|
1
|
/* |
|
2
|
** Copyright (c) 2006 D. Richard Hipp |
|
3
|
** |
|
4
|
** This program is free software; you can redistribute it and/or |
|
5
|
** modify it under the terms of the Simplified BSD License (also |
|
6
|
** known as the "2-Clause License" or "FreeBSD License".) |
|
7
|
|
|
8
|
** This program is distributed in the hope that it will be useful, |
|
9
|
** but without any warranty; without even the implied warranty of |
|
10
|
** merchantability or fitness for a particular purpose. |
|
11
|
** |
|
12
|
** Author contact information: |
|
13
|
** [email protected] |
|
14
|
** http://www.hwaci.com/drh/ |
|
15
|
** |
|
16
|
******************************************************************************* |
|
17
|
** |
|
18
|
** Routines for encoding and decoding text. |
|
19
|
*/ |
|
20
|
#include "config.h" |
|
21
|
#include "encode.h" |
|
22
|
|
|
23
|
/* |
|
24
|
** Make the given string safe for HTML by converting every "<" into "<", |
|
25
|
** every ">" into ">" and every "&" into "&". Return a pointer |
|
26
|
** to a new string obtained from malloc(). |
|
27
|
** |
|
28
|
** We also encode " as " and ' as ' so they can appear as an argument |
|
29
|
** to markup. |
|
30
|
*/ |
|
31
|
char *htmlize(const char *z, int n){ |
|
32
|
unsigned char c; |
|
33
|
int i = 0; |
|
34
|
int count = 0; |
|
35
|
unsigned char *zOut; |
|
36
|
const unsigned char *zIn = (const unsigned char*)z; |
|
37
|
|
|
38
|
if( n<0 ) n = strlen(z); |
|
39
|
while( i<n ){ |
|
40
|
switch( zIn[i] ){ |
|
41
|
case '<': count += 3; break; |
|
42
|
case '>': count += 3; break; |
|
43
|
case '&': count += 4; break; |
|
44
|
case '"': count += 5; break; |
|
45
|
case '\'': count += 4; break; |
|
46
|
case 0: n = i; break; |
|
47
|
} |
|
48
|
i++; |
|
49
|
} |
|
50
|
i = 0; |
|
51
|
zOut = fossil_malloc( count+n+1 ); |
|
52
|
if( count==0 ){ |
|
53
|
memcpy(zOut, zIn, n); |
|
54
|
zOut[n] = 0; |
|
55
|
return (char*)zOut; |
|
56
|
} |
|
57
|
while( n-->0 ){ |
|
58
|
c = *(zIn++); |
|
59
|
switch( c ){ |
|
60
|
case '<': |
|
61
|
zOut[i++] = '&'; |
|
62
|
zOut[i++] = 'l'; |
|
63
|
zOut[i++] = 't'; |
|
64
|
zOut[i++] = ';'; |
|
65
|
break; |
|
66
|
case '>': |
|
67
|
zOut[i++] = '&'; |
|
68
|
zOut[i++] = 'g'; |
|
69
|
zOut[i++] = 't'; |
|
70
|
zOut[i++] = ';'; |
|
71
|
break; |
|
72
|
case '&': |
|
73
|
zOut[i++] = '&'; |
|
74
|
zOut[i++] = 'a'; |
|
75
|
zOut[i++] = 'm'; |
|
76
|
zOut[i++] = 'p'; |
|
77
|
zOut[i++] = ';'; |
|
78
|
break; |
|
79
|
case '"': |
|
80
|
zOut[i++] = '&'; |
|
81
|
zOut[i++] = 'q'; |
|
82
|
zOut[i++] = 'u'; |
|
83
|
zOut[i++] = 'o'; |
|
84
|
zOut[i++] = 't'; |
|
85
|
zOut[i++] = ';'; |
|
86
|
break; |
|
87
|
case '\'': |
|
88
|
zOut[i++] = '&'; |
|
89
|
zOut[i++] = '#'; |
|
90
|
zOut[i++] = '3'; |
|
91
|
zOut[i++] = '9'; |
|
92
|
zOut[i++] = ';'; |
|
93
|
break; |
|
94
|
default: |
|
95
|
zOut[i++] = c; |
|
96
|
break; |
|
97
|
} |
|
98
|
} |
|
99
|
zOut[i] = 0; |
|
100
|
return (char*)zOut; |
|
101
|
} |
|
102
|
|
|
103
|
/* |
|
104
|
** Append HTML-escaped text to a Blob. |
|
105
|
*/ |
|
106
|
void htmlize_to_blob(Blob *p, const char *zIn, int n){ |
|
107
|
int c, i, j; |
|
108
|
if( n<0 ) n = strlen(zIn); |
|
109
|
for(i=j=0; i<n; i++){ |
|
110
|
c = zIn[i]; |
|
111
|
switch( c ){ |
|
112
|
case '<': |
|
113
|
if( j<i ) blob_append(p, zIn+j, i-j); |
|
114
|
blob_append(p, "<", 4); |
|
115
|
j = i+1; |
|
116
|
break; |
|
117
|
case '>': |
|
118
|
if( j<i ) blob_append(p, zIn+j, i-j); |
|
119
|
blob_append(p, ">", 4); |
|
120
|
j = i+1; |
|
121
|
break; |
|
122
|
case '&': |
|
123
|
if( j<i ) blob_append(p, zIn+j, i-j); |
|
124
|
blob_append(p, "&", 5); |
|
125
|
j = i+1; |
|
126
|
break; |
|
127
|
case '"': |
|
128
|
if( j<i ) blob_append(p, zIn+j, i-j); |
