Close#43.
The method allows to fail with arbitrary collection of
messages. ‘unexpected’ is not defined in terms of ‘failure’. One
consequence of this design decision is that ‘failure’ is now method of
‘MonadParsec’, while ‘unexpected’ is not.
Close#47, close#57.
This commit introduces ‘runParser'’ and ‘runParserT'’ functions that
take and return parser state. This makes it possible to partially parse
input, resume parsing, specify non-standard initial textual position,
etc.
Internal changes involve some refactoring to make ‘Reply’ more
readable and facilitate extraction of complete parser state on failure
as well as success.
The commit adds basic tests for the new functionality as well.
Close#65.
Previously we had 5 nearly identical definitions of the function,
varying only in type-specific ‘readFile’ function. Now the problem is
solved by introduction of ‘StorableStream’ type class. All supported
stream types are instances of the class out of box and thus we have
polymorphic version of ‘parseFromFile’.
Close#62.
Apart from some refactoring, the following important changes were
introduced:
* ‘ParseError’ is now a monoid.
* Added functions ‘addErrorMessages’ and ‘newErrorMessages’.
Now it's impossible to create ‘SourcePos’ with non-positive line number
or column number. Unfortunately we cannot use ‘Numeric.Natural’ because
we need to support older versions of ‘base’.
Indented text returned by ‘showMessages’ may be undesirable, but we
cannot add indentation outside of the function (edge case: strings
including newline are displayed in the messages).
Closes#56.
In particular, file name and textual position are represented like this:
filename.hs:5:6:
error message
This format should be more conventional, so various tools will be able
to parse it and provide some support (for example, Emacs can work with
this format).
‘Text.Megaparsec’ and ‘Text.Megaparsec.Prim’ do not export these data
types and their constructors anymore. These data types are rather
low-level implementation detail that should not be visible to
end-user. They are also subject to certain changes in future.
This patch introduces compatibility to base-4.7.0.x. It was tested
on Win 8.1 x86_64, using GHC 7.8.4. It mainly consists of a bunch
of #if !MIN_VERSION(4,8,0) ... #endif additions and a lower bound
on base in the cabal file as well as a general introduction of the
CPP extension via default-extensions.
It also removes a potential error source in tests/Util.hs, since
the backslash in /=\ can lead to strange quirks on certain systems
(backslash and newline only separated by whitespace).
Other, squashed commits:
- Remove 'recent version of base' from Readme
- Change necessary version of GHC
‘Text.Megaparsec.Prim’ cannot be considered portable since it uses
multi-parameter type classes and functional dependencies.
Other modules that depend on these non-portable features from
‘Text.Megaparsec.Prim’ should be considered non-portable too.
Closes#37.
Most part of these changes is proposed by @neongreen. To apply precisely
what I deem acceptable, correct some of them in other way, and add some
other things, I've manually re-edited this.
Closes#38.
Now tab width can be manipulated with via the following functions:
* ‘getTabWidth’
* ‘setTabWidth’
Other auxiliary changes were performed, such as updating of
‘updatePosChar’.
This also corrects a bit obsolete descriptions of some functions.
Closes#36.
We should try to preserve original information where possible. User then
can convert case of parsed string if necessary. Previous implementation
discarded actually parsed string and returned argument of the
function — this can be considered as data loss of a sort.
Closes#35.
Since ‘many’ (and thus ‘some’) are the only combinator that can succeed
consuming input and produce hints at the same time we can conclude that
‘cok'’ continuation in ‘pLabel’ combinator is only called when ‘many’ is
labelled. By correcting label in this case prepending the phrase “rest
of ” to actual label we can greatly improve result error message.
Close # 27.
Backtracking user state can be achieved via combination of ‘StateT’
monad transformer and ‘ParsecT’:
StateT StateType (ParsecT s m a)
This user state can be more flexible. This fact renders current built-in
user state redundant.
To help work with this new approach (combining monad transformers more
freely) we introduce ‘MonadParsec’ MTL-style type class. All tools that
come with Megaparsec library were modified to work smoothly with any
instance of ‘MonadParsec’, not only ‘ParsecT’.
Now all the combinators in ‘Text.Megaparsec.Combinator’ are defined for
any instance of ‘Control.Alternative’ (sometimes ‘Control.Applicative’).
Some combinators are inlined.
Closes#29.
Now testing function can return ‘Either [Message] a’ so it can construct
full list of error messages. This may be useful in some cases when
tokens are more complex than simple characters.
Multi-character operators should use ‘try’ in order to be reported
correctly (as “operator”). I've mentioned it in doc-string of
‘makeExprParser’.
It's tempting to include ‘try’ directly in expression parsing code, but
following general spirit of Parsec toward ‘try’, I think current
solution is the best.
