* Tokens are now immutable. The progress of transformations is tracked on
chunks instead of tokens. Tokenizer output is cached and can be directly
returned without a need for cloning. Transforms are required to clone or
newly create tokens they are modifying.
* Expansions per chunk are now shared between equivalent frames via a cache
stored on the chunk itself. Equivalence of frames is not yet ideal though,
as right now a hash tree of *unexpanded* arguments is used. This should be
switched to a hash of the fully expanded local parameters instead.
* There is now a vastly improved maybeSyncReturn wrapper for async transforms
that either forwards processing to the iterative transformTokens if the
current transform is still ongoing, or manages a recursive transformation if
needed.
* Parameters for parser functions are now wrapped in abstract Params and
ParserValue objects, which support some handy on-demand *value* expansions.
Keys are always expanded. Parser functions are converted to use these
interfaces, and now properly expand their values in the correct frame.
Making this expansion lazier is certainly possible, but would complicate
transformTokens and other token-handling machinery. Need to investigate if
it would really be worth it. Dead branch elimination is certainly a bigger
win overall.
* Complex recursive asynchronous expansions should now be closer to correct
for both the iterative (transformTokens) and recursive (maybeSyncReturn
after transformTokens has returned) code paths.
* Performance degraded slightly. There are no micro-optimizations done yet
and the shared expansion cache still has a low hit rate. The progress
tracking on chunks is not yet perfect, so there are likely a lot of unneeded
re-expansions that can be easily eliminated. There is also more debug
tracing right now. Obama currently expands in 54 seconds on my laptop.
Change-Id: I4a603f3d3c70ca657ebda9fbb8570269f943d6b6
* [[:en:Barack Obama]] can now be expanded in 77 seconds using 330MB RAM,
while it would prevously run out of RAM after ~30 minutes. Wohoooo!
The token transform framework rework really paid off.
* 303 parser tests are passing in the new record time of 5.5 seconds. Two more
tests are passing since these tests expect the day of the week to be
Thursday. Won't be the case tomorrow.
Change-Id: I56e850838476b546df10c6a239c8c9e29a1a3136
* All parser pipelines including tokenizer and DOM stuff are now constructed
from a 'recipe' data structure in a ParserPipelineFactory.
* All sub-pipelines of these can now be cached
* Event registrations to a pipeline are directly forwarded to the last
pipeline member to save relatively expensive event forwarding.
* Some APIs for on-demand expansion / format conversion of parameters from
parser functions are added:
param.to('tokens/expanded', cb)
param.to('text/wiki', cb) (this does not work yet)
All parameters are additionally wrapped into a Param object that provides
method for positional parameter naming (.named() or conversion to a dict
(.dict()).
* The async token transform manager is now separated from a frame object, with
the frame holding arguments, an on-demand expansion method and loop checks.
* Only keys of template parameters are now expanded. Parser functions or
template arguments trigger an expansion on-demand. This (unsurprisingly)
makes a big performance difference with typical switch-heavy template
systems.
* Return values from async transforms are no longer used in favor of plain
callbacks. This saves the complication of having to maintain two code paths.
A trick in transformTokens still avoids the construction of unneeded
TokenAccumulators.
* The results of template expansions are no longer buffered.
* 301 parser tests are passing
Known issues:
* Cosmetic cleanup remains to do
* Some parser functions do not support async expansions yet, and need to be
modified.
Change-Id: I1a7690baffbe8141cadf67270904a1b2e1df879a
This is a bit better than cloning tokens wholesale, but not by much. There is
a lot of potential for much better per-token caching with reduced token
cloning. Need to map out all dependencies besides token attributes expanded
from template parameters or other scoped state. Even if tokens themselves
don't need transformation, they might still need to be considered for other
token transformers, so simply keeping the final rank won't quite work even if
the token itself is fully transformed. As a minimum, a shallow clone would
need to be made and the rank reset (as in env.cloneTokens).
Change-Id: I4329113bb21750bae9a635229ed1b08da75dc614
* Added an LRU cache (using the lru-cache node module) for tokenizer output
* Mutation of nested attributes now replaces the containers. A shallow copy of
tokens is sufficient to isolate token transformations. Need to investigate
if we can actually get away without isolation and re-transformation for most
ordinary tokens.