|
129
|
blob_append(p, """, 6); |
|
130
|
j = i+1; |
|
131
|
break; |
|
132
|
case '\'': |
|
133
|
if( j<i ) blob_append(p, zIn+j, i-j); |
|
134
|
blob_append(p, "'", 5); |
|
135
|
j = i+1; |
|
136
|
break; |
|
137
|
case '\r': |
|
138
|
if( j<i ) blob_append(p, zIn+j, i-j); |
|
139
|
blob_append(p, " ", 1); |
|
140
|
j = i+1; |
|
141
|
break; |
|
142
|
} |
|
143
|
} |
|
144
|
if( j<i ) blob_append(p, zIn+j, i-j); |
|
145
|
} |
|
146
|
|
|
147
|
/* |
|
148
|
** Make the given string safe for HTML by converting syntax characters |
|
149
|
** into alternatives that do not have the special syntactic meaning. |
|
150
|
** |
|
151
|
** < --> U+227a |
|
152
|
** > --> U+227b |
|
153
|
** & --> + |
|
154
|
** " --> U+201d |
|
155
|
** ' --> U+2019 |
|
156
|
** |
|
157
|
** Return a pointer to a new string obtained from fossil_malloc(). |
|
158
|
*/ |
|
159
|
char *html_lookalike(const char *z, int n){ |
|
160
|
unsigned char c; |
|
161
|
int i = 0; |
|
162
|
int count = 0; |
|
163
|
unsigned char *zOut; |
|
164
|
const unsigned char *zIn = (const unsigned char*)z; |
|
165
|
|
|
166
|
if( n<0 ) n = strlen(z); |
|
167
|
while( i<n ){ |
|
168
|
switch( zIn[i] ){ |
|
169
|
case '<': count += 3; break; |
|
170
|
case '>': count += 3; break; |
|
171
|
case '"': count += 3; break; |
|
172
|
case '\'': count += 3; break; |
|
173
|
case 0: n = i; break; |
|
174
|
} |
|
175
|
i++; |
|
176
|
} |
|
177
|
i = 0; |
|
178
|
zOut = fossil_malloc( count+n+1 ); |
|
179
|
if( count==0 ){ |
|
180
|
memcpy(zOut, zIn, n); |
|
181
|
zOut[n] = 0; |
|
182
|
return (char*)zOut; |
|
183
|
} |
|
184
|
while( n-->0 ){ |
|
185
|
c = *(zIn++); |
|
186
|
switch( c ){ |
|
187
|
case '<': |
|
188
|
zOut[i++] = 0xe2; |
|
189
|
zOut[i++] = 0x89; |
|
190
|
zOut[i++] = 0xba; |
|
191
|
break; |
|
192
|
case '>': |
|
193
|
zOut[i++] = 0xe2; |
|
194
|
zOut[i++] = 0x89; |
|
195
|
zOut[i++] = 0xbb; |
|
196
|
break; |
|
197
|
case '&': |
|
198
|
zOut[i++] = '+'; |
|
199
|
break; |
|
200
|
case '"': |
|
201
|
zOut[i++] = 0xe2; |
|
202
|
zOut[i++] = 0x80; |
|
203
|
zOut[i++] = 0x9d; |
|
204
|
break; |
|
205
|
case '\'': |
|
206
|
zOut[i++] = 0xe2; |
|
207
|
zOut[i++] = 0x80; |
|
208
|
zOut[i++] = 0x99; |
|
209
|
break; |
|
210
|
default: |
|
211
|
zOut[i++] = c; |
|
212
|
break; |
|
213
|
} |
|
214
|
} |
|
215
|
zOut[i] = 0; |
|
216
|
return (char*)zOut; |
|
217
|
} |
|
218
|
|
|
219
|
|
|
220
|
/* |
|
221
|
** Encode a string for HTTP. This means converting lots of |
|
222
|
** characters into the "%HH" where H is a hex digit. It also |
|
223
|
** means converting spaces to "+". |
|
224
|
** |
|
225
|
** This is the opposite of DeHttpizeString below. |
|
226
|
*/ |
|
227
|
static char *EncodeHttp(const char *zIn, int n, int encodeSlash){ |
|
228
|
int c; |
|
229
|
int i = 0; |
|
230
|
int count = 0; |
|
231
|
char *zOut; |
|
232
|
# define IsSafeChar(X) \ |
|
233
|
(fossil_isalnum(X) || (X)=='.' || (X)=='$' \ |
|
234
|
|| (X)=='~' || (X)=='-' || (X)=='_' \ |
|
235
|
|| (!encodeSlash && ((X)=='/' || (X)==':'))) |
|
236
|
|
|
237
|
if( zIn==0 ) return 0; |
|
238
|
if( n<0 ) n = strlen(zIn); |
|
239
|
while( i<n && (c = zIn[i])!=0 ){ |
|
240
|
if( IsSafeChar(c) || c==' ' ){ |
|
241
|
count++; |
|
242
|
}else{ |
|
243
|
count += 3; |
|
244
|
} |
|
245
|
i++; |
|
246
|
} |
|
247
|
i = 0; |
|
248
|
zOut = fossil_malloc( count+1 ); |
|
249
|
while( n-->0 && (c = *zIn)!=0 ){ |
|
250
|
if( IsSafeChar(c) ){ |
|
251
|
zOut[i++] = c; |
|
252
|
}else if( c==' ' ){ |
|
253
|
zOut[i++] = '+'; |
|
254
|
}else{ |
|
255
|
zOut[i++] = '%'; |
|
256
|
zOut[i++] = "0123456789ABCDEF"[(c>>4)&0xf]; |
|
257
|
zOut[i++] = "0123456789ABCDEF"[c&0xf]; |
|
258
|
} |
|
259
|
zIn++; |
|
260
|
} |
|
261
|
zOut[i] = 0; |
|
262
|
#undef IsSafeChar |
