by Ken Dickey
This copy of the SRFI 48 specification document is distributed as part of the Racket package srfi-doc.
The canonical source of this document is https://srfi.schemers.org/srfi-48/srfi-48.html.
This SRFI is currently in final status. Here is an explanation of each status that a SRFI can hold. To provide input on this SRFI, please send email to srfi-48@nospamsrfi.schemers.org. To subscribe to the list, follow these instructions. You can access previous messages via the mailing list archive.
~&) directive, suggesting a change when it
is not known whether the previous character was a newline.
This is not an error in the SRFI, but deviation from it that
Per recommends. SRFI 48 author Ken Dickey responded with a
patch to the sample implementation that could be used to
implement Per's change, if desired.This document specifies Format Strings, a method of interpreting a Scheme string which contains a number of format directives that are replaced with other string data according to the semantics of each directive. This SRFI extends SRFI-28 in being more generally useful but is less general than advanced format strings in that it does not allow, aside from ~F, for controlled positioning of text within fields.
Some may disagree with specific escape options or return values. For those who desire complex options as implemented by SLIB or Common Lisp's FORMAT, an upwards compatible "Advanced Format" SRFI should be proposed.
In particular, the reference implementation given here does not accept numeric arguments (aside from ~F). Hence it does not support SRFI-29.
It is highly desireable that baseline library code be small, attempt to eliminiate heap allocation and bound stack usage. This is especially important in embedded systems. This can be accomplished by writing directly to a port, rather than a string, by not supporting ~W or ~F, and by replacing (display (number->string n r) p) with a carefully written (display:number->string n r p) which does not build intermediate strings.
As this is intermediate format, it was felt that ~F and ~W are too highly useful to elide. The ~H option is helpful to users, allows for programattic query, and makes clear which format directives are supported.
Inheriting from MacLisp, nearly all Lisp and Scheme implementations support some form of FORMAT function with support for various numbers of format directives. By agreeing to the options here, we raise the bar for portable code.
The reference implementation is R5RS compliant and easy to port. In not requiring advanced features (aside from ~W and ~F) small implementations are possible. E.g. the reference code does not use side effects (assignment) and is less than a third the source size of the latest SLIB implementation of FORMAT (less than a tenth if ~F support is elided).
The optional port argument allows for compatibility with older code written for, e.g. scheme48, MIT Scheme, T, et cetera, which required a port argument. It is also useful in cases where a synoptic implementation of Scheme and CommonLisp is maintained.
Accepts a format template (a Scheme String), and processes it, replacing any format directives in order with one or more characters, the characters themselves dependent on the semantics of the format directive encountered. Each directive may consume one obj. It is an error if fewer or more obj values are provided than format directives that require them.
When a port is specified it must be either an output port or a boolean. If an output-port is specified, the formatted output is output into that port. If the port argument is #t, output is to the current-output-port. If the port is #f or no port is specified, the output is returned as a string. If the port is specified and is #t or an output-port, the result of the format function is unspecified.
It is unspecified which encoding is used (e.g. ASCII, EBCDIC, UNICODE). A given implementation must specify which encoding is used. The implementation may or may not allow the encoding to be selected or changed.
It is an error if an format directive consumes an obj argument and that argument does not confirm to a required type as noted in the table below.
It is permissible, but highly discouraged, to implement pretty-print as (define pretty-print write).
An format directive is a two character sequence in the string where the first character is a tilde '~'. Directive characters are case-independent, i.e. upper and lower case characters are interpreted the same. Each directive code's meaning is described in the following table:
DIRECTIVE MNEMONIC ACTION CONSUMES? ~a Any (display obj) for humans yes ~s Slashified (write obj) for parsers yes ~w WriteCircular (write-with-shared-structure obj) like ~s, but handles recursive structures yes ~d Decimal the obj is a number which is output in decimal radix yes ~x heXadecimal the obj is a number which is output in hexdecimal radix yes ~o Octal the obj is a number which is output in octal radix yes ~b Binary the obj is a number which is output in binary radix yes ~c Character the single charater obj is output by write-char yes ~y Yuppify the list obj is pretty-printed to the output yes ~? Indirection the obj is another format-string and the following obj is a list of arguments; format is called recursively yes ~K Indirection the same as ~? for backward compatability with some existing implementations yes ~[w[,d]]F Fixed ~w,dF outputs a number with width w and d digits after the decimal; ~wF outputs a string or number with width w. yes ~~ Tilde output a tilde no ~t Tab output a tab character no ~% Newline output a newline character no ~& Freshline output a newline character if it is known that the previous output was not a newline no ~_ Space a single space character is output no ~h Help outputs one line of call synopsis, one line of comment, and one line of synopsis for each format directive, starting with the directive (e.g. "~t") no
The ~F, fixed format, directive requires some elucidation.
