(ql:quickload "str")
also on Ultralisp.
Why ?
-
modernity, simplicity and discoverability:
(str:trim s)instead of(string-trim '(#\Backspace #\Tab #\Linefeed #\Newline #\Vt #\Page #\Return #\Space #\Rubout #\Next-Line #\No-break_space) s)), orstr:concat stringsinstead of an unusualformatconstruct; one discoverable library instead of many;
-
consistence and composability, where
sis always the last argument, which makes it easier to feed pipes and arrows. -
fixing built-in surprises:
(string-downcase nil) =>"nil"the string, whereas(str:downcase nil)=>nil.
The only dependency is cl-ppcre.
Table of Contents
- A modern and consistent Common Lisp string manipulation library
- Install
- Global parameters
- Functions
- Tweak whitespace
- To longer strings
- join
(separator list-of-strings) - concat
(&rest strings) - ensure
(s &key wrapped-in prefix suffix)NEW in March, 2023 - ensure-prefix, ensure-suffix
(start/end s)NEW in March, 2023 - ensure-wrapped-in
(start/end s) - insert
(string/char index s) - repeat
(count s) - add-prefix, add-suffix
(items s) - pad
(len s &key (pad-side :right) (pad-char #\Space)), pad-left, pad-right, pad-center (new in 0.16, 2019/12)
- join
- To shorter strings
- To a fixed length
- To and from lists
- To and from files
- Predicates
- Case
- Others
- Macros
- Changelog
- Dev and test
- See also
Install with Quicklisp:
(ql:quickload :str)
Add it in your .asd's project dependencies, and call functions with the str prefix. It is not recommended to :use :str in a package. It's safer to use the str prefix.
Check its version:
(str:version)
To get a newer version, you need to update the Quicklisp dist (think of QL as Debian's apt rather than pip/npm/etc):
(ql:update-dist "quicklisp")
Don't have a full Common Lisp development environment yet ? Get Portacle, a portable and multiplatform development environment shipping Emacs, Quicklisp, SBCL and Git. See also editor support (Vim, Lem, Atom, Eclipse,…).
Some parameters are common to various functions and often used:
:ignore-case and :omit-nulls.
Consequently we can also manage them with global parameters:
(let ((str:*ignore-case* t))
(str:ends-with-p "BAR" "foobar"))is equivalent to
(str:ends-with-p "BAR" "foobar" :ignore-case t)Removes all characters in char-bag (default: whitespaces) at the beginning and end of s.
If supplied, char-bag has to be a sequence (e.g. string or list of characters).
(str:trim " rst ") ;; => "rst"
(str:trim "+-*foo-bar*-+" :char-bag "+-*") => "foo-bar"
(str:trim "afood" :char-bag (concat "a" "d")) => "foo""
(str:trim "cdoooh" :char-bag (str:concat "c" "d" "h")) => "ooo"Also trim-left and trim-right.
Uses the built-in
string-trim
where whitespaces are '(#\Space #\Newline #\Backspace #\Tab #\Linefeed #\Page #\Return #\Rubout).
Ensure there is only one space character between words. Remove newlines.
(str:collapse-whitespaces "foo bar
baz")
;; "foo bar baz"
;;TJoin strings in list list-of-strings with separator (either a string or a char) in between.
(join " " '("foo" "bar" "baz")) ;; => "foo bar baz"
(join #\Space '("foo" "bar" "baz")) ;; => "foo bar baz"Join strings into one.
(concat "f" "o" "o") ;; => "foo"Simple call of the built-in concatenate.
We actually also have uiop:strcat.
The "ensure-" functions return a string that has the specified prefix or suffix, appended if necessary.
This str:ensure function looks for the following key parameters, in order:
:wrapped-in: if non nil, callstr:ensure-wrapped-in. This checks thatsboth starts and ends with the supplied string or character.:prefixand:suffix: if both are supplied and non-nil, callstr:ensure-suffixfollowed bystr:ensure-prefix.:prefix: callstr:ensure-prefix:suffix: callstr:ensure-suffix.