Change-Id: I9136b1d7a1fbcc538183a319d4ecaa290d616fdf
* Ignore safesubst for now
* Remove an unneeded whitelist entry
* Make sure the caption is not lost for thumbs (fix to last commit) and remove
debug print
Change-Id: I243584ed0838cf7c3b4110fe9cdf869272477312
* less verbose logging in noinclude processing and template expansion
* Give priority to the processing of templates transcluded from transclusions
to get closer to depth-first processing. This serves to minimize memory
usage from queued-up tokens.
* Increase the maximum outstanding requests per template retrieval. 10000
amazingly proved too low a limit on some big pages.
* Only process a single template request callback at a time for now
* Add a debug print in the treebuilder wrapper
* Don't treat multiple comments on a single line as a single comment to match
the PHP parser's behavior
Change-Id: I9a86b6d7bec3b9e1f17415daf1bf74170240721a
In experiments this dropped the memory consumption further, and reduces the
queuing overhead in the node reactor.
Change-Id: I9409b6ca863b43b7557663bbec9572365059c078
Only call back a few callbacks per reactor iteration from the template fetch
request queue. This changes the expansion pattern from a (memory intensive)
breadth-first expansion to something quite close to depth-first expansion.
Additionally, retrieved pages are quickly added to the page cache so that a
lot of request queuing is avoided in favor of synchronous expansion from the
cache. On pages like Barack Obama that previously ran out of memory after
consuming node's 1.6G heap limit, expansion now runs in relatively constant
100-300M resident (so far, still running).
Change-Id: Ie34a1eeff00d868416de45ef8d289898258f560c
* add past paths for empty arguments etc
* cache attribute token transform pipelines
* fix bugs in TokenCollector and NoIncludeOnly handler, and improve its
efficiency by only registering for 'end' tokens on demand
* Remove empty reset methods from a few handlers
* Add a simple 'ap' debug print function that makes it easy to only print some
debug prints by temporarily changing 'dp' to 'ap'
* Improvements and bug fixes in AttributeExpander
Change-Id: Ie69729c8f62d48bba922712e44ebce484c621c50
* DOM based on Wikia's thumb output: HTML5, clean caption without magnify
icon.
* basic RDFa annotations, but most options additionally in data-mw object-
might want to move more (or all?) of those into RDFa data using meta tags.
* no support yet for framed or other formats, image scaling etc
* also tweaked some config options in the environment
Change-Id: Ie461fcdce060cfc2dec65cc057709ae650ef3368
only expand used branches selected by parser functions. Template (and
-argument) expansion is simply registered before general expansion.
Additionally, a few more simple time-based magic words are added in
ParserFunctions.
values. This includes comments, templates and template arguments.
This also replaces the specialized expansion logic in the TemplateHandler. The
removal of link validation lets one more parser test fail for now. External
link target validation will need to be implemented in the token stream handler
for links. This is noted as TODO in
https://www.mediawiki.org/wiki/Future/Parser_development#Token_stream_transforms.
functionality (comments, templates, template arguments) in arbitrary
attributes. The grammar for this is still quite rough, will need to
consolidate that area.
other tokens. This is only the first half of the conversion. The next step is
to drop the type attribute on most tokens and match on the constructor in the
token transform machinery.
improvements to parser functions on the way to support the cite extensions.
Preparation for generic template and template arg in attribute support. 222
parser tests now passing.
page like this:
cd extensions/VisualEditor/modules/parser
echo '{{:Main Page}}' | node parse.js
echo '{{:Main Page}}' | node parse.js --html
echo '{{:Main Page}}' | node parse.js --debug
Even the date-based includes work somewhat, although they don't yet accept
passed-in dates.
directly to WikiDom from enwiki using a commandline like this:
echo '{{User:GWicke/Test}}' | node parse.js
Wohoo!
Complex pages with templates won't render properly yet, as noinclude /
includeonly and parser functions are not yet implemented. As a result, the
parser will run out of memory or hit the currently low expansion depth limit
as it tries to expand documentation for all templates.
disable it by default in parserTests as it tries to fetch all sorts of parser
functions and is not yet fully supported in parserTests. The next step will be
to build a list of parser functions (to avoid fetching them as templates) and
pushing the event interface into parserTests.
