• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 // ImGui - binary_to_compressed_c.cpp
2 // Helper tool to turn a file into a C array, if you want to embed font data in your source code.
3 
4 // The data is first compressed with stb_compress() to reduce source code size,
5 // then encoded in Base85 to fit in a string so we can fit roughly 4 bytes of compressed data into 5 bytes of source code (suggested by @mmalex)
6 // (If we used 32-bits constants it would require take 11 bytes of source code to encode 4 bytes, and be endianness dependent)
7 // Note that even with compression, the output array is likely to be bigger than the binary file..
8 // Load compressed TTF fonts with ImGui::GetIO().Fonts->AddFontFromMemoryCompressedTTF()
9 
10 // Build with, e.g:
11 //   # cl.exe binary_to_compressed_c.cpp
12 //   # gcc binary_to_compressed_c.cpp
13 // You can also find a precompiled Windows binary in the binary/demo package available from https://github.com/ocornut/imgui
14 
15 // Usage:
16 //   binary_to_compressed_c.exe [-base85] [-nocompress] <inputfile> <symbolname>
17 // Usage example:
18 //   # binary_to_compressed_c.exe myfont.ttf MyFont > myfont.cpp
19 //   # binary_to_compressed_c.exe -base85 myfont.ttf MyFont > myfont.cpp
20 
21 #define _CRT_SECURE_NO_WARNINGS
22 #include <stdio.h>
23 #include <string.h>
24 #include <stdlib.h>
25 #include <assert.h>
26 
27 // stb_compress* from stb.h - declaration
28 typedef unsigned int stb_uint;
29 typedef unsigned char stb_uchar;
30 stb_uint stb_compress(stb_uchar *out,stb_uchar *in,stb_uint len);
31 
32 static bool binary_to_compressed_c(const char* filename, const char* symbol, bool use_base85_encoding, bool use_compression);
33 
main(int argc,char ** argv)34 int main(int argc, char** argv)
35 {
36     if (argc < 3)
37     {
38         printf("Syntax: %s [-base85] [-nocompress] <inputfile> <symbolname>\n", argv[0]);
39         return 0;
40     }
41 
42     int argn = 1;
43     bool use_base85_encoding = false;
44     bool use_compression = true;
45     if (argv[argn][0] == '-')
46     {
47         if (strcmp(argv[argn], "-base85") == 0) { use_base85_encoding = true; argn++; }
48         else if (strcmp(argv[argn], "-nocompress") == 0) { use_compression = false; argn++; }
49         else
50         {
51             printf("Unknown argument: '%s'\n", argv[argn]);
52             return 1;
53         }
54     }
55 
56     return binary_to_compressed_c(argv[argn], argv[argn+1], use_base85_encoding, use_compression) ? 0 : 1;
57 }
58 
Encode85Byte(unsigned int x)59 char Encode85Byte(unsigned int x)
60 {
61     x = (x % 85) + 35;
62     return (x>='\\') ? x+1 : x;
63 }
64 
binary_to_compressed_c(const char * filename,const char * symbol,bool use_base85_encoding,bool use_compression)65 bool binary_to_compressed_c(const char* filename, const char* symbol, bool use_base85_encoding, bool use_compression)
66 {
67     // Read file
68     FILE* f = fopen(filename, "rb");
69     if (!f) return false;
70     int data_sz;
71     if (fseek(f, 0, SEEK_END) || (data_sz = (int)ftell(f)) == -1 || fseek(f, 0, SEEK_SET)) { fclose(f); return false; }
72     char* data = new char[data_sz+4];
73     if (fread(data, 1, data_sz, f) != (size_t)data_sz) { fclose(f); delete[] data; return false; }
74     memset((void*)(((char*)data) + data_sz), 0, 4);
75     fclose(f);
76 
77     // Compress
78     int maxlen = data_sz + 512 + (data_sz >> 2) + sizeof(int); // total guess
79     char* compressed = use_compression ? new char[maxlen] : data;
80     int compressed_sz = use_compression ? stb_compress((stb_uchar*)compressed, (stb_uchar*)data, data_sz) : data_sz;
81     if (use_compression)
82 		memset(compressed + compressed_sz, 0, maxlen - compressed_sz);
83 
84     // Output as Base85 encoded
85     FILE* out = stdout;
86     fprintf(out, "// File: '%s' (%d bytes)\n", filename, (int)data_sz);
87     fprintf(out, "// Exported using binary_to_compressed_c.cpp\n");
88 	const char* compressed_str = use_compression ? "compressed_" : "";
89     if (use_base85_encoding)
90     {
91         fprintf(out, "static const char %s_%sdata_base85[%d+1] =\n    \"", symbol, compressed_str, (int)((compressed_sz+3)/4)*5);
92         char prev_c = 0;
93         for (int src_i = 0; src_i < compressed_sz; src_i += 4)
94         {
95             // This is made a little more complicated by the fact that ??X sequences are interpreted as trigraphs by old C/C++ compilers. So we need to escape pairs of ??.
