39 KiB

Raw Permalink Blame History

Changelog

This document contains details of the various releases and their release dates. Dates are in the format yyyy-mm-dd.

3.4 - 2022-08-02

This release includes a change that when setting the child nodes of a node A, node A takes ownership over the entire new child tree. See merge request https://gitlab.com/yorickpeterse/oga/-/merge_requests/194 for more details.

3.3 - 2020-07-27

This release adds to_s as an alias for to_xml, thanks to Roy Zwambag. See merge request https://gitlab.com/yorickpeterse/oga/-/merge_requests/192 for more information.

3.2 - 2020-01-10

This release fixes a few warnings that would show up when using Oga on Ruby 2.7.0. See https://gitlab.com/yorickpeterse/oga/merge_requests/190 for more information.

3.1 - 2020-01-08

This release fixes a bug in the XML lexer that prevented the parsing of doctypes using "public" or "system" instead of "PUBLIC"/"SYSTEM". See issue https://gitlab.com/yorickpeterse/oga/issues/199 for more information.

3.0 - 2019-12-03

This release bumps the Ruby version requirement to Ruby 2.3.0, as we haven't supported older versions for several years now. We also no longer officially support Rubinius.

2.17 - 2019-12-02

Elements using the default XML namespace can now be queried using XPath queries, which was broken for quite a while.

See commit https://gitlab.com/yorickpeterse/oga/commit/95da93949bf613612981f5cd7decc0d2c2a60e15 for more information.

2.16 - 2019-11-29

XPath namespace aliases can now be used when querying elements using XPath expressions.
Several RDOc and RubyGems deprecation warnings have been resolved.

See the following commits for more information:

2.15 - 2018-04-11

The HTML parser now allows th elements to occur in thead, tbody, and tfoot elements. See issue https://gitlab.com/yorickpeterse/oga/issues/190 for more information.

2.14 - 2018-01-30

Various methods that yield a block now return an Enumerator when no block is provided. See merge request https://gitlab.com/yorickpeterse/oga/merge_requests/184 for more information.

2.13 - 2018-01-05

Leading and trailing whitespace is now removed from CSS selectors. See https://gitlab.com/yorickpeterse/oga/merge_requests/183 for more information.

2.12 - 2017-12-29

Element start tags containing other start tags (e.g. <script<script>) are now parsed correctly.

See f574197ea6 for more information.

2.11 - 2017-09-07

Various Ruby warnings have been resolved by Loic Nageleisen. See pull request https://gitlab.com/yorickpeterse/oga/pull/180 for more information.

2.10 - 2017-04-18

Fix `Element#attribute` for HTML documents when using Symbol arguments

You can now pass a Symbol to Oga::XML::Element#attribute for both XML and HTML documents, previously this only worked for XML documents. See PR #174 for more information.

2.9 - 2017-02-10

Closing tags for HTML void elements

Certain HTML elements such as <img> and <link> (called "void elements" in Oga) are now closed using a > tag instead of />. In other words, instead of outputting <img src="..." /> Oga now outputs <img src="...">.

Doctypes are now Nodes

Each Doctype now inherits from Oga::XML::Node. This makes it possible to parse documents where a doctype is located in a child node. However, in these cases Oga will not populate Oga::XML::Document#doctype as this can not be done in an efficient way.

2.8 - 2017-01-04

Ruby 2.4 deprecates Fixnum in favour of Integer, producing warnings whenever Fixnum is used. Oga 2.8 contains a fix contributed by Po Shan Cheah to remove these deprecation warnings. See commit c75ca96d22 for more information.

2.7 - 2016-09-27

Closing Elements When Generating XML

When generating XML Oga now properly closes elements with siblings but without children. See commit e0e0687dc2 for more information.

Newlines After Doctypes

When generating XML a newline would be inserted after a doctype. If another newline would follow in a text node this would lead to multiple newlines being present. Oga now ensures there is only 1 newline following a doctype. See commit e0e0687dc2 for more information.

Processing Instructions With Namespace Prefixes

The XML lexer now supports processing instructions containing namespace prefixes such as <?xml:foo ?>. See commit 01fa1513f4 for more information.

XML Declarations Are Now Processing Instructions

The class Oga::XML::XmlDeclaration now extends Oga::XML::ProcessingInstruction. This allows documents to contain XML declarations in nested elements, instead of only allowing this at the root of the document. See commit 116b9b0ceb for more information.

