wgit 0.0.15 → 0.0.16

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: b83cdb3dd0deec7e706c77a8302eb9a5d52d9ec930fbfab43dc2790334521160
-  data.tar.gz: a1ac1ae6d151a64db95aaee386457fddfce2b76f3d0b824b360926e689ceb64a
+  metadata.gz: 269236ab66e07aaabe01d61f765689e3d997628ad76d5f61a9c477e35d67880b
+  data.tar.gz: 5fd11a994c23cd9569099109f8e2236873cf2d6267ea38bd661329620ece50b0
 SHA512:
-  metadata.gz: 7b3831ba9f3d60810507ebf49ec19e2ca533da188af84657552371823b1924c2357bdf2581a34e3526f4c160aff6dc876e2d5eba927ff793b5b4a0d6d9ff4221
-  data.tar.gz: 2e6b4fc6f36f97f6d400488f52df69323666186d03fab596a8f1895cdd76b1a9d7df573fdc1336c4df759259a1aa309f9e55e5488c2118fb9fb8c0ab307c5dc7
+  metadata.gz: 1c97aab9a225690205fcf10a99d2f632c45c08e7c3c5a543a0d374eb0595a6953baf77acd16eebf032c741d671d9e0fece030b01578af14fabc2acfa446734aa
+  data.tar.gz: eefb60a4462142fce4643dc12edac1fa11951c32f7e2d72f3295369fc8db83b3b126ea5e51410bc9ca9955cc1d7c386ac3c0aac77b1c6eaf9bc89ffc517f44ee

data/lib/wgit/crawler.rb

@@ -196,7 +196,7 @@ module Wgit
         raise 'Too many redirects' if redirect_count >= redirect_limit
         redirect_count += 1
 
-        response = Net::HTTP.get_response(url.normalise.to_uri)
+        response = Net::HTTP.get_response(url.to_uri)
         location = Wgit::Url.new(response.fetch('location', ''))
 
         if not location.empty?

data/lib/wgit/url.rb

@@ -8,7 +8,7 @@ module Wgit
   # Class modeling a web based URL.
   # Can be an internal/relative link e.g. "about.html" or a full URL
   # e.g. "http://www.google.co.uk". Is a subclass of String and uses
-  # 'addressable/uri' internally.
+  # 'uri' and 'addressable/uri' internally.
   class Url < String
     include Assertable
 
@@ -51,18 +51,27 @@ module Wgit
       super(url)
     end
 
+    # A class alias for Url.new.
+    #
+    # @param str [String] The URL string to parse.
+    # @return [Wgit::Url] The parsed Url object.
+    def self.parse(str)
+      self.new(str)
+    end
+
     # Raises an exception if url is not a valid HTTP URL.
     #
     # @param url [Wgit::Url, String] The Url to validate.
     # @raise [RuntimeError] If url is invalid.
     def self.validate(url)
-      if Wgit::Url.relative_link?(url)
+      url = Wgit::Url.new(url)
+      if url.relative_link?
         raise "Invalid url (or a relative link): #{url}"
       end
       unless url.start_with?("http://") or url.start_with?("https://")
         raise "Invalid url (missing protocol prefix): #{url}"
       end
-      if URI.regexp.match(url).nil?
+      if URI.regexp.match(url.normalise).nil?
         raise "Invalid url: #{url}"
       end
     end
@@ -96,20 +105,31 @@ module Wgit
       url
     end
 
