Tokenizing A String
From Rosetta Code
Programming Task
This is a programming task. It lays out a problem which Rosetta Code users are encouraged to solve, using languages they know.
Separate the string "Hello,How,Are,You,Today" by commas into an array (or list) so that each element of it stores a different word. Display the words to the 'user', in the simplest manner possible, separated by a period. To simplify, you may display a trailing period.
[edit] ActionScript
var hello:String = "Hello,How,Are,You,Today";
var tokens:Array = hello.split(",");
trace(tokens.join("."));
// Or as a one-liner
trace("Hello,How,Are,You,Today".split(",").join("."));
[edit] Ada
with Ada.Strings.Fixed; use Ada.Strings.Fixed;
with Ada.Text_Io; use Ada.Text_Io;
procedure Parse_Commas is
Source_String : String := "Hello,How,Are,You,Today";
Index_List : array(1..256) of Natural;
Next_Index : Natural := 1;
begin
Index_List(Next_Index) := 1;
while Index_List(Next_Index) < Source_String'Last loop
Next_Index := Next_Index + 1;
Index_List(Next_Index) := 1 + Index(Source_String(Index_List(Next_Index - 1)..Source_String'Last), ",");
if Index_List(Next_Index) = 1 then
Index_List(Next_Index) := Source_String'Last + 2;
end if;
Put(Source_String(Index_List(Next_Index - 1)..Index_List(Next_Index)-2) & ".");
end loop;
end Parse_Commas;
[edit] ALGOL 68
main:(
OP +:= = (REF FLEX[]STRING in out, STRING item)VOID:(
[LWB in out: UPB in out+1]STRING new;
new[LWB in out: UPB in out]:=in out;
new[UPB new]:=item;
in out := new
);
PROC string split = (REF STRING beetles, STRING substr)[]STRING:(
""" Split beetles where substr is found """;
FLEX[1:0]STRING out;
INT start := 1, pos;
WHILE string in string(substr, pos, beetles[start:]) DO
out +:= STRING(beetles[start:start+pos-2]);
start +:= pos + UPB substr - 1
OD;
IF start > LWB beetles THEN
out +:= STRING(beetles[start:])
FI;
out
);
PROC char split = (REF STRING beetles, STRING chars)[]STRING: (
""" Split beetles where character is found in chars """;
FLEX[1:0]STRING out;
FILE beetlef;
associate(beetlef, beetles); # associate a FILE handle with a STRING #
make term(beetlef, chars); # make term: assign CSV string terminator #
PROC raise logical file end = (REF FILE f)BOOL: except logical file end;
on logical file end(beetlef, raise logical file end);
STRING solo;
DO
getf(beetlef, ($g$, solo));
out+:=solo;
getf(beetlef, ($x$)) # skip CHAR separator #
OD;
except logical file end:
SKIP;
out
);
STRING beetles := "John Lennon, Paul McCartney, George Harrison, Ringo Starr";
printf(($g"."$, string split(beetles, ", "),$l$));
printf(($g"."$, char split(beetles, ", "),$l$))
)
Output:John Lennon.Paul McCartney.George Harrison.Ringo Starr. John.Lennon..Paul.McCartney..George.Harrison..Ringo.Starr.
[edit] AutoHotkey
string := "Hello,How,Are,You,Today"
stringsplit, string, string, `,
loop, % string0
{
msgbox % string%A_Index%
}
[edit] AWK
BEGIN {
s = "Hello,How,Are,You,Today"
split(s, arr, ",")
for(i=1; i < length(arr); i++) {
printf arr[i] "."
}
print
}
A more idiomatic way for AWK is
BEGIN { FS = "," }
{
for(i=1; i <= NF; i++) printf $i ".";
print ""
}
which "tokenize" each line of input and this is achieved by using "," as field separator
[edit] C
Works with: ANSI C Works with: gcc version 3.3.3 Library: POSIX
This example uses the strtok() function to separate the tokens. This function is destructive (replacing token separators with '\0'), so we have to make a copy of the string (using strdup()) before tokenizing. strdup() is not part of ANSI C, but is available on most platforms. It can easily be implemented with a combination of strlen(), malloc(), and strcpy().
#include<string.h>
#include<stdio.h>
#include<stdlib.h>
int main(void)
{
char *a[5];
const char *s="Hello,How,Are,You,Today";
int n=0, nn;
char *ds=strdup(s);
a[n]=strtok(ds, ",");
while(a[n] && n<4) a[++n]=strtok(NULL, ",");
for(nn=0; nn<=n; ++nn) printf("%s.", a[nn]);
putchar('\n');
free(ds);
return 0;
}
[edit] C#
string str = "Hello,How,Are,You,Today";
// or Regex.Split ( "Hello,How,Are,You,Today", "," );
// (Regex is in System.Text.RegularExpressions namespace
string[] strings = str.Split(',');
foreach (string s in strings)
{
Console.WriteLine (s + ".");
}
[edit] C++
Works with: ANSI C++ Works with: g++ version 3.4.4 (cygming special) Library: STL
This is not the most efficient method as it involves redundant copies in the background, but it is very easy to use. In most cases it will be a good choice as long as it is not used as an inner loop in a performance critical system.
Note doxygen tags in comments before function, describing details of interface.
#include <string>
#include <vector>
/// \brief convert input string into vector of string tokens
///
/// \note consecutive delimiters will be treated as single delimiter
/// \note delimiters are _not_ included in return data
///
/// \param input string to be parsed
/// \param delims list of delimiters.
std::vector<std::string> tokenize_str(const std::string & str,
const std::string & delims=", \t")
{
using namespace std;
// Skip delims at beginning, find start of first token
string::size_type lastPos = str.find_first_not_of(delims, 0);
// Find next delimiter @ end of token
string::size_type pos = str.find_first_of(delims, lastPos);
// output vector
vector<string> tokens;
while (string::npos != pos || string::npos != lastPos)
{
// Found a token, add it to the vector.
tokens.push_back(str.substr(lastPos, pos - lastPos));
// Skip delims. Note the "not_of". this is beginning of token
lastPos = str.find_first_not_of(delims, pos);
// Find next delimiter at end of token.
pos = str.find_first_of(delims, lastPos);
}
return tokens;
}
here is sample usage code:
#include <iostream>
int main() {
using namespace std;
string s("Hello,How,Are,You,Today");
vector<string> v(tokenize_str(s));
for (unsigned i = 0; i < v.size(); i++)
cout << v[i] << ".";
cout << endl;
return 0;
}
[edit] Clojure
(apply str (interpose "." (seq (.split #"," "Hello,How,Are,You,Today"))))
[edit] D
writefln( "Hello,How,Are,You,Today".split(",").join(".") );
[edit] E
".".rjoin("Hello,How,Are,You,Today".split(","))
[edit] Erlang
-module(tok).
-export([start/0]).
start() ->
Lst = string:tokens("Hello,How,Are,You,Today",","),
io:fwrite("~s~n", [string:join(Lst,".")]),
ok.
[edit] Forth
There is no standard string split routine, but it is easily written. The results are saved temporarily to the dictionary.
: split ( str len separator len -- tokens count )
here >r 2swap
begin
2dup 2, \ save this token ( addr len )
2over search \ find next separator
while
dup negate here 2 cells - +! \ adjust last token length
2over nip /string \ start next search past separator
repeat
2drop 2drop
r> here over - ( tokens length )
dup negate allot \ reclaim dictionary
2 cells / ; \ turn byte length into token count
: .tokens ( tokens count -- )
1 ?do dup 2@ type ." ." cell+ cell+ loop 2@ type ;
s" Hello,How,Are,You,Today" s" ," split .tokens \ Hello.How.Are.You.Today
[edit] Fortran
Works with: Fortran version 90 and later
PROGRAM Example
CHARACTER(23) :: str = "Hello,How,Are,You,Today"
CHARACTER(5) :: word(5)
INTEGER :: pos1 = 1, pos2, n = 0, i
DO
pos2 = INDEX(str(pos1:), ",")
IF (pos2 == 0) THEN
n = n + 1
word(n) = str(pos1:)
EXIT
END IF
n = n + 1
word(n) = str(pos1:pos1+pos2-2)
pos1 = pos2+pos1
END DO
DO i = 1, n
WRITE(*,"(2A)", ADVANCE="NO") TRIM(word(i)), "."
END DO
END PROGRAM Example
[edit] Haskell
The necessary operations are unfortunately not in the standard library (yet), but simple to write:
splitBy :: (a -> Bool) -> [a] -> <nowiki>[[a]]</nowiki>
splitBy _ [] = []
splitBy f list = first : splitBy f (dropWhile f rest) where
(first, rest) = break f list
splitRegex :: Regex -> String -> [String]
joinWith :: [a] -> <nowiki>[[a]]</nowiki> -> [a]
joinWith d xs = concat $ List.intersperse d xs
-- "concat $ intersperse" can be replaced with "intercalate" from the Data.List in GHC 6.8 and later
putStrLn $ joinWith "." $ splitBy (== ',') $ "Hello,How,Are,You,Today"
-- using regular expression to split:
import Text.Regex
putStrLn $ joinWith "." $ splitRegex (mkRegex ',') $ "Hello,How,Are,You,Today"
[edit] Groovy
println 'Hello,How,Are,You,Today'.split(',').join('.')
[edit] Io
"Hello,How,Are,You,Today" split(",") join(".") println
[edit] J
s=: 'Hello,How,Are,You,Today' ] t=: <;._1 ',',s +-----+---+---+---+-----+ |Hello|How|Are|You|Today| +-----+---+---+---+-----+ ; t,&.>'.' Hello.How.Are.You.Today. '.' (I.','=s)}s NB. two steps combined Hello.How.Are.You.Today
[edit] Java
Works with: Java version 1.0+
There are multiple ways to tokenize a String in Java. The first is by splitting the String into an array of Strings, and the other way is to use StringTokenizer with a delimiter. The second way given here will skip any empty tokens. So if two commas are given in line, there will be an empty string in the array given by the split function, but no empty string with the StringTokenizer object.
String toTokenize = "Hello,How,Are,You,Today";
//First way
String word[] = toTokenize.split(",");
for(int i=0; i<word.length; i++) {
System.out.print(word[i] + ".");
}
//Second way
StringTokenizer tokenizer = new StringTokenizer(toTokenize, ",");
while(tokenizer.hasMoreTokens()) {
System.out.print(tokenizer.nextToken() + ".");
}
[edit] JavaScript
Works with: Firefox version 2.0
alert( "Hello,How,Are,You,Today".split(",").join(".") );
[edit] Logo
Works with: UCB Logo
to split :str :sep output parse map [ifelse ? = :sep ["| |] [?]] :str end
? show split "Hello,How,Are,You,Today ", [Hello How Are You Today]
[edit] MAXScript
output = ""
for word in (filterString "Hello,How,Are,You,Today" ",") do
(
output += (word + ".")
)
format "%\n" output
[edit] Modula-3
MODULE Tokenize EXPORTS Main;
IMPORT IO, TextConv;
TYPE Texts = REF ARRAY OF TEXT;
VAR tokens: Texts;
string := "Hello,How,Are,You,Today";
sep := SET OF CHAR {','};
BEGIN
tokens := NEW(Texts, TextConv.ExplodedSize(string, sep));
TextConv.Explode(string, tokens^, sep);
FOR i := FIRST(tokens^) TO LAST(tokens^) DO
IO.Put(tokens[i] & ".");
END;
IO.Put("\n");
END Tokenize.
[edit] Objective-C
Works with: GNUstep Works with: Cocoa
NSString *text = @"Hello,How,Are,You,Today";
NSArray *tokens = [text componentsSeparatedByString:@","];
NSString *result = [tokens componentsJoinedByString:@"."];
NSLog(result);
[edit] OCaml
To split on a single-character separator:
let rec split_char sep str =
try
let i = String.index str sep in
String.sub str 0 i ::
split_char sep (String.sub str (i+1) (String.length str - i - 1))
with Not_found ->
[str]
Or the tail-recursive equivalent:
let split_char sep str =
let rec aux acc str =
try
let i = String.index str sep in
let this = String.sub str 0 i
and next = String.sub str (i+1) (String.length str - i - 1) in
aux (this::acc) next
with Not_found ->
List.rev(str::acc)
in
aux [] str
;;
But both of these will process extraneous String.sub (so one string alloc). For N tokens there will be (N - 2) unneeded allocs. To resolve this here is a version which first gets the indices, and then extracts the tokens:
let split_char sep str =
let rec indices acc i =
try
let i = succ(String.index_from str i sep) in
indices (i::acc) i
with Not_found ->
(String.length str + 1) :: acc
in
let is = indices [0] 0 in
let rec aux acc = function
| last::start::tl ->
let w = String.sub str start (last-start-1) in
aux (w::acc) (start::tl)
| _ -> acc
in
aux [] is
Splitting on a string separator using the regular expressions library:
#load "str.cma";;
let split_str sep str =
Str.split (Str.regexp_string sep) str
There is already a library function for joining:
String.concat sep strings
[edit] Perl
Works with: Perl version 5.X
As a one liner without a trailing period, and most efficient way of doing it as you don't have to define an array.
print join('.', split(/,/, "Hello,How,Are,You,Today"));
If you needed to keep an array for later use, again no trailing period
my @words = split(/,/, "Hello,How,Are,You,Today");
print join('.', @words);
If you really want a trailing period, here is an example
my @words = split(/,/, "Hello,How,Are,You,Today");
print $_.'.' for (@words);
[edit] PHP
Works with: PHP version 5.x
<?php
$str = 'Hello,How,Are,You,Today';
echo implode('.', explode(',', $str));
?>
[edit] Pop11
The natural solution in Pop11 uses lists.
There are built in libraries for tokenising strings, illustrated below, along with code that the user could create for the task.
First show the use of sysparse_string to break up a string and make a list of strings.
;;; Make a list of strings from a string using space as separator
lvars list;
sysparse_string('the cat sat on the mat') -> list;
;;; print the list of strings
list =>
** [the cat sat on the mat]
By giving it an extra parameter 'true' we can make it recognize numbers and produce a list of strings and numbers
lvars list;
sysparse_string('one 1 two 2 three 3 four 4', true) -> list;
;;; print the list of strings and numbers
list =>
** [one 1 two 2 three 3 four 4]
;;; check that first item is a string and second an integer
isstring(list(1))=>
** <true>
isinteger(list(2))=>
** <true>
Now show some uses of the built in procedure sys_parse_string, which allows more options:
;;; Make pop-11 print strings with quotes true -> pop_pr_quotes; ;;; ;;; Create a string of tokens using comma as token separator lvars str='Hello,How,Are,You,Today'; ;;; ;;; Make a list of strings by applying sys_parse_string ;;; to str, using the character `,` as separator (the default ;;; separator, if none is provided, is the space character). lvars strings; [% sys_parse_string(str, `,`) %] -> strings; ;;; ;;; print the list of strings strings => ** ['Hello' 'How' 'Are' 'You' 'Today']
If {% ... %} were used instead of [% ... %] the result would be a vector (i.e. array) of strings rather than a list of strings.
{% sys_parse_string(str, `,`) %} -> strings;
;;; print the vector
strings =>
** {'Hello' 'How' 'Are' 'You' 'Today'}
It is also possible to give sys_parse_string a 'conversion' procedure, which is applied to each of the tokens. E.g. it could be used to produce a vector of numbers, using the conversion procedure 'strnumber', which converts a string to a number:
lvars numbers;
{% sys_parse_string('100 101 102 103 99.9 99.999', strnumber) %} -> numbers;
;;; the result is a vector containing integers and floats,
;;; which can be printed thus:
numbers =>
** {100 101 102 103 99.9 99.999}
Using lower level pop-11 facilities to tokenise the string:
;;; Declare and initialize variables
lvars str='Hello,How,Are,You,Today';
;;; Iterate over string
lvars ls = [], i, j = 1;
for i from 1 to length(str) do
;;; If comma
if str(i) = `,` then
;;; Prepend word (substring) to list
cons(substring(j, i - j, str), ls) -> ls;
i + 1 -> j;
endif;
endfor;
;;; Prepend final word (if needed)
if j <= length(str) then
cons(substring(j, length(str) - j + 1, str), ls) -> ls;
endif;
;;; Reverse the list
rev(ls) -> ls;
Since the task requires to use array we convert list to array
;;; Put list elements and lenght on the stack
destlist(ls);
;;; Build a vector from them
lvars ar = consvector();
;;; Display in a loop, putting trailing period
for i from 1 to length(ar) do
printf(ar(i), '%s.');
endfor;
printf('\n');
We could use list directly for printing:
for i in ls do
printf(i, '%s.');
endfor;
so the conversion to vector is purely to satisfy task formulation.
[edit] Python
Works with: Python version 2.5
text = "Hello,How,Are,You,Today"
tokens = text.split(',')
print '.'.join(tokens)
If you want to print each word on its own line:
for token in tokens:
print token
or
print "\n".join(tokens)
or the one liner
print '.'.join('Hello,How,Are,You,Today'.split(','))
[edit] Raven
'Hello,How,Are,You,Today' ',' split '.' join print
[edit] Ruby
string = "Hello,How,Are,You,Today".split(',')
string.each do |w|
print "#{w}."
end
puts "Hello,How,Are,You,Today".split(',').join('.')
[edit] Seed7
var array string: tokens is 0 times "";
tokens := split("Hello,How,Are,You,Today", ",");
[edit] Slate
('Hello,How,Are,You,Today' splitWith: $,) join &separator: '.'.
[edit] Smalltalk
|array |
array := 'Hello,How,Are,You,Today' subStrings: $,.
array fold: [:concatenation :string | concatenation, '.', string ]
Some implementations also have a join: convenience method that allows the following shorter solution:
('Hello,How,Are,You,Today' subStrings: $,) join: '.'
The solution displaying a trailing period would be:
|array |
array := 'Hello,How,Are,You,Today' subStrings: $,.
array inject: '' into: [:concatenation :string | concatenation, string, '.' ]
[edit] Standard ML
val splitter = String.tokens (fn c => c = #","); val main = (String.concatWith ".") o splitter;
Test:
- main "Hello,How,Are,You,Today" val it = "Hello.How.Are.You.Today" : string
[edit] Tcl
Generating a list form a string by splitting on a comma:
split string ,
Joining the elements of a list by a period:
join list .
Thus the whole thing would look like this:
puts [join [split "Hello,How,Are,You,Today" ,] .]
If you'd like to retain the list in a variable with the name "words", it would only be marginally more complex:
puts [join [set words [split "Hello,How,Are,You,Today" ,]] .]
(In general, the regexp command is also used in Tcl for tokenization of strings, but this example does not need that level of complexity.)
[edit] tr
echo 'Hello,How,Are,You,Today' | tr ',' '.'
[edit] UnixPipes
rtoken() {
(IFS=\ read A B ; echo $A; test -n "$B" && (echo $B | token) )
}
tokens() {
IFS=, read A ; echo $A | rtoken
}
echo "Hello,How,Are,You" | tokens
[edit] Ursala
A list of strings is made by separating at the commas using the library function, sep. A single string is then made by joining the list of strings with periods using the library function, mat. Each of these is a second order function parameterized by the delimiter. Character literals are preceded by a backquote.
#import std token_list = sep`, 'Hello,How,Are,You,Today' #cast %s main = mat`. token_list
output:
'Hello.How.Are.You.Today'
[edit] Vedit macro language
Vedit does not use the concepts of array or list. Normally, the text is processed as text in an edit buffer.
However, this example shows how to split the text into multiple text registers (10, 11, 12 etc.). The contents of each text register is then displayed to user, separated by a period.
Buf_Switch(Buf_Free)
Ins_Text("Hello,How,Are,You,Today")
// Split the text into text registers 10, 11, ...
BOF
#1 = 9
Repeat(ALL) {
#1++
#2 = Cur_Pos
Search(",", ADVANCE+ERRBREAK)
Reg_Copy_Block(#1, #2, Cur_Pos-1)
}
Reg_Copy_Block(#1, #2, EOB_Pos)
// Display the list
for (#3 = 10; #3 <= #1; #3++) {
Reg_Type(#3) Message(".")
}
Buf_Quit(OK)
