| line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
|
1
|
|
|
|
|
|
|
/* |
|
2
|
|
|
|
|
|
|
* pdfmake_parser.c — PDF object parser + cross-reference reader |
|
3
|
|
|
|
|
|
|
* |
|
4
|
|
|
|
|
|
|
* Implementation of recursive-descent parser for PDF objects and |
|
5
|
|
|
|
|
|
|
* xref table/stream parsing. |
|
6
|
|
|
|
|
|
|
*/ |
|
7
|
|
|
|
|
|
|
|
|
8
|
|
|
|
|
|
|
#include "pdfmake_parser.h" |
|
9
|
|
|
|
|
|
|
#include "pdfmake_filter.h" |
|
10
|
|
|
|
|
|
|
#include |
|
11
|
|
|
|
|
|
|
#include |
|
12
|
|
|
|
|
|
|
#include |
|
13
|
|
|
|
|
|
|
#include |
|
14
|
|
|
|
|
|
|
|
|
15
|
|
|
|
|
|
|
/*============================================================================ |
|
16
|
|
|
|
|
|
|
* Internal helpers |
|
17
|
|
|
|
|
|
|
*==========================================================================*/ |
|
18
|
|
|
|
|
|
|
|
|
19
|
|
|
|
|
|
|
/* Set error state */ |
|
20
|
60
|
|
|
|
|
|
static void set_error(pdfmake_parser_t *p, pdfmake_err_t err, |
|
21
|
|
|
|
|
|
|
size_t offset, const char *msg) { |
|
22
|
60
|
|
|
|
|
|
p->last_err = err; |
|
23
|
60
|
|
|
|
|
|
p->err_offset = offset; |
|
24
|
60
|
50
|
|
|
|
|
if (msg) { |
|
25
|
60
|
|
|
|
|
|
strncpy(p->err_msg, msg, sizeof(p->err_msg) - 1); |
|
26
|
60
|
|
|
|
|
|
p->err_msg[sizeof(p->err_msg) - 1] = '\0'; |
|
27
|
|
|
|
|
|
|
} else { |
|
28
|
0
|
|
|
|
|
|
p->err_msg[0] = '\0'; |
|
29
|
|
|
|
|
|
|
} |
|
30
|
60
|
|
|
|
|
|
} |
|
31
|
|
|
|
|
|
|
|
|
32
|
|
|
|
|
|
|
/* Ensure xref table has capacity for object num */ |
|
33
|
1617
|
|
|
|
|
|
static int ensure_xref_cap(pdfmake_parser_t *p, uint32_t num) { |
|
34
|
|
|
|
|
|
|
size_t new_cap; |
|
35
|
|
|
|
|
|
|
pdfmake_xref_entry_t *new_xref; |
|
36
|
|
|
|
|
|
|
|
|
37
|
1617
|
100
|
|
|
|
|
if (num < p->xref_cap) return 1; |
|
38
|
|
|
|
|
|
|
|
|
39
|
117
|
50
|
|
|
|
|
new_cap = p->xref_cap ? p->xref_cap * 2 : 64; |
|
40
|
117
|
50
|
|
|
|
|
while (new_cap <= num) new_cap *= 2; |
|
41
|
|
|
|
|
|
|
|
|
42
|
117
|
|
|
|
|
|
new_xref = realloc(p->xref, new_cap * sizeof(pdfmake_xref_entry_t)); |
|
43
|
117
|
50
|
|
|
|
|
if (!new_xref) return 0; |
|
44
|
|
|
|
|
|
|
|
|
45
|
|
|
|
|
|
|
/* Zero new entries */ |
|
46
|
117
|
|
|
|
|
|
memset(new_xref + p->xref_cap, 0, (new_cap - p->xref_cap) * sizeof(pdfmake_xref_entry_t)); |
|
47
|
117
|
|
|
|
|
|
p->xref = new_xref; |
|
48
|
117
|
|
|
|
|
|
p->xref_cap = new_cap; |
|
49
|
117
|
|
|
|
|
|
return 1; |
|
50
|
|
|
|
|
|
|
} |
|
51
|
|
|
|
|
|
|
|
|
52
|
|
|
|
|
|
|
/* Check if offset was already visited in /Prev chain (cycle detection) */ |
|
53
|
48
|
|
|
|
|
|
static int prev_visited(pdfmake_parser_t *p, uint64_t offset) { |
|
54
|
|
|
|
|
|
|
size_t i; |
|
55
|
96
|
100
|
|
|
|
|
for (i = 0; i < p->prev_count; i++) { |
|
56
|
80
|
100
|
|
|
|
|
if (p->prev_offsets[i] == offset) return 1; |
|
57
|
|
|
|
|
|
|
} |
|
58
|
16
|
|
|
|
|
|
return 0; |
|
59
|
|
|
|
|
|
|
} |
|
60
|
|
|
|
|
|
|
|
|
61
|
|
|
|
|
|
|
/* Add offset to /Prev visited list */ |
|
62
|
228
|
|
|
|
|
|
static int prev_add(pdfmake_parser_t *p, uint64_t offset) { |
|
63
|
228
|
100
|
|
|
|
|
if (p->prev_count >= p->prev_cap) { |
|
64
|
116
|
50
|
|
|
|
|
size_t new_cap = p->prev_cap ? p->prev_cap * 2 : 8; |
|
65
|
116
|
|
|
|
|
|
uint64_t *new_arr = realloc(p->prev_offsets, new_cap * sizeof(uint64_t)); |
|
66
|
116
|
50
|
|
|
|
|
if (!new_arr) return 0; |
|
67
|
116
|
|
|
|
|
|
p->prev_offsets = new_arr; |
|
68
|
116
|
|
|
|
|
|
p->prev_cap = new_cap; |
|
69
|
|
|
|
|
|
|
} |
|
70
|
228
|
|
|
|
|
|
p->prev_offsets[p->prev_count++] = offset; |
|
71
|
228
|
|
|
|
|
|
return 1; |
|
72
|
|
|
|
|
|
|
} |
|
73
|
|
|
|
|
|
|
|
|
74
|
|
|
|
|
|
|
/* Get interned name ID from arena */ |
|
75
|
2043
|
|
|
|
|
|
static uint32_t intern_name(pdfmake_parser_t *p, const char *bytes, size_t len) { |
|
76
|
2043
|
|
|
|
|
|
return pdfmake_arena_intern_name(p->doc->arena, bytes, len); |
|
77
|
|
|
|
|
|
|
} |
|
78
|
|
|
|
|
|
|
|
|
79
|
|
|
|
|
|
|
/* Decode hex digit */ |
|
80
|
14583
|
|
|
|
|
|
static int hex_digit(int c) { |
|
81
|
14583
|
50
|
|
|
|
|
if (c >= '0' && c <= '9') return c - '0'; |
|
|
|
100
|
|
|
|
|
|
|
82
|
5873
|
100
|
|
|
|
|
if (c >= 'a' && c <= 'f') return c - 'a' + 10; |
|
|
|
50
|
|
|
|
|
|
|
83
|
3951
|
50
|
|
|
|
|
if (c >= 'A' && c <= 'F') return c - 'A' + 10; |
|
|
|
50
|
|
|
|
|
|
|
84
|
0
|
|
|
|
|
|
return -1; |
|
85
|
|
|
|
|
|
|
} |
|
86
|
|
|
|
|
|
|
|
|
87
|
|
|
|
|
|
|
/*============================================================================ |
|
88
|
|
|
|
|
|
|
* Parser lifecycle |
|
89
|
|
|
|
|
|
|
*==========================================================================*/ |
|
90
|
|
|
|
|
|
|
|
|
91
|
122
|
|
|
|
|
|
pdfmake_parser_t *pdfmake_parser_new(const uint8_t *buf, size_t len) { |
|
92
|
122
|
|
|
|
|
|
pdfmake_parser_t *p = calloc(1, sizeof(pdfmake_parser_t)); |
|
93
|
122
|
50
|
|
|
|
|
if (!p) return NULL; |
|
94
|
|
|
|
|
|
|
|
|
95
|
122
|
|
|
|
|
|
p->buf = buf; |
|
96
|
122
|
|
|
|
|
|
p->buf_len = len; |
|
97
|
122
|
|
|
|
|
|
pdfmake_tokenizer_init(&p->tok, buf, len); |
|
98
|
|
|
|
|
|
|
|
|
99
|
122
|
|
|
|
|
|
p->doc = pdfmake_doc_new(); |
|
100
|
122
|
50
|
|
|
|
|
if (!p->doc) { |
|
101
|
0
|
|
|
|
|
|
free(p); |
|
102
|
0
|
|
|
|
|
|
return NULL; |
|
103
|
|
|
|
|
|
|
} |
|
104
|
|
|
|
|
|
|
|
|
105
|
122
|
|
|
|
|
|
return p; |
|
106
|
|
|
|
|
|
|
} |
|
107
|
|
|
|
|
|
|
|
|
108
|
122
|
|
|
|
|
|
void pdfmake_parser_free(pdfmake_parser_t *parser) { |
|
109
|
122
|
50
|
|
|
|
|
if (!parser) return; |
|
110
|
122
|
|
|
|
|
|
free(parser->xref); |
|
111
|
122
|
|
|
|
|
|
free(parser->prev_offsets); |
|
112
|
|
|
|
|
|
|
/* Note: does NOT free doc - caller owns it */ |
|
113
|
122
|
|
|
|
|
|
free(parser); |
|
114
|
|
|
|
|
|
|
} |
|
115
|
|
|
|
|
|
|
|
|
116
|
118
|
|
|
|
|
|
void pdfmake_parser_set_repair(pdfmake_parser_t *parser, int enable) { |
|
117
|
118
|
50
|
|
|
|
|
if (parser) parser->repair = enable ? 1 : 0; |
|
118
|
118
|
|
|
|
|
|
} |
|
119
|
|
|
|
|
|
|
|
|
120
|
0
|
|
|
|
|
|
const char *pdfmake_parser_errmsg(pdfmake_parser_t *parser) { |
|
121
|
0
|
0
|
|
|
|
|
return parser ? parser->err_msg : "NULL parser"; |
|
122
|
|
|
|
|
|
|
} |
|
123
