ohloh_scm 2.5.1 → 3.0.1

data/lib/ohloh_scm.rb

@@ -1,39 +1,33 @@
+# frozen_string_literal: true
+
 module OhlohScm
 end
 
-#require 'rbconfig'
-
-#$: << File.join(File.dirname(__FILE__),"..")
+require 'time'
+require 'tmpdir'
+require 'forwardable'
+require 'nokogiri'
 
-require_relative 'ohloh_scm/shellout'
-require_relative 'ohloh_scm/scratch_dir'
-require_relative 'ohloh_scm/commit'
-require_relative 'ohloh_scm/diff'
+require 'ohloh_scm/string_extensions'
+require 'ohloh_scm/version'
+require 'ohloh_scm/system'
+require 'ohloh_scm/diff'
+require 'ohloh_scm/commit'
+require 'ohloh_scm/parser'
+require 'ohloh_scm/core'
+require 'ohloh_scm/scm'
+require 'ohloh_scm/activity'
+require 'ohloh_scm/status'
+require 'ohloh_scm/validation'
+require 'ohloh_scm/py_bridge'
 
-require_relative 'ohloh_scm/adapters/abstract_adapter'
-require_relative 'ohloh_scm/adapters/cvs_adapter'
-require_relative 'ohloh_scm/adapters/svn_adapter'
-require_relative 'ohloh_scm/adapters/svn_chain_adapter'
-require_relative 'ohloh_scm/adapters/git_adapter'
-require_relative 'ohloh_scm/adapters/hg_adapter'
-require_relative 'ohloh_scm/adapters/hglib_adapter'
-require_relative 'ohloh_scm/adapters/bzr_adapter'
-require_relative 'ohloh_scm/adapters/bzrlib_adapter'
-require_relative 'ohloh_scm/adapters/git_svn_adapter'
-require_relative 'ohloh_scm/adapters/factory'
+require 'ohloh_scm/hg'
+require 'ohloh_scm/git'
+require 'ohloh_scm/bzr'
+require 'ohloh_scm/cvs'
+require 'ohloh_scm/svn'
+require 'ohloh_scm/git_svn'
 
-require_relative 'ohloh_scm/parsers/parser'
-require_relative 'ohloh_scm/parsers/branch_number'
-require_relative 'ohloh_scm/parsers/cvs_parser'
-require_relative 'ohloh_scm/parsers/svn_parser'
-require_relative 'ohloh_scm/parsers/svn_xml_parser'
-require_relative 'ohloh_scm/parsers/git_parser'
-require_relative 'ohloh_scm/parsers/git_styled_parser'
-require_relative 'ohloh_scm/parsers/hg_parser'
-require_relative 'ohloh_scm/parsers/hg_styled_parser'
-require_relative 'ohloh_scm/parsers/bzr_xml_parser'
-require_relative 'ohloh_scm/parsers/bzr_parser'
+require 'ohloh_scm/factory'
 
-require_relative 'ohloh_scm/parsers/array_writer'
-require_relative 'ohloh_scm/parsers/xml_writer'
-require_relative 'ohloh_scm/parsers/human_writer'
+`#{File.expand_path('../.bin/check_scm_version', __dir__)}`

data/lib/ohloh_scm/activity.rb

@@ -0,0 +1,38 @@
+# frozen_string_literal: true
+
+module OhlohScm
+  class Activity
+    include OhlohScm::System
+    extend Forwardable
+    def_delegators :@core, :scm, :status
+    def_delegators :scm, :url
+
+    def initialize(core)
+      @core = core
+    end
+
+    def log_filename
+      File.join(Dir.tmpdir, url.gsub(/\W/, '') + '.log')
+    end
+
+    def tags; end
+
+    def export; end
+
+    def export_tag; end
+
+    def head_token; end
+
+    def each_commit; end
+
+    def commits; end
+
+    def commit_tokens; end
+
+    def commit_count; end
+
+    def diffs; end
+
+    def cleanup; end
+  end
+end

data/lib/ohloh_scm/bzr.rb

@@ -0,0 +1,12 @@
+# frozen_string_literal: true
+
+require_relative 'parser/bzr_xml_parser'
+require_relative 'bzr/scm'
+require_relative 'bzr/status'
+require_relative 'bzr/activity'
+require_relative 'bzr/validation'
+
+module OhlohScm
+  module Bzr
+  end
+end

