|
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
|
** A Blob is a variable-length containers for arbitrary string |
|
19
|
** or binary data. |
|
20
|
*/ |
|
21
|
#include "config.h" |
|
22
|
#include <zlib.h> |
|
23
|
#include "blob.h" |
|
24
|
#if defined(_WIN32) |
|
25
|
#include <fcntl.h> |
|
26
|
#include <io.h> |
|
27
|
#endif |
|
28
|
|
|
29
|
#if INTERFACE |
|
30
|
/* |
|
31
|
** A Blob can hold a string or a binary object of arbitrary size. The |
|
32
|
** size changes as necessary. |
|
33
|
*/ |
|
34
|
struct Blob { |
|
35
|
unsigned int nUsed; /* Number of bytes used in aData[] */ |
|
36
|
unsigned int nAlloc; /* Number of bytes allocated for aData[] */ |
|
37
|
unsigned int iCursor; /* Next character of input to parse */ |
|
38
|
unsigned int blobFlags; /* One or more BLOBFLAG_* bits */ |
|
39
|
char *aData; /* Where the information is stored */ |
|
40
|
void (*xRealloc)(Blob*,u64); /* Function to reallocate the buffer */ |
|
41
|
}; |
|
42
|
|
|
43
|
/* |
|
44
|
** Allowed values for Blob.blobFlags |
|
45
|
*/ |
|
46
|
#define BLOBFLAG_NotSQL 0x0001 /* Non-SQL text */ |
|
47
|
|
|
48
|
/* |
|
49
|
** The current size of a Blob |
|
50
|
*/ |
|
51
|
#define blob_size(X) ((X)->nUsed) |
|
52
|
|
|
53
|
/* |
|
54
|
** The buffer holding the blob data |
|
55
|
*/ |
|
56
|
#define blob_buffer(X) ((X)->aData) |
|
57
|
|
|
58
|
/* |
|
59
|
** Number of elements that fits into the current blob's size |
|
60
|
*/ |
|
61
|
#define blob_count(X,elType) (blob_size(X)/sizeof(elType)) |
|
62
|
|
|
63
|
/* |
|
64
|
** Append blob contents to another |
|
65
|
*/ |
|
66
|
#define blob_appendb(dest, src) \ |
|
67
|
blob_append((dest), blob_buffer(src), blob_size(src)) |
|
68
|
|
|
69
|
/* |
|
70
|
** Append a string literal to a blob |
|
71
|
** TODO: Consider renaming to blob_appendl() |
|
72
|
*/ |
|
73
|
#define blob_append_literal(blob, literal) \ |
|
74
|
blob_append((blob), "" literal, (sizeof literal)-1) |
|
75
|
/* |
|
76
|
* The empty string in the second argument leads to a syntax error |
|
77
|
* when the macro is not used with a string literal. Unfortunately |
|
78
|
* the error is not overly explicit. |
|
79
|
*/ |
|
80
|
|
|
81
|
/* |
|
82
|
** TODO: Suggested for removal because the name seems misleading. |
|
83
|
*/ |
|
84
|
#define blob_append_string blob_append_literal |
|
85
|
|
|
86
|
/* |
|
87
|
** Seek whence parameter values |
|
88
|
*/ |
|
89
|
#define BLOB_SEEK_SET 1 |
|
90
|
#define BLOB_SEEK_CUR 2 |
|
91
|
|
|
92
|
#endif /* INTERFACE */ |
|
93
|
|
|
94
|
/* |
|
95
|
** Make sure a blob is initialized |
|
96
|
*/ |
|
97
|
#define blob_is_init(x) \ |
|
98
|
assert((x)->xRealloc==blobReallocMalloc || (x)->xRealloc==blobReallocStatic) |
|
99
|
|
|
100
|
/* |
|
101
|
** Make sure a blob does not contain malloced memory. |
|
102
|
** |
|
103
|
** This might fail if we are unlucky and x is uninitialized. For that |
|
104
|
** reason it should only be used locally for debugging. Leave it turned |
|
105
|
** off for production. |
|
106
|
*/ |
|
107
|
#if 0 /* Enable for debugging only */ |
|
108
|
#define assert_blob_is_reset(x) assert(blob_is_reset(x)) |
|
109
|
#else |
|
110
|
#define assert_blob_is_reset(x) |
|
111
|
#endif |
|
112
|
|
|
113
|
|
|
114
|
|
|
115
|
/* |
|
116
|
** We find that the built-in isspace() function does not work for |
|
117
|
** some international character sets. So here is a substitute. |
|
118
|
*/ |
|
119
|
int fossil_isspace(char c){ |
|
120
|
return c==' ' || (c<='\r' && c>='\t'); |
|
121
|
} |
|
122
|
|
|
123
|
/* |
|
124
|
** Other replacements for ctype.h functions. |
|
125
|
*/ |
|
126
|
int fossil_islower(char c){ return c>='a' && c<='z'; } |
|
127
|
int fossil_isupper(char c){ return c>='A' && c<='Z'; } |
|
128
|
int fossil_isdigit(char c){ return c>='0' && c<='9'; } |
|
129
|
int fossil_isxdigit(char c){ return (c>='0' && c<='9') || (c>='a' && c<='f'); } |
|
130
|
int fossil_isXdigit(char c){ |
|
131
|
return (c>='0' && c<='9') || (c>='A' && c<='F') || (c>='a' && c<='f'); |
|
132
|
} |
|
133
|
int fossil_tolower(char c){ |
|
134
|
return fossil_isupper(c) ? c - 'A' + 'a' : c; |
|
135
|
} |
|
136
|
int fossil_toupper(char c){ |
|
137
|
return fossil_islower(c) ? c - 'a' + 'A' : c; |
|
138
|
} |
|
139
|
int fossil_isalpha(char c){ |
|
140
|
return (c>='a' && c<='z') || (c>='A' && c<='Z'); |
|
141
|
} |
|
142
|
int fossil_isalnum(char c){ |
|
143
|
return (c>='a' && c<='z') || (c>='A' && c<='Z') || (c>='0' && c<='9'); |
|
144
|
} |
|
145
|
|
|
146
|
/* Return true if and only if the entire string consists of only |
|
147
|
** alphanumeric characters. |
|
148
|
*/ |
|
149
|
int fossil_no_strange_characters(const char *z){ |
|
150
|
while( z && (fossil_isalnum(z[0]) || z[0]=='_' || z[0]=='-') ) z++; |
|
151
|
return z[0]==0; |
|
152
|
} |
|
153
|
|
|
154
|
|
|
155
|
/* |
|
156
|
** COMMAND: test-isspace |
|
157
|
** |
|
158
|
** Verify that the fossil_isspace() routine is working correctly by |
|
159
|
** testing it on all possible inputs. |
|
160
|
*/ |
|
161
|
void isspace_cmd(void){ |
|
162
|
int i; |
|
163
|
for(i=0; i<=255; i++){ |
|
164
|
if( i==' ' || i=='\n' || i=='\t' || i=='\v' |
|
165
|
|| i=='\f' || i=='\r' ){ |
|
166
|
assert( fossil_isspace((char)i) ); |
|
167
|
}else{ |
|
168
|
assert( !fossil_isspace((char)i) ); |
|
169
|
} |
|
170
|
} |
|
171
|
fossil_print("All 256 characters OK\n"); |
|
172
|
} |
|
173
|
|
|
174
|
/* |
|
175
|
** This routine is called if a blob operation fails because we |
|
176
|
** have run out of memory. |
|
177
|
*/ |
|
178
|
static void blob_panic(void){ |
|
179
|
static const char zErrMsg[] = "out of memory\n"; |
|
180
|
fputs(zErrMsg, stderr); |
|
181
|
fossil_exit(1); |
|
182
|
} |
|
183
|
|
|
184
|
/* |
|
185
|
** Maximum size of a Blob's managed memory. This is ~2GB, largely for |
|
186
|
** historical reasons. |
|
187
|
** |
|
188
|
*/ |
|
189
|
#define MAX_BLOB_SIZE 0x7fff0000 |
|
190
|
|
|
191
|
/* |
|
192
|
** If n >= MAX_BLOB_SIZE, calls blob_panic(), |
|
193
|
** else this is a no-op. |
|
194
|
*/ |
|
195
|
static void blob_assert_safe_size(u64 n){ |
|
196
|
if( n>=(i64)MAX_BLOB_SIZE ){ |
|
197
|
blob_panic(); |
|
198
|
} |
|
199
|
} |
|
200
|
|
|
201
|
/* |
|
202
|
** A reallocation function that assumes that aData came from malloc(). |
|
203
|
** This function attempts to resize the buffer of the blob to hold |
|
204
|
** newSize bytes. |
|
205
|
** |
|
206
|
** No attempt is made to recover from an out-of-memory error. |
|
207
|
** If an OOM error occurs, an error message is printed on stderr |
|
208
|
** and the program exits. |
|
209
|
*/ |
|
210
|
void blobReallocMalloc(Blob *pBlob, u64 newSize){ |
|
211
|
if( newSize==0 ){ |
|
212
|
free(pBlob->aData); |
|
213
|
pBlob->aData = 0; |
|
214
|
pBlob->nAlloc = 0; |
|
215
|
pBlob->nUsed = 0; |
|
216
|
pBlob->iCursor = 0; |
|
217
|
pBlob->blobFlags = 0; |
|
218
|
}else if( newSize>pBlob->nAlloc || newSize+4000<pBlob->nAlloc ){ |
|
219
|
char *pNew; |
|
220
|
blob_assert_safe_size(newSize); |
|
221
|
pNew = fossil_realloc(pBlob->aData, newSize); |
|
222
|
pBlob->aData = pNew; |
|
223
|
pBlob->nAlloc = (unsigned int)newSize; |
|
224
|
if( pBlob->nUsed>pBlob->nAlloc ){ |
|
225
|
pBlob->nUsed = pBlob->nAlloc; |
|
226
|
} |
|
227
|
} |
|
228
|
} |
|
229
|
|
|
230
|
/* |
|
231
|
** An initializer for Blobs |
|
232
|
*/ |
|
233
|
#if INTERFACE |
|
234
|
#define BLOB_INITIALIZER {0,0,0,0,0,blobReallocMalloc} |
|
235
|
#endif |
|
236
|
const Blob empty_blob = BLOB_INITIALIZER; |
|
237
|
|
|
238
|
/* |
|
239
|
** A reallocation function for when the initial string is in unmanaged |
|
240
|
** space. Copy the string to memory obtained from malloc(). |
|
241
|
*/ |
|
242
|
static void blobReallocStatic(Blob *pBlob, u64 newSize){ |
|
243
|
if( newSize==0 ){ |
|
244
|
*pBlob = empty_blob; |
|
245
|
}else{ |
|
246
|
char *pNew; |
|
247
|
blob_assert_safe_size(newSize); |
|
248
|
pNew = fossil_malloc( newSize ); |
|
249
|
if( pBlob->nUsed>newSize ) pBlob->nUsed = newSize; |
|
250
|
memcpy(pNew, pBlob->aData, pBlob->nUsed); |
|
251
|
pBlob->aData = pNew; |
|
252
|
pBlob->xRealloc = blobReallocMalloc; |
|
253
|
pBlob->nAlloc = (unsigned int)newSize; |
|
254
|
} |
|
255
|
} |
|
256
|
|
|
257
|
/* |
|
258
|
** Reset a blob to be an empty container. |
|
259
|
*/ |
|
260
|
void blob_reset(Blob *pBlob){ |
|
261
|
blob_is_init(pBlob); |
|
262
|
pBlob->xRealloc(pBlob, 0); |
|
263
|
} |
|
264
|
|
|
265
|
|
|
266
|
/* |
|
267
|
** Return true if the blob has been zeroed - in other words if it contains |
|
268
|
** no malloced memory. This only works reliably if the blob has been |
|
269
|
** initialized - it can return a false negative on an uninitialized blob. |
|
270
|
*/ |
|
271
|
int blob_is_reset(Blob *pBlob){ |
|
272
|
if( pBlob==0 ) return 1; |
|
273
|
if( pBlob->nUsed ) return 0; |
|
274
|
if( pBlob->xRealloc==blobReallocMalloc && pBlob->nAlloc ) return 0; |
|
275
|
return 1; |
|
276
|
} |
|
277
|
|
|
278
|
/* |
|
279
|
** Initialize a blob to a string or byte-array constant of a specified length. |
|
280
|
** Any prior data in the blob is discarded. |
|
281
|
*/ |
|
282
|
void blob_init(Blob *pBlob, const char *zData, int size){ |
|
283
|
assert_blob_is_reset(pBlob); |
|
284
|
if( zData==0 ){ |
|
285
|