|
|
|
|
|
|
|
|
|
124
|
0
|
|
|
|
|
|
size_t pdfmake_parser_erroffset(pdfmake_parser_t *parser) { |
|
125
|
0
|
0
|
|
|
|
|
return parser ? parser->err_offset : 0; |
|
126
|
|
|
|
|
|
|
} |
|
127
|
|
|
|
|
|
|
|
|
128
|
|
|
|
|
|
|
/*============================================================================ |
|
129
|
|
|
|
|
|
|
* PDF header validation |
|
130
|
|
|
|
|
|
|
*==========================================================================*/ |
|
131
|
|
|
|
|
|
|
|
|
132
|
218
|
|
|
|
|
|
pdfmake_err_t pdfmake_check_header(pdfmake_parser_t *p, int *major, int *minor) { |
|
133
|
|
|
|
|
|
|
/* Look for %PDF-N.M in first 1024 bytes */ |
|
134
|
218
|
|
|
|
|
|
const char *needle = "%PDF-"; |
|
135
|
218
|
|
|
|
|
|
size_t search_len = p->buf_len < 1024 ? p->buf_len : 1024; |
|
136
|
218
|
|
|
|
|
|
const uint8_t *found = NULL; |
|
137
|
218
|
|
|
|
|
|
int maj = 0, min = 0; |
|
138
|
|
|
|
|
|
|
const uint8_t *ptr; |
|
139
|
|
|
|
|
|
|
size_t i; |
|
140
|
|
|
|
|
|
|
|
|
141
|
218
|
50
|
|
|
|
|
for (i = 0; i + 7 < search_len; i++) { |
|
142
|
218
|
50
|
|
|
|
|
if (memcmp(p->buf + i, needle, 5) == 0) { |
|
143
|
218
|
|
|
|
|
|
found = p->buf + i; |
|
144
|
218
|
|
|
|
|
|
break; |
|
145
|
|
|
|
|
|
|
} |
|
146
|
|
|
|
|
|
|
} |
|
147
|
|
|
|
|
|
|
|
|
148
|
218
|
50
|
|
|
|
|
if (!found) { |
|
149
|
0
|
|
|
|
|
|
set_error(p, PDFMAKE_EHEADER, 0, "PDF header not found"); |
|
150
|
0
|
|
|
|
|
|
return PDFMAKE_EHEADER; |
|
151
|
|
|
|
|
|
|
} |
|
152
|
|
|
|
|
|
|
|
|
153
|
|
|
|
|
|
|
/* Parse version */ |
|
154
|
218
|
|
|
|
|
|
ptr = found + 5; |
|
155
|
436
|
100
|
|
|
|
|
while (isdigit(*ptr)) maj = maj * 10 + (*ptr++ - '0'); |
|
156
|
218
|
50
|
|
|
|
|
if (*ptr == '.') ptr++; |
|
157
|
436
|
100
|
|
|
|
|
while (isdigit(*ptr)) min = min * 10 + (*ptr++ - '0'); |
|
158
|
|
|
|
|
|
|
|
|
159
|
218
|
50
|
|
|
|
|
if (major) *major = maj; |
|
160
|
218
|
50
|
|
|
|
|
if (minor) *minor = min; |
|
161
|
|
|
|
|
|
|
|
|
162
|
218
|
|
|
|
|
|
return PDFMAKE_OK; |
|
163
|
|
|
|
|
|
|
} |
|
164
|
|
|
|
|
|
|
|
|
165
|
|
|
|
|
|
|
/*============================================================================ |
|
166
|
|
|
|
|
|
|
* Object parsing — recursive descent |
|
167
|
|
|
|
|
|
|
*==========================================================================*/ |
|
168
|
|
|
|
|
|
|
|
|
169
|
|
|
|
|
|
|
/* Forward declaration for mutual recursion */ |
|
170
|
|
|
|
|
|
|
static pdfmake_obj_t parse_object_internal(pdfmake_parser_t *p); |
|
171
|
|
|
|
|
|
|
|
|
172
|
|
|
|
|
|
|
/* Parse name token - decode #XX escapes */ |
|
173
|
7003
|
|
|
|
|
|
static pdfmake_obj_t parse_name(pdfmake_parser_t *p, pdfmake_tok_t *tok) { |
|
174
|
|
|
|
|
|
|
/* Skip leading / - use tok.buf since it may be repositioned */ |
|
175
|
7003
|
|
|
|
|
|
const uint8_t *src = p->tok.buf + tok->offset + 1; |
|
176
|
7003
|
|
|
|
|
|
size_t src_len = tok->length - 1; |
|
177
|
7003
|
|
|
|
|
|
int has_escape = 0; |
|
178
|
|
|
|
|
|
|
uint8_t *decoded; |
|
179
|
|
|
|
|
|
|
size_t out_len; |
|
180
|
|
|
|
|
|
|
size_t i; |
|
181
|
|
|
|
|
|
|
|
|
182
|
|
|
|
|
|
|
/* Check if we need to decode #XX */ |
|
183
|
49242
|
100
|
|
|
|
|
for (i = 0; i < src_len; i++) { |
|
184
|
42240
|
100
|
|
|
|
|
if (src[i] == '#') { has_escape = 1; break; } |
|
185
|
|
|
|
|
|
|
} |
|
186
|
|
|
|
|
|
|
|
|
187
|
7003
|
100
|
|
|
|
|
if (!has_escape) { |
|
188
|
7002
|
|
|
|
|
|
return pdfmake_name(p->doc->arena, (const char *)src, src_len); |
|
189
|
|
|
|
|
|
|
} |
|
190
|
|
|
|
|
|
|
|
|
191
|
|
|
|
|
|
|
/* Decode escapes */ |
|
192
|
1
|
|
|
|
|
|
decoded = pdfmake_arena_alloc(p->doc->arena, src_len); |
|
193
|
1
|
50
|
|
|
|
|
if (!decoded) return pdfmake_null(); |
|
194
|
|
|
|
|
|
|
|
|
195
|
1
|
|
|
|
|
|
out_len = 0; |
|
196
|
6
|
100
|
|
|
|
|
for (i = 0; i < src_len; i++) { |
|
197
|
5
|
100
|
|
|
|
|
if (src[i] == '#' && i + 2 < src_len) { |
|
|
|
50
|
|
|
|
|
|
|
198
|
1
|
|
|
|
|
|
int hi = hex_digit(src[i + 1]); |
|
199
|
1
|
|
|
|
|
|
int lo = hex_digit(src[i + 2]); |
|
200
|
1
|
50
|
|
|
|
|
if (hi >= 0 && lo >= 0) { |
|
|
|
50
|
|
|
|
|
|
|
201
|
1
|
|
|
|
|
|
decoded[out_len++] = (uint8_t)((hi << 4) | lo); |
|
202
|
1
|
|
|
|
|
|
i += 2; |
|
203
|
1
|
|
|
|
|
|
continue; |
|
204
|
|
|
|
|
|
|
} |
|
205
|
|
|
|
|
|
|
} |
|
206
|
4
|
|
|
|
|
|
decoded[out_len++] = src[i]; |
|
207
|
|
|
|
|
|
|
} |
|
208
|
|
|
|
|
|
|
|
|
209
|
1
|
|
|
|
|
|
return pdfmake_name(p->doc->arena, (const char *)decoded, out_len); |
|
210
|
|
|
|
|
|
|
} |
|
211
|
|
|
|
|
|
|
|
|
212
|
|
|
|
|
|
|
/* Parse literal string - decode escape sequences */ |
|
213
|
47
|
|
|
|
|
|
static pdfmake_obj_t parse_literal_string(pdfmake_parser_t *p, pdfmake_tok_t *tok) { |
|
214
|
|
|
|
|
|
|
/* Content is between ( and ) - skip outer parens - use tok.buf */ |
|
215
|
47
|
|
|
|
|
|
const uint8_t *src = p->tok.buf + tok->offset + 1; |
|
216
|
47
|
|
|
|
|
|
size_t src_len = tok->length - 2; |
|
217
|
|
|
|
|
|
|
uint8_t *decoded; |
|
218
|
|
|
|
|
|
|
size_t out_len; |
|
219
|
|
|
|
|
|
|
int paren_depth; |
|
220
|
|
|
|
|
|
|
size_t i; |
|
221
|
|
|
|
|
|
|
|
|
222
|
|
|
|
|
|
|
/* Decode string escapes */ |
|
223
|
47
|
|
|
|
|
|
decoded = pdfmake_arena_alloc(p->doc->arena, src_len + 1); |
|
224
|
47
|
50
|
|
|
|
|
if (!decoded) return pdfmake_null(); |
|
225
|
|
|
|
|
|
|
|
|
226
|
47
|
|
|
|
|
|
out_len = 0; |
|
227
|
47
|
|
|
|
|
|
paren_depth = 0; |
|
228
|
|
|
|
|
|
|
|
|
229
|
1244
|
100
|
|
|
|
|
for (i = 0; i < src_len; i++) { |
|
230
|
1197
|
|
|
|
|
|
uint8_t c = src[i]; |
|
231
|
|
|
|
|
|
|
|
|
232
|
1197
|
100
|
|
|
|
|
if (c == '\\' && i + 1 < src_len) { |
|
|
|
50
|
|
|
|
|
|
|
233
|
3
|
|
|
|
|
|
uint8_t next = src[i + 1]; |
|
234
|
3
|
|
|
|
|
|
switch (next) { |
|
235
|
1
|
|
|
|
|
|
case 'n': decoded[out_len++] = '\n'; i++; continue; |
|
236
|
0
|
|
|
|
|
|
case 'r': decoded[out_len++] = '\r'; i++; continue; |
|
237
|
0
|
|
|
|
|
|
case 't': decoded[out_len++] = '\t'; i++; continue; |
|
238
|
0
|
|
|
|
|
|
case 'b': decoded[out_len++] = '\b'; i++; continue; |
|
239
|
0
|
|
|
|
|
|
case 'f': decoded[out_len++] = '\f'; i++; continue; |
|
240
|
1
|
|
|
|
|
|
case '(': decoded[out_len++] = '('; i++; continue; |
|
241
|
1
|
|
|
|
|
|
case ')': decoded[out_len++] = ')'; i++; continue; |
|
242
|
0
|
|
|
|
|
|
case '\\': decoded[out_len++] = '\\'; i++; continue; |
|
243
|
0
|
|
|
|
|
|
case '\n': i++; continue; /* Line continuation LF */ |
|
244
|
0
|
|
|
|
|
|
case '\r': /* Line continuation CR or CRLF */ |
|
245
|
0
|
|
|
|
|
|
i++; |
|
246
|
0
|
0
|
|
|
|
|
if (i + 1 < src_len && src[i + 1] == '\n') i++; |
|
|
|
0
|
|
|
|
|
|
|
247
|
0
|
|
|
|
|
|
continue; |
|
248
|
0
|
|
|
|
|
|
default: |
|
249
|
|
|
|
|
|
|
/* Octal escape? */ |
|
250
|
0
|
0
|
|
|
|
|
if (next >= '0' && next <= '7') { |
|
|
|
0
|
|
|
|
|
|
|
251
|
0
|
|
|
|
|
|
int val = next - '0'; |
|
252
|
0
|
|
|
|
|
|
i++; |
|
253
|
0
|
0
|
|
|
|
|
if (i + 1 < src_len && src[i + 1] >= '0' && src[i + 1] <= '7') { |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
254
|
0
|
|
|
|
|
|
val = val * 8 + (src[++i] - '0'); |
|
255
|
0
|
0
|
|
|
|
|
if (i + 1 < src_len && src[i + 1] >= '0' && src[i + 1] <= '7') { |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
256
|
0
|
|
|
|
|
|
val = val * 8 + (src[++i] - '0'); |
|
257
|
|
|
|
|
|
|
} |
|
258
|
|
|
|
|
|
|
} |
|
259
|
0
|
|
|
|
|
|
decoded[out_len++] = (uint8_t)(val & 0xFF); |
|
260
|
0
|
|
|
|
|
|
continue; |
|
261
|
|
|
|
|
|
|
} |
|
262
|
|
|
|
|
|
|
/* Unknown escape - just keep the backslash per PDF spec */ |
|
263
|
0
|
|
|
|
|
|
decoded[out_len++] = c; |
|
264
|
0
|
|
|
|
|
|
continue; |
|
265
|
|
|
|
|
|
|
} |
|
266
|
|
|
|
|
|
|
} |
|
267
|
|
|
|
|
|
|
|
|
268
|
|
|
|
|
|
|
/* Track balanced parens (tokenizer already validated) */ |
|
269
|
1194
|
50
|
|
|
|
|
if (c == '(') paren_depth++; |
|
270
|
1194
|
50
|
|
|
|
|
else if (c == ')') paren_depth--; |
|
271
|
|
|
|
|
|
|
|
|
272
|
1194
|
|
|
|
|
|
decoded[out_len++] = c; |
|
273
|
|
|
|
|
|
|
} |
|
274
|
|
|
|
|
|
|
|
|
275
|
|
|
|
|
|
|
(void)paren_depth; /* Validated by tokenizer */ |
|
276
|
47
|
|
|
|
|
|
return pdfmake_str(p->doc->arena, (const char *)decoded, out_len); |
|
277
|
|
|
|
|
|
|
} |
|
278
|
|
|
|
|
|
|
|
|
279
|
|
|
|
|
|
|
/* Parse hex string - decode hex pairs */ |
|
280
|
420
|
|
|
|
|
|
static pdfmake_obj_t parse_hex_string(pdfmake_parser_t *p, pdfmake_tok_t *tok) { |
|
281
|
|
|
|
|
|
|
/* Content is between < and > - use tok.buf */ |
|
282
|
420
|
|
|
|
|
|
const uint8_t *src = p->tok.buf + tok->offset + 1; |
|
283
|
420
|
|
|
|
|
|
size_t src_len = tok->length - 2; |
|
284
|
|
|
|
|
|
|
uint8_t *decoded; |
|
285
|
|
|
|
|
|
|
size_t out_len; |
|
286
|
|
|
|
|
|
|
int high; |
|
287
|
|
|
|
|
|
|
size_t i; |
|
288
|
|
|
|
|
|
|
|
|
289
|
|
|
|
|
|
|
/* Decode hex pairs (skip whitespace) */ |
|
290
|
420
|
|
|
|
|
|
decoded = pdfmake_arena_alloc(p->doc->arena, (src_len + 1) / 2); |
|
291
|
420
|
50
|
|
|
|
|
if (!decoded) return pdfmake_null(); |
|
292
|
|
|
|
|
|
|
|
|
293
|
420
|
|
|
|
|
|
out_len = 0; |
|
294
|
420
|
|
|
|
|
|
high = -1; |
|
295
|
|
|
|
|
|
|
|
|
296
|
15001
|
100
|
|
|
|
|
for (i = 0; i < src_len; i++) { |
|
297
|
14581
|
|
|
|
|
|
uint8_t c = src[i]; |
|
298
|
|
|
|
|
|
|
int digit; |
|
299
|
14581
|
50
|
|
|
|
|
if (pdfmake_is_whitespace(c)) continue; |
|
300
|
|
|
|
|
|
|
|
|
301
|
14581
|
|
|
|
|
|
digit = hex_digit(c); |
|
302
|
14581
|
50
|
|
|
|
|
if (digit < 0) continue; /* Invalid hex digit - skip */ |
|
303
|
|
|
|
|
|
|
|
|
304
|
14581
|
100
|
|
|
|
|
if (high < 0) { |
|
305
|
7291
|
|
|
|
|
|
high = digit; |
|
306
|
|
|
|
|
|
|
} else { |
|
307
|
7290
|
|
|
|
|
|
decoded[out_len++] = (uint8_t)((high << 4) | digit); |
|
308
|
7290
|
|
|
|
|
|
high = -1; |
|
309
|
|
|
|
|
|
|
} |
|
310
|
|
|
|
|
|
|
} |
|
311
|
|
|
|
|
|
|
|
|
312
|
|
|
|
|
|
|
/* Odd number of digits - pad with 0 */ |
|
313
|
420
|
100
|
|
|
|
|
if (high >= 0) { |
|
314
|
1
|
|
|
|
|
|
decoded[out_len++] = (uint8_t)(high << 4); |
|
315
|
|
|
|
|
|
|
} |
|
316
|
|
|
|
|
|
|
|
|
317
|
420
|
|
|
|
|
|
return pdfmake_hexstr(p->doc->arena, decoded, out_len); |
|
318
|
|
|
|
|
|
|
} |
|
319
|
|
|
|
|
|
|
|
|
320
|
|
|
|
|
|
|
/* Parse array */ |
|
321
|
706
|
|
|
|
|
|
static pdfmake_obj_t parse_array(pdfmake_parser_t *p) { |
|
322
|
706
|
|
|
|
|
|
pdfmake_obj_t arr = pdfmake_array_new(p->doc->arena); |
|
323
|
|
|
|
|
|
|
pdfmake_tok_t tok; |
|
324
|
|
|
|
|
|
|
pdfmake_obj_t item; |
|
325
|
|
|
|
|
|
|
|
|
326
|
706
|
50
|
|
|
|
|
if (arr.kind != PDFMAKE_ARRAY) return pdfmake_null(); |
|
327
|
|
|
|
|
|
|
|
|
328
|
|
|
|
|
|
|
while (1) { |
|
329
|
6542
|
|
|
|
|
|
pdfmake_tok_peek_significant(&p->tok, &tok); |
|
330
|
|
|
|
|
|
|
|
|
331
|
6542
|
100
|
|
|
|
|
if (tok.kind == PDFMAKE_TOK_ARR_CLOSE) { |
|
332
|
706
|
|
|
|
|
|
pdfmake_tok_next_significant(&p->tok); /* Consume ] */ |
|
333
|
706
|
|
|
|
|
|
break; |
|
334
|
|
|
|
|
|
|
} |
|
335
|
5836
|
50
|
|
|
|
|
if (tok.kind == PDFMAKE_TOK_EOF) { |
|
336
|
0
|
|
|
|
|
|
set_error(p, PDFMAKE_EPARSE, tok.offset, "Unterminated array"); |
|
337
|
0
|
|
|
|
|
|
return pdfmake_null(); |
|
338
|
|
|
|
|
|
|
} |
|
339
|
|
|
|
|
|
|
|
|
340
|
5836
|
|
|
|
|
|
item = parse_object_internal(p); |
|
341
|
5836
|
50
|
|
|
|
|
if (p->last_err != PDFMAKE_OK) return pdfmake_null(); |
|
342
|
|
|
|
|
|
|
|
|
343
|
5836
|
50
|
|
|
|
|
if (!pdfmake_array_push(p->doc->arena, &arr, item)) { |
|
344
|
0
|
|
|
|
|
|
set_error(p, PDFMAKE_ENOMEM, tok.offset, "Array push failed"); |
|
345
|
0
|
|
|
|
|
|
return pdfmake_null(); |
|
346
|
|
|
|
|
|
|
} |
|
347
|
|
|
|
|
|
|
} |
|
348
|
|
|
|
|
|
|
|
|
349
|
706
|
|
|
|
|
|
return arr; |
|
350
|
|
|
|
|
|
|
} |
|
351
|
|
|
|
|
|
|
|
|
352
|
|
|
|
|
|
|
/* Parse dictionary (and possibly stream) */ |
|
353
|
1526
|
|
|
|
|
|
static pdfmake_obj_t parse_dict(pdfmake_parser_t *p) { |
|
354
|
1526
|
|
|
|
|
|
pdfmake_obj_t dict = pdfmake_dict_new(p->doc->arena); |
|
355
|
|
|
|
|
|
|
pdfmake_tok_t tok; |
|
356
|
|
|
|
|
|
|
pdfmake_obj_t key_obj; |
|
357
|
|
|
|
|
|
|
pdfmake_obj_t value; |
|
358
|
|
|
|
|
|
|
pdfmake_tok_t next; |
|
359
|
|
|
|
|
|
|
pdfmake_obj_t stream; |
|
360
|
|
|
|
|
|
|
pdfmake_obj_t stream_dict_wrapper; |
|
361
|
|
|
|
|
|
|
pdfmake_dict_iter_t iter; |
|
362
|
|
|
|
|
|
|
const uint8_t *body_data; |
|
363
|
|
|
|
|
|
|
size_t body_len; |
|
364
|
|
|
|
|
|
|
pdfmake_err_t err; |
