Menu
▾
▴
[etch-devel] SF.net SVN: etch:[267] trunk
|
From: <jh...@us...> - 2011-04-12 01:23:59
|
Revision: 267
http://etch.svn.sourceforge.net/etch/?rev=267&view=rev
Author: jheiss
Date: 2011-04-12 01:23:53 +0000 (Tue, 12 Apr 2011)
Log Message:
-----------
Add support for nokogiri as a choice of XML libraries in addition to the
existing support for libxml and rexml.
Modified Paths:
--------------
trunk/server/lib/etch.rb
trunk/server/lib/etchserver.rb
trunk/test/README
Modified: trunk/server/lib/etch.rb
===================================================================
--- trunk/server/lib/etch.rb 2011-04-12 01:21:17 UTC (rev 266)
+++ trunk/server/lib/etch.rb 2011-04-12 01:23:53 UTC (rev 267)
@@ -3,6 +3,19 @@
require 'digest/sha1' # hexdigest
require 'base64' # decode64, encode64
require 'fileutils' # mkdir_p
+require 'erb'
+require 'versiontype' # Version
+require 'logger'
+
+class Etch
+ def self.xmllib
+ @@xmllib
+ end
+ def self.xmllib=(lib)
+ @@xmllib=lib
+ end
+end
+
# By default we try to use libxml, falling back to rexml if it is not
# available. The xmllib environment variable can be used to force one library
# or the other, mostly for testing purposes.
@@ -10,24 +23,24 @@
if !ENV['xmllib'] || ENV['xmllib'] == 'libxml'
require 'rubygems' # libxml is a gem
require 'libxml'
- @@xmllib = :libxml
+ Etch.xmllib = :libxml
+ elsif ENV['xmllib'] == 'nokogiri'
+ require 'rubygems' # nokogiri is a gem
+ require 'nokogiri'
+ Etch.xmllib = :nokogiri
else
raise LoadError
end
rescue LoadError
if !ENV['xmllib'] || ENV['xmllib'] == 'rexml'
require 'rexml/document'
- @@xmllib = :rexml
+ Etch.xmllib = :rexml
else
raise
end
end
-require 'erb'
-require 'versiontype' # Version
-require 'logger'
class Etch
-
# FIXME: I'm not really proud of this, it seems like there ought to be a way
# to just use one logger. The problem is that on the server we'd like to
# use RAILS_DEFAULT_LOGGER for general logging (which is logging to
@@ -270,8 +283,10 @@
end
# Validate the filtered file against config.dtd
- if !Etch.xmlvalidate(config_xml, @config_dtd)
- raise "Filtered config.xml for #{file} fails validation"
+ begin
+ Etch.xmlvalidate(config_xml, @config_dtd)
+ rescue Exception => e
+ raise Etch.wrap_exception(e, "Filtered config.xml for #{file} fails validation:\n" + e.message)
end
generation_status = :unknown
@@ -846,8 +861,10 @@
end
# Validate the filtered file against commands.dtd
- if !Etch.xmlvalidate(commands_xml, @commands_dtd)
- raise "Filtered commands.xml for #{command} fails validation"
+ begin
+ Etch.xmlvalidate(commands_xml, @commands_dtd)
+ rescue Exception => e
+ raise Etch.wrap_exception(e, "Filtered commands.xml for #{command} fails validation:\n" + e.message)
end
generation_status = :unknown
@@ -1076,102 +1093,154 @@
end
end
+ # These methods provide an abstraction from the underlying XML library in
+ # use, allowing us to use whatever the user has available and switch between
+ # libraries easily.
+
def self.xmlnewdoc
- case @@xmllib
+ case Etch.xmllib
when :libxml
LibXML::XML::Document.new
+ when :nokogiri
+ Nokogiri::XML::Document.new
when :rexml
REXML::Document.new
else
- raise "Unknown @xmllib #{@xmllib}"
+ raise "Unknown XML library #{Etch.xmllib}"
end
end
def self.xmlroot(doc)
- case @@xmllib
+ case Etch.xmllib
when :libxml
doc.root
+ when :nokogiri
+ doc.root
when :rexml
doc.root
else
- raise "Unknown @xmllib #{@xmllib}"
+ raise "Unknown XML library #{Etch.xmllib}"
end
end
def self.xmlsetroot(doc, root)
- case @@xmllib
+ case Etch.xmllib
when :libxml
doc.root = root
+ when :nokogiri
+ doc.root = root
when :rexml
doc << root
else
- raise "Unknown @xmllib #{@xmllib}"
+ raise "Unknown XML library #{Etch.xmllib}"
end
end
def self.xmlload(file)
- case @@xmllib
+ case Etch.xmllib
when :libxml
LibXML::XML::Document.file(file)
+ when :nokogiri
+ Nokogiri::XML(File.open(file)) do |config|
+ # Nokogiri is tolerant of malformed documents by default. Good when
+ # parsing HTML, but there's no reason for us to tolerate errors. We
+ # want to ensure that the user's instructions to us are clear.
