xmlscan

The fastest XML parser written in 100% pure Ruby.

Abstract

XMLscan is one of non-validating XML parser written in 100% pure Ruby.

XMLscan’s features are as follows:

  • 100% pure Ruby

XMLscan doesn’t require any extension libraries, so it completely works only with a Ruby interpreter version 1.6 or above. (It also needs no standard-bundled extension library.)

  • Compliant to the specification

XMLscan has been developed to satisfy all conditions, described in XML 1.0 Specification and required to a non-validating XML processor

  • High-speed

XMLscan is, probably, the fastest parser among all existing XML/HTML parsers written in pure Ruby.

  • Support for various CES.

XMLscan can parse an XML document encoded in at least iso-8859-*, EUC-*, Shift_JIS, and UTF-8 as it is. UTF-16 is not supported directly, though.

  • Just parsing

The role of xmlscan is just to parse an XML document. XMLscan doesn’t provide high-level features to easily handle an XML document. XMLscan is assumed to be used as a core part of a library providing such features.

  • HTML

XMLscan contains htmlscan, an HTML parser.

New Scanner Translater Feature with 0.3.0

Created for a use case in Wagn (wagn.org)

The fragment is pretty much the whole parsing as used in Wagn:

pairs = XMLScan::XMLProcessor.process(io, {:key=>:name, :element=>:card,
                  :substitute=>":transclude|{{:name}}", :extras=>[:type]})

The ‘io’ object is directly passed from the request.body in a XML POST action.

ref: github.com/GerryG/wagn/commit/46e57dfd88cde45d33fe7f167d5803ab819ceef3#commitcomment-1012403

See the tests cases in test/integration for some examples.

Character encodings

By default, the value of global variable $KCODE decides which CES (character encoding scheme) is assumed for xmlscan to parse an XML document.

You need to set $KCODE or XMLScan::XMLScanner#kcode= an appropriate value to parse an XML document encoded in EUC-*, Shift_JIS, or UTF-8.

UTF-16 is not supported directly. You should convert it into UTF-8 before parsing.

Character encoding needs more work with 0.3.0+. The character maps were taken out in favor of p{} RE classes in ruby, but it needs attention still.

XML Namespaces

XML Namespaces have been already implemented in xmlscan/namespace.rb. However, since its interface is going to be modified, this feature is undocumented now.

Class Reference

XMLScan::Error The superclass for all exceptions related to xmlscan. These exceptions are raised by XMLScan::Visitor by default when it receives an error report from a parser, such as XMLScan::XMLScanner or XMLScan::XMLParser. Each parser never raises these exceptions by itself.

XMLScan::ParseError

An error except a constraint violation, for example, an XML document is unmatched with a production.

XMLScan::NotWellFormedError

Raised when an XML document violates an well-formedness constraint.

XMLScan::NotValidError

Raised when an XML document violates an validity constraint.

XMLScan::Visitor

Mix-in for receiving the result of parsing an XML document. Each parser included in xmlscan parses an XML document from the beginning, and calls each specific method of given instance of XMLScan::Visitor for each syntactic element, such as a tag. It is ensured that these calls is in order of the appearance in the document from the beginning.

Changes: The scanner/parser now sends some additional arguments with each event, primarily the original string that was parsed for the even, but also a hash of attributes for stag_end. This supports scanning for replacement of some elements very simple.

Methods:

Without special notice, the following methods do nothing by default.

XMLScan::Visitor#parse_error(msg)

Called when the parser meets an error except a constraint violation, for example, an XML document is unmatched with a production. By default, this method raises XMLScan::ParseError exception. If no exception is raised and this method returns normally, the parser recovers the error and continues to parse.

XMLScan::Visitor#wellformed_error(msg)

Called when the parser meets an well-formedness constraint violation. By default, this method raises XMLScan::NotWellFormedError exception. If no exception is raised and this method returns normally, the parser recovers the error and continues to parse.

