yasuri 0.0.7 → 0.0.8

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 18c8b6da6ca1f9d5433adc128b83ed5d5a8e353e
-  data.tar.gz: 07ba467f8d62982e4a8e969da42b839d8ee07664
+  metadata.gz: f667fce1e95bc5a16e350a1227abdde1a6db8514
+  data.tar.gz: 68cb3c5754f9636f9ae2c677feccfd62183c0af5
 SHA512:
-  metadata.gz: 232a5893c4511b0ef80b34a95d58af4d1bb8683512a1cb41b7f9a6d19def75d3c825b12c383bf51bab09e12fe8bd54cd91b79b32640319b60d6185a46ed7f086
-  data.tar.gz: 4cb31e01b60861d13770b8b9d033cef35fe6bbdd226d11703826040840ac979fcfd98c162b18322baa8f9102fd2ed059ac0c15c277c49217feaf39173800dea6
+  metadata.gz: ddb2760495645509953e05748ac617148f1a69916f955f4781fef4565dcb030d44a204d2999e5510c3dc43af6500c654886fe8eeb5694d339c76e8eb7fa195ba
+  data.tar.gz: 2ae5d067c108a7137739c3df51001abfbc6002bda5eeeec1e3632e3e57154e8f856effd0c0017b587768feb84c76b73889582ec526e07697c95d8ba140bb8e9a

data/lib/yasuri/version.rb

@@ -1,3 +1,3 @@
 module Yasuri
-  VERSION = "0.0.7"
+  VERSION = "0.0.8"
 end

data/lib/yasuri/yasuri.rb

@@ -15,7 +15,7 @@ require_relative 'yasuri_node_generator'
 module Yasuri
 
   def self.json2tree(json_string)
-    json = JSON.parse(json_string)
+    json = JSON.parse(json_string, {symbolize_names: true})
     Yasuri.hash2node(json)
   end
 
@@ -30,26 +30,31 @@ module Yasuri
 
   private
   Text2Node = {
-    "text"   => Yasuri::TextNode,
-    "struct" => Yasuri::StructNode,
-    "links"  => Yasuri::LinksNode,
-    "pages"  => Yasuri::PaginateNode
+    text:   Yasuri::TextNode,
+    struct: Yasuri::StructNode,
+    links:  Yasuri::LinksNode,
+    pages:  Yasuri::PaginateNode
   }
   Node2Text = Text2Node.invert
 
-  ReservedKeys = %w|node name path children|
+  ReservedKeys = %i|node name path children|
   def self.hash2node(node_h)
     node, name, path, children = ReservedKeys.map do |key|
       node_h[key]
     end
     children ||= []
 
+    fail "Not found 'node' value in json" if node.nil?
+    fail "Not found 'name' value in json" if name.nil?
+    fail "Not found 'path' value in json" if path.nil?
+
     childnodes = children.map{|c| Yasuri.hash2node(c) }
     ReservedKeys.each{|key| node_h.delete(key)}
     opt = node_h
 
-    klass = Text2Node[node]
-    klass ? klass.new(path, name, childnodes, opt: opt) : nil
+    klass = Text2Node[node.to_sym]
+    fail "Undefined node type #{node}" if klass.nil?
+    klass.new(path, name, childnodes, opt)
   end
 
   def self.node2hash(node)
@@ -73,6 +78,10 @@ module Yasuri
     json
   end
 
+  def self.NodeName(name, symbolize_names:false)
+    symbolize_names ? name.to_sym : name
+  end
+
   def self.with_retry(retry_count = 5)
     begin
       return yield() if block_given?

data/lib/yasuri/yasuri_links_node.rb

@@ -6,14 +6,17 @@ require_relative 'yasuri_node'
 module Yasuri
   class LinksNode
     include Node
-    def inject(agent, page, retry_count = 5)
+    def inject(agent, page, opt = {})
+      retry_count = opt[:retry_count] || 5
+
       links = page.search(@xpath) || [] # links expected
       links.map do |link|
         link_button = Mechanize::Page::Link.new(link, agent, page)
         child_page = Yasuri.with_retry(retry_count) { link_button.click }
 
         child_results_kv = @children.map do |child_node|
-          [child_node.name, child_node.inject(agent, child_page, retry_count)]
+          child_name = Yasuri.NodeName(child_node.name, opt)
+          [child_name, child_node.inject(agent, child_page, opt)]
         end
 
         Hash[child_results_kv]

data/lib/yasuri/yasuri_node.rb

@@ -11,7 +11,7 @@ module Yasuri
       @xpath, @name, @children = xpath, name, children
     end
 
-    def inject(agent, page)
+    def inject(agent, page, opt = {})
       fail "#{Kernel.__method__} is not implemented."
     end
     def opts

data/lib/yasuri/yasuri_paginate_node.rb

@@ -7,18 +7,20 @@ module Yasuri
   class PaginateNode
     include Node
 
-    def initialize(xpath, name, children = [], limit: nil, opt: {})
+    def initialize(xpath, name, children = [], limit: nil)
       super(xpath, name, children)
-      @limit = limit || opt["limit"]
+      @limit = limit
     end
 