Example:
(str:ensure "abc" :wrapped-in "/") ;; => "/abc/"
(str:ensure "/abc" :prefix "/") ;; => "/abc" => no change, still one "/"
(str:ensure "/abc" :suffix "/") ;; => "/abc/" => added a "/" suffix.These functions accept strings and characters:
(str:ensure "/abc" :prefix #\/)warn: if both :wrapped-in and :prefix (and/or :suffix) are supplied together, :wrapped-in takes precedence and :prefix (and/or :suffix) is ignored.
Ensure that s starts with start/end (or ends with start/end, respectively).
Return a new string with its prefix (or suffix) added, if necessary.
Example:
(str:ensure-prefix "/" "abc/") => "/abc/" (a prefix was added)
;; and
(str:ensure-prefix "/" "/abc/") => "/abc/" (does nothing)Ensure that s both starts and ends with start/end.
Return a new string with the necessary added bits, if required.
It simply calls str:ensure-suffix followed by str:ensure-prefix.
See also str:wrapped-in-p and uiop:string-enclosed-p prefix s suffix.
(str:ensure-wrapped-in "/" "abc") ;; => "/abc/" (added both a prefix and a suffix)
(str:ensure-wrapped-in "/" "/abc/") ;; => "/abc/" (does nothing)Insert the given string (or character) at the index index into s and return a
new string.
If index is out of bounds, just return s.
(str:insert "l" 2 "helo") ; => "hello"
(str:insert "o" 99 "hell") : => "hell"Make a string of s repeated count times.
(repeat 3 "foo") ;; => "foofoofoo"Respectively prepend or append s to the front of each item.
pad (len s &key (pad-side :right) (pad-char #\Space)), pad-left, pad-right, pad-center (new in 0.16, 2019/12)
Fill s with characters until it is of the given length. By default,
add spaces on the right:
(str:pad 10 "foo")
"foo "pad-side: one of:right(the default),:leftor:center. See*pad-side*.pad-char: the padding character (or string of one character). Defaults to a space. See*pad-char*.
(str:pad 10 "foo" :pad-side :center :pad-char "+")
"+++foo++++"If the given length is smaller than the length o s, return s.
Filling with spaces can easily be done with format:
(format nil "~va" len s) ;; => "foo "
(format nil "~v@a" 10 "foo") ;; => " foo" (with @)Return the substring of s from start to end.
It uses subseq with differences:
- argument order, s at the end
startandendcan be lower than 0 or bigger than the length of s.- for convenience
endcan be nil or t to denote the end of the string.
Examples:
(is "abcd" (substring 0 t "abcd") "t denotes the end of the string")
(is "abcd" (substring 0 nil "abcd") "nil too")
(is "abcd" (substring 0 100 "abcd") "end can be too large")
(is "abc" (substring 0 -1 "abcd") "end can be negative. Counts from the end.")
(is "" (substring 0 -100 "abcd") "end can be negative and too low")
(is "" (substring 100 1 "abcd") "start can be too big")
(is "abcd" (substring -100 4 "abcd") "start can also be too low")
(is "" (substring 2 1 "abcd") "start is bigger than end")Return the first letter of s.
Examples:
(s-first "foobar") ;; => "f"
(s-first "") ;; => ""Return the last letter of s.
Return the rest substring of s.
Examples:
(s-rest "foobar") ;; => "oobar"
(s-rest "") ;; => ""Return the nth letter of s.
Examples:
(s-nth 3 "foobar") ;; => "b"
(s-nth 3 "") ;; => ""You could also use
(elt "test" 1)
;; => #\e
(string (elt "test" 1))
;; => "e"If s is longer than len, truncate it and add an ellipsis at the
end (... by default). s is cut down to len minus the length of
the ellipsis (3 by default).
Optionally, give an :ellipsis keyword argument. Also set it globally
with *ellipsis*.
(shorten 8 "hello world")
;; => "hello..."