|
365
|
|
|
|
|
|
|
size_t endstream_pos; |
|
366
|
|
|
|
|
|
|
const char *endstream_kw; |
|
367
|
|
|
|
|
|
|
|
|
368
|
1526
|
50
|
|
|
|
|
if (dict.kind != PDFMAKE_DICT) return pdfmake_null(); |
|
369
|
|
|
|
|
|
|
|
|
370
|
|
|
|
|
|
|
while (1) { |
|
371
|
6946
|
|
|
|
|
|
pdfmake_tok_peek_significant(&p->tok, &tok); |
|
372
|
|
|
|
|
|
|
|
|
373
|
6946
|
100
|
|
|
|
|
if (tok.kind == PDFMAKE_TOK_DICT_CLOSE) { |
|
374
|
1526
|
|
|
|
|
|
pdfmake_tok_next_significant(&p->tok); /* Consume >> */ |
|
375
|
1526
|
|
|
|
|
|
break; |
|
376
|
|
|
|
|
|
|
} |
|
377
|
5420
|
50
|
|
|
|
|
if (tok.kind == PDFMAKE_TOK_EOF) { |
|
378
|
0
|
|
|
|
|
|
set_error(p, PDFMAKE_EPARSE, tok.offset, "Unterminated dictionary"); |
|
379
|
0
|
|
|
|
|
|
return pdfmake_null(); |
|
380
|
|
|
|
|
|
|
} |
|
381
|
|
|
|
|
|
|
|
|
382
|
|
|
|
|
|
|
/* Key must be a name */ |
|
383
|
5420
|
50
|
|
|
|
|
if (tok.kind != PDFMAKE_TOK_NAME) { |
|
384
|
0
|
|
|
|
|
|
set_error(p, PDFMAKE_EPARSE, tok.offset, "Dictionary key must be a name"); |
|
385
|
0
|
|
|
|
|
|
return pdfmake_null(); |
|
386
|
|
|
|
|
|
|
} |
|
387
|
|
|
|
|
|
|
|
|
388
|
5420
|
|
|
|
|
|
pdfmake_tok_next_significant(&p->tok); /* Consume key */ |
|
389
|
5420
|
|
|
|
|
|
key_obj = parse_name(p, &tok); |
|
390
|
5420
|
50
|
|
|
|
|
if (key_obj.kind != PDFMAKE_NAME) return pdfmake_null(); |
|
391
|
|
|
|
|
|
|
|
|
392
|
|
|
|
|
|
|
/* Value */ |
|
393
|
5420
|
|
|
|
|
|
value = parse_object_internal(p); |
|
394
|
5420
|
50
|
|
|
|
|
if (p->last_err != PDFMAKE_OK) return pdfmake_null(); |
|
395
|
|
|
|
|
|
|
|
|
396
|
5420
|
50
|
|
|
|
|
if (!pdfmake_dict_set(p->doc->arena, &dict, key_obj.as.name.id, value)) { |
|
397
|
0
|
|
|
|
|
|
set_error(p, PDFMAKE_ENOMEM, tok.offset, "Dict set failed"); |
|
398
|
0
|
|
|
|
|
|
return pdfmake_null(); |
|
399
|
|
|
|
|
|
|
} |
|
400
|
|
|
|
|
|
|
} |
|
401
|
|
|
|
|
|
|
|
|
402
|
|
|
|
|
|
|
/* Check if followed by stream keyword */ |
|
403
|
1526
|
|
|
|
|
|
pdfmake_tok_peek_significant(&p->tok, &next); |
|
404
|
|
|
|
|
|
|
|
|
405
|
1526
|
100
|
|
|
|
|
if (next.kind == PDFMAKE_TOK_KW_STREAM) { |
|
406
|
217
|
|
|
|
|
|
pdfmake_tok_next_significant(&p->tok); /* Consume stream */ |
|
407
|
|
|
|
|
|
|
|
|
408
|
|
|
|
|
|
|
/* Create stream object */ |
|
409
|
217
|
|
|
|
|
|
stream = pdfmake_stream_new(p->doc->arena); |
|
410
|
217
|
50
|
|
|
|
|
if (stream.kind != PDFMAKE_STREAM) return pdfmake_null(); |
|
411
|
|
|
|
|
|
|
|
|
412
|
|
|
|
|
|
|
/* Copy dict entries to stream dict - need a wrapper pdfmake_obj_t */ |
|
413
|
217
|
|
|
|
|
|
stream_dict_wrapper.kind = PDFMAKE_DICT; |
|
414
|
217
|
|
|
|
|
|
stream_dict_wrapper.as.dict = stream.as.stream->dict; |
|
415
|
|
|
|
|
|
|
|
|
416
|
217
|
|
|
|
|
|
pdfmake_dict_iter_init(&iter, &dict); |
|
417
|
813
|
100
|
|
|
|
|
while (pdfmake_dict_iter_next(&iter)) { |
|
418
|
596
|
|
|
|
|
|
pdfmake_dict_set(p->doc->arena, &stream_dict_wrapper, |
|
419
|
596
|
|
|
|
|
|
iter.current_key, *iter.current_value); |
|
420
|
|
|
|
|
|
|
} |
|
421
|
|
|
|
|
|
|
|
|
422
|
|
|
|
|
|
|
/* Extract stream body */ |
|
423
|
217
|
|
|
|
|
|
err = pdfmake_extract_stream_body(p, &dict, next.payload.body_offset, |
|
424
|
|
|
|
|
|
|
&body_data, &body_len); |
|
425
|
217
|
50
|
|
|
|
|
if (err != PDFMAKE_OK) return pdfmake_null(); |
|
426
|
|
|
|
|
|
|
|
|
427
|
|
|
|
|
|
|
/* Copy stream data to arena */ |
|
428
|
217
|
|
|
|
|
|
pdfmake_stream_set_data(p->doc->arena, &stream, body_data, body_len); |
|
429
|
|
|
|
|
|
|
|
|
430
|
|
|
|
|
|
|
/* Skip tokenizer past stream body to find endstream. |
|
431
|
|
|
|
|
|
|
* The tokenizer can't parse arbitrary binary stream content, |
|
432
|
|
|
|
|
|
|
* so we manually position it after the stream body. */ |
|
433
|
217
|
|
|
|
|
|
endstream_pos = next.payload.body_offset + body_len; |
|
434
|
|
|
|
|
|
|
|
|
435
|
|
|
|
|
|
|
/* Skip optional EOL before endstream */ |
|
436
|
431
|
50
|
|
|
|
|
while (endstream_pos < p->buf_len && |
|
437
|
431
|
100
|
|
|
|
|
(p->buf[endstream_pos] == '\r' || p->buf[endstream_pos] == '\n')) { |
|
|
|
100
|
|
|
|
|
|
|
438
|
214
|
|
|
|
|
|
endstream_pos++; |
|
439
|
|
|
|
|
|
|
} |
|
440
|
|
|
|
|
|
|
|
|
441
|
|
|
|
|
|
|
/* Verify endstream keyword */ |
|
442
|
217
|
|
|
|
|
|
endstream_kw = "endstream"; |
|
443
|
217
|
50
|
|
|
|
|
if (endstream_pos + 9 > p->buf_len || |
|
444
|
217
|
50
|
|
|
|
|
memcmp(p->buf + endstream_pos, endstream_kw, 9) != 0) { |
|
445
|
0
|
|
|
|
|
|
set_error(p, PDFMAKE_ESTREAM, endstream_pos, "Expected endstream"); |
|
446
|
0
|
|
|
|
|
|
return pdfmake_null(); |
|
447
|
|
|
|
|
|
|
} |
|
448
|
|
|
|
|
|
|
|
|
449
|
|
|
|
|
|
|
/* Position tokenizer after endstream */ |
|
450
|
217
|
|
|
|
|
|
p->tok.pos = endstream_pos + 9; |
|
451
|
217
|
|
|
|
|
|
p->tok.has_peeked = 0; |
|
452
|
|
|
|
|
|
|
|
|
453
|
217
|
|
|
|
|
|
return stream; |
|
454
|
|
|
|
|
|
|
} |
|
455
|
|
|
|
|
|
|
|
|
456
|
1309
|
|
|
|
|
|
return dict; |
|
457
|
|
|
|
|
|
|
} |
|
458
|
|
|
|
|
|
|
|
|
459
|
|
|
|
|
|
|
/* Parse any object at current position */ |
|
460
|
12411
|
|
|
|
|
|
static pdfmake_obj_t parse_object_internal(pdfmake_parser_t *p) { |
|
461
|
12411
|
|
|
|
|
|
pdfmake_tok_t tok = pdfmake_tok_next_significant(&p->tok); |
|
462
|
|
|
|
|
|
|
|
|
463
|
12411
|
|
|
|
|
|
switch (tok.kind) { |
|
464
|
2
|
|
|
|
|
|
case PDFMAKE_TOK_KW_NULL: |
|
465
|
2
|
|
|
|
|
|
return pdfmake_null(); |
|
466
|
|
|
|
|
|
|
|
|
467
|
3
|
|
|
|
|
|
case PDFMAKE_TOK_KW_TRUE: |
|
468
|
3
|
|
|
|
|
|
return pdfmake_bool(1); |
|
469
|
|
|
|
|
|
|
|
|
470
|
7
|
|
|
|
|
|
case PDFMAKE_TOK_KW_FALSE: |
|
471
|
7
|
|
|
|
|
|
return pdfmake_bool(0); |
|
472
|
|
|
|
|
|
|
|
|
473
|
8108
|
|
|
|
|
|
case PDFMAKE_TOK_INT: { |
|
474
|
|
|
|
|
|
|
/* Check for indirect reference: INT INT R */ |
|
475
|
|
|
|
|
|
|
pdfmake_tok_t peek1, peek2; |
|
476
|
8108
|
|
|
|
|
|
pdfmake_tok_peek_significant(&p->tok, &peek1); |
|
477
|
|
|
|
|
|
|
|
|
478
|
8108
|
100
|
|
|
|
|
if (peek1.kind == PDFMAKE_TOK_INT) { |
|
479
|
|
|
|
|
|
|
/* Save tokenizer state */ |
|
480
|
6561
|
|
|
|
|
|
size_t saved_pos = p->tok.pos; |
|
481
|
6561
|
|
|
|
|
|
int saved_peeked = p->tok.has_peeked; |
|
482
|
6561
|
|
|
|
|
|
