mbutterick
/
pollen
3
3
Fork 0
Code Issues 1 Pull Requests Projects Releases Wiki Activity

streamline decode

Browse Source
pull/9/head
Matthew Butterick 11 years ago
parent 95478ba824
commit 7c138d2915
3 changed files with 71 additions and 94 deletions
Show Stats Download Patch File Download Diff File

89
decode.rkt
Unescape Escape View File

@ -1,30 +1,61 @@
#lang racket/base #lang racket/base
(require racket/contract xml txexpr) (require xml txexpr sugar/define)
(require "decode/fast.rkt" "predicates.rkt" "decode/typography-fast.rkt") (require "predicates.rkt" "decode/typography.rkt")
(provide to-string (contract-out [register-block-tag (symbol? . -> . void?)] (provide (all-from-out "decode/typography.rkt"))
[decode ((xexpr/c) ;; use xexpr/c for contract on nx because it gives better error messages
;; todo: how to write more specific contracts for these procedures? (define+provide (to-string x)
;; e.g., string-proc should be restricted to procs that accept a string as input (if (string? x)
;; and return a string as output x ; fast exit for strings
(#:exclude-xexpr-tags list? (with-handlers ([exn:fail? (λ(exn) (error (format "Pollen parser: can't convert ~v to ~a" x 'string)))])
#:xexpr-tag-proc procedure? (cond
#:xexpr-attrs-proc procedure? [(equal? '() x) ""]
#:xexpr-elements-proc procedure? [(symbol? x) (symbol->string x)]
#:block-xexpr-proc procedure? [(number? x) (number->string x)]
#:inline-xexpr-proc procedure? [(path? x) (path->string x)]
#:string-proc procedure?) [(char? x) (format "~a" x)]
. ->* . txexpr?)] [else (error)])))) ; put this last so other xexprish things don't get caught
[typogrify (string? . -> . string?)] ;; add a block tag to the list
[nonbreaking-last-space ((txexpr?) (#:nbsp string? #:minimum-word-length integer?) . ->* . txexpr?)] (define+provide/contract (register-block-tag tag)
[wrap-hanging-quotes ((txexpr?) (#:single-prepend list? #:double-prepend list?) . ->* . txexpr?)] (symbol? . -> . void?)
[convert-linebreaks ((txexpr-elements?) (#:newline string?) . ->* . txexpr-elements?)] (append-block-tag tag))
[whitespace? (any/c . -> . boolean?)]
[paragraph-break? ((any/c) (#:pattern pregexp?) . ->* . boolean?)]
[merge-newlines (list? . -> . list?)] ;; decoder wireframe
[prep-paragraph-flow (txexpr-elements? . -> . txexpr-elements?)] (define+provide/contract (decode nx
[wrap-paragraph ((txexpr-elements?) (#:tag symbol?) . ->* . block-xexpr?)] #:exclude-xexpr-tags [excluded-xexpr-tags '()]
[detect-paragraphs (txexpr-elements? . -> . txexpr-elements?)] #:xexpr-tag-proc [xexpr-tag-proc (λ(x)x)]
)) #:xexpr-attrs-proc [xexpr-attrs-proc (λ(x)x)]
#:xexpr-elements-proc [xexpr-elements-proc (λ(x)x)]
#:block-xexpr-proc [block-xexpr-proc (λ(x)x)]
#:inline-xexpr-proc [inline-xexpr-proc (λ(x)x)]
#:string-proc [string-proc (λ(x)x)])
((xexpr/c)
(#:exclude-xexpr-tags list?
#:xexpr-tag-proc procedure?
#:xexpr-attrs-proc procedure?
#:xexpr-elements-proc procedure?
#:block-xexpr-proc procedure?
#:inline-xexpr-proc procedure?
#:string-proc procedure?) . ->* . txexpr?)
(let loop ([x (validate-txexpr? nx)])
(cond
[(txexpr? x) (let-values([(tag attr elements) (txexpr->values x)])
(if (member tag excluded-xexpr-tags)
x ; let x pass through untouched
(let ([decoded-xexpr (apply make-txexpr (map loop (list tag attr elements)))])
((if (block-xexpr? decoded-xexpr)
block-xexpr-proc
inline-xexpr-proc) decoded-xexpr))))]
[(txexpr-tag? x) (xexpr-tag-proc x)]
[(txexpr-attrs? x) (xexpr-attrs-proc x)]
;; need this for operations that may depend on context in list
[(txexpr-elements? x) (map loop (xexpr-elements-proc x))]
[(string? x) (string-proc x)]
;; if something has made it through undecoded, that's a problem
[else (error "decode: can't decode" x)])))