(shorten 3 "hello world")
;; => "..."
(shorten 8 "hello world" :ellipsis "-")
;; => "hello w-"
(let ((*ellipsis* "-"))
(shorten 8 "hello world"))
;; => "hello w-"Fit this string to the given length:
- if it's too long, shorten it (showing the
ellipsis), - if it's too short, add paddding (to the side
pad-side, adding the characterpad-char).
As such, it accepts the same key arguments as str:shorten and
str:pad: ellipsis, pad-side, pad-char…
CL-USER> (str:fit 10 "hello" :pad-char "+")
"hello+++++"
CL-USER> (str:fit 10 "hello world" :ellipsis "…")
"hello wor…"If, like me, you want to print a list of data as a table, see:
CL-USER> (ql:quickload "cl-ansi-term")
CL-USER> (term:table '(("name" "age" "email")
("me" 7 "some@blah")
("me" 7 "[email protected]"))
:column-width '(10 4 20))
+---------+---+-------------------+
|name |age|email |
+---------+---+-------------------+
|me |7 |some@blah |
+---------+---+-------------------+
|me |7 |some@with-some-l(…)|
+---------+---+-------------------+CL-USER> (ql:quickload "cl-ascii-table")
CL-USER> (let ((table (ascii-table:make-table '("Id" "Name" "Amount") :header "Infos")))
(ascii-table:add-row table '(1 "Bob" 150))
(ascii-table:add-row table '(2 "Joe" 200))
(ascii-table:add-separator table)
(ascii-table:add-row table '("" "Total" 350))
(ascii-table:display table))
.---------------------.
| Infos |
+----+-------+--------+
| Id | Name | Amount |
+----+-------+--------+
| 1 | Bob | 150 |
| 2 | Joe | 200 |
+----+-------+--------+
| | Total | 350 |
+----+-------+--------+
NILReturn list of words, which were delimited by whitespace.
Join the list of strings with a whitespace.
Split string by newline character and return list of lines.
A terminal newline character does not result in an extra empty string (new in v0.14, october 2019).
Join the list of strings with a newline character.
Split into subtrings. If
omit-nulls is non-nil, zero-length substrings are omitted.
By default, metacharacters are treated as normal characters.
If regex is not nil, then separator is treated as regular expression.
(split "+" "foo++bar") ;; => ("foo" "" "bar")
(split #\+ "foo++bar") ;; => ("foo" "" "bar")
(split "+" "foo++bar" :omit-nulls t) ;; => ("foo" "bar")
(split "[,|;]" "foo,bar;baz") ;; => ("foo,bar;baz")
(split "[,|;]" "foo,bar;baz" :regex t) ;; => ("foo" "bar" "baz")cl-ppcre has an inconsistency such that when the separator appears at the end, it doesn't return a trailing empty string. But we do since v0.14 (october, 2019).
Similar to split, but split from the end. In particular, this will
be different from split when a :limit is provided, but in more
obscure cases it can be different when there are multiple different
ways to split the string.
(rsplit "/" "/var/log/mail.log" :limit 2) ;; => ("/var/log" "mail.log")(cl-ppcre:split " " "a b c ")
("a" "b" "c")
(str:split " " "a b c ")
("a" "b" "c" "")Because it is a common pattern and it can be clearer than an option coming after many parenthesis.
Read the file and return its content as a string.
Example: (str:from-file "path/to/file.txt").
:external-format: if nil, the system default. Can be bound to :utf-8.
But you might just call
uiop's uiop:read-file-string
directly.
There is also uiop:read-file-lines.
Write the string s to the file filename. If the file does not
exist, create it, if it already exists, replace it.
Options:
:if-does-not-exist::create(default),:error:if-exists::supersede(default),:append,:overwrite,:rename,:error,...
Returns the string written to file.
True if s is nil or the empty string:
(emptyp nil) ;; => T
(emptyp "") ;; => T
(emptyp " ") ;; => NILSee also str:non-empty-string-p, which adds a stringp check.
