darcs support for ohloh

.gitignore

@@ -2,3 +2,4 @@
 *.pyc
 pkg/
 *.cache
+TAGS

lib/scm.rb

@@ -19,6 +19,7 @@ module Scm
 require 'lib/scm/adapters/hglib_adapter'
 require 'lib/scm/adapters/bzr_adapter'
 require 'lib/scm/adapters/bzrlib_adapter'
+require 'lib/scm/adapters/darcs_adapter'
 require 'lib/scm/adapters/factory'
 
 require 'lib/scm/parsers/parser'
@@ -32,6 +33,7 @@ module Scm
 require 'lib/scm/parsers/hg_styled_parser'
 require 'lib/scm/parsers/bzr_xml_parser'
 require 'lib/scm/parsers/bzr_parser'
+require 'lib/scm/parsers/darcs_parser'
 
 require 'lib/scm/parsers/array_writer'
 require 'lib/scm/parsers/xml_writer'

lib/scm/adapters/darcs/cat_file.rb

@@ -0,0 +1,28 @@
+module Scm::Adapters
+	class DarcsAdapter < AbstractAdapter
+		def cat_file(commit, diff)
+			cat(commit.token, diff.path)
+		end
+
+		def cat_file_parent(commit, diff)
+			p = parent_tokens(commit)
+			cat(p.first, diff.path) if p.first
+		end
+
+		def cat(revision, path)
+			out, err = run_with_err("cd '#{url}' && darcs show contents -p '#{revision}' #{escape(path)}")
+			# show contents gives no error for non-existent paths
+			#return nil if err =~ /No such file in rev/
+			raise RuntimeError.new(err) unless err.to_s == ''
+			return nil if out == ''
+			out
+		end
+
+		# Escape bash-significant characters in the filename
+		# Example:
+		#     "Foo Bar & Baz" => "Foo\ Bar\ \&\ Baz"
+		def escape(path)
+			path.gsub(/[ '"&()<>|]/) { |c| '\\' + c }
+		end
+	end
+end

lib/scm/adapters/darcs/commits.rb

@@ -0,0 +1,87 @@
+module Scm::Adapters
+	class DarcsAdapter < AbstractAdapter
+
+		# Return the number of commits in the repository following +since+.
+		def commit_count(since=nil)
+			commit_tokens(since).size
+		end
+
+		# Return the list of commit tokens following +since+.
+		def commit_tokens(since=nil, up_to=nil)
+      from = since ? " --from-patch #{since}" : ""
+      to   = up_to ? " --to-patch #{up_to}" : ""
+			tokens = string_to_patch_names(run("cd '#{self.url}' && darcs changes#{from}#{to}")).reverse
+
+			# Darcs returns everything after *and including* since.
+			# We want to exclude it.
+			if tokens.any? && tokens.first == since
+				tokens[1..-1]
+			else
+				tokens
+			end
+		end
+
+		# Returns a list of shallow commits (i.e., the diffs are not populated).
+		# Not including the diffs is meant to be a memory savings when we encounter massive repositories.
+		# If you need all commits including diffs, you should use the each_commit() iterator, which only holds one commit
+		# in memory at a time.
+		def commits(since=nil)
+			from = since ? " --from-patch #{since}" : ""
+			log = run("cd '#{self.url}' && darcs changes#{from} --reverse")
+			a = Scm::Parsers::DarcsParser.parse(log)
+			if a.any? && a.first.token == since
+				a[1..-1]
+			else
+				a
+			end
+		end
+
+		# Returns a single commit, including its diffs
+		def verbose_commit(token)
+			log = run("cd '#{self.url}' && darcs changes -v -p '#{token}'")
+			Scm::Parsers::DarcsParser.parse(log).first
+		end
+
+		# Yields each commit after +since+, including its diffs.
+		# The log is stored in a temporary file.
+		# This is designed to prevent excessive RAM usage when we encounter a massive repository.
+		# Only a single commit is ever held in memory at once.
+		def each_commit(since=nil)
+			open_log_file(since) do |io|
+				Scm::Parsers::DarcsParser.parse(io) do |commit|
+					yield commit if block_given? && commit.token != since
+				end
+			end
+		end
+
+		# Not used by Ohloh proper, but handy for debugging and testing
+		def log(since=nil)
+      from = since ? " --from-patch #{since}" : ""
+			run "cd '#{url}' && darcs changes -s#{from}"
+		end
+
+		# Returns a file handle to the log.
+		# In our standard, the log should include everything AFTER +since+. However, darcs doesn't work that way;
+		# it returns everything after and INCLUDING +since+. Therefore, consumers of this file should check for
+		# and reject the duplicate commit.
+		def open_log_file(since=nil)
+			begin
+				if since == head_token # There are no new commits
+					# As a time optimization, just create an empty file rather than fetch a log we know will be empty.
+					File.open(log_filename, 'w') { }
+				else
+          from = since ? " --from-patch #{since}" : ""
+					run "cd '#{url}' && darcs changes --reverse -v#{from} > #{log_filename}"
+				end
+				File.open(log_filename, 'r') { |io| yield io }
+			ensure
+				File.delete(log_filename) if FileTest.exist?(log_filename)
+			end
+		end
+
+		def log_filename
+		  File.join('/tmp', (self.url).gsub(/\W/,'') + '.log')
+		end
+
+	end
+end