|
263
|
return zOut; |
|
264
|
} |
|
265
|
|
|
266
|
/* |
|
267
|
** Convert the input string into a form that is suitable for use as |
|
268
|
** a token in the HTTP protocol. Spaces are encoded as '+' and special |
|
269
|
** characters are encoded as "%HH" where HH is a two-digit hexadecimal |
|
270
|
** representation of the character. The "/" character is encoded |
|
271
|
** as "%2F". |
|
272
|
*/ |
|
273
|
char *httpize(const char *z, int n){ |
|
274
|
return EncodeHttp(z, n, 1); |
|
275
|
} |
|
276
|
|
|
277
|
/* |
|
278
|
** Convert the input string into a form that is suitable for use as |
|
279
|
** a token in the HTTP protocol. Spaces are encoded as '+' and special |
|
280
|
** characters are encoded as "%HH" where HH is a two-digit hexidecimal |
|
281
|
** representation of the character. The "/" character is not encoded |
|
282
|
** by this routine. |
|
283
|
*/ |
|
284
|
char *urlize(const char *z, int n){ |
|
285
|
return EncodeHttp(z, n, 0); |
|
286
|
} |
|
287
|
|
|
288
|
/* |
|
289
|
** If input string does not contain quotes (neither ' nor ") |
|
290
|
** then return the argument itself. Otherwise return a newly allocated |
|
291
|
** copy of input with all quotes %-escaped. |
|
292
|
*/ |
|
293
|
const char* escape_quotes(const char *zIn){ |
|
294
|
char *zRet, *zOut; |
|
295
|
size_t i, n = 0; |
|
296
|
for(i=0; zIn[i]; i++){ |
|
297
|
if( zIn[i]== '"' || zIn[i]== '\'' ) n++; |
|
298
|
} |
|
299
|
if( !n ) return zIn; |
|
300
|
zRet = zOut = fossil_malloc( i + 2*n + 1 ); |
|
301
|
for(i=0; zIn[i]; i++){ |
|
302
|
if( zIn[i]=='"' ){ |
|
303
|
*(zOut++) = '%'; |
|
304
|
*(zOut++) = '2'; |
|
305
|
*(zOut++) = '2'; |
|
306
|
}else if( zIn[i]=='\'' ){ |
|
307
|
*(zOut++) = '%'; |
|
308
|
*(zOut++) = '2'; |
|
309
|
*(zOut++) = '7'; |
|
310
|
}else{ |
|
311
|
*(zOut++) = zIn[i]; |
|
312
|
} |
|
313
|
} |
|
314
|
*zOut = 0; |
|
315
|
return zRet; |
|
316
|
} |
|
317
|
|
|
318
|
/* |
|
319
|
** Convert a single HEX digit to an integer |
|
320
|
*/ |
|
321
|
int fossil_hexvalue(int c){ |
|
322
|
if( c>='a' && c<='f' ){ |
|
323
|
c += 10 - 'a'; |
|
324
|
}else if( c>='A' && c<='F' ){ |
|
325
|
c += 10 - 'A'; |
|
326
|
}else if( c>='0' && c<='9' ){ |
|
327
|
c -= '0'; |
|
328
|
}else{ |
|
329
|
c = 0; |
|
330
|
} |
|
331
|
return c; |
|
332
|
} |
|
333
|
|
|
334
|
/* |
|
335
|
** Remove the HTTP encodings from a string. The conversion is done |
|
336
|
** in-place. Return the length of the string after conversion. |
|
337
|
*/ |
|
338
|
int dehttpize(char *z){ |
|
339
|
int i, j; |
|
340
|
|
|
341
|
/* Treat a null pointer as a zero-length string. */ |
|
342
|
if( !z ) return 0; |
|
343
|
|
|
344
|
i = j = 0; |
|
345
|
while( z[i] ){ |
|
346
|
switch( z[i] ){ |
|
347
|
case '%': |
|
348
|
if( z[i+1] && z[i+2] ){ |
|
349
|
z[j] = fossil_hexvalue(z[i+1]) << 4; |
|
350
|
z[j] |= fossil_hexvalue(z[i+2]); |
|
351
|
i += 2; |
|
352
|
} |
|
353
|
break; |
|
354
|
case '+': |
|
355
|
z[j] = ' '; |
|
356
|
break; |
|
357
|
default: |
|
358
|
z[j] = z[i]; |
|
359
|
break; |
|
360
|
} |
|
361
|
i++; |
|
362
|
j++; |
|
363
|
} |
|
364
|
z[j] = 0; |
|
365
|
return j; |
|
366
|
} |
|
367
|
|
|
368
|
/* |
|
369
|
** The "fossilize" encoding is used in the headers of records |
|
370
|
** (aka "content files") to escape special characters. The |
|
371
|
** fossilize encoding passes most characters through unchanged. |
|
372
|
** The changes are these: |
|
373
|
** |
|
374
|
** space -> \s |
|
375
|
** tab -> \t |
|
376
|
** newline -> \n |
|
377
|
** cr -> \r |
|
378
|
** formfeed -> \f |
|
379
|
** vtab -> \v |
|
380
|
** nul -> \0 |
|
381
|
** \ -> \\ |
|
382
|
** |
|
383
|
** The fossilize() routine does an encoding of its input and |
|
384
|
