Convert a string into Morse code [closed]

Question

C (131 characters)

Yes, 131!

main(c){for(;c=c?c:(c=toupper(getch())-32)?
"•ƒŒKa`^ZRBCEIQiw#S#nx(37+$6-2&@/4)'18=,*%.:0;?5"
[c-12]-34:-3;c/=2)putch(c/2?46-c%2:0);}

I eeked out a few more characters by combining the logic from the while and for loops into a single for loop, and by moving the declaration of the c variable into the main definition as an input parameter. This latter technique I borrowed from strager's answer to another challenge.

---

For those trying to verify the program with GCC or with ASCII-only editors, you may need the following, slightly longer version:

main(c){for(;c=c?c:(c=toupper(getchar())-32)?c<0?1:
"\x95#\x8CKa`^ZRBCEIQiw#S#nx(37+$6-2&@/4)'18=,*%.:0;?5"
[c-12]-34:-3;c/=2)putchar(c/2?46-c%2:32);}

This version is 17 characters longer (weighing in at a comparatively huge 148), due to the following changes:

• +4: getchar() and putchar() instead of the non-portable getch() and putch()
• +6: escape codes for two of the characters instead of non-ASCII characters
• +1: 32 instead of 0 for space character
• +6: added "c<0?1:" to suppress garbage from characters less than ASCII 32 (namely, from '\n'). You'll still get garbage from any of !"#$%&'()*+[\]^_`{|}~, or anything above ASCII 126.

This should make the code completely portable. Compile with:

gcc -std=c89 -funsigned-char morse.c

The -std=c89 is optional. The -funsigned-char is necessary, though, or you will get garbage for comma and full stop.

---

135 characters

c;main(){while(c=toupper(getch()))for(c=c-32?
"•ƒŒKa`^ZRBCEIQiw#S#nx(37+$6-2&@/4)'18=,*%.:0;?5"
[c-44]-34:-3;c;c/=2)putch(c/2?46-c%2:0);}

In my opinion, this latest version is much more visually appealing, too. And no, it's not portable, and it's no longer protected against out-of-bounds input. It also has a pretty bad UI, taking character-by-character input and converting it to Morse Code and having no exit condition (you have to hit Ctrl+Break). But portable, robust code with a nice UI wasn't a requirement.

A brief-as-possible explanation of the code follows:

main(c){
    while(c = toupper(getch())) /* well, *sort of* an exit condition */
        for(c =
            c - 32 ? // effectively: "if not space character"
            "•ƒŒKa`^ZRBCEIQiw#S#nx(37+$6-2&@/4)'18=,*%.:0;?5"[c - 44] - 34
            /* This array contains a binary representation of the Morse Code
             * for all characters between comma (ASCII 44) and capital Z.
             * The values are offset by 34 to make them all representable
             * without escape codes (as long as chars > 127 are allowed).
             * See explanation after code for encoding format.
             */
            : -3; /* if input char is space, c = -3
                   * this is chosen because -3 % 2 = -1 (and 46 - -1 = 47)
                   * and -3 / 2 / 2 = 0 (with integer truncation)
                   */
            c; /* continue loop while c != 0 */
            c /= 2) /* shift down to the next bit */
                putch(c / 2 ? /* this will be 0 if we're down to our guard bit */
                    46 - c % 2 /* We'll end up with 45 (-), 46 (.), or 47 (/).
                                * It's very convenient that the three characters
                                * we need for this exercise are all consecutive.
                                */
                    : 0 /* we're at the guard bit, output blank space */
                );
}

Each character in the long string in the code contains the encoded Morse Code for one text character. Each bit of the encoded character represents either a dash or a dot. A one represents a dash, and a zero represents a dot. The least significant bit represents the first dash or dot in the Morse Code. A final "guard" bit determines the length of the code. That is, the highest one bit in each encoded character represents end-of-code and is not printed. Without this guard bit, characters with trailing dots couldn't be printed correctly.

For instance, the letter 'L' is ".-.." in Morse Code. To represent this in binary, we need a 0, a 1, and two more 0s, starting with the least significant bit: 0010. Tack one more 1 on for a guard bit, and we have our encoded Morse Code: 10010, or decimal 18. Add the +34 offset to get 52, which is the ASCII value of the character '4'. So the encoded character array has a '4' as the 33rd character (index 32).

This technique is similar to that used to encode characters in ACoolie's, strager's⁽²⁾, Miles's, pingw33n's, Alec's, and Andrea's solutions, but is slightly simpler, requiring only one operation per bit (shifting/dividing), rather than two (shifting/dividing and decrementing).

EDIT:
Reading through the rest of the implementations, I see that Alec and Anon came up with this encoding scheme—using the guard bit—before I did. Anon's solution is particularly interesting, using Python's bin function and stripping off the "0b" prefix and the guard bit with [3:], rather than looping, anding, and shifting, as Alec and I did.