*pBlob = empty_blob; |
|
286
|
}else{ |
|
287
|
if( size<=0 ) size = strlen(zData); |
|
288
|
pBlob->nUsed = pBlob->nAlloc = size; |
|
289
|
pBlob->aData = (char*)zData; |
|
290
|
pBlob->iCursor = 0; |
|
291
|
pBlob->blobFlags = 0; |
|
292
|
pBlob->xRealloc = blobReallocStatic; |
|
293
|
} |
|
294
|
} |
|
295
|
|
|
296
|
/* |
|
297
|
** Initialize a blob to a nul-terminated string. |
|
298
|
** Any prior data in the blob is discarded. |
|
299
|
*/ |
|
300
|
void blob_set(Blob *pBlob, const char *zStr){ |
|
301
|
blob_init(pBlob, zStr, -1); |
|
302
|
} |
|
303
|
|
|
304
|
/* |
|
305
|
** Initialize a blob to a nul-terminated string obtained from fossil_malloc(). |
|
306
|
** The blob will take responsibility for freeing the string. |
|
307
|
*/ |
|
308
|
void blob_set_dynamic(Blob *pBlob, char *zStr){ |
|
309
|
blob_init(pBlob, zStr, -1); |
|
310
|
pBlob->xRealloc = blobReallocMalloc; |
|
311
|
} |
|
312
|
|
|
313
|
/* |
|
314
|
** Initialize a blob to an empty string. |
|
315
|
*/ |
|
316
|
void blob_zero(Blob *pBlob){ |
|
317
|
static const char zEmpty[] = ""; |
|
318
|
assert_blob_is_reset(pBlob); |
|
319
|
pBlob->nUsed = 0; |
|
320
|
pBlob->nAlloc = 1; |
|
321
|
pBlob->aData = (char*)zEmpty; |
|
322
|
pBlob->iCursor = 0; |
|
323
|
pBlob->blobFlags = 0; |
|
324
|
pBlob->xRealloc = blobReallocStatic; |
|
325
|
} |
|
326
|
|
|
327
|
/* |
|
328
|
** Append text or data to the end of a blob. Or, if pBlob==NULL, send |
|
329
|
** the text to standard output in terminal mode, or to standard CGI output |
|
330
|
** in CGI mode. |
|
331
|
** |
|
332
|
** If nData<0 then output all of aData up to the first 0x00 byte. |
|
333
|
** |
|
334
|
** Use the blob_append() routine in all application code. The blob_append() |
|
335
|
** routine is faster, but blob_append_full() handles all the corner cases. |
|
336
|
** The blob_append() routine automatically calls blob_append_full() if |
|
337
|
** necessary. |
|
338
|
*/ |
|
339
|
static void blob_append_full(Blob *pBlob, const char *aData, int nData){ |
|
340
|
u64 nNew; |
|
341
|
/* assert( aData!=0 || nData==0 ); // omitted for speed */ |
|
342
|
/* blob_is_init(pBlob); // omitted for speed */ |
|
343
|
if( nData<0 ) nData = strlen(aData); |
|
344
|
if( nData==0 ) return; |
|
345
|
if( pBlob==0 ){ |
|
346
|
if( g.cgiOutput ){ |
|
347
|
pBlob = cgi_output_blob(); |
|
348
|
}else{ |
|
349
|
fossil_puts(aData, 0, nData); |
|
350
|
return; |
|
351
|
} |
|
352
|
} |
|
353
|
nNew = pBlob->nUsed; |
|
354
|
nNew += nData; |
|
355
|
if( nNew >= pBlob->nAlloc ){ |
|
356
|
nNew += pBlob->nAlloc; |
|
357
|
nNew += 100; |
|
358
|
blob_assert_safe_size(nNew); |
|
359
|
pBlob->xRealloc(pBlob, (unsigned)nNew); |
|
360
|
if( pBlob->nUsed + nData >= pBlob->nAlloc ){ |
|
361
|
blob_panic(); |
|
362
|
} |
|
363
|
} |
|
364
|
memcpy(&pBlob->aData[pBlob->nUsed], aData, nData); |
|
365
|
pBlob->nUsed += nData; |
|
366
|
pBlob->aData[pBlob->nUsed] = 0; /* Blobs are always nul-terminated */ |
|
367
|
} |
|
368
|
void blob_append(Blob *pBlob, const char *aData, int nData){ |
|
369
|
u64 nUsed; |
|
370
|
/* assert( aData!=0 || nData==0 ); // omitted for speed */ |
|
371
|
if( nData<=0 || pBlob==0 || pBlob->nUsed + nData >= pBlob->nAlloc ){ |
|
372
|
blob_append_full(pBlob, aData, nData); |
|
373
|
return; |
|
374
|
} |
|
375
|
nUsed = pBlob->nUsed; |
|
376
|
pBlob->nUsed += nData; |
|
377
|
pBlob->aData[pBlob->nUsed] = 0; |
|
378
|
memcpy(&pBlob->aData[nUsed], aData, nData); |
|
379
|
} |
|
380
|
|
|
381
|
/* |
|
382
|
** Append a single character to the blob. If pBlob is zero then the |
|
383
|
** character is written directly to stdout. |
|
384
|
*/ |
|
385
|
void blob_append_char(Blob *pBlob, char c){ |
|
386
|
if( pBlob==0 || pBlob->nUsed+1 >= pBlob->nAlloc ){ |
|
387
|
blob_append_full(pBlob, &c, 1); |
|
388
|
}else{ |
|
389
|
pBlob->aData[pBlob->nUsed++] = c; |
|
390
|
} |
|
391
|
} |
|
392
|
|
|
393
|
/* |
|
394
|
** Copy a blob. pTo is reinitialized to be a copy of pFrom. |
|
395
|
*/ |
|
396
|
void blob_copy(Blob *pTo, Blob *pFrom){ |
|
397
|
blob_is_init(pFrom); |
|
398
|
blob_zero(pTo); |
|
399
|
blob_append(pTo, blob_buffer(pFrom), blob_size(pFrom)); |
|
400
|
} |
|
401
|
|
|
402
|
/* |
|
403
|
** Append the second blob onto the end of the first blob and reset the |
|
404
|
** second blob. If the first blob (pTo) is NULL, then the content |
|
405
|
** of the second blob is written to stdout or to CGI depending on if the |
|
406
|
** Fossil is running in terminal or CGI mode. |
|
407
|
*/ |
|
408
|
void blob_append_xfer(Blob *pTo, Blob *pFrom){ |
|
409
|
blob_append(pTo, blob_buffer(pFrom), blob_size(pFrom)); |
|
410
|
blob_reset(pFrom); |
|
411
|
} |
|
412
|
|
|
413
|
/* |
|
414
|
** Write into pOut, a string literal representation for the first n bytes |
|
415
|
** of z[]. The string literal representation is compatible with C, TCL, |
|
416
|
** and JSON. Double-quotes are added to both ends. Double-quote and |
|
417
|
** backslash characters are escaped. |
|
418
|
*/ |
|
419
|
void blob_append_tcl_literal(Blob *pOut, const char *z, size_t n){ |
|
420
|
size_t i; |
|
421
|
blob_append_char(pOut, '"'); |
|
422
|
for(i=0; i<n; i++){ |
|
423
|
char c = z[i]; |
|
424
|
switch( c ){ |
|
425
|
case '\r': c = 'r'; |
|
426
|
case '[': |
|
427
|
case ']': |
|
428
|
case '$': |
|
429
|
case '"': |
|
430
|
case '\\': |
|
431
|
blob_append_char(pOut, '\\'); |
|
432
|
default: |
|
433
|
blob_append_char(pOut, c); |
|
434
|
} |
|
435
|
} |
|
436
|
blob_append_char(pOut, '"'); |
|
437
|
} |
|
438
|
void blob_append_json_literal(Blob *pOut, const char *z, size_t n){ |
|
439
|
size_t i; |
|
440
|
blob_append_char(pOut, '"'); |
|
441
|
for(i=0; i<n; i++){ |
|
442
|
char c = z[i]; |
|
443
|
switch( c ){ |
|
444
|
case 0x00: |
|
445
|
case 0x01: |
|
446
|
case 0x02: |
|
447
|
case 0x03: |
|
448
|
case 0x04: |
|
449
|
case 0x05: |
|
450
|
case 0x06: |
|
451
|
case 0x07: c += '0' - 0x00; blob_append(pOut, "\\u000",5); break; |
|
452
|
case 0x0b: |
|
453
|
case 0x0e: |
|
454
|
case 0x0f: c += 'a' - 0x0a; blob_append(pOut, "\\u000",5); break; |
|
455
|
case 0x10: |
|
456
|
case 0x11: |
|
457
|
case 0x12: |
|
458
|
case 0x13: |
|
459
|
case 0x14: |
|
460
|
case 0x15: |
|
461
|
case 0x16: |
|
462
|
case 0x17: |
|
463
|
case 0x18: |
|
464
|
case 0x19: c += '0' - 0x10; blob_append(pOut, "\\u001",5); break; |
|
465
|
case 0x1a: |
|
466
|
case 0x1b: |
|
467
|
case 0x1c: |
|
468
|
case 0x1d: |
|
469
|
case 0x1e: |
|
470
|
case 0x1f: c += 'a' - 0x1a; blob_append(pOut, "\\u001",5); break; |
|
471
|
case '\b': c = 'b'; blob_append_char(pOut, '\\'); break; |
|
472
|
case '\t': c = 't'; blob_append_char(pOut, '\\'); break; |
|
473
|
case '\r': c = 'r'; blob_append_char(pOut, '\\'); break; |
|
474
|
case '\n': c = 'n'; blob_append_char(pOut, '\\'); break; |
|
475
|
case '\f': c = 'f'; blob_append_char(pOut, '\\'); break; |
|
476
|
case '"': blob_append_char(pOut, '\\'); break; |
|
477
|
case '\\': blob_append_char(pOut, '\\'); break; |
|
478
|
default: break; |
|
479
|
} |
|
480
|
blob_append_char(pOut, c); |
|
481
|
} |
|
482
|
blob_append_char(pOut, '"'); |
|
483
|
} |
|
484
|
|
|
485
|
|
|
486
|
/* |
|
487
|
** Return a pointer to a null-terminated string for a blob. |
|
488
|
*/ |
|
489
|
char *blob_str(Blob *p){ |
|
490
|
blob_is_init(p); |
|
491
|
if( p->nUsed==0 ){ |
|
492
|
blob_append_char(p, 0); /* NOTE: Changes nUsed. */ |
|
493
|
p->nUsed = 0; |
|
494
|
} |
|
495
|
if( p->nUsed<p->nAlloc ){ |
|
496
|
p->aData[p->nUsed] = 0; |
|
497
|
}else{ |
|
498
|
blob_materialize(p); |
|
499
|
} |
|
500
|
return p->aData; |
|
501
|
} |
|
502
|
|
|
503
|
/* |
|
504
|
** Compute the string length of a Blob. If there are embedded |
|
505
|
** nul characters, truncate the to blob at the first nul. |
|
506
|
*/ |
|
507
|
int blob_strlen(Blob *p){ |
|
508
|
char *z = blob_str(p); |
|
509
|
if( z==0 ) return 0; |
|
510
|
p->nUsed = (int)strlen(p->aData); |
|
511
|
return p->nUsed; |
|
512
|
} |
|
513
|
|
|
514
|
/* |
|
515
|
** Return a pointer to a null-terminated string for a blob that has |
|
516
|
** been created using blob_append_sql() and not blob_appendf(). If |
|
517
|
** text was ever added using blob_appendf() then throw an error. |
|
518
|
*/ |
|
519
|
char *blob_sql_text(Blob *p){ |
|
520
|
blob_is_init(p); |
|
521
|
if( (p->blobFlags & BLOBFLAG_NotSQL) ){ |
|
522
|
fossil_panic("use of blob_appendf() to construct SQL text"); |
|
523
|
} |
|
524
|
return blob_str(p); |
|
525
|
} |
|
526
|
|
|
527
|
|
|
528
|
/* |
|
529
|
** Return a pointer to a null-terminated string for a blob. |
|
530
|
** |
|
531
|
** WARNING: If the blob is ephemeral, it might cause a '\000' |
|
532
|
** character to be inserted into the middle of the parent blob. |
|
533
|
** Example: Suppose p is a token extracted from some larger |
|
534
|
** blob pBig using blob_token(). If you call this routine on p, |
|
535
|
** then a '\000' character will be inserted in the middle of |
|
536
|
** pBig in order to cause p to be nul-terminated. If pBig |
|
537
|
** should not be modified, then use blob_str() instead of this |
|
538
|
** routine. blob_str() will make a copy of the p if necessary |
|
539
|
** to avoid modifying pBig. |
|
540
|
*/ |
|
541
|
char *blob_terminate(Blob *p){ |
|
542
|
blob_is_init(p); |