** returns a pointer to the encoding in space obtained from |
|
385
|
** malloc. |
|
386
|
*/ |
|
387
|
char *fossilize(const char *zIn, int nIn){ |
|
388
|
int n, i, j, c; |
|
389
|
char *zOut; |
|
390
|
if( nIn<0 ) nIn = strlen(zIn); |
|
391
|
for(i=n=0; i<nIn; i++){ |
|
392
|
c = zIn[i]; |
|
393
|
if( c==0 || c==' ' || c=='\n' || c=='\t' || c=='\r' || c=='\f' || c=='\v' |
|
394
|
|| c=='\\' ) n++; |
|
395
|
} |
|
396
|
n += nIn; |
|
397
|
zOut = fossil_malloc( n+1 ); |
|
398
|
if( zOut ){ |
|
399
|
for(i=j=0; i<nIn; i++){ |
|
400
|
int c = zIn[i]; |
|
401
|
if( c==0 ){ |
|
402
|
zOut[j++] = '\\'; |
|
403
|
zOut[j++] = '0'; |
|
404
|
}else if( c=='\\' ){ |
|
405
|
zOut[j++] = '\\'; |
|
406
|
zOut[j++] = '\\'; |
|
407
|
}else if( fossil_isspace(c) ){ |
|
408
|
zOut[j++] = '\\'; |
|
409
|
switch( c ){ |
|
410
|
case '\n': c = 'n'; break; |
|
411
|
case ' ': c = 's'; break; |
|
412
|
case '\t': c = 't'; break; |
|
413
|
case '\r': c = 'r'; break; |
|
414
|
case '\v': c = 'v'; break; |
|
415
|
case '\f': c = 'f'; break; |
|
416
|
} |
|
417
|
zOut[j++] = c; |
|
418
|
}else{ |
|
419
|
zOut[j++] = c; |
|
420
|
} |
|
421
|
} |
|
422
|
zOut[j] = 0; |
|
423
|
} |
|
424
|
return zOut; |
|
425
|
} |
|
426
|
|
|
427
|
/* |
|
428
|
** Decode a fossilized string in-place. |
|
429
|
*/ |
|
430
|
void defossilize(char *z){ |
|
431
|
int i, j, c; |
|
432
|
char *zSlash = strchr(z, '\\'); |
|
433
|
if( zSlash==0 ) return; |
|
434
|
i = zSlash - z; |
|
435
|
for(j=i; (c=z[i])!=0; i++){ |
|
436
|
if( c=='\\' && z[i+1] ){ |
|
437
|
i++; |
|
438
|
switch( z[i] ){ |
|
439
|
case 'n': c = '\n'; break; |
|
440
|
case 's': c = ' '; break; |
|
441
|
case 't': c = '\t'; break; |
|
442
|
case 'r': c = '\r'; break; |
|
443
|
case 'v': c = '\v'; break; |
|
444
|
case 'f': c = '\f'; break; |
|
445
|
case '0': c = 0; break; |
|
446
|
case '\\': c = '\\'; break; |
|
447
|
default: c = z[i]; break; |
|
448
|
} |
|
449
|
} |
|
450
|
z[j++] = c; |
|
451
|
} |
|
452
|
if( z[j] ) z[j] = 0; |
|
453
|
} |
|
454
|
|
|
455
|
|
|
456
|
/* |
|
457
|
** The *pz variable points to a UTF8 string. Read the next character |
|
458
|
** off of that string and return its codepoint value. Advance *pz to the |
|
459
|
** next character |
|
460
|
*/ |
|
461
|
u32 fossil_utf8_read( |
|
462
|
const unsigned char **pz /* Pointer to string from which to read char */ |
|
463
|
){ |
|
464
|
unsigned int c; |
|
465
|
|
|
466
|
/* |
|
467
|
** This lookup table is used to help decode the first byte of |
|
468
|
** a multi-byte UTF8 character. |
|
469
|
*/ |
|
470
|
static const unsigned char utf8Trans1[] = { |
|
471
|
0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, |
|
472
|
0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f, |
|
473
|
0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16, 0x17, |
|
474
|
0x18, 0x19, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f, |
|
475
|
0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, |
|
476
|
0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f, |
|
477
|
0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, |
|
478
|
0x00, 0x01, 0x02, 0x03, 0x00, 0x01, 0x00, 0x00, |
|
479
|
}; |
|
480
|
|
|
481
|
c = *((*pz)++); |
|
482
|
if( c>=0xc0 ){ |
|
483
|
c = utf8Trans1[c-0xc0]; |
|
484
|
while( (*(*pz) & 0xc0)==0x80 ){ |
|
485
|
c = (c<<6) + (0x3f & *((*pz)++)); |
|
486
|
} |
|
487
|
if( c<0x80 |
|
488
|
|| (c&0xFFFFF800)==0xD800 |
|
489
|
|| (c&0xFFFFFFFE)==0xFFFE ){ c = 0xFFFD; } |
|
490
|
} |
|
491
|
return c; |
|
492
|
} |
|
493
|
|
|
494
|
/* |
|
495
|
** Encode a UTF8 string as a JSON string literal (with or without the |
|
496
|
** surrounding "...", depending on whether the 2nd argument is true or |
|
497
|
** false) and return a pointer to the encoding. Space to hold the |
|