XMLScan::Visitor#valid_error(msg)

Called when the parser meets validity constraint violation. By default, this method raises XMLScan::NotValidError exception. If no exception is raised and this method returns normally, the parser recovers the error and continues to parse. FYI, current version of xmlscan includes no validating XML processor. This method is reserved for future versions.

XMLScan::Visitor#warning(msg)

Called when the parser meets a non-error but unrecommended thing or a syntax which xmlscan is not able to parse.

XMLScan::Visitor#on_start_document

Called just before the parser starts parsing an XML document.

After this method is called, corresponding XMLScan::Visitor#on_end_document method is always called.

XMLScan::Visitor#on_end_document

Called after the parser reaches the end of an XML document.

XMLScan::Visitor#on_xmldecl XMLScan::Visitor#on_xmldecl_version(str) XMLScan::Visitor#on_xmldecl_encoding(str) XMLScan::Visitor#on_xmldecl_standalone(str) XMLScan::Visitor#on_xmldecl_other(name, value) XMLScan::Visitor#on_xmldecl_end

Called when the parser meets an XML declaration. <?xml version=“1.0” encoding=“euc-jp” standalone=“yes” ?> ^ ^ ^ ^ ^ 1 2 3 4 5

    method                 argument
--------------------------------------
 1: on_xmldecl
 2: on_xmldecl_version     ("1.0")
 3: on_xmldecl_encoding    ("euc-jp")
 4: on_xmldecl_standalone  ("yes")
 5: on_xmldecl_end

When an XML declaration is found, both on_xmldecl and on_xmldecl_end method are always called. Any other methods are called only when the corresponding syntaxes are found.

When a declaration except version, encoding, and standalone is found in an XML declaration, on_xmldecl_other method is called. Since such a declaration is not permitted, note that the parser always calls XMLScan::Visitor#parse_error method before calling on_xmldecl_other method.

XMLScan::Visitor#on_doctype(root, pubid, sysid)

Called when the parser meets a document type declaration.

document argument


1: <!DOCTYPE foo>                      ('foo', nil,   nil)
2: <!DOCTYPE foo SYSTEM "bar">         ('foo', nil,   'bar')
3: <!DOCTYPE foo PUBLIC "bar">         ('foo', 'bar',  nil )
4: <!DOCTYPE foo PUBLIC "bar" "baz">   ('foo', 'bar', 'baz')

XMLScan::Visitor#on_prolog_space(str)

Called when the parser meets whitespaces in prolog.

XMLScan::Visitor#on_comment(str)

Called when the parser meets a comment.

XMLScan::Visitor#on_pi(target, pi)

Called when the parser meets a processing instruction.

XMLScan::Visitor#on_chardata(str)

Called when the parser meets character data.

XMLScan::Visitor#on_cdata(str)

Called when the parser meets a CDATA section.

XMLScan::Visitor#on_entityref(ref)

Called when the parser meets a general entity reference in a place except an attribute value.

XMLScan::Visitor#on_charref(code) XMLScan::Visitor#on_charref_hex(code)

Called when the parser meets a character reference in a place except an attribute value.

When the character code is represented by decimals, on_charref is called. When by hexadecimals, on_charref_hex is called. code is an integer.

XMLScan::Visitor#on_stag(name) XMLScan::Visitor#on_attribute(name) XMLScan::Visitor#on_attr_value(str) XMLScan::Visitor#on_attr_entityref(ref) XMLScan::Visitor#on_attr_charref(code) XMLScan::Visitor#on_attr_charref_hex(code) XMLScan::Visitor#on_attribute_end(name) XMLScan::Visitor#on_stag_end_empty(name) XMLScan::Visitor#on_stag_end(name)

Called when the parser meets an XML declaration. <hoge fuga=“foo&bar;&&baz” > ^ ^ ^ ^ ^ ^ ^ ^ ^ 1 2 3 4 5 6 7 8 9

    method                 argument
------------------------------------
 1: on_stag                ('hoge')
 2: on_attribute           ('fuga')
 3: on_attr_value          ('foo')
 4: on_attr_entityref      ('bar')
 5: on_attr_charref        (38)
 6: on_attr_charref_hex    (38)
 7: on_attr_value          ('baz')
 8: on_attribute_end       ('fuga')
 9: on_stag_end            ('hoge')
     or
    on_stag_end_empty      ('hoge')

When a start tag is found, both on_stag and corresponding either on_stag_end or on_stag_end_empty method are always called. Any other methods are called only when at least one attribute is found in the start tag.

