digger 0.1.3 → 0.1.7

checksums.yaml

@@ -1,7 +1,7 @@
 ---
-SHA1:
-  metadata.gz: 26deb68788670976da4b7edc2170236d35e695b5
-  data.tar.gz: 8e1c83697d912e8b401c4933c8d7e437024559c8
+SHA256:
+  metadata.gz: 2f506fd615df8b9d732d6b67bedc72644df26dc3f5725cfe5dba10c1098bae0b
+  data.tar.gz: 0e9afcb19ba0be5ce4a90787d54c3b43d58100bac3ec45bf5ef93781a60b6eb1
 SHA512:
-  metadata.gz: 47d08104b079873296e1c94cdb6ecd384be7e02e38343eba91c25dbfa27be067bd62084d7347b4c9dd0662ea17cbbacfa3196c6f2bcbf2a18e4919996b613e91
-  data.tar.gz: b29784d26b7044e6201ad9322fa932d7226ac68bffc2bcf6f802dee98f79097ed88c2e0c4dbe0bc4e99fd5e4dc4c7bf0a6562d25687b616477973f90bd91f514
+  metadata.gz: 8608a2ee8e06ddd846772d40dc3e417560229729b7b736eda8a7f50977a7d2c6fc523f86fe64480b5172c5295137eae7c85f945f7ca310502c54c8b90dd75e8d
+  data.tar.gz: 59fef1a13adc8f983c16428ee4d08d6ccdecea09b7583452b6ca07689727c3d6a386f5a5b767a20d117c8c4bc9775a2a6b32fd4caff1c1a5e85ee2af82e39d1b

data/digger.gemspec

@@ -18,8 +18,8 @@ Gem::Specification.new do |spec|
   spec.test_files    = spec.files.grep(%r{^(test|spec|features)/})
   spec.require_paths = ["lib"]
 
-  spec.add_development_dependency "bundler", "~> 1.7"
-  spec.add_development_dependency "rake", "~> 10.0"
+  spec.add_development_dependency "bundler", "~> 2.0"
+  spec.add_development_dependency "rake", ">= 12.3.3"
 
   spec.add_runtime_dependency 'nokogiri', '~> 1.6'
   spec.add_runtime_dependency 'http-cookie', '~> 1.0'

data/lib/digger/http.rb

@@ -49,7 +49,7 @@ module Digger
       url = URI(url)
       pages = []
       get(url, referer) do |response, code, location, redirect_to, response_time|
-        handle_compression response
+        handle_compression response if handle_compression?
         pages << Page.new(location, body: response.body,
                                     code: code,
                                     headers: response.to_hash,
@@ -70,6 +70,13 @@ module Digger
       [Page.new(url, error: e, referer: referer, depth: depth)]
     end
 
+    #
+    # Accept response compression, may bring encoding error if true
+    # 
+    def handle_compression?
+      @opts[:handle_compression]
+    end
+
     #
     # The maximum number of redirects to follow
     #
@@ -185,7 +192,7 @@ module Digger
       opts['User-Agent'] = user_agent if user_agent
       opts['Referer'] = referer.to_s if referer
       opts['Cookie']  = ::HTTP::Cookie.cookie_value(cookie_jar.cookies(url)) if accept_cookies?
-      opts['Accept-Encoding'] = 'gzip,deflate'
+      opts['Accept-Encoding'] = 'gzip,deflate' if handle_compression?
 
       retries = 0
       begin

data/lib/digger/index.rb

@@ -8,33 +8,34 @@ module Digger
 
     def urls
       @urls ||= begin
-        args = self.args.map{|a| (a.respond_to? :each) ? a.to_a : [a]}
-        args.shift.product(*args).map{|arg| pattern_applied_url(arg)}
+        args = self.args.map { |a| a.respond_to?(:each) ? a.to_a : [a] }
+        args.shift.product(*args).map { |arg| pattern_applied_url(arg) }
       end
     end
 
     def pattern_applied_url(arg)
-      pattern.gsub('*').each_with_index{|_, i| arg[i]}
+      pattern.gsub('*').each_with_index { |_, i| arg[i] }
     end
 
     def self.batch(entities, cocurrence = 1, &block)
-      raise NoBlockError, "No block given" unless block
+      raise NoBlockError, 'No block given' unless block
 