498
|
** encoding is obtained from fossil_malloc() and must be freed by the |
|
499
|
** caller. |
|
500
|
** |
|
501
|
** If nOut is not NULL then it is assigned to the length, in bytes, of |
|
502
|
** the returned string (its strlen(), not counting the terminating |
|
503
|
** NUL). |
|
504
|
*/ |
|
505
|
char *encode_json_string_literal(const char *zStr, int fAddQuotes, |
|
506
|
int * nOut){ |
|
507
|
const unsigned char *z; |
|
508
|
char *zOut; |
|
509
|
u32 c; |
|
510
|
int n, i, j; |
|
511
|
z = (const unsigned char*)zStr; |
|
512
|
n = 0; |
|
513
|
while( (c = *(z++))!=0 ){ |
|
514
|
if( c=='\\' || c=='"' ){ |
|
515
|
n += 2; |
|
516
|
}else if( c<' ' ){ |
|
517
|
if( c=='\n' || c=='\r' ){ |
|
518
|
n += 2; |
|
519
|
}else{ |
|
520
|
n += 6; |
|
521
|
} |
|
522
|
}else{ |
|
523
|
n++; |
|
524
|
} |
|
525
|
} |
|
526
|
if(fAddQuotes){ |
|
527
|
n += 2; |
|
528
|
} |
|
529
|
zOut = fossil_malloc(n+1); |
|
530
|
if( zOut==0 ) return 0; |
|
531
|
z = (const unsigned char*)zStr; |
|
532
|
i = 0; |
|
533
|
if(fAddQuotes){ |
|
534
|
zOut[i++] = '"'; |
|
535
|
} |
|
536
|
while( (c = *(z++))!=0 ){ |
|
537
|
if( c=='\\' || c=='"' ){ |
|
538
|
zOut[i++] = '\\'; |
|
539
|
zOut[i++] = c; |
|
540
|
}else if( c<' ' ){ |
|
541
|
zOut[i++] = '\\'; |
|
542
|
if( c=='\n' ){ |
|
543
|
zOut[i++] = 'n'; |
|
544
|
}else if( c=='\r' ){ |
|
545
|
zOut[i++] = 'r'; |
|
546
|
}else{ |
|
547
|
zOut[i++] = 'u'; |
|
548
|
for(j=3; j>=0; j--){ |
|
549
|
zOut[i+j] = "0123456789abcdef"[c&0xf]; |
|
550
|
c >>= 4; |
|
551
|
} |
|
552
|
i += 4; |
|
553
|
} |
|
554
|
}else{ |
|
555
|
zOut[i++] = c; |
|
556
|
} |
|
557
|
} |
|
558
|
if(fAddQuotes){ |
|
559
|
zOut[i++] = '"'; |
|
560
|
} |
|
561
|
zOut[i] = 0; |
|
562
|
if(nOut!=0){ |
|
563
|
*nOut = i; |
|
564
|
} |
|
565
|
return zOut; |
|
566
|
} |
|
567
|
|
|
568
|
/* |
|
569
|
** The characters used for HTTP base64 encoding. |
|
570
|
*/ |
|
571
|
static unsigned char zBase[] = |
|
572
|
"ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/"; |
|
573
|
|
|
574
|
/* |
|
575
|
** Translate nData bytes of content from zData into |
|
576
|
** ((nData+2)/3)*4) bytes of base64 encoded content and |
|
577
|
** put the result in z64. Add a zero-terminator at the end. |
|
578
|
*/ |
|
579
|
int translateBase64(const char *zData, int nData, char *z64){ |
|
580
|
int i, n; |
|
581
|
for(i=n=0; i+2<nData; i+=3){ |
|
582
|
z64[n++] = zBase[ (zData[i]>>2) & 0x3f ]; |
|
583
|
z64[n++] = zBase[ ((zData[i]<<4) & 0x30) | ((zData[i+1]>>4) & 0x0f) ]; |
|
584
|
z64[n++] = zBase[ ((zData[i+1]<<2) & 0x3c) | ((zData[i+2]>>6) & 0x03) ]; |
|
585
|
z64[n++] = zBase[ zData[i+2] & 0x3f ]; |
|
586
|
} |
|
587
|
if( i+1<nData ){ |
|
588
|
z64[n++] = zBase[ (zData[i]>>2) & 0x3f ]; |
|
589
|
z64[n++] = zBase[ ((zData[i]<<4) & 0x30) | ((zData[i+1]>>4) & 0x0f) ]; |
|
590
|
z64[n++] = zBase[ ((zData[i+1]<<2) & 0x3c) ]; |
|
591
|
z64[n++] = '='; |
|
592
|
}else if( i<nData ){ |
|
593
|
z64[n++] = zBase[ (zData[i]>>2) & 0x3f ]; |
|
594
|
z64[n++] = zBase[ ((zData[i]<<4) & 0x30) ]; |
|
595
|
z64[n++] = '='; |
|
596
|
z64[n++] = '='; |
|
597
|
} |
|
598
|
z64[n] = 0; |
|
599
|
return n; |
|
600
|
} |
|
601
|
|
|
602
|
/* |
|
603
|
** Encode a string using a base-64 encoding. |
|
604
|
** The encoding can be reversed using the <b>decode64</b> function. |
|
605
|
** |
|
606
|
** Space to hold the result comes from malloc(). |
|
607
|
*/ |
|
608
|
char *encode64(const char *zData, int nData){ |
|
609
|
char *z64; |
|
610
|
if( nData<=0 ){ |
|
611
|
nData = strlen(zData); |
|
612
|
} |
|
613
|
z64 = fossil_malloc( (nData*4)/3 + 8 ); |
|
614
|
translateBase64(zData, nData, z64); |
|
615
|
return z64; |
|
616
|
} |
|
617
|
|
|
618
|
/* |
|
619
|
