This site is a static rendering of the Trac instance that was used by R7RS-WG1 for its work on R7RS-small (PDF), which was ratified in 2013. For more information, see Home. For a version of this page that may be more recent, see StringSlicesCowan in WG2's repo for R7RS-large.

String­Slices­Cowan

cowan
2014-12-18 09:45:45
27history
source

Character span library

This is a library for manipulating textual content based on character spans, also known as just spans. These are conceptually references to a part of a string delimited using two string cursors, which are references to the characters of a string. It is not defined whether the character span type is disjoint from strings. Character spans are immutable, and except as noted below, it is an error to mutate the string(s) that underly a span.

Issues

  1. Is character span a satisfactory name?
  1. Allow negative indices in make-span and subspan? Convenient, but irregular.
  1. Titlecase doesn't really fit; keep it?
  1. (resolved)
  1. Keep string trees?
  1. Keep compatibility routines, possibly in a different package?
  1. I have made the argument order of string-tabulate compatible with SRFI 1 list-tabulate rather than SRFI 13's string-tabulate; the discrepancy was accidental. Revert to SRFI 13 argument order?
  1. Currently string-mismatch returns a cursor into its second argument. Should it return a cursor into the first argument, or into both arguments? (In Chibi it makes no difference.)
  1. Keep in-char-set?? It creates a dependency on SRFI 14, but is essential for SRFI 13 emulation.
  1. Bring back -ci variants of the prefix and suffix operations and span-compare from SRFI 13?

Rationale

When SRFI 13 was defined in 1999, it was intended to provide efficient string operations on both whole strings and substrings. At that time, only Guile and T provided true shared copy-on-write substrings, and SRFI 13 could not reasonably require them of a Scheme implementation. Consequently, almost all the SRFI 13 procedures accept optional start and end arguments for each of the string arguments, indexing the beginning and the end of the substring(s) to be operated on.

Unfortunately, variable-arity procedures are often slow and may not interact well with type checking in Schemes that provide it. In addition, it is now fairly common to store strings internally as UTF-8 or UTF-16 code unit sequences, which means that indexing operations are often O(n) rather than O(1), and string mutation can be extremely expensive if the storage used for the string needs to be expanded and the implementation does not use an indirect pointer to it (as in Chicken).

As for shared substrings, they are no more common in 2015 than they were in 1999. Fortunately, however, since then it has become normal for Schemes to provide user-defined records, and they are required by both R6RS and R7RS. This makes it possible to portably provide a representation for a segment of a string, provided the string is never mutated. The most portable such record consists of a string and two indexes, but other more efficient representations may be used instead.

This proposal, therefore, is intended to help move the practice of Scheme programming away from mutable strings, string indexes, and SRFI 13, while maintaining as much backward compatibility as is consistent with these goals. It does not require any particular run-time efficiencies from its procedures. The string procedures, as well as string-transform, make it possible to migrate a code base gradually.

It is also possible to implement character spans as ropes (trees of strings), which makes concatenation more efficient at the expense of more complex cursor objects and/or slower conversion to strings. For this reason, as well as for security and efficiency reasons, there is no operation to retrieve an underlying string from a character span, as there may be more than one such string or none at all. The operations provided here (with the exception of those in the Compatibility section) are entirely independent of the character repertoire supported by the implementation. In particular, this means that the case-insensitive procedures of SRFI 13 are excluded. There is also no provision for R6RS normalization procedures or for an string->integer procedure that was proposed for SRFI 13 but not included. These may appear in future SRFIs.

Specification

String cursors are pointers into strings or spans, and are not necessarily disjoint from other Scheme types. For example, they may be exact integers that are character-based indexes into strings. Alternatively, in an implementation whose internal representation of strings is UTF-8, string cursors may be indexes of individual bytes in the string. It is also possible to implement string cursors as objects of a disjoint type. The string-cursor procedures of this proposal are mostly taken from Chibi Scheme.

Given a span of length n, there are n+2 possible cursors that refer to it: one for each character in the span, one for the position just before the first character, and one for the position just after the last character. These additional positions are provided for backward and forward iteration respectively, and also because when creating a span from cursors the second cursor argument is exclusive.