       if cocurrence > 1
-        results = {}
-        entities.each_slice(cocurrence) do |group|
+        results = Array.new(entities.size)
+        entities.each_slice(cocurrence).with_index do |group, idx1|
           threads = []
-          group.each do |entity|
+          group.each_with_index do |entity, idx2|
+            index = idx1 * cocurrence + idx2
             threads << Thread.new(entity) do |ent|
-              results[ent] = block.call(ent)
+              results[index] = block.call(ent)
             end
           end
-          threads.each{|thread| thread.join}
+          threads.each(&:join)
         end
-        entities.map{|ent| results[ent]}
+        results
       else
-        entities.map{|ent| block.call(ent) }
+        entities.map { |ent| block.call(ent) }
       end
     end
   end
-end
+end

data/lib/digger/model.rb

@@ -1,16 +1,19 @@
 
 module Digger
   class Model
-    @@digger_config = {'pattern'=>{}, 'index'=>{}}
+    @@digger_config = {
+      'pattern' => {},
+      'index' => {}
+    }
 
     class << self
       # patterns
       def pattern_config
-        @@digger_config['pattern'][self.name] ||= {}
+        @@digger_config['pattern'][name] ||= {}
       end
 
       Pattern::TYPES.each do |method|
-        define_method method, ->(pairs, &block){
+        define_method method, -> (pairs, &block) {
           pairs.each_pair do |key, value|
             pattern_config[key] = Pattern.new(type: method, value: value, block: block)
           end
@@ -18,21 +21,22 @@ module Digger
       end
 
       def validate_presence(*keys)
-        keys_all = pattern_config.keys
-        raise "Pattern keys #{(keys - keys_all).join(', ')} should be present" unless keys.all?{|k| keys_all.include?(k) }
+        is_all = pattern_config.keys.all? { |k| keys.include?(k) }
+        raise "Pattern keys #{(keys - keys_all).join(', ')} should be present" unless is_all
       end
 
       def validate_includeness(*keys)
-        raise "Pattern keys #{(pattern_config.keys - keys).join(', ')} should not be included" unless pattern_config.keys.all?{|k| keys.include?(k)}
+        is_all = pattern_config.keys.all? { |k| keys.include?(k) }
+        raise "Pattern keys #{(pattern_config.keys - keys).join(', ')} should not be included" if is_all
       end
 
       # index page
       def index_config
-        @@digger_config['index'][self.name]
+        @@digger_config['index'][name]
       end
 
       def index_page(pattern, *args)
-         @@digger_config['index'][self.name] = Index.new(pattern, args)
+        @@digger_config['index'][name] = Index.new(pattern, args)
       end
 
       def index_page?
@@ -55,13 +59,15 @@ module Digger
     end
 
     def dig_urls(urls, cocurrence = 1, opts = {})
-      Index.batch(urls, cocurrence){|url| dig_url(url, opts) }
+      Index.batch(urls, cocurrence) { |url| dig_url(url, opts) }
     end
 
     def dig(cocurrence = 1)
       if self.class.index_page?
-        self.class.index_config.process(cocurrence){|url| dig_url(url) }
+        self.class.index_config.process(cocurrence) do |url|
+          dig_url(url)
+        end
       end
     end
   end
-end
+end

data/lib/digger/page.rb

@@ -3,6 +3,7 @@ require 'json'
 require 'ostruct'
 require 'set'
 require 'kconv'
+require 'uri'
 
 # https://github.com/taganaka/polipus/blob/master/lib/polipus/page.rb
 module Digger
@@ -27,16 +28,12 @@ module Digger
     # OpenStruct it holds users defined data
     attr_accessor :user_data
 
-    attr_accessor :aliases
-
-    attr_accessor :domain_aliases
+    attr_accessor :aliases, :domain_aliases, :fetched_at
 
     # Whether the current page should be stored
     # Default: true
     attr_accessor :storable
 
-    attr_accessor :fetched_at
-
     #
     # Create a new page
     #
@@ -60,7 +57,7 @@ module Digger
     end
 
     def title
-      doc.title if doc
+      doc&.title
     end
 
     #
@@ -74,6 +71,7 @@ module Digger
         doc.search('//a[@href]').each do |a|
           u = a['href']
           next if u.nil? || u.empty?
+
           abs = to_absolute(u) rescue next
           @links << abs if abs && in_domain?(abs)
         end
@@ -95,7 +93,13 @@ module Digger
       end
     end
 
