Perl 5 -p, 100 bytes

$a="{,{one,two,three,four,five,six,seven,eight,nine}";$_=(<"$a-hundred }$a-ten }$a}">)[$_];s/one-//g 

Try it online!

Readable version:

#!perl -pl $a="{,{one,two,three,four,five,six,seven,eight,nine}"; $_=(<"$a-hundred }$a-ten }$a}">)[$_]; s/one-//g 

The idea here is simple: instead of trying to generate the number on the fly, generate a list of all 2050 numbers using a glob expression (which is much easier) and then index into that list. If $n is the list of numbers, we construct the glob {,{$n}-hundred }{,{$n}-ten }{,{$n}}. The only problem is that this glob expression will generate one-ten and one-hundred, which we can simply strip at the end with s/one-//g.

Python + num2words, 162 138 137 bytes

lambda N:' '.join([((num2words(n)+'-'*(w>''),'')[n=='1'<w]+w)for n,w in zip('%3s'%N,['hundred','ten',''])if'0'<n]) from num2words import*

Attempt This Online!

Requires the package num2words to be installed, so won't run in the ATO link.

-24 bytes thans to Jonathan Allan

-1 byte, replaced '%03s'%N with '%3s'%N

Charcoal, 74 bytes

⪫Φ⮌Ｅ⮌Ｓ⪫Φ⟦§⪪”↶⌈↶？#λ<∕CＳＧ∨⮌№¤"~^hφ﹪▶U⊕hU⦄w↷” ∧∨⊖ι¬κＩι§⪪”↶↧WaＫπ⁵1✂” ∧Ｉικ⟧λ-ι 

Try it online! Link is to verbose version of code. Explanation:

 Ｓ Input as a string ⮌ Reversed Ｅ Map over digits ⟦ ⟧ Tuple of ”...” Compressed digits string ⪪ Split on spaces § Indexed by ι Current digit ⊖ Decremented ∨ Logical Or κ Current power of 10 ¬ Is zero ∧ Logical And ι Current digit Ｉ Cast to integer ”...” Compressed places string ⪪ Split on spaces § Indexed by ι Current digit Ｉ Cast to integer ∧ Logical And κ Current power of 10 Φ Filtered where λ String is not empty ⪫ Joined with - Literal string `-` ⮌ Reversed Φ Filtered where ι String is not empty ⪫ Joined with spaces Implicitly print 

C (gcc), 193 171 169 bytes

• -2 thanks to ceilingcat

Recursively divides the input number, looking up each digit (except for 1 in non-ones position and 0) and and appending the suffix (except for ones position.)

g(v,s,i){v&&g(v/10,s?"hundred":"ten"),(i=v%10)&&printf("%3$.5s%2$s%s "+(i<2&&s)*10,s?:"",!s+"-","one\0 two\0 threefour\0five\0six\0 seveneightnine"+~-i*5);}f(v){g(v,0);} 

Try it online!

Factor + math.text.english math.text.utils, 138 136 126 bytes

[ 1 digit-groups [ 10^ [ number>text ] bi@ "-"glue R/ one[- ]|-one$|zero.*/ ""re-replace ] map-index reverse harvest " "join ] 

Try it online!

Explanation

Code | Data stack | Comment ============================================================================================================ ! 478 1 digit-groups ! { 8 7 4 } <<inside the map-index quotation now; first iteration>> ! 8 0 (digit, index) 10^ ! 8 1 [ number>text ] bi@ ! "eight" "one" "-"glue ! "eight-one" R/ one[- ]|-one$|zero.*/ ! "eight-one" R/ one[- ]|-one$|zero.*/ (push a regexp literal on the data stack) ""re-replace ! "eight" <<outside the map-index quotation now>> map-index ! { "eight" "seven-ten" "four-hundred" } reverse ! { "four-hundred" "seven-ten" "eight" } harvest ! { "four-hundred" "seven-ten" "eight" } (remove empty strings) " "join ! "four-hundred seven-ten eight" 

JavaScript (V8), 162 153 151 148 bytes

-3 bytes thanks to Arnauld.
-6 bytes by merging the handling of 10's and 100's.
-2 bytes thanks to Arnauld, again.
-3 bytes by moving the floor operation, thanks to Shaggy.

f=(n,p)=>n>9?f(n/(m=n>99?100:10)|0,m%4?'ten ':'hundred ')+f(n%m):n?--n|!p?'one,two,three,four,five,six,seven,eight,nine'.split`,`[n]+[p&&'-'+p]:p:'' 

Try it online!