|
543
|
if( p->nUsed==0 ) return ""; |
|
544
|
p->aData[p->nUsed] = 0; |
|
545
|
return p->aData; |
|
546
|
} |
|
547
|
|
|
548
|
/* |
|
549
|
** Compare two blobs. Return negative, zero, or positive if the first |
|
550
|
** blob is less then, equal to, or greater than the second. |
|
551
|
*/ |
|
552
|
int blob_compare(const Blob *pA, const Blob *pB){ |
|
553
|
int szA, szB, sz, rc; |
|
554
|
blob_is_init(pA); |
|
555
|
blob_is_init(pB); |
|
556
|
szA = blob_size(pA); |
|
557
|
szB = blob_size(pB); |
|
558
|
sz = szA<szB ? szA : szB; |
|
559
|
rc = memcmp(blob_buffer(pA), blob_buffer(pB), sz); |
|
560
|
if( rc==0 ){ |
|
561
|
rc = szA - szB; |
|
562
|
} |
|
563
|
return rc; |
|
564
|
} |
|
565
|
|
|
566
|
/* |
|
567
|
** Compare two blobs in constant time and return zero if they are equal. |
|
568
|
** Constant time comparison only applies for blobs of the same length. |
|
569
|
** If lengths are different, immediately returns 1. |
|
570
|
*/ |
|
571
|
int blob_constant_time_cmp(Blob *pA, Blob *pB){ |
|
572
|
int szA, szB, i; |
|
573
|
unsigned char *buf1, *buf2; |
|
574
|
unsigned char rc = 0; |
|
575
|
|
|
576
|
blob_is_init(pA); |
|
577
|
blob_is_init(pB); |
|
578
|
szA = blob_size(pA); |
|
579
|
szB = blob_size(pB); |
|
580
|
if( szA!=szB || szA==0 ) return 1; |
|
581
|
|
|
582
|
buf1 = (unsigned char*)blob_buffer(pA); |
|
583
|
buf2 = (unsigned char*)blob_buffer(pB); |
|
584
|
|
|
585
|
for( i=0; i<szA; i++ ){ |
|
586
|
rc = rc | (buf1[i] ^ buf2[i]); |
|
587
|
} |
|
588
|
|
|
589
|
return rc; |
|
590
|
} |
|
591
|
|
|
592
|
/* |
|
593
|
** Compare a blob to a string. Return TRUE if they are equal. |
|
594
|
*/ |
|
595
|
int blob_eq_str(Blob *pBlob, const char *z, int n){ |
|
596
|
Blob t; |
|
597
|
blob_is_init(pBlob); |
|
598
|
if( n<=0 ) n = (int)strlen(z); |
|
599
|
t.aData = (char*)z; |
|
600
|
t.nUsed = n; |
|
601
|
t.xRealloc = blobReallocStatic; |
|
602
|
return blob_compare(pBlob, &t)==0; |
|
603
|
} |
|
604
|
|
|
605
|
/* |
|
606
|
** This macro compares a blob against a string constant. We use the sizeof() |
|
607
|
** operator on the string constant twice, so it really does need to be a |
|
608
|
** string literal or character array - not a character pointer. |
|
609
|
*/ |
|
610
|
#if INTERFACE |
|
611
|
# define blob_eq(B,S) \ |
|
612
|
((B)->nUsed==sizeof(S"")-1 && memcmp((B)->aData,S,sizeof(S)-1)==0) |
|
613
|
#endif |
|
614
|
|
|
615
|
|
|
616
|
/* |
|
617
|
** Attempt to resize a blob so that its internal buffer is |
|
618
|
** nByte in size. The blob is truncated if necessary. |
|
619
|
*/ |
|
620
|
void blob_resize(Blob *pBlob, u64 newSize){ |
|
621
|
pBlob->xRealloc(pBlob, newSize+1); |
|
622
|
pBlob->nUsed = newSize; |
|
623
|
pBlob->aData[newSize] = 0; |
|
624
|
} |
|
625
|
|
|
626
|
/* |
|
627
|
** Ensures that the given blob has at least the given amount of memory |
|
628
|
** allocated to it. Does not modify pBlob->nUsed nor will it reduce |
|
629
|
** the currently-allocated amount of memory. |
|
630
|
** |
|
631
|
** For semantic compatibility with blob_append_full(), if newSize is |
|
632
|
** >=MAX_BLOB_SIZE then this function will trigger blob_panic(). If it |
|
633
|
** didn't, it would be possible to bypass that hard-coded limit via |
|
634
|
** this function. |
|
635
|
** |
|
636
|
** We've had at least one report: |
|
637
|
** https://fossil-scm.org/forum/forumpost/b7bbd28db4 |
|
638
|
** which implies that this is unconditionally failing on mingw 32-bit |
|
639
|
** builds. |
|
640
|
*/ |
|
641
|
void blob_reserve(Blob *pBlob, u64 newSize){ |
|
642
|
blob_assert_safe_size( newSize ); |
|
643
|
if(newSize>pBlob->nAlloc){ |
|
644
|
pBlob->xRealloc(pBlob, newSize+1); |
|
645
|
pBlob->aData[newSize] = 0; |
|
646
|
} |
|
647
|
} |
|
648
|
|
|
649
|
/* |
|
650
|
** Make sure a blob is nul-terminated and is not a pointer to unmanaged |
|
651
|
** space. Return a pointer to the data. |
|
652
|
*/ |
|
653
|
char *blob_materialize(Blob *pBlob){ |
|
654
|
blob_resize(pBlob, pBlob->nUsed); |
|
655
|
return pBlob->aData; |
|
656
|
} |
|
657
|
|
|
658
|
/* |
|
659
|
** Call dehttpize on a blob. This causes an ephemeral blob to be |
|
660
|
** materialized. |
|
661
|
*/ |
|
662
|
void blob_dehttpize(Blob *pBlob){ |
|
663
|
blob_materialize(pBlob); |
|
664
|
pBlob->nUsed = dehttpize(pBlob->aData); |
|
665
|
} |
|
666
|
|
|
667
|
/* |
|
668
|
** Extract N bytes from blob pFrom and use it to initialize blob pTo. |
|
669
|
** Return the actual number of bytes extracted. The cursor position |
|
670
|
** is advanced by the number of bytes extracted. |
|
671
|
** |
|
672
|
** After this call completes, pTo will be an ephemeral blob. |
|
673
|
*/ |
|
674
|
int blob_extract(Blob *pFrom, int N, Blob *pTo){ |
|
675
|
blob_is_init(pFrom); |
|
676
|
assert_blob_is_reset(pTo); |
|
677
|
if( pFrom->iCursor + N > pFrom->nUsed ){ |
|
678
|
N = pFrom->nUsed - pFrom->iCursor; |
|
679
|
if( N<=0 ){ |
|
680
|
blob_zero(pTo); |
|
681
|
return 0; |
|
682
|
} |
|
683
|
} |
|
684
|
pTo->nUsed = N; |
|
685
|
pTo->nAlloc = N; |
|
686
|
pTo->aData = &pFrom->aData[pFrom->iCursor]; |
|
687
|
pTo->iCursor = 0; |
|
688
|
pTo->xRealloc = blobReallocStatic; |
|
689
|
pFrom->iCursor += N; |
|
690
|
return N; |
|
691
|
} |
|
692
|
|
|
693
|
/* |
|
694
|
** Extract N **lines** of text from blob pFrom beginning at the current |
|
695
|
** cursor position and use that text to initialize blob pTo. Unlike the |
|
696
|
** blob_extract() routine, the cursor position is unchanged. |
|
697
|
** |
|
698
|
** pTo is assumed to be uninitialized. |
|
699
|
** |
|
700
|
** After this call completes, pTo will be an ephemeral blob. |
|
701
|
*/ |
|
702
|
int blob_extract_lines(Blob *pFrom, int N, Blob *pTo){ |
|
703
|
int i; |
|
704
|
int mx; |
|
705
|
int iStart; |
|
706
|
int n; |
|
707
|
const char *z; |
|
708
|
|
|
709
|
blob_zero(pTo); |
|
710
|
z = pFrom->aData; |
|
711
|
i = pFrom->iCursor; |
|
712
|
mx = pFrom->nUsed; |
|
713
|
while( N>0 ){ |
|
714
|
while( i<mx && z[i]!='\n' ){ i++; } |
|
715
|
if( i>=mx ) break; |
|
716
|
i++; |
|
717
|
N--; |
|
718
|
} |
|
719
|
iStart = pFrom->iCursor; |
|
720
|
n = blob_extract(pFrom, i-pFrom->iCursor, pTo); |
|
721
|
pFrom->iCursor = iStart; |
|
722
|
return n; |
|
723
|
} |
|
724
|
|
|
725
|
/* |
|
726
|
** Return the number of lines of text in the blob. If the last |
|
727
|
** line is incomplete (if it does not have a \n at the end) then |
|
728
|
** it still counts. |
|
729
|
*/ |
|
730
|
int blob_linecount(Blob *p){ |
|
731
|
int n = 0; |
|
732
|
int i; |
|
733
|
for(i=0; i<p->nUsed; i++){ |
|
734
|
if( p->aData[i]=='\n' ) n++; |
|
735
|
} |
|
736
|
if( p->nUsed>0 && p->aData[p->nUsed-1]!='\n' ) n++; |
|
737
|
return n; |
|
738
|
} |
|
739
|
|
|
740
|
/* |
|
741
|
** Rewind the cursor on a blob back to the beginning. |
|
742
|
*/ |
|
743
|
void blob_rewind(Blob *p){ |
|
744
|
p->iCursor = 0; |
|
745
|
} |
|
746
|
|
|
747
|
/* |
|
748
|
** Truncate a blob to the specified length in bytes. |
|
749
|
*/ |
|
750
|
void blob_truncate(Blob *p, int sz){ |
|
751
|
if( sz>=0 && sz<(int)(p->nUsed) ) p->nUsed = sz; |
|
752
|
} |
|
753
|
|
|
754
|
/* |
|
755
|
** Truncate a blob to the specified length in bytes. If truncation |
|
756
|
** results in an incomplete UTF-8 sequence at the end, remove up |
|
757
|
** to three more bytes back to the last complete UTF-8 sequence. |
|
758
|
*/ |
|
759
|
void blob_truncate_utf8(Blob *p, int sz){ |
|
760
|
if( sz>=0 && sz<(int)(p->nUsed) ){ |
|
761
|
p->nUsed = utf8_nearest_codepoint(p->aData,sz); |
|
762
|
} |
|
763
|
} |
|
764
|
|
|
765
|
/* |
|
766
|
** Seek the cursor in a blob to the indicated offset. |
|
767
|
*/ |
|
768
|
int blob_seek(Blob *p, int offset, int whence){ |
|
769
|
if( whence==BLOB_SEEK_SET ){ |
|
770
|
p->iCursor = offset; |
|
771
|
}else if( whence==BLOB_SEEK_CUR ){ |
|
772
|
p->iCursor += offset; |
|
773
|
} |
|
774
|
if( p->iCursor>p->nUsed ){ |
|
775
|
p->iCursor = p->nUsed; |
|
776
|
} |
|
777
|
return p->iCursor; |
|
778
|
} |
|
779
|
|
|
780
|
/* |
|
781
|
** Return the current offset into the blob |
|
782
|
*/ |
|
783
|
int blob_tell(Blob *p){ |
|
784
|
return p->iCursor; |
|
785
|
} |
|
786
|
|
|
787
|
/* |
|
788
|
** Extract a single line of text from pFrom beginning at the current |
|
789
|
** cursor location and use that line of text to initialize pTo. |
|
790
|
** pTo will include the terminating \n. Return the number of bytes |
|
791
|
** in the line including the \n at the end. 0 is returned at |
|
792
|
** end-of-file. |
|
793
|
** |
|
794
|
** The cursor of pFrom is left pointing at the first byte past the |
|
795
|
** \n that terminated the line. |
|
796
|
** |
|
797
|
** pTo will be an ephemeral blob. If pFrom changes, it might alter |
|
798
|
** pTo as well. |
|
799
|
*/ |
|
800
|
int blob_line(Blob *pFrom, Blob *pTo){ |
|
801
|
char *aData = pFrom->aData; |
|
802
|
int n = pFrom->nUsed; |
|
803
|
int i = pFrom->iCursor; |
|
804
|
|
|
805
|
while( i<n && aData[i]!='\n' ){ i++; } |
|
806
|
if( i<n ){ |
|
807
|
assert( aData[i]=='\n' ); |
|
808
|
i++; |
|
809
|
} |
|
810
|
blob_extract(pFrom, i-pFrom->iCursor, pTo); |
|
811
|
return pTo->nUsed; |
|
812
|
} |
|
813
|
|
|
814
|
/* |
|
815
|
** Trim whitespace off of the end of a blob. Return the number |
|
