- Regular expressions on empty strings weren't properly returning a failure.
- Updated tests to use Clojure 1.6.0's final release.
- Added
:ci-string trueflag toinsta/parser.
- Calling
emptyon a FlattenOnDemandVector now returns [].
- :auto-whitespace can now take the keyword :standard or :comma to access one of the predefined whitespace parsers.
- Fixed newline problem visualizing parse trees on Linux.
- Fixed problem with visualizing rootless trees.
- Further refinements to the way ordered choice interacts with epsilon parsers.
- Fixed bug introduced by 1.2.9 affecting ordered choice.
- Fixed bug where ordered choice was ignoring epsilon parser.
- Fixed bug introduced by 1.2.7, affecting printing of grammars with regexes.
- Parser printing format now includes <> hidden information and tags.
- Fixed bug when regular expression contains | character.
- Changed pre-condition assertion for auto-whitespace option which was causing a problem with "lein jar".
- Improved handling of unusual characters in ABNF grammars.
- When parsing in :total mode with :enlive as the output format, changed the content of failure node from vector to list to match the rest of the enlive output.
- Fixed problem when epsilon was the only thing in a nonterminal, e.g., "S = epsilon"
- Added experimental
:auto-whitespacefeature. See the Experimental Features Document for more details.
- Fixed reflection warning.
- I had accidentally left a dependency on tools.trace in the repeat.clj file, used while I was debugging that namespace. Removed it.
spanfunction returns substring indexes into the parsed text for a portion of the parse tree.visualizefunction draws the parse tree, using rhizome and graphviz if installed.:optimize :memoryflag that, for suitable parsers, will perform the parsing in discrete chunks, using less memory.- New parsing flag to undo the effect of the <> hide notation.
(my-parser text :unhide :tags)- reveals tags, i.e.,<>applied on the left-hand sides of rules.(my-parser text :unhide :content)- reveals content hidden on the right-hand side of rules with<>(my-parser text :unhide :all)- reveals both tags and content.
- Dramatic performance improvement (quadratic time reduced to linear) when repetition parsers (+ or *) operate on text whose parse tree contains a large number of repetitions.
- Performance improvement for regular expressions.
- Added more support to IncrementalVector for a wider variety of vector operations, including subvec, nth, and vec.
- When you run a parser in "total" mode, the failure node is no longer tagged with
:failure, but instead is tagged with:instaparse/failure.
- Comments now supported in CFGs. Use (* and *) notation.
- Added
ebnfcombinator to theinstaparse/combinatorsnamespace. This new combinator converts string specifications to the combinator-built equivalent. See combinator section of the updated tutorial for details. - ABNF: can now create a parser from a specification using
:input-format :abnffor ABNF parser syntax.- New combinators related to ABNF:
abnf-- converts ABNF string fragments to combinators.string-ci-- case-insensitive strings.rep-- between m and n repetitions.
- New core function related to ABNF:
set-default-input-format!-- initially defaults to :ebnf
- New combinators related to ABNF:
- Added comments to regexes used by the parser that processes the context-free grammar syntax, improving the readability of error messages if you have a faulty grammar specification.
- Backslashes in front of quotation mark were escaping the quotation mark, even if the backslash itself was escaped.
- Unescaped double-quote marks weren't properly handled, e.g., (parser "A = '"'").
- Nullable Plus: ((parser "S = ('a'?)+") "") previously returned a failure, now returns [:S]
- Fixed problem with failure reporting that would occur if parse failed on an input that ended with a newline character.