Ungolfed

f = (n, p) => n > 9 ? // check if we should handle 10's or 100's f( n / (m = n > 99 ? 100 : 10) | 0, // check what mod to use, then floor divide m % 4 ? 'ten ' : 'hundred ' // check what place to use ) + f(n % m) : n ? // if 9 >= n > 0 --n | !p ? // if n > 1 or p ('ten' or 'hundred') is not defined we need to find the name of n 'one,two,three,four,five,six,seven,eight,nine'.split`,`[n] + [p && '-' + p] : // add p if we have it p : // return p if n = 1 '' // return '' if n = 0 

Python, 198 188 bytes

lambda m:" ".join([r[int(i)]for i,r in zip('%3s'%m,[[c+s for c in",one,two,three,four,five,six,seven,eight,nine".split(",")]for s in['-hundred','-ten','']])if"0"<i>" "]).replace("one-","")

Attempt This Online!

Inspired by Sisyphus's answer. Generates three lists of the number string for each of the hundreds, tens and ones. Indexes each list by it's respective value in the input, then gets rid of one-

JavaScript (Node.js), 147 bytes

n=>['hundred','ten',''].flatMap(t=>(c=',,two,three,four,five,six,seven,eight,nine'.split`,`[p=(n*=10)/1e3%10|0])?t?c+'-'+t:c:p?t||'one':[]).join` ` 

Try it online!

C (gcc), 223 bytes

o,i,z;main(a,c)char**c;{for(o=strlen(c[1]);z=c[1][i]-49,printf("%s%s%s ",z|o-i==1?"zero\0one\0two\0three\0four\0five\0six\0seven\0eight\0nine"+"059=CHMQW]"[z+1]-48:"",z&&o+~i?"-":"",o-i-2?o-i-3?"":"hundred":"ten"),++i-o;);} 

Try it online!

Thanks to Mukul Kumar for basis, fixed what I perceived to be incorrect output, and shaved down by switching to C and embracing the sketchiness of golf C.

Updates based on issues found and golf contributed by Arnauld

Thanks to ceilingcat for some more C golf sketchiness.

Python 3.8, no library, 184 bytes

Late to the party, but here's my answer. jezza_99 answer using library is unbeatable!

n=input() for i,o in zip([' one two three four five six seven eight nine'.split(' ')[int(d)]for d in n],['-hundred','-ten',''][-len(n):]): if i:print((i+o).replace('one-',''),end=' ') 

Try it online!

Commented code

# input: get number to be rewritten in words # n will be a string n = input('Input number n to be translated into words (0 < n < 1000): ') # list of literal digit # specify split(' ') to have empty string corresponding to 0 literal_digit = ' one two three four five six seven eight nine'.split(' ') # we will get # ['', 'one', 'two', 'three', 'four', 'five', 'six', 'seven', 'eight', 'nine'] # list of literal digit corresponding to the input number # expanded list comprehension [literal_digit[int(d)] for d in n] literal_number = [] for d in n: literal_number.append(literal_digit[int(d)]) # for example '438' -> ['four', 'three', 'eight'] # list of literal orders of magnitude # we have to consider only the elements according to the length of the number ord_mag = ['-hundred', '-ten', ''][-len(n):] # for example '43' -> len = 2 -> start at -2 # we will get only ['-ten', ''] # build the number string translated into words # loop over the literal number and the corresponding order of magnitude for i, o in zip(literal_number, ord_mag): # check if digit is 0 -> '': don't print if i: # not 0 # we must print the literal digit and its order of magnitude string = i + o # if digit is one # we must print only the order of magnitude, # otherwise only 'one' if the magnitude is unit string = string.replace('one-', '') # this can be done because the '-' is defined in the order of magnitude # print the string, concatenating the digit on the same line print(string, end=' ') 

05AB1E, 53 bytes

“0€µ‚•„í†ìˆÈŒšï¿Ÿ¯¥Š“#ISèāR… —¿°¡#sèøʒнa}€á'-ýðý„€µ-K 

Try it online or verify all test cases.

Explanation:

“0€µ‚•„í†ìˆÈŒšï¿Ÿ¯¥Š“ # Push dictionary string "0 one two ... eight nine" # # Split it on spaces I # Push the input-integer S # Convert it to a list of digits è # Index those into the list ā # Push a list in the range [1,length] (without popping) R # Reverse it to range [length,1] … —¿°¡ # Push dictionary string " ten hundred" # # Split it on spaces: ["","","ten","hundred"] s # Swap so the [length,1] integer-list is at the top è # 0-based index into the ["","","ten","hundred"] ø # Create pairs of the two lists ʒ # Filter, to only keep the pairs where: н # the first item a # only contains letters # (this removes ["0","hundred"], ["0","ten"], and ["0",""] pairs) }€ # After the filter: map over each remaining pair: á # Only keep strings consisting solely of letters # (this removes the "" from the last pair) '-ý '# Join each inner pair with "-" delimiter ðý # Then join these strings with " " delimiter „€µ- # Push dictionary string "one-" K # Remove all those substrings # (after which the result is output implicitly) 

See this 05AB1E tip of mine (section How to use the dictionary?) to understand why “0€µ‚•„í†ìˆÈŒšï¿Ÿ¯¥Š“ is "0 one two three four five six seven eight nine"; … —¿°¡ is " ten hundred"; and „€µ- is "one-".