+ config.options = Nokogiri::XML::ParseOptions::STRICT
+ end
when :rexml
REXML::Document.new(File.open(file))
else
- raise "Unknown @xmllib #{@xmllib}"
+ raise "Unknown XML library #{Etch.xmllib}"
end
end
def self.xmlloaddtd(dtdfile)
- case @@xmllib
+ case Etch.xmllib
when :libxml
LibXML::XML::Dtd.new(IO.read(dtdfile))
+ when :nokogiri
+ # For some reason there isn't a straightforward way to load a standalone
+ # DTD in Nokogiri
+ dtddoctext = '<!DOCTYPE dtd [' + File.read(dtdfile) + ']'
+ dtddoc = Nokogiri::XML(dtddoctext)
+ dtddoc.children.first
when :rexml
nil
else
- raise "Unknown @xmllib #{@xmllib}"
+ raise "Unknown XML library #{Etch.xmllib}"
end
end
+ # Returns true if validation is successful, or if validation is not
+ # supported by the XML library in use. Raises an exception if validation
+ # fails.
def self.xmlvalidate(xmldoc, dtd)
- case @@xmllib
+ case Etch.xmllib
when :libxml
- xmldoc.validate(dtd)
+ result = xmldoc.validate(dtd)
+ # LibXML::XML::Document#validate is documented to return false if
+ # validation fails. However, as currently implemented it raises an
+ # exception instead. Just in case that behavior ever changes raise an
+ # exception if a false value is returned.
+ if result
+ true
+ else
+ raise "Validation failed"
+ end
+ when :nokogiri
+ errors = dtd.validate(xmldoc)
+ if errors.empty?
+ true
+ else
+ raise errors.join('|')
+ end
when :rexml
true
else
- raise "Unknown @xmllib #{@xmllib}"
+ raise "Unknown XML library #{Etch.xmllib}"
end
end
- def self.xmlnewelem(name)
- case @@xmllib
+ def self.xmlnewelem(name, doc)
+ case Etch.xmllib
when :libxml
LibXML::XML::Node.new(name)
+ when :nokogiri
+ Nokogiri::XML::Element.new(name, doc)
when :rexml
REXML::Element.new(name)
else
- raise "Unknown @xmllib #{@xmllib}"
+ raise "Unknown XML library #{Etch.xmllib}"
end
end
def self.xmleach(xmldoc, xpath, &block)
- case @@xmllib
+ case Etch.xmllib
when :libxml
xmldoc.find(xpath).each(&block)
+ when :nokogiri
+ xmldoc.xpath(xpath).each(&block)
when :rexml
xmldoc.elements.each(xpath, &block)
else
- raise "Unknown @xmllib #{@xmllib}"
+ raise "Unknown XML library #{Etch.xmllib}"
end
end
def self.xmleachall(xmldoc, &block)
- case @@xmllib
+ case Etch.xmllib
when :libxml
if xmldoc.kind_of?(LibXML::XML::Document)
xmldoc.root.each_element(&block)
else
xmldoc.each_element(&block)
end
+ when :nokogiri
+ if xmldoc.kind_of?(Nokogiri::XML::Document)
+ xmldoc.root.element_children.each(&block)
+ else
+ xmldoc.element_children.each(&block)
+ end
when :rexml
if xmldoc.node_type == :document
xmldoc.root.elements.each(&block)
@@ -1179,23 +1248,25 @@
xmldoc.elements.each(&block)
end
else
- raise "Unknown @xmllib #{@xmllib}"
+ raise "Unknown XML library #{Etch.xmllib}"
end
end
def self.xmleachattrall(elem, &block)
- case @@xmllib
+ case Etch.xmllib
when :libxml
elem.attributes.each(&block)
+ when :nokogiri
+ elem.attribute_nodes.each(&block)
when :rexml
elem.attributes.each_attribute(&block)
else
- raise "Unknown @xmllib #{@xmllib}"
+ raise "Unknown XML library #{Etch.xmllib}"
end
end
def self.xmlarray(xmldoc, xpath)
- case @@xmllib
+ case Etch.xmllib
when :libxml
elements = xmldoc.find(xpath)
if elements
@@ -1203,28 +1274,34 @@
else
[]
end
+ when :nokogiri
+ xmldoc.xpath(xpath).to_a
when :rexml
xmldoc.elements.to_a(xpath)
else
- raise "Unknown @xmllib #{@xmllib}"
+ raise "Unknown XML library #{Etch.xmllib}"
end
end
def self.xmlfindfirst(xmldoc, xpath)
- case @@xmllib
+ case Etch.xmllib
when :libxml
xmldoc.find_first(xpath)
+ when :nokogiri
+ xmldoc.at_xpath(xpath)
when :rexml
xmldoc.elements[xpath]
else
- raise "Unknown @xmllib #{@xmllib}"
