barx 0.1.0

data/lib/xri_parser.rb

@@ -0,0 +1,163 @@
+module XriParser
+  class AuthorityParser
+    ## AuthorityParser.parse(xri) returns an AuthorityParser object
+    ## Pass in <false> as a second arg to parse a local (rootless) XRI
+    class <<self
+      def parse(xri, get_root=true)
+        AuthorityParser.new(xri, get_root)
+      end
+    end
+    
+    attr_reader :root, :subsegments, :authority, :local, :query, :qxri
+    attr_accessor :path
+
+    def initialize(xri, get_root=true)
+      @root = String.new
+      @subsegments = Array.new
+      xri = xri.strip rescue nil
+      xri = xri_strip_scheme(xri) rescue nil
+      @qxri = xri
+      raise 'No XRI Received' if xri.nil? or xri.empty?
+      #get root
+      if get_root
+        gcs_chars = Regexp.escape("@=!+*()")
+        gcs_connectors = Regexp.escape("@=!+*")
+        dns_rgx = Regexp.new("\\A(\\$dns\\*[^#{gcs_chars}]+)([#{gcs_connectors}].*)\\z", true)
+        ## equivalent to:  /\A(\$dns\*[^@=!\+\*\(\)]+)([@=!\+\*].*)\z/i
+        ip_rgx = Regexp.new("\\A(\\$ip\\*[^#{gcs_chars}]+)([#{gcs_connectors}].*)\\z", true)
+        ## equivalent to:  /\A(\$ip\*[^@=!\+\*\(\)]+)([@=!\+\*].*)\z/i
+        simple_roots = [?=, ?@, ?!, ?+]
+        if simple_roots.include? xri[0]
+          @root = xri[0].chr
+          xri = xri[1..-1] # xri is now the rest of xri after the root
+        elsif xri[0,2] == '$('
+          paren = 0
+          pointer = 0
+          
+          @root = '$'
+          xri = xri[1..-1]
+          xri.each_byte do |c|
+            paren += 1 if c == ?(
+            paren -= 1 if c == ?)
+            pointer += 1
+            @root << c.chr
+            break if paren.zero?
+          end
+
+          xri = xri[pointer..-1] # xri is now the rest of xri after the root
+        elsif xri =~ dns_rgx
+          @root = $1
+          xri = $2
+        elsif xri =~ ip_rgx
+          @root = $1
+          xri = $2
+        else
+          raise XRIRootMissing, 'No XRI root'
+        end
+      end
+
+      if xri.length < 1
+        return true
+      end
+      if (@root == '!') and (xri[0] != ?!)
+        raise 'XRIs in the ! registry must start with !'
+      end
+      
+      subsegment_delimiters = [?@, ?=, ?!, ?+, ?*]
+      #add implicit * after root if neccessary
+      xri.insert(0,'*') unless subsegment_delimiters.include? xri[0] 
+      open_paren = 0
+      close_paren = 0
+      segment_buf = String.new
+      
+      # get subsegments
+      xri.each_byte do |c|
+        open_paren += 1 if c == ?(
+        close_paren += 1 if c == ?)
+        # if we're inside an xref keep adding to buffer
+        if (open_paren > 0) and (open_paren != close_paren)
+          segment_buf << c.chr
+          next
+        end
+        # starting a new subsegment
+        if subsegment_delimiters.include? c
+          # flush the buffer
+          @subsegments << segment_buf unless segment_buf.empty?
+          # initialize a new buffer
+          segment_buf = ""
+          # add the delimiter to the buffer
+          segment_buf << c.chr
+        elsif (c == ?/) or (c == ??) or (c == ?#) #end of authority parsing
+          break
+        else
+          # add the character to the buffer
+          segment_buf << c.chr
+        end
+      end
+
+      @subsegments << segment_buf # flush buffer at the end
+      @local = xri[@subsegments.to_s.length..-1] # the rest of xri
+      @path = @local.index('?') ? @local[/^(.*)\?/, 1] : @local # path is everything up to a question mark
+      @query = @local.index('?') ? @local[/\?(.*)$/, 1] : '' # query is everything after a question mark
+      @authority = self.normalized
+      true
+    end
+
+    def xri_strip_scheme(xri)
+      if xri[0..5] =~ /xri:\/\//i
+        xri[6..-1]
+      elsif xri[0..3] =~ /xri:/i
+        xri[4..-1]
+      else
+        xri
+      end
+    end
+
+    def is_inumber?
+      @subsegments.each {|seg| return false if seg[0] != ?!}
+      true
+    end
+
+    def community
+      all_but_last_seg = @subsegments[0..-2].to_s
+      # strip off leading * if i-name and root is = or @
+      if @root == '@' or @root == '='
+        all_but_last_seg = all_but_last_seg[1..-1] if all_but_last_seg[0] == ?*
+      end
+      unless all_but_last_seg.nil? or all_but_last_seg.empty?
+        @root + all_but_last_seg
+      else
+        @root
+      end
+    end
+
+    def local_iname
+      unless self.is_inumber?
+        if @subsegments.last[0] == ?* # strip off leading * on local i-name
+          @subsegments.last[1..-1]
+        else
+          @subsegments.last
+        end
+      end
+    end
+    
+    def local_inumber
+      @subsegments.last[0..-1] if self.is_inumber?
+    end
+    
+    def fully_qualified
+      @root + @subsegments.to_s
+    end
+    
+    def normalized
+      if @root == '@' or @root == '='
+        @root + @subsegments.to_s[1..-1]
+      else
+        @root + @subsegments.to_s
+      end
+    end
+    
+  end
+  
+  class XRIRootMissing < RuntimeError; end
+end