-    def inject(agent, page, retry_count = 5)
+    def inject(agent, page, opt = {})
+      retry_count = opt[:retry_count] || 5
 
       child_results = []
       limit = @limit.nil? ? Float::MAX : @limit
       while page
         child_results_kv = @children.map do |child_node|
-          [child_node.name, child_node.inject(agent, page, retry_count)]
+          child_name = Yasuri.NodeName(child_node.name, opt)
+          [child_name, child_node.inject(agent, page, opt)]
         end
         child_results << Hash[child_results_kv]
 

data/lib/yasuri/yasuri_struct_node.rb

@@ -6,11 +6,12 @@ require_relative 'yasuri_node'
 module Yasuri
   class StructNode
     include Node
-    def inject(agent, page, retry_count = 5)
+    def inject(agent, page, opt = {})
       sub_tags = page.search(@xpath)
       sub_tags.map do |sub_tag|
         child_results_kv = @children.map do |child_node|
-          [child_node.name, child_node.inject(agent, sub_tag, retry_count)]
+          child_name = Yasuri.NodeName(child_node.name, opt)
+          [child_name, child_node.inject(agent, sub_tag, opt)]
         end
         Hash[child_results_kv]
       end

data/lib/yasuri/yasuri_text_node.rb

@@ -6,18 +6,17 @@ require_relative 'yasuri_node'
 module Yasuri
   class TextNode
     include Node
-    def initialize(xpath, name, children = [], truncate: nil, opt: {})
-      super(xpath, name, children)
 
-      truncate_opt = opt["truncate"] #str
-      truncate_opt = Regexp.new(truncate_opt) if not truncate_opt.nil? # regexp or nil
+    def initialize(xpath, name, children = [], truncate: nil)
+      super(xpath, name, children)
 
-      @truncate = truncate || truncate_opt || nil # regexp or nil
+      truncate = Regexp.new(truncate) if not truncate.nil? # regexp or nil
 
+      @truncate = truncate
       @truncate = Regexp.new(@truncate.to_s) if not @truncate.nil?
-
     end
-    def inject(agent, page, retry_count = 5)
+
+    def inject(agent, page, opt = {})
       node = page.search(@xpath)
       text = node.text.to_s
 

data/spec/yasuri_links_node_spec.rb

@@ -90,5 +90,19 @@ describe 'Yasuri' do
       ])
       compare_generated_vs_original(generated, original, @index_page)
     end
+
+    it 'return child node as symbol' do
+      root_node = Yasuri::LinksNode.new('/html/body/a', "root", [
+        Yasuri::TextNode.new('/html/body/p', "content"),
+      ])
+
+      actual = root_node.inject(@agent, @index_page, symbolize_names: true )
+      expected = [
+        {:content => "Child 01 page."},
+        {:content => "Child 02 page."},
+        {:content => "Child 03 page."},
+      ]
+      expect(actual).to match expected
+    end
   end
 end

data/spec/yasuri_paginate_node_spec.rb

@@ -81,5 +81,19 @@ describe 'Yasuri' do
       ], limit: 2)
       compare_generated_vs_original(generated, original, @page)
     end
+
+    it "return child node as symbol" do
+      root_node = Yasuri::PaginateNode.new("/html/body/nav/span/a[@class='next']", "root", [
+        Yasuri::TextNode.new('/html/body/p', "content"),
+      ])
+      actual = root_node.inject(@agent, @page, symbolize_names:true)
+      expected = [
+        {:content => "PaginationTest01"},
+        {:content => "PaginationTest02"},
+        {:content => "PaginationTest03"},
+        {:content => "PaginationTest04"},
+      ]
+      expect(actual).to match expected
+    end
   end
 end

data/spec/yasuri_spec.rb

@@ -20,9 +20,8 @@ describe 'Yasuri' do
   # json2tree #
   #############
   describe '.json2tree' do
-    it "return empty tree" do
-      tree = Yasuri.json2tree("{}")
-      expect(tree).to be_nil
+    it "fail if empty json" do
+      expect { Yasuri.json2tree("{}") }.to raise_error
     end
 
     it "return TextNode" do
@@ -32,6 +31,7 @@ describe 'Yasuri' do
                 }|
       generated = Yasuri.json2tree(src)
       original  = Yasuri::TextNode.new('/html/body/p[1]', "content")
+
       compare_generated_vs_original(generated, original, @index_page)
     end
 

data/spec/yasuri_struct_node_spec.rb

@@ -110,5 +110,16 @@ describe 'Yasuri' do
       ])
       compare_generated_vs_original(generated, original, @page)
     end
+
+    it 'return child node as symbol' do
+      node = Yasuri::StructNode.new('/html/body/table[1]/tr', "table", [
+        Yasuri::TextNode.new('./td[1]', "title"),
+        Yasuri::TextNode.new('./td[2]', "pub_date"),
+      ])
+      expected = @table_1996.map{|h| h.map{|k,v| [k.to_sym, v] }.to_h }
+      actual = node.inject(@agent, @page, symbolize_names:true)
+      expect(actual).to match expected
+    end
+
   end
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: yasuri
 version: !ruby/object:Gem::Version
-  version: 0.0.7
+  version: 0.0.8
 platform: ruby
 authors:
 - TAC
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2015-02-26 00:00:00.000000000 Z
+date: 2015-03-01 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler