Tokenize a string with escaping: Difference between revisions

Content added Content deleted

Inline

Revision as of 18:24, 20 May 2022

Task

Write a function or program that can split a string at each non-escaped occurrence of a separator character.

It should accept three input parameters:

The string
The separator character
The escape character

It should output a list of strings.

Details

Rules for splitting:

The fields that were separated by the separators, become the elements of the output list.
Empty fields should be preserved, even at the start and end.

Rules for escaping:

"Escaped" means preceded by an occurrence of the escape character that is not already escaped itself.
When the escape character precedes a character that has no special meaning, it still counts as an escape (but does not do anything special).
Each occurrence of the escape character that was used to escape something, should not become part of the output.

Test case

Demonstrate that your function satisfies the following test-case:

Input

Output

string:	one^\|uno\|\|three^^^^\|four^^^\|^cuatro\|
separator character:	\|
escape character:	^

one|uno

three^^

four^|cuatro

(Print the output list in any format you like, as long as it is it easy to see what the fields are.)

Other tasks related to string operations:

Metrics

Counting

Remove/replace

Anagrams/Derangements/shuffling

Find/Search/Determine

Formatting

Song lyrics/poems/Mad Libs/phrases

Tokenize

Sequences

11l

Translation of: Python

<lang 11l>F token_with_escape(a, escape = ‘^’, separator = ‘|’)

  [String] result
  V token = ‘’
  V state = 0
  L(c) a
     I state == 0
        I c == escape
           state = 1
        E I c == separator
           result.append(token)
           token = ‘’
        E
           token ‘’= c
     E I state == 1
        token ‘’= c
        state = 0
  result.append(token)
  R result

print(token_with_escape(‘one^|uno||three^^^^|four^^^|^cuatro|’).map(s -> ‘'’s‘'’).join(‘, ’))</lang>

Output:

'one|uno', '', 'three^^', 'four^|cuatro', ''

<lang 8080asm> org 100h jmp demo ;;; Routine to split a 0-terminated string ;;; Input: B=separator, C=escape, HL=string pointer. ;;; Output: DE=end of list of strings ;;; The split strings are stored in place. split: mov d,h ; Set DE = output pointer mov e,l snext: mov a,m ; Get current input character inx h ; Advance input pointer stax d ; Write character at output pointer ana a ; If zero, we are done rz cmp c ; Is it the escape character? jz sesc cmp b ; Is it the separator character? jz ssep inx d ; Otherwise, advance output pointer, jmp snext ; and get the next character sesc: mov a,m ; Store the escaped character without inx h ; checking for anything except zero. stax d inx d ana a ; Zero is still end of string rz jmp snext ssep: xra a ; End of string, write zero terminator stax d inx d jmp snext ;;; Use the routine to split the test-case string demo: mvi b,'|' ; Separator character mvi c,'^' ; Escape character lxi h,test ; Pointer to test string call split ;;; Print each string on its own line lxi h,test str: call puts ; Print string call cmp16 ; Are we there yet? jnc str ; If not, print the next string ret ;;; 16-bit compare cmp16: mov a,d cmp h rnz mov a,e cmp l ret ;;; Print zero-terminated string with newline puts: push d ; Keep DE registers push h ; Keep pointer lxi d,pfx ; Print prefix mvi c,9 call 5 pop h ; Restore pointer ploop: mov e,m ; Get current character push h ; Keep pointer mvi c,2 ; CP/M print character call 5 pop h ; Restore pointer mov a,m ; Is character zero? ora a inx h ; Increment pointer jnz ploop ; If not, there are more characters push h ; Keep pointer lxi d,nl ; Write newline mvi c,9 ; CP/M print string call 5 pop h pop d ; Restore DE registers ret pfx: db '> $' ; Prefix to make the output more obvious nl: db 13,10,'$' test: db 'one^|uno||three^^^^|four^^^|^cuatro|',0</lang>

Output:

> one|uno
>
> three^^
> four^|cuatro
>

Action!

<lang Action!>DEFINE PTR="CARD"

TYPE Tokens=[

 PTR buf   ;BYTE ARRAY
 PTR arr   ;CARD ARRAY
 PTR endPtr
 BYTE count]

PROC Init(Tokens POINTER t BYTE ARRAY b PTR ARRAY a)

 t.buf=b
 t.arr=a
 t.endPtr=b
 t.count=0

RETURN

PROC AddToken(Tokens POINTER t CHAR ARRAY s)

 PTR ARRAY a
 CHAR ARRAY tmp

 a=t.arr
 tmp=t.endPtr
 SCopy(tmp,s)
 a(t.count)=tmp
 t.count==+1
 t.endPtr=t.endPtr+s(0)+1

RETURN

PROC PrintTokens(Tokens POINTER t)

 BYTE i
 PTR ARRAY a
 
 a=t.arr
 FOR i=0 TO t.count-1
 DO
   PrintF("""%S""%E",a(i))
 OD

RETURN

PROC Append(CHAR ARRAY s CHAR c)

 s(0)==+1
 s(s(0))=c

RETURN

PROC Tokenize(CHAR ARRAY s CHAR sep,esc Tokens POINTER res)

 BYTE ARRAY b(200)
 PTR ARRAY a(20)
 CHAR ARRAY tmp(255)
 BYTE i,isEsc
 CHAR c

 Init(res,b,a)
 isEsc=0
 tmp(0)=0
 FOR i=1 TO s(0)
 DO
   c=s(i)
   IF isEsc THEN
     isEsc=0
     Append(tmp,c)
   ELSE
     IF c=esc THEN
       isEsc=1
     ELSEIF c=sep THEN
       AddToken(res,tmp)
       tmp(0)=0
     ELSE
       Append(tmp,c)
     FI
   FI
 OD
 AddToken(res,tmp)

RETURN

PROC Main()

 Tokens t

 Tokenize("one^|uno||three^^^^|four^^^|^cuatro|",'|,'^,t)
 PrintTokens(t)

RETURN</lang>

Output:

Screenshot from Atari 8-bit computer

"one|uno"
""
"three^^"
"four^|cuatro"
""

Ada

<lang Ada>with Ada.Text_Io; with Ada.Containers.Indefinite_Vectors; with Ada.Strings.Unbounded;

procedure Tokenize is

  package String_Vectors is
    new Ada.Containers.Indefinite_Vectors (Positive, String);
  use String_Vectors;

  function Split (Text      : String;
                  Separator : Character := '|';
                  Escape    : Character := '^') return Vector
  is
     use Ada.Strings.Unbounded;
     Result  : Vector;
     Escaped : Boolean := False;
     Accu    : Unbounded_String;
  begin

     for Char of Text loop

        case Escaped is

           when False =>
              if Char = Escape then
                 Escaped := True;
              elsif Char = Separator then
                 Append (Result, To_String (Accu));
                 Accu := Null_Unbounded_String;
              else
                 Append (Accu, Char);
              end if;

           when True =>
              Append (Accu, Char);
              Escaped := False;

        end case;

     end loop;
     Append (Result, To_String (Accu));

     return Result;
  end Split;

  procedure Put_Vector (List : Vector) is
     use Ada.Text_Io;
  begin
     for Element of List loop
        Put ("'"); Put (Element); Put ("'"); New_Line;
     end loop;
  end Put_Vector;

begin

  Put_Vector (Split ("one^|uno||three^^^^|four^^^|^cuatro|"));

end Tokenize;</lang>

Output:

'one|uno'
''
'three^^'
'four^|cuatro'
''

ALGOL 68

<lang algol68>BEGIN

   # returns s parsed according to delimiter and escape                      #
   PROC parse with escapes = ( STRING s, CHAR delimiter, escape )[]STRING:
        IF ( UPB s - LWB s ) + 1 < 1 THEN
           # empty string                                                    #
           [ 1 : 0 ]STRING empty array;
           empty array
        ELSE
           # at least one character                                          #
           # allow for a string composed entirely of delimiter characters    #
           [ 1 : ( UPB s - LWB s ) + 3 ]STRING result;
           INT r pos := 1;
           INT s pos := LWB s;
           result[ r pos ] := "";
           WHILE s pos <= UPB s DO
               CHAR c = s[ s pos ];
               IF  c = delimiter THEN
                   # start a new element                                     #
                   result[ r pos +:= 1 ] := ""
               ELIF c = escape THEN
                   # use the next character even if it is an escape          #
                   s pos +:= 1;
                   IF s pos < UPB s THEN
                       # the escape is not the last character                #
                       result[ r pos ] +:= s[ s pos ]
                   FI
               ELSE
                   # normal character                                        #
                   result[ r pos ] +:= c
               FI;
               s pos +:= 1
           OD;
           result[ 1 : r pos ]
        FI; # parse with escapes #
   # task test case                                                          #
   []STRING tokens = parse with escapes( "one^|uno||three^^^^|four^^^|^cuatro|", "|", "^" );
   FOR t pos FROM LWB tokens TO UPB tokens DO print( ( "[", tokens[ t pos ], "]", newline ) ) OD

END</lang>

Output:

[one|uno]
[]
[three^^]
[four^|cuatro]
[]

AppleScript

Translation of: JavaScript

<lang AppleScript>------------------ TOKENIZE WITH ESCAPING ----------------

-- tokenize :: String -> Character -> Character -> [String] on tokenize(str, delimChar, chrEsc)

   script charParse
       -- Record: {esc:Bool, token:String, tokens:[String]}
       -- charParse :: Record -> Character -> Record
       on |λ|(a, x)
           set blnEsc to esc of a
           set blnEscChar to ((not blnEsc) and (x = chrEsc))
           
           if ((not blnEsc) and (x = delimChar)) then
               set k to ""
               set ks to (tokens of a) & token of a
           else
               set k to (token of a) & cond(blnEscChar, "", x)
               set ks to tokens of (a)
           end if
           
           {esc:blnEscChar, token:k, tokens:ks}
       end |λ|
   end script
   
   set recParse to foldl(charParse, ¬
       {esc:false, token:"", tokens:[]}, splitOn("", str))
   
   tokens of recParse & token of recParse

end tokenize

TEST -------------------------

on run

   script numberedLine
       on |λ|(a, s)
           set iLine to lineNum of a
           {lineNum:iLine + 1, report:report of a & iLine & ":" & tab & s & linefeed}
       end |λ|
   end script
   
   report of foldl(numberedLine, {lineNum:1, report:""}, ¬
       tokenize("one^|uno||three^^^^|four^^^|^cuatro|", "|", "^"))

end run

GENERIC FUNCTIONS -------------------

-- foldl :: (a -> b -> a) -> a -> [b] -> a on foldl(f, startValue, xs)

   tell mReturn(f)
       set v to startValue
       set lng to length of xs
       repeat with i from 1 to lng
           set v to |λ|(v, item i of xs, i, xs)
       end repeat
       return v
   end tell

end foldl

-- Lift 2nd class handler function into 1st class script wrapper -- mReturn :: Handler -> Script on mReturn(f)

   if class of f is script then
       f
   else
       script
           property |λ| : f
       end script
   end if

end mReturn

-- splitOn :: String -> String -> [String] on splitOn(pat, src)

   set {dlm, my text item delimiters} to ¬
       {my text item delimiters, pat}
   set xs to text items of src
   set my text item delimiters to dlm
   return xs

end splitOn

-- cond :: Bool -> a -> a -> a on cond(bool, f, g)

   if bool then
       f
   else
       g
   end if

end cond</lang>

Output:

1:    one|uno
2:    
3:    three^^
4:    four^|cuatro
5:

Arturo

<lang rebol>tokenize: function [s sep esc][ escaping: 0

loop 0..(size s)-1 [i][ chr: get split s i

if? escaping=1 [ prints chr escaping: 0 ] else [ case [chr] when? [=sep] [print ""] when? [=esc] [escaping: 1] else [prints chr] ] ] print "" ]

str: "one^|uno||three^^^^|four^^^|^cuatro|" tokenize str "|" "^"</lang>

Output:

one|uno

three^^
four^|cuatro

AutoHotkey

<lang AutoHotkey>Tokenize(s,d,e){ for i,v in x:=StrSplit(StrReplace(StrReplace(StrReplace(s,e e,Chr(0xFFFE)),e d,Chr(0xFFFF)),e),d) x[i]:=StrReplace(StrReplace(v,Chr(0xFFFE),e),Chr(0xFFFF),d) return x }</lang> Examples:<lang AutoHotkey>str := "one^|uno||three^^^^|four^^^|^cuatro|" for i, v in Tokenize(str, "|", "^") output .= i " : " v "`n" MsgBox % output</lang>

Output:

1 : one|uno
2 : 
3 : three^^
4 : four^|cuatro
5 :

BBC BASIC

<lang bbcbasic>REM >tokenizer PROC_tokenize("one^|uno||three^^^^|four^^^|^cuatro|", "|", "^") END

DEF PROC_tokenize(src$, sep$, esc$) LOCAL field%, char$, escaping%, i% field% = 1 escaping% = FALSE PRINT field%; " "; FOR i% = 1 TO LEN src$

 char$ = MID$(src$, i%, 1)
 IF escaping% THEN
   PRINT char$;
   escaping% = FALSE
 ELSE
   CASE char$ OF
   WHEN sep$
     PRINT
     field% += 1
     PRINT field%; " ";
   WHEN esc$
     escaping% = TRUE
   OTHERWISE
     PRINT char$;
   ENDCASE
 ENDIF

NEXT PRINT ENDPROC</lang>

Output:

         1 one|uno
         2 
         3 three^^
         4 four^|cuatro
         5

BQN

<lang bqn>str ← "one^|uno||three^^^^|four^^^|^cuatro|" Split ← ((⊢-˜+`×¬)∘=⊔⊢) SplitE ← {

 esc ← <`'^'=𝕩
 rem ← »esc
 spl ← (¬rem)∧'|'=𝕩 
 𝕩⊔˜(⊢-(esc∨spl)×1⊸+)+`spl

}

•Show SplitE str</lang> <lang>⟨ "one|uno" ⟨⟩ "three^^" "four^|cuatro" ⟩</lang>

C

Works with: C

<lang c>#include <stdlib.h>

include <stdio.h>

define STR_DEMO "one^|uno||three^^^^|four^^^|^cuatro|"
define SEP '|'
define ESC '^'

typedef char* Str; /* just for an easier reading */

/* ===> FUNCTION PROTOTYPES <================================================ */ unsigned int ElQ( const char *s, char sep, char esc ); Str *Tokenize( char *s, char sep, char esc, unsigned int *q );

/*============================================================================== Main function. Just passes a copy of the STR_DEMO string to the tokenization function and shows the results. ==============================================================================*/

int main() {

   char s[] = STR_DEMO;
   unsigned int i, q;

   Str *list = Tokenize( s, SEP, ESC, &q );

   if( list != NULL ) {
       printf( "\n Original string: %s\n\n", STR_DEMO );
       printf( " %d tokens:\n\n", q );

       for( i=0; i<q; ++i )
           printf( " %4d. %s\n", i+1, list[i] );

       free( list );
   }

   return 0;

}

/*============================================================================== "ElQ" stands for "Elements Quantity". Counts the amount of valid element in the string s, according to the separator character provided in sep and the escape character provided in esc. ==============================================================================*/

unsigned int ElQ( const char *s, char sep, char esc ) {

   unsigned int q, e;
   const char *p;

   for( e=0, q=1, p=s; *p; ++p ) {
       if( *p == esc )
           e = !e;
       else if( *p == sep )
           q += !e;
       else e = 0;
   }

   return q;

}

/*============================================================================== The actual tokenization function. Allocates as much dynamic memory as needed to contain the pointers to the tokenized portions of the string passed as the "s" parameter, then looks for the separators characters sep, paying attention to the occurrences of the escape character provided in esc. When a valid separator is found, the function swaps it with a '\0' terminator character and stores the pointer to the next string into the array of pointers in dynamic memory. On output, the value of *q is the number of pointers in the array. The caller is responsible for deallocating with free() the returned array of pointers when it is no longer needed. In case of failure, NULL is returned. ==============================================================================*/

Str *Tokenize( char *s, char sep, char esc, unsigned int *q ) {

   Str *list = NULL;

   *q = ElQ( s, sep, esc );
   list = malloc( *q * sizeof(Str) );

   if( list != NULL ) {
       unsigned int e, i;
       char *p;

       i = 0;
       list[i++] = s;

       for( e=0, p=s; *p; ++p ) {
           if( *p == esc ) {
               e = !e;
           }
           else if( *p == sep && !e ) {
               list[i++] = p+1;
               *p = '\0';
           }
           else {
               e = 0;
           }
       }
   }

   return list;

}</lang>

Output:

 Original string: one^|uno||three^^^^|four^^^|^cuatro|

 5 tokens:

    1. one^|uno
    2.
    3. three^^^^
    4. four^^^|^cuatro
    5.

C#

<lang csharp>using System; using System.Text; using System.Collections.Generic;

public class TokenizeAStringWithEscaping {

   public static void Main() {
       string testcase = "one^|uno||three^^^^|four^^^|^cuatro|";
       foreach (var token in testcase.Tokenize(separator: '|', escape: '^')) {
           Console.WriteLine(": " + token); //Adding a : so we can see empty lines
       }
   }

}

public static class Extensions {

   public static IEnumerable<string> Tokenize(this string input, char separator, char escape) {
       if (input == null) yield break;
       var buffer = new StringBuilder();
       bool escaping = false;
       foreach (char c in input) {
           if (escaping) {
               buffer.Append(c);
               escaping = false;
           } else if (c == escape) {
               escaping = true;
           } else if (c == separator) {
               yield return buffer.Flush();
           } else {
               buffer.Append(c);
           }
       }
       if (buffer.Length > 0 || input[input.Length-1] == separator) yield return buffer.Flush();
   }
   
   public static string Flush(this StringBuilder stringBuilder) {
       string result = stringBuilder.ToString();
       stringBuilder.Clear();
       return result;
   }

}</lang>

Output:

: one|uno
: 
: three^^
: four^|cuatro
:

C++

<lang cpp>#include <iostream>

include <stdexcept>
include <string>
include <vector>

using namespace std;

vector<string> tokenize(const string& input, char seperator, char escape) {

   vector<string> output;
   string token;

   bool inEsc = false;
   for (char ch : input) {
       if (inEsc) {
           inEsc = false;
       } else if (ch == escape) {
           inEsc = true;
           continue;
       } else if (ch == seperator) {
           output.push_back(token);
           token = "";
           continue;
       }
       token += ch;
   }
   if (inEsc)
       throw new invalid_argument("Invalid terminal escape");

   output.push_back(token);
   return output;

}

int main() {

   string sample = "one^|uno||three^^^^|four^^^|^cuatro|";

   cout << sample << endl;
   cout << '[';
   for (auto t : tokenize(sample, '|', '^')) {
       cout << '"' << t << "\", ";
   }
   cout << ']' << endl;

   return 0;

}</lang>

Output:

one^|uno||three^^^^|four^^^|^cuatro|
["one|uno", "", "three^^", "four^|cuatro", "", ]

CLU

<lang clu>tokenize = iter (sep, esc: char, s: string) yields (string)

   escape: bool := false
   part: array[char] := array[char]$[]
   for c: char in string$chars(s) do   
       if escape then 
           escape := false
           array[char]$addh(part,c)
       elseif c=esc then
           escape := true
       elseif c=sep then
           yield(string$ac2s(part))
           part := array[char]$[]
       else
           array[char]$addh(part,c)
       end
   end
   yield(string$ac2s(part))

end tokenize

start_up = proc ()

   po: stream := stream$primary_output()
   testcase: string := "one^|uno||three^^^^|four^^^|^quatro|"
   
   for part: string in tokenize('|', '^', testcase) do
       stream$putl(po, "\"" || part || "\"")
   end

end start_up</lang>

Output:

"one|uno"
""
"three^^"
"four^|quatro"
""

COBOL

<lang cobol> >>SOURCE FORMAT FREE identification division. program-id. 'tokenizewithescaping'. environment division. configuration section. repository.

   function all intrinsic.

data division. working-storage section.

01 escape-char pic x value '^'. 01 separator-char pic x value '|'. 01 reference-string pic x(64) value

  'one^|uno||three^^^^|four^^^|^cuatro|'.

01 input-string pic x(64). 01 c pic 99. 01 escaped pic x.

01 t pic 99. 01 t-max pic 99. 01 t-lim pic 99 value 32. 01 token-entry occurs 32.

  03  token-len pic 99.
  03  token pic x(16).

01 l pic 99. 01 l-lim pic 99 value 16.

01 error-found pic x.

procedure division. start-tokenize-with-escaping.

   move reference-string to input-string
   perform tokenize

   move 'token' to input-string
   perform tokenize
  
   move '^^^^^^^^' to input-string
   perform tokenize
  
   move '||||||||' to input-string
   perform tokenize

   move all 'token' to input-string
   perform tokenize

   move all 't|' to input-string
   perform tokenize

   move spaces to input-string
   perform tokenize

   display space

   stop run
   .

tokenize.

   display space
   display 'string:'
   display input-string

   move 'N' to escaped error-found
   move 1 to t-max
   initialize token-entry(t-max)
   move 0 to l

   perform varying c from 1 by 1 until
   c > length(input-string)
   or input-string(c:) = spaces

       evaluate escaped also input-string(c:1)
       when 'N' also escape-char
           move 'Y' to escaped
       when 'N' also separator-char
           perform increment-t-max
           if error-found = 'Y'
               exit paragraph
           end-if
       when 'N' also any
           perform move-c
           if error-found = 'Y'
               exit paragraph
           end-if
       when 'Y' also any
           perform move-c
           if error-found = 'Y'
               exit paragraph
           end-if
           move 'N' to escaped
       end-evaluate

   end-perform
   if l > 0
       move l to token-len(t-max)
   end-if

   if c = 1
       display 'no tokens'
   else
       display 'tokens:'
       perform varying t from 1 by 1 until t > t-max
           if token-len(t) > 0
               display t ': ' token-len(t) space token(t)
           else
               display t ': ' token-len(t)
           end-if
       end-perform
   end-if
   .

increment-t-max.

   if t-max >= t-lim
       display 'error: at ' c ' number of tokens exceeds ' t-lim
       move 'Y' to error-found
   else
       move l to token-len(t-max)
       add 1 to t-max
       initialize token-entry(t-max)
       move 0 to l
       move 'N' to error-found
   end-if
   .

move-c.

   if l >= l-lim
       display 'error: at ' c ' token length exceeds ' l-lim
       move 'Y' to error-found
   else
       add 1 to l
       move input-string(c:1) to token(t-max)(l:1)
       move 'N' to error-found
   end-if
   .

end program 'tokenizewithescaping'. </lang>

Output:

$ cobc -xj tokenizewithescaping.cbl 
 
string:
one^|uno||three^^^^|four^^^|^cuatro|                            
tokens:
01: 07 one|uno         
02: 00
03: 07 three^^         
04: 12 four^|cuatro    
05: 00
 
string:
token                                                           
tokens:
01: 05 token           
 
string:
^^^^^^^^                                                        
tokens:
01: 04 ^^^^            
 
string:
||||||||                                                        
tokens:
01: 00
02: 00
03: 00
04: 00
05: 00
06: 00
07: 00
08: 00
09: 00
 
string:
tokentokentokentokentokentokentokentokentokentokentokentokentoke
error: at 17 token length exceeds 16
 
string:
t|t|t|t|t|t|t|t|t|t|t|t|t|t|t|t|t|t|t|t|t|t|t|t|t|t|t|t|t|t|t|t|
error: at 64 number of tokens exceeds 32
 
string:
                                                                
no tokens

Common Lisp

<lang lisp>(defun split (input separator escape)

 (flet ((make-string-buffer ()
          (make-array 0 :element-type 'character :adjustable t :fill-pointer t)))
   (loop with token = (make-string-buffer)
         with result = nil
         with to-be-escaped = nil
         for ch across input
         do (cond (to-be-escaped
                   (vector-push-extend ch token)
                   (setf to-be-escaped nil))
                  ((char= ch escape)
                   (setf to-be-escaped t))
                  ((char= ch separator)
                   (push token result)
                   (setf token (make-string-buffer)))
                  (t
                   (vector-push-extend ch token)))
         finally (push token result)
                 (return (nreverse result)))))

(defun main ()

 (dolist (token (split "one^|uno||three^^^^|four^^^|^cuatro|" #\| #\^))
   (format t "'~A'~%" token)))</lang>

Output:

'one|uno'
''
'three^^'
'four^|cuatro'
''

D

Translation of: Java

<lang D>import std.stdio;

void main() {

   string sample = "one^|uno||three^^^^|four^^^|^cuatro|";

   writeln(sample);
   writeln(tokenizeString(sample, '|', '^'));

}

auto tokenizeString(string source, char seperator, char escape) {

   import std.array : appender;
   import std.exception : enforce;

   auto output = appender!(string[]);
   auto token = appender!(char[]);

   bool inEsc;
   foreach(ch; source) {
       if (inEsc) {
           inEsc = false;
       } else if (ch == escape) {
           inEsc = true;
           continue;
       } else if (ch == seperator) {
           output.put(token.data.idup);
           token.clear();
           continue;
       }
       token.put(ch);
   }
   enforce(!inEsc, "Invalid terminal escape");

   output.put(token.data.idup);
   return output.data;

}</lang>

Output:

one^|uno||three^^^^|four^^^|^cuatro|
["one|uno", "", "three^^", "four^|cuatro", ""]

Dyalect

Translation of: C#

<lang dyalect>func String.Tokenize(separator, escape) {

   var buffer = []
   var escaping = false
   for c in this {
       if escaping {
           buffer.Add(c)
           escaping = false
       } else if c == escape {
           escaping = true
       } else if c == separator {
           yield buffer.Flush();
       } else {
           buffer.Add(c);
       }
   }

   if buffer.Length() > 0 || this[this.Length() - 1] == separator {
       yield buffer.Flush()
   }

}

func Array.Flush() {

   var str = String.Concat(values: this)
   this.Clear()
   str

}

let testcase = "one^|uno||three^^^^|four^^^|^cuatro|"; for token in testcase.Tokenize(separator: '|', escape: '^') {

   print(": \(token)")

}</lang>

Output:

: one|uno
:
: three^^
: four^|cuatro
:

Elena

Translation of: C#

ELENA 4.x : <lang elena>import extensions; import extensions'routines; import system'collections; import system'routines; import system'text;

extension op : String {

   tokenize(separator,escape)
   {
       auto buffer := new TextBuilder();
       auto list := new ArrayList();

       bool escaping := false;
       self.forEach:(ch)
       {
           if (escaping)
           {
               buffer.write:ch;
               escaping := false
           }
           else if (ch == escape)
           {
               escaping := true
           }
           else if (ch == separator)
           {
               list.append(buffer.Value);
               buffer.clear()
           }
           else
           {
               buffer.write:ch
           }
       };

       ^ list
   }

}

const string testcase = "one^|uno||three^^^^|four^^^|^cuatro|";

public program() {

   testcase.tokenize("|", "^").forEach:printingLn

}</lang>

Output:

one|uno

three^^
four^|cuatro

F#

<lang fsharp>open System open System.Text.RegularExpressions

(*

   .NET regexes have unlimited look-behind, so we can look for separators
   which are preceeded by an even number of (or no) escape characters

)

let split esc sep s =

   Regex.Split (
       s,
       String.Format("(?<=(?:\b|[^{0}])(?:{0}{0})*){1}", Regex.Escape(esc), Regex.Escape(sep))
       )

let unescape esc s =

   Regex.Replace(
       s,
       Regex.Escape(esc) + "(.)",
       "$1"
       )

[<EntryPoint>] let main argv =

   let (esc, sep) = ("^", "|")
   "one^|uno||three^^^^|four^^^|^cuatro|"
   |> split esc sep
   |> Seq.map (unescape esc)
   |> Seq.iter (fun s -> printfn "'%s'" s)
   0</lang>

Output:

'one|uno'
''
'three^^'
'four^|cuatro'
''

Factor

This example uses Factor's parser-combinators vocabulary, which is modeled after Haskell's parser combinators. Page 51 of this pdf contains a useful introduction to this vocabulary.

Works with: Factor version 0.99 2019-10-06

<lang factor>USING: accessors kernel lists literals namespaces parser-combinators prettyprint sequences strings ;

SYMBOLS: esc sep ;

set-chars ( m n -- ) [ sep set ] [ esc set ] bi* ;

escape ( -- parser ) esc get 1token ;

escaped ( -- parser ) escape any-char-parser &> ;

separator ( -- parser ) sep get 1token ;

character ( -- parser )

   ${ esc get sep get } [ member? not ] curry satisfy ;

my-token ( -- parser ) escaped character <|> <*> ;

token-list ( -- parser )

   my-token separator list-of [ [ >string ] map ] <@ ;

tokenize ( str sep-char esc-char -- seq )

   set-chars token-list parse car parsed>> ;

"one^|uno||three^^^^|four^^^|^cuatro|" CHAR: | CHAR: ^ tokenize .</lang>

Output:

{ "one|uno" "" "three^^" "four^|cuatro" "" }

Forth

<lang forth>variable 'src variable #src variable offset

advance 1 offset +! ;

chr@ offset @ 'src @ + c@ ;

nextchr advance chr@ ;

bound offset @ #src @ u< ;

separator? dup [char] | = if drop cr else emit then ;

escape? dup [char] ^ = if drop nextchr emit else separator? then ;

tokenize 0 offset ! begin bound while nextchr escape? repeat ;

\ Test of function Here 'src ! ," one^|uno||three^^^^|four^^^|^cuatro|" here 'src @ - #src ! page cr ." #### start ####" cr tokenize cr ." #### End ####" cr </lang>

Output:

#### start ####
one|uno

three^^
four^|cuatro

#### End  ####

Fortran

First Fortran (1958) offered no facilities for inspecting or manipulating text, until Fortran IV when the A format code was introduced whereby text could be read or written from numeric variables. The difficulties and incompatibilities between different computers were eased with F77 that offered CHARACTER*n variables, though they are not quite strings that have a varying length. F95 introduces the ability to define a compound entity such as a string and F2003 standardised a version of strings whereby with each assignment to such a variable, it would be re-allocated with the required amount of storage. Otherwise, one proceeds with CHARACTER variables and an associated variable containing its current length as with TOKEN and L. However, when passed to subroutines (or functions) as a parameter, a CHARACTER variable is supplied along with a secret additional parameter giving the size of the variable, and this is stringlike, so long as there is no need to change the length. Thus, the length of parameter TEXT to subroutine SPLIT can be found via LEN(TEXT).

The source style is F90 simply for the convenience of having subroutine SPLOT defined within subroutine SPLIT so as to gain access to certain variables. If separate subroutines were to be used, then there would have to be parameters or COMMON variables, or, one could just replicate the code within SPLIT. A further F90 feature involves declaring the size of internal variable TOKEN to be LEN(TEXT), which is surely the largest it could be. Otherwise, one would have to select some "surely big enough" value.<lang Fortran> SUBROUTINE SPLIT(TEXT,SEP,ESC) !Identifies and prints tokens from within a text.

      CHARACTER*(*) TEXT	!To be scanned.
      CHARACTER*(1) SEP	!The only separator for tokens.
      CHARACTER*(1) ESC	!Miscegnator.
      CHARACTER*(LEN(TEXT)) TOKEN	!Surely sufficient space.
      INTEGER N	!Counts the tokens as they're found.
      INTEGER I	!Steps through the text.
      INTEGER L	!Length of the token so far accumulated.
      LOGICAL ESCAPING	!Miscegnatory state.
       N = 0		!No tokens so far.
       L = 0		!Nor any text for the first.
       ESCAPING = .FALSE.	!And the state is good.
       DO I = 1,LEN(TEXT)	!Step through the text.
         IF (ESCAPING) THEN	!Are we in a mess?
           L = L + 1			!Yes. An ESC character had been seen.
           TOKEN(L:L) = TEXT(I:I)	!So, whatever follows is taken as itself.
           ESCAPING = .FALSE.		!There are no specially-recognised names.
          ELSE			!Otherwise, we're in text to inspect.
           IF (TEXT(I:I).EQ.ESC) THEN	!So, is it a troublemaker?
            ESCAPING = .TRUE.			!Yes! Trouble is to follow.
           ELSE IF (TEXT(I:I).EQ.SEP) THEN	!If instead a separator,
            CALL SPLOT				!Then the token up to it is complete.
           ELSE			!Otherwise, a simple constituent character.
            L = L + 1				!So, count it in.
            TOKEN(L:L) = TEXT(I:I)		!And copy it in.
           END IF			!So much for grist.
         END IF		!So much for that character.
       END DO			!On to the next.

Completes on end-of-text with L > 0, or, if the last character had been SEP, a null token is deemed to be following.

       CALL SPLOT	!Tail end.
      CONTAINS	!Save on having two copies of this code.
       SUBROUTINE SPLOT	!Show the token and scrub.
        N = N + 1			!Another one.
        WRITE (6,1) N,TOKEN(1:L)	!Reveal.
   1    FORMAT ("Token ",I0," >",A,"<")!Fancy layout.
        L = 0				!Prepare for a fresh token.
       END SUBROUTINE SPLOT	!A brief life.
     END SUBROUTINE SPLIT	!And then oblivion.

     PROGRAM POKE

     CALL SPLIT("one^|uno||three^^^^|four^^^|^cuatro|","|","^")

     END</lang>

The output has the text of the tokens marked >thus<

Token 1 >one|uno<
Token 2 ><
Token 3 >three^^<
Token 4 >four^|cuatro<
Token 5 ><

The terminating separator character is deemed to mark the start of a null token in the problem's specification. If the text ends without one, then the end-of-text ends the token and the DO-loop quits with L > 0 and so something for SPLOT. If the penultimate character were an ESC followed by a SEP, then the loop also ends with L > 0. If the text ends with a SEP but not preceded by an ESC (as in the example) then L = 0 - but SPLOT is invoked unconditionally. A SEP at the start of the text will also elicit a null token, as will an entirely null text.

If the text ends with an ESC, then this is surely a mistake and could be caught via a test that ESCAPING was true on exit from the loop. But no error messages are called for...

In this example the DO-loop relentlessly steps through the text, and in general this would not be convenient. Normally, token identification proceeds within a much larger context where one would not discard the token immediately after it is isolated, and rather than copying the text hither and thither, one might prefer to identify it in-place, say with variables L1 and L2 identifying the start and end positions within the working area. In such a case there would no longer be a need for a variable TOKEN and the angst of deciding on a suitable maximum size. This would also make it easier in any error messages to show context and provenance. However, the bizarre miscegnation of "escape" sequences (especially confusing within text literals), means that the source text does not necessarily constitute the text of the token.

FreeBASIC

Translation of: Ring

<lang freebasic>Sub tokenize(cadena As String, separador As String, escape As String)

   Dim As Integer campo = 1
   Dim As Boolean escapando = false
   Dim As String char
   Print ""; campo; " ";
   
   For i As Integer = 1 To Len(cadena)
       char = Mid(cadena, i, 1)
       If escapando  Then
           Print char;
           escapando = false
       Else
           Select Case char 
           Case separador
               Print
               campo += 1
               Print ""; campo; " ";
           Case escape
               escapando = true
           Case Else
               Print char;
           End Select
       End If
   Next i
   Print

End Sub

tokenize("one^|uno||three^^^^|four^^^|^cuatro|", "|", "^") Sleep</lang>

Output:

Igual que la entrada de Ring.

Go

<lang go>package main

import ( "errors" "fmt" )

func TokenizeString(s string, sep, escape rune) (tokens []string, err error) { var runes []rune inEscape := false for _, r := range s { switch { case inEscape: inEscape = false fallthrough default: runes = append(runes, r) case r == escape: inEscape = true case r == sep: tokens = append(tokens, string(runes)) runes = runes[:0] } } tokens = append(tokens, string(runes)) if inEscape { err = errors.New("invalid terminal escape") } return tokens, err }

func main() { const sample = "one^|uno||three^^^^|four^^^|^cuatro|" const separator = '|' const escape = '^'

fmt.Printf("Input: %q\n", sample) tokens, err := TokenizeString(sample, separator, escape) if err != nil { fmt.Println("error:", err) } else { fmt.Printf("Tokens: %q\n", tokens) } }</lang>

Output:

Input:   "one^|uno||three^^^^|four^^^|^cuatro|"
Tokens: ["one|uno" "" "three^^" "four^|cuatro" ""]

Haskell

Deterministic Finite Automaton

<lang haskell>splitEsc :: (Foldable t1, Eq t) => t -> t -> t1 t -> t splitEsc sep esc = reverse . map reverse . snd . foldl process (0, [[]])

 where process (st, r:rs) ch
         | st == 0 && ch == esc               = (1,      r:rs)
         | st == 0 && ch == sep               = (0,   []:r:rs)
         | st == 1 && sep == esc && ch /= sep = (0, [ch]:r:rs)
         | otherwise                          = (0, (ch:r):rs)</lang>

Output:

λ> splitEsc '|' '^' "one^|uno||three^^^^|four^^^|^cuatro|"
["one|uno","","three^^","four^|cuatro",""]

The solution works with any foldable structures.

λ> splitEsc 11 0 [2,3,11,3,4,5,11,0,11,2,3,4]
[[2,3],[3,4,5],[11,2,3,4]]

It handles pathological case when separator and escape are the same:

λ> split '|' '|' "one^|uno||three^^^^|four^^^|^cuatro|"
["one^","uno|three^^^^","four^^^","^cuatro"]

For splitting lists without escaping see Data.List.Split package.

Counduit-based solution

Constant in space (~ O(k), where k -- is token length), as fast as DFA-based solution.

<lang haskell>{-#Language LambdaCase #-} import Conduit

splitEscC :: (Monad m, Eq t) => t -> t -> Conduit t m [t] splitEscC sep esc = mapOutput reverse $ go True []

 where
   go notEsc b = await >>= \case 
     Nothing -> yield b
     Just ch | notEsc && ch == esc -> go False b
             | notEsc && ch == sep -> yield b >> go True []
             | otherwise -> go True (ch:b)</lang>

This new conduit could be used in a pipeline as follows:

<lang haskell>main = runConduit $

 yieldMany "one^|uno||three^^^^|four^^^|^cuatro|"
 .| splitEscC '|' '^'
 .| mapM_C print</lang>

λ> main
"one|uno"
""
"three^^"
"four^|cuatro"
""

Alternative

This is essentially equivalent to the first (DFA) example, but, though possibly less elegant than the guard idiom, appears to be fractionally faster with larger (eg 180k) test strings. <lang haskell>import Data.Bool (bool)

TOKENIZE WITH ESCAPING ----------------

tokenize :: Char -> Char -> String -> [String] tokenize delim esc str =

 reverse $
   reverse <$> (token : list)
 where
   (token, list, _) =
     foldr
       ( \x (aToken, aList, aEsc) ->
           let literal = not aEsc
               isEsc = literal && (x == esc)
            in bool
                 ( bool (x : aToken) aToken isEsc,
                   aList,
                   isEsc
                 )
                 ([], aToken : aList, isEsc)
                 (literal && x == delim)
       )
       ([], [], False)
       (reverse str)

TEST -------------------------

main :: IO () main =

 mapM_ print $
   tokenize
     '|'
     '^'
     "one^|uno||three^^^^|four^^^|^cuatro|"</lang>

Output:

"one|uno"
""
"three^^"
"four^|cuatro"
""

J

From the python example: <lang J> tokenize1=: tokenize =: '^|'&$: :(4 : 0)

'ESC SEP' =. x
STATE =. 0
RESULT =. 0 $ a:
TOKEN =. 
for_C. y do.
 if. STATE do.
  TOKEN =. TOKEN , C
  STATE =. 0
 else.
  if. C = ESC do.
   STATE =. 1
  elseif. C = SEP do.
   RESULT =. RESULT , < TOKEN
   TOKEN =. 
  elseif. do.
   TOKEN =. TOKEN , C
  end.
 end.
end.
RESULT =. RESULT , < TOKEN

) </lang>

tokenize 'one^|uno||three^^^^|four^^^|^cuatro|'
┌───────┬┬───────┬────────────┬┐
│one|uno││three^^│four^|cuatro││
└───────┴┴───────┴────────────┴┘

Here's a somewhat more efficient approach (over 100 times faster on a 100k textual example):

<lang J>tokenize2=: tokenize=:3 :0

 '^|' tokenize2 y  NB. task default escape and separator

 'ESC SEP'=. x
 E=. 18 b./\.&.|.ESC=y NB. escape positions
 S=. (SEP=y)>_1}.0,E NB. separator positions
 K=. -.E+.S NB. keep positions
 T=. (#y){. 1,}.S NB. token beginnings
 (T<;.1 K)#&.>T<;.1 y

)</lang>

Example use:

<lang J> '^|' tokenize 'one^|uno||three^^^^|four^^^|^cuatro|' ┌───────┬┬───────┬────────────┬┐ │one|uno││three^^│four^|cuatro││ └───────┴┴───────┴────────────┴┘</lang>

Solution invoking the sequential machine primitive verb.[this thread.]<lang J>charTokens =: (0;(3 2 2$(2 1 1 1 2 2 1 2 1 0 1 0));<<'^')&;: NB. sequential machine splitTokens =: ((<,'|')&= <;._1 ])@:((<,'|'),]) removeExtra =: (}.^:(1<#)) L:0 tokenize3=: tokenize=: ; each @: (removeExtra @: splitTokens @: charTokens)</lang>Example use:<lang J> t=: 'one^|uno||three^^^^|four^^^|^cuatro|'

 tokenize t

┌───────┬┬───────┬────────────┬┐ │one|uno││three^^│four^|cuatro││ └───────┴┴───────┴────────────┴┘

  $tokenize t

5</lang>

Relative efficiencies:

<lang J> txt=: 1e5$'one^|uno||three^^^^|four^^^|^cuatro|'

  (%"1 <./) timespacex every 'tokenize1 txt';'tokenize2 txt';'tokenize3 txt'

132.856 1

     1 7.73534

8.29568 19.9766</lang>

So tokenize2 is the fastest, while tokenize1 uses the least amount of memory. Also, tokenize1 is the slowest and tokenize3 uses the most memory. (First column is relative time used, second is relative space used, rows correspond to implementations.)

Java

Translation of: Go

Works with: Java version 7

<lang java>import java.util.*;

public class TokenizeStringWithEscaping {

   public static void main(String[] args) {
       String sample = "one^|uno||three^^^^|four^^^|^cuatro|";
       char separator = '|';
       char escape = '^';

       System.out.println(sample);
       try {
           System.out.println(tokenizeString(sample, separator, escape));
       } catch (Exception e) {
           System.out.println(e);
       }
   }

   public static List<String> tokenizeString(String s, char sep, char escape)
           throws Exception {
       List<String> tokens = new ArrayList<>();
       StringBuilder sb = new StringBuilder();

       boolean inEscape = false;
       for (char c : s.toCharArray()) {
           if (inEscape) {
               inEscape = false;
           } else if (c == escape) {
               inEscape = true;
               continue;
           } else if (c == sep) {
               tokens.add(sb.toString());
               sb.setLength(0);
               continue;
           }
           sb.append(c);
       }
       if (inEscape)
           throw new Exception("Invalid terminal escape");

       tokens.add(sb.toString());

       return tokens;
   }

}</lang>

[one|uno, , three^^, four^|cuatro, ]

JavaScript

ES5

Iterative

<lang JavaScript>function tokenize(s, esc, sep) { for (var a=[], t=, i=0, e=s.length; i<e; i+=1) { var c = s.charAt(i) if (c == esc) t+=s.charAt(++i) else if (c != sep) t+=c else a.push(t), t= } a.push(t) return a }

var s = 'one^|uno||three^^^^|four^^^|^cuatro|' document.write(s, '
') for (var a=tokenize(s,'^','|'), i=0; i<a.length; i+=1) document.write(i, ': ', a[i], '
')</lang>

Output:

one^|uno||three^^^^|four^^^|^cuatro|
0: one|uno
1: 
2: three^^
3: four^|cuatro
4:

Functional

<lang JavaScript>(function () {

   'use strict';

   // tokenize :: String -> Character -> Character -> [String]
   function tokenize(str, charDelim, charEsc) {
       var dctParse = str.split()
           .reduce(function (a, x) {

               var blnEsc = a.esc,
                   blnBreak = !blnEsc && x === charDelim,
                   blnEscChar = !blnEsc && x === charEsc;

               return {
                   esc: blnEscChar,
                   token: blnBreak ?  : (
                       a.token + (blnEscChar ?  : x)
                   ),
                   list: a.list.concat(blnBreak ? a.token : [])
               };
           }, {
               esc: false,
               token: ,
               list: []
           });

       return dctParse.list.concat(
           dctParse.token
       );
   }

   return tokenize(
           'one^|uno||three^^^^|four^^^|^cuatro|',
           '|','^'
       )
       .join('\n');

})();</lang>

Output:

one|uno

three^^
four^|cuatro

ES6

Hand-parsed

Translation of: Haskell

(Single fold version)

<lang JavaScript>((() => {

   // tokenize :: String -> Character -> Character -> [String]
   const tokenize = (charDelim, charEsc, str) => {
       const [token, list, _] = str.split()
           .reduce(([aToken, aList, aEsc], x) => {
               const
                   blnBreak = !aEsc && x === charDelim,
                   blnEscChar = !aEsc && x === charEsc;

               return [
                   blnBreak ?  : (
                       aToken + (blnEscChar ?  : x)
                   ),
                   aList.concat(blnBreak ? aToken : []),
                   blnEscChar
               ];
           }, [, [], false]);

       return list.concat(token);
   };

   // splitEsc :: String -> [String]
   const splitEsc = str => tokenize('|', '^', str);

   // TEST
   // show :: a -> String
   const show = x => JSON.stringify(x, null, 2);

   return splitEsc(
           'one^|uno||three^^^^|four^^^|^cuatro|',
       )
       .map(show)
       .join('\n');

}))();</lang>

Output:

"one|uno"
""
"three^^"
"four^|cuatro"
""

Parser combinators

Defining the function as a composition of generics from a parser combinator library:

<lang javascript>(() => {

   'use strict';

   // ------ TOKENIZATION OF A STRING WITH ESCAPES ------

   // tokenizedWithEscapes :: Char -> Char -> 
   // String -> [String]
   const tokenizedWithEscapes = esc =>
       // A list of tokens in a given string, 
       // where the separator character is sep
       // and any character may be escaped by 
       // a preceding esc character.
       sep => compose(
           concatMap(fst),
           parse(
               sepBy(
                   takeWhileEscP(esc)(
                       constant(true)
                   )(
                       ne(sep)
                   )
               )(char(sep))
           )
       );

   // ---------------------- TEST -----------------------
   // main :: IO ()
   const main = () =>
       JSON.stringify(
           tokenizedWithEscapes('^')('|')(
               'one^|uno||three^^^^|four^^^|^cuatro|'
           ),
           null, 2
       );
   // -->
   // [
   //     "one|uno",
   //     "",
   //     "three^^",
   //     "four^|cuatro",
   //     ""
   // ]

   // ----------- GENERIC PARSER COMBINATORS ------------

   // Parser :: String -> [(a, String)] -> Parser a
   const Parser = f =>
       // A function lifted into a Parser object.
       ({
           type: 'Parser',
           parser: f
       });

   // altP (<|>) :: Parser a -> Parser a -> Parser a
   const altP = p =>
       // p, or q if p doesn't match.
       q => Parser(s => {
           const xs = parse(p)(s);
           return 0 < xs.length ? (
               xs
           ) : parse(q)(s);
       });

   // anyChar :: () -> Parser Char
   const anyChar = () =>
       // A single character.
       Parser(
           s => 0 < s.length ? [
               Tuple(s[0])(
                   s.slice(1)
               )
           ] : []
       );

   // apP <*> :: Parser (a -> b) -> Parser a -> Parser b
   const apP = pf =>
       // A new parser obtained by the application 
       // of a Parser-wrapped function,
       // to a Parser-wrapped value.
       p => Parser(
           s => parse(pf)(s).flatMap(
               vr => parse(
                   fmapP(vr[0])(p)
               )(vr[1])
           )
       );

   // bindP (>>=) :: Parser a -> 
   // (a -> Parser b) -> Parser b
   const bindP = p =>
       // A new parser obtained by the application of 
       // a function to a Parser-wrapped value.
       // The function must enrich its output, lifting it 
       // into a new Parser.
       // Allows for the nesting of parsers.
       f => Parser(
           s => parse(p)(s).flatMap(
               tpl => parse(f(tpl[0]))(tpl[1])
           )
       );

   // char :: Char -> Parser Char
   const char = x =>
       // A particular single character.
       satisfy(c => x == c);

   // fmapP :: (a -> b) -> Parser a -> Parser b  
   const fmapP = f =>
       // A new parser derived by the structure-preserving 
       // application of f to the value in p.
       p => Parser(
           s => parse(p)(s).flatMap(
               first(f)
           )
       );

   // liftA2P :: (a -> b -> c) -> 
   // Parser a -> Parser b -> Parser c
   const liftA2P = op =>
       // The binary function op, lifted
       // to a function over two parsers.
       p => apP(fmapP(op)(p));

   // many :: Parser a -> Parser [a]
   const many = p => {
       // Zero or more instances of p.
       // Lifts a parser for a simple type of value 
       // to a parser for a list of such values.
       const some_p = p =>
           liftA2P(
               x => xs => [x].concat(xs)
           )(p)(many(p));
       return Parser(
           s => parse(
               0 < s.length ? (
                   altP(some_p(p))(pureP())
               ) : pureP()
           )(s)
       );
   };

   // parse :: Parser a -> String -> [(a, String)]
   const parse = p =>
       // The result of parsing a string with p.
       p.parser;

   // pureP :: a -> Parser a
   const pureP = x =>
       // The value x lifted, unchanged, 
       // into the Parser monad.
       Parser(s => [Tuple(x)(s)]);

   // satisfy :: (Char -> Bool) -> Parser Char
   const satisfy = test =>
       // Any character for which the 
       // given predicate returns true.
       Parser(
           s => 0 < s.length ? (
               test(s[0]) ? [
                   Tuple(s[0])(s.slice(1))
               ] : []
           ) : []
       );

   // sepBy :: Parser a -> Parser b -> Parser [a]
   const sepBy = p =>
       // Zero or more occurrences of p, as 
       // separated by (discarded) instances of sep.
       sep => altP(
           sepBy1(p)(sep)
       )(
           pureP([])
       );

   // sepBy1 :: Parser a -> Parser b -> Parser [a]
   const sepBy1 = p =>
       // One or more occurrences of p, as 
       // separated by (discarded) instances of sep.
       sep => bindP(
           p
       )(x => bindP(
           many(
               thenP(sep)(
                   bindP(p)(pureP)
               )
           )
       )(xs => pureP([x].concat(xs))));

   // takeWhileEscP :: Char -> (Char -> Bool) -> 
   // (Char -> Bool) -> Parser Text
   const takeWhileEscP = esc =>
       escTest => test => {
           // Longest prefix, including any escaped
           // characters, in which escTest returns
           // true for all escaped characters, and
           // test returns true for all other chars.
           const plain = takeWhileP(
               c => (esc !== c) && test(c)
           );
           const escaped = thenBindP(
               char(esc)
           )(
               anyChar()
           )(x => bindP(
               plain
           )(
               compose(pureP, cons(x))
           ));
           return bindP(
               plain
           )(x => bindP(
               many(escaped)
           )(xs => pureP(concat([x].concat(xs)))));
       };

   // takeWhileP :: (Char -> Bool) -> Parser String
   const takeWhileP = p =>
       // The largest prefix in which p is
       // true over all the characters.
       Parser(
           compose(
               pureList,
               first(concat),
               span(p)
           )
       );

   // thenBindP :: Parser a -> Parser b -> 
   // (b -> Parser c) Parser c
   const thenBindP = o =>
       // A combination of thenP and bindP in which a 
       // preliminary  parser consumes text and discards
       // its output, before any output of a subsequent
       // parser is bound.
       p => f => Parser(
           s => parse(o)(s).flatMap(
               vr => parse(p)(vr[1]).flatMap(
                   tpl => parse(f(tpl[0]))(tpl[1])
               )
           )
       );

   // thenP (>>) :: Parser a -> Parser b -> Parser b
   const thenP = o =>
       // A composite parser in which o just consumes text
       // and then p consumes more and returns a value.
       p => Parser(
           s => parse(o)(s).flatMap(
               vr => parse(p)(vr[1])
           )
       );

   // --------------------- GENERIC ---------------------

   // Tuple (,) :: a -> b -> (a, b)
   const Tuple = a =>
       b => ({
           type: 'Tuple',
           '0': a,
           '1': b,
           length: 2
       });

   // compose (<<<) :: (b -> c) -> (a -> b) -> a -> c
   const compose = (...fs) =>
       // A function defined by the right-to-left
       // composition of all the functions in fs.
       fs.reduce(
           (f, g) => x => f(g(x)),
           x => x
       );

   // concat :: a -> [a]
   // concat :: [String] -> String
   const concat = xs => (
       ys => 0 < ys.length ? (
           ys.every(Array.isArray) ? (
               []
           ) : 
       ).concat(...ys) : ys
   )(list(xs));

   // concatMap :: (a -> [b]) -> [a] -> [b]
   const concatMap = f =>
       // List monad bind operator.
       xs => xs.flatMap(f);

   // cons :: a -> [a] -> [a]
   const cons = x =>
       // A list constructed from the item x,
       // followed by the existing list xs.
       xs => Array.isArray(xs) ? (
           [x].concat(xs)
       ) : 'GeneratorFunction' !== xs
       .constructor.constructor.name ? (
           x + xs
       ) : ( // cons(x)(Generator)
           function* () {
               yield x;
               let nxt = xs.next();
               while (!nxt.done) {
                   yield nxt.value;
                   nxt = xs.next();
               }
           }
       )();

   // constant :: a -> b -> a
   const constant = k =>
       _ => k;

   // first :: (a -> b) -> ((a, c) -> (b, c))
   const first = f =>
       // A simple function lifted to one which applies
       // to a tuple, transforming only its first item.
       xy => Tuple(f(xy[0]))(
           xy[1]
       );

   // fst :: (a, b) -> a
   const fst = tpl =>
       // First member of a pair.
       tpl[0];

   // list :: StringOrArrayLike b => b -> [a]
   const list = xs =>
       // xs itself, if it is an Array,
       // or an Array derived from xs.
       Array.isArray(xs) ? (
           xs
       ) : Array.from(xs || []);

   // map :: (a -> b) -> [a] -> [b]
   const map = f =>
       // The list obtained by applying f
       // to each element of xs.
       // (The image of xs under f).
       xs => [...xs].map(f);

   // ne :: a -> a -> Bool
   const ne = a =>
       b => a !== b;

   // pureList :: a -> [a]
   const pureList = x => [x];

   // span p xs is equivalent to (takeWhile p xs, dropWhile p xs) 
   // span :: (a -> Bool) -> [a] -> ([a], [a])
   const span = p =>
       // Longest prefix of xs consisting of elements which
       // all satisfy p, tupled with the remainder of xs.
       xs => {
           const
               ys = 'string' !== typeof xs ? (
                   list(xs)
               ) : xs,
               iLast = ys.length - 1;
           return splitAt(
               until(
                   i => iLast < i || !p(ys[i])
               )(i => 1 + i)(0)
           )(ys);
       };

   // splitAt :: Int -> [a] -> ([a], [a])
   const splitAt = n =>
       xs => Tuple(xs.slice(0, n))(
           xs.slice(n)
       );

   // unlines :: [String] -> String
   const unlines = xs =>
       // A single string formed by the intercalation
       // of a list of strings with the newline character.
       xs.join('\n');

   // until :: (a -> Bool) -> (a -> a) -> a -> a
   const until = p =>
       f => x => {
           let v = x;
           while (!p(v)) v = f(v);
           return v;
       };

   // MAIN ---
   return main();

})();</lang>

Output:

[
  "one|uno",
  "",
  "three^^",
  "four^|cuatro",
  "",
  ""
]

jq

Works with: jq version 1.5

<lang jq># Tokenize the input using the string "escape" as the prefix escape string def tokenize(separator; escape):

 # Helper functions:
 # mapper/1 is like map/1, but for each element, $e, in the input array,
 # if $e is an array, then it is inserted,
 # otherwise the elements of ($e|f) are inserted.
 def mapper(f): reduce .[] as $e
   ( [];
     if ($e|type) == "array" then . + [$e] else . + ($e | f) end ) ;
 
 # interpolate x
 def interpolate(x):
   reduce .[] as $i ([]; . +  [$i, x]) | .[0:-1];
 
 def splitstring(s; twixt):
   if type == "string" then split(s) | interpolate(twixt)
   else .
   end;

 # concatenate sequences of non-null elements:
 def reform:
   reduce .[] as $x ([];
     if $x == null and .[-1] == null then .[0:-1] + ["", null]
     elif $x == null then . + [null]
     elif .[-1] == null then .[0:-1] + [$x]
     else .[0:-1] +  [ .[-1] + $x ]
     end)
   | if .[-1] == null then .[-1] = "" else . end;
 
 splitstring(escape + escape; [escape])
 | mapper( splitstring( escape + separator; [separator]) )
 | mapper( splitstring( separator; null ) )
 | map( if type == "string" then split(escape) else . end)
 | flatten
 | reform ;</lang>

Example: <lang jq>"one^|uno||three^^^^|four^^^|^cuatro|" | tokenize("|"; "^")</lang>

Output:

<lang sh>$ jq -n -f tokenize.jq [

 "one|uno",
 "",
 "three^^",
 "four^|cuatro",
 ""

]</lang>

Julia

Works with: Julia version 0.6

Translation of: Kotlin

<lang julia>function tokenize2(s::AbstractString, sep::Char, esc::Char)

   SPE = "\ufffe"
   SPF = "\uffff"
   s = replace(s, "$esc$esc", SPE) |>
       s -> replace(s, "$esc$sep", SPF) |>
       s -> last(s) == esc ? string(replace(s[1:end-1], esc, ""), esc) : replace(s, esc, "")
   return map(split(s, sep)) do token
       token = replace(token, SPE, esc)
       return replace(token, SPF, sep)
   end

end

@show tokenize2("one^|uno||three^^^^|four^^^|^cuatro|", '|', '^')</lang>

Output:

tokenize2("one^|uno||three^^^^|four^^^|^cuatro|", '|', '^') = String["one|uno", "", "three^^", "four^|cuatro", ""]

Kotlin

<lang scala>// version 1.1.3

const val SPE = "\ufffe" // unused unicode char in Specials block const val SPF = "\uffff" // ditto

fun tokenize(str: String, sep: Char, esc: Char): List<String> {

   var s = str.replace("$esc$esc", SPE).replace("$esc$sep", SPF)
   s = if (s.last() == esc) // i.e. 'esc' not escaping anything
       s.dropLast(1).replace("$esc", "") + esc
   else 
       s.replace("$esc", "")
   return s.split(sep).map { it.replace(SPE, "$esc").replace(SPF, "$sep") }

}

fun main(args: Array<String>) {

   var str = "one^|uno||three^^^^|four^^^|^cuatro|"
   val sep = '|'
   val esc = '^'
   val items = tokenize(str, sep, esc) 
   for (item in items) println(if (item.isEmpty()) "(empty)" else item)

}</lang>

Output:

one|uno
(empty)
three^^
four^|cuatro
(empty)

Lingo

<lang lingo>-- in some movie script

on tokenize (str, sep, esc)

 l = []
 _player.itemDelimiter = sep
 cnt = str.item.count
 repeat with i = 1 to cnt
   prev = l.getLast() -- can be VOID
   if _trailEscCount(prev, esc) mod 2 then
     l[l.count] = prev.char[1..prev.length-1]&sep&str.item[i]
   else
     l.add(str.item[i])
   end if
 end repeat
 -- remove escape characters from tokens
 cnt = l.count
 repeat with i = 1 to cnt
   l[i] = _removeEsc(l[i], esc)
 end repeat
 return l

end

-- counts number of trailing escape characters on _trailEscCount (str, esc)

 n = 0
 repeat with i = str.length down to 1
   if str.char[i]=esc then n=n+1
   else exit repeat
 end repeat
 return n

end

-- could be implemented more efficiently by using offset() on _removeEsc (str, esc)

 cnt = str.length-1
 repeat with i = 1 to cnt
   if str.char[i]=esc then
     delete char i of str
     cnt = cnt-1
   end if
 end repeat
 return str

end</lang>

<lang lingo>str = "one^|uno||three^^^^|four^^^|^cuatro|" sep = "|" esc = "^" put tokenize(str, sep, esc) -- ["one|uno", "", "three^^", "four^|cuatro", ""]</lang>

Lua

<lang Lua>function tokenise (str, sep, esc)

   local strList, word, escaped, ch = {}, "", false
   for pos = 1, #str do
       ch = str:sub(pos, pos)
       if ch == esc then
           if escaped then
               word = word .. ch
               escaped = false
           else
               escaped = true
           end
       elseif ch == sep then
           if escaped then
               word = word .. ch
               escaped = false
           else
               table.insert(strList, word)
               word = ""
           end
       else
           escaped = false
           word = word .. ch
       end
   end
   table.insert(strList, word)
   return strList

end

local testStr = "one^|uno||three^^^^|four^^^|^cuatro|" local testSep, testEsc = "|", "^" for k, v in pairs(tokenise(testStr, testSep, testEsc)) do

   print(k, v)

end</lang>

Output:

1       one|uno
2
3       three^^
4       four^|cuatro
5

Mathematica / Wolfram Language

<lang Mathematica>ClearAll[Tokenize] Tokenize[str_String, escape_String : "^", sep_String : "|"] :=

Module[{results = {}, token = "", state = 0, a},
 a = Characters[str];
 Do[
  If[state == 0,
   Switch[c,
    escape,
    state = 1
    ,
    sep,
    AppendTo[results, token];
    token = "";
    ,
    _,
    token = token <> c;
    ]
   ,
   If[state == 1,
    token = token <> c;
    state = 0;
    ]
   ]
  
  ,
  {c, a}
  ];
 AppendTo[results, token];
 results
]

Tokenize["one^|uno||three^^^^|four^^^|^cuatro|"]</lang>

Output:

{"one|uno", "", "three^^", "four^|cuatro", ""}

Nim

<lang nim>import streams

proc tokenize(s: Stream, sep: static[char] = '|', esc: static[char] = '^'): seq[string] =

 var buff = ""
 while not s.atEnd():
   let c = s.readChar
   case c
   of sep:
     result.add buff
     buff = ""
   of esc:
     buff.add s.readChar
   else:
     buff.add c
 result.add buff

for i, s in tokenize(newStringStream "one^|uno||three^^^^|four^^^|^cuatro|"):

   echo i, ":", s

</lang>

Output:

0:one|uno
1:
2:three^^
3:four^|cuatro
4:

OCaml

<lang ocaml>let split_with_escaping ~esc ~sep s =

 let len = String.length s in
 let buf = Buffer.create 16 in
 let rec loop i =
   if i = len then [Buffer.contents buf]
   else if s.[i] = esc && i + 1 < len then begin
     Buffer.add_char buf s.[i + 1];
     loop (i + 2)
   end else if s.[i] = sep then begin
     let s = Buffer.contents buf in
     Buffer.clear buf;
     s :: loop (i + 1)
   end else begin
     Buffer.add_char buf s.[i];
     loop (i + 1)
   end
 in
 loop 0</lang>

Example: <lang ocaml>let res = split_with_escaping ~esc:'^' ~sep:'|' "one^|uno||three^^^^|four^^^|^cuatro|";; val res : string list = ["one|uno"; ""; "three^^"; "four^|cuatro"; ""]</lang>

Perl

The built-in split function can be used with a regex that matches the delimiter (although advanced backtracking control verbs are needed to skip escaped delimiters):

<lang perl>sub tokenize {

   my ($string, $sep, $esc) = (shift, quotemeta shift, quotemeta shift);
   
   my @fields = split /$esc . (*SKIP)(*FAIL) | $sep/sx, $string, -1;
   return map { s/$esc(.)/$1/gsr } @fields;

}</lang>

A more traditional approach is to parse the input string step by step (using a repeatedly-matching regex of the form /\G.../g), and throw away the separators (which can be done implicitly using \K):

<lang perl> my @fields = $string =~ /\G (?:^ | $sep) \K (?: [^$sep$esc] | $esc .)*/gsx;</lang>

In both cases, stripping the escape characters happens as a separate step.

Testing:

<lang perl>print "'$_'\n" for tokenize("one^|uno||three^^^^|four^^^|^cuatro|", '|', '^');</lang>

Output:

'one|uno'
''
'three^^'
'four^|cuatro'
''

Phix

function tokenize(string s, integer sep, integer esc)
sequence ret = {}
string word = ""
integer skip = 0
 
    if length(s)!=0 then
        for i=1 to length(s) do
            integer si = s[i]
            if skip then
                word &= si
                skip = 0
            elsif si=esc then
                skip = 1
            elsif si=sep then
                ret = append(ret,word)
                word = ""
            else
                word &= si
            end if
        end for
        ret = append(ret,word)
    end if
    return ret
end function
 
?tokenize("one^|uno||three^^^^|four^^^|^cuatro|",'|','^')

Output:

{"one|uno","","three^^","four^|cuatro",""}

PicoLisp

<lang PicoLisp>(de tokenize (Str Sep Esc)

  (split
     (make
        (for (L (chop Str)  L)
           (let C (pop 'L)
              (cond
                 ((= C Esc) (link (pop 'L)))
                 ((= C Sep) (link 0))
                 (T (link C)) ) ) ) )
     0 ) )</lang>

Test: <lang PicoLisp>(for (I . S) (tokenize "one\^|uno||three\^\^\^\^|four\^\^\^|\^cuatro|" "|" "\^")

  (prinl I ": " S) )</lang>

Output:

1: one|uno
2:
3: three^^
4: four^|cuatro
5:

PowerShell

<lang PowerShell> function Split-String ([string]$String, [char]$Separator, [char]$Escape) {

   if ($String -notmatch "\$Separator|\$Escape") {return $String}

   [bool]$escaping = $false
   [string]$output = ""

   for ($i = 0; $i -lt $String.Length; $i++)
   { 
       [char]$character = $String.Substring($i,1)

       if ($escaping)
       {
           $output += $character
           $escaping = $false
       }
       else
       {
           switch ($character)
           {
               {$_ -eq $Separator} {$output; $output = ""; break}
               {$_ -eq $Escape}    {$escaping = $true    ; break}
               Default             {$output += $character}
           }
       }
   }

   if ($String[-1] -eq $Separator) {[String]::Empty}

} </lang> <lang PowerShell> Split-String "one^|uno||three^^^^|four^^^|^cuatro|" -Separator "|" -Escape "^" | ForEach-Object `

                                                                                       -Begin   {$n = 0} `
                                                                                       -Process {$n+= 1; "{0}: {1}" -f $n, $_}

</lang>

Output:

1: one|uno
2: 
3: three^^
4: four^|cuatro
5:

Python

Procedural

<lang python>def token_with_escape(a, escape = '^', separator = '|'):

   
       Issue  python -m doctest thisfile.py  to run the doctests.

       >>> print(token_with_escape('one^|uno||three^^^^|four^^^|^cuatro|'))
       ['one|uno', , 'three^^', 'four^|cuatro', ]
   
   result = []
   token = 
   state = 0
   for c in a:
       if state == 0:
           if c == escape:
               state = 1
           elif c == separator:
               result.append(token)
               token = 
           else:
               token += c
       elif state == 1:
           token += c
           state = 0
   result.append(token)
   return result</lang>

Functional

Works with: Python version 3

<lang python>Tokenize a string with escaping

from functools import reduce

tokenize :: Char -> Char -> String -> [String]

def tokenize(delim):

   A list of the tokens in a string, given
      a delimiting char and an escape char.
   
   def go(esc, s):
       def chop(a, x):
           tkn, xs, escaped = a
           literal = not escaped
           isEsc = literal and (esc == x)
           return ([], [tkn] + xs, isEsc) if (
               literal and (delim == x)
           ) else (tkn if isEsc else [x] + tkn, xs, isEsc)

       tkn, xs, _ = reduce(chop, list(s), ([], [], False))

       return list(reversed(
           [.join(reversed(x)) for x in [tkn] + xs]
       ))
   return lambda esc: lambda s: go(esc, s)

--------------------------TEST---------------------------
main :: IO ()

def main():

   Test
   
   print(
       tokenize('|')('^')(
           "one^|uno||three^^^^|four^^^|^cuatro|"
       )
   )

MAIN ---

if __name__ == '__main__':

   main()</lang>

Output:

['one|uno', '', 'three^^', 'four^|cuatro', '']

Regex-based

Using `Scanner`

The python re library has a handy class Scanner which is intended precisely for this use-case. It takes a list of pairs regex, action and whenever it encounters regex in the input, it executes action. This allows us to solve this task very efficiently with minimum effort, the hardest part being the correct definition of the regular expressions.

The following code also illustrates an important feature of Python ‒ nested functions with closures. Owing to this feature, the inner functions, such as start_new_token, are able to access the local variable tokens of their enclosing function tokenize. For the inner function, the name tokens is nonlocal, and is in the enclosing scope of the inner function (as opposed to the parameters scanner and substring, which are in the local scope).

<lang python>import re

STRING = 'one^|uno||three^^^^|four^^^|^cuatro|'

def tokenize(string=STRING, escape='^', separator='|'):

   escape, separator = map(re.escape, (escape, separator))

   tokens = []

   def start_new_token(scanner, substring):
       tokens.append()

   def add_escaped_char(scanner, substring):
       char = substring[1]
       tokens[-1] += char

   def add_substring(scanner, substring):
       tokens[-1] += substring

   re.Scanner([
       # an escape followed by a character produces that character
       (fr'{escape}.', add_escaped_char),

       # when encountering a separator not preceded by an escape,
       # start a new token
       (fr'{separator}', start_new_token),

       # a sequence of regular characters (i.e. not escape or separator)
       # is just appended to the token
       (fr'[^{escape}{separator}]+', add_substring),
   ]).scan(string)

   return tokens

if __name__ == '__main__':

   print(list(tokenize()))</lang>

Output is the same as in the functional Python version above.

Simpler version with preprocessing

This version does not require any extra state, such as the token list in the Scanner-based version above. It first preprocesses the input, since Python does not support variable-length lookbehind assertions. Then it works only with the primitive regex operations re.findall and re.sub. Note that the regex used here is compiled with the re.VERBOSE flag. This allows us to write the regex on several lines (since unescaped whitespace is ignored in this mode), and use comments inside the regex (starting with #).

<lang python>import re

STRING = 'one^|uno||three^^^^|four^^^|^cuatro|'

def tokenize(string=STRING, escape='^', separator='|'):

   re_escape, re_separator = map(re.escape, (escape, separator))

   # token regex
   regex = re.compile(fr
       # lookbehind: a token must be preceded by a separator
       # (note that `(?<=^|{re_separator})` doesn't work in Python)
       (?<={re_separator})

       # a token consists either of an escape sequence,
       # or a regular (non-escape, non-separator) character,
       # repeated arbitrarily many times (even zero)
       (?:{re_escape}.|[^{re_escape}{re_separator}])*
     ,
     flags=re.VERBOSE
   )

   # since each token must start with a separator,
   # we must add an extra separator at the beginning of input
   preprocessed_string = separator + string

   for almost_token in regex.findall(preprocessed_string):
     # now get rid of escape characters: '^^' -> '^' etc.
     token = re.sub(fr'{re_escape}(.)', r'\1', almost_token)
     yield token

if __name__ == '__main__':

   print(list(tokenize()))</lang>

Racket

<lang racket>#lang racket/base (require racket/match)

Returns a tokenising function based on sep and esc

(define ((tokenise-with-escape sep esc) str)

 (define tsil->string (compose list->string reverse))
 (define (inr rem l-acc acc)
   (match rem
     ['() (if (and (null? acc) (null? l-acc)) null (reverse (cons (tsil->string l-acc) acc)))]
     [(list (== sep)   tl ...) (inr tl null (cons (tsil->string l-acc) acc))]
     [(list (== esc) c tl ...) (inr tl (cons c l-acc) acc)]
     [(list c          tl ...) (inr tl (cons c l-acc) acc)]))
 (inr (string->list str) null null))

This is the tokeniser that matches the parameters in the task

(define task-tokeniser (tokenise-with-escape #\| #\^))

(define (report-input-output str)

 (printf "Input:  ~s~%Output: ~s~%~%" str (task-tokeniser str)))

(report-input-output "one^|uno||three^^^^|four^^^|^cuatro|") (report-input-output "") (report-input-output "|") (report-input-output "^") (report-input-output ".")</lang>

Output:

Input:  "one^|uno||three^^^^|four^^^|^cuatro|"
Output: ("one|uno" "" "three^^" "four^|cuatro" "")

Input:  ""
Output: ()

Input:  "|"
Output: ("" "")

Input:  "^"
Output: ("^")

Input:  "."
Output: (".")

Raku

(formerly Perl 6)

<lang perl6>sub tokenize ($string, :$sep!, :$esc!) {

   return $string.match(/([ <!before $sep | $esc> . | $esc . ]*)+ % $sep/)\
                 .[0].map(*.subst: /$esc )> ./, , :g);

}

say "'$_'" for tokenize 'one^|uno||three^^^^|four^^^|^cuatro|', sep => '|', esc => '^';</lang>

Output:

'one|uno'
''
'three^^'
'four^|cuatro'
''

Notable Raku innovations that make this different from the equivalent #Perl solution:

string variables can be safely interpolated into regexes without having to 'quotemeta' them
regexes matches return a nested Match object which allows retrieving all results for a given capture group (rather than just the last thing that it matched), thus getting rid of the need for repeated global matching
the <field>+ % <delimiter> regex construct allows handling the delimiters in a more idiomatic way
the )> regex construct can be used to exclude anything that follows it from the returned match result

REXX

IF/THEN logic

<lang rexx>/*REXX program demonstrates tokenizing and displaying a string with escaping sequences. */

 str = 'one^|uno||three^^^^|four^^^|^cuatro|'   /*the character string to be tokenized.*/
 esc = '^'                                      /* "    escape  character to be used.  */
 sep = '|'                                      /* "  separator     "      "  "   "    */
 out =                                          /* "  output string  (so far).         */

eMode = 0 /*a flag, escape is in progress. */

 do j=1  for length(str);  _=substr(str, j, 1)  /*parse a single character at a time.  */
 if eMode   then do; out=out || _;  eMode=0;  iterate;  end   /*are we in escape mode? */
 if _==esc  then do;                eMode=1;  iterate;  end   /*is it an escape char ? */
 if _==sep  then do; call show;               iterate;  end   /* "  " a separator char?*/
 out=out || _                                                 /*append the character.  */
 end   /*j*/

if out\== | _==sep then call show /*handle a residual str or a separator.*/ exit /*stick a fork in it, we're all done. */ /*──────────────────────────────────────────────────────────────────────────────────────*/ show: say '[length'right(length(out),4)"]" out; out=; return</lang> output

[length   7] one|uno
[length   0]
[length   7] three^^
[length  12] four^|cuatro
[length   0]

SELECT logic

This REXX version also shows a scale in the output. <lang rexx>/*REXX program demonstrates tokenizing and displaying a string with escaping sequences. */

 str = 'one^|uno||three^^^^|four^^^|^cuatro|'   /*the character string to be tokenized.*/
 esc = '^'                                      /* "    escape  character to be used.  */
 sep = '|'                                      /* "  separator     "      "  "   "    */
 $   =                                          /* "  output string  (so far).         */

eMode = 0 /*a flag, escape is in progress. */ say ' output len output' /*title verbiage used for the output.*/ say '──────────── ────────────────────' /* " separator " " " " */

 do j=1  for length(str);  _=substr(str, j, 1)  /*parse a single character at a time.  */
     select
     when eMode   then do; $=$ || _;  eMode=0; end      /*are we in in escape  mode?   */
     when _==esc  then                eMode=1           /*is it an  escape  character? */
     when _==sep  then do; call show;          end      /* "  " a separator character? */
     otherwise             $=$ || _                     /*append the single character. */
     end   /*select*/
 end       /*j*/

if $\== | _==sep then call show /*handle a residual str or a separator.*/ say '──────────── ────────────────────' /*the foot separator for the output. */ say ' ····^····1····^····2' /*show the top part of the scale.*/ say ' {scale} 12345678901234567890' /* " " bottom " " " " */ exit /*stick a fork in it, we're all done. */ /*──────────────────────────────────────────────────────────────────────────────────────*/ show: say '[length'right(length($),4)"]" $; $=; return</lang> output

 output len        output
──────────── ────────────────────
[length   7] one|uno
[length   0]
[length   7] three^^
[length  12] four^|cuatro
[length   0]
──────────── ────────────────────
             ····^····1····^····2
  {scale}    12345678901234567890

Ring

<lang ring> tokenize("one^|uno||three^^^^|four^^^|^cuatro|", "|", "^")

func tokenize(src, sep, esc) field = 1 escaping = false see "" + field + " " for i = 1 to len(src)

   char = substr(src, i, 1)
   if escaping 
      see char
      escaping = false
   else
      switch char 
             on sep
                see nl
                field = field + 1
                see "" + field + " "
             on esc
                escaping = true
             other
                see char 
      off
   ok

next see nl </lang> Output:

1 one|uno
2 
3 three^^
4 four^|cuatro
5

Ruby

I had to drop the \K flag and instead drop the seperator at the beginning manually. I am not sure if \K is broken or works different than Perl

Translation of: Perl

<lang ruby> def tokenize(string, sep, esc)

 sep = Regexp.escape(sep)
 esc = Regexp.escape(esc)
 string.scan(/\G (?:^ | #{sep}) (?: [^#{sep}#{esc}] | #{esc} .)*/x).collect do |m|
   m.gsub(/#{esc}(.)/, '\1').gsub(/^#{sep}/, )
 end

end

p tokenize('one^|uno||three^^^^|four^^^|^cuatro|', '|', '^')

</lang>

Rust

<lang rust>const SEPARATOR: char = '|'; const ESCAPE: char = '^'; const STRING: &str = "one^|uno||three^^^^|four^^^|^cuatro|";

fn tokenize(string: &str) -> Vec<String> {

   let mut token = String::new();
   let mut tokens: Vec<String> = Vec::new();
   let mut chars = string.chars();
   while let Some(ch) = chars.next() {
       match ch {
           SEPARATOR => {
               tokens.push(token);
               token = String::new();
           },
           ESCAPE => {
               if let Some(next) = chars.next() {
                   token.push(next);
               }
           },
           _ => token.push(ch),
       }
   }
   tokens.push(token);
   tokens

}

fn main() {

   println!("{:#?}", tokenize(STRING));

}</lang>

Output:

[
    "one|uno",
    "",
    "three^^",
    "four^|cuatro",
    "",
]

Scala

Old fashioned Imperative

Imperative with removed (ugly) mutable variables.

Translation of: Kotlin

<lang Scala>object TokenizeStringWithEscaping0 extends App {

 val (markerSpE,markerSpF) = ("\ufffe" , "\uffff")

 def tokenize(str: String, sep: String, esc: String): Array[String] = {

   val s0 = str.replace( esc + esc, markerSpE).replace(esc + sep, markerSpF)
   val s = if (s0.last.toString == esc) s0.replace(esc, "") + esc else s0.replace(esc, "")
   s.split(sep.head).map (_.replace(markerSpE, esc).replace(markerSpF, sep))
 }

 def str = "one^|uno||three^^^^|four^^^|^cuatro|"

 tokenize(str, "|", "^").foreach(it => println(if (it.isEmpty) "<empty token>" else it))

}</lang>

Idiomatic

Functional with Tail recursion

<lang Scala>import scala.annotation.tailrec

object TokenizeStringWithEscaping1 extends App {

 def tokenize(str: String, sep: String, esc: String): Seq[String] = {
   @tailrec
   def loop(accu: Seq[String], s: String): Seq[String] = {
     def append2StringInList(char: String): Seq[String] =
       accu.init :+ (accu.last + char)

     s.length match {
       case 0 => accu
       case 1 => if (s.head.toString == sep) accu :+ "" else append2StringInList(s)
       case _ => (s.head.toString, s.tail.head.toString) match {
         case c@((`esc`, `sep`) | (`esc`, `esc`)) => loop(append2StringInList(c._2), s.tail.tail)
         case (`sep`, _)                          => loop(accu :+ "", s.tail)
         case (`esc`, _)                          => loop(accu, s.tail)
         case (sub, _)                            => loop(append2StringInList(sub.head.toString), s.tail)
       }
     }
   }

   loop(Seq(""), str)
 }

 def str = "one^|uno||three^^^^|four^^^|^cuatro|"

 tokenize(str, "|", "^")
   .foreach(it =>
     println(
       f"[length:${it.length}%3d] ${if (it.isEmpty) "<empty token>" else it}"))

}</lang>

Output:

See it in running in your browser by ScalaFiddle (JavaScript) or by Scastie (JVM).

Sidef

Translation of: Perl

<lang ruby>func tokenize(string, sep, esc) {

   var fields = string.split(
       Regex(esc.escape + '.(*SKIP)(*FAIL)|' + sep.escape, 's'), -1
   )
   fields.map{.gsub(Regex(esc.escape + '(.)'), {|s1| s1 }) }

}

tokenize("one^|uno||three^^^^|four^^^|^cuatro|", '|', '^').each { |str|

   say str.dump

}</lang>

Output:

"one|uno"
""
"three^^"
"four^|cuatro"
""

Simula

<lang simula> SIMSET BEGIN

   LINK CLASS ITEM(TXT); TEXT TXT;;

   REF(HEAD) PROCEDURE SPLIT(TXT, SEP, ESC); TEXT TXT; CHARACTER SEP, ESC;
   BEGIN
       REF(HEAD) PARTS;
       CHARACTER CH;
       TEXT PART;

       PART :- BLANKS(TXT.LENGTH);
       PARTS :- NEW HEAD;
       TXT.SETPOS(1);
       WHILE TXT.MORE DO BEGIN
           CH := TXT.GETCHAR;
           IF CH = ESC THEN BEGIN
               IF TXT.MORE THEN BEGIN
                   CH := TXT.GETCHAR;
                   PART.PUTCHAR(CH);
               END ELSE BEGIN
                   ERROR("SPLIT: ESCAPE CHAR AT END OF STRING");
               END;
           END ELSE IF CH = SEP THEN BEGIN
               NEW ITEM(COPY(PART.SUB(1,PART.POS-1))).INTO(PARTS);
               PART.SETPOS(1);
           END ELSE BEGIN
               PART.PUTCHAR(CH);
           END;
       END;
       NEW ITEM(COPY(PART.SUB(1,PART.POS-1))).INTO(PARTS);

       SPLIT :- PARTS;
   END SPLIT;

   TEXT EXAMPLE;
   REF(HEAD) RESULT;
   REF(ITEM) PART;
   INTEGER NO;

   FOR EXAMPLE :- "ONE^|UNO||THREE^^^^|FOUR^^^|^CUATRO|" DO
   BEGIN
       OUTTEXT("INPUT: '");
       OUTTEXT(EXAMPLE);
       OUTTEXT("'");
       OUTIMAGE;
       RESULT :- SPLIT(EXAMPLE, '|', '^');
       PART :- RESULT.FIRST;
       NO := 0;
       WHILE PART =/= NONE DO
       BEGIN
           NO := NO + 1;
           OUTTEXT("PART");
           OUTINT(NO, 0);
           OUTTEXT(": '");
           OUTTEXT(PART.TXT);
           OUTTEXT("'");
           OUTIMAGE;
           PART :- PART.SUC;
       END;
   END;

END. </lang>

Output:

INPUT: 'ONE^|UNO||THREE^^^^|FOUR^^^|^CUATRO|'
PART1: 'ONE|UNO'
PART2: ''
PART3: 'THREE^^'
PART4: 'FOUR^|CUATRO'
PART5: ''

Swift

Translation of: Rust

<lang swift>extension String {

 func tokenize(separator: Character, escape: Character) -> [String] {
   var token = ""
   var tokens = [String]()
   var chars = makeIterator()

   while let char = chars.next() {
     switch char {
     case separator:
       tokens.append(token)
       token = ""
     case escape:
       if let next = chars.next() {
         token.append(next)
       }
     case _:
       token.append(char)
     }
   }

   tokens.append(token)

   return tokens
 }

}

print("one^|uno||three^^^^|four^^^|^cuatro|".tokenize(separator: "|", escape: "^"))</lang>

Output:

["one|uno", "", "three^^", "four^|cuatro", ""]

Tcl

Putting a coroutine in a TclOO object following the "generator pattern" gives a nice structure: <lang Tcl>oo::class create tokens {

   constructor {s} {
       puts [coroutine Next my Iter $s]
       oo::objdefine [self] forward next Next
   }
   method Iter {s} {
       yield [info coroutine]
       for {set i 0} {$i < [string length $s]} {incr i} {
           yield [string index $s $i]
       }
       return -code break
   }

}

proc tokenize {s {sep |} {escape ^}} {

   set part ""
   set parts ""
   set iter [tokens new $s]
   while {1} {
       set c [$iter next]
       if {$c eq $escape} {
           append part [$iter next]
       } elseif {$c eq $sep} {
           lappend parts $part
           set part ""
       } else {
           append part $c
       }
   }
   lappend parts $part
   return $parts

}

puts [tokenize one^|uno||three^^^^|four^^^|^cuatro| | ^]</lang>

Output:

one|uno {} three^^ four^|cuatro {}

TMG

Unix TMG: <lang UnixTMG>prog: char(sep) *

       char(esc) *

str: smark token: forw/outp

       ( [ch==esc?] char(ch) any(!<<>>) token
       | [ch==sep?] char(ch) outp str
       | any(!<<>>) token );

outp: parse(( scopy = { <"> 1 <"> * } )); forw: peek/chkeof; peek: [ch=0] char(ch) fail; chkeof: ( [ch?] succ | fail );

ch: 0; sep: 0; esc: 0;</lang>

Input:

|
^
one^|uno||three^^^^|four^^^|^cuatro|

Output:

"one|uno"
""
"three^^"
"four^|cuatro"
""

VBA

Translation of: Phix

<lang vb>Private Function tokenize(s As String, sep As String, esc As String) As Collection

   Dim ret As New Collection
   Dim this As String
   Dim skip As Boolean
    
   If Len(s) <> 0 Then
       For i = 1 To Len(s)
           si = Mid(s, i, 1)
           If skip Then
               this = this & si
               skip = False
           Else
               If si = esc Then
                   skip = True
               Else
                   If si = sep Then
                       ret.Add this
                       this = ""
                   Else
                       this = this & si
                   End If
               End If
           End If
       Next i
       ret.Add this
   End If
   Set tokenize = ret

End Function

Public Sub main()

   Dim out As Collection
   Set out = tokenize("one^|uno||three^^^^|four^^^|^cuatro|", "|", "^")
   Dim outstring() As String
   ReDim outstring(out.Count - 1)
   For i = 0 To out.Count - 1
       outstring(i) = out(i + 1)
   Next i
   Debug.Print Join(outstring, ", ")

End Sub</lang>

Output:

one|uno, , three^^, four^|cuatro,

Vlang

Translation of: Go

<lang ecmascript>fn tokenize_string(s string, sep u8, escape u8) ?[]string {

   mut tokens := []string{}

mut runes := []u8{} mut in_escape := false for r in s { if in_escape { in_escape = false runes << r

       } else if r == escape {

in_escape = true

       } else if r == sep {

tokens << runes.bytestr() runes = runes[..0] } else {

           runes << r
       }

} tokens << runes.bytestr() if in_escape { return error("invalid terminal escape") } return tokens }

const sample = "one^|uno||three^^^^|four^^^|^cuatro|" const separator = `|` const escape = `^` fn main() { println("Input: $sample") tokens := tokenize_string(sample, separator, escape)? println("Tokens: $tokens") }</lang>

Output:

Input:   one^|uno||three^^^^|four^^^|^cuatro|
Tokens: ['one|uno', '', 'three^^', 'four^|cuatro', '']
)

Wren

Translation of: Kotlin

<lang ecmascript>var SPE = "\ufffe" // unused unicode character in Specials block var SPF = "\uffff" // ditto

var tokenize = Fn.new { |str, sep, esc|

   str = str.replace(esc + esc, SPE).replace(esc + sep, SPF)
   str = (str[-1] == esc) ? str[0...-1].replace(esc, "") + esc : str.replace(esc, "")
   return str.split(sep).map { |s| s.replace(SPE, esc).replace(SPF, sep) }.toList

}

var str = "one^|uno||three^^^^|four^^^|^cuatro|" var sep = "|" var esc = "^" var items = tokenize.call(str, sep, esc) for (item in items) System.print((item == "") ? "(empty)" : item)</lang>

Output:

one|uno
(empty)
three^^
four^|cuatro
(empty)

zkl

Two simplifying assumptions (since their behavior is undefined): A string ending with an un-escaped escape is an error and 0xff is not an allowed character in the string. <lang zkl>fcn tokenize(str,sep,esc){

  sink:=Sink(String);
  foreach c in (str){
     switch(c){
        case(esc){ sink.write(__cWalker.next()); }  // error if ^EoS

case(sep){ sink.write("\xff"); } else { sink.write(c) }

     }
  }
  sink.close().split("\xff");

}</lang> Or, if you prefer brevity: <lang zkl>fcn tokenize(str,sep,esc){

  sink:=Sink(String);
  foreach c in (str){ sink.write( (c==esc and __cWalker.next()) or (c==sep and "\xff") or c ) }
  sink.close().split("\xff");

}</lang> <lang zkl>tokenize("one^|uno||three^^^^|four^^^|^cuatro|", "|","^").println();</lang>

Output:

L("one|uno","","three^^","four^|cuatro","")

Revision as of 18:24, 20 May 2022

Deterministic Finite Automaton

Counduit-based solution

Alternative

ES5

Iterative

Functional

ES6

Hand-parsed

Parser combinators

Mathematica / Wolfram Language

Procedural

Functional

Regex-based

Using Scanner

Simpler version with preprocessing

IF/THEN logic

SELECT logic

Old fashioned Imperative

Idiomatic

Functional with Tail recursion

Using `Scanner`