pdfmake_tok_t saved_peek = p->tok.peeked; |
|
483
|
|
|
|
|
|
|
|
|
484
|
6561
|
|
|
|
|
|
pdfmake_tok_next_significant(&p->tok); /* Consume second int */ |
|
485
|
6561
|
|
|
|
|
|
pdfmake_tok_peek_significant(&p->tok, &peek2); |
|
486
|
|
|
|
|
|
|
|
|
487
|
6561
|
100
|
|
|
|
|
if (peek2.kind == PDFMAKE_TOK_KW_R) { |
|
488
|
1893
|
|
|
|
|
|
pdfmake_tok_next_significant(&p->tok); /* Consume R */ |
|
489
|
1893
|
|
|
|
|
|
return pdfmake_ref((uint32_t)tok.payload.int_val, |
|
490
|
1893
|
|
|
|
|
|
(uint16_t)peek1.payload.int_val); |
|
491
|
|
|
|
|
|
|
} |
|
492
|
|
|
|
|
|
|
|
|
493
|
|
|
|
|
|
|
/* Not a reference - restore state */ |
|
494
|
4668
|
|
|
|
|
|
p->tok.pos = saved_pos; |
|
495
|
4668
|
|
|
|
|
|
p->tok.has_peeked = saved_peeked; |
|
496
|
4668
|
|
|
|
|
|
p->tok.peeked = saved_peek; |
|
497
|
|
|
|
|
|
|
} |
|
498
|
6215
|
|
|
|
|
|
return pdfmake_int(tok.payload.int_val); |
|
499
|
|
|
|
|
|
|
} |
|
500
|
|
|
|
|
|
|
|
|
501
|
9
|
|
|
|
|
|
case PDFMAKE_TOK_REAL: |
|
502
|
9
|
|
|
|
|
|
return pdfmake_real(tok.payload.real_val); |
|
503
|
|
|
|
|
|
|
|
|
504
|
1583
|
|
|
|
|
|
case PDFMAKE_TOK_NAME: |
|
505
|
1583
|
|
|
|
|
|
return parse_name(p, &tok); |
|
506
|
|
|
|
|
|
|
|
|
507
|
47
|
|
|
|
|
|
case PDFMAKE_TOK_LSTR: |
|
508
|
47
|
|
|
|
|
|
return parse_literal_string(p, &tok); |
|
509
|
|
|
|
|
|
|
|
|
510
|
420
|
|
|
|
|
|
case PDFMAKE_TOK_HSTR: |
|
511
|
420
|
|
|
|
|
|
return parse_hex_string(p, &tok); |
|
512
|
|
|
|
|
|
|
|
|
513
|
706
|
|
|
|
|
|
case PDFMAKE_TOK_ARR_OPEN: |
|
514
|
706
|
|
|
|
|
|
return parse_array(p); |
|
515
|
|
|
|
|
|
|
|
|
516
|
1526
|
|
|
|
|
|
case PDFMAKE_TOK_DICT_OPEN: |
|
517
|
1526
|
|
|
|
|
|
return parse_dict(p); |
|
518
|
|
|
|
|
|
|
|
|
519
|
0
|
|
|
|
|
|
case PDFMAKE_TOK_EOF: |
|
520
|
0
|
|
|
|
|
|
set_error(p, PDFMAKE_EPARSE, tok.offset, "Unexpected end of input"); |
|
521
|
0
|
|
|
|
|
|
return pdfmake_null(); |
|
522
|
|
|
|
|
|
|
|
|
523
|
0
|
|
|
|
|
|
default: |
|
524
|
0
|
|
|
|
|
|
set_error(p, PDFMAKE_EPARSE, tok.offset, "Unexpected token"); |
|
525
|
0
|
|
|
|
|
|
return pdfmake_null(); |
|
526
|
|
|
|
|
|
|
} |
|
527
|
|
|
|
|
|
|
} |
|
528
|
|
|
|
|
|
|
|
|
529
|
0
|
|
|
|
|
|
pdfmake_obj_t pdfmake_parse_object(pdfmake_parser_t *parser) { |
|
530
|
0
|
|
|
|
|
|
parser->last_err = PDFMAKE_OK; |
|
531
|
0
|
|
|
|
|
|
return parse_object_internal(parser); |
|
532
|
|
|
|
|
|
|
} |
|
533
|
|
|
|
|
|
|
|
|
534
|
|
|
|
|
|
|
/*============================================================================ |
|
535
|
|
|
|
|
|
|
* Stream body extraction |
|
536
|
|
|
|
|
|
|
*==========================================================================*/ |
|
537
|
|
|
|
|
|
|
|
|
538
|
217
|
|
|
|
|
|
pdfmake_err_t pdfmake_extract_stream_body(pdfmake_parser_t *p, |
|
539
|
|
|
|
|
|
|
pdfmake_obj_t *stream_dict, |
|
540
|
|
|
|
|
|
|
size_t body_offset, |
|
541
|
|
|
|
|
|
|
const uint8_t **out_data, |
|
542
|
|
|
|
|
|
|
size_t *out_len) { |
|
543
|
|
|
|
|
|
|
/* Get /Length from dict */ |
|
544
|
217
|
|
|
|
|
|
uint32_t length_id = intern_name(p, "Length", 6); |
|
545
|
217
|
|
|
|
|
|
pdfmake_obj_t *length_obj = pdfmake_dict_get(stream_dict, length_id); |
|
546
|
217
|
|
|
|
|
|
size_t declared_len = 0; |
|
547
|
217
|
|
|
|
|
|
int have_length = 0; |
|
548
|
|
|
|
|
|
|
const char *endstream_marker; |
|
549
|
|
|
|
|
|
|
size_t endstream_len; |
|
550
|
|
|
|
|
|
|
size_t scan_start; |
|
551
|
|
|
|
|
|
|
size_t scan_end; |
|
552
|
|
|
|
|
|
|
size_t i; |
|
553
|
|
|
|
|
|
|
|
|
554
|
217
|
50
|
|
|
|
|
if (length_obj) { |
|
555
|
217
|
50
|
|
|
|
|
if (length_obj->kind == PDFMAKE_INT) { |
|
556
|
217
|
|
|
|
|
|
declared_len = (size_t)length_obj->as.i; |
|
557
|
217
|
|
|
|
|
|
have_length = 1; |
|
558
|
0
|
0
|
|
|
|
|
} else if (length_obj->kind == PDFMAKE_REF) { |
|
559
|
|
|
|
|
|
|
/* /Length is indirect - need to resolve */ |
|
560
|
0
|
|
|
|
|
|
pdfmake_obj_t *resolved = pdfmake_parser_resolve(p, length_obj->as.ref); |
|
561
|
0
|
0
|
|
|
|
|
if (resolved && resolved->kind == PDFMAKE_INT) { |
|
|
|
0
|
|
|
|
|
|
|
562
|
0
|
|
|
|
|
|
declared_len = (size_t)resolved->as.i; |
|
563
|
0
|
|
|
|
|
|
have_length = 1; |
|
564
|
|
|
|
|
|
|
} |
|
565
|
|
|
|
|
|
|
} |
|
566
|
|
|
|
|
|
|
} |
|
567
|
|
|
|
|
|
|
|
|
568
|
|
|
|
|
|
|
/* Find endstream */ |
|
569
|
217
|
|
|
|
|
|
endstream_marker = "endstream"; |
|
570
|
217
|
|
|
|
|
|
endstream_len = 9; |
|
571
|
217
|
|
|
|
|
|
scan_start = body_offset; |
|
572
|
217
|
|
|
|
|
|
scan_end = p->buf_len; |
|
573
|
|
|
|
|
|
|
|
|
574
|
|
|
|
|
|
|
/* If we have declared length, try it first */ |
|
575
|
217
|
50
|
|
|
|
|
if (have_length && body_offset + declared_len + endstream_len <= p->buf_len) { |
|
|
|
50
|
|
|
|
|
|
|
576
|
217
|
|
|
|
|
|
size_t expected_end = body_offset + declared_len; |
|
577
|
|
|
|
|
|
|
|
|
578
|
|
|
|
|
|
|
/* Skip optional EOL before endstream */ |
|
579
|
429
|
50
|
|
|
|
|
while (expected_end < p->buf_len && |
|
580
|
429
|
100
|
|
|
|
|
(p->buf[expected_end] == '\r' || p->buf[expected_end] == '\n')) { |
|
|
|
100
|
|
|
|
|
|
|
581
|
212
|
|
|
|
|
|
expected_end++; |
|
582
|
|
|
|
|
|
|
} |
|
583
|
|
|
|
|
|
|
|
|
584
|
217
|
50
|
|
|
|
|
if (expected_end + endstream_len <= p->buf_len && |
|
585
|
217
|
100
|
|
|
|
|
memcmp(p->buf + expected_end, endstream_marker, endstream_len) == 0) { |
|
586
|
215
|
|
|
|
|
|
*out_data = p->buf + body_offset; |
|
587
|
215
|
|
|
|
|
|
*out_len = declared_len; |
|
588
|
215
|
|
|
|
|
|
return PDFMAKE_OK; |
|
589
|
|
|
|
|
|
|
} |
|
590
|
|
|
|
|
|
|
} |
|
591
|
|
|
|
|
|
|
|
|
592
|
|
|
|
|
|
|
/* Fallback: scan for endstream */ |
|
593
|
76
|
50
|
|
|
|
|
for (i = scan_start; i + endstream_len <= scan_end; i++) { |
|
594
|
76
|
100
|
|
|
|
|
if (memcmp(p->buf + i, endstream_marker, endstream_len) == 0) { |
|
595
|
|
|
|
|
|
|
/* Check preceding bytes for proper delimiter */ |
|
596
|
2
|
|
|
|
|
|
size_t body_end = i; |
|
597
|
|
|
|
|
|
|
|
|
598
|
|
|
|
|
|
|
/* Back up over optional EOL */ |
|
599
|
2
|
50
|
|
|
|
|
if (body_end > body_offset && p->buf[body_end - 1] == '\n') { |
|
|
|
50
|
|
|
|
|
|
|
600
|
2
|
|
|
|
|
|
body_end--; |
|
601
|
|
|
|
|
|
|
} |
|
602
|
2