~wF is useful for strings or numbers. Where the string (or number->string of the number) has fewer characters than the integer width w, the string is padded on the left with space characters.
~w,dF is typically used only on numbers. For strings, the d specifier is ignored. For numbers, the integer d specifies the number of decimal digits after the decimal place. Both w and d must be zero or positive.
If d is specified, the number is processed as if added to 0.0, i.e. it is converted to an inexact value.
(format "~8,2F" 1/3) => " 0.33"If no d is specified, the number is not coerced to inexact.
(format "~6F" 32) => " 32"Digits are padded to the right with zeros
(format "~8,2F" 32) => " 32.00"If the number it too large to fit in the width specified, a string longer than the width is returned
(format "~1,2F" 4321) => "4321.00"If the number is complex, d is applied to both real and imaginal parts
(format "~1,2F" (sqrt -3.9)) => "0.00+1.97i"For very large or very small numbers, the point where exponential notation is used is implementation defined.
(format "~8F" 32e5) => " 3.2e6" or "3200000.0"
(format "~h") ; => "(format [<port>] <format-string> [<arg>...]) -- <port> is #t, #f or an output-port OPTION [MNEMONIC] DESCRIPTION -- This implementation Assumes ASCII Text Encoding ~H [Help] output this text ~A [Any] (display arg) for humans ~S [Slashified] (write arg) for parsers ~~ [tilde] output a tilde ~T [Tab] output a tab character ~% [Newline] output a newline character ~& [Freshline] output a newline character if the previous output was not a newline ~D [Decimal] the arg is a number which is output in decimal radix ~X [heXadecimal] the arg is a number which is output in hexdecimal radix ~O [Octal] the arg is a number which is output in octal radix ~B [Binary] the arg is a number which is output in binary radix ~w,dF [Fixed] the arg is a string or number which has width w and d digits after the decimal ~C [Character] charater arg is output by write-char ~_ [Space] a single space character is output ~Y [Yuppify] the list arg is pretty-printed to the output ~? [Indirection] recursive format: next arg is a format-string and the following arg a list of arguments ~K [Indirection] same as ~? " (format "Hello, ~a" "World!") ; => "Hello, World!" (format "Error, list is too short: ~s" '(one "two" 3)) ; => "Error, list is too short: (one \"two\" 3)" (format "test me") ; => "test me" (format "~a ~s ~a ~s" 'this 'is "a" "test") ; => "this is a \"test\"" (format #t "#d~d #x~x #o~o #b~b~%" 32 32 32 32) ;; Prints: #d32 #x20 #o40 #b100000 ; => <unspecified> (format "~a ~? ~a" 'a "~s" '(new) 'test) ; =>"a new test" (format #f "~&1~&~&2~&~&~&3~%") ; => " 1 2 3 " (format #f "~a ~? ~a ~%" 3 " ~s ~s " '(2 2) 3) ; => "3 2 2 3 " (format "~w" (let ( (c '(a b c)) ) (set-cdr! (cddr c) c) c)) ; => "#1=(a b c . #1#)" (format "~8,2F" 32) ; => " 32.00" (format "~8,3F" (sqrt -3.8)) ; => "0.000+1.949i" (format "~8,2F" 3.4567e11) ; => " 3.45e11" (format "~6,3F" 1/3) ; => " 0.333" (format "~4F" 12) ; => " 12" (format "~8,3F" 123.3456) ; => " 123.346" (format "~6,3F" 123.3456) ; => "123.346" (format "~2,3F" 123.3456) ; => "123.346" (format "~8,3F" "foo") ; => " foo" (format "~a~a~&" (list->string (list #\newline)) "") ; => " "
;; IMPLEMENTATION DEPENDENT options
(define ascii-tab (integer->char 9)) ;; NB: assumes ASCII encoding
(define dont-print (if (eq? #t #f) 1))
;;(define DONT-PRINT (string->symbol ""))
;;(define DONT-PRINT (void))
;;(define DONT-PRINT #!void)
(define pretty-print write) ; ugly but permitted
;; (require 'srfi-38) ;; write-with-shared-structure
;; Following three procedures are used by format ~F .