816
|
** of characters remaining. |
|
817
|
** |
|
818
|
** All this does is reduce the length counter. This routine does |
|
819
|
** not insert a new zero terminator. |
|
820
|
*/ |
|
821
|
int blob_trim(Blob *p){ |
|
822
|
char *z = p->aData; |
|
823
|
int n = p->nUsed; |
|
824
|
while( n>0 && fossil_isspace(z[n-1]) ){ n--; } |
|
825
|
p->nUsed = n; |
|
826
|
return n; |
|
827
|
} |
|
828
|
|
|
829
|
/* |
|
830
|
** Extract a single token from pFrom and use it to initialize pTo. |
|
831
|
** Return the number of bytes in the token. If no token is found, |
|
832
|
** return 0. |
|
833
|
** |
|
834
|
** A token consists of one or more non-space characters. Leading |
|
835
|
** whitespace is ignored. |
|
836
|
** |
|
837
|
** The cursor of pFrom is left pointing at the first character past |
|
838
|
** the end of the token. |
|
839
|
** |
|
840
|
** pTo will be an ephemeral blob. If pFrom changes, it might alter |
|
841
|
** pTo as well. |
|
842
|
*/ |
|
843
|
int blob_token(Blob *pFrom, Blob *pTo){ |
|
844
|
char *aData = pFrom->aData; |
|
845
|
int n = pFrom->nUsed; |
|
846
|
int i = pFrom->iCursor; |
|
847
|
while( i<n && fossil_isspace(aData[i]) ){ i++; } |
|
848
|
pFrom->iCursor = i; |
|
849
|
while( i<n && !fossil_isspace(aData[i]) ){ i++; } |
|
850
|
blob_extract(pFrom, i-pFrom->iCursor, pTo); |
|
851
|
while( i<n && fossil_isspace(aData[i]) ){ i++; } |
|
852
|
pFrom->iCursor = i; |
|
853
|
return pTo->nUsed; |
|
854
|
} |
|
855
|
|
|
856
|
/* |
|
857
|
** Extract a single SQL token from pFrom and use it to initialize pTo. |
|
858
|
** Return the number of bytes in the token. If no token is found, |
|
859
|
** return 0. |
|
860
|
** |
|
861
|
** An SQL token consists of one or more non-space characters. If the |
|
862
|
** first character is ' then the token is terminated by a matching ' |
|
863
|
** (ignoring double '') or by the end of the string |
|
864
|
** |
|
865
|
** The cursor of pFrom is left pointing at the first character past |
|
866
|
** the end of the token. |
|
867
|
** |
|
868
|
** pTo will be an ephemeral blob. If pFrom changes, it might alter |
|
869
|
** pTo as well. |
|
870
|
*/ |
|
871
|
int blob_sqltoken(Blob *pFrom, Blob *pTo){ |
|
872
|
char *aData = pFrom->aData; |
|
873
|
int n = pFrom->nUsed; |
|
874
|
int i = pFrom->iCursor; |
|
875
|
while( i<n && fossil_isspace(aData[i]) ){ i++; } |
|
876
|
pFrom->iCursor = i; |
|
877
|
if( aData[i]=='\'' ){ |
|
878
|
i++; |
|
879
|
while( i<n ){ |
|
880
|
if( aData[i]=='\'' ){ |
|
881
|
if( aData[++i]!='\'' ) break; |
|
882
|
} |
|
883
|
i++; |
|
884
|
} |
|
885
|
}else{ |
|
886
|
while( i<n && !fossil_isspace(aData[i]) ){ i++; } |
|
887
|
} |
|
888
|
blob_extract(pFrom, i-pFrom->iCursor, pTo); |
|
889
|
while( i<n && fossil_isspace(aData[i]) ){ i++; } |
|
890
|
pFrom->iCursor = i; |
|
891
|
return pTo->nUsed; |
|
892
|
} |
|
893
|
|
|
894
|
/* |
|
895
|
** Extract everything from the current cursor to the end of the blob |
|
896
|
** into a new blob. The new blob is an ephemeral reference to the |
|
897
|
** original blob. The cursor of the original blob is unchanged. |
|
898
|
*/ |
|
899
|
int blob_tail(Blob *pFrom, Blob *pTo){ |
|
900
|
int iCursor = pFrom->iCursor; |
|
901
|
blob_extract(pFrom, pFrom->nUsed-pFrom->iCursor, pTo); |
|
902
|
pFrom->iCursor = iCursor; |
|
903
|
return pTo->nUsed; |
|
904
|
} |
|
905
|
|
|
906
|
/* |
|
907
|
** Copy N lines of text from pFrom into pTo. The copy begins at the |
|
908
|
** current cursor position of pIn. The pIn cursor is left pointing |
|
909
|
** at the first character past the last \n copied. |
|
910
|
** |
|
911
|
** If pTo==NULL then this routine simply skips over N lines. |
|
912
|
*/ |
|
913
|
void blob_copy_lines(Blob *pTo, Blob *pFrom, int N){ |
|
914
|
char *z = pFrom->aData; |
|
915
|
int i = pFrom->iCursor; |
|
916
|
int n = pFrom->nUsed; |
|
917
|
int cnt = 0; |
|
918
|
|
|
919
|
if( N==0 ) return; |
|
920
|
while( i<n ){ |
|
921
|
if( z[i]=='\n' ){ |
|
922
|
cnt++; |
|
923
|
if( cnt==N ){ |
|
924
|
i++; |
|
925
|
break; |
|
926
|
} |
|
927
|
} |
|
928
|
i++; |
|
929
|
} |
|
930
|
if( pTo ){ |
|
931
|
blob_append(pTo, &pFrom->aData[pFrom->iCursor], i - pFrom->iCursor); |
|
932
|
} |
|
933
|
pFrom->iCursor = i; |
|
934
|
} |
|
935
|
|
|
936
|
/* |
|
937
|
** Remove comment lines (starting with '#') from a blob pIn. |
|
938
|
** Keep lines starting with "\#" but remove the initial backslash. |
|
939
|
** |
|
940
|
** Store the result in pOut. It is ok for pIn and pOut to be the same blob. |
|
941
|
** |
|
942
|
** pOut must either be the same as pIn or else uninitialized. |
|
943
|
*/ |
|
944
|
void blob_strip_comment_lines(Blob *pIn, Blob *pOut){ |
|
945
|
char *z = pIn->aData; |
|
946
|
unsigned int i = 0; |
|
947
|
unsigned int n = pIn->nUsed; |
|
948
|
unsigned int lineStart = 0; |
|
949
|
unsigned int copyStart = 0; |
|
950
|
int doCopy = 1; |
|
951
|
Blob temp; |
|
952
|
blob_zero(&temp); |
|
953
|
|
|
954
|
while( i<n ){ |
|
955
|
if( i==lineStart && z[i]=='#' ){ |
|
956
|
copyStart = i; |
|
957
|
doCopy = 0; |
|
958
|
}else if( i==lineStart && z[i]=='\\' && z[i+1]=='#' ){ |
|
959
|
/* keep lines starting with an escaped '#' (and unescape it) */ |
|
960
|
copyStart = i + 1; |
|
961
|
} |
|
962
|
if( z[i]=='\n' ){ |
|
963
|
if( doCopy ) blob_append(&temp,&pIn->aData[copyStart], i - copyStart + 1); |
|
964
|
lineStart = copyStart = i + 1; |
|
965
|
doCopy = 1; |
|
966
|
} |
|
967
|
i++; |
|
968
|
} |
|
969
|
/* Last line */ |
|
970
|
if( doCopy ) blob_append(&temp, &pIn->aData[copyStart], i - copyStart); |
|
971
|
|
|
972
|
if( pOut==pIn ) blob_reset(pOut); |
|
973
|
*pOut = temp; |
|
974
|
} |
|
975
|
|
|
976
|
/* |
|
977
|
** COMMAND: test-strip-comment-lines |
|
978
|
** |
|
979
|
** Usage: %fossil test-strip-comment-lines ?OPTIONS? INPUTFILE |
|
980
|
** |
|
981
|
** Read INPUTFILE and print it without comment lines (starting with '#'). |
|
982
|
** Keep lines starting with "\\#" but remove the initial backslash. |
|
983
|
** |
|
984
|
** This is used to test and debug the blob_strip_comment_lines() routine. |
|
985
|
** |
|
986
|
** Options: |
|
987
|
** -y|--side-by-side Show diff of INPUTFILE and output side-by-side |
|
988
|
** -W|--width N Width of lines in side-by-side diff |
|
989
|
*/ |
|
990
|
void test_strip_comment_lines_cmd(void){ |
|
991
|
Blob f, h; /* uninitialized */ |
|
992
|
Blob out; |
|
993
|
DiffConfig dCfg; |
|
994
|
int sbs = 0; |
|
995
|
const char *z; |
|
996
|
int w = 0; |
|
997
|
|
|
998
|
memset(&dCfg, 0, sizeof(dCfg)); |
|
999
|
|
|
1000
|
sbs = find_option("side-by-side","y",0)!=0; |
|
1001
|
if( (z = find_option("width","W",1))!=0 && (w = atoi(z))>0 ){ |
|
1002
|
dCfg.wColumn = w; |
|
1003
|
} |
|
1004
|
verify_all_options(); |
|
1005
|
if( g.argc!=3 ) usage("INPUTFILE"); |
|
1006
|
|
|
1007
|
blob_read_from_file(&f, g.argv[2], ExtFILE); |
|
1008
|
blob_strip_comment_lines(&f, &h); |
|
1009
|
|
|
1010
|
if ( !sbs ){ |
|
1011
|
blob_write_to_file(&h, "-"); |
|
1012
|
}else{ |
|
1013
|
blob_zero(&out); |
|
1014
|
dCfg.nContext = -1; /* whole content */ |
|
1015
|
dCfg.diffFlags = DIFF_SIDEBYSIDE | DIFF_CONTEXT_EX | DIFF_STRIP_EOLCR; |
|
1016
|
diff_begin(&dCfg); |
|
1017
|
text_diff(&f, &h, &out, &dCfg); |
|
1018
|
blob_write_to_file(&out, "-"); |
|
1019
|
diff_end(&dCfg, 0); |
|
1020
|
} |
|
1021
|
} |
|
1022
|
|
|
1023
|
/* |
|
1024
|
** Ensure that the text in pBlob ends with '\n' |
|
1025
|
*/ |
|
1026
|
void blob_add_final_newline(Blob *pBlob){ |
|
1027
|
if( pBlob->nUsed<=0 ) return; |
|
1028
|
if( pBlob->aData[pBlob->nUsed-1]!='\n' ){ |
|
1029
|
blob_append_char(pBlob, '\n'); |
|
1030
|
} |
|
1031
|
} |
|
1032
|
|
|
1033
|
/* |
|
1034
|
** Return true if the blob contains a valid base16 identifier artifact hash. |
|
1035
|
** |
|
1036
|
** The value returned is actually one of HNAME_SHA1 OR HNAME_K256 if the |
|
1037
|
** hash is valid. Both of these are non-zero and therefore "true". |
|
1038
|
** If the hash is not valid, then HNAME_ERROR is returned, which is zero or |
|
1039
|
** false. |
|
1040
|
*/ |
|
1041
|
int blob_is_hname(Blob *pBlob){ |
|
1042
|
return hname_validate(blob_buffer(pBlob), blob_size(pBlob)); |
|
1043
|
} |
|
1044
|
|
|
1045
|
/* |
|
1046
|
** Return true if the blob contains a valid filename |
|
1047
|
*/ |
|
1048
|
int blob_is_filename(Blob *pBlob){ |
|
1049
|
return file_is_simple_pathname(blob_str(pBlob), 1); |
|
1050
|
} |
|
1051
|
|
|
1052
|
/* |
|
1053
|
** Return true if the blob contains a valid non-negative 32-bit integer |
|
1054
|
** and store the integer value in *pValue. If the blob is not a valid |
|
1055
|
** non-negative 32-bit integer, return false and leave *pValue unchanged. |
|
1056
|
*/ |
|
1057
|
int blob_is_int(Blob *pBlob, int *pValue){ |
|
1058
|
const char *z = blob_buffer(pBlob); |
|
1059
|
int i, n, c; |
|
1060
|
sqlite3_uint64 v; |
|
1061
|
n = blob_size(pBlob); |
|
1062
|
v = 0; |
|
1063
|
for(i=0; i<n && (c = z[i])!=0 && c>='0' && c<='9'; i++){ |
|
1064
|
v = v*10 + c - '0'; |
|
1065
|
if( v>0x7fffffff ) return 0; |
|
1066
|
} |
|
1067
|
if( i==n ){ |
|
1068
|
*pValue = (int)v; |
|
1069
|
return 1; |
|
1070
|
}else{ |
|
1071
|
return 0; |
|
1072
|
} |
|
1073
|
} |
|
1074
|
|
|
1075
|
/* |
|
1076
|
