serialized into a single data-mw-rt attribute if present. Update parserTests
to ignore this attribute for comparisons with expected parser output.
A few more tweaks and notes are thrown into this commit too. 233 tests are
passing now.
are now merged with specific registrations by rank. Not yet clear if that is a
good idea overall, need to check use cases when implementing template expansion
and other functionality.
183 parser test now passing.
The TokenTransformDispatcher now actually implements an asynchronous, phased
token transformation framework as described in
https://www.mediawiki.org/wiki/Future/Parser_development/Token_stream_transformations.
Additionally, the parser pipeline is now mostly held together using events.
The tokenizer still emits a lame single events with all tokens, as block-level
emission failed with scoping issues specific to the PEGJS parser generator.
All stages clean up when receiving the end tokens, so that the full pipeline
can be used for repeated parsing.
The QuoteTransformer is not yet 100% fixed to work with the new interface, and
the Cite extension is disabled for now pending adaptation. Bold-italic related
tests are failing currently.
tests now passing.
Link trails depend on language-dependent positive character classes in the PHP
parser. These classes all seem to disallow punctuation implicitly and list
differing plain text characters instead, so it might be possible to get away
with identifying a common class of non-trail punctuation instead. This would
help to keep the tokenizer independent of configurations, which is very
desirable for caching and simplified external parsing.
start / row / table end). The old productions are not deleted yet to make it
easy to compare the output on more complex articles. 181 tests passing after
adding two table tests with whitespace-only differences to the whitelist.
This required a few further additions to the TokenTransformDispatcher. In
particular, there is now an 'any' token match whose callbacks are executed
before more specific callbacks. This is used by the Cite extension to eat all
tokens between ref and /ref tags. This need is very common, so should be
broken out to an intermediate layer in the future.
In general, the requirements for the TokenTransformDispatcher API are now
clearer, and the API should likely be cleaned up / simplified.
token stream. This is the first token transformation exercising the
TokenTransformer class as its dispatcher. Template expansions, wiki link
formatting, tag sanitation and extensions should be able to use the same
dispatcher by registering for specific token types.
The parser performance is very slightly improved as the token stream is only
traversed once.
Handle arguments and options properly by using the 'optimist' node module.
Please note wordwrapping in usage does not seem to work on my setup :(
Only --help implemented yet.
Example:
$ node parserTests.js --help
Starting up JS parser tests
Usage: node ./parserTests.js
Options:
--filter, --regex Only run tests whose descriptions which match given regex (option not implemented)
--help, -h Show this help message
--disabled Run disabled tests (default false) (option not implemented) [boolean]
html markup handling.
* Remove global 'use strict' declarations from html5 parser.
* Add trailing whitespace handling in dt
Overall, 55 parser tests are now passing.
HTML is parsed using a HTML parser and re-serialized, and the output compared
to the serialization of the new parser's dom. Newline normalization is a
cheap hack for now, need to improve that later.
Builds a DOM tree (jsdom) from the tokens and then serializes that using
document.innerHTML. This is all very experimental, so don't be surprised by
rough edges.
tokens, which for now is still completely built before parsing can proceed.
For each top-level block, the source start/end positions are added as
attributes to the top-most tokens. No tracking of wiki vs. html syntax yet.
tests/parser/parserTests.js.
* Removed var from es in es.js to allow node.js to access it as global. Only
alternative solution appears to be a node-specific 'exports' construct:
http://nodejs.org/docs/v0.3.1/api/modules.html
* Added es.Document.js and es.Document.Serializer.js in es/bases. Not sure if
this is the desired location.
* Changed es.extend to es.extendClass in the serializers
* Modified the first parser test to include the WikiDom modules and call the
new HTML serializer