Retina 0.8.2, 102 bytes

\B -ten¶ ten(¶..) hundred$1 A`0 ¶   1- 1 one 2 two 3 three 4 four 5 five 6 six 7 seven 8 eight 9 nine 

Try it online! Link includes test cases. Explanation:

\B -ten¶ 

If there is more than one digit then assume the leading digit is a ten. Also split the digits onto their own line.

ten(¶..) hundred$1 

If there's a second digit then correct it to a hundred.

A`0 

Delete any lines containing a 0.

¶   

Join the lines together again.

1- 

10 and 100 should just be ten and hundred.

1 one 2 two 3 three 4 four 5 five 6 six 7 seven 8 eight 9 nine 

Turn the remaining digits into words.

Bonus: Here's a version that works up to 9999 for 116 bytes: Try it online! Link includes test cases. (Sadly the three digit version is 103 bytes.)

Python 2, 209 bytes

i=input();t,n="0 one two three four five six seven eight nine".split(),map(int,"00"+`i`)[-3:] print ("",t[n[0]]+"-hundred ")[n[0]and i>99]+("",(t[n[1]]+"-","")[n[1]<2]+"ten ")[n[1]and i>9]+("",t[n[2]])[n[2]>0] 

Try it online!

Python 2 because:

1. Accepts integers directly as input

2. `` converts integer to string

3. Can directly index the result of map without converting to a list

4. No () required by the print

PHP, 195 bytes

<?php foreach(str_split(substr("00".fgets(STDIN),-3))as$i=>$c)echo trim($c>0?[0,$i>1?"one":"","two","three","four","five","six","seven","eight","nine"][$c].["-hundred","-ten",""][$i]." ":"","-"); 

Input via STDIN, output via echo

Try it online!

De-golf:

<?php $inp = substr("00".fgets(STDIN),-3); // here I take input and append leading zeros, so as the total lengh is 3 // (for example, 13 becomes 013) $arr = str_split($inp); // string to array of chars foreach($arr as $index => $digit){ $num_array = [0, $index > 1 ? "one":"","two","three","four","five","six","seven","eight","nine"]; // here one is hidden unless it's on the last position $name_array = ["-hundred","-ten",""]; $chunk = $num_array[$digit].$name_array[$index]." "; $chunk = $digit > 0 ? $chunk : ""; // empty string if it's zero $trimmed = trim($chunk, "-"); // to dispose of "-" at the beginning of the string (this could happen if we drop "one" in "one-ten", so just "-ten" left) echo $trimmed; } ?> 

Fun note - I didn't expect PHP doesn't complain here:

C++, 277 bytes

#include <iostream> main(int a,char*c[]){char s[9];int q[]={0,4,7,10,15,19,23,26,31,36,40},n,m,o=strlen(c[1]),i=0;do{m=c[1][i]-48;n=q[m];strcpy_s(s,"zeroonetwothreefourfivesixseveneightnine"+n,q[m+1]-n);std::cout<<((o-i==2)?"ten-":(o-i==3)?"hundred-":"")<<s<<' ';}while(++i<o);} 

Ungolfed

#include <iostream> int main(int a, char* c[]) { char s[9]; int q[] = { 0,4,7,10,15,19,23,26,31,36,40 }, n, m, o = strlen(c[1]), i = 0; do { m = c[1][i] - 48; n = q[m]; strncpy_s(s, 9, "zeroonetwothreefourfivesixseveneightnine" + n, q[m + 1] - n); std::cout << ((o - i == 2) ? "ten-" : (o - i == 3) ? "hundred-" : "") << s << ' '; } while (++i < o); } 

Input comes from command-line arguments

Reason for using C++ over C

• apparantly some compilers include string.h header in iostream

Julia, ¹⁸⁵ 158 bytes

n+m+s=!(n÷m)*"-$s "*!(n%m) !n=replace(n<10 ? split(" one two three four five six seven eight nine"," ")[-~n] : n<100 ? n+10+:ten : n+100+:hundred,"one-"=>"")

Attempt This Online!