|
50
|
|
|
|
|
if (body_end > body_offset && p->buf[body_end - 1] == '\r') { |
|
|
|
50
|
|
|
|
|
|
|
603
|
0
|
|
|
|
|
|
body_end--; |
|
604
|
|
|
|
|
|
|
} |
|
605
|
|
|
|
|
|
|
|
|
606
|
2
|
|
|
|
|
|
*out_data = p->buf + body_offset; |
|
607
|
2
|
|
|
|
|
|
*out_len = body_end - body_offset; |
|
608
|
2
|
|
|
|
|
|
return PDFMAKE_OK; |
|
609
|
|
|
|
|
|
|
} |
|
610
|
|
|
|
|
|
|
} |
|
611
|
|
|
|
|
|
|
|
|
612
|
0
|
|
|
|
|
|
set_error(p, PDFMAKE_ESTREAM, body_offset, "endstream not found"); |
|
613
|
0
|
|
|
|
|
|
return PDFMAKE_ESTREAM; |
|
614
|
|
|
|
|
|
|
} |
|
615
|
|
|
|
|
|
|
|
|
616
|
|
|
|
|
|
|
/*============================================================================ |
|
617
|
|
|
|
|
|
|
* Indirect object parsing |
|
618
|
|
|
|
|
|
|
*==========================================================================*/ |
|
619
|
|
|
|
|
|
|
|
|
620
|
911
|
|
|
|
|
|
pdfmake_err_t pdfmake_parse_indirect_object(pdfmake_parser_t *p) { |
|
621
|
|
|
|
|
|
|
pdfmake_tok_t num_tok; |
|
622
|
|
|
|
|
|
|
pdfmake_tok_t gen_tok; |
|
623
|
|
|
|
|
|
|
pdfmake_tok_t obj_tok; |
|
624
|
|
|
|
|
|
|
uint32_t num; |
|
625
|
|
|
|
|
|
|
uint16_t gen; |
|
626
|
|
|
|
|
|
|
pdfmake_obj_t obj; |
|
627
|
|
|
|
|
|
|
pdfmake_tok_t endobj_tok; |
|
628
|
|
|
|
|
|
|
size_t idx; |
|
629
|
|
|
|
|
|
|
|
|
630
|
|
|
|
|
|
|
/* Clear any residual error from previous operations */ |
|
631
|
911
|
|
|
|
|
|
p->last_err = PDFMAKE_OK; |
|
632
|
|
|
|
|
|
|
|
|
633
|
911
|
|
|
|
|
|
num_tok = pdfmake_tok_next_significant(&p->tok); |
|
634
|
911
|
50
|
|
|
|
|
if (num_tok.kind != PDFMAKE_TOK_INT) { |
|
635
|
0
|
|
|
|
|
|
set_error(p, PDFMAKE_EPARSE, num_tok.offset, "Expected object number"); |
|
636
|
0
|
|
|
|
|
|
return PDFMAKE_EPARSE; |
|
637
|
|
|
|
|
|
|
} |
|
638
|
|
|
|
|
|
|
|
|
639
|
911
|
|
|
|
|
|
gen_tok = pdfmake_tok_next_significant(&p->tok); |
|
640
|
911
|
50
|
|
|
|
|
if (gen_tok.kind != PDFMAKE_TOK_INT) { |
|
641
|
0
|
|
|
|
|
|
set_error(p, PDFMAKE_EPARSE, gen_tok.offset, "Expected generation number"); |
|
642
|
0
|
|
|
|
|
|
return PDFMAKE_EPARSE; |
|
643
|
|
|
|
|
|
|
} |
|
644
|
|
|
|
|
|
|
|
|
645
|
911
|
|
|
|
|
|
obj_tok = pdfmake_tok_next_significant(&p->tok); |
|
646
|
911
|
50
|
|
|
|
|
if (obj_tok.kind != PDFMAKE_TOK_KW_OBJ) { |
|
647
|
0
|
|
|
|
|
|
set_error(p, PDFMAKE_EPARSE, obj_tok.offset, "Expected 'obj' keyword"); |
|
648
|
0
|
|
|
|
|
|
return PDFMAKE_EPARSE; |
|
649
|
|
|
|
|
|
|
} |
|
650
|
|
|
|
|
|
|
|
|
651
|
911
|
|
|
|
|
|
num = (uint32_t)num_tok.payload.int_val; |
|
652
|
911
|
|
|
|
|
|
gen = (uint16_t)gen_tok.payload.int_val; |
|
653
|
|
|
|
|
|
|
|
|
654
|
|
|
|
|
|
|
/* Parse the object value */ |
|
655
|
911
|
|
|
|
|
|
obj = parse_object_internal(p); |
|
656
|
911
|
50
|
|
|
|
|
if (p->last_err != PDFMAKE_OK) return p->last_err; |
|
657
|
|
|
|
|
|
|
|
|
658
|
|
|
|
|
|
|
/* Expect endobj */ |
|
659
|
911
|
|
|
|
|
|
endobj_tok = pdfmake_tok_next_significant(&p->tok); |
|
660
|
911
|
50
|
|
|
|
|
if (endobj_tok.kind != PDFMAKE_TOK_KW_ENDOBJ) { |
|
661
|
|
|
|
|
|
|
/* Some PDFs omit endobj before stream - tolerate */ |
|
662
|
0
|
0
|
|
|
|
|
if (endobj_tok.kind != PDFMAKE_TOK_KW_STREAM) { |
|
663
|
0
|
|
|
|
|
|
set_error(p, PDFMAKE_EPARSE, endobj_tok.offset, "Expected 'endobj' keyword"); |
|
664
|
0
|
|
|
|
|
|
return PDFMAKE_EPARSE; |
|
665
|
|
|
|
|
|
|
} |
|
666
|
|
|
|
|
|
|
} |
|
667
|
|
|
|
|
|
|
|
|
668
|
|
|
|
|
|
|
/* Store in document */ |
|
669
|
|
|
|
|
|
|
/* Grow objects array if needed - use num directly as we need capacity for num items */ |
|
670
|
|
|
|
|
|
|
/* pdfmake_doc uses 1-based numbering with object N at index N-1 */ |
|
671
|
911
|
50
|
|
|
|
|
while (num > p->doc->obj_cap) { |
|
672
|
0
|
0
|
|
|
|
|
size_t new_cap = p->doc->obj_cap ? p->doc->obj_cap * 2 : PDFMAKE_DOC_INIT_CAP; |
|
673
|
|
|
|
|
|
|
pdfmake_indirect_t *new_objs; |
|
674
|
0
|
0
|
|
|
|
|
while (new_cap < num) new_cap *= 2; |
|
675
|
0
|
|
|
|
|
|
new_objs = realloc(p->doc->objects, |
|
676
|
|
|
|
|
|
|
new_cap * sizeof(pdfmake_indirect_t)); |
|
677
|
0
|
0
|
|
|
|
|
if (!new_objs) { |
|
678
|
0
|
|
|
|
|
|
set_error(p, PDFMAKE_ENOMEM, 0, "Failed to grow object table"); |
|
679
|
0
|
|
|
|
|
|
return PDFMAKE_ENOMEM; |
|
680
|
|
|
|
|
|
|
} |
|
681
|
0
|
|
|
|
|
|
memset(new_objs + p->doc->obj_cap, 0, |
|
682
|
0
|
|
|
|
|
|
(new_cap - p->doc->obj_cap) * sizeof(pdfmake_indirect_t)); |
|
683
|
0
|
|
|
|
|
|
p->doc->objects = new_objs; |
|
684
|
0
|
|
|
|
|
|
p->doc->obj_cap = new_cap; |
|
685
|
|
|
|
|
|
|
} |
|
686
|
|
|
|
|
|
|
|
|
687
|
|
|
|
|
|
|
/* Store at index num-1 (1-based numbering) */ |
|
688
|
911
|
|
|
|
|
|
idx = num - 1; |
|
689
|
911
|
|
|
|
|
|
p->doc->objects[idx].num = num; |
|
690
|
911
|
|
|
|
|
|
p->doc->objects[idx].gen = gen; |
|
691
|
911
|
|
|
|
|
|
p->doc->objects[idx].obj = obj; |
|
692
|
911
|
|
|
|
|
|
p->doc->objects[idx].in_use = 1; |
|
693
|
911
|
|
|
|
|
|
p->doc->objects[idx].byte_offset = num_tok.offset; |
|
694
|
|
|
|
|
|
|
|
|
695
|
911
|
100
|
|
|
|
|
if (num > p->doc->obj_count) { |
|
696
|
244
|
|
|
|
|
|
p->doc->obj_count = num; |
|
697
|
|
|
|
|
|
|
} |
|
698
|
|
|
|
|
|
|
|
|
699
|
|
|
|
|
|
|
/* Update xref if tracking */ |
|
700
|
911
|
50
|
|
|
|
|
if (ensure_xref_cap(p, num)) { |
|
701
|
911
|
|
|
|
|
|
p->xref[num].num = num; |
|
702
|
911
|
|
|
|
|
|
p->xref[num].gen = gen; |
|
703
|
911
|
|
|
|
|
|
p->xref[num].type = PDFMAKE_XREF_UNCOMPRESSED; |
|
704
|
911
|
|
|
|
|
|
p->xref[num].loc.offset = num_tok.offset; |
|
705
|
911
|
|
|
|
|
|
p->xref[num].loaded = 1; |
|
706
|
911
|
50
|
|
|
|
|
if (num >= p->xref_size) p->xref_size = num + 1; |
|
707
|
|
|
|
|
|
|
} |
|
708
|
|
|
|
|
|
|
|
|
709
|
911
|
|
|
|
|
|
return PDFMAKE_OK; |
|
710
|
|
|
|
|
|
|
} |
|
711
|
|
|
|
|
|
|
|
|
712
|
|
|
|
|
|
|
/*============================================================================ |
|
713
|
|
|
|
|
|
|
* startxref locator |
|
714
|
|
|
|
|
|
|
*==========================================================================*/ |
|
715
|
|
|
|
|
|
|
|
|
716
|
218
|
|
|
|
|
|
pdfmake_err_t pdfmake_locate_startxref(pdfmake_parser_t *p, uint64_t *offset) { |
|
717
|
|
|
|
|
|
|