Aliases For Getting & Setting Attributes

The methods Oga::XML::Element#get and Oga::XML::Element#set are now aliased as #[] and #[]= respectively. See d40baf0c72 for more information and thanks to Scott Wheeler for contributing the patch.

2.6 - 2016-09-10

This release fixes a bug in the XML generation code that would cause it to get stuck in the generation loop. See issue https://gitlab.com/yorickpeterse/oga/issues/161 and commit 38284278d5 for more information.

2.5 - 2016-09-06

This release fixes a bug in the XML parser that would prevent it from parsing doctypes that contain a mixture of public/system IDs, a name, and inline rules.

See issue https://gitlab.com/yorickpeterse/oga/issues/159 and commit 68f1f9f660 for more information.

2.4 - 2016-09-04

Serialising Large Documents

Oga can now serialise large documents without causing the call stack to overflow thanks to the new Oga::XML::Generator class. This class can generate XML without using a stack at all.

See issue https://gitlab.com/yorickpeterse/oga/issues/158 and commit dd138981f6 for more information.

Faster retrieval of previous/next nodes

The methods Oga::XML::Node#previous and Oga::XML::Node#next now simply return the value of an instance variable instead of calculating the previous/next node on the fly. This greatly improves the performance of these methods at the cost of a bit of extra work when adding or removing nodes from a NodeSet.

See commit 5a58b14137 for more information.

2.3 - 2016-07-13

Thanks to various changes provided by Erik Michaels-Ober Oga can now be used to parse XML input from a pipe (as returned by for example IO.pipe). See the following pull request for more information: https://gitlab.com/yorickpeterse/oga/pull/154.

2.2 - 2016-02-23

XPath support for nested pipe operators

Nested pipe operators such as a | b | c are now supported as XPath expressions. See issue https://gitlab.com/yorickpeterse/oga/issues/149 and commit 6d3c5c2ce9 for more information.

2.1 - 2016-02-09

Preserving entities that can't be decoded

Decoding of invalid XML/HTML entities now results in these entities being preserved as-is, instead of raising an EncodingError in certain places. See https://gitlab.com/yorickpeterse/oga/issues/143 and commit 5bfc2d50f2 for more information.

New Versioning Format

Starting with this release the patch number is dropped from the version. This means version numbers are now in the format MAJOR.MINOR. See the README for more information.

2.0.0 - 2015-12-26

Fixed parsing HTML identifiers

HTML identifiers are now parsed correctly. This means that for the element <foo:bar /> the element name is now "bar" without a namespace prefix ever being set. For the element <foo bar:baz="10" /> the attribute name is now "bar:baz" instead of just "baz".

This particular change may break existing applications, hence the version bump to 2.0.

See commit 66fc4b1dfc for more information.

Slightly improved performance of checking XPath booleans

Performance of checking if certain XPath values are booleans has been improved somewhat. See commit 9bb908f8b1 for more information.

1.3.1 - 2015-09-07

Race condition in the XPath compiler

This release fixes a race condition in the XPath compiler. The XPath::Compiler#compile method would compile Procs using its own Binding, this in turn would lead to race conditions when using the compiled Procs concurrently.

See commit bd48dc15cc for more information.

1.3.0 - 2015-09-06

XPath query evaluation rewritten

The system used for evaluating XPath and CSS queries has been rewritten from the ground up, resulting in much better performance. Prior to 1.3.0 Oga would evaluate queries by iterating over the Abstract Syntax Tree (AST) produced by the XPath/CSS parser. This setup could lead to lots of object allocations and method calls, even for small queries.

Starting with 1.3.0 Oga instead generates Ruby code based on XPath expressions. The generated code relies on nesting of conditionals (instead of method calls) and allocates far fewer objects (partially as a result of this). The generated code is cached based on the input expression, removing the need for recompiling the same expression over and over. The result of all this greatly improved querying performance.