96             unsigned int d = *(unsigned int*)(compressed + src_i);
97             for (unsigned int n5 = 0; n5 < 5; n5++, d /= 85)
98             {
99                 char c = Encode85Byte(d);
100                 fprintf(out, (c == '?' && prev_c == '?') ? "\\%c" : "%c", c);
101                 prev_c = c;
102             }
103             if ((src_i % 112) == 112-4)
104                 fprintf(out, "\"\n    \"");
105         }
106         fprintf(out, "\";\n\n");
107     }
108     else
109     {
110         fprintf(out, "static const unsigned int %s_%ssize = %d;\n", symbol, compressed_str, (int)compressed_sz);
111         fprintf(out, "static const unsigned int %s_%sdata[%d/4] =\n{", symbol, compressed_str, (int)((compressed_sz+3)/4)*4);
112         int column = 0;
113         for (int i = 0; i < compressed_sz; i += 4)
114         {
115             unsigned int d = *(unsigned int*)(compressed + i);
116             if ((column++ % 12) == 0)
117                 fprintf(out, "\n    0x%08x, ", d);
118             else
119                 fprintf(out, "0x%08x, ", d);
120         }
121         fprintf(out, "\n};\n\n");
122     }
123 
124     // Cleanup
125     delete[] data;
126     if (use_compression)
127 	    delete[] compressed;
128     return true;
129 }
130 
131 // stb_compress* from stb.h - definition
132 
133 ////////////////////           compressor         ///////////////////////
134 
stb_adler32(stb_uint adler32,stb_uchar * buffer,stb_uint buflen)135 static stb_uint stb_adler32(stb_uint adler32, stb_uchar *buffer, stb_uint buflen)
136 {
137     const unsigned long ADLER_MOD = 65521;
138     unsigned long s1 = adler32 & 0xffff, s2 = adler32 >> 16;
139     unsigned long blocklen, i;
140 
141     blocklen = buflen % 5552;
142     while (buflen) {
143         for (i=0; i + 7 < blocklen; i += 8) {
144             s1 += buffer[0], s2 += s1;
145             s1 += buffer[1], s2 += s1;
146             s1 += buffer[2], s2 += s1;
147             s1 += buffer[3], s2 += s1;
148             s1 += buffer[4], s2 += s1;
149             s1 += buffer[5], s2 += s1;
150             s1 += buffer[6], s2 += s1;
151             s1 += buffer[7], s2 += s1;
152 
153             buffer += 8;
154         }
155 
156         for (; i < blocklen; ++i)
157             s1 += *buffer++, s2 += s1;
158 
159         s1 %= ADLER_MOD, s2 %= ADLER_MOD;
160         buflen -= blocklen;
161         blocklen = 5552;
162     }
163     return (s2 << 16) + s1;
164 }
165 
stb_matchlen(stb_uchar * m1,stb_uchar * m2,stb_uint maxlen)166 static unsigned int stb_matchlen(stb_uchar *m1, stb_uchar *m2, stb_uint maxlen)
167 {
168     stb_uint i;
169     for (i=0; i < maxlen; ++i)
170         if (m1[i] != m2[i]) return i;
171     return i;
172 }
173 
174 // simple implementation that just takes the source data in a big block
175 
176 static stb_uchar *stb__out;
177 static FILE      *stb__outfile;
178 static stb_uint   stb__outbytes;
179 
stb__write(unsigned char v)180 static void stb__write(unsigned char v)
181 {
182     fputc(v, stb__outfile);
183     ++stb__outbytes;
184 }
185 
186 //#define stb_out(v)    (stb__out ? *stb__out++ = (stb_uchar) (v) : stb__write((stb_uchar) (v)))
187 #define stb_out(v)    do { if (stb__out) *stb__out++ = (stb_uchar) (v); else stb__write((stb_uchar) (v)); } while (0)
188 