+ raise "Unknown XML library #{Etch.xmllib}"
end
end
def self.xmltext(elem)
- case @@xmllib
+ case Etch.xmllib
when :libxml
elem.content
+ when :nokogiri
+ elem.content
when :rexml
text = elem.text
# REXML returns nil rather than '' if there is no text
@@ -1234,57 +1311,67 @@
''
end
else
- raise "Unknown @xmllib #{@xmllib}"
+ raise "Unknown XML library #{Etch.xmllib}"
end
end
def self.xmlsettext(elem, text)
- case @@xmllib
+ case Etch.xmllib
when :libxml
elem.content = text
+ when :nokogiri
+ elem.content = text
when :rexml
elem.text = text
else
- raise "Unknown @xmllib #{@xmllib}"
+ raise "Unknown XML library #{Etch.xmllib}"
end
end
def self.xmladd(xmldoc, xpath, name, contents=nil)
- case @@xmllib
+ case Etch.xmllib
when :libxml
elem = LibXML::XML::Node.new(name)
if contents
elem.content = contents
end
xmldoc.find_first(xpath) << elem
- elem
+ when :nokogiri
+ elem = Nokogiri::XML::Node.new(name, xmldoc)
+ if contents
+ elem.content = contents
+ end
+ xmldoc.at_xpath(xpath) << elem
when :rexml
elem = REXML::Element.new(name)
if contents
elem.text = contents
end
xmldoc.elements[xpath].add_element(elem)
- elem
else
- raise "Unknown @xmllib #{@xmllib}"
+ raise "Unknown XML library #{Etch.xmllib}"
end
end
def self.xmlcopyelem(elem, destelem)
- case @@xmllib
+ case Etch.xmllib
when :libxml
destelem << elem.copy(true)
+ when :nokogiri
+ destelem << elem.dup
when :rexml
- destelem.add_element(elem.dup)
+ destelem.add_element(elem.clone)
else
- raise "Unknown @xmllib #{@xmllib}"
+ raise "Unknown XML library #{Etch.xmllib}"
end
end
def self.xmlremove(xmldoc, element)
- case @@xmllib
+ case Etch.xmllib
when :libxml
element.remove!
+ when :nokogiri
+ element.remove
when :rexml
if xmldoc.node_type == :document
xmldoc.root.elements.delete(element)
@@ -1292,40 +1379,51 @@
xmldoc.elements.delete(element)
end
else
- raise "Unknown @xmllib #{@xmllib}"
+ raise "Unknown XML library #{Etch.xmllib}"
end
end
def self.xmlremovepath(xmldoc, xpath)
- case @@xmllib
+ case Etch.xmllib
when :libxml
xmldoc.find(xpath).each { |elem| elem.remove! }
+ when :nokogiri
+ xmldoc.xpath(xpath).each { |elem| elem.remove }
when :rexml
- xmldoc.delete_element(xpath)
+ elem = nil
+ # delete_element only removes the first match, so call it in a loop
+ # until it returns nil to indicate no matching element remain
+ begin
+ elem = xmldoc.delete_element(xpath)
+ end while elem != nil
else
- raise "Unknown @xmllib #{@xmllib}"
+ raise "Unknown XML library #{Etch.xmllib}"
end
end
def self.xmlattradd(elem, attrname, attrvalue)
- case @@xmllib
+ case Etch.xmllib
when :libxml
elem.attributes[attrname] = attrvalue
+ when :nokogiri
+ elem[attrname] = attrvalue
when :rexml
elem.add_attribute(attrname, attrvalue)
else
- raise "Unknown @xmllib #{@xmllib}"
+ raise "Unknown XML library #{Etch.xmllib}"
end
end
def self.xmlattrremove(elem, attribute)
- case @@xmllib
+ case Etch.xmllib
when :libxml
attribute.remove!
+ when :nokogiri
+ attribute.remove
when :rexml
elem.attributes.delete(attribute)
else
- raise "Unknown @xmllib #{@xmllib}"
+ raise "Unknown XML library #{Etch.xmllib}"
end
end
Modified: trunk/server/lib/etchserver.rb
===================================================================
--- trunk/server/lib/etchserver.rb 2011-04-12 01:21:17 UTC (rev 266)
+++ trunk/server/lib/etchserver.rb 2011-04-12 01:23:53 UTC (rev 267)
@@ -3,25 +3,6 @@
require 'openssl'
require 'time' # Time.parse
require 'fileutils' # mkdir_p
-# By default we try to use libxml, falling back to rexml if it is not
-# available. The xmllib environment variable can be used to force one library
-# or the other, mostly for testing purposes.