Most of the character span procedures in this proposal also have string equivalents. In order to make the specification more concise, the string procedures are listed but don't have detailed explanations, except for the constructors. Procedures with the same names and basic functions as SRFI 13 procedures are marked [SRFI 13]. However, this proposal contains only a subset of SRFI 13. In particular, the string procedures of this proposal do not accept start and end arguments, as their function is subsumed by spans, nor is SRFI 13's separation between sharable and non-sharable results supported. In addition, the SRFI 13 low-level procedures and macros are not provided, nor are there any mutators.

All the R7RS-small string procedures are included here, with the exception of the string mutators string-set!, string-copy!, and string-fill!. They are marked [R7RS-small], and are not exported by implementations of this proposal meant for R7RS systems. They are included only for clarity and completeness.

All predicates passed to procedures defined in this proposal may be called in any order and any number of times, except as otherwise noted. In SRFI 13, there is no such provision, and so character sets are inherently more efficient than predicates because testing them is fast and free of side effects, though how fast character sets are if they support full Unicode is implementation-dependent.

Character span constructors

(make-whole-span string)

Returns a character span which contains all the characters of string in order.

(make-span string start end ] ])

Returns a character span which contains the characters of string in order from indexes start (inclusive) to end (exclusive).

(span char ...)

Returns a character span which contains the characters char in order.

(string->span string)

Returns a character span which contains the characters of string in order. Later mutation of string will not affect the value of span.

(span-transform proc span obj ...)

Proc is a procedure which accepts a string as its first argument and returns a string. It is invoked on a string which contains the characters of span in order plus the obj arguments, if any. The resulting string is returned as a character span by span-transform. This procedure allows string-based procedures to be easily used in a context that provides and expects spans.

String constructors

(make-string k [ char ]) [R7RS-small]

Returns a string containing k characters, all of which are char. If char is omitted, the contents of the string are implementation-dependent.

(string char ...) [R7RS-small]

Returns a string consisting of the char arguments.

(string-unfold stop? mapper successor [ seed ])

(string-unfold-right stop? mapper successor [ seed ])

Returns a newly allocated string, whose characters are generated in forward/reverse order using the following algorithm: If the result of applying the predicate stop? to seed is true, the string is complete and is returned. Otherwise, apply the procedure mapper to seed. The value that mapper returns becomes the next character of the string. Then a new seed is obtained by applying the procedure successor to seed, and this algorithm is repeated.

(span->string span)

Returns a newly allocated string which contains the characters of span in order.

(string-tabulate len proc)

Invokes proc for all exact integers between 0 (inclusive) and len (exclusive), and returns a newly allocated string containing the characters returned by the invocations.

Compatibility note: The argument order here agrees with the list-tabulate procedure of SRFI 1 rather than SRFI 13's string-tabulate procedure. The discrepancy was unintentional, but was discovered too late to fix.

Predicates

(span? obj)

(string? obj) [R7RS-small]

Returns #t if obj is a character span, and #f otherwise.

(span-null? span)

(string-null? string) [SRFI 13]

Returns #t if span contains zero characters, and #f otherwise.

(span-every pred span)

(string-every pred string) [SRFI 13]

Returns #t if pred returns true for every character in span, and #f otherwise.

(span-any pred span)

(string-any pred string) [SRFI 13]

Returns #f if pred returns false for each character in span, and #t otherwise.

(is-char? char)

Returns a predicate which accepts one argument. This predicate returns #t if the argument is the same as char (in the sense of char=?) and #f otherwise.

(in-char-set? char-set)

Returns a predicate which accepts one argument. This predicate returns #t if the argument is an element of char-set, a SRFI 14 character set, and #f otherwise.

Selection

(span-ref span k)

(string-ref string k) [R7RS-small]

Returns the kth character of span, starting with 0. It is an error if k is not a non-negative exact integer less than the length of span.

(span-take span n)

(string-take string n) [SRFI 13]

Returns a character span which contains the first n characters of span.