When an attribute is found, both on_attribute and on_attribute_end method are always called. If the attribute value is empty, only these two methods are called.

When the parser meets a general entity reference in an attribute value, it calls on_attr_entityref method.

When the parser meets a character reference in an attribute value, it calls either on_charref or on_charref_hex method. If the tag is an empty element tag, on_stag_end_empty method is called instead of on_stag_end method.

XMLScan::Visitor#on_etag(name)

Called when the parser meets an end tag.

XMLScan::XMLScanner The scanner which tokenizes an XML document and recognize tags, and so on.

The conformance of XMLScan::XMLScanner to the specification is described in another document. SuperClass:

Object

Class Methods:

XMLScan::XMLScanner.new(visitor[, option …])

Creates an instance. visitor is a instance of XMLScan::Visitor and receives the result of parsing from the XMLScan::Scanner object.

You can specify one of more option as a string or symbol.

XMLScan::Scanner’s options are as follows:

‘strict_char’

This option is enabled after require ‘xmlscan/xmlchar’.

XMLScan::Scanner checks whether an XML document includes an illegal character. The performance decreases sharply.

Methods:

XMLScan::XMLScanner#kcode= arg

Sets CES. Available values for code are same as $KCODE except nil. If code is nil, $KCODE decides the CES.

XMLScan::XMLScanner#kcode

Returns CES. The format of the return value is same as Regexp#kcode. If this method returns nil, it represents that $KCODE decides the CES.

XMLScan::XMLScanner#parse(source)

Parses source as an XML document. source must be a string, an array of strings, or an object which responds to gets method which behaves same as IO#gets does.

XMLScan::XMLParser

The non-validating XML parser.

The conformance of XMLScan::XMLParser to the specification is described in another document.

SuperClass:

XMLScan::XMLScanner

Class Methods:

XMLScan::XMLParser.new(visitor[, option …])

XMLScan::XMLParser makes sure the following for each method of visitor:

XMLScan::Visitor#on_stag

After calling this method, XMLScan::Parser always call corresponding XMLScan::Visitor#on_etag method.

In addition, if you never intend error recovery, method calls which must not be occurred in a well-formed XML document are all suppressed.

XMLScan::HTMLScanner An HTML parser based on XMLScan::XMLScanner. The conformance of XMLScan::HTMLScanner to the specification is described in another document. SuperClass:

XMLScan::XMLScanner

Class Methods:

XMLScan::HTMLScanner.new(visitor[, option …])

XMLScan::HTMLScanner makes sure the following for each method of visitor:

XMLScan::Visitor#on_xmldecl XMLScan::Visitor#on_xmldecl_version XMLScan::Visitor#on_xmldecl_encoding XMLScan::Visitor#on_xmldecl_standalone XMLScan::Visitor#on_xmldecl_end

An XML declaration never appears in an HTML document, so XMLScan::HTMLScanner never calls these methods.

XMLScan::Visitor#on_stag_end_empty

An empty element tag never appears in an HTML document, so XMLScan::HTMLScanner never calls this method. An empty element tag causes a parse error.

XMLScan::Visitor#wellformed_error

There is no well-formedness constraint for HTML, so XMLScan::HTMLScanner never calls this method.

LICENCE

This module’s license is the same of Ruby. You can redistribute it and/or modify it under the same term as Ruby.

* Japanese: http://www.ruby-lang.org/ja/LICENSE.txt
* English:  http://www.ruby-lang.org/en/LICENSE.txt

UENO Katsuhiro