As an example, lets look at the benchmark benchmark/xpath/compiler/big_xml_average_bench.rb. When using Oga 1.2.3 the output is as following:

Iteration: 1: 3.292
Iteration: 2: 2.71
Iteration: 3: 2.747
Iteration: 4: 2.752
Iteration: 5: 2.776
Iteration: 6: 2.735
Iteration: 7: 2.761
Iteration: 8: 2.741
Iteration: 9: 2.791
Iteration: 10: 2.787

Iterations: 10
Average:    2.809 sec

Using Oga 1.3.0 we instead get the following output:

Iteration: 1: 0.639
Iteration: 2: 0.422
Iteration: 3: 0.428
Iteration: 4: 0.47
Iteration: 5: 0.443
Iteration: 6: 0.445
Iteration: 7: 0.51
Iteration: 8: 0.485
Iteration: 9: 0.506
Iteration: 10: 0.547

Iterations: 10
Average:    0.489 sec

Here Oga 1.3.0 is about 5.7 times faster compared to version 1.2.3.

In the coming days I'll work on writing a blog post that explains more about the new compiler setup, how it works, how it performs, etc.

In the mean time, see the following issues/pull requests for more information:

Escaping of characters in CSS expressions

CSS expressions now allow querying of nodes having dots in the element name or namespace. This can be done by escaping the dot using a backslash. For example:

Oga.parse_xml('<foo.bar />').css('foo\.bar') # => NodeSet(Element(name: "foo.bar"))

See issue https://gitlab.com/yorickpeterse/oga/issues/124 for more information.

Support for the CSS :not() pseudo class

CSS expressions can now use the :not() pseudo class.

See issue https://gitlab.com/yorickpeterse/oga/issues/125 for more information.

Improved parsing of CSS expressions

CSS expressions such as foo>bar and foo > .bar are now supported, previously these would result in parser errors.

See the following issues for more information:

Unicode support for CSS/XPath

CSS and XPath expressions can now contain Unicode characters, previously only ASCII characters were allowed for identifiers (node tests, attribute names, etc).

See issue https://gitlab.com/yorickpeterse/oga/issues/140 for more information.

1.2.3 - 2015-08-19

NodeSet performance improvements

Performance of the NodeSet class has been improved, especially when used in concurrent environments. See commit 4f94d03a85 for more information.

Comparing names in the XPath evaluator

Performance of comparing names of nodes in the XPath evaluator has been improved thanks to Daniel Fockler.

See the following commits for more information:

1.2.2 - 2015-08-14

Race condition in the LRU class

A race condition in the LRU class has been resolved. This race condition would result in errors such as "ConcurrencyError: Detected invalid array contents due to unsynchronized modifications with concurrent users" on JRuby or "ArgumentError: negative array size" on Rubinius.

See commit 32b75bf62c for more information.

Lexing of void elements with explicit closing tags

Void elements followed by an explicit closing tag (e.g. <param></param>) are now lexed properly. Thanks to Jakub Pawlowicz for fixing this.

See commit ed3cbe7975 for more information.

1.2.1 - 2015-07-01

Better support for decoding unrecognized XML/HTML entities

Jakub Pawlowicz improved the process of decoding XML/HTML entities so that it handles unrecognized entities better. Previously Oga would raise an error when trying to decode entities such as &#TAB; instead of just leaving them as-is.

See issue https://gitlab.com/yorickpeterse/oga/issues/118 and pull request https://gitlab.com/yorickpeterse/oga/pull/122 for more information.

1.2.0 - 2015-06-30

Support for Unicode in XML/HTML identifiers

XML/HTML element and attribute names can now contain Unicode characters. While the HTML specification states only ASCII may be used Oga still supports Unicode identifiers for HTML.

See commit dde644cd79 for more information.

Support for dots in XML/HTML identifiers

XML/HTML element and attribute names can now contain dots such as <foo.bar>baz</foo.bar>. Thanks to Laurence Lee for adding this.

See commit b7771ed5fe for more information.

Support for the CSS :nth() pseudo class

Oga now supports the :nth() CSS pseudo class. This pseudo class can be used to select the Nth element in a set regardless of any preceding/following siblings.

See commit 71960fff87 for more information.

Support for commas in CSS expressions

CSS expressions such as foo, bar and #hello, #world are now supported.

See commit d26b48feb4 for more information.

1.1.0 - 2015-06-29

Better support for unquoted HTML attribute values

Oga can now parse HTML such as <a href=foo("bar","baz")></a> and basically any other kind of value as long as it does not contain a > or whitespace.

See commit 3b633ff41c for more information.

Support for replacing of DOM nodes

The newly added method Oga::XML::Node#replace can be used to replace an existing node with another node or with a String (which will result in it being replaced with a Text node). For example:

p   = Oga::XML::Element.new(:name => 'p')
div = Oga::XML::Element.new(:name => 'div', :children => [p])

puts div.to_xml # => "<div><p /></div>"

p.replace('Hello world!')

puts div.to_xml # => "<div>Hello world!</div>"

Thanks to Tero Tasanen for adding this.