;; 'inexact-number->string' determines whether output is fixed-point
;; notation or exponential notation. In the current definition,
;; the notation depends on the implementation of 'number->string'.
;; 'exact-number->string' is expected to output only numeric characters
;; (not including such as '#', 'e', '.', '/') if the input is an positive
;; integer or zero.
;; 'real-number->string' is used when the digits of ~F is not specified.
(define (inexact-number->string x) (number->string (exact->inexact x)))
(define (exact-number->string x) (number->string (inexact->exact x)))
(define (real-number->string x) (number->string x))
;; FORMAT
(define (format . args)
(cond
((null? args)
(error "FORMAT: required format-string argument is missing")
)
((string? (car args))
(apply format (cons #f args)))
((< (length args) 2)
(error (format #f "FORMAT: too few arguments ~s" (cons 'format args)))
)
(else
(let ( (output-port (car args))
(format-string (cadr args))
(args (cddr args))
)
(letrec ( (port
(cond ((output-port? output-port) output-port)
((eq? output-port #t) (current-output-port))
((eq? output-port #f) (open-output-string))
(else (error
(format #f "FORMAT: bad output-port argument: ~s"
output-port)))
) )
(return-value
(if (eq? output-port #f) ;; if format into a string
(lambda () (get-output-string port)) ;; then return the string
(lambda () dont-print)) ;; else do something harmless
)
)
(define (string-index str c)
(let ( (len (string-length str)) )
(let loop ( (i 0) )
(cond ((= i len) #f)
((eqv? c (string-ref str i)) i)
(else (loop (+ i 1)))))))
(define (string-grow str len char)
(let ( (off (- len (string-length str))) )
(if (positive? off)
(string-append (make-string off char) str)
str)))
(define (compose-with-digits digits pre-str frac-str exp-str)
(let ( (frac-len (string-length frac-str)) )
(cond
((< frac-len digits) ;; grow frac part, pad with zeros
(string-append pre-str "."
frac-str (make-string (- digits frac-len) #\0)
exp-str)
)
((= frac-len digits) ;; frac-part is exactly the right size
(string-append pre-str "."
frac-str
exp-str)
)
(else ;; must round to shrink it
(let* ( (minus-flag (and (> (string-length pre-str) 0)
(char=? (string-ref pre-str 0) #\-)))
(pre-str* (if minus-flag
(substring pre-str 1 (string-length pre-str))
pre-str))
(first-part (substring frac-str 0 digits))
(last-part (substring frac-str digits frac-len))
(temp-str
(string-grow
(exact-number->string
(round (string->number
(string-append pre-str* first-part "." last-part))))
digits
#\0))
(temp-len (string-length temp-str))
(new-pre (substring temp-str 0 (- temp-len digits)))
(new-frac (substring temp-str (- temp-len digits) temp-len))
)
(string-append
(if minus-flag "-" "")
(if (string=? new-pre "")
;; check if the system displays integer part of numbers
;; whose absolute value is 0 < x < 1.
(if (and (string=? pre-str* "")
(> digits 0)
(not (= (string->number new-frac) 0)))
"" "0")
new-pre)
"."
new-frac
exp-str)))
) ) )
(define (format-fixed number-or-string width digits) ; returns a string
(cond
((string? number-or-string)
(string-grow number-or-string width #\space)
)
((number? number-or-string)
(let ( (real (real-part number-or-string))
(imag (imag-part number-or-string))
)
(cond
((not (zero? imag))
(string-grow
(string-append (format-fixed real 0 digits)
(if (negative? imag) "" "+")
(format-fixed imag 0 digits)
"i")
width
#\space)
)
(digits
(let* ( (num-str (inexact-number->string real))
(dot-index (string-index num-str #\.))