As a bonus, this version also handles hyphen (-....-), slash (-..-.), colon (---...), semicolon (-.-.-.), equals (-...-), and at sign (.--.-.). As long as 8-bit characters are allowed, these characters require no extra code bytes to support. No more characters can be supported with this version without adding length to the code (unless there's Morse Codes for greater/less than signs).

Because I find the old implementations still interesting, and the text has some caveats applicable to this version, I've left the previous content of this post below.

---

Okay, presumably, the user interface can suck, right? So, borrowing from strager, I've replaced gets(), which provides buffered, echoed line input, with getch(), which provides unbuffered, unechoed character input. This means that every character you type gets translated immediately into Morse Code on the screen. Maybe that's cool. It no longer works with either stdin or a command-line argument, but it's pretty damn small.

I've kept the old code below, though, for reference. Here's the new.

New code, with bounds checking, 171 characters:

W(i){i?W(--i/2),putch(46-i%2):0;}c;main(){while(c=toupper(getch())-13)
c=c-19?c>77|c<31?0:W("œ*~*hXPLJIYaeg*****u*.AC5+;79-@6=0/8?F31,2:4BDE"
[c-31]-42):putch(47),putch(0);}

Enter breaks the loop and exits the program.

New code, without bounds checking, 159 characters:

W(i){i?W(--i/2),putch(46-i%2):0;}c;main(){while(c=toupper(getch())-13)
c=c-19?W("œ*~*hXPLJIYaeg*****u*.AC5+;79-@6=0/8?F31,2:4BDE"[c-31]-42):
putch(47),putch(0);}

Below follows the old 196/177 code, with some explanation:

W(i){i?W(--i/2),putch(46-i%2):0;}main(){char*p,c,s[99];gets(s);
for(p=s;*p;)c=*p++,c=toupper(c),c=c-32?c>90|c<44?0:W(
"œ*~*hXPLJIYaeg*****u*.AC5+;79-@6=0/8?F31,2:4BDE"[c-44]-42):
putch(47),putch(0);}

This is based on Andrea's Python answer, using the same technique for generating the morse code as in that answer. But instead of storing the encodable characters one after another and finding their indexes, I stored the indexes one after another and look them up by character (similarly to my earlier answer). This prevents the long gaps near the end that caused problems for earlier implementors.

As before, I've used a character that's greater than 127. Converting it to ASCII-only adds 3 characters. The first character of the long string must be replaced with \x9C. The offset is necessary this time, otherwise a large number of characters are under 32, and must be represented with escape codes.

Also as before, processing a command-line argument instead of stdin adds 2 characters, and using a real space character between codes adds 1 character.

On the other hand, some of the other routines here don't deal with input outside the accepted range of [ ,.0-9\?A-Za-z]. If such handling were removed from this routine, then 19 characters could be removed, bringing the total down as low as 177 characters. But if this is done, and invalid input is fed to this program, it may crash and burn.

The code in this case could be:

W(i){i?W(--i/2),putch(46-i%2):0;}main(){char*p,s[99];gets(s);
for(p=s;*p;p++)*p=*p-32?W(
"œ*~*hXPLJIYaeg*****u*.AC5+;79-@6=0/8?F31,2:4BDE"
[toupper(*p)-44]-42):putch(47),putch(0);}

Using a Morse Code Font?

Console.Write(params[0]);

Perl, 170 characters (with a little help from accomplished golfer mauke). Wrapped for clarity; all newlines are removable.

$_=uc<>;y,. ,|/,;s/./$& /g;@m{A..Z,0..9,qw(| , ?)}=
".-NINNN..]IN-NII..AMN-AI---.M-ANMAA.I.-].AIAA-NANMMIOMAOUMSMSAH.B.MSOIONARZMIZ"
=~/../g;1while s![]\w|,?]!$m{$&}!;print

Explanation:

1. Extract the morse dictionary. Each symbol is defined in terms of two chars, which can be either literal dots or dashes, or a reference to the value of another defined char. E and T contain dummy chars to avoid desyncing the decoder; we'll remove them later.
2. Read and format the input. "Hello world" becomes "H E L L O / W O R L D"
3. The next step depends on the input and output dictionaries being distinct, so turn dots in the input to an unused char (vertical bar, |)
4. Replace any char in the input that occurs in the morse dictionary with its value in the dictionary, until no replacements occur.
5. Remove the dummy char mentioned in step 1.
6. Print the output.

In the final version, the dictionary is optimized for runtime efficiency:

• All one-symbol characters (E and T) and two-symbol characters (A, I, M, and N) are defined directly and decode in one pass.
• All three-symbol characters are defined in terms of a two-symbol character and a literal symbol, decoding in two passes.
• All four-symbol characters are defined in terms of two two-symbol characters, decoding in two passes with three replacements.
• The five- and six-symbol characters (numbers and punctuation) decode in three passes, with four or five replacements respectively.