** COMMAND: test-encode64 |
|
620
|
** |
|
621
|
** Usage: %fossil test-encode64 STRING |
|
622
|
*/ |
|
623
|
void test_encode64_cmd(void){ |
|
624
|
char *z; |
|
625
|
int i; |
|
626
|
for(i=2; i<g.argc; i++){ |
|
627
|
z = encode64(g.argv[i], -1); |
|
628
|
fossil_print("%s\n", z); |
|
629
|
free(z); |
|
630
|
} |
|
631
|
} |
|
632
|
|
|
633
|
|
|
634
|
/* Decode base64 text. Write the output into zData. The caller |
|
635
|
** must ensure that zData is large enough. It is ok for z64 and |
|
636
|
** zData to be the same buffer. In other words, it is ok to decode |
|
637
|
** in-place. A zero terminator is always placed at the end of zData. |
|
638
|
*/ |
|
639
|
void decodeBase64(const char *z64, int *pnByte, char *zData){ |
|
640
|
const unsigned char *zIn = (const unsigned char*)z64; |
|
641
|
int i, j, k; |
|
642
|
int x[4]; |
|
643
|
static int isInit = 0; |
|
644
|
static signed char trans[256]; |
|
645
|
|
|
646
|
if( !isInit ){ |
|
647
|
for(i=0; i<256; i++){ trans[i] = -1; } |
|
648
|
for(i=0; zBase[i]; i++){ trans[zBase[i] & 0x7f] = i; } |
|
649
|
isInit = 1; |
|
650
|
} |
|
651
|
for(j=k=0; zIn[0]; zIn++){ |
|
652
|
int v = trans[zIn[0]]; |
|
653
|
if( v>=0 ){ |
|
654
|
x[k++] = v; |
|
655
|
if( k==4 ){ |
|
656
|
zData[j++] = ((x[0]<<2) & 0xfc) | ((x[1]>>4) & 0x03); |
|
657
|
zData[j++] = ((x[1]<<4) & 0xf0) | ((x[2]>>2) & 0x0f); |
|
658
|
zData[j++] = ((x[2]<<6) & 0xc0) | (x[3] & 0x3f); |
|
659
|
k = 0; |
|
660
|
} |
|
661
|
} |
|
662
|
} |
|
663
|
if( k>=2 ){ |
|
664
|
zData[j++] = ((x[0]<<2) & 0xfc) | ((x[1]>>4) & 0x03); |
|
665
|
if( k==3 ){ |
|
666
|
zData[j++] = ((x[1]<<4) & 0xf0) | ((x[2]>>2) & 0x0f); |
|
667
|
} |
|
668
|
} |
|
669
|
zData[j] = 0; |
|
670
|
*pnByte = j; |
|
671
|
} |
|
672
|
|
|
673
|
|
|
674
|
/* |
|
675
|
** This function treats its input as a base-64 string and returns the |
|
676
|
** decoded value of that string. Characters of input that are not |
|
677
|
** valid base-64 characters (such as spaces and newlines) are ignored. |
|
678
|
** |
|
679
|
** Space to hold the decoded string is obtained from malloc(). |
|
680
|
** |
|
681
|
** The number of bytes decoded is returned in *pnByte |
|
682
|
*/ |
|
683
|
char *decode64(const char *z64, int *pnByte){ |
|
684
|
char *zData; |
|
685
|
int n64 = (int)strlen(z64); |
|
686
|
while( n64>0 && z64[n64-1]=='=' ) n64--; |
|
687
|
zData = fossil_malloc( (n64*3)/4 + 4 ); |
|
688
|
decodeBase64(z64, pnByte, zData); |
|
689
|
return zData; |
|
690
|
} |
|
691
|
|
|
692
|
/* |
|
693
|
** COMMAND: test-decode64 |
|
694
|
** |
|
695
|
** Usage: %fossil test-decode64 STRING |
|
696
|
*/ |
|
697
|
void test_decode64_cmd(void){ |
|
698
|
char *z; |
|
699
|
int i, n; |
|
700
|
for(i=2; i<g.argc; i++){ |
|
701
|
z = decode64(g.argv[i], &n); |
|
702
|
fossil_print("%d: %s\n", n, z); |
|
703
|
fossil_free(z); |
|
704
|
} |
|
705
|
} |
|
706
|
|
|
707
|
/* |
|
708
|
** The base-16 encoding using the following characters: |
|
709
|
** |
|
710
|
** 0123456789abcdef |
|
711
|
** |
|
712
|
*/ |
|
713
|
|
|
714
|
/* |
|
715
|
** The array used for encoding |
|
716
|
*/ /* 123456789 12345 */ |
|
717
|
static const char zEncode[] = "0123456789abcdef"; |
|
718
|
|
|
719
|
/* |
|
720
|
** Encode a N-digit base-256 in base-16. Return zero on success |
|
721
|
** and non-zero if there is an error. |
|
722
|
*/ |
|
723
|
int encode16(const unsigned char *pIn, unsigned char *zOut, int N){ |
|
724
|
int i; |
|
725
|
for(i=0; i<N; i++){ |
|
726
|
*(zOut++) = zEncode[pIn[i]>>4]; |
|
727
|
*(zOut++) = zEncode[pIn[i]&0xf]; |
|
728
|
} |
|
729
|
*zOut = 0; |
|
730
|
return 0; |
|
731
|
} |
|
732
|
|
|
733
|
/* |
|
734
|
** An array for translating single base-16 characters into a value. |
|
735
|
