aranha-parsers 0.23.1 → 0.24.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 6a199311d583a58c2c57bd9440cf4427cacb84b2dea969c576f1be3e0fba1a33
-  data.tar.gz: 6f8ec7116bdeca022910267fcc27ebd64295fd70f93f99ceb0d409859c64601e
+  metadata.gz: 0e9f2435af805f1aaac850b47ade4df067af4517e9a8d44e39cc0aadc12ef8fe
+  data.tar.gz: 6e1146e598ca952c5f15f0d0c83deaec7382b6487996b489727642ec3ba5a553
 SHA512:
-  metadata.gz: 75994fd4861bf9dab93c2fa9196ebb022004c9dcdc2d88702385d62d0b55fb81b985df82b42d980decd410e5c8d0dd8664e0e3bfe9b83c27f056f560b61da247
-  data.tar.gz: 5798eebdd9d63c3cd6717bbc33a58feec336e0d033df20181db42a49fb19b70e01b7264d8cdaa303bb1c649ed950d64793b612425ada4d65f29afeb1c84b7f64
+  metadata.gz: 72d12c157d957b8238d2c9f8d39a545d06aa974e9ac49e7103b64c69648e946ecd8c439b626f1196b789a8eb30c17bc3fdf0fb46956999ddacdbbf08d37f4de0
+  data.tar.gz: ee5fe7392f18a318a64ad35bd16341118c4254d978288c64a62ddbaa85e6ff7997018311317614e5c48f1c27f159d547c33151f8ae1c5f768b3a191e343a5faa

data/lib/aranha/parsers/firefox/request_from_firefox.rb

@@ -0,0 +1,46 @@
+# frozen_string_literal: true
+
+require 'eac_ruby_utils/core_ext'
+
+module Aranha
+  module Parsers
+    module Firefox
+      class RequestFromFirefox
+        BASE_URL_SUBPATH = 'base_url'
+        BODY_SUBPATH = 'body'
+        REQUEST_SUBPATH = 'request'
+
+        class << self
+          def from_directory(path)
+            path = path.to_pathname
+            body_path = path.join(BODY_SUBPATH)
+            new(
+              path.join(BASE_URL_SUBPATH).read.strip,
+              ::Aranha::Parsers::Firefox::RequestHeaderFromFirefox
+              .from_file(path.join(REQUEST_SUBPATH)),
+              body_path.file? ? body_path.read : nil
+            )
+          end
+        end
+
+        enable_simple_cache
+        common_constructor :the_base_uri, :header, :body, default: [nil] do
+          self.the_base_uri = the_base_uri.to_uri
+        end
+
+        def to_uri_source
+          {
+            method: header.verb,
+            url: url,
+            headers: header.headers,
+            body: body
+          }
+        end
+
+        def url
+          (the_base_uri + header.uri).to_s
+        end
+      end
+    end
+  end
+end

data/lib/aranha/parsers/firefox/request_header_from_firefox.rb

@@ -0,0 +1,57 @@
+# frozen_string_literal: true
+
+module Aranha
+  module Parsers
+    module Firefox
+      class RequestHeaderFromFirefox
+        class << self
+          def from_file(path)
+            new(path.to_pathname.read)
+          end
+        end
+
+        FIRST_LINE_PARSER = /\A(\S+)\s(\S+)\s(\S+)\z/.to_parser do |m|
+          { verb: m[1], uri: m[2], version: m[3] }
+        end
+
+        HEADER_LINE_PARSER = /\A([^:]+):\s+(.+)\z/.to_parser do |m|
+          m[1..2]
+        end
+
+        enable_simple_cache
+
+        common_constructor :string
+
+        def to_h
+          %w[verb uri headers].index_with { |m| send(m) }
+        end
+
+        def headers
+          all_except_first_line.map { |line| HEADER_LINE_PARSER.parse!(line) }.to_h # rubocop:disable Style/MapToHash
+        end
+
+        def verb
+          parsed_first_line.fetch(:verb)
+        end
+
+        def uri
+          parsed_first_line.fetch(:uri)
+        end
+
+        private
+
+        def all_lines_uncached
+          string.each_line.map(&:strip)
+        end
+
+        def parsed_first_line_uncached
+          FIRST_LINE_PARSER.parse!(all_lines.first)
+        end
+
+        def all_except_first_line
+          all_lines[1..-1] # rubocop:disable Style/SlicingWithRange
+        end
+      end
+    end
+  end
+end