Since the golfed code only replaces one character per loop (to save one character of code!) the number of loops is limited to five times the length of the input (three times the length of the input if only alphabetics are used). But by adding a g to the s/// operation, the number of loops is limited to three (two if only alphabetics are used).

Example transformation:

Hello 123
H E L L O / 1 2 3
II .] AI AI M- / AO UM SM
.... . .-.. .-.. --- / .-M- .A-- I.--
.... . .-.. .-.. --- / .---- ..--- ...--

Python list comprehension, 159-character one-liner

for c in raw_input().upper():print c<","and"/"or bin(ord("•ƒwTaQIECBRZ^`šŒ#S#n|':<.$402&9/6)(18?,*%+3-;=>"[ord(c)-44])-34)[3:].translate(" "*47+"/.-"+" "*206),

Uses the similar data packing to P Daddy's C implementation, but does not store the bits in reverse order and uses bin() to extract the data rather than arithmetic. Note also that spaces are detected using inequality; it considers every character "less than comma" to be a space.

Python for loop, 205 chars including newlines

for a in raw_input().upper():
 q='_ETIANMSURWDKGOHVF_L_PJBXCYZQ__54_3___2__+____16=/_____7___8_90'.find(a);s=''
 while q>0:s='-.'[q%2]+s;q=~-q/2
 print['/','--..--','..--..','.-.-.-',''][' ,?.'.find(a)]+s,

I was dorking around with a compact coding for the symbols, but I don't see if getting any better than the implicit trees already in use, so I present the coding here in case some one else can use it.

Consider the string:

 --..--..-.-.-..--...----.....-----.--/

which contains all the needed sequences as substrings. We could code the symbols by offset and length like this:

       ET  RRRIIGGGJJJJ    
--..--..-.-.-..--...----.....-----.--/
          CCCC  DD WWW       00000
,,,,,,   AALLLL BBBB        11111
--..--..-.-.-..--...----.....-----.--/
  ??????  KKK  MMSSS       22222   
        FFFF  PPPP        33333
--..--..-.-.-..--...----.....-----.--/
        UUU XXXX         44444       
          NN  PPPP  OOO 55555
--..--..-.-.-..--...----.....-----.--/
               ZZZZ    66666
                      77777      YYYY
--..--..-.-.-..--...----.....-----.--/
       ......        88888 HHHH
                    99999 VVVV  QQQQ
--..--..-.-.-..--...----.....-----.--/

with the space (i.e. word boundary) starting and ending on the final character (the '/'). Feel free to use it, if you see a good way.

Most of the shorter symbols have several possible codings, of course.

---

P Daddy found a shorter version of this trick (and I can now see at least some of the redundancy here) and did a nice c implementation. Alec did a python implementation with the first (buggy and incomplete) version. Hobbs did a pretty compact perl version that I don't understand at all.

J, 124 130 134 characters

'.- /'{~;2,~&.>(]`(<&3:)@.(a:=])"0)}.&,&#:&.></.40-~a.i.')}ggWOKIHX`dfggggggg-@B4*:68,?5</.7>E20+193ACD'{~0>.45-~a.i.toupper

J beats C! Awesome!

Usage:

   '.- /'{~;2,~&.>(]`(<&3:)@.(a:=])"0)}.&,&#:&.></.40-~a.i.')}ggWOKIHX`dfggggggg-@B4*:68,?5</.7>E20+193ACD'{~0>.45-~a.i.toupper 'Hello World'
.... . .-.. .-.. --- / .-- --- .-. .-.. -.. 

   '.- /'{~;2,~&.>(]`(<&3:)@.(a:=])"0)}.&,&#:&.></.40-~a.i.')}ggWOKIHX`dfggggggg-@B4*:68,?5</.7>E20+193ACD'{~0>.45-~a.i.toupper 'Hello, Stackoverflow.'
.... . .-.. .-.. --- .-.-.- / ... - .- -.-. -.- --- ...- . .-. ..-. .-.. --- .-- --..-- 

Python 3 One Liner: 172 characters

print(' '.join('/'if c==' 'else''.join('.'if x=='0'else'-'for x in bin(ord("ijÁĕÁÿïçãáàðøüþÁÁÁÁÁČÁÅ×ÚÌÂÒÎÐÄ×ÍÔÇÆÏÖÝÊÈÃÉÑËÙÛÜ"[ord(c)-44])-192)[3:])for c in input().upper()))

(Encoding the tranlation table into unicode code points. Works fine, and they display here fine in my test on my Windows Vista machine.)

Edited to pare down to 184 characters by removing some unnecessary spaces and brackets (making list comps gen exps).

Edit again: More spaces removed that I didn't even know was possible before seeing other answers here - so down to 176.

Edit again down to 172 (woo woo!) by using ' '.join instead of ''.join and doing the spaces separately. (duh!)