True if s is empty or only contains whitespaces.
(blankp "") ;; => T
(blankp " ") ;; => T
(emptyp " ") ;; => NIL
See also str:non-blank-string-p.
True if s starts with the substring start, nil otherwise. Ignore
case by default.
(starts-with-p "foo" "foobar") ;; => T
(starts-with-p "FOO" "foobar") ;; => NIL
(starts-with-p "FOO" "foobar" :ignore-case t) ;; => T
Calls string= or string-equal depending on the case, with their
:start and :end delimiters.
True if s ends with the substring end. Ignore case by default.
(ends-with-p "bar" "foobar") ;; => T
end can be a string or a character.
Return true if s contains substring, nil otherwise. Ignore the
case with :ignore-case t (don't ignore by default).
Based on a simple call to the built-in search (which returns the
position of the substring).
Return T if s' is a member of list'. Do not ignore case by default.
NOTE: s-member's arguments' order is the reverse of CL's member.
If :ignore-case or *ignore-case* are not nil, ignore case (using
string-equal instead of string=).
Unlike CL's member, s-member returns T or NIL, instead of the tail of LIST whose first element satisfies the test.
Return s if all items start (or end) with it.
See also uiop:string-prefix-p prefix s, which returns t if
prefix is a prefix of s,
and uiop:string-enclosed-p prefix s suffix, which returns t if s
begins with prefix and ends with suffix.
Does s start and end with `start/end'?
If true, return s. Otherwise, return nil.
Example:
(str:wrapped-in-p "/" "/foo/" ;; => "/foo/"
(str:wrapped-in-p "/" "/foo" ;; => nilSee also: UIOP:STRING-ENCLOSED-P (prefix s suffix).
We use
cl-change-case (go
thank him and star the repo!).
We adapt these functions to also accept symbols and characters (like the inbuilt casing functions).
Also the functions return nil when argument is nil.
The available functions are:
:no-case (s &key replacement)
:camel-case (s &key merge-numbers)
:dot-case
:header-case
:param-case
:pascal-case
:path-case
:sentence-case
:snake-case
:swap-case
:title-case
:constant-case
More documentation and examples are there.
The functions str:downcase, str:upcase and str:capitalize return
a new string. They call the built-in string-downcase,
string-upcase and string-capitalize respectively, but they fix
something surprising. When the argument is nil, the built-ins return
"nil" or "NIL" or "Nil", a string. Indeed, they work on anything:
(string-downcase nil) ;; => "nil" the string !
(str:downcase nil) ;; nil
(string-downcase :FOO) ;; => "foo"
These functions return t if the given string contains at least one
letter and all its letters are lowercase or uppercase, respectively.
(is (downcasep " a+,. ") t "downcasep with one letter and punctuation is true.")
(is (downcasep " +,. ") nil "downcasep with only punctuation or spaces is false")alphap returns t if s contains at least one character and all characters are
alpha (as in "^[a-zA-Z]+$").
lettersp works for unicode letters too.
(is (alphap "abcdeé") nil "alphap is nil with accents")
(is (lettersp "éß") t "lettersp is t with accents and ß")alphanump returns t if s contains at least one character and all characters are alphanumeric (as in ^[a-zA-Z0-9]+$).
lettersnump also works on unicode letters (as in ^[\\p{L}a-zA-Z0-9]+$).
Return t if the character / string is an ASCII character / is composed of ASCII characters.
An ASCII character has a char-code inferior to 128.
Returns t if s contains at least one character and all characters are numerical (as for digit-char-p).
Return t if s has at least one alpha, letter, alphanum character (as with alphanumericp).
Replace the first occurence of old by new in s.
By default, metacharacters are treated as normal characters.
If regex is not nil, then old is treated as regular expression.
(replace-first "a" "o" "faa") ;; => "foa"
(replace-first "fo+" "frob" "foofoo bar" :regex t) ;; => "frobfoo bar"Uses cl-ppcre:regex-replace but quotes the user input to not treat it as a regex (if regex is nil).