(exp-index (string-index num-str #\e))
(length (string-length num-str))
(pre-string
(if dot-index
(substring num-str 0 dot-index)
(if exp-index
(substring num-str 0 exp-index)
num-str))
)
(exp-string
(if exp-index
(substring num-str exp-index length)
"")
)
(frac-string
(if dot-index
(if exp-index
(substring num-str (+ dot-index 1) exp-index)
(substring num-str (+ dot-index 1) length))
"")
)
)
;; check +inf.0, -inf.0, +nan.0, -nan.0
(if (string-index num-str #\n)
(string-grow num-str width #\space)
(string-grow
(compose-with-digits digits
pre-string
frac-string
exp-string)
width
#\space))
))
(else ;; no digits
(string-grow (real-number->string real) width #\space)))
))
(else
(error
(format "FORMAT: ~F requires a number or a string, got ~s" number-or-string)))
))
(define documentation-string
"(format [<port>] <format-string> [<arg>...]) -- <port> is #t, #f or an output-port
OPTION [MNEMONIC] DESCRIPTION -- Implementation Assumes ASCII Text Encoding
~H [Help] output this text
~A [Any] (display arg) for humans
~S [Slashified] (write arg) for parsers
~W [WriteCircular] like ~s but outputs circular and recursive data structures
~~ [tilde] output a tilde
~T [Tab] output a tab character
~% [Newline] output a newline character
~& [Freshline] output a newline character if the previous output was not a newline
~D [Decimal] the arg is a number which is output in decimal radix
~X [heXadecimal] the arg is a number which is output in hexdecimal radix
~O [Octal] the arg is a number which is output in octal radix
~B [Binary] the arg is a number which is output in binary radix
~w,dF [Fixed] the arg is a string or number which has width w and d digits after the decimal
~C [Character] charater arg is output by write-char
~_ [Space] a single space character is output
~Y [Yuppify] the list arg is pretty-printed to the output
~? [Indirection] recursive format: next 2 args are format-string and list of arguments
~K [Indirection] same as ~?
"
)
(define (require-an-arg args)
(if (null? args)
(error "FORMAT: too few arguments" ))
)
(define (format-help format-strg arglist)
(letrec (
(length-of-format-string (string-length format-strg))
(anychar-dispatch
(lambda (pos arglist last-was-newline)
(if (>= pos length-of-format-string)
arglist ; return unused args
(let ( (char (string-ref format-strg pos)) )
(cond
((eqv? char #\~)
(tilde-dispatch (+ pos 1) arglist last-was-newline))
(else
(write-char char port)
(anychar-dispatch (+ pos 1) arglist #f)
))
))
)) ; end anychar-dispatch
(has-newline?
(lambda (whatever last-was-newline)
(or (eqv? whatever #\newline)
(and (string? whatever)
(let ( (len (string-length whatever)) )
(if (zero? len)
last-was-newline
(eqv? #\newline (string-ref whatever (- len 1)))))))
)) ; end has-newline?
(tilde-dispatch
(lambda (pos arglist last-was-newline)
(cond
((>= pos length-of-format-string)
(write-char #\~ port) ; tilde at end of string is just output
arglist ; return unused args
)
(else
(case (char-upcase (string-ref format-strg pos))
((#\A) ; Any -- for humans
(require-an-arg arglist)
(let ( (whatever (car arglist)) )
(display whatever port)
(anychar-dispatch (+ pos 1)
(cdr arglist)
(has-newline? whatever last-was-newline))
))
((#\S) ; Slashified -- for parsers
(require-an-arg arglist)
(let ( (whatever (car arglist)) )
(write whatever port)
(anychar-dispatch (+ pos 1)
(cdr arglist)
(has-newline? whatever last-was-newline))
))
((#\W)
(require-an-arg arglist)
(let ( (whatever (car arglist)) )
(write-with-shared-structure whatever port) ;; srfi-38
(anychar-dispatch (+ pos 1)
(cdr arglist)
(has-newline? whatever last-was-newline))
))
((#\D) ; Decimal
(require-an-arg arglist)
(display (number->string (car arglist) 10) port)
(anychar-dispatch (+ pos 1) (cdr arglist) #f)
)
((#\X) ; HeXadecimal
(require-an-arg arglist)
(display (number->string (car arglist) 16) port)
(anychar-dispatch (+ pos 1) (cdr arglist) #f)
)
((#\O) ; Octal
(require-an-arg arglist)
(display (number->string (car arglist) 8) port)
(anychar-dispatch (+ pos 1) (cdr arglist) #f)
)
((#\B) ; Binary
(require-an-arg arglist)
(display (number->string (car arglist) 2) port)
(anychar-dispatch (+ pos 1) (cdr arglist) #f)
)
((#\C) ; Character
(require-an-arg arglist)
(write-char (car arglist) port)
(anychar-dispatch (+ pos 1) (cdr arglist) (eqv? (car arglist) #\newline))
)
((#\~) ; Tilde
(write-char #\~ port)
(anychar-dispatch (+ pos 1) arglist #f)
)
((#\%) ; Newline
(newline port)
(anychar-dispatch (+ pos 1) arglist #t)
)
((#\&) ; Freshline
(if (not last-was-newline) ;; (unless last-was-newline ..