(span-take-right span n)

(string-take-right string n) [SRFI 13]

Returns a character span which contains the last n characters of span.

(span-drop span n)

(string-drop string n) [SRFI 13]

Returns a character span which contains all but the first n characters of span.

(span-drop-right span n)

(string-drop-right string n) [SRFI 13]

Returns a character span which contains all but the last n characters of span.

(span-split-at span n)

(string-split-at string n) [SRFI 13]

Returns two values, a character span containing the first n characters of span, and another character span containing the remaining characters of span.

(span-replicate span from to)

(string-replicate string from to)

Span is conceptually replicated an infinite number of times to both left and right, and this doubly infinite sequence is then truncated to form a span starting at index from (inclusive) and ending at index to (exclusive). Negative indexes are allowed in order to access the infinite left extension.

Compatibility note: This procedure is the same as the SRFI 13 procedure xsubstring, except that the to argument is required.

Examples:

(string-replicate "abcdef" 2 7) => "cdefab" ; rotate left (string-replicate "abcdef" -2 4) => "efabcd" ; rotate right (string-replicate "abc" 0 7) => "abcabca" ; replicate

(subspan span start end)

(substring string start end) [R7RS-small]

(subspan/cursors span start end)

(substring/cursors string start end)

Returns a character span which contains the characters in span between the indexes/cursors start (inclusive) and end (exclusive).

Padding, trimming, and compressing

(span-pad span len [ char ])

(string-pad string len [ char ]) [SRFI 13]

(span-pad-right span len [ char ])

(string-pad-right string len [ char ]) [SRFI 13]

Returns a span of length len consisting of span padded on the left/right by as many occurrences of the character char as needed. If span has more than len characters, it is truncated on the left (right) to length len. If char is omitted, #\space is used.

(span-trim span [ pred'' ])

(string-trim string [ pred'' ]) [SRFI 13]

(span-trim-right span [ pred ][ char'' ])

(string-trim-right string [ pred'' ]) [SRFI 13]

(span-trim-both span [ pred'' ])

(string-trim-both string [ pred'' ]) [SRFI 13]

Trim span by skipping over all characters on the left / on the right / on both sides that satisfy pred and returning the resulting span.

(span-compress span [ char ])

(string-compress string [ char ])

Return a span which differs from span in that every sequence of consecutive occurrences of char has been replaced by a single char. If char is omitted, #\space is used.

Prefixes and suffixes

(span-prefix span1 span2)

(string-prefix string1 string2)

(span-suffix span1 span2)

(string-suffix string1 string2)

Returns a span containing the characters in the common prefix/suffix of span1 and span2. If there is no common prefix/suffix, returns an empty span.

(span-prefix-length span1 span2)

(string-prefix-length string1 string2) [SRFI 13]

(span-suffix-length span1 span2)

(string-suffix-length string1 string2) [SRFI 13]

Returns the length (in characters) of the span that would be returned by span-prefix and friends.

(span-mismatch span1 span2)

(string-mismatch string1 string2)

Returns a cursor referring to the first character in span2 that is not equal to the corresponding character in span1. If the spans are equal, the cursor referring to the position after the last character in span2 is returned.

(span-mismatch-right span1 span2)

(string-mismatch-right string1 string2)

Returns a cursor referring to the first character in span2, scanning in reverse order, that is not equal to the corresponding character in span1, also processed in reverse order. If the spans are equal, the cursor referring to the position before the first character in span2 is returned.

(span-prefix? span1 span2)

(string-prefix? string1 string2) [SRFI 13]

(span-suffix? span1 span2)

(string-suffix? string1 string2) [SRFI 13]

Returns #t if span1 is a prefix/suffix of span2, and #f otherwise.

Searching

(span-count pred span)

(string-count pred string) [SRFI 13]

Returns the number of characters in span which satisfy predas an exact integer.

(span-find pred span)

(string-find pred string)

Returns a cursor referring to the first character in span that satisfies pred, or the cursor referring to the position after the last character if there is none.

(span-find-right pred span)

(string-find-right pred string)

Returns a cursor referring to the first character in span that satisfies pred', processing span'' in reverse order, or the cursor referring to the position before the first character if there is none.