See commit 0b4791b277 and https://gitlab.com/yorickpeterse/oga/pull/116 for more information.

Encoding quotes in attribute values

When serializing elements back to XML Oga now properly encodes single/double quotes in attribute values. See commit 074b53c18c for more information.

1.0.3 - 2015-06-16

Strict XML parsing support

Oga can now parse XML documents in "strict mode". This mode currently only disables the automatic insertion of missing closing tags. This feature can be used as following:

document = Oga.parse_xml('<foo>bar</foo>', :strict => true)

This works for all 3 (DOM, SAX and pull) parsing APIs.

See commit 2c18a51ba9 for more information.

Support for HTML attribute values without starting quotes

Oga can now parse HTML such as <foo bar=baz" />. This will be parsed as if the input were <foo bar='baz"' />.

See commit fd307a0fcc for more information.

Support for spaces around attribute equal signs

Previously XML/HTML such as <foo bar = "baz" /> would not be parsed correctly as Oga didn't support spaces around the = sign. Commit a76286b973 added support for input like this.

Decoding entities with numbers

Oga can now decode entities such as ½. Due to an incorrect regular expression these entities would not be decoded.

See commit af7f2674af for more information.

1.0.2 - 2015-06-03

Fix for requiring extensions on certain platforms

The loading of files has been changed to use require so that native extensions are loaded properly even when a platform decides not to store in in the lib directory.

See commit 4bfeea2590 for more information.

Better closing of HTML tags

Closing of HTML tags has been improved so Oga can parse HTML such as this:

<div>
    <ul>
        <li>foo
    </ul>
    inside div
</div>
outside div

See the following commits for more information:

Whitespace support in closing tags

Oga can now lex HTML/XML such as the following:

<p>hello<p
>

See commit d2523a1082 for more information.

1.0.1 - 2015-05-21

Encoding quotes in XML

Oga no longer encodes single/double quotes as XML entities when serializing a document back to XML. This ensures that input such as <foo>a"b</foo> doesn't get turned into <foo>a"b</foo>.

HTML Entity Encoding

HTML entities are now generated using pack('U*') instead of pack('U') ensuring the correct characters/codepoints are produced.

1.0.0 - 2015-05-20

This marks the first stable release (API wise) for Oga. It's been quite the ride since the very first commit from February 26, 2014. In the releases following 1.0 I plan to focus mainly on performance as both XMl/HTML parsing and XPath evaluation performance is not quite as fast as I'd like it to be.

License Change

Up until 1.0.0 Oga was licensed under the MIT license. Since this license does fairly little to protect authors (especially regarding patents) I've decided to change the license to the Mozilla Public License 2.0. More information on this can be found in commit 0a7242aed4.

XPath Performance Improvements

With 1.0 the evaluator received further performance improvements that should be especially noticable when querying large XML/HTML documents. Improving XPath performance is an ongoing task so expect similar improvements in upcoming releases.

See the following commits for more information:

Full HTML5 Support

With 1.0 Oga finally supports parsing of HTML5 according to the official specification. This means that Oga is now capable of parsing HTML such as the following:

<p>Hello, this is a list:</p>

<ul>
    <li>First item
    </li>Second item
</ul>

This would be parsed as if the HTML were as following instead:

<p>Hello, this is a list:</p>

<ul>
    <li>First item</li>
    </li>Second item</li>
</ul>

See the following commits for more information:

The following issues are also worth checking out:

Handling of invalid XML/HTML

Oga can now handle most forms of invalid XML/HTML by automatically inserting missing closing tags and ignoring stray opening tags where possible. This allows Oga to parse XML such as the following:

<root>
    <person>
        <name>Alice</name>
    </person>

See commit 13e2c3d82f for more information.

Decoding zero padded XML/HTML entities

Oga can now decode zero padded XML/HTML entities such as &. See commit 853d804f34 for more information.

0.3.4 - 2015-04-19

XML and HTML entities are decoded in the SAX parser before data is passed to a custom handler class.

See commit da62fcd75d for more information.

