You cannot select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
txexpr/main.rkt

238 lines
8.1 KiB
Racket

11 years ago
#lang racket/base
(require (for-syntax racket/base))
11 years ago
(require racket/match xml)
11 years ago
(module+ safe (require racket/contract))
(define-syntax (define+provide+safe stx)
(syntax-case stx ()
[(_ (proc arg ... . rest-arg) contract body ...)
#'(define+provide+safe proc contract
(λ(arg ... . rest-arg) body ...))]
[(_ name contract body ...)
#'(begin
(define name body ...)
(provide name)
(module+ safe
(provide (contract-out [name contract]))))]))
11 years ago
11 years ago
(define+provide+safe (txexpr-tag? x)
11 years ago
(any/c . -> . boolean?)
(symbol? x))
11 years ago
(define+provide+safe (txexpr-attr? x)
11 years ago
(any/c . -> . boolean?)
(match x
[(list (? symbol?) (? string?)) #t]
[else #f]))
11 years ago
(define+provide+safe (txexpr-attrs? x)
(any/c . -> . boolean?)
(match x
11 years ago
[(list (? txexpr-attr?) ...) #t]
11 years ago
[else #f]))
11 years ago
(define+provide+safe (txexpr-element? x)
11 years ago
(any/c . -> . boolean?)
11 years ago
(or (string? x) (txexpr? x) (symbol? x)
11 years ago
(valid-char? x) (cdata? x)))
11 years ago
(define+provide+safe (txexpr-elements? x)
11 years ago
(any/c . -> . boolean?)
(match x
11 years ago
[(list elem ...) (andmap txexpr-element? elem)]
11 years ago
[else #f]))
;; is it a named x-expression?
;; todo: rewrite this recurively so errors can be pinpointed (for debugging)
11 years ago
(define+provide+safe (txexpr? x)
11 years ago
(any/c . -> . boolean?)
(and (xexpr? x) ; meets basic xexpr contract
(match x
11 years ago
[(list (? symbol? name) rest ...) ;; is a list starting with a symbol
11 years ago
(or (null? rest)
11 years ago
(andmap txexpr-element? rest) ;; the rest is content or ...
(and (txexpr-attrs? (car rest)) (andmap txexpr-element? (cdr rest))))] ;; attr + content
11 years ago
[else #f])))
11 years ago
(define+provide+safe (make-txexpr tag [attrs null] [elements null])
11 years ago
;; todo?: use xexpr/c provides a nicer error message
11 years ago
((symbol?) (txexpr-attrs? (listof txexpr-element?))
. ->* . txexpr?)
11 years ago
(filter (compose1 not null?) `(,tag ,attrs ,@elements)))
11 years ago
11 years ago
(define+provide+safe (txexpr->values x)
11 years ago
(txexpr? . -> .
(values symbol? txexpr-attrs? (listof txexpr-element?)))
11 years ago
(match
11 years ago
; txexpr may or may not have attr
11 years ago
; if not, add null attr so that decomposition only handles one case
11 years ago
(match x
11 years ago
[(list _ (? txexpr-attrs?) _ ...) x]
11 years ago
[else `(,(car x) ,null ,@(cdr x))])
11 years ago
[(list tag attr content ...) (values tag attr content)]))
11 years ago
(define+provide+safe (txexpr->list x)
11 years ago
(txexpr? . -> . list?)
(define-values (tag attrs content) (txexpr->values x))
(list tag attrs content))
11 years ago
11 years ago
11 years ago
;; convenience functions to retrieve only one part of txexpr
11 years ago
(define+provide+safe (get-tag x)
11 years ago
(txexpr? . -> . txexpr-tag?)
11 years ago
(car x))
11 years ago
11 years ago
11 years ago
(define+provide+safe (get-attrs x)
11 years ago
(txexpr? . -> . txexpr-attrs?)
(define-values (tag attrs content) (txexpr->values x))
attrs)
11 years ago
11 years ago
11 years ago
(define+provide+safe (get-elements x)
11 years ago
(txexpr? . -> . (listof txexpr-element?))
(define-values (tag attrs elements) (txexpr->values x))
11 years ago
elements)
11 years ago
;; helpers. we are getting a string or symbol
11 years ago
(define+provide+safe (->txexpr-attr-key x)
11 years ago
(can-be-txexpr-attr-key? . -> . txexpr-attr-key?)
(if (string? x) (string->symbol x) x))
11 years ago
11 years ago
(define+provide+safe (->txexpr-attr-value x)
11 years ago
(can-be-txexpr-attr-value? . -> . txexpr-attr-value?)
(->string x))
11 years ago
(define+provide+safe (txexpr-attr-key? x)
11 years ago
(any/c . -> . boolean?)
(symbol? x))
11 years ago
(define+provide+safe (can-be-txexpr-attr-key? x)
11 years ago
(any/c . -> . boolean?)
(or (symbol? x) (string? x)))
11 years ago
(define+provide+safe (txexpr-attr-value? x)
11 years ago
(any/c . -> . boolean?)
(string? x))
11 years ago
(define+provide+safe (can-be-txexpr-attr-value? x)
11 years ago
(any/c . -> . boolean?)
(can-be-txexpr-attr-key? x))
(define (->string x)
(if (symbol? x) (symbol->string x) x))
11 years ago
(define+provide+safe (can-be-txexpr-attrs? x)
11 years ago
(any/c . -> . boolean?)