stb_out2(stb_uint v)189 static void stb_out2(stb_uint v) { stb_out(v >> 8); stb_out(v); }
stb_out3(stb_uint v)190 static void stb_out3(stb_uint v) { stb_out(v >> 16); stb_out(v >> 8); stb_out(v); }
stb_out4(stb_uint v)191 static void stb_out4(stb_uint v) { stb_out(v >> 24); stb_out(v >> 16); stb_out(v >> 8 ); stb_out(v); }
192 
outliterals(stb_uchar * in,int numlit)193 static void outliterals(stb_uchar *in, int numlit)
194 {
195     while (numlit > 65536) {
196         outliterals(in,65536);
197         in     += 65536;
198         numlit -= 65536;
199     }
200 
201     if      (numlit ==     0)    ;
202     else if (numlit <=    32)    stb_out (0x000020 + numlit-1);
203     else if (numlit <=  2048)    stb_out2(0x000800 + numlit-1);
204     else /*  numlit <= 65536) */ stb_out3(0x070000 + numlit-1);
205 
206     if (stb__out) {
207         memcpy(stb__out,in,numlit);
208         stb__out += numlit;
209     } else
210         fwrite(in, 1, numlit, stb__outfile);
211 }
212 
213 static int stb__window = 0x40000; // 256K
214 
stb_not_crap(int best,int dist)215 static int stb_not_crap(int best, int dist)
216 {
217     return   ((best > 2  &&  dist <= 0x00100)
218         || (best > 5  &&  dist <= 0x04000)
219         || (best > 7  &&  dist <= 0x80000));
220 }
221 
222 static  stb_uint stb__hashsize = 32768;
223 
224 // note that you can play with the hashing functions all you
225 // want without needing to change the decompressor
226 #define stb__hc(q,h,c)      (((h) << 7) + ((h) >> 25) + q[c])
227 #define stb__hc2(q,h,c,d)   (((h) << 14) + ((h) >> 18) + (q[c] << 7) + q[d])
228 #define stb__hc3(q,c,d,e)   ((q[c] << 14) + (q[d] << 7) + q[e])
229 
230 static unsigned int stb__running_adler;
231 
stb_compress_chunk(stb_uchar * history,stb_uchar * start,stb_uchar * end,int length,int * pending_literals,stb_uchar ** chash,stb_uint mask)232 static int stb_compress_chunk(stb_uchar *history,
233     stb_uchar *start,
234     stb_uchar *end,
235     int length,
236     int *pending_literals,
237     stb_uchar **chash,
238     stb_uint mask)
239 {
240     (void)history;
241     int window = stb__window;
242     stb_uint match_max;
243     stb_uchar *lit_start = start - *pending_literals;
244     stb_uchar *q = start;
245 
246 #define STB__SCRAMBLE(h)   (((h) + ((h) >> 16)) & mask)
247 
248     // stop short of the end so we don't scan off the end doing
249     // the hashing; this means we won't compress the last few bytes
250     // unless they were part of something longer
251     while (q < start+length && q+12 < end) {
252         int m;
253         stb_uint h1,h2,h3,h4, h;
254         stb_uchar *t;
255         int best = 2, dist=0;
256 
257         if (q+65536 > end)
258             match_max = end-q;
259         else
260             match_max = 65536;
261 
262 #define stb__nc(b,d)  ((d) <= window && ((b) > 9 || stb_not_crap(b,d)))
263 
264 #define STB__TRY(t,p)  /* avoid retrying a match we already tried */ \
265     if (p ? dist != q-t : 1)                             \
266     if ((m = stb_matchlen(t, q, match_max)) > best)     \
267     if (stb__nc(m,q-(t)))                                \
268     best = m, dist = q - (t)
269 
270         // rather than search for all matches, only try 4 candidate locations,
271         // chosen based on 4 different hash functions of different lengths.