0.3.3 - 2015-04-18

Improved lexer support for script/style tags

Commit 73fbbfbdbd improved support for lexing HTML script and style tags, ensuring that HTML such as the following is processed correctly:

<script>
var foo = "</style>"
</script>

<style>
/* </script> */
</style>

Lexing of extra quotes

The XML lexer can now handle stray quotes that reside in the open tag of an element, for example:

<a href="foo""></a>

While technically invalid HTML certain websites such as http://yahoo.com contain HTML like this.

See commit 6b779d7883 for more information.

Lexing of doctypes containing newlines

The XML lexer is now capable of lexing doctypes that contain newlines such as:

<!DOCTYPE
    html>

See commit 9a0e31d0ae for more information.

0.3.2 - 2015-04-15

Support for unquoted HTML attribute values

Oga can now lex/parse HTML attribute values that don't use quotes. For example, the following is valid HTML:

<a href=foo>Foo</a>

And so is this:

<a href=foo/bar>Foo/bar</a>

See GitLab issue https://gitlab.com/yorickpeterse/oga/issues/94 and the following commits for more information:

Counting newlines in XML declarations

The XML lexer has been adjusted so that it counts newlines when processing XML declarations. While these newlines are not exposed to the resulting Oga::XML::* instances they are used when reporting errors. Previously the lexer wouldn't count newlines in XML declarations, leading to error messages referring to incorrect line numbers.

This was fixed in commit e942086f2d.

Better lexer support for CDATA, comments and processing instructions

The XML lexer has been tweaked so it can handle multi-line CDATA tags, comments and processing instructions, both when using a String and IO (or similar) as input.

See GitLab issue https://gitlab.com/yorickpeterse/oga/issues/93 and the following commits for more information:

Performance Improvements

To improve performance of the XPath evaluator (as well as generic code using Oga) the following methods now cache their return values:

Oga::XML::Element#available_namespaces
Oga::XML::Element#namespace
Oga::XML::Node#html?

These cache of these methods is flushed automatically when needed. For example, registering a new namespace will flush the cache for Element#available_namespaces and Element#namespace.

The performance of Oga::XML::Traversal#each_node has also been optimized, cutting down the amount of object allocations significantly.

Combined these improvements should make XPath evaluation roughly 4 times faster.

See the following commits for more information:

0.3.1 - 2015-04-08

Oga no longer decodes any HTML entities that appear inside the body of a <script> or <style> tag. For example, this HTML:

<script>foo&bar;</script>

Would effectively be turned into:

<script>foo</script>

See commit 4bdc8a3fdc for more information.

0.3.0 - 2015-04-03

Lexing of carriage returns

Oga can now lex and parse XML documents using carriage returns for newlines. This was added in commit 0800654c96.

Improved handling of HTML namespaces

Oga now ignores any declared namespaces when parsing HTML documents as HTML5 does not allow one to register custom namespaces.

See commit 3176459307 for more information.

Improved handling of explicitly declared default XML namespaces

In the past explicitly defining the default XML namespace in a document would lead to Oga's XPath evaluator not being able to match any nodes. This has been fixed in commit 5adeae18d0.

Caching of XPath/CSS expressions

The CSS and XPath parsers now cache the ASTs of an expression used when querying a document using CSS or XPath. This can give a pretty noticable speed improvement, especially when running the same expression in a loop (or just many different times).

Parsed expressions are stored in an LRU to prevent memory from growing forever. Currently the capacity is set to 1024 values but this can be changed as following:

Oga::XPath::Parser::CACHE.maximum = 2048
Oga::CSS::Parser::CACHE.maximum   = 2048

The LRU synchronizes method calls to allow safe usage from multiple threads.

See the following commits for more info:

Windows support

While Oga for the most part already supported Windows a few changes for the extension compilation process were required to allow users to install Oga on Windows. Tests are run on AppVeyor (a continuous integration service for Windows platforms).

Oga requires devkit (http://rubyinstaller.org/add-ons/devkit/) to be installed on non Cygwin/MinGW environments. Cygwin/MinGW environments probably already work, although I do not run any tests on these environments.

SAX parsing of XML attributes

Parsing of XML attributes using the SAX API was overhauled quite a bit. As these changes are not backwards compatible it's likely that existing SAX parsers will break.

See commit d8b9725b82 for more information.

Parser callbacks for XML attributes

The XML parser has an extra callback method called on_attribute which is used to create a new attribute. This callback can be used in custom SAX parsers just like the other callbacks.