11 years ago
(ormap (λ(test) (test x)) (list txexpr-attr? txexpr-attrs? can-be-txexpr-attr-key? can-be-txexpr-attr-value?)))
(define (flatten orig-sexp)
(let loop ([sexp orig-sexp] [acc null])
(cond [(null? sexp) acc]
[(pair? sexp) (loop (car sexp) (loop (cdr sexp) acc))]
[else (cons sexp acc)])))
11 years ago
(define+provide+safe (attrs->hash . items)
11 years ago
(() #:rest (listof can-be-txexpr-attrs?) . ->* . hash?)
;; can be liberal with input because they're all just nested key/value pairs
;; but still need this function to make sure that 'foo and "foo" are treated as the same hash key
(define (make-key-value-list items)
11 years ago
(if (null? items)
null
(let ([key (->txexpr-attr-key (car items))]
[value (->txexpr-attr-value (cadr items))]
[rest (cddr items)])
11 years ago
(cons (cons key value) (make-key-value-list rest)))))
(make-immutable-hash (make-key-value-list (flatten items))))
11 years ago
(define+provide+safe (hash->attrs hash)
11 years ago
(hash? . -> . txexpr-attrs?)
(hash-map hash list))
(define+provide+safe (attrs-have-key? x key)
((or/c txexpr-attrs? txexpr?) can-be-txexpr-attr-key? . -> . boolean?)
(define attrs (if (txexpr-attrs? x) x (get-attrs x)))
(hash-has-key? (attrs->hash attrs) (->txexpr-attr-key key)))
11 years ago
(define+provide+safe (attr-set tx key value)
11 years ago
(txexpr? can-be-txexpr-attr-key? can-be-txexpr-attr-value? . -> . txexpr?)
(define new-attrs
(hash->attrs (hash-set (attrs->hash (get-attrs tx)) (->txexpr-attr-key key) (->txexpr-attr-value value))))
(make-txexpr (get-tag tx) new-attrs (get-elements tx)))
11 years ago
(define+provide+safe (attr-ref tx key)
11 years ago
(txexpr? can-be-txexpr-attr-key? . -> . txexpr-attr-value?)
(with-handlers ([exn:fail? (λ(e) (error (format "attr-ref: no value found for key ~v" key)))])
(hash-ref (attrs->hash (get-attrs tx)) key)))
;; convert list of alternating keys & values to attr
11 years ago
(define+provide+safe (merge-attrs . items)
11 years ago
(() #:rest (listof can-be-txexpr-attrs?) . ->* . txexpr-attrs?)
(define attrs-hash (apply attrs->hash items))
;; sort needed for predictable results for unit tests
(define sorted-hash-keys (sort (hash-keys attrs-hash) (λ(a b) (string<? (->string a) (->string b)))))
`(,@(map (λ(key) (list key (hash-ref attrs-hash key))) sorted-hash-keys)))
11 years ago
11 years ago
(define+provide+safe (remove-attrs x)
11 years ago
(txexpr? . -> . txexpr?)
(cond
[(txexpr? x) (let-values ([(tag attr elements) (txexpr->values x)])
11 years ago
(make-txexpr tag null (remove-attrs elements)))]
11 years ago
[(txexpr-elements? x) (map remove-attrs x)]
11 years ago
[else x]))
11 years ago
11 years ago
;; todo: exclude-proc will keep things out, but is there a way to keep things in?
11 years ago
(define+provide+safe (map-elements/exclude proc x exclude-test)
11 years ago
(procedure? txexpr? procedure? . -> . txexpr?)
(cond
[(txexpr? x)
(if (exclude-test x)
x
(let-values ([(tag attr elements) (txexpr->values x)])
(make-txexpr tag attr
(map (λ(x)(map-elements/exclude proc x exclude-test)) elements))))]
;; externally the function only accepts txexpr,
;; but internally we don't care
[else (proc x)]))
11 years ago
(define+provide+safe (map-elements proc x)
11 years ago
(procedure? txexpr? . -> . txexpr?)
(map-elements/exclude proc x (λ(x) #f)))
11 years ago
;; function to split tag out of txexpr
11 years ago
(define+provide+safe (splitf-txexpr tx proc)
(txexpr? procedure? . -> . (values txexpr? (listof txexpr-element?)))
11 years ago
(define matches null)
11 years ago
(define (do-extraction x)
(cond
11 years ago
[(proc x) (begin ; store matched item but return null value
11 years ago
(set! matches (cons x matches))
11 years ago
null)]
11 years ago
[(txexpr? x) (let-values([(tag attr body) (txexpr->values x)])
(make-txexpr tag attr (do-extraction body)))]
11 years ago
[(txexpr-elements? x) (filter (compose1 not null?) (map do-extraction x))]
11 years ago
[else x]))
(define tx-extracted (do-extraction tx)) ;; do this first to fill matches
(values tx-extracted (reverse matches)))
(define+provide+safe (txexpr->html x)
(txexpr? . -> . string?)
(define (->cdata x)
(if (cdata? x) x (cdata #f #f x)))
(xexpr->string (let loop ([x x])
(cond
[(txexpr? x) (if (member (get-tag x) '(script style))
(make-txexpr (get-tag x) (get-attrs x) (map ->cdata (get-elements x)))
(make-txexpr (get-tag x) (get-attrs x) (map loop (get-elements x))))]
[else x]))))