** Return true if the blob contains a valid non-negative 64-bit integer |
|
1077
|
** and store the integer value in *pValue. If the blob is not a valid |
|
1078
|
** non-negative 64-bit integer, return false and leave *pValue unchanged. |
|
1079
|
*/ |
|
1080
|
int blob_is_int64(Blob *pBlob, sqlite3_int64 *pValue){ |
|
1081
|
const char *z = blob_buffer(pBlob); |
|
1082
|
int i, n, c; |
|
1083
|
sqlite3_uint64 v; |
|
1084
|
n = blob_size(pBlob); |
|
1085
|
v = 0; |
|
1086
|
for(i=0; i<n && (c = z[i])!=0 && c>='0' && c<='9'; i++){ |
|
1087
|
if( v<922337203685477580ULL || (v==922337203685477580ULL && c<='7') ){ |
|
1088
|
v = v*10 + c - '0'; |
|
1089
|
}else{ |
|
1090
|
return 0; |
|
1091
|
} |
|
1092
|
} |
|
1093
|
if( i==n ){ |
|
1094
|
*pValue = (sqlite3_int64)v; |
|
1095
|
return 1; |
|
1096
|
}else{ |
|
1097
|
return 0; |
|
1098
|
} |
|
1099
|
} |
|
1100
|
|
|
1101
|
/* |
|
1102
|
** COMMAND: test-atoi |
|
1103
|
** |
|
1104
|
** Use the blob_is_int() and blob_is_int64() routines to convert arguments |
|
1105
|
** into integers. Used for unit testing of those routines. |
|
1106
|
*/ |
|
1107
|
void blob_is_int_cmd(void){ |
|
1108
|
int i; |
|
1109
|
for(i=2; i<g.argc; i++){ |
|
1110
|
Blob x; |
|
1111
|
int i32 = 0; |
|
1112
|
sqlite3_int64 i64 = 0; |
|
1113
|
int rc; |
|
1114
|
blob_init(&x, g.argv[i], -1); |
|
1115
|
rc = blob_is_int(&x, &i32); |
|
1116
|
fossil_print("%20s: 32-bit %d %-10d", g.argv[i], rc, i32); |
|
1117
|
rc = blob_is_int64(&x, &i64); |
|
1118
|
fossil_print(" 64-bit %d %lld\n", rc, i64); |
|
1119
|
} |
|
1120
|
} |
|
1121
|
|
|
1122
|
/* |
|
1123
|
** Zero or reset an array of Blobs. |
|
1124
|
*/ |
|
1125
|
void blobarray_zero(Blob *aBlob, int n){ |
|
1126
|
int i; |
|
1127
|
for(i=0; i<n; i++) blob_zero(&aBlob[i]); |
|
1128
|
} |
|
1129
|
void blobarray_reset(Blob *aBlob, int n){ |
|
1130
|
int i; |
|
1131
|
for(i=0; i<n; i++) blob_reset(&aBlob[i]); |
|
1132
|
} |
|
1133
|
/* |
|
1134
|
** Allocate array of n blobs and initialize each element with `empty_blob` |
|
1135
|
*/ |
|
1136
|
Blob* blobarray_new(int n){ |
|
1137
|
int i; |
|
1138
|
Blob *aBlob = fossil_malloc(sizeof(Blob)*n); |
|
1139
|
for(i=0; i<n; i++) aBlob[i] = empty_blob; |
|
1140
|
return aBlob; |
|
1141
|
} |
|
1142
|
/* |
|
1143
|
** Free array of n blobs some of which may be empty (have NULL buffer) |
|
1144
|
*/ |
|
1145
|
void blobarray_delete(Blob *aBlob, int n){ |
|
1146
|
int i; |
|
1147
|
for(i=0; i<n; i++){ |
|
1148
|
if( blob_buffer(aBlob+i) ) blob_reset(aBlob+i); |
|
1149
|
} |
|
1150
|
fossil_free(aBlob); |
|
1151
|
} |
|
1152
|
|
|
1153
|
/* |
|
1154
|
** Parse a blob into space-separated tokens. Store each token in |
|
1155
|
** an element of the blobarray aToken[]. aToken[] is nToken elements in |
|
1156
|
** size. Return the number of tokens seen. |
|
1157
|
*/ |
|
1158
|
int blob_tokenize(Blob *pIn, Blob *aToken, int nToken){ |
|
1159
|
int i; |
|
1160
|
for(i=0; i<nToken && blob_token(pIn, &aToken[i]); i++){} |
|
1161
|
return i; |
|
1162
|
} |
|
1163
|
|
|
1164
|
/* |
|
1165
|
** Do printf-style string rendering and append the results to a blob. Or |
|
1166
|
** if pBlob==0, do printf-style string rendering directly to stdout. |
|
1167
|
** |
|
1168
|
** The blob_appendf() version sets the BLOBFLAG_NotSQL bit in Blob.blobFlags |
|
1169
|
** whereas blob_append_sql() does not. |
|
1170
|
*/ |
|
1171
|
void blob_appendf(Blob *pBlob, const char *zFormat, ...){ |
|
1172
|
va_list ap; |
|
1173
|
va_start(ap, zFormat); |
|
1174
|
vxprintf(pBlob, zFormat, ap); |
|
1175
|
va_end(ap); |
|
1176
|
if( pBlob ) pBlob->blobFlags |= BLOBFLAG_NotSQL; |
|
1177
|
} |
|
1178
|
void blob_append_sql(Blob *pBlob, const char *zFormat, ...){ |
|
1179
|
va_list ap; |
|
1180
|
va_start(ap, zFormat); |
|
1181
|
vxprintf(pBlob, zFormat, ap); |
|
1182
|
va_end(ap); |
|
1183
|
} |
|
1184
|
void blob_vappendf(Blob *pBlob, const char *zFormat, va_list ap){ |
|
1185
|
vxprintf(pBlob, zFormat, ap); |
|
1186
|
} |
|
1187
|
|
|
1188
|
/* |
|
1189
|
** Initialize a blob to the data on an input channel. Return |
|
1190
|
** the number of bytes read into the blob. Any prior content |
|
1191
|
** of the blob is discarded, not freed. |
|
1192
|
*/ |
|
1193
|
int blob_read_from_channel(Blob *pBlob, FILE *in, int nToRead){ |
|
1194
|
size_t n; |
|
1195
|
blob_zero(pBlob); |
|
1196
|
if( nToRead<0 ){ |
|
1197
|
char zBuf[10000]; |
|
1198
|
while( !feof(in) ){ |
|
1199
|
n = fread(zBuf, 1, sizeof(zBuf), in); |
|
1200
|
if( n>0 ){ |
|
1201
|
blob_append(pBlob, zBuf, n); |
|
1202
|
} |
|
1203
|
} |
|
1204
|
}else{ |
|
1205
|
blob_resize(pBlob, nToRead); |
|
1206
|
n = fread(blob_buffer(pBlob), 1, nToRead, in); |
|
1207
|
blob_resize(pBlob, n); |
|
1208
|
} |
|
1209
|
return blob_size(pBlob); |
|
1210
|
} |
|
1211
|
|
|
1212
|
/* |
|
1213
|
** Initialize a blob to the data read from HTTP input. Return |
|
1214
|
** the number of bytes read into the blob. Any prior content |
|
1215
|
** of the blob is discarded, not freed. |
|
1216
|
*/ |
|
1217
|
int blob_read_from_cgi(Blob *pBlob, int nToRead){ |
|
1218
|
size_t n; |
|
1219
|
blob_zero(pBlob); |
|
1220
|
if( nToRead<0 ){ |
|
1221
|
char zBuf[10000]; |
|
1222
|
while( !cgi_feof() ){ |
|
1223
|
n = cgi_fread(zBuf, sizeof(zBuf)); |
|
1224
|
if( n>0 ){ |
|
1225
|
blob_append(pBlob, zBuf, n); |
|
1226
|
} |
|
1227
|
} |
|
1228
|
}else{ |
|
1229
|
blob_resize(pBlob, nToRead); |
|
1230
|
n = cgi_fread(blob_buffer(pBlob), nToRead); |
|
1231
|
blob_resize(pBlob, n); |
|
1232
|
} |
|
1233
|
return blob_size(pBlob); |
|
1234
|
} |
|
1235
|
|
|
1236
|
/* |
|
1237
|
** Initialize a blob to be the content of a file. If the filename |
|
1238
|
** is blank or "-" then read from standard input. |
|
1239
|
** |
|
1240
|
** If zFilename is a symbolic link, behavior depends on the eFType |
|
1241
|
** parameter: |
|
1242
|
** |
|
1243
|
** * If eFType is ExtFILE or allow-symlinks is OFF, then the |
|
1244
|
** pBlob is initialized to the *content* of the object to which |
|
1245
|
** the zFilename symlink points. |
|
1246
|
** |
|
1247
|
** * If eFType is RepoFILE and allow-symlinks is ON, then the |
|
1248
|
** pBlob is initialized to the *name* of the object to which |
|
1249
|
** the zFilename symlink points. |
|
1250
|
** |
|
1251
|
** Any prior content of the blob is discarded, not freed. |
|
1252
|
** |
|
1253
|
** Return the number of bytes read. Calls fossil_fatal() on error (i.e. |
|
1254
|
** it exit()s and does not return). |
|
1255
|
*/ |
|
1256
|
sqlite3_int64 blob_read_from_file( |
|
1257
|
Blob *pBlob, /* The blob to be initialized */ |
|
1258
|
const char *zFilename, /* Extract content from this file */ |
|
1259
|
int eFType /* ExtFILE or RepoFILE - see above */ |
|
1260
|
){ |
|
1261
|
sqlite3_int64 size, got; |
|
1262
|
FILE *in; |
|
1263
|
if( zFilename==0 || zFilename[0]==0 |
|
1264
|
|| (zFilename[0]=='-' && zFilename[1]==0) ){ |
|
1265
|
return blob_read_from_channel(pBlob, stdin, -1); |
|
1266
|
} |
|
1267
|
if( file_islink(zFilename) ){ |
|
1268
|
return blob_read_link(pBlob, zFilename); |
|
1269
|
} |
|
1270
|
size = file_size(zFilename, eFType); |
|
1271
|
blob_zero(pBlob); |
|
1272
|
if( size<0 ){ |
|
1273
|
fossil_fatal("no such file: %s", zFilename); |
|
1274
|
} |
|
1275
|
if( size==0 ){ |
|
1276
|
return 0; |
|
1277
|
} |
|
1278
|
blob_resize(pBlob, size); |
|
1279
|
in = fossil_fopen(zFilename, "rb"); |
|
1280
|
if( in==0 ){ |
|
1281
|
fossil_fatal("cannot open %s for reading", zFilename); |
|
1282
|
} |
|
1283
|
got = fread(blob_buffer(pBlob), 1, size, in); |
|
1284
|
fclose(in); |
|
1285
|
if( got<size ){ |
|
1286
|
blob_resize(pBlob, got); |
|
1287
|
} |
|
1288
|
return got; |
|
1289
|
} |
|
1290
|
|
|
1291
|
/* |
|
1292
|
** Reads symlink destination path and puts int into blob. |
|
1293
|
** Any prior content of the blob is discarded, not freed. |
|
1294
|
** |
|
1295
|
** Returns length of destination path. |
|
1296
|
** |
|
1297
|
** On windows, zeros blob and returns 0. |
|
1298
|
*/ |
|
1299
|
int blob_read_link(Blob *pBlob, const char *zFilename){ |
|
1300
|
#if !defined(_WIN32) |
|
1301
|
char zBuf[1024]; |
|
1302
|
ssize_t len = readlink(zFilename, zBuf, 1023); |
|
1303
|
if( len < 0 ){ |
|
1304
|
fossil_fatal("cannot read symbolic link %s", zFilename); |
|
1305
|
} |
|
1306
|
zBuf[len] = 0; /* null-terminate */ |
|
1307
|
blob_zero(pBlob); |
|
1308
|
blob_appendf(pBlob, "%s", zBuf); |
|
1309
|
return len; |
|
1310
|
#else |
|
1311
|
blob_zero(pBlob); |
|
1312
|
return 0; |
|
1313
|
#endif |
|
1314
|
} |
|
1315
|
|
|
1316
|
/* |
|
1317
|
** Write the content of a blob into a file. |
|
1318
|
** |
|
1319
|
** If the filename is blank or "-" then write to standard output. |
|
1320
|
** |
|
1321
|
** This routine always assumes ExtFILE. If zFilename is a symbolic link |
|
1322
|
** then the content is written into the object that symbolic link points |
|
1323
|
** to, not into the symbolic link itself. This is true regardless of |
|
1324
|
** the allow-symlinks setting. |
|
1325
|
** |
|
1326
|
** Return the number of bytes written. |
|
1327
|
*/ |
|
1328
|
int blob_write_to_file(Blob *pBlob, const char *zFilename){ |
|
1329
|
FILE *out; |
|
1330
|
int nWrote; |
|
1331
|
|
|
1332
|
if( zFilename[0]==0 || (zFilename[0]=='-' && zFilename[1]==0) ){ |