lib/scm/adapters/darcs/head.rb

@@ -0,0 +1,23 @@
+module Scm::Adapters
+	class DarcsAdapter < AbstractAdapter
+		def head_token
+		  string_to_patch_names(run("cd '#{url}' && darcs changes --last 1"))[0]
+		end
+
+		def head
+			verbose_commit(head_token)
+		end
+
+		def parent_tokens(commit)
+		  string_to_patch_names(run("cd '#{url}' && darcs changes --to-patch #{commit.token}"))[1..-1]
+		end
+
+		def parents(commit)
+			parent_tokens(commit).map {|token| verbose_commit(token)}
+		end
+
+		def string_to_patch_names(s)
+		  s.split(/\n/).select {|s| s =~ /^  \* /}.map {|s| s.sub(/^  \* /,'')}
+		end
+	end
+end

lib/scm/adapters/darcs/misc.rb

@@ -0,0 +1,21 @@
+module Scm::Adapters
+	class DarcsAdapter < AbstractAdapter
+		def exist?
+			begin
+				!!(head_token)
+			rescue
+				logger.debug { $! }
+				false
+			end
+		end
+
+		def ls_tree(token)
+			run("cd '#{path}' && darcs show files -p '#{token}'").split("\n")
+		end
+
+		def export(dest_dir, token=nil)
+			p = token ? " -p '#{token}'" : ""
+			run("cd '#{path}' && darcs dist#{p} && mv darcs.tar.gz '#{dest_dir}'")
+		end
+	end
+end

lib/scm/adapters/darcs/patch.rb

@@ -0,0 +1,7 @@
+module Scm::Adapters
+  class DarcsAdapter < AbstractAdapter
+    def patch_for_commit(commit)
+      run("cd '#{url}' && darcs changes -p'#{commit.token}' -v")
+    end
+  end
+end

lib/scm/adapters/darcs/pull.rb

@@ -0,0 +1,23 @@
+module Scm::Adapters
+	class DarcsAdapter < AbstractAdapter
+
+		def pull(from, &block)
+			raise ArgumentError.new("Cannot pull from #{from.inspect}") unless from.is_a?(DarcsAdapter)
+			logger.info { "Pulling #{from.url}" }
+
+			yield(0,1) if block_given? # Progress bar callback
+
+			unless self.exist?
+				run "mkdir -p '#{self.url}'"
+				run "rm -rf '#{self.url}'"
+				run "darcs get '#{from.url}' '#{self.url}'"
+			else
+				# might also need to unpull for an exact copy
+				run "cd '#{self.url}' && darcs revert --all && darcs pull --dont-allow-conflicts -a '#{from.url}'"
+			end
+
+			yield(1,1) if block_given? # Progress bar callback
+		end
+
+	end
+end