-    # Returns if link is a relative or absolute Url.
+    # Concats the host and link Strings and returns the result.
+    #
+    # @param host [Wgit::Url, String] The Url host.
+    # @param link [Wgit::Url, String] The link to add to the host prefix.
+    # @return [Wgit::Url] host + "/" + link
+    def self.concat(host, link)
+      host = Wgit::Url.new(host).without_trailing_slash
+      link = Wgit::Url.new(link).without_leading_slash
+      separator = (link.start_with?('#') or link.start_with?('?')) ? '' : '/'
+      Wgit::Url.new(host + separator + link)
+    end
+
+    # Returns true if self is a relative Url.
+    #
     # All external links in a page are expected to have a protocol prefix e.g.
     # "http://", otherwise the link is treated as an internal link (regardless
-    # of whether it is valid or not). The only exception is if base is provided
-    # and link is a page within that site; then the link is relative.
+    # of whether it's valid or not). The only exception is if base is provided
+    # and self is a page within that site; then the link is relative.
     #
-    # @param link [Wgit::Url, String] The url to test if relative or not.
-    # @param base [String] The Url base e.g. http://www.google.co.uk.
+    # @param base [Wgit::Url, String] The Url base e.g. http://www.google.com.
     # @return [Boolean] True if relative, false if absolute.
-    # @raise [RuntimeError] If the link is invalid.
-    def self.relative_link?(link, base: nil)
-      raise "Invalid link: #{link}" if link.nil? or link.empty?
+    # @raise [RuntimeError] If self is invalid e.g. empty.
+    def is_relative?(base: nil)
+      raise "Invalid link: #{self}" if nil? or empty?
 
-      link = Wgit::Url.new(link)
       if base
         base = Wgit::Url.new(base)
         if base.to_scheme.nil?
@@ -117,35 +137,13 @@ module Wgit
         end
       end
 
-      if link.to_uri.relative?
+      if @uri.relative?
         true
       else
-        base ? link.to_host == base.to_host : false
+        base ? to_host == base.to_host : false
       end
     end
 
-    # Concats the host and link Strings and returns the result.
-    #
-    # @param host [Wgit::Url, String] The Url host.
-    # @param link [Wgit::Url, String] The link to add to the host prefix.
-    # @return [Wgit::Url] host + "/" + link
-    def self.concat(host, link)
-      host = Wgit::Url.new(host).without_trailing_slash
-      link = Wgit::Url.new(link).without_leading_slash
-      separator = (link.start_with?('#') or link.start_with?('?')) ? '' : '/'
-      Wgit::Url.new(host + separator + link)
-    end
-
-    # Returns if self is a relative or absolute Url. If base is provided and
-    # self is a page within that site then the link is relative.
-    # See Wgit.relative_link? for more information.
-    #
-    # @return [Boolean] True if relative, false if absolute.
-    # @raise [RuntimeError] If the link is invalid.
-    def relative_link?(base: nil)
-      Wgit::Url.relative_link?(self, base: base)
-    end
-
     # Determines if self is a valid Url or not.
     #
     # @return [Boolean] True if valid, otherwise false.
@@ -170,18 +168,18 @@ module Wgit
       @date_crawled = bool ? Wgit::Utils.time_stamp : nil
     end
 
-    # Normalises/encodes self and returns a new Wgit::Url.
+    # Normalises/escapes self and returns a new Wgit::Url.
     #
     # @return [Wgit::Url] An encoded version of self.
     def normalise
       Wgit::Url.new(@uri.normalize.to_s)
     end
 
-    # Returns the @uri instance var of this URL.
+    # Returns a normalised URI object for this URL.
     #
-    # @return [Addressable::URI] The URI object of self.
+    # @return [URI::HTTP, URI::HTTPS] The URI object of self.
     def to_uri
-      @uri
+      URI(normalise)
     end
 
     # Returns self.
@@ -361,9 +359,9 @@ module Wgit
     alias :fragment :to_anchor
     alias :extension :to_extension
     alias :without_fragment :without_anchor
-    alias :internal_link? :relative_link?
-    alias :is_relative? :relative_link?
-    alias :is_internal? :relative_link?
+    alias :relative_link? :is_relative?
+    alias :internal_link? :is_relative?
+    alias :is_internal? :is_relative?
     alias :crawled? :crawled
     alias :normalize :normalise
   end

data/lib/wgit/version.rb

@@ -3,5 +3,5 @@
 # @author Michael Telford
 module Wgit
   # The current gem version of Wgit.
-  VERSION = "0.0.15".freeze
+  VERSION = "0.0.16".freeze
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: wgit
 version: !ruby/object:Gem::Version
-  version: 0.0.15
+  version: 0.0.16
 platform: ruby
 authors:
 - Michael Telford