'
 end

data/spec/shopee_spec.rb

@@ -157,7 +157,7 @@ end
 
 VCR.use_cassette('shopee') do
   # //It should be placed outside.
-  shopeecate = ShopeeScrape::ShopeeListGoodsByCate.new('電腦資訊')
+  shopeecate = ShopeeScrape::ShopeeListGoodsByCate.new('電腦資訊', 1)
   describe 'Shopee testament using vcr' do
      it 'should return an array of string and name of categories' do
        good_list = shopeecate.goods

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: shopee
 version: !ruby/object:Gem::Version
-  version: 0.2.1
+  version: 1.0.0
 platform: ruby
 authors:
 - Sheng Jung Wu
@@ -125,7 +125,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
       version: '0'
 requirements: []
 rubyforge_project: 
-rubygems_version: 2.4.6
+rubygems_version: 2.5.1
 signing_key: 
 specification_version: 4
 summary: Scrape categories of mobile01!