Compatibility note: These procedures are analogous to SRFI 13's string-index procedures, but return cursors rather than indexes or #f on failure.

(span-bypass pred span)

(string-bypass pred string)

(span-bypass-right pred span)

(string-bypass-right pred string)

Returns a cursor pointing to the first/last character in span that does not satisfy pred, or the end/start cursor if there is none.

Compatibility note: These procedures are analogous to SRFI 13's string-skip procedures, but return cursors rather than indexes or #f on failure.

(span-take-while pred span)

(string-take-while pred string)

Returns the longest initial prefix of span whose characters all satisfy pred.

(span-drop-while pred span)

(string-drop-while pred string)

Drops the longest initial prefix of span whose characters all satisfy pred, and returns the rest of span.

(span-span pred span)

(string-span pred string) [SRFI 13]

(span-break pred span)

(string-break pred string) [SRFI 13]

The span-span procedure splits span, returning two values: the longest initial prefix whose characters all satisfy pred, and the remainder of span. The span-break procedure inverts the sense of the predicate: the remainder commences with the first character of span that satisfies pred. In other words: span-span finds the intial span of characters satisfying pred, and span-break breaks span at the first character satisfying pred.

The name "span-span" is unfortunate but unavoidable.

(span-contains haystack needle)

(string-contains haystack needle) [SRFI 13]

It is an error if needle and haystack are not both spans. Returns a cursor referring to haystack indicating the first position in which the characters of needle appear. If there is no such position, #f is returned.

For compatibility with SRFI 13, string-contains returns an index rather than a cursor.

The whole character span or string

(span-length span)

(string-length string) [R7RS-small]

Returns the number of characters in span.

(span-reverse span)

(string-reverse string) [SRFI 13]

Returns a span containing the characters of span in reverse order.

(span-append span ...)

(string-append string ...) [R7RS-small]

Returns a span containing the characters of the spans in order.

(span-concatenate list)

(string-concatenate list) [SRFI 13]

Returns a span containing the characters of the spans and/or strings enumerated in list in order. This procedure will succeed even if (apply string-append list-of-strings) fails due to an implementation limit on the number of arguments a procedure may receive. For convenience, the elements of list may be spans or strings or both.

(span-concatenate-reverse list)

(string-concatenate-reverse list) [SRFI 13]

The same as span-concatenate, except that list is processed in reverse order. Note that the individual spans and strings are not processed in reverse order.

This procedure is useful in the construction of procedures that accumulate character data into lists of string buffers, and wish to convert the accumulated data into a single string when done.

Folding and mapping

(span-map proc span ...)

(string-map proc string ...) [R7RS-small]

It is an error if proc does not accept as many arguments as there are spans and return a single character.

Applies proc element-wise to the characters of the spans and returns a span of the results, in order. If more than one span is given and not all spans have the same length, span-map terminates when the shortest span runs out. The dynamic order in which proc is applied to the characters of the spans is unspecified. If multiple returns occur from span-map, the values returned by earlier returns are not mutated.

(span-for-each proc span ...)

(string-for-each proc string ...) [R7RS-small]

It is an error if proc does not accept as many arguments as there are strings.

The arguments to span-for-each are like the arguments to span-map, but it calls proc for its side effects rather than for its values. Unlike span-map, span-for-each is guaranteed to call proc exactly once on each of the characters of the spans in order from the first character(s) to the last. The value returned by string-for-each is unspecified. If more than one span is given and not all spans have the same length, span-for-each terminates when the shortest string runs out.

(span-fold proc nil span)

(string-fold proc nil string) [SRFI 13]

(span-fold-right proc nil span)

(string-fold-right proc nil string) [SRFI 13]

Invokes proc on each member of span in forward/reverse order, passing the result of the previous invocation as a second argument. For the first invocation, nil is used as the second argument. Returns the result of the last invocation, or nil if there was no invocation.