272         // this strategy is inspired by LZO; hashing is unrolled here using the
273         // 'hc' macro
274         h = stb__hc3(q,0, 1, 2); h1 = STB__SCRAMBLE(h);
275         t = chash[h1]; if (t) STB__TRY(t,0);
276         h = stb__hc2(q,h, 3, 4); h2 = STB__SCRAMBLE(h);
277         h = stb__hc2(q,h, 5, 6);        t = chash[h2]; if (t) STB__TRY(t,1);
278         h = stb__hc2(q,h, 7, 8); h3 = STB__SCRAMBLE(h);
279         h = stb__hc2(q,h, 9,10);        t = chash[h3]; if (t) STB__TRY(t,1);
280         h = stb__hc2(q,h,11,12); h4 = STB__SCRAMBLE(h);
281         t = chash[h4]; if (t) STB__TRY(t,1);
282 
283         // because we use a shared hash table, can only update it
284         // _after_ we've probed all of them
285         chash[h1] = chash[h2] = chash[h3] = chash[h4] = q;
286 
287         if (best > 2)
288             assert(dist > 0);
289 
290         // see if our best match qualifies
291         if (best < 3) { // fast path literals
292             ++q;
293         } else if (best > 2  &&  best <= 0x80    &&  dist <= 0x100) {
294             outliterals(lit_start, q-lit_start); lit_start = (q += best);
295             stb_out(0x80 + best-1);
296             stb_out(dist-1);
297         } else if (best > 5  &&  best <= 0x100   &&  dist <= 0x4000) {
298             outliterals(lit_start, q-lit_start); lit_start = (q += best);
299             stb_out2(0x4000 + dist-1);
300             stb_out(best-1);
301         } else if (best > 7  &&  best <= 0x100   &&  dist <= 0x80000) {
302             outliterals(lit_start, q-lit_start); lit_start = (q += best);
303             stb_out3(0x180000 + dist-1);
304             stb_out(best-1);
305         } else if (best > 8  &&  best <= 0x10000 &&  dist <= 0x80000) {
306             outliterals(lit_start, q-lit_start); lit_start = (q += best);
307             stb_out3(0x100000 + dist-1);
308             stb_out2(best-1);
309         } else if (best > 9                      &&  dist <= 0x1000000) {
310             if (best > 65536) best = 65536;
311             outliterals(lit_start, q-lit_start); lit_start = (q += best);
312             if (best <= 0x100) {
313                 stb_out(0x06);
314                 stb_out3(dist-1);
315                 stb_out(best-1);
316             } else {
317                 stb_out(0x04);
318                 stb_out3(dist-1);
319                 stb_out2(best-1);
320             }
321         } else {  // fallback literals if no match was a balanced tradeoff
322             ++q;
323         }
324     }
325 
326     // if we didn't get all the way, add the rest to literals
327     if (q-start < length)
328         q = start+length;
329 
330     // the literals are everything from lit_start to q
331     *pending_literals = (q - lit_start);
332 
333     stb__running_adler = stb_adler32(stb__running_adler, start, q - start);
334     return q - start;
335 }
336 
stb_compress_inner(stb_uchar * input,stb_uint length)337 static int stb_compress_inner(stb_uchar *input, stb_uint length)
338 {
339     int literals = 0;
340     stb_uint len,i;
341 
342     stb_uchar **chash;
343     chash = (stb_uchar**) malloc(stb__hashsize * sizeof(stb_uchar*));
344     if (chash == NULL) return 0; // failure
345     for (i=0; i < stb__hashsize; ++i)
346         chash[i] = NULL;
347 
348     // stream signature
349     stb_out(0x57); stb_out(0xbc);
350     stb_out2(0);
351 
352     stb_out4(0);       // 64-bit length requires 32-bit leading 0
353     stb_out4(length);
354     stb_out4(stb__window);
355 
356     stb__running_adler = 1;
357 
358     len = stb_compress_chunk(input, input, input+length, length, &literals, chash, stb__hashsize-1);
359     assert(len == length);
360 
361     outliterals(input+length - literals, literals);
362 
363     free(chash);
364 
365     stb_out2(0x05fa); // end opcode
366 
367     stb_out4(stb__running_adler);
368 
369     return 1; // success
370 }
371 
stb_compress(stb_uchar * out,stb_uchar * input,stb_uint length)372 stb_uint stb_compress(stb_uchar *out, stb_uchar *input, stb_uint length)
373 {
374     stb__out = out;
375     stb__outfile = NULL;
376 
377     stb_compress_inner(input, length);
378 
379     return stb__out - out;
380 }
381