(newline port))
(anychar-dispatch (+ pos 1) arglist #t)
)
((#\_) ; Space
(write-char #\space port)
(anychar-dispatch (+ pos 1) arglist #f)
)
((#\T) ; Tab -- IMPLEMENTATION DEPENDENT ENCODING
(write-char ascii-tab port)
(anychar-dispatch (+ pos 1) arglist #f)
)
((#\Y) ; Pretty-print
(pretty-print (car arglist) port) ;; IMPLEMENTATION DEPENDENT
(anychar-dispatch (+ pos 1) (cdr arglist) #f)
)
((#\F)
(require-an-arg arglist)
(display (format-fixed (car arglist) 0 #f) port)
(anychar-dispatch (+ pos 1) (cdr arglist) #f)
)
((#\0 #\1 #\2 #\3 #\4 #\5 #\6 #\7 #\8 #\9) ;; gather "~w[,d]F" w and d digits
(let loop ( (index (+ pos 1))
(w-digits (list (string-ref format-strg pos)))
(d-digits '())
(in-width? #t)
)
(if (>= index length-of-format-string)
(error
(format "FORMAT: improper numeric format directive in ~s" format-strg))
(let ( (next-char (string-ref format-strg index)) )
(cond
((char-numeric? next-char)
(if in-width?
(loop (+ index 1)
(cons next-char w-digits)
d-digits
in-width?)
(loop (+ index 1)
w-digits
(cons next-char d-digits)
in-width?))
)
((char=? (char-upcase next-char) #\F)
(let ( (width (string->number (list->string (reverse w-digits))))
(digits (if (zero? (length d-digits))
#f
(string->number (list->string (reverse d-digits)))))
)
(display (format-fixed (car arglist) width digits) port)
(anychar-dispatch (+ index 1) (cdr arglist) #f))
)
((char=? next-char #\,)
(if in-width?
(loop (+ index 1)
w-digits
d-digits
#f)
(error
(format "FORMAT: too many commas in directive ~s" format-strg)))
)
(else
(error (format "FORMAT: ~~w.dF directive ill-formed in ~s" format-strg))))))
))
((#\? #\K) ; indirection -- take next arg as format string
(cond ; and following arg as list of format args
((< (length arglist) 2)
(error
(format "FORMAT: less arguments than specified for ~~?: ~s" arglist))
)
((not (string? (car arglist)))
(error
(format "FORMAT: ~~? requires a string: ~s" (car arglist)))
)
(else
(format-help (car arglist) (cadr arglist))
(anychar-dispatch (+ pos 1) (cddr arglist) #f)
)))
((#\H) ; Help
(display documentation-string port)
(anychar-dispatch (+ pos 1) arglist #t)
)
(else
(error (format "FORMAT: unknown tilde escape: ~s"
(string-ref format-strg pos))))
)))
)) ; end tilde-dispatch
) ; end letrec
; format-help main
(anychar-dispatch 0 arglist #f)
)) ; end format-help
; format main
(let ( (unused-args (format-help format-string args)) )
(if (not (null? unused-args))
(error
(format "FORMAT: unused arguments ~s" unused-args)))
(return-value))
)) ; end letrec, if
))) ; end format
Copyright (C) Kenneth A Dickey (2003). All Rights Reserved.
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.