|
1333
|
blob_is_init(pBlob); |
|
1334
|
#if defined(_WIN32) |
|
1335
|
nWrote = fossil_utf8_to_console(blob_buffer(pBlob), blob_size(pBlob), 0); |
|
1336
|
if( nWrote>=0 ) return nWrote; |
|
1337
|
fflush(stdout); |
|
1338
|
_setmode(_fileno(stdout), _O_BINARY); |
|
1339
|
#endif |
|
1340
|
nWrote = fwrite(blob_buffer(pBlob), 1, blob_size(pBlob), stdout); |
|
1341
|
#if defined(_WIN32) |
|
1342
|
fflush(stdout); |
|
1343
|
_setmode(_fileno(stdout), _O_TEXT); |
|
1344
|
#endif |
|
1345
|
}else{ |
|
1346
|
file_mkfolder(zFilename, ExtFILE, 1, 0); |
|
1347
|
out = fossil_fopen(zFilename, "wb"); |
|
1348
|
if( out==0 ){ |
|
1349
|
#if defined(_WIN32) |
|
1350
|
const char *zReserved = file_is_win_reserved(zFilename); |
|
1351
|
if( zReserved ){ |
|
1352
|
fossil_fatal("cannot open \"%s\" because \"%s\" is " |
|
1353
|
"a reserved name on Windows", zFilename, zReserved); |
|
1354
|
} |
|
1355
|
#endif |
|
1356
|
fossil_fatal_recursive("unable to open file \"%s\" for writing", |
|
1357
|
zFilename); |
|
1358
|
return 0; |
|
1359
|
} |
|
1360
|
blob_is_init(pBlob); |
|
1361
|
nWrote = fwrite(blob_buffer(pBlob), 1, blob_size(pBlob), out); |
|
1362
|
fclose(out); |
|
1363
|
if( nWrote!=(int)blob_size(pBlob) ){ |
|
1364
|
fossil_fatal_recursive("short write: %d of %d bytes to %s", nWrote, |
|
1365
|
blob_size(pBlob), zFilename); |
|
1366
|
} |
|
1367
|
} |
|
1368
|
return nWrote; |
|
1369
|
} |
|
1370
|
|
|
1371
|
/* |
|
1372
|
** Compress a blob pIn. Store the result in pOut. It is ok for pIn and |
|
1373
|
** pOut to be the same blob. |
|
1374
|
** |
|
1375
|
** pOut must either be the same as pIn or else uninitialized. |
|
1376
|
*/ |
|
1377
|
void blob_compress(Blob *pIn, Blob *pOut){ |
|
1378
|
unsigned int nIn = blob_size(pIn); |
|
1379
|
unsigned int nOut = 13 + nIn + (nIn+999)/1000; |
|
1380
|
unsigned long int nOut2; |
|
1381
|
unsigned char *outBuf; |
|
1382
|
Blob temp; |
|
1383
|
blob_zero(&temp); |
|
1384
|
blob_resize(&temp, nOut+4); |
|
1385
|
outBuf = (unsigned char*)blob_buffer(&temp); |
|
1386
|
outBuf[0] = nIn>>24 & 0xff; |
|
1387
|
outBuf[1] = nIn>>16 & 0xff; |
|
1388
|
outBuf[2] = nIn>>8 & 0xff; |
|
1389
|
outBuf[3] = nIn & 0xff; |
|
1390
|
nOut2 = (long int)nOut; |
|
1391
|
compress(&outBuf[4], &nOut2, |
|
1392
|
(unsigned char*)blob_buffer(pIn), blob_size(pIn)); |
|
1393
|
if( pOut==pIn ) blob_reset(pOut); |
|
1394
|
assert_blob_is_reset(pOut); |
|
1395
|
*pOut = temp; |
|
1396
|
blob_resize(pOut, nOut2+4); |
|
1397
|
} |
|
1398
|
|
|
1399
|
/* |
|
1400
|
** COMMAND: test-compress |
|
1401
|
** |
|
1402
|
** Usage: %fossil test-compress INPUTFILE OUTPUTFILE |
|
1403
|
** |
|
1404
|
** Run compression on INPUTFILE and write the result into OUTPUTFILE. |
|
1405
|
** |
|
1406
|
** This is used to test and debug the blob_compress() routine. |
|
1407
|
*/ |
|
1408
|
void compress_cmd(void){ |
|
1409
|
Blob f; |
|
1410
|
if( g.argc!=4 ) usage("INPUTFILE OUTPUTFILE"); |
|
1411
|
blob_read_from_file(&f, g.argv[2], ExtFILE); |
|
1412
|
blob_compress(&f, &f); |
|
1413
|
blob_write_to_file(&f, g.argv[3]); |
|
1414
|
} |
|
1415
|
|
|
1416
|
/* |
|
1417
|
** Compress the concatenation of a blobs pIn1 and pIn2. Store the result |
|
1418
|
** in pOut. |
|
1419
|
** |
|
1420
|
** pOut must be either uninitialized or must be the same as either pIn1 or |
|
1421
|
** pIn2. |
|
1422
|
*/ |
|
1423
|
void blob_compress2(Blob *pIn1, Blob *pIn2, Blob *pOut){ |
|
1424
|
unsigned int nIn = blob_size(pIn1) + blob_size(pIn2); |
|
1425
|
unsigned int nOut = 13 + nIn + (nIn+999)/1000; |
|
1426
|
unsigned char *outBuf; |
|
1427
|
z_stream stream; |
|
1428
|
Blob temp; |
|
1429
|
blob_zero(&temp); |
|
1430
|
blob_resize(&temp, nOut+4); |
|
1431
|
outBuf = (unsigned char*)blob_buffer(&temp); |
|
1432
|
outBuf[0] = nIn>>24 & 0xff; |
|
1433
|
outBuf[1] = nIn>>16 & 0xff; |
|
1434
|
outBuf[2] = nIn>>8 & 0xff; |
|
1435
|
outBuf[3] = nIn & 0xff; |
|
1436
|
stream.zalloc = (alloc_func)0; |
|
1437
|
stream.zfree = (free_func)0; |
|
1438
|
stream.opaque = 0; |
|
1439
|
stream.avail_out = nOut; |
|
1440
|
stream.next_out = &outBuf[4]; |
|
1441
|
deflateInit(&stream, 9); |
|
1442
|
stream.avail_in = blob_size(pIn1); |
|
1443
|
stream.next_in = (unsigned char*)blob_buffer(pIn1); |
|
1444
|
deflate(&stream, 0); |
|
1445
|
stream.avail_in = blob_size(pIn2); |
|
1446
|
stream.next_in = (unsigned char*)blob_buffer(pIn2); |
|
1447
|
deflate(&stream, 0); |
|
1448
|
deflate(&stream, Z_FINISH); |
|
1449
|
blob_resize(&temp, stream.total_out + 4); |
|
1450
|
deflateEnd(&stream); |
|
1451
|
if( pOut==pIn1 ) blob_reset(pOut); |
|
1452
|
if( pOut==pIn2 ) blob_reset(pOut); |
|
1453
|
assert_blob_is_reset(pOut); |
|
1454
|
*pOut = temp; |
|
1455
|
} |
|
1456
|
|
|
1457
|
/* |
|
1458
|
** COMMAND: test-compress-2 |
|
1459
|
** |
|
1460
|
** Usage: %fossil test-compress-2 IN1 IN2 OUT |
|
1461
|
** |
|
1462
|
** Read files IN1 and IN2, concatenate the content, compress the |
|
1463
|
** content, then write results into OUT. |
|
1464
|
** |
|
1465
|
** This is used to test and debug the blob_compress2() routine. |
|
1466
|
*/ |
|
1467
|
void compress2_cmd(void){ |
|
1468
|
Blob f1, f2; |
|
1469
|
if( g.argc!=5 ) usage("INPUTFILE1 INPUTFILE2 OUTPUTFILE"); |
|
1470
|
blob_read_from_file(&f1, g.argv[2], ExtFILE); |
|
1471
|
blob_read_from_file(&f2, g.argv[3], ExtFILE); |
|
1472
|
blob_compress2(&f1, &f2, &f1); |
|
1473
|
blob_write_to_file(&f1, g.argv[4]); |
|
1474
|
} |
|
1475
|
|
|
1476
|
/* |
|
1477
|
** Uncompress blob pIn and store the result in pOut. It is ok for pIn and |
|
1478
|
** pOut to be the same blob. |
|
1479
|
** |
|
1480
|
** pOut must be either uninitialized or the same as pIn. |
|
1481
|
*/ |
|
1482
|
int blob_uncompress(Blob *pIn, Blob *pOut){ |
|
1483
|
unsigned int nOut; |
|
1484
|
unsigned char *inBuf; |
|
1485
|
unsigned int nIn = blob_size(pIn); |
|
1486
|
Blob temp; |
|
1487
|
int rc; |
|
1488
|
unsigned long int nOut2; |
|
1489
|
if( nIn<=4 ){ |
|
1490
|
return 0; |
|
1491
|
} |
|
1492
|
inBuf = (unsigned char*)blob_buffer(pIn); |
|
1493
|
nOut = (inBuf[0]<<24) + (inBuf[1]<<16) + (inBuf[2]<<8) + inBuf[3]; |
|
1494
|
blob_zero(&temp); |
|
1495
|
blob_resize(&temp, nOut+1); |
|
1496
|
nOut2 = (long int)nOut; |
|
1497
|
rc = uncompress((unsigned char*)blob_buffer(&temp), &nOut2, |
|
1498
|
&inBuf[4], nIn - 4); |
|
1499
|
if( rc!=Z_OK ){ |
|
1500
|
blob_reset(&temp); |
|
1501
|
return 1; |
|
1502
|
} |
|
1503
|
blob_resize(&temp, nOut2); |
|
1504
|
if( pOut==pIn ) blob_reset(pOut); |
|
1505
|
assert_blob_is_reset(pOut); |
|
1506
|
*pOut = temp; |
|
1507
|
return 0; |
|
1508
|
} |
|
1509
|
|
|
1510
|
/* |
|
1511
|
** COMMAND: test-uncompress |
|
1512
|
** |
|
1513
|
** Usage: %fossil test-uncompress IN OUT |
|
1514
|
** |
|
1515
|
** Read the content of file IN, uncompress that content, and write the |
|
1516
|
** result into OUT. This command is intended for testing of the |
|
1517
|
** blob_compress() function. |
|
1518
|
*/ |
|
1519
|
void uncompress_cmd(void){ |
|
1520
|
Blob f; |
|
1521
|
if( g.argc!=4 ) usage("INPUTFILE OUTPUTFILE"); |
|
1522
|
blob_read_from_file(&f, g.argv[2], ExtFILE); |
|
1523
|
blob_uncompress(&f, &f); |
|
1524
|
blob_write_to_file(&f, g.argv[3]); |
|
1525
|
} |
|
1526
|
|
|
1527
|
/* |
|
1528
|
** COMMAND: test-cycle-compress |
|
1529
|
** |
|
1530
|
** Compress and uncompress each file named on the command line. |
|
1531
|
** Verify that the original content is recovered. |
|
1532
|
*/ |
|
1533
|
void test_cycle_compress(void){ |
|
1534
|
int i; |
|
1535
|
Blob b1, b2, b3; |
|
1536
|
for(i=2; i<g.argc; i++){ |
|
1537
|
blob_read_from_file(&b1, g.argv[i], ExtFILE); |
|
1538
|
blob_compress(&b1, &b2); |
|
1539
|
blob_uncompress(&b2, &b3); |
|
1540
|
if( blob_compare(&b1, &b3) ){ |
|
1541
|
fossil_fatal("compress/uncompress cycle failed for %s", g.argv[i]); |
|
1542
|
} |
|
1543
|
blob_reset(&b1); |
|
1544
|
blob_reset(&b2); |
|
1545
|
blob_reset(&b3); |
|
1546
|
} |
|
1547
|
fossil_print("ok\n"); |
|
1548
|
} |
|
1549
|
|
|
1550
|
/* |
|
1551
|
** Convert every \n character in the given blob into \r\n. |
|
1552
|
*/ |
|
1553
|
void blob_add_cr(Blob *p){ |
|
1554
|
char *z = p->aData; |
|
1555
|
int j = p->nUsed; |
|
1556
|
int i, n; |
|
1557
|
for(i=n=0; i<j; i++){ |
|
1558
|
if( z[i]=='\n' ) n++; |
|
1559
|
} |
|
1560
|
j += n; |
|
1561
|
if( j>=(int)(p->nAlloc) ){ |
|
1562
|
blob_resize(p, j); |
|
1563
|
z = p->aData; |
|
1564
|
} |
|
1565
|
p->nUsed = j; |
|
1566
|
z[j] = 0; |
|
1567
|
while( j>i ){ |
|
1568
|
if( (z[--j] = z[--i]) =='\n' ){ |
|
1569
|
z[--j] = '\r'; |
|
1570
|
} |
|
1571
|
} |
|
1572
|
} |
|
1573
|
|
|
1574
|
/* |
|
1575
|
** Remove every \r character from the given blob, replacing each one with |
|
1576
|
** a \n character if it was not already part of a \r\n pair. |
|
1577
|
*/ |
|
1578
|
void blob_to_lf_only(Blob *p){ |
|
1579
|
int i, j; |
|
1580
|
char *z = blob_materialize(p); |
|
1581
|
for(i=j=0; z[i]; i++){ |
|
1582
|
if( z[i]!='\r' ) z[j++] = z[i]; |
|
1583
|
else if( z[i+1]!='\n' ) z[j++] = '\n'; |
|
1584
|
} |
|
1585
|
z[j] = 0; |
|
1586
|
p->nUsed = j; |
|
1587
|
} |
|
1588
|
|
|
1589
|
/* |
|
1590
|
** Convert blob from cp1252 to UTF-8. As cp1252 is a superset |
|
1591
|
** of iso8859-1, this is useful on UNIX as well. |
|