+    def json
+      @json ||= JSON.parse body
+    end
 
+    def jsonp
+      @jsonp ||= JSON.parse body.match(/^[^(]+?\((.+)\)[^)]*$/)[1]
+    end
 
     #
     # Discard links, a next call of page.links will return an empty array
@@ -156,7 +160,7 @@ module Digger
     # returns +false+ otherwise.
     #
     def not_found?
-      404 == @code
+      @code == 404
     end
 
     #
@@ -170,6 +174,7 @@ module Digger
               end unless @base
 
       return nil if @base && @base.to_s.empty?
+
       @base
     end
 
@@ -180,16 +185,7 @@ module Digger
     def to_absolute(link)
       return nil if link.nil?
 
-      # link = link.to_s.encode('utf-8', 'binary', :invalid => :replace, :undef => :replace, :replace => '')
-
-      # remove anchor
-      link =
-        begin
-          URI.encode(URI.decode(link.to_s.gsub(/#[a-zA-Z0-9_-]*$/, '')))
-        rescue URI::Error
-          return nil
-        end
-
+      link = link.to_s.encode('utf-8', 'binary', invalid: :replace, undef: :replace, replace: '').gsub(/#[\w]*$/, '')
       relative = begin
                    URI(link)
                  rescue URI::Error
@@ -247,6 +243,7 @@ module Digger
 
     def expired?(ttl)
       return false if fetched_at.nil?
+
       (Time.now.to_i - ttl) > fetched_at
     end
 

data/lib/digger/pattern.rb

@@ -1,91 +1,113 @@
 require 'nokogiri'
 
 module Digger
+  # Extractor patterns definition
   class Pattern
     attr_accessor :type, :value, :block
 
     def initialize(hash = {})
-      hash.each_pair{|key, value| send("#{key}=", value) if %w{type value block}.include?(key.to_s)}
+      hash.each_pair { |key, value| send("#{key}=", value) if %w[type value block].include?(key.to_s)}
     end
 
-    def safe_block
-      block && begin
-        if block.respond_to?(:call)
-          block
-        elsif block.strip == '' # 
-          nil
-        else
-          proc{ $SAFE = 2; eval block }.call
-        end
-      rescue StandardError
-        nil
+    def safe_block(&default_block)
+      if block.nil? || (block.is_a?(String) && block.strip.empty?)
+        default_block
+      elsif block.respond_to?(:call)
+        block
+      else
+        proc {
+          $SAFE = 2
+          eval block
+        }.call
       end
     end
 
     def self.wrap(hash)
-      Hash[hash.map{|key, value| [key, value.is_a?(Pattern) ? value : Pattern.new(value)]}]
+      hash.transform_values { |value| value.is_a?(Pattern) ? value : Pattern.new(value) }
     end
 
     MATCH_MAX = 3
-    
-    TYPES = 0.upto(MATCH_MAX).map{|i| "match_#{i}"} + %w{match_many css_one css_many}
 
-    def regexp?
-      TYPES.index(type) <= MATCH_MAX + 1 # match_many in addition
+    TYPES_REGEXP = 0.upto(MATCH_MAX).map { |i| "match_#{i}" } + %w[match_many]
+    TYPES_CSS = %w[css_one css_many].freeze
+    TYPES_JSON = %w[json jsonp].freeze
+
+    TYPES = TYPES_REGEXP + TYPES_CSS + TYPES_JSON
+
+    def match_page(page)
+      return unless page.success?
+      if TYPES_REGEXP.include?(type) # regular expression
+        regexp_match(page.body)
+      elsif TYPES_CSS.include?(type) # css expression
+        css_match(page.doc)
+      elsif TYPES_JSON.include?(type)
+        json_match(page)
+      end
     end
 