/* Scan backward from EOF - check last 1024 bytes per spec */ |
|
718
|
218
|
|
|
|
|
|
size_t search_start = p->buf_len > 1024 ? p->buf_len - 1024 : 0; |
|
719
|
218
|
|
|
|
|
|
const char *keyword = "startxref"; |
|
720
|
218
|
|
|
|
|
|
size_t keyword_len = 9; |
|
721
|
218
|
|
|
|
|
|
const uint8_t *found = NULL; |
|
722
|
|
|
|
|
|
|
const uint8_t *ptr; |
|
723
|
|
|
|
|
|
|
const uint8_t *end; |
|
724
|
|
|
|
|
|
|
uint64_t off; |
|
725
|
|
|
|
|
|
|
size_t i; |
|
726
|
|
|
|
|
|
|
|
|
727
|
|
|
|
|
|
|
/* Find last occurrence of startxref */ |
|
728
|
4432
|
100
|
|
|
|
|
for (i = p->buf_len; i > search_start + keyword_len; ) { |
|
729
|
4426
|
|
|
|
|
|
i--; |
|
730
|
4426
|
100
|
|
|
|
|
if (memcmp(p->buf + i, keyword, keyword_len) == 0) { |
|
731
|
212
|
|
|
|
|
|
found = p->buf + i; |
|
732
|
212
|
|
|
|
|
|
break; |
|
733
|
|
|
|
|
|
|
} |
|
734
|
|
|
|
|
|
|
} |
|
735
|
|
|
|
|
|
|
|
|
736
|
218
|
100
|
|
|
|
|
if (!found) { |
|
737
|
6
|
|
|
|
|
|
set_error(p, PDFMAKE_EXREF, 0, "startxref not found"); |
|
738
|
6
|
|
|
|
|
|
return PDFMAKE_EXREF; |
|
739
|
|
|
|
|
|
|
} |
|
740
|
|
|
|
|
|
|
|
|
741
|
|
|
|
|
|
|
/* Parse offset after startxref */ |
|
742
|
212
|
|
|
|
|
|
ptr = found + keyword_len; |
|
743
|
212
|
|
|
|
|
|
end = p->buf + p->buf_len; |
|
744
|
|
|
|
|
|
|
|
|
745
|
|
|
|
|
|
|
/* Skip whitespace */ |
|
746
|
424
|
50
|
|
|
|
|
while (ptr < end && (*ptr == ' ' || *ptr == '\t' || *ptr == '\r' || *ptr == '\n')) { |
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
100
|
|
|
|
|
|
|
|
|
100
|
|
|
|
|
|
|
747
|
212
|
|
|
|
|
|
ptr++; |
|
748
|
|
|
|
|
|
|
} |
|
749
|
|
|
|
|
|
|
|
|
750
|
|
|
|
|
|
|
/* Parse number */ |
|
751
|
212
|
|
|
|
|
|
off = 0; |
|
752
|
914
|
50
|
|
|
|
|
while (ptr < end && isdigit(*ptr)) { |
|
|
|
100
|
|
|
|
|
|
|
753
|
702
|
|
|
|
|
|
off = off * 10 + (*ptr - '0'); |
|
754
|
702
|
|
|
|
|
|
ptr++; |
|
755
|
|
|
|
|
|
|
} |
|
756
|
|
|
|
|
|
|
|
|
757
|
212
|
|
|
|
|
|
*offset = off; |
|
758
|
212
|
|
|
|
|
|
return PDFMAKE_OK; |
|
759
|
|
|
|
|
|
|
} |
|
760
|
|
|
|
|
|
|
|
|
761
|
|
|
|
|
|
|
/*============================================================================ |
|
762
|
|
|
|
|
|
|
* Classic xref table parser (§7.5.4) |
|
763
|
|
|
|
|
|
|
*==========================================================================*/ |
|
764
|
|
|
|
|
|
|
|
|
765
|
|
|
|
|
|
|
/* Consume a "first_num count" pair that introduces a classic-xref |
|
766
|
|
|
|
|
|
|
* subsection. Caller must have already confirmed via peek that the next |
|
767
|
|
|
|
|
|
|
* token is not KW_TRAILER. */ |
|
768
|
|
|
|
|
|
|
static pdfmake_err_t |
|
769
|
206
|
|
|
|
|
|
parse_xref_subsection_header(pdfmake_parser_t *p, uint64_t table_offset, |
|
770
|
|
|
|
|
|
|
uint32_t *out_first, uint32_t *out_count) |
|
771
|
|
|
|
|
|
|
{ |
|
772
|
|
|
|
|
|
|
pdfmake_tok_t first_tok; |
|
773
|
|
|
|
|
|
|
pdfmake_tok_t count_tok; |
|
774
|
|
|
|
|
|
|
|
|
775
|
206
|
|
|
|
|
|
first_tok = pdfmake_tok_next_significant(&p->tok); |
|
776
|
206
|
50
|
|
|
|
|
if (first_tok.kind != PDFMAKE_TOK_INT) { |
|
777
|
0
|
|
|
|
|
|
set_error(p, PDFMAKE_EXREF, table_offset + first_tok.offset, |
|
778
|
|
|
|
|
|
|
"Expected subsection start object number"); |
|
779
|
0
|
|
|
|
|
|
return PDFMAKE_EXREF; |
|
780
|
|
|
|
|
|
|
} |
|
781
|
|
|
|
|
|
|
|
|
782
|
206
|
|
|
|
|
|
count_tok = pdfmake_tok_next_significant(&p->tok); |
|
783
|
206
|
50
|
|
|
|
|
if (count_tok.kind != PDFMAKE_TOK_INT) { |
|
784
|
0
|
|
|
|
|
|
set_error(p, PDFMAKE_EXREF, table_offset + count_tok.offset, |
|
785
|
|
|
|
|
|
|
"Expected subsection entry count"); |
|
786
|
0
|
|
|
|
|
|
return PDFMAKE_EXREF; |
|
787
|
|
|
|
|
|
|
} |
|
788
|
|
|
|
|
|
|
|
|
789
|
206
|
|
|
|
|
|
*out_first = (uint32_t)first_tok.payload.int_val; |
|
790
|
206
|
|
|
|
|
|
*out_count = (uint32_t)count_tok.payload.int_val; |
|
791
|
206
|
|
|
|
|
|
return PDFMAKE_OK; |
|
792
|
|
|
|
|
|
|
} |
|
793
|
|
|
|
|
|
|
|
|
794
|
|
|
|
|
|
|
/* Parse one classic-xref entry (offset, gen, n|f) from the token stream |
|
795
|
|
|
|
|
|
|
* and store it into p->xref[obj_num] unless the slot is already loaded |
|
796
|
|
|
|
|
|
|
* from a later xref section. The type byte is a single-character ERROR |
|
797
|
|
|
|
|
|
|
* token rather than a keyword — the tokenizer does not recognise 'n'/'f' |
|
798
|
|
|
|
|
|
|
* specially. */ |
|
799
|
|
|
|
|
|
|
static pdfmake_err_t |
|
800
|
3046
|
|
|
|
|
|
parse_xref_entry(pdfmake_parser_t *p, uint64_t table_offset, uint32_t obj_num) |
|
801
|
|
|
|
|
|
|
{ |
|
802
|
|
|
|
|
|
|
pdfmake_tok_t off_tok; |
|
803
|
|
|
|
|
|
|
pdfmake_tok_t gen_tok; |
|
804
|
|
|
|
|
|
|
pdfmake_tok_t type_tok; |
|
805
|
|
|
|
|
|
|
char type_char; |
|
806
|
|
|
|
|
|
|
|
|
807
|
3046
|
|
|
|
|
|
off_tok = pdfmake_tok_next_significant(&p->tok); |
|
808
|
3046
|
50
|
|
|
|
|
if (off_tok.kind != PDFMAKE_TOK_INT) { |
|
809
|
0
|
|
|
|
|
|
set_error(p, PDFMAKE_EXREF, table_offset + off_tok.offset, |
|
810
|
|
|
|
|
|
|
"Expected xref entry offset"); |
|
811
|
0
|
|
|
|
|
|
return PDFMAKE_EXREF; |
|
812
|
|
|
|
|
|
|
} |
|
813
|
|
|
|
|
|
|
|
|
814
|
3046
|
|
|
|
|
|
gen_tok = pdfmake_tok_next_significant(&p->tok); |
|
815
|
3046
|
50
|
|
|
|
|
if (gen_tok.kind != PDFMAKE_TOK_INT) { |
|
816
|
0
|
|
|
|
|
|
set_error(p, PDFMAKE_EXREF, table_offset + gen_tok.offset, |
|
817
|
|
|
|
|
|
|
"Expected xref entry generation"); |
|
818
|
0
|
|
|
|
|
|
return PDFMAKE_EXREF; |
|
819
|
|
|
|
|
|
|
} |
|
820
|
|
|
|
|
|
|
|
|
821
|
3046
|
|
|
|
|
|
type_tok = pdfmake_tok_next(&p->tok); |
|
822
|
6092
|
100
|
|
|
|
|
while (type_tok.kind == PDFMAKE_TOK_WS || type_tok.kind == PDFMAKE_TOK_COMMENT) { |
|
|
|
50
|
|
|
|
|
|
|
823
|
3046
|
|
|
|
|
|
type_tok = pdfmake_tok_next(&p->tok); |
|
824
|
|
|
|
|
|
|
} |
|
825
|
|
|
|
|
|
|
|
|
826
|
3046
|
|
|
|
|
|
type_char = 'n'; |
|
827
|
3046
|
50
|
|
|
|
|
if (type_tok.kind == PDFMAKE_TOK_ERROR && type_tok.length == 1) { |
|
|
|
50
|
|
|
|
|
|
|
828
|
3046
|
|
|
|
|
|
type_char = (char)p->buf[table_offset + type_tok.offset]; |
|
829
|
|
|
|
|
|
|
} |
|
830
|
|
|
|
|
|
|
|
|
831
|
3046
|
50
|
|
|
|
|
if (!p->xref[obj_num].loaded) { |
|
832
|
3046
|
|
|
|
|
|
p->xref[obj_num].num = obj_num; |
|
833