1592
|
** |
|
1593
|
** This table contains the character translations for 0x80..0xA0. |
|
1594
|
*/ |
|
1595
|
|
|
1596
|
static const unsigned short cp1252[32] = { |
|
1597
|
0x20ac, 0x81, 0x201A, 0x0192, 0x201E, 0x2026, 0x2020, 0x2021, |
|
1598
|
0x02C6, 0x2030, 0x0160, 0x2039, 0x0152, 0x8D, 0x017D, 0x8F, |
|
1599
|
0x90, 0x2018, 0x2019, 0x201C, 0x201D, 0x2022, 0x2013, 0x2014, |
|
1600
|
0x2DC, 0x2122, 0x0161, 0x203A, 0x0153, 0x9D, 0x017E, 0x0178 |
|
1601
|
}; |
|
1602
|
|
|
1603
|
void blob_cp1252_to_utf8(Blob *p){ |
|
1604
|
unsigned char *z = (unsigned char *)p->aData; |
|
1605
|
int j = p->nUsed; |
|
1606
|
int i, n; |
|
1607
|
for(i=n=0; i<j; i++){ |
|
1608
|
if( z[i]>=0x80 ){ |
|
1609
|
if( (z[i]<0xa0) && (cp1252[z[i]&0x1f]>=0x800) ){ |
|
1610
|
n++; |
|
1611
|
} |
|
1612
|
n++; |
|
1613
|
} |
|
1614
|
} |
|
1615
|
j += n; |
|
1616
|
if( j>=(int)(p->nAlloc) ){ |
|
1617
|
blob_resize(p, j); |
|
1618
|
z = (unsigned char *)p->aData; |
|
1619
|
} |
|
1620
|
p->nUsed = j; |
|
1621
|
z[j] = 0; |
|
1622
|
while( j>i ){ |
|
1623
|
if( z[--i]>=0x80 ){ |
|
1624
|
if( z[i]<0xa0 ){ |
|
1625
|
unsigned short sym = cp1252[z[i]&0x1f]; |
|
1626
|
if( sym>=0x800 ){ |
|
1627
|
z[--j] = 0x80 | (sym&0x3f); |
|
1628
|
z[--j] = 0x80 | ((sym>>6)&0x3f); |
|
1629
|
z[--j] = 0xe0 | (sym>>12); |
|
1630
|
}else{ |
|
1631
|
z[--j] = 0x80 | (sym&0x3f); |
|
1632
|
z[--j] = 0xc0 | (sym>>6); |
|
1633
|
} |
|
1634
|
}else{ |
|
1635
|
z[--j] = 0x80 | (z[i]&0x3f); |
|
1636
|
z[--j] = 0xC0 | (z[i]>>6); |
|
1637
|
} |
|
1638
|
}else{ |
|
1639
|
z[--j] = z[i]; |
|
1640
|
} |
|
1641
|
} |
|
1642
|
} |
|
1643
|
|
|
1644
|
/* |
|
1645
|
** ASCII (for reference): |
|
1646
|
** x0 x1 x2 x3 x4 x5 x6 x7 x8 x9 xa xb xc xd xe xf |
|
1647
|
** 0x ^` ^a ^b ^c ^d ^e ^f ^g \b \t \n () \f \r ^n ^o |
|
1648
|
** 1x ^p ^q ^r ^s ^t ^u ^v ^w ^x ^y ^z ^{ ^| ^} ^~ ^ |
|
1649
|
** 2x () ! " # $ % & ' ( ) * + , - . / |
|
1650
|
** 3x 0 1 2 3 4 5 6 7 8 9 : ; < = > ? |
|
1651
|
** 4x @ A B C D E F G H I J K L M N O |
|
1652
|
** 5x P Q R S T U V W X Y Z [ \ ] ^ _ |
|
1653
|
** 6x ` a b c d e f g h i j k l m n o |
|
1654
|
** 7x p q r s t u v w x y z { | } ~ ^_ |
|
1655
|
*/ |
|
1656
|
|
|
1657
|
/* |
|
1658
|
** Meanings for bytes in a filename: |
|
1659
|
** |
|
1660
|
** 0 Ordinary character. No encoding required |
|
1661
|
** 1 Needs to be escaped |
|
1662
|
** 2 Illegal character. Do not allow in a filename |
|
1663
|
** 3 First byte of a 2-byte UTF-8 |
|
1664
|
** 4 First byte of a 3-byte UTF-8 |
|
1665
|
** 5 First byte of a 4-byte UTF-8 |
|
1666
|
*/ |
|
1667
|
static const char aSafeChar[256] = { |
|
1668
|
#ifdef _WIN32 |
|
1669
|
/* Windows |
|
1670
|
** Prohibit: all control characters, including tab, \r and \n. |
|
1671
|
** Escape: (space) " # $ % & ' ( ) * ; < > ? [ ] ^ ` { | } |
|
1672
|
*/ |
|
1673
|
/* x0 x1 x2 x3 x4 x5 x6 x7 x8 x9 xa xb xc xd xe xf */ |
|
1674
|
2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, /* 0x */ |
|
1675
|
2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, /* 1x */ |
|
1676
|
1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, /* 2x */ |
|
1677
|
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 1, 1, /* 3x */ |
|
1678
|
1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 4x */ |
|
1679
|
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 1, 1, 0, /* 5x */ |
|
1680
|
1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 6x */ |
|
1681
|
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 0, 1, /* 7x */ |
|
1682
|
#else |
|
1683
|
/* Unix |
|
1684
|
** Prohibit: all control characters, including tab, \r and \n |
|
1685
|
** Escape: (space) ! " # $ % & ' ( ) * ; < > ? [ \ ] ^ ` { | } |
|
1686
|
*/ |
|
1687
|
/* x0 x1 x2 x3 x4 x5 x6 x7 x8 x9 xa xb xc xd xe xf */ |
|
1688
|
2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, /* 0x */ |
|
1689
|
2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, /* 1x */ |
|
1690
|
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, /* 2x */ |
|
1691
|
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 1, 1, /* 3x */ |
|
1692
|
1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 4x */ |
|
1693
|
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 0, /* 5x */ |
|
1694
|
1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 6x */ |
|
1695
|
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 0, 1, /* 7x */ |
|
1696
|
#endif |
|
1697
|
/* all bytes 0x80 through 0xbf are unescaped, being secondary |
|
1698
|
** bytes to UTF8 characters. Bytes 0xc0 through 0xff are the |
|
1699
|
** first byte of a UTF8 character and do get escaped */ |
|
1700
|
2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, /* 8x */ |
|
1701
|
2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, /* 9x */ |
|
1702
|
2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, /* ax */ |
|
1703
|
2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, /* bx */ |
|
1704
|
3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, /* cx */ |
|
1705
|
3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, /* dx */ |
|
1706
|
4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, /* ex */ |
|
1707
|
5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5 /* fx */ |
|
1708
|
}; |
|
1709
|
|
|
1710
|
/* |
|
1711
|
** pBlob is a shell command under construction. This routine safely |
|
1712
|
** appends filename argument zIn. |
|
1713
|
** |
|
1714
|
** The argument is escaped if it contains white space or other characters |
|
1715
|
** that need to be escaped for the shell. If zIn contains characters |
|
1716
|
** that cannot be safely escaped, then throw a fatal error. |
|
1717
|
** |
|
1718
|
** If the isFilename argument is true, then the argument is expected |
|
1719
|
** to be a filename. As shell commands commonly have command-line |
|
1720
|
** options that begin with "-" and since we do not want an attacker |
|
1721
|
** to be able to invoke these switches using filenames that begin |
|
1722
|
** with "-", if zIn begins with "-", prepend an additional "./" |
|
1723
|
** (or ".\\" on Windows). |
|
1724
|
*/ |
|
1725
|
void blob_append_escaped_arg(Blob *pBlob, const char *zIn, int isFilename){ |
|
1726
|
int i; |
|
1727
|
unsigned char c; |
|
1728
|
int needEscape = 0; |
|
1729
|
int n = blob_size(pBlob); |
|
1730
|
char *z = blob_buffer(pBlob); |
|
1731
|
|
|
1732
|
/* Look for illegal byte-sequences and byte-sequences that require |
|
1733
|
** escaping. No control-characters are allowed. All spaces and |
|
1734
|
** non-ASCII unicode characters and some punctuation characters require |
|
1735
|
** escaping. */ |
|
1736
|
for(i=0; (c = (unsigned char)zIn[i])!=0; i++){ |
|
1737
|
if( aSafeChar[c] ){ |
|
1738
|
unsigned char x = aSafeChar[c]; |
|
1739
|
needEscape = 1; |
|
1740
|
if( x==2 ){ |
|
1741
|
Blob bad; |
|
1742
|
blob_token(pBlob, &bad); |
|
1743
|
fossil_fatal("the [%s] argument to the \"%s\" command contains " |
|
1744
|
"a character (ascii 0x%02x) that is not allowed in " |
|
1745
|
"filename arguments", |
|
1746
|
zIn, blob_str(&bad), c); |
|
1747
|
}else if( x>2 ){ |
|
1748
|
if( (zIn[i+1]&0xc0)!=0x80 |
|
1749
|
|| (x>=4 && (zIn[i+2]&0xc0)!=0x80) |
|
1750
|
|| (x==5 && (zIn[i+3]&0xc0)!=0x80) |
|
1751
|
){ |
|
1752
|
Blob bad; |
|
1753
|
blob_token(pBlob, &bad); |
|
1754
|
fossil_fatal("the [%s] argument to the \"%s\" command contains " |
|
1755
|
"an illegal UTF-8 character", |
|
1756
|
zIn, blob_str(&bad)); |
|
1757
|
} |
|
1758
|
i += x-2; |
|
1759
|
} |
|
1760
|
} |
|
1761
|
} |
|
1762
|
|
|
1763
|
/* Separate from the previous argument by a space */ |
|
1764
|
if( n>0 && !fossil_isspace(z[n-1]) ){ |
|
1765
|
blob_append_char(pBlob, ' '); |
|
1766
|
} |
|
1767
|
|
|
1768
|
/* Check for characters that need quoting */ |
|
1769
|
if( !needEscape ){ |
|
1770
|
if( isFilename && zIn[0]=='-' ){ |
|
1771
|
blob_append_char(pBlob, '.'); |
|
1772
|
#if defined(_WIN32) |
|
1773
|
blob_append_char(pBlob, '\\'); |
|
1774
|
#else |
|
1775
|
blob_append_char(pBlob, '/'); |
|
1776
|
#endif |
|
1777
|
} |
|
1778
|
blob_append(pBlob, zIn, -1); |
|
1779
|
}else{ |
|
1780
|
#if defined(_WIN32) |
|
1781
|
/* Quoting strategy for windows: |
|
1782
|
** Put the entire name inside of "...". Any " characters within |
|
1783
|
** the name get doubled. |
|
1784
|
*/ |
|
1785
|
blob_append_char(pBlob, '"'); |
|
1786
|
if( isFilename && zIn[0]=='-' ){ |
|
1787
|
blob_append_char(pBlob, '.'); |
|
1788
|
blob_append_char(pBlob, '\\'); |
|
1789
|
}else if( zIn[0]=='/' ){ |
|
1790
|
blob_append_char(pBlob, '.'); |
|
1791
|
} |
|
1792
|
for(i=0; (c = (unsigned char)zIn[i])!=0; i++){ |
|
1793
|
blob_append_char(pBlob, (char)c); |
|
1794
|
if( c=='"' ) blob_append_char(pBlob, '"'); |
|
1795
|
if( c=='\\' ) blob_append_char(pBlob, '\\'); |
|
1796
|
if( c=='%' && isFilename ) blob_append(pBlob, "%cd:~,%", 7); |
|
1797
|
} |
|
1798
|
blob_append_char(pBlob, '"'); |
|
1799
|
#else |
|
1800
|
/* Quoting strategy for unix: |
|
1801
|
** If the name does not contain ', then surround the whole thing |
|
1802
|
** with '...'. If there is one or more ' characters within the |
|
1803
|
** name, then put \ before each special character. |
|
1804
|
*/ |
|
1805
|
if( strchr(zIn,'\'') ){ |
|
1806
|
if( isFilename && zIn[0]=='-' ){ |