-begin
- if !ENV['xmllib'] || ENV['xmllib'] == 'libxml'
- require 'rubygems' # libxml is a gem
- require 'libxml'
- @@xmllib = :libxml
- else
- raise LoadError
- end
-rescue LoadError
- if !ENV['xmllib'] || ENV['xmllib'] == 'rexml'
- require 'rexml/document'
- @@xmllib = :rexml
- else
- raise
- end
-end
require 'logger'
require 'etch'
@@ -372,11 +353,11 @@
# Generate the XML document to return to the client
response_xml = Etch.xmlnewdoc
- responseroot = Etch.xmlnewelem('files')
+ responseroot = Etch.xmlnewelem('files', response_xml)
Etch.xmlsetroot(response_xml, responseroot)
# Add configs for files we generated
if response[:configs]
- configs_xml = Etch.xmlnewelem('configs')
+ configs_xml = Etch.xmlnewelem('configs', response_xml)
response[:configs].each do |file, config_xml|
# Update the stored record of the config
# Exclude configs which correspond to files for which we're requesting
@@ -408,18 +389,18 @@
end
end
if !need_sum.empty?
- need_sums_xml = Etch.xmlnewelem('need_sums')
+ need_sums_xml = Etch.xmlnewelem('need_sums', response_xml)
need_sum.each do |need|
- need_xml = Etch.xmlnewelem('need_sum')
+ need_xml = Etch.xmlnewelem('need_sum', response_xml)
Etch.xmlsettext(need_xml, need)
need_sums_xml << need_xml
end
responseroot << need_sums_xml
end
if !need_orig.empty?
- need_origs_xml = Etch.xmlnewelem('need_origs')
+ need_origs_xml = Etch.xmlnewelem('need_origs', response_xml)
need_orig.each do |need|
- need_xml = Etch.xmlnewelem('need_orig')
+ need_xml = Etch.xmlnewelem('need_orig', response_xml)
Etch.xmlsettext(need_xml, need)
need_origs_xml << need_xml
end
@@ -431,7 +412,7 @@
# "commands", so we have to use something different here as the XML
# element we insert all of those into as part of the response.
if response[:allcommands]
- commands_xml = Etch.xmlnewelem('allcommands')
+ commands_xml = Etch.xmlnewelem('allcommands', response_xml)
response[:allcommands].each do |commandname, command_xml|
# Update the stored record of the command
config = EtchConfig.find_or_create_by_client_id_and_file(:client_id => @client.id, :file => commandname, :config => command_xml.to_s)
@@ -444,19 +425,19 @@
responseroot << commands_xml
end
if response[:retrycommands]
- retrycommands_xml = Etch.xmlnewelem('retrycommands')
+ retrycommands_xml = Etch.xmlnewelem('retrycommands', response_xml)
response[:retrycommands].each_key do |commandname|
- retry_xml = Etch.xmlnewelem('retrycommand')
+ retry_xml = Etch.xmlnewelem('retrycommand', response_xml)
Etch.xmlsettext(retry_xml, commandname)
retrycommands_xml << retry_xml
end
responseroot << retrycommands_xml
end
- # FIXME: clean up XML formatting
+ # Clean up XML formatting
# But only if we're in debug mode, in regular mode nobody but the
# machines will see the XML and they don't care if it is pretty.
- # Tidy's formatting breaks things, it inserts leading/trailing whitespace into text nodes
+ # FIXME: Tidy's formatting breaks things, it inserts leading/trailing whitespace into text nodes
if @debug && false
require 'tidy'
Tidy.path = '/sw/lib/libtidy.dylib'
Modified: trunk/test/README
===================================================================
--- trunk/test/README 2011-04-12 01:21:17 UTC (rev 266)
+++ trunk/test/README 2011-04-12 01:23:53 UTC (rev 267)
@@ -3,8 +3,8 @@
To execute a specific test method run
"rake test TEST=test/file.rb TESTOPTS='--name=test_files'"
-To force a particular XML library set xmllib=libxml or xmllib=rexml in
-your environment before running the tests.
+To force a particular XML library set xmllib=libxml or xmllib=nokogiri or
+xmllib=rexml in your environment before running the tests.
Some of the older files here have all of their tests in one method. Over
time I'm breaking those up into multiple methods so that it is easier to
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|