Replace all occurences of old by new in s.
By default, metacharacters are treated as normal characters.
If regex is not nil, old is treated as regular expression.
(replace-all "a" "o" "faa") ;; => "foo"
(replace-all "fo+" "frob" "foofoo bar" :regex t) ;; => "frobfrob bar"Uses cl-ppcre:regex-replace-all but quotes the user input to not treat it as a regex (if regex is nil).
If the replacement is only one character, you can use substitute:
(substitute #\+ #\Space "foo bar baz")
;; "foo+bar+baz"
Replace all associations given by pairs in a replacement-list and return a new string.
The replacement-list alternates a string to replace (case sensitive) and its replacement.
By default, metacharacters in the string to replace are treated as normal characters.
If regex is not nil, strings to replace are treated as regular expression.
Example:
(replace-using (list "%phone%" "987")
"call %phone%")
;; => "call 987"
(replace-using (list "fo+" "frob"
"ba+" "Bob")
"foo bar"
:regex t)
;; => "frob Bobr"Remove the punctuation characters from s, replace them with
replacement (defaults to a space) and strip continuous whitespace.
(str:remove-punctuation "I say: - 'Hello, world?'") ;; => "I say Hello world"Use str:no-case to remove punctuation and return the string as lower-case.
(renamed from common-prefix in v0.9)
Find the common prefix between strings.
Example: (str:prefix '(\"foobar\" \"foozz\")) => "foo"
Uses the built-in mismatch, that returns the position at which
the strings fail to match.
Return a string or nil when the input is the void list.
Find the common suffix between strings.
Counts the non-overlapping occurrences of substring in s.
You could also count only the ocurrencies between start and end.
Examples:
(count-substring "abc" "abcxabcxabc")
;; => 3(count-substring "abc" "abcxabcxabc" :start 3 :end 7)
;; => 1Returns the value of a cons cell in alist with key key, when key is a string.
The second return value is the cons cell, if any was matched.
The arguments are in the opposite order of cl:assoc's, but are consistent
with alexandria:assoc-value (and str).
(s-assoc-value '(("hello" . 1)) "hello")
;; 1
;; ("hello" . 1)
(alexandria:assoc-value '(("hello" . 1)) "hello")
;; NIL
(alexandria:assoc-value '(("hello" . 1)) "hello" :test #'string=)
;; 1
;; ("hello" . 1)
(assoc "hello" '(("hello" . 1)))
;; NIL
(assoc "hello" '(("hello" . 1)) :test #'string=)
;; ("hello" . 1)
(cdr *)
;; 1A case-like macro that works with strings (CL case's test function is
eql, and that isn't enough for strings).
Example:
(str:string-case "hello"
("foo" 1)
(("hello" "test") 5)
(nil (print "input is nil"))
(otherwise (print "non of the previous forms was caught.")))You might also like pattern matching. The example below with trivia is very similar:
(trivia:match "hey"
("hey" (print "it matched"))
(otherwise :nothing))Note that there is also http://quickdocs.org/string-case/.
A COND-like macro to match substrings and bind variables to matches. Regular expressions are allowed for matches.
_ is a placeholder that is ignored.
THIS MACRO IS EXPERIMENTAL and might break in future releases.
Example:
(str:match "a 1 b 2 d"
(("a " x " b " y " d") ;; => matched
(+ (parse-integer x) (parse-integer y)))
(t
'default-but-not-for-this-case)) ;; default branch
;; => 3
(str:match "a 1 b c d"
(("a 2 b" _ "d") ;; => not matched
(print "pass"))
(("a " _ " b c d") ;; => matched
"here we go")
(t 'default-but-not-for-this-case)) ;; default branch
;; => "here we go"Match with regexs:
(str:match "123 hello 456"
(("\\d+" s "\\d+")
s)
(t "nothing"))
;; => " hello "- Feb, 2024:
- added the
matchmacro. It is EXPERIMENTAL and might change in future versions. We welcome your bug reports and feedback.