63
decode/fast.rkt
Unescape Escape View File

@ -1,63 +0,0 @@
#lang racket/base
(require racket/match xml)
(require "../tools.rkt" "../predicates.rkt" txexpr "typography-fast.rkt")
(provide (all-defined-out) (all-from-out "typography-fast.rkt"))
;; general way of coercing to string
(define (to-string x)
(if (string? x)
x ; fast exit for strings
(with-handlers ([exn:fail? (λ(exn) (error (format "Pollen parser: can't convert ~v to ~a" x 'string)))])
(cond
[(equal? '() x) ""]
[(symbol? x) (symbol->string x)]
[(number? x) (number->string x)]
[(path? x) (path->string x)]
[(char? x) (format "~a" x)]
[else (error)])))) ; put this last so other xexprish things don't get caught
;; add a block tag to the list
;; this function is among the predicates because it alters a predicate globally.
(define (register-block-tag tag)
(append-block-tag tag))
;; decoder wireframe
(define (decode nx
#:exclude-xexpr-tags [excluded-xexpr-tags '()]
#:xexpr-tag-proc [xexpr-tag-proc (λ(x)x)]
#:xexpr-attrs-proc [xexpr-attrs-proc (λ(x)x)]
#:xexpr-elements-proc [xexpr-elements-proc (λ(x)x)]
#:block-xexpr-proc [block-xexpr-proc (λ(x)x)]
#:inline-xexpr-proc [inline-xexpr-proc (λ(x)x)]
#:string-proc [string-proc (λ(x)x)])
(when (not (txexpr? nx))
(error (format "decode: ~v not a full txexpr" nx)))
(define (&decode x)
(cond
[(txexpr? x) (let-values([(tag attr elements) (txexpr->values x)])
(if (member tag excluded-xexpr-tags)
x ; let x pass through untouched
(let ([decoded-xexpr (apply make-txexpr
(map &decode (list tag attr elements)))])
((if (block-xexpr? decoded-xexpr)
block-xexpr-proc
inline-xexpr-proc) decoded-xexpr))))]
[(txexpr-tag? x) (xexpr-tag-proc x)]
[(txexpr-attrs? x) (xexpr-attrs-proc x)]
;; need this for operations that may depend on context in list
[(txexpr-elements? x) (map &decode (xexpr-elements-proc x))]
[(string? x) (string-proc x)]
;; if something has made it through undecoded, that's a problem
[else (error "Can't decode" x)]))
(&decode nx))

13
decode/typography-fast.rkt → decode/typography.rkt
Unescape Escape View File

@ -3,8 +3,17 @@
(require "../tools.rkt" "../predicates.rkt" sugar txexpr) (require "../tools.rkt" "../predicates.rkt" sugar txexpr)
(provide (all-defined-out)) (provide (contract-out
[typogrify (string? . -> . string?)]
[nonbreaking-last-space ((txexpr?) (#:nbsp string? #:minimum-word-length integer?) . ->* . txexpr?)]
[wrap-hanging-quotes ((txexpr?) (#:single-prepend list? #:double-prepend list?) . ->* . txexpr?)]
[convert-linebreaks ((txexpr-elements?) (#:newline string?) . ->* . txexpr-elements?)]
[whitespace? (any/c . -> . boolean?)]
[paragraph-break? ((any/c) (#:pattern pregexp?) . ->* . boolean?)]
[merge-newlines (list? . -> . list?)]
[prep-paragraph-flow (txexpr-elements? . -> . txexpr-elements?)]
[wrap-paragraph ((txexpr-elements?) (#:tag symbol?) . ->* . block-xexpr?)]
[detect-paragraphs (txexpr-elements? . -> . txexpr-elements?)]))
;; This module is a library of functions to be used in building pollen decoders. ;; This module is a library of functions to be used in building pollen decoders.
Write Preview
Loading…
Cancel
Save