-    def match_page(page, &callback)
-      blk = callback || safe_block
-      if regexp? # regular expression
-        index = TYPES.index(type)
-        blk ||= ->(text){text.strip}
-        # content is String
-        if type == 'match_many'
-          match = page.body.gsub(value).to_a
-        else
-          matches = page.body.match(value)
-          match = matches.nil? ? nil : matches[index]
-        end
-      else # css expression
-        blk ||= ->(node){node.content.strip}
-        # content is Nokogiri::HTML::Document
-        if type == 'css_one'
-          match = page.doc.css(value).first
-        elsif type == 'css_many' # css_many
-          match = page.doc.css(value)
-        end
+    def json_match(page)
+      block = safe_block { |j| j }
+      json = page.send(type)
+      keys = json_index_keys(value)
+      match = json_fetch(json, keys)
+      block.call(match)
+    end
+
+    def css_match(doc)
+      block = safe_block { |node| node.content.strip }
+      # content is Nokogiri::HTML::Document
+      contents = doc.css(value)
+      if type == 'css_many'
+        contents.map { |node| block.call(node) }.uniq
+      else
+        block.call(contents.first)
+      end
+    end
+
+    def regexp_match(body)
+      block = safe_block(&:strip)
+      # content is String
+      if type == 'match_many'
+        body.gsub(value).to_a.map { |node| block.call(node) }.uniq
+      else
+        index = TYPES_REGEXP.index(type)
+        matches = body.match(value)
+        block.call(matches[index]) unless matches.nil?
       end
-      if match.nil?
-        nil
-      elsif %w{css_many match_many}.include? type
-        match.map{|node| blk.call(node) }.uniq
+    end
+
+    def json_fetch(json, keys)
+      if keys.empty?
+        json
       else
-        blk.call(match)
+        pt = keys.shift
+        json_fetch(json[pt[:index] || pt[:key]], keys)
       end
-    rescue
-      nil
     end
 
+    def json_index_keys(keys)
+      keys.to_s.match(/^\$\S*$/)[0].scan(/(\.(\w+)|\[(\d+)\])/).map do |p|
+        p[1].nil? ? { index: p[2].to_i } : { key: p[1] }
+      end
+    end
+
+    private :json_index_keys, :json_fetch
+
+    # Nokogiri node methods
     class Nokogiri::XML::Node
-      %w{one many}.each do |name|
-        define_method "inner_#{name}" do |css, &block| 
-          callback = ->(node) do
-            if node
-              (block || ->(n){n.text.strip}).call(node)
-            else
-              nil
-            end
-          end
+      %w[one many].each do |name|
+        define_method "inner_#{name}" do |css, &block|
+          callback = ->(node) { (block || ->(n) { n.text.strip }).call(node) if node }
           if name == 'one' # inner_one
             callback.call(self.css(css).first)
           else # inner_many
-            self.css(css).map{|node| callback.call(node)}
+            self.css(css).map { |node| callback.call(node) }
           end
         end
       end
       def source
         to_xml
       end
-    end # nokogiri
+    end
   end
-end
+end

data/lib/digger/version.rb

@@ -1,3 +1,3 @@
 module Digger
-  VERSION = "0.1.3"
+  VERSION = '0.1.7'.freeze
 end

data/spec/digger_spec.rb

@@ -1,15 +1,14 @@
 require 'digger'
 
 http = Digger::HTTP.new
-page = http.fetch_page('http://nan.so/')
+page = http.fetch_page('http://www.baidu.com/')
 
-pattern = Digger::Pattern.new({type: 'css_many', value: '.sites>a>span' })
+pattern = Digger::Pattern.new({ type: 'css_many', value: '#s-top-left>a' })
 
 class Item < Digger::Model
-  css_many sites: '.sites>a>span'
-  css_one logo: '.logo'
+  css_many sites: '#s-top-left>a'
   validate_presence :sites
-  validate_includeness :sites, :logo
+  validate_includeness :sites
 end
 
 describe Digger do
@@ -19,12 +18,12 @@ describe Digger do
 
   it "pattern should match content" do
     sites = pattern.match_page(page)
-    expect(sites.include?('百度网盘')).to eq(true)
+    expect(sites.include?('新闻')).to eq(true)
   end
 
   it "model should dig content" do
     item = Item.new.match_page(page)
-    expect(item[:sites].include?('读远')).to be(true)
+    expect(item[:sites].include?('新闻')).to be(true)
   end
 