- added the
- 0.21, November, 2023:
- added the
regexkey argument tosplit,rsplit,split-omit-nulls.
- added the
- August, 2023:
- added the
regexkey argument to thereplace-*functions.
- added the
- March, 2023:
- added
str:ensure,str:ensure-prefix,str:ensure-suffix,str:ensure-wrapped-inandstr:wrapped-in-p.
- added
- January, 2023: added the
:char-bargparameter totrim,trim-left,trim-right.- minor:
ends-with-pnow works with a character.
- minor:
- June, 2022: small breaking change: fixed
prefixpwhen used with a smaller prefix: "f" was not recognized as a prefix of "foobar" and "foobuz", only "foo" was. Now it is fixed. Same forsuffixp. - Feb, 2022: added
fit: fit the string to the given length: either shorten it, either padd padding. - 0.20, May, 2021: added
ascii-p. - 0.19.1, May, 2021: speed up
join(by a factor of 4). - 0.19, October, 2020: added s-member *0.18.1, September, 2020: fix replace-all edge case when the replacement string ends with two backslashes and a single quote.
- 0.18, June, 2020: added
replace-using. - 0.17, April 2020:
- added
collapse-whitespaces joinandsplitalso accept a char as separator- fixed
remove-punctuationthat did not respect the case. Useno-casefor this - fixed
from-file"odd number of arguments" error.
- added
- 0.16, November 2019: added
pad,pad-[left, right, center]. - 0.15, October 2019: added functions to change case (based on cl-change-case). added remove-punctuation.
- 0.14, October, 2019: fixed the cl-ppcre inconsistency in
splitandlines. A trailing separator now returns a trailing empty string.
Before:
(str:split " " "a b c ")
("a" "b" "c") ;; like cl-ppcre:splitNow:
(str:split " " "a b c ")
("a" "b" "c" "")- august, 2019: deprecated
prune, renamed toshorten. - added
:limittosplit. - 0.13 june, 2019
- added
insert
- added
- 0.12
- added case predicates (
downcasep,alphap,has-xand friends).
- added case predicates (
- 0.11 (Quicklisp end of march, 2019, also in Ultralisp)
- added
str:downcase,str:upcaseandstr:capitalize, that fix thenilargument surprise.
- added
- 0.10
splitdoesn't fix cl-ppcre's inconsistency anymore (when the separator appears at the end). See issue #18. So(str:split "xx" "fooxxbarxx")doesn't return a trailing"".- added
s-last s-firstand friends returnnilwhen appropriate, not"".
- 0.9
- added
s-first,s-restands-nth - added
prefixandsuffixfunctions and predicates. - added
prune.
- added
- 0.8 added
string-case - 0.7 added
version - 0.6 added
split-omit-nulls(QL, january 2018) - 0.5 added
common-prefix - 0.4 added
from-fileandto-file. - 0.3 added
substring.
Regression testing is implemented with fiveam.
Either use
(asdf:test-system :str)or load the test package str.test and then
(fiveam:run! 'test-str:str) (fiveam:run! 'test-str:replace-functions)Test suite names:
- replace-functions
- lengthen-functions
- ensure-functions
- pad-functions
- substring-functions
- list-functions
- from-list-to-string
- from-list-to-list
- from-string-to-list
- predicates, case-functions
- miscellaneous
(fiveam:run! 'test-str::downcase) ;; (test symbols are unexported)First you need to
(setf fiveam:*run-test-when-defined* t)then the test is run after each definition / compilation. This can be done with C-c C-c on emacs.
- the Common Lisp Cookbook, strings page.
- my Common Lisp course on Udemy: from novice to effective developer. Check out my blog for regular coupons.
- https://lisp-journey.gitlab.io/
- video: how to create a Common Lisp project from scratch with our project generator: it sums up in 5 minutes what took me a much longer time to gather.
Inspired by the famous Emacs Lisp's s.el.
