crustimoney.data-grammar — nl.functionalbytes/crustimoney 0.0.0-165-0x9388

Create a parser based on a data grammar. The data is translated into combinators.

create-parser clj

(create-parser data)

(create-parser data other-parsers)

Create a parser based on a data grammar definition. If a map with rules is supplied, a map of parsers is returned. Otherwise a single parser is returned. The following example shows what a data grammar looks like:

{;; terminals
 literal            "foo"
 character          \c
 regex              #"[a-z]"
 regex-tag          #crusti/regex "[a-z]" ; EDN support
 eof                $

 ;; refs, chains, choices and grouping
 reference          literal
 chain              (literal regex)
 choices            (literal / regex / "alice" "bob")
 named-group        (:my-name literal / "the end" $)
 auto-named-group=  (literal / "the end" $)

 ;; quantifiers
 zero-to-many       (literal *)
 one-to-many        ("bar"+)
 zero-to-one        ("foo" "bar"?) ; bar is optional here

 ;; lookaheads
 lookahead          (& regex)
 negative-lookahead (!"alice")

 ;; cuts
 soft-cut           ('[' > expr? ']') ; note the >
 hard-cut           ((class-open class class-close >>)*) ; note the >>

 ;; direct combinator calls
 combinator-call    [:with-error :fail #crusti/parser ("fooba" #"r|z")]
 custom-combinator  [:my.app/my-combinator ...]}

To capture nodes in the parse result, you need to use named groups. If you postfix a rule name with =, the expression is automatically captured using the rule's name (without the postfix). Please read up on this at crustimoney.combinators/grammar.

Optionally an existing map of parsers can be supplied, which can refered to by the data grammar. For example:

(create-parser '{root ("Hello " email)}
               {:email (regex "...")})

If you want to use an EDN grammar file or string, you can use #crusti/regex tagged literal for regular expressions. To read this, use the following:

(clojure.edn/read-string {:readers *data-readers*} ...)

DataGrammar clj protocol

vector-tree clj

(vector-tree data)

Low-level protocol function which translates the data type into an intermediary vector-based representation. See crustimoney.vector-grammar for more on this format. This can be useful for debugging, or adding your own data type.

In the latter case, add your type like so:

(extend-type java.util.Date
  DataGrammar
  (vector-tree [date]
    [:my-namespace/my-flexible-date-parser date]))

To see which data types are already supported, use (-> DataGrammar :impls keys)