Parser rewritten using ruby-ll

The XML, CSS and XPath parsers have been re-written using ruby-ll (https://gitlab.com/yorickpeterse/ruby-ll). While Racc served its purpose (until now) it has three main problems:

Performance is not as good as it should be.
The codebase is dated and generally hard to deal with, as such it's quite difficult to optimize in reasonable time.
LALR parser errors can be incredibly painful to debug.

For this reason I wrote ruby-ll and replaced Oga's Racc based parsers with ruby-ll parsers. These parsers are LL(1) parsers which makes them a lot easier to debug. Performance is currently a tiny bit faster than the old Racc parsers, but this will be improved in the coming releases of both Oga and ruby-ll.

See pull request https://gitlab.com/yorickpeterse/oga/pull/78 for more information.

Lazy decoding of XML/HTML entities

In the past XML/HTML entities were decoded in the lexer, adding overhead even when not needed. This has been changed so that the decoding of entities only occurs when calling XML::Text#text. With this particular change also comes support for HTML entities and codepoint based XML/HTML entities.

See commit 2ec91f130f for more information.

0.2.3 - 2015-03-04

This release adds support for lexing HTML <style> tags similar to how <script> tags are handled. This ensures that the contents of these tags are treated as-is without any HTML entity conversion being applied.

See commits 78e40b55c0 and 3b2055a30b for more information.

0.2.2 - 2015-03-03

This release fixes a bug where setting the text of an element using Oga::XML::Element#inner_text= would not set the parent element of the newly created text node. This would result in the following:

some_element.inner_text = 'foo'

some_element.children[0].parent # => nil

Here parent is supposed to return some_element instead. See commit 142b467277 for more information.

0.2.1 - 2015-03-02

Proper HTML serializing support for script tags

When serializing an HTML document back to HTML (as a String) the contents of <script> tags are serialized correctly. Previously XML unsafe characters (e.g. <) would be converted to XML entities, which results in invalid Javascript syntax. This has been changed so that <script> tags in HTML documents don't have their contents converted, ensuring proper Javascript syntax upon output.

See commit 874d7124af and issue https://gitlab.com/yorickpeterse/oga/issues/79 for more information.

Proper lexing support for script tags

When lexing HTML documents the XML lexer is now capable of lexing the contents of <script> tags properly. Previously input such as <script>x >y</script> would result in incorrect tokens being emitted. See commit ba2177e2cf and issue https://gitlab.com/yorickpeterse/oga/issues/70 for more information.

Element Inner Text

When setting the inner text of an element using Oga::XML::Element#inner_text= all child nodes of the element are now removed first, instead of only text nodes being removed.

See https://gitlab.com/yorickpeterse/oga/issues/64 for more information.

Support for extra XML entities

Support for encoding/decoding extra XML entities was added by Dmitry Krasnoukhov. This includes entities such as &#60, &#34, etc. See commit 26baf89440 for more information.

Support for inline doctypes with newlines in IO input

The XML lexer (and thus the parser) can now handle inline doctypes containing newlines when using an IO object as the input. For example:

<!DOCTYPE html[foo
bar]>

Previously this would result in incorrect tokens being emitted by the lexer. See commit cbb2815146 for more information.

0.2.0 - 2014-11-17

CSS Selector Support

Probably the biggest feature of this release: support for querying documents using CSS selectors. Oga supports a subset of the CSS3 selector specification, in particular the following selectors are supported:

Element, class and ID selectors
Attribute selectors (e.g. foo[x ~= "y"])

The following pseudo classes are supported:

:root
:nth-child(n)
:nth-last-child(n)
:nth-of-type(n)
:nth-last-of-type(n)
:first-child
:last-child
:first-of-type
:last-of-type
:only-child
:only-of-type
:empty

You can use CSS selectors using the methods css and at_css on an instance of Oga::XML::Document or Oga::XML::Element. For example:

document = Oga.parse_xml('<people><person>Alice</person></people>')

document.css('people person') # => NodeSet(Element(name: "person" ...))

The architecture behind this is quite similar to parsing XPath. There's a lexer (Oga::CSS::Lexer) and a parser (Oga::CSS::Parser). Unlike Nokogiri (and perhaps other libraries) the parser does not output XPath expressions as a String or a CSS specific AST. Instead it directly emits an XPath AST. This allows the resulting AST to be directly evaluated by Oga::XPath::Evaluator.