data/lib/aranha/parsers/firefox/uri_from_har.rb

@@ -0,0 +1,25 @@
+# frozen_string_literal: true
+
+require 'json'
+
+module Aranha
+  module Parsers
+    module Firefox
+      class UriFromHar
+        class << self
+          def from_file(path)
+            new(::JSON.parse(path.to_pathname.read))
+          end
+        end
+
+        common_constructor :data
+
+        def result
+          data.fetch('log').fetch('entries').map { |e| e.fetch('request').fetch('url') }
+        end
+
+        def request_data; end
+      end
+    end
+  end
+end

data/lib/aranha/parsers/firefox.rb

@@ -0,0 +1,11 @@
+# frozen_string_literal: true
+
+require 'eac_ruby_utils/require_sub'
+
+module Aranha
+  module Parsers
+    module Firefox
+      ::EacRubyUtils.require_sub __FILE__
+    end
+  end
+end

data/lib/aranha/parsers/source_address/base.rb

@@ -0,0 +1,55 @@
+# frozen_string_literal: true
+
+require 'eac_ruby_utils/core_ext'
+
+module Aranha
+  module Parsers
+    class SourceAddress
+      class Base
+        acts_as_abstract
+        common_constructor :source
+        compare_by :source
+
+        # @return [String]
+        def content
+          raise_abstract_method __method__
+        end
+
+        # @return [Addressable::URI]
+        def uri
+          raise_abstract_method __method__
+        end
+
+        # @return [String]
+        def url
+          uri.to_s
+        end
+
+        # @return [Hash]
+        def source_as_hash
+          source_as_hash? ? source.with_indifferent_access : raise('source is not a Hash')
+        end
+
+        # @return [Boolean]
+        def source_as_hash?
+          source.is_a?(::Hash)
+        end
+
+        # @|return [Hash]
+        def source_as_uri
+          source_as_uri? ? source.to_uri : raise('source is not a URI')
+        end
+
+        # @return [Boolean]
+        def source_as_uri?
+          source.to_uri.scheme.present?
+        end
+
+        # @return [Boolean]
+        def valid?
+          raise_abstract_method __method__
+        end
+      end
+    end
+  end
+end

data/lib/aranha/parsers/source_address/file.rb

@@ -1,6 +1,7 @@
 # frozen_string_literal: true
 
 require 'aranha/parsers/source_address/http_get'
+require 'eac_ruby_utils/core_ext'
 
 module Aranha
   module Parsers
@@ -8,23 +9,23 @@ module Aranha
       class File < ::Aranha::Parsers::SourceAddress::HttpGet
         SCHEME = 'file://'
 
-        class << self
-          def valid_source?(source)
-            source.to_s.start_with?("#{SCHEME}/", '/')
-          end
-        end
-
         def initialize(source)
           super(source.to_s.gsub(/\A#{Regexp.quote(SCHEME)}/, ''))
         end
 
-        def url
-          "#{SCHEME}#{source}"
-        end
-
         def content
           ::File.read(source)
         end
+
+        # @return [Addressable::URI]
+        def uri
+          source_as_uri? ? source_as_uri : "#{SCHEME}#{source}".to_uri
+        end
+
+        # @return [Boolean]
+        def valid?
+          source.to_s.start_with?("#{SCHEME}/", '/')
+        end
       end
     end
   end

data/lib/aranha/parsers/source_address/hash_http_base.rb

@@ -1,7 +1,7 @@
 # frozen_string_literal: true
 
 require 'aranha/parsers/source_address/fetch_content_error'
-require 'aranha/parsers/source_address/hash_http_base'
+require 'aranha/parsers/source_address/base'
 require 'eac_envs/http/error'
 require 'eac_envs/http/request'
 require 'eac_ruby_utils/core_ext'
@@ -10,16 +10,11 @@ require 'yaml'
 module Aranha
   module Parsers
     class SourceAddress
-      class HashHttpBase
+      class HashHttpBase < ::Aranha::Parsers::SourceAddress::Base
         class << self
           def http_method
             const_get 'HTTP_METHOD'
           end
-
-          def valid_source?(source)
-            source.is_a?(::Hash) &&
-              source.with_indifferent_access[:method].to_s.downcase.strip == http_method.to_s
-          end
         end
 
         DEFAULT_BODY = ''
@@ -31,11 +26,6 @@ module Aranha
 
         enable_simple_cache
 
-        common_constructor :source do
-          self.source = source.with_indifferent_access
-        end
-        compare_by :source
-
         def body
           param(:body, DEFAULT_BODY)
         end
@@ -48,12 +38,8 @@ module Aranha
           param(:headers, DEFAULT_HEADERS)
         end
 