Parsing and unparsing

(span-split span [separator [ limit'' ] ])

(span-split span [separator [ limit'' ] ])

Returns a list of the words contained in span. If separator (which is also a character span) is omitted, then the words are separated by one or more whitespace characters (those on which char-whitespace? returns #t). If separator is supplied, it specifies a string to be used as the word separator. The returned list will then have one more item than the number of non-overlapping occurrences of the separator in the string. If separator is an empty span, then the returned list contains a list of the characters in span.

If limit is provided, at most that many splits occur, and the remainder of span is returned as the final element of the list (thus, the result will have at most limit+1 elements). If limit is not specified, then as many splits as possible are made. It is an error if limit is not a positive exact integer.

(span-join list [ delim [ grammar ] ])

(string-join list [ delim [ grammar ] ])

This procedure pastes the elements of list together using delimiter, which is a span. For convenience, list elements are allowed to be strings or spans or both. If delimiter is omitted, it is a single space. The grammar argument is a symbol that determines how the delimiter is used, and defaults to infix. The following values are understood:

Filtering and partitioning

(span-filter pred span) [SRFI 13]

(string-filter pred string)

Returns a span containing the characters of span which satisfy pred.

(span-remove pred span)

(string-remove pred string)

Returns a span containing the characters of span which do not satisfy pred.

Compatibility note: The SRFI 13 variant of this procedure is called string-delete, which is inconsistent with the conventions of SRFI 1 and other SRFIs.

(span-partition pred span)

(string-partition pred string)

Returns two values, a span containing the characters of span which satisfy pred and another span containing those which do not.

Copying and conversion

(span-copy span)

(string-copy string [ start [ end ] ]) [R7RS-small]

Makes a copy of span such that any future mutation of any string underlying span does not affect the characters of span. The start and end arguments are for compatibility with R7RS-small.

(span->list span)

(string->list string [ start [ end ] ]) [R7RS-small]

(span->vector span)

(string->vector string[ start [ end ] ]) [R7RS-small]

Returns a newly allocated list/vector containing the characters of span in order. The start and end arguments are for compatibility with R7RS-small.

(list->string list) [R7RS-small]

(vector->string vector [ start [ end ] ]) [R7RS-small]

Returns a newly allocated string whose characters are the elements of list/vector in order. It is an error if the elements are not characters. The start and end arguments are for compatibility with R7RS-small.

(reverse-list->string list) [SRFI 13]

Returns a newly allocated string whose characters are the elements of list in reverse order. It is an error if the elements are not characters.

String cursors

(span-cursor-start span)

(string-cursor-start string)

Returns a cursor referring to the first character in span.

(span-cursor-end span)

(string-cursor-end string)

Returns a cursor referring to the position following the last character in span.

(span-cursor-ref span cursor)

(string-cursor-ref string cursor)

Returns the character referred to by cursor. It is an error if cursor does not refer to a character in span.

(span-cursor-next span cursor)

(string-cursor-next string cursor)

Returns the cursor following cursor. It is an error if cursor does not refer to a character in span.

(span-cursor-prev span cursor)

(string-cursor-prev string cursor)

Returns the cursor following cursor. It is an error if cursor does not refer either to a character in span or the position following the last character in span.

(span-cursor-forward span cursor n)

(string-cursor-forward string cursor n)

(span-cursor-backward span cursor n)

(string-cursor-backward string cursor n)

Iterates span-cursor-next or span-cursor-prev n times.

(span-cursor-forward-until span cursor n)

(string-cursor-forward-until string cursor n)

Iterates span-cursor-next until it refers to a character that satisfies pred or the position following the last character of span is reached, and returns that cursor.

(span-cursor-backward-until span cursor n)

(string-cursor-backward-until string cursor n)

Iterates span-cursor-prev until it refers to a character that satisfies pred or the position preceding the first character of span is reached, and returns that cursor.

(span-cursor=? span cursor1 cursor2)

(string-cursor=? string cursor1 cursor2)

(span-cursor<? span cursor1 cursor2)

(string-cursor<? string cursor1 cursor2)

(span-cursor>? span cursor1 cursor2)

(string-cursor>? string cursor1 cursor2)

(span-cursor<=? span cursor1 cursor2)

(string-cursor<=? string cursor1 cursor2)

(span-cursor>=? span cursor1 cursor2)

(string-cursor>=? string cursor1 cursor2)

Compare cursor1 and cursor2. It is an error if the cursors do not refer to positions in span.