data/lib/ohloh_scm/bzr/activity.rb

@@ -0,0 +1,191 @@
+# frozen_string_literal: true
+
+# rubocop:disable Metrics/ClassLength
+module OhlohScm
+  module Bzr
+    class Activity < OhlohScm::Activity
+      # rubocop:disable Metrics/MethodLength
+      def tags
+        tag_strings.map do |tag_string|
+          parse_tag_names_and_revision = tag_string.split(/\s+/)
+          if parse_tag_names_and_revision.size > 1
+            tag_name = parse_tag_names_and_revision[0..-2].join(' ')
+            rev = parse_tag_names_and_revision.last
+          else
+            tag_name = parse_tag_names_and_revision.first
+            rev = nil
+          end
+          next if rev == '?' || tag_name == '....'
+
+          [tag_name, rev, Time.parse(time_string(rev))]
+        end.compact
+      end
+      # rubocop:enable Metrics/MethodLength
+
+      def export_tag(dest_dir, tag_name)
+        run "cd '#{path}' && bzr export -r #{tag_name} #{dest_dir}"
+      end
+
+      def export(dest_dir, token = head_token)
+        # Unlike other SCMs, Bzr doesn't simply place the contents into dest_dir.
+        # It actually *creates* dest_dir. Since it should already exist at this point,
+        # first we have to delete it.
+        Dir.delete(dest_dir) if File.exist?(dest_dir)
+
+        run "cd '#{url}' && bzr export --format=dir -r #{to_rev_param(token)} '#{dest_dir}'"
+      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(opts = {})
+        after = opts[:after]
+        log = run("#{rev_list_command(opts)} | cat")
+        a = OhlohScm::BzrXmlParser.parse(log)
+
+        if after && (i = a.index { |commit| commit.token == after })
+          a[(i + 1)..-1]
+        else
+          a
+        end
+      end
+
+      # Return the list of commit tokens following +after+.
+      def commit_tokens(opts = {})
+        commits(opts).map(&:token)
+      end
+
+      # Return the number of commits in the repository following +after+.
+      def commit_count(opts = {})
+        commit_tokens(opts).size
+      end
+
+      def each_commit(opts = {})
+        after = opts[:after]
+        skip_commits = !after.nil? # Don't emit any commits until the 'after' resume point passes
+
+        safe_open_log_file(opts) do |io|
+          OhlohScm::BzrXmlParser.parse(io) do |commit|
+            yield remove_directories(commit) if block_given? && !skip_commits
+            skip_commits = false if commit.token == after
+          end
+        end
+      end
+
+      def head_token
+        run("cd '#{url}' && bzr log --limit 1 --show-id #{url} 2> /dev/null"\
+              " | grep ^revision-id | cut -f2 -d' '").strip
+      end
+
+      def head
+        verbose_commit(head_token)
+      end
+
+      def parent_tokens(commit)
+        bzr_client.parent_tokens(commit.token)
+      end
+
+      def parents(commit)
+        parent_tokens(commit).collect { |token| verbose_commit(token) }
+      end
+
+      def cat_file(commit, diff)
+        cat(commit.token, diff.path)
+      end
+
+      def cat_file_parent(commit, diff)
+        first_parent_token = parent_tokens(commit).first
+        cat(first_parent_token, diff.path) if first_parent_token
+      end
+
+      def cleanup
+        bzr_client.shutdown
+      end
+
+      private
+
+      # Returns a single commit, including its diffs
+      def verbose_commit(token)
+        cmd = "cd '#{url}' && bzr xmllog --show-id -v --limit 1 -c #{to_rev_param(token)}"
+        log = run(cmd)
+        OhlohScm::BzrXmlParser.parse(log).first
+      end
+
+      def to_rev_param(rev = nil)
+        case rev
+        when nil
+          1
+        when Integer
+          rev.to_s
+        when /^\d+$/
+          rev
+        else
+          "'revid:#{rev}'"
+        end
+      end
+
+      def rev_list_command(opts = {})
+        after = opts[:after]
+        trunk_only = opts[:trunk_only] ? '--levels=1' : '--include-merges'
+        "cd '#{url}' && bzr xmllog --show-id --forward #{trunk_only} -r #{to_rev_param(after)}.."
+      end
+
+      # Returns a file handle to the log.
+      # In our standard, the log should include everything AFTER
+      # +after+. However, bzr doesn't work that way; it returns
+      # everything after and INCLUDING +after+. Therefore, consumers
+      # of this file should check for and reject the duplicate commit.
+      def safe_open_log_file(opts = {})
+        return '' if opts[:after] && opts[:after] == head_token
+
+        open_log_file(opts) { |io| yield io }
+      end
+
+      def open_log_file(opts = {})
+        cmd = "#{rev_list_command(opts)} -v > #{log_filename}"
+        run cmd
+        File.open(log_filename, 'r') { |io| yield io }
+      ensure
+        File.delete(log_filename) if File.exist?(log_filename)
+      end
+
+      # Ohloh tracks only files, not directories. This function removes directories
+      # from the commit diffs.
+      def remove_directories(commit)
+        commit.diffs.delete_if { |d| d.path[-1..-1] == '/' }
+        commit
+      end
+
+      def cat(revision, path)
+        bzr_client.cat_file(revision, path)
+      end
+
+      # Bzr doesn't like it when the filename includes a colon
+      # Also, fix the case where the filename includes a single quote
+      def escape(path)
+        path.gsub(/[:]/) { |c| '\\' + c }.gsub("'", "''")
+      end
+
+      def tag_strings
+        run("cd '#{url}' && bzr tags").split(/\n/)
+      end
+
+      def time_string(rev)
+        run("cd '#{url}' && bzr log -r #{rev} | grep 'timestamp:' | sed 's/timestamp://'")
+      end
+
+      def bzr_client
+        @bzr_client ||= setup_bzr_client
+      end
+
+      def setup_bzr_client
+        bzr_client = PyBridge::BzrClient.new(url)
+        bzr_client.start
+        bzr_client
+      end
+    end
+  end
+end
+# rubocop:enable Metrics/ClassLength