|
1807
|
blob_append_char(pBlob, '.'); |
|
1808
|
blob_append_char(pBlob, '/'); |
|
1809
|
} |
|
1810
|
for(i=0; (c = (unsigned char)zIn[i])!=0; i++){ |
|
1811
|
if( aSafeChar[c] && aSafeChar[c]!=2 ) blob_append_char(pBlob, '\\'); |
|
1812
|
blob_append_char(pBlob, (char)c); |
|
1813
|
} |
|
1814
|
}else{ |
|
1815
|
blob_append_char(pBlob, '\''); |
|
1816
|
if( isFilename && zIn[0]=='-' ){ |
|
1817
|
blob_append_char(pBlob, '.'); |
|
1818
|
blob_append_char(pBlob, '/'); |
|
1819
|
} |
|
1820
|
blob_append(pBlob, zIn, -1); |
|
1821
|
blob_append_char(pBlob, '\''); |
|
1822
|
} |
|
1823
|
#endif |
|
1824
|
} |
|
1825
|
} |
|
1826
|
|
|
1827
|
/* |
|
1828
|
** COMMAND: test-escaped-arg |
|
1829
|
** |
|
1830
|
** Usage %fossil ARGS ... |
|
1831
|
** |
|
1832
|
** Run each argument through blob_append_escaped_arg() and show the |
|
1833
|
** result. Append each argument to "fossil test-echo" and run that |
|
1834
|
** using fossil_system() to verify that it really does get escaped |
|
1835
|
** correctly. |
|
1836
|
** |
|
1837
|
** Other options: |
|
1838
|
** |
|
1839
|
** --filename-args BOOL Subsequent arguments are assumed to be |
|
1840
|
** filenames if BOOL is true, or not if BOOL |
|
1841
|
** is false. Defaults on. |
|
1842
|
** |
|
1843
|
** --hex HEX Skip the --hex flag and instead decode HEX |
|
1844
|
** into ascii. This provides a way to insert |
|
1845
|
** unusual characters as an argument for testing. |
|
1846
|
** |
|
1847
|
** --compare HEX ASCII Verify that argument ASCII is identical to |
|
1848
|
** to decoded HEX. |
|
1849
|
** |
|
1850
|
** --fuzz N Run N fuzz cases. Each case is a call |
|
1851
|
** to "fossil test-escaped-arg --compare HEX ARG" |
|
1852
|
** where HEX and ARG are the same argument. |
|
1853
|
** The argument is chosen at random. |
|
1854
|
*/ |
|
1855
|
void test_escaped_arg_command(void){ |
|
1856
|
int i; |
|
1857
|
Blob x; |
|
1858
|
const char *zArg; |
|
1859
|
int isFilename = 1; |
|
1860
|
char zBuf[100]; |
|
1861
|
blob_init(&x, 0, 0); |
|
1862
|
for(i=2; i<g.argc; i++){ |
|
1863
|
zArg = g.argv[i]; |
|
1864
|
if( fossil_strcmp(zArg, "--hex")==0 && i+1<g.argc ){ |
|
1865
|
size_t n = strlen(g.argv[++i]); |
|
1866
|
if( n>=(sizeof(zBuf)-1)*2 ){ |
|
1867
|
fossil_fatal("Argument to --hex is too big"); |
|
1868
|
} |
|
1869
|
memset(zBuf, 0, sizeof(zBuf)); |
|
1870
|
decode16((const unsigned char*)g.argv[i], (unsigned char*)zBuf, (int)n); |
|
1871
|
zArg = zBuf; |
|
1872
|
}else if( fossil_strcmp(zArg, "--compare")==0 && i+2<g.argc ){ |
|
1873
|
size_t n = strlen(g.argv[++i]); |
|
1874
|
if( n>=(sizeof(zBuf)-1)*2 ){ |
|
1875
|
fossil_fatal("HEX argument to --compare is too big"); |
|
1876
|
} |
|
1877
|
memset(zBuf, 0, sizeof(zBuf)); |
|
1878
|
if( decode16((const unsigned char*)g.argv[i], (unsigned char*)zBuf, |
|
1879
|
(int)n) ){ |
|
1880
|
fossil_fatal("HEX decode of %s failed", g.argv[i]); |
|
1881
|
} |
|
1882
|
zArg = g.argv[++i]; |
|
1883
|
if( zArg[0]=='-' ){ |
|
1884
|
fossil_fatal("filename argument \"%s\" begins with \"-\"", zArg); |
|
1885
|
} |
|
1886
|
#ifdef _WIN32 |
|
1887
|
if( zBuf[0]=='-' && zArg[0]=='.' && zArg[1]=='\\' ) zArg += 2; |
|
1888
|
#else |
|
1889
|
if( zBuf[0]=='-' && zArg[0]=='.' && zArg[1]=='/' ) zArg += 2; |
|
1890
|
#endif |
|
1891
|
if( strcmp(zBuf, zArg)!=0 ){ |
|
1892
|
fossil_fatal("argument disagree: \"%s\" (%s) versus \"%s\"", |
|
1893
|
zBuf, g.argv[i-1], zArg); |
|
1894
|
} |
|
1895
|
continue; |
|
1896
|
}else if( fossil_strcmp(zArg, "--fuzz")==0 && i+1<g.argc ){ |
|
1897
|
int n = atoi(g.argv[++i]); |
|
1898
|
int j; |
|
1899
|
for(j=0; j<n; j++){ |
|
1900
|
unsigned char m, k; |
|
1901
|
int rc; |
|
1902
|
unsigned char zWord[100]; |
|
1903
|
sqlite3_randomness(sizeof(m), &m); |
|
1904
|
m = (m%40)+5; |
|
1905
|
sqlite3_randomness(m, zWord); /* Between 5 and 45 bytes of randomness */ |
|
1906
|
for(k=0; k<m; k++){ |
|
1907
|
unsigned char cx = zWord[k]; |
|
1908
|
if( cx<0x20 || cx>=0x7f ){ |
|
1909
|
/* Translate illegal bytes into various non-ASCII unicode |
|
1910
|
** characters in order to exercise those code paths */ |
|
1911
|
unsigned int u; |
|
1912
|
if( cx>=0x7f ){ |
|
1913
|
u = cx; |
|
1914
|
}else if( cx>=0x08 ){ |
|
1915
|
u = 0x800 + cx; |
|
1916
|
}else{ |
|
1917
|
u = 0x10000 + cx; |
|
1918
|
} |
|
1919
|
if( u<0x00080 ){ |
|
1920
|
zWord[k] = u & 0xFF; |
|
1921
|
}else if( u<0x00800 ){ |
|
1922
|
zWord[k++] = 0xC0 + (u8)((u>>6)&0x1F); |
|
1923
|
zWord[k] = 0x80 + (u8)(u & 0x3F); |
|
1924
|
}else if( u<0x10000 ){ |
|
1925
|
zWord[k++] = 0xE0 + (u8)((u>>12)&0x0F); |
|
1926
|
zWord[k++] = 0x80 + (u8)((u>>6) & 0x3F); |
|
1927
|
zWord[k] = 0x80 + (u8)(u & 0x3F); |
|
1928
|
}else{ |
|
1929
|
zWord[k++] = 0xF0 + (u8)((u>>18) & 0x07); |
|
1930
|
zWord[k++] = 0x80 + (u8)((u>>12) & 0x3F); |
|
1931
|
zWord[k++] = 0x80 + (u8)((u>>6) & 0x3F); |
|
1932
|
zWord[k] = 0x80 + (u8)(u & 0x3F); |
|
1933
|
} |
|
1934
|
} |
|
1935
|
} |
|
1936
|
zWord[k] = 0; |
|
1937
|
encode16(zWord, (unsigned char*)zBuf, (int)k); |
|
1938
|
blob_appendf(&x, "%$ test-escaped-arg --compare %s %$", |
|
1939
|
g.nameOfExe, zBuf,zWord); |
|
1940
|
rc = fossil_system(blob_str(&x)); |
|
1941
|
if( rc ) fossil_fatal("failed test (%d): %s\n", rc, blob_str(&x)); |
|
1942
|
blob_reset(&x); |
|
1943
|
} |
|
1944
|
continue; |
|
1945
|
}else if( fossil_strcmp(zArg, "--filename-args")==0 ){ |
|
1946
|
if( i+1<g.argc ){ |
|
1947
|
i++; |
|
1948
|
isFilename = is_truth(g.argv[i]); |
|
1949
|
} |
|
1950
|
continue; |
|
1951
|
} |
|
1952
|
fossil_print("%3d [%s]: ", i, zArg); |
|
1953
|
if( isFilename ){ |
|
1954
|
blob_appendf(&x, "%$ test-echo %$", g.nameOfExe, zArg); |
|
1955
|
}else{ |
|
1956
|
blob_appendf(&x, "%$ test-echo %!$", g.nameOfExe, zArg); |
|
1957
|
} |
|
1958
|
fossil_print("%s\n", blob_str(&x)); |
|
1959
|
fossil_system(blob_str(&x)); |
|
1960
|
blob_reset(&x); |
|
1961
|
} |
|
1962
|
} |
|
1963
|
|
|
1964
|
/* |
|
1965
|
** A read(2)-like impl for the Blob class. Reads (copies) up to nLen |
|
1966
|
** bytes from pIn, starting at position pIn->iCursor, and copies them |
|
1967
|
** to pDest (which must be valid memory at least nLen bytes long). |
|
1968
|
** |
|
1969
|
** Returns the number of bytes read/copied, which may be less than |
|
1970
|
** nLen (if end-of-blob is encountered). |
|
1971
|
** |
|
1972
|
** Updates pIn's cursor. |
|
1973
|
** |
|
1974
|
** Returns 0 if pIn contains no data. |
|
1975
|
*/ |
|
1976
|
unsigned int blob_read(Blob *pIn, void * pDest, unsigned int nLen ){ |
|
1977
|
if( !pIn->aData || (pIn->iCursor >= pIn->nUsed) ){ |
|
1978
|
return 0; |
|
1979
|
} else if( (pIn->iCursor + nLen) > (unsigned int)pIn->nUsed ){ |
|
1980
|
nLen = (unsigned int) (pIn->nUsed - pIn->iCursor); |
|
1981
|
} |
|
1982
|
assert( pIn->nUsed > pIn->iCursor ); |
|
1983
|
assert( (pIn->iCursor+nLen) <= pIn->nUsed ); |
|
1984
|
if( nLen ){ |
|
1985
|
memcpy( pDest, pIn->aData, nLen ); |
|
1986
|
pIn->iCursor += nLen; |
|
1987
|
} |
|
1988
|
return nLen; |
|
1989
|
} |
|
1990
|
|
|
1991
|
/* |
|
1992
|
** Swaps the contents of the given blobs. Results |
|
1993
|
** are unspecified if either value is NULL or both |
|
1994
|
** point to the same blob. |
|
1995
|
*/ |
|
1996
|
void blob_swap( Blob *pLeft, Blob *pRight ){ |
|
1997
|
Blob swap = *pLeft; |
|
1998
|
*pLeft = *pRight; |
|
1999
|
*pRight = swap; |
|
2000
|
} |
|
2001
|
|
|
2002
|
/* |
|
2003
|
** Strip a possible byte-order-mark (BOM) from the blob. On Windows, if there |
|
2004
|
** is either no BOM at all or an (le/be) UTF-16 BOM, a conversion to UTF-8 is |
|
2005
|
** done. If useMbcs is false and there is no BOM, the input string is assumed |
|
2006
|
** to be UTF-8 already, so no conversion is done. |
|
2007
|
*/ |
|
2008
|
void blob_to_utf8_no_bom(Blob *pBlob, int useMbcs){ |
|
2009
|
char *zUtf8; |
|
2010
|
int bomSize = 0; |
|
2011
|
int bomReverse = 0; |
|
2012
|
if( starts_with_utf8_bom(pBlob, &bomSize) ){ |
|
2013
|
struct Blob temp; |
|
2014
|
zUtf8 = blob_str(pBlob) + bomSize; |
|
2015
|
blob_zero(&temp); |
|
2016
|
blob_append(&temp, zUtf8, -1); |
|
2017
|
blob_swap(pBlob, &temp); |
|
2018
|
blob_reset(&temp); |
|
2019
|
}else if( starts_with_utf16_bom(pBlob, &bomSize, &bomReverse) ){ |
|
2020
|
zUtf8 = blob_buffer(pBlob); |
|
2021
|
if( bomReverse ){ |
|
2022
|
/* Found BOM, but with reversed bytes */ |
|
2023
|
unsigned int i = blob_size(pBlob); |
|
2024
|
while( i>1 ){ |
|
2025
|
/* swap bytes of unicode representation */ |
|
2026
|
char zTemp = zUtf8[--i]; |
|
2027
|
zUtf8[i] = zUtf8[i-1]; |
|
2028
|
zUtf8[--i] = zTemp; |
|
2029
|
} |
|
2030
|
} |
|
2031
|
/* Make sure the blob contains two terminating 0-bytes */ |
|
2032
|
blob_append(pBlob, "\000\000", 3); |
|
2033
|
zUtf8 = blob_str(pBlob) + bomSize; |
|
2034
|
zUtf8 = fossil_unicode_to_utf8(zUtf8); |
|
2035
|
blob_reset(pBlob); |
|
2036
|
blob_set_dynamic(pBlob, zUtf8); |
|
2037
|
}else if( useMbcs && invalid_utf8(pBlob) ){ |
|
2038
|
#if defined(_WIN32) |
|
2039
|
zUtf8 = fossil_mbcs_to_utf8(blob_str(pBlob)); |
|
2040
|
blob_reset(pBlob); |
|
2041
|
blob_append(pBlob, zUtf8, -1); |
|
2042
|
fossil_mbcs_free(zUtf8); |
|
2043
|
#else |
|
2044
|
blob_cp1252_to_utf8(pBlob); |
|
2045
|
#endif /* _WIN32 */ |
|
2046
|
} |
|
2047
|
} |
|
2048
|
|