(span-cursor->index span cursor)

(string-cursor->index string cursor)

Return the character index into span corresponding to cursor. It is an error if cursor does not refer to a position in span.

(span-index->cursor span index)

(string-index->cursor string index)

Return the cursor referring to span that corresponds to cursor. It is an error if index is less than zero or greater than the length of span.

(span-cursor-difference span cursor1 cursor2)

(string-cursor-difference string cursor1 cursor2)

Return the difference in characters between cursor2 and cursor1. It is an error if the cursors do not refer to positions in span.

Output

(write-string-tree obj [ port ])

It is an error if port is not a textual output port. If port is omitted, the value of (current-output-port) is used.

If obj is a string or character span, its characters are output to port. If obj is a character, it is output to port. If obj is a number, it is converted to a string as if by number->string and the characters of the string are output to port. If obj is a pair or vector, its components are processed recursively by write-string-tree. Otherwise, write-string-tree does nothing.

(tree->span obj)

(tree->string obj)

Behaves as if write-string-tree were applied to obj and a newly allocated string output port. When obj has been completely output, the port's string is returned as a span or a string.

Compatibility

(span-upcase span)

(string-upcase sv) [R7RS-small]

(span-downcase span)

(string-downcase string) [R7RS-small]

(span-foldcase span)

(string-foldcase string) [R7RS-small]

For the behavior of the string procedures, see R7RS-small. In any implementation of this proposal based on R7RS, the span procedures must behave analogously to the string procedures. That is, if a call to string procedure x on a string containing characters y0 ... yn produces a string containing characters z0 ... zn, then a call to the analogous span procedure x′ on a span containing characters y0 ... yn must produce a span containing characters z0 ... zn.

(span=? span1 span2 span ...)

(string=? string1 string2 string ...) [R7RS-small]

(span<? span1 span2 span ...)

(string<? string1 string2 string ...) [R7RS-small]

(span>? span1 span2 span ...)

(string>? string1 string2 string ...) [R7RS-small]

(span<=? span1 span2 span ...)

(string<=? string1 string2 string ...) [R7RS-small]

(span>=? span1 span2 span ...)

(string>=? string1 string2 string ...) [R7RS-small]

(span-ci=? span1 span2 span ...)

(string-ci=? string1 string2 string ...) [R7RS-small]

(span-ci<? span1 span2 span ...)

(string-ci<? string1 string2 string ...) [R7RS-small]

(span-ci>? span1 span2 span ...)

(string-ci>? string1 string2 string ...) [R7RS-small]

(span-ci<=? span1 span2 span ...)

(string-ci<=? string1 string2 string ...) [R7RS-small]

(span-ci>=? span1 span2 span ...)

(string-ci>=? string1 string2 sv ...) [R7RS-small]

For the behavior of the string procedures, see R7RS-small. In any implementation of this proposal based on R7RS, the span procedures must behave analogously to the string procedures.

(span-titlecase ''span'')`

(string-titlecase string)[SRFI 13]

For every character c in span: if c is preceded by a character with case, it is downcased; otherwise it is replaced by its titlecase equivalent, if any. Other characters are unchanged. Note that most lowercase characters have the same character as both uppercase and titlecase equivalents.

Examples:

(string-titlecase "--capitalize tHIS sentence.") => "--Capitalize This Sentence." (string-titlecase "see Spot run. see Nix run.") => "See Spot Run. See Nix Run." (string-titlecase "3com makes routers.") => "3Com Makes Routers."

Comparator

span-comparator

This is a SRFI 114 comparator for comparing strings. Its procedures behave as if their arguments are converted to strings and then passed to the procedures of string-comparator.

Sample Implementation

The sample implementation (which is not yet written) represents spans as records containing a string and two string cursors, and provides two implementations of string cursors, one using string indexes directly and one that layers UTF-8 character spans on top of single-byte native strings.