See https://gitlab.com/yorickpeterse/oga/issues/11 for more information.

Mutli-line Attribute Support

Oga can now lex/parse elements that have attributes with newlines in them. Previously this would trigger memory allocation errors.

See https://gitlab.com/yorickpeterse/oga/issues/58 for more information.

SAX after_element

The after_element method in the SAX parsing API now always takes two arguments: the namespace name and element name. Previously this method would always receive a single nil value as its argument, which is rather pointless.

See https://gitlab.com/yorickpeterse/oga/issues/54 for more information.

XPath Grouping

XPath expressions can now be grouped together using parenthesis. This allows one to specify a custom operator precedence.

Enumerator Parsing Input

Enumerator instances can now be used as input for Oga.parse_xml and friends. This can be used to download and parse XML files on the fly. For example:

enum = Enumerator.new do |yielder|
  HTTPClient.get('http://some-website.com/some-big-file.xml') do |chunk|
    yielder << chunk
  end
end

document = Oga.parse_xml(enum)

See https://gitlab.com/yorickpeterse/oga/issues/48 for more information.

Removing Attributes

Element attributes can now be removed using Oga::XML::Element#unset:

element = Oga::XML::Element.new(:name => 'foo')

element.set('class', 'foo')
element.unset('class')

XPath Attributes

XPath predicates are now evaluated for every context node opposed to being evaluated once for the entire context. This ensures that expressions such as descendant-or-self::node()/foo[1] are evaluated correctly.

Available Namespaces

When calling Oga::XML::Element#available_namespaces the Hash returned by Oga::XML::Element#namespaces would be modified in place. This was a bug that has been fixed in this release.

NodeSets

NodeSet instances can now be compared with each other using ==. Previously this would always consider two instances to be different from each other due to the usage of the default Object#== method.

XML Entities

XML entities such as & and < are now encoded/decoded by the lexer, string and text nodes.

See https://gitlab.com/yorickpeterse/oga/issues/49 for more information.

General

Source lines are no longer included in error messages generated by the XML parser. This simplifies the code and removes the need of re-reading the input (in case of IO/Enumerable inputs).

XML Lexer Newlines

Newlines in the XML lexer are now counted in native code (C/Java). On MRI and JRuby the improvement is quite small, but on Rubinius it's a massive improvement. See commit 8db77c0a09bf6c996dd2856a6dbe1ad076b1d30a for more information.

HTML Void Element Performance

Performance for detecting HTML void elements (e.g. <br> and <link>) has been improved by removing String allocations that were not needed.

0.1.3 - 2014-09-24

This release fixes a problem with serializing attributes using the namespace prefix "xmlns". See https://gitlab.com/yorickpeterse/oga/issues/47 for more information.

0.1.2 - 2014-09-23

SAX API

A SAX parser/API has been added. This API is useful when even the overhead of the pull-parser is too much memory wise. Example:

class ElementNames
  attr_reader :names

  def initialize
    @names = []
  end

  def on_element(namespace, name, attrs = {})
    @names << name
  end
end

handler = ElementNames.new

Oga.sax_parse_xml(handler, '<foo><bar></bar></foo>')

handler.names # => ["foo", "bar"]

Racc Gem

Oga will now always use the Racc gem instead of the version shipped with the Ruby standard library.

Error Reporting

XML parser errors have been made a little bit more user friendly, though they can still be quite cryptic.

Serializing Elements

Elements serialized to XML/HTML will use self-closing tags whenever possible. When parsing HTML documents only HTML void elements will use self-closing tags (e.g. <link> tags). Example:

Oga.parse_xml('<foo></foo>').to_xml        # => "<foo />"
Oga.parse_html('<script></script>').to_xml # => "<script></script>"

Default Namespaces

Namespaces are no longer removed from the attributes list when an element is created.