** Disallowed input characters have a value of 64. Upper and lower |
|
736
|
** case is the same. |
|
737
|
*/ |
|
738
|
static const char zDecode[] = { |
|
739
|
64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, |
|
740
|
64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, |
|
741
|
64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, |
|
742
|
0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 64, 64, 64, 64, 64, 64, |
|
743
|
64, 10, 11, 12, 13, 14, 15, 64, 64, 64, 64, 64, 64, 64, 64, 64, |
|
744
|
64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, |
|
745
|
64, 10, 11, 12, 13, 14, 15, 64, 64, 64, 64, 64, 64, 64, 64, 64, |
|
746
|
64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, |
|
747
|
64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, |
|
748
|
64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, |
|
749
|
64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, |
|
750
|
64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, |
|
751
|
64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, |
|
752
|
64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, |
|
753
|
64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, |
|
754
|
64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, |
|
755
|
}; |
|
756
|
|
|
757
|
/* |
|
758
|
** Decode a N-character base-16 number into base-256. N must be a |
|
759
|
** multiple of 2. The output buffer must be at least N/2 characters |
|
760
|
** in length |
|
761
|
*/ |
|
762
|
int decode16(const unsigned char *zIn, unsigned char *pOut, int N){ |
|
763
|
int i, j; |
|
764
|
if( (N&1)!=0 ) return 1; |
|
765
|
for(i=j=0; i<N; i += 2, j++){ |
|
766
|
int v1, v2, a; |
|
767
|
a = zIn[i]; |
|
768
|
if( (a & 0x80)!=0 || (v1 = zDecode[a])==64 ) return 1; |
|
769
|
a = zIn[i+1]; |
|
770
|
if( (a & 0x80)!=0 || (v2 = zDecode[a])==64 ) return 1; |
|
771
|
pOut[j] = (v1<<4) + v2; |
|
772
|
} |
|
773
|
return 0; |
|
774
|
} |
|
775
|
|
|
776
|
|
|
777
|
/* |
|
778
|
** Return true if the input string contains only valid base-16 digits. |
|
779
|
** If any invalid characters appear in the string, return false. |
|
780
|
*/ |
|
781
|
int validate16(const char *zIn, int nIn){ |
|
782
|
int i; |
|
783
|
if( nIn<0 ) nIn = (int)strlen(zIn); |
|
784
|
if( zIn[nIn]==0 ){ |
|
785
|
return (int)strspn(zIn,"0123456789abcdefABCDEF")==nIn; |
|
786
|
} |
|
787
|
for(i=0; i<nIn; i++, zIn++){ |
|
788
|
if( zDecode[zIn[0]&0xff]>63 ){ |
|
789
|
return zIn[0]==0; |
|
790
|
} |
|
791
|
} |
|
792
|
return 1; |
|
793
|
} |
|
794
|
|
|
795
|
/* |
|
796
|
** The input string is a base16 value. Convert it into its canonical |
|
797
|
** form. This means that digits are all lower case and that conversions |
|
798
|
** like "l"->"1" and "O"->"0" occur. |
|
799
|
*/ |
|
800
|
void canonical16(char *z, int n){ |
|
801
|
while( *z && n-- ){ |
|
802
|
*z = zEncode[zDecode[(*z)&0x7f]&0x1f]; |
|
803
|
z++; |
|
804
|
} |
|
805
|
} |
|
806
|
|
|
807
|
/* |
|
808
|
** Decode hexadecimal into a string and return the new string. Space to |
|
809
|
** hold the string is obtained from fossil_malloc() and should be released |
|
810
|
** by the caller. |
|
811
|
** |
|
812
|
** If the input is not hex, return NULL. |
|
813
|
*/ |
|
814
|
char *decode16_dup(const char *zIn){ |
|
815
|
int nIn = (int)strlen(zIn); |
|
816
|
char *zOut; |
|
817
|
if( !validate16(zIn, nIn) ) return 0; |
|
818
|
zOut = fossil_malloc(nIn/2+1); |
|
819
|
decode16((const u8*)zIn, (u8*)zOut, nIn); |
|
820
|
zOut[nIn/2] = 0; |
|
821
|
return zOut; |
|
822
|
} |
|
823
|
|
|
824
|
|
|
825
|
/* |
|
826
|
** Decode a string encoded using "quoted-printable". |
|
827
|
** |
|
828
|
** (1) "=" followed by two hex digits becomes a single |
|
829
|
** byte specified by the two digits |
|