data/lib/ohloh_scm/bzr/scm.rb

@@ -0,0 +1,24 @@
+# frozen_string_literal: true
+
+module OhlohScm
+  module Bzr
+    class Scm < OhlohScm::Scm
+      def pull(from, callback)
+        callback.update(0, 1)
+
+        if status.exist?
+          run "cd '#{url}' && bzr revert && bzr pull --overwrite '#{from.url}'"
+        else
+          run "rm -rf '#{url}'"
+          run "bzr branch '#{from.url}' '#{url}'"
+        end
+
+        callback.update(1, 1)
+      end
+
+      def vcs_path
+        "#{url}/.bzr"
+      end
+    end
+  end
+end

data/lib/ohloh_scm/bzr/status.rb

@@ -0,0 +1,8 @@
+# frozen_string_literal: true
+
+module OhlohScm
+  module Bzr
+    class Status < OhlohScm::Status
+    end
+  end
+end

data/lib/ohloh_scm/bzr/validation.rb

@@ -0,0 +1,18 @@
+# frozen_string_literal: true
+
+module OhlohScm
+  module Bzr
+    class Validation < OhlohScm::Validation
+      private
+
+      def validate_server_connection
+        msg = "The server did not respond to the 'bzr revno' command. Is the URL correct?"
+        @errors << [:failed, msg] unless status.exist?
+      end
+
+      def public_url_regex
+        %r{^(((http|https|bzr)://(\w+@)?[\w\-\.]+(:\d+)?/)|(lp:[\w\-\.\~]))[\w\-\./\~\+]*$}
+      end
+    end
+  end
+end

data/lib/ohloh_scm/commit.rb

@@ -1,49 +1,52 @@
+# frozen_string_literal: true
+
 module OhlohScm
-	# A commit is a collection of diffs united by a single timestamp, author, and
-	# message.
-	#
-	# Ohloh's internal data model assumes that commits are immutable, and exist
-	# in a singly-linked list. That is, commits can be nicely numbered a la
-	# Subversion, and new commits are always added to the end of the list.
-	#
-	# This works for CVS and Subversion, but unfortunately, does not at all map
-	# to the DAG used by Git, which allows a commit to have multiple parents and
-	# children, and which allows new commits to appear during a pull which have
-	# timestamps older than previously known commits.
-	#
-	# This means that Ohloh's support for systems like Git is crude at best. For
-	# 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.
-	#
-	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
-		# prefers author, but will fall back to committer if no author is given.
-		attr_accessor :author_name, :author_email, :author_date, :committer_name, :committer_email, :committer_date
+  # A commit is a collection of diffs united by a single timestamp, author, and
+  # message.
+  #
+  # OpenHub's internal data model assumes that commits are immutable, and exist
+  # in a singly-linked list. That is, commits can be nicely numbered a la
+  # Subversion, and new commits are always added to the end of the list.
+  #
+  # This works for CVS and Subversion, but unfortunately, does not at all map
+  # to the DAG used by Git, which allows a commit to have multiple parents and
+  # children, and which allows new commits to appear during a pull which have
+  # timestamps older than previously known commits.
+  #
+  # This means that OpenHub's support for systems like Git is crude at best. For
+  # 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.
+  #
+  class Commit
+    # This object supports the idea of distinct authors and committers, a la
+    # Git.  However, OpenHub will retain only one of them in its database. It
+    # prefers author, but will fall back to committer if no author is given.
+    attr_accessor :author_name, :author_email, :author_date,
+                  :committer_name, :committer_email, :committer_date
 
-		attr_accessor :message
+    attr_accessor :message
 
-		attr_accessor :diffs
+    attr_accessor :diffs
 
-		# The token is used to uniquely identify a commit, and can be any type of
-		# adapter-specific data.
-		#
-		# For Subversion, the token is the revision number.
-		# 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.
-		attr_accessor :token
+    # The token is used to uniquely identify a commit, and can be any type of
+    # adapter-specific data.
+    #
+    # For Subversion, the token is the revision number.
+    # 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.
+    attr_accessor :token
 
-		# A pointer back to the adapter that contains this commit.
-		attr_accessor :scm
+    # A pointer back to the adapter that contains this commit.
+    attr_accessor :scm
 
-		# Hack. To optimize CVS updates, we will store the names of all the
-		# directories that require updating during this commit. Ohloh itself never
-		# actually sees this.
-		attr_accessor :directories
+    # Hack. To optimize CVS updates, we will store the names of all the
+    # directories that require updating during this commit. OpenHub itself never
+    # actually sees this.
+    attr_accessor :directories
 
-		def initialize(params={})
-			params.each { |k,v| send(k.to_s + '=', v) if respond_to?(k.to_s + '=') }
-		end
-	end
+    def initialize(params = {})
+      params.each { |k, v| send(k.to_s + '=', v) if respond_to?(k.to_s + '=') }
+    end
+  end
 end