Default XML namespaces can now be registered using xmlns="...". Previously this would be ignored. Example:

document = Oga.parse_xml('<root xmlns="baz"></root>')
root     = document.children[0]

root.namespace # => Namespace(name: "xmlns" uri: "baz")

Lexing Incomplete Input

Oga can now lex input such as </ without entering an infinite loop. Example:

Oga.parse_xml('</') # => Document(children: NodeSet(Text("</")))

Absolute XPath Paths

Oga can now parse and evaluate the XPath expression "/" (that is, just "/"). This will return the root node (usually a Document instance). Example:

document = Oga.parse_xml('<root></root>')

document.xpath('/') # => NodeSet(Document(children: NodeSet(Element(name: "root"))))

Namespace Ordering

Namespaces available to an element are now returned in the correct order. Previously outer namespaces would take precedence over inner namespaces, instead of it being the other way around. Example:

document = Oga.parse_xml <<-EOF
<root xmlns:foo="bar">
  <container xmlns:foo="baz">
    <foo:text>Text!</foo:text>
  </container>
</root>
EOF

foo = document.at_xpath('root/container/foo:text')

foo.namespace # => Namespace(name: "foo" uri: "baz")

Parsing Capitalized HTML Void Elements

Oga is now capable of parsing capitalized HTML void elements (e.g. <BR>). Previously it could only parse lower-cased void elements. Thanks to Tero Tasanen for fixing this. Example:

Oga.parse_html('<BR>') # => Document(children: NodeSet(Element(name: "BR")))

Node Type Method Removed

The node_type method has been removed and its purpose has been moved into the XML::PullParser class itself. This method was solely used by the pull parser to provide shorthands for node classes. As such it doesn't make sense to expose this as a method to the outside world as a public method.

0.1.1 - 2014-09-13

This release fixes a problem where element attributes were not separated by spaces. Thanks to Jonathan Rochkind for reporting it and Bill Dueber providing an initial patch for this problem.

0.1.0 - 2014-09-12

The first public release of Oga. This release contains support for parsing XML, basic support for parsing HTML, support for querying documents using XPath and more.

39 KiB Raw Permalink Blame History

Changelog

3.4 - 2022-08-02

3.3 - 2020-07-27

3.2 - 2020-01-10

3.1 - 2020-01-08

3.0 - 2019-12-03

2.17 - 2019-12-02

2.16 - 2019-11-29

2.15 - 2018-04-11

2.14 - 2018-01-30

2.13 - 2018-01-05

2.12 - 2017-12-29

2.11 - 2017-09-07

2.10 - 2017-04-18

Fix Element#attribute for HTML documents when using Symbol arguments

2.9 - 2017-02-10

Closing tags for HTML void elements

Doctypes are now Nodes

2.8 - 2017-01-04

2.7 - 2016-09-27

Closing Elements When Generating XML

Newlines After Doctypes

Processing Instructions With Namespace Prefixes

XML Declarations Are Now Processing Instructions

Aliases For Getting & Setting Attributes

2.6 - 2016-09-10

2.5 - 2016-09-06

2.4 - 2016-09-04

Serialising Large Documents

Faster retrieval of previous/next nodes

2.3 - 2016-07-13

2.2 - 2016-02-23

XPath support for nested pipe operators

2.1 - 2016-02-09

Preserving entities that can't be decoded

New Versioning Format

2.0.0 - 2015-12-26

Fixed parsing HTML identifiers

Slightly improved performance of checking XPath booleans

1.3.1 - 2015-09-07

Race condition in the XPath compiler

1.3.0 - 2015-09-06

XPath query evaluation rewritten

Escaping of characters in CSS expressions

Support for the CSS :not() pseudo class

Improved parsing of CSS expressions

Unicode support for CSS/XPath

1.2.3 - 2015-08-19

NodeSet performance improvements

Comparing names in the XPath evaluator

1.2.2 - 2015-08-14

Race condition in the LRU class

Lexing of void elements with explicit closing tags

1.2.1 - 2015-07-01

Better support for decoding unrecognized XML/HTML entities

1.2.0 - 2015-06-30

Support for Unicode in XML/HTML identifiers

Support for dots in XML/HTML identifiers

Support for the CSS :nth() pseudo class

Support for commas in CSS expressions

1.1.0 - 2015-06-29

Better support for unquoted HTML attribute values

Support for replacing of DOM nodes

Encoding quotes in attribute values

1.0.3 - 2015-06-16

Strict XML parsing support

Support for HTML attribute values without starting quotes

Support for spaces around attribute equal signs

Decoding entities with numbers

1.0.2 - 2015-06-03

Fix for requiring extensions on certain platforms

Better closing of HTML tags

Whitespace support in closing tags

1.0.1 - 2015-05-21

Encoding quotes in XML

HTML Entity Encoding

1.0.0 - 2015-05-20

License Change

XPath Performance Improvements

39 KiB

Raw Permalink Blame History

Fix `Element#attribute` for HTML documents when using Symbol arguments