830
|
** |
|
831
|
** The decoding is done in-place. |
|
832
|
*/ |
|
833
|
void decodeQuotedPrintable(char *z, int *pnByte){ |
|
834
|
int i, j, c; |
|
835
|
for(i=j=0; (c = z[i])!=0; i++){ |
|
836
|
if( c=='=' ){ |
|
837
|
if( z[i+1]!='\r' ){ |
|
838
|
decode16((unsigned char*)&z[i+1], (unsigned char*)&z[j], 2); |
|
839
|
j++; |
|
840
|
} |
|
841
|
i += 2; |
|
842
|
}else{ |
|
843
|
z[j++] = c; |
|
844
|
} |
|
845
|
} |
|
846
|
if( pnByte ) *pnByte = j; |
|
847
|
z[j] = 0; |
|
848
|
} |
|
849
|
|
|
850
|
/* Randomness used for XOR-ing by the obscure() and unobscure() routines */ |
|
851
|
static const unsigned char aObscurer[16] = { |
|
852
|
0xa7, 0x21, 0x31, 0xe3, 0x2a, 0x50, 0x2c, 0x86, |
|
853
|
0x4c, 0xa4, 0x52, 0x25, 0xff, 0x49, 0x35, 0x85 |
|
854
|
}; |
|
855
|
|
|
856
|
|
|
857
|
/* |
|
858
|
** Obscure plain text so that it is not easily readable. |
|
859
|
** |
|
860
|
** This is used for storing sensitive information (such as passwords) in a |
|
861
|
** way that prevents their exposure through idle browsing. This is not |
|
862
|
** encryption. Anybody who really wants the password can still get it. |
|
863
|
** |
|
864
|
** The text is XOR-ed with a repeating pattern then converted to hex. |
|
865
|
** Space to hold the returned string is obtained from malloc and should |
|
866
|
** be freed by the caller. |
|
867
|
*/ |
|
868
|
char *obscure(const char *zIn){ |
|
869
|
int n, i; |
|
870
|
unsigned char salt; |
|
871
|
char *zOut; |
|
872
|
|
|
873
|
if( zIn==0 ) return 0; |
|
874
|
n = strlen(zIn); |
|
875
|
zOut = fossil_malloc( n*2+3 ); |
|
876
|
sqlite3_randomness(1, &salt); |
|
877
|
zOut[n+1] = (char)salt; |
|
878
|
for(i=0; i<n; i++) zOut[i+n+2] = zIn[i]^aObscurer[i&0x0f]^salt; |
|
879
|
encode16((unsigned char*)&zOut[n+1], (unsigned char*)zOut, n+1); |
|
880
|
return zOut; |
|
881
|
} |
|
882
|
|
|
883
|
/* |
|
884
|
** Undo the obscuring of text performed by obscure(). Or, if the input is |
|
885
|
** not hexadecimal (meaning the input is not the output of obscure()) then |
|
886
|
** do the equivalent of strdup(). |
|
887
|
** |
|
888
|
** The result is memory obtained from malloc that should be freed by the caller. |
|
889
|
*/ |
|
890
|
char *unobscure(const char *zIn){ |
|
891
|
int n, i; |
|
892
|
unsigned char salt; |
|
893
|
char *zOut; |
|
894
|
|
|
895
|
if( zIn==0 ) return 0; |
|
896
|
n = strlen(zIn); |
|
897
|
zOut = fossil_malloc( n + 1 ); |
|
898
|
if( n<2 |
|
899
|
|| decode16((unsigned char*)zIn, &salt, 2) |
|
900
|
|| decode16((unsigned char*)&zIn[2], (unsigned char*)zOut, n-2) |
|
901
|
){ |
|
902
|
memcpy(zOut, zIn, n+1); |
|
903
|
}else{ |
|
904
|
n = n/2 - 1; |
|
905
|
for(i=0; i<n; i++) zOut[i] = zOut[i]^aObscurer[i&0x0f]^salt; |
|
906
|
zOut[n] = 0; |
|
907
|
} |
|
908
|
return zOut; |
|
909
|
} |
|
910
|
|
|
911
|
/* |
|
912
|
** Command to test obscure() and unobscure(). These commands are also useful |
|
913
|
** utilities for decoding passwords found in the database. |
|
914
|
** |
|
915
|
** COMMAND: test-obscure |
|
916
|
** |
|
917
|
** For each command-line argument X, run both obscure(X) and |
|
918
|
** unobscure(obscure(X)) and print the results. This is used for testing |
|
919
|
** and debugging of the obscure() and unobscure() functions. |
|
920
|
*/ |
|
921
|
void test_obscure_cmd(void){ |
|
922
|
int i; |
|
923
|
char *z, *z2; |
|
924
|
for(i=2; i<g.argc; i++){ |
|
925
|
z = obscure(g.argv[i]); |
|
926
|
z2 = unobscure(z); |
|
927
|
fossil_print("OBSCURE: %s -> %s (%s)\n", g.argv[i], z, z2); |
|
928
|
free(z); |
|
929
|
free(z2); |
|
930
|
z = unobscure(g.argv[i]); |
|
931
|
fossil_print("UNOBSCURE: %s -> %s\n", g.argv[i], z); |
|
932
|
free(z); |
|
933
|
} |
|
934
|
} |
|
935
|
|