   it "validation support" do

data/spec/index_spec.rb

@@ -0,0 +1,12 @@
+require 'digger'
+
+describe Digger::Index do
+  it 'batch digger' do
+    list = [1, 2, 3, 4, 5, 6, 7, 8]
+    pt = Digger::Index.batch(list, 3) do |num|
+      sleep(rand(1..3))
+      "##{num}"
+    end
+    expect(pt.join).to eq(list.map { |num| "##{num}" }.join)
+  end
+end

data/spec/page_spec.rb

@@ -0,0 +1,27 @@
+require 'digger'
+require 'json'
+require 'uri'
+
+describe Digger::Page do
+  it 'page json' do
+    json_str = '{"a":1,"b":[1,2,3]}'
+    j1 = Digger::Page.new('', body: json_str)
+    j2 = Digger::Page.new('', body: "hello(#{json_str});")
+    expect(j1.json['a']).to eq(1)
+    expect(j2.jsonp['a']).to eq(1)
+    expect(j1.json['b'][0]).to eq(1)
+    expect(j2.jsonp['b'][1]).to eq(2)
+  end
+
+  it 'fetch baidu' do
+    http = Digger::HTTP.new
+    page = http.fetch_page('http://www.baidu.com/')
+    expect(page.code).to eq(200)
+  end
+
+  it 'page uri' do
+    link ='https://www.baidu.com/s?wd=%E5%93%88%E5%93%88#hello'
+    link = link.to_s.encode('utf-8', 'binary', invalid: :replace, undef: :replace, replace: '').gsub(/#[\w]*$/, '')
+    p link
+  end
+end

data/spec/pattern_spec.rb

@@ -0,0 +1,15 @@
+require 'digger'
+require 'json'
+
+describe Digger::Pattern do
+  it 'json fetch' do
+    json = JSON.parse('[{"a":1,"b":[1,2,3]}]')
+    pt = Digger::Pattern.new
+    expect(pt.json_fetch(json, '$[0]')['a']).to eq(1)
+    expect(pt.json_fetch(json, '$[0].a')).to eq(1)
+    expect(pt.json_fetch(json, '$[0].b').length).to eq(3)
+    expect(pt.json_fetch(json, '$[0].b[2]')).to eq(3)
+  end
+
+
+end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: digger
 version: !ruby/object:Gem::Version
-  version: 0.1.3
+  version: 0.1.7
 platform: ruby
 authors:
 - binz
-autorequire: 
+autorequire:
 bindir: bin
 cert_chain: []
-date: 2015-05-22 00:00:00.000000000 Z
+date: 2021-12-26 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler
@@ -16,28 +16,28 @@ dependencies:
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: '1.7'
+        version: '2.0'
   type: :development
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: '1.7'
+        version: '2.0'
 - !ruby/object:Gem::Dependency
   name: rake
   requirement: !ruby/object:Gem::Requirement
     requirements:
-    - - "~>"
+    - - ">="
       - !ruby/object:Gem::Version
-        version: '10.0'
+        version: 12.3.3
   type: :development
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
-    - - "~>"
+    - - ">="
       - !ruby/object:Gem::Version
-        version: '10.0'
+        version: 12.3.3
 - !ruby/object:Gem::Dependency
   name: nokogiri
   requirement: !ruby/object:Gem::Requirement
@@ -87,12 +87,15 @@ files:
 - lib/digger/pattern.rb
 - lib/digger/version.rb
 - spec/digger_spec.rb
+- spec/index_spec.rb
+- spec/page_spec.rb
+- spec/pattern_spec.rb
 - spec/validate_spec.rb
 homepage: ''
 licenses:
 - MIT
 metadata: {}
-post_install_message: 
+post_install_message:
 rdoc_options: []
 require_paths:
 - lib
@@ -107,11 +110,13 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubyforge_project: 
-rubygems_version: 2.2.2
-signing_key: 
+rubygems_version: 3.2.32
+signing_key:
 specification_version: 4
 summary: Dig need stractual infomation from web page.
 test_files:
 - spec/digger_spec.rb
+- spec/index_spec.rb
+- spec/page_spec.rb
+- spec/pattern_spec.rb
 - spec/validate_spec.rb