lib/scm/adapters/darcs/push.rb

@@ -0,0 +1,50 @@
+module Scm::Adapters
+	class DarcsAdapter < AbstractAdapter
+
+		def push(to, &block)
+			raise ArgumentError.new("Cannot push to #{to.inspect}") unless to.is_a?(DarcsAdapter)
+			logger.info { "Pushing to #{to.url}" }
+
+			yield(0,1) if block_given? # Progress bar callback
+
+			unless to.exist?
+				if to.local?
+					# Create a new repo on the same local machine. Just use existing pull code in reverse.
+					to.pull(self)
+				else
+					run "cd '#{self.url}' && darcs put #{to.hostname}:#{to.path}"
+				end
+			else
+				run "cd '#{self.url}' && darcs push -a '#{to.url}'"
+			end
+
+			yield(1,1) if block_given? # Progress bar callback
+		end
+
+		def local?
+			return true if hostname == Socket.gethostname
+			return true if url =~ /^file:\/\//
+			return true if url !~ /:/
+			false
+		end
+
+		def hostname
+			$1 if url =~ /^ssh:\/\/([^\/]+)/
+		end
+
+		def path
+			case url
+			when /^file:\/\/(.+)$/
+				$1
+			when /^ssh:\/\/[^\/]+(\/.+)$/
+				$1
+			when /^[^:]*$/
+				url
+			end
+		end
+
+		def darcs_path
+			path && File.join(path, '.darcs')
+		end
+	end
+end

lib/scm/adapters/darcs/validation.rb

@@ -0,0 +1,26 @@
+module Scm::Adapters
+	class DarcsAdapter < AbstractAdapter
+		def self.url_regex
+			/^((http|https|ssh|file):\/\/((\w+@)?[A-Za-z0-9_\-\.]+(:\d+)?\/)?)?[A-Za-z0-9_\-\.\/\~\+]*$/
+		end
+
+		def self.public_url_regex
+			/^(http|https):\/\/(\w+@)?[A-Za-z0-9_\-\.]+(:\d+)?\/[A-Za-z0-9_\-\.\/\~\+]*$/
+		end
+
+		def validate_server_connection
+			return unless valid?
+			@errors << [:failed, "The server did not respond to the 'darcs id' command. Is the URL correct?"] unless self.exist?
+		end
+
+		def guess_forge
+			u = @url =~ /:\/\/(.*\.?darcs\.)?([^\/^:]+)(:\d+)?\// ? $2 : nil
+			case u
+			when /(sourceforge\.net$)/
+				$1
+			else
+				u
+			end
+		end
+	end
+end

lib/scm/adapters/darcs_adapter.rb

@@ -0,0 +1,16 @@
+module Scm::Adapters
+	class DarcsAdapter < AbstractAdapter
+		def english_name
+			"Darcs"
+		end
+	end
+end
+
+require 'lib/scm/adapters/darcs/validation'
+require 'lib/scm/adapters/darcs/cat_file'
+require 'lib/scm/adapters/darcs/commits'
+require 'lib/scm/adapters/darcs/misc'
+require 'lib/scm/adapters/darcs/pull'
+require 'lib/scm/adapters/darcs/push'
+require 'lib/scm/adapters/darcs/head'
+require 'lib/scm/adapters/darcs/patch'

lib/scm/adapters/factory.rb

@@ -11,6 +11,9 @@ def self.from_path(path)
 			elsif FileTest.exist?(File.join(path, '.hg'))
 				HgAdapter.new(:url => File.expand_path(path)).normalize
 
+			elsif FileTest.exist?(File.join(path, '_darcs'))
+				DarcsAdapter.new(:url => File.expand_path(path)).normalize
+
 			elsif FileTest.exist?(File.join(path, '.bzr'))
 				BzrAdapter.new(:url => File.expand_path(path)).normalize
 

lib/scm/commit.rb

@@ -15,6 +15,8 @@ module Scm
 	# the near future, it is the job of the adapter to make the Git commit chain
 	# appear as much like a single array as possible.
 	#
+	# For Darcs, it is assumed the repo's patch ordering is never changed.
+	#
 	class Commit
 		# This object supports the idea of distinct authors and committers, a la
 		# Git.  However, Ohloh will retain only one of them in its database. It
@@ -32,6 +34,7 @@ class Commit
 		# For Git, the token is the commit SHA1 hash.
 		# For CVS, which does not support atomic commits with unique IDs, we use
 		# the approximate timestamp of the change.
+		# For Darcs, the token is the patch name, and it may not be unique. XXX
 		attr_accessor :token
 
 		# A pointer back to the adapter that contains this commit.