-        def url
-          source.fetch(:url)
-        end
-
         def serialize
-          source.to_yaml
+          source_as_hash.to_yaml
         end
 
         def content
@@ -64,11 +50,22 @@ module Aranha
         end
 
         def param(key, default_value)
-          source[key] || params[key] || default_value
+          source_as_hash[key] || params[key] || default_value
         end
 
         def params
-          source[:params].if_present(DEFAULT_PARAMS)
+          source_as_hash[:params].if_present(DEFAULT_PARAMS)
+        end
+
+        # @return [Addressable::URI]
+        def uri
+          ::Addressable::URI.parse(source_as_hash.fetch(:url))
+        end
+
+        # @return [Boolean]
+        def valid?
+          source_as_hash? &&
+            source_as_hash[:method].to_s.downcase.strip == self.class.http_method.to_s
         end
 
         private

data/lib/aranha/parsers/source_address/http_get.rb

@@ -1,38 +1,26 @@
 # frozen_string_literal: true
 
 require 'addressable'
-require 'aranha/parsers/source_address/fetch_content_error'
+require 'aranha/parsers/source_address/base'
 require 'eac_envs/http/error'
 require 'eac_envs/http/request'
 
 module Aranha
   module Parsers
     class SourceAddress
-      class HttpGet
+      class HttpGet < ::Aranha::Parsers::SourceAddress::Base
         class << self
           def location_uri(source_uri, location)
             ::Addressable::URI.join(source_uri, location).to_s
           end
-
-          def valid_source?(source)
-            source.to_s =~ %r{\Ahttps?://}
-          end
         end
 
-        attr_reader :source
-
-        def initialize(source)
-          @source = source.to_s
-        end
+        common_constructor :source, super_args: -> { [source.to_s] }
 
         def ==(other)
           self.class == other.class && source == other.source
         end
 
-        def url
-          source
-        end
-
         def final_url
           content unless @final_url
           @final_url
@@ -51,6 +39,16 @@ module Aranha
         def serialize
           url
         end
+
+        # @return [Addressable::URI]
+        def uri
+          source_as_uri
+        end
+
+        # @return [Boolean]
+        def valid?
+          source.to_s =~ %r{\Ahttps?://}
+        end
       end
     end
   end

data/lib/aranha/parsers/source_address.rb

@@ -20,8 +20,10 @@ module Aranha
         def detect_sub(source)
           return source.sub if source.is_a?(self)
 
-          SUBS.each do |sub|
-            return sub.new(source) if sub.valid_source?(source)
+          SUBS.each do |sub_class|
+            sub_class.new(source).then do |sub|
+              return sub if sub.valid?
+            end
           end
           raise "No content fetcher found for source \"#{source}\""
         end

data/lib/aranha/parsers/version.rb

@@ -2,6 +2,6 @@
 
 module Aranha
   module Parsers
-    VERSION = '0.23.1'
+    VERSION = '0.24.0'
   end
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: aranha-parsers
 version: !ruby/object:Gem::Version
-  version: 0.23.1
+  version: 0.24.0
 platform: ruby
 authors:
 - Esquilo Azul Company
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2024-10-03 00:00:00.000000000 Z
+date: 2024-10-25 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: activesupport
@@ -51,6 +51,9 @@ dependencies:
     - - "~>"
       - !ruby/object:Gem::Version
         version: '0.6'
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: 0.6.1
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
@@ -58,6 +61,9 @@ dependencies:
     - - "~>"
       - !ruby/object:Gem::Version
         version: '0.6'
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: 0.6.1
 - !ruby/object:Gem::Dependency
   name: eac_ruby_utils
   requirement: !ruby/object:Gem::Requirement
@@ -128,6 +134,10 @@ extra_rdoc_files: []
 files:
 - lib/aranha/parsers.rb
 - lib/aranha/parsers/base.rb
+- lib/aranha/parsers/firefox.rb
+- lib/aranha/parsers/firefox/request_from_firefox.rb
+- lib/aranha/parsers/firefox/request_header_from_firefox.rb
+- lib/aranha/parsers/firefox/uri_from_har.rb
 - lib/aranha/parsers/html.rb
 - lib/aranha/parsers/html/base.rb
 - lib/aranha/parsers/html/item.rb
@@ -145,6 +155,7 @@ files:
 - lib/aranha/parsers/patches.rb
 - lib/aranha/parsers/patches/ofx_parser.rb
 - lib/aranha/parsers/source_address.rb
+- lib/aranha/parsers/source_address/base.rb
 - lib/aranha/parsers/source_address/fetch_content_error.rb
 - lib/aranha/parsers/source_address/file.rb
 - lib/aranha/parsers/source_address/hash_http_base.rb