|
3046
|
|
|
|
|
|
p->xref[obj_num].gen = (uint16_t)gen_tok.payload.int_val; |
|
834
|
|
|
|
|
|
|
|
|
835
|
3046
|
100
|
|
|
|
|
if (type_char == 'n') { |
|
836
|
2868
|
|
|
|
|
|
p->xref[obj_num].type = PDFMAKE_XREF_UNCOMPRESSED; |
|
837
|
2868
|
|
|
|
|
|
p->xref[obj_num].loc.offset = (uint64_t)off_tok.payload.int_val; |
|
838
|
|
|
|
|
|
|
} else { |
|
839
|
178
|
|
|
|
|
|
p->xref[obj_num].type = PDFMAKE_XREF_FREE; |
|
840
|
178
|
|
|
|
|
|
p->xref[obj_num].loc.next_free = (uint32_t)off_tok.payload.int_val; |
|
841
|
|
|
|
|
|
|
} |
|
842
|
|
|
|
|
|
|
} |
|
843
|
|
|
|
|
|
|
|
|
844
|
3046
|
100
|
|
|
|
|
if (obj_num >= p->xref_size) p->xref_size = obj_num + 1; |
|
845
|
3046
|
|
|
|
|
|
return PDFMAKE_OK; |
|
846
|
|
|
|
|
|
|
} |
|
847
|
|
|
|
|
|
|
|
|
848
|
206
|
|
|
|
|
|
pdfmake_err_t pdfmake_parse_xref_table(pdfmake_parser_t *p, uint64_t offset) { |
|
849
|
|
|
|
|
|
|
pdfmake_tok_t tok; |
|
850
|
|
|
|
|
|
|
pdfmake_tok_t next; |
|
851
|
|
|
|
|
|
|
uint32_t first_num, count; |
|
852
|
|
|
|
|
|
|
pdfmake_err_t err; |
|
853
|
|
|
|
|
|
|
uint32_t i; |
|
854
|
|
|
|
|
|
|
|
|
855
|
206
|
50
|
|
|
|
|
if (offset >= p->buf_len) { |
|
856
|
0
|
|
|
|
|
|
set_error(p, PDFMAKE_EXREF, (size_t)offset, "Xref offset beyond EOF"); |
|
857
|
0
|
|
|
|
|
|
return PDFMAKE_EXREF; |
|
858
|
|
|
|
|
|
|
} |
|
859
|
|
|
|
|
|
|
|
|
860
|
206
|
|
|
|
|
|
pdfmake_tokenizer_init(&p->tok, p->buf + offset, p->buf_len - offset); |
|
861
|
|
|
|
|
|
|
|
|
862
|
206
|
|
|
|
|
|
tok = pdfmake_tok_next_significant(&p->tok); |
|
863
|
206
|
50
|
|
|
|
|
if (tok.kind != PDFMAKE_TOK_KW_XREF) { |
|
864
|
0
|
|
|
|
|
|
set_error(p, PDFMAKE_EXREF, offset + tok.offset, "Expected 'xref' keyword"); |
|
865
|
0
|
|
|
|
|
|
return PDFMAKE_EXREF; |
|
866
|
|
|
|
|
|
|
} |
|
867
|
|
|
|
|
|
|
|
|
868
|
|
|
|
|
|
|
for (;;) { |
|
869
|
412
|
|
|
|
|
|
pdfmake_tok_peek_significant(&p->tok, &next); |
|
870
|
412
|
100
|
|
|
|
|
if (next.kind == PDFMAKE_TOK_KW_TRAILER) break; /* trailer follows */ |
|
871
|
|
|
|
|
|
|
|
|
872
|
206
|
|
|
|
|
|
err = parse_xref_subsection_header(p, offset, &first_num, &count); |
|
873
|
206
|
50
|
|
|
|
|
if (err != PDFMAKE_OK) return err; |
|
874
|
|
|
|
|
|
|
|
|
875
|
206
|
50
|
|
|
|
|
if (!ensure_xref_cap(p, first_num + count)) { |
|
876
|
0
|
|
|
|
|
|
set_error(p, PDFMAKE_ENOMEM, 0, "Failed to allocate xref table"); |
|
877
|
0
|
|
|
|
|
|
return PDFMAKE_ENOMEM; |
|
878
|
|
|
|
|
|
|
} |
|
879
|
|
|
|
|
|
|
|
|
880
|
3252
|
100
|
|
|
|
|
for (i = 0; i < count; i++) { |
|
881
|
3046
|
|
|
|
|
|
err = parse_xref_entry(p, offset, first_num + i); |
|
882
|
3046
|
50
|
|
|
|
|
if (err != PDFMAKE_OK) return err; |
|
883
|
|
|
|
|
|
|
} |
|
884
|
|
|
|
|
|
|
} |
|
885
|
|
|
|
|
|
|
|
|
886
|
206
|
|
|
|
|
|
return PDFMAKE_OK; |
|
887
|
|
|
|
|
|
|
} |
|
888
|
|
|
|
|
|
|
|
|
889
|
|
|
|
|
|
|
/*============================================================================ |
|
890
|
|
|
|
|
|
|
* Trailer parsing |
|
891
|
|
|
|
|
|
|
*==========================================================================*/ |
|
892
|
|
|
|
|
|
|
|
|
893
|
244
|
|
|
|
|
|
pdfmake_err_t pdfmake_parse_trailer(pdfmake_parser_t *p, pdfmake_obj_t *trailer) { |
|
894
|
|
|
|
|
|
|
pdfmake_tok_t tok; |
|
895
|
|
|
|
|
|
|
uint32_t root_id, info_id, size_id, prev_id, encrypt_id; |
|
896
|
|
|
|
|
|
|
pdfmake_obj_t *root; |
|
897
|
|
|
|
|
|
|
pdfmake_obj_t *info; |
|
898
|
|
|
|
|
|
|
pdfmake_obj_t *encrypt; |
|
899
|
|
|
|
|
|
|
uint32_t id_id; |
|
900
|
|
|
|
|
|
|
pdfmake_obj_t *id_arr; |
|
901
|
|
|
|
|
|
|
pdfmake_obj_t *size; |
|
902
|
|
|
|
|
|
|
pdfmake_obj_t *prev; |
|
903
|
|
|
|
|
|
|
|
|
904
|
244
|
|
|
|
|
|
tok = pdfmake_tok_next_significant(&p->tok); |
|
905
|
244
|
50
|
|
|
|
|
if (tok.kind != PDFMAKE_TOK_KW_TRAILER) { |
|
906
|
0
|
|
|
|
|
|
set_error(p, PDFMAKE_ETRAILER, tok.offset, "Expected 'trailer' keyword"); |
|
907
|
0
|
|
|
|
|
|
return PDFMAKE_ETRAILER; |
|
908
|
|
|
|
|
|
|
} |
|
909
|
|
|
|
|
|
|
|
|
910
|
244
|
|
|
|
|
|
*trailer = parse_object_internal(p); |
|
911
|
244
|
50
|
|
|
|
|
if (p->last_err != PDFMAKE_OK) return p->last_err; |
|
912
|
|
|
|
|
|
|
|
|
913
|
244
|
50
|
|
|
|
|
if (trailer->kind != PDFMAKE_DICT) { |
|
914
|
0
|
|
|
|
|
|
set_error(p, PDFMAKE_ETRAILER, tok.offset, "Trailer must be a dictionary"); |
|
915
|
0
|
|
|
|
|
|
return PDFMAKE_ETRAILER; |
|
916
|
|
|
|
|
|
|
} |
|
917
|
|
|
|
|
|
|
|
|
918
|
|
|
|
|
|
|
/* Extract trailer values */ |
|
919
|
244
|
|
|
|
|
|
root_id = intern_name(p, "Root", 4); |
|
920
|
244
|
|
|
|
|
|
info_id = intern_name(p, "Info", 4); |
|
921
|
244
|
|
|
|
|
|
size_id = intern_name(p, "Size", 4); |
|
922
|
244
|
|
|
|
|
|
prev_id = intern_name(p, "Prev", 4); |
|
923
|
244
|
|
|
|
|
|
encrypt_id = intern_name(p, "Encrypt", 7); |
|
924
|
|
|
|
|
|
|
|
|
925
|
244
|
|
|
|
|
|
root = pdfmake_dict_get(trailer, root_id); |
|
926
|
244
|
100
|
|
|
|
|
if (root && root->kind == PDFMAKE_REF) { |
|
|
|
50
|
|
|
|
|
|
|
927
|
228
|
|
|
|
|
|
p->root_num = root->as.ref.num; |
|
928
|
228
|
|
|
|
|
|
p->root_gen = root->as.ref.gen; |
|
929
|
|
|
|
|
|
|
} |
|
930
|
|
|
|
|
|
|
|
|
931
|
244
|
|
|
|
|
|
info = pdfmake_dict_get(trailer, info_id); |
|
932
|
244
|
100
|
|
|
|
|
if (info && info->kind == PDFMAKE_REF) { |
|
|
|
50
|
|
|
|
|
|
|
933
|
172
|
|
|
|
|
|
p->info_num = info->as.ref.num; |
|
934
|
172
|
|
|
|
|
|
p->info_gen = info->as.ref.gen; |
|
935
|
|
|
|
|
|
|
} |
|
936
|
|
|
|
|
|
|
|
|
937
|
244
|
|
|
|
|
|
encrypt = pdfmake_dict_get(trailer, encrypt_id); |
|
938
|
244
|
100
|
|
|
|
|
if (encrypt && encrypt->kind == PDFMAKE_REF) { |
|
|
|
50
|
|
|
|
|
|
|
939
|
66
|
|
|
|
|
|
p->encrypt_num = encrypt->as.ref.num; |
|
940
|
66
|
|
|
|
|
|
p->encrypt_gen = encrypt->as.ref.gen; |
|
941
|
|
|
|
|
|
|
} |
|
942
|
|
|
|
|
|
|
|
|
943
|
|
|
|
|
|
|
/* Extract /ID array (needed for encryption key derivation) */ |
|
944
|
244
|
|
|
|
|
|
id_id = intern_name(p, "ID", 2); |
|
945
|
244
|
|
|
|
|
|
id_arr = pdfmake_dict_get(trailer, id_id); |
|
946
|
244
|
100
|
|
|
|
|
if (id_arr && id_arr->kind == PDFMAKE_ARRAY && pdfmake_array_len(id_arr) >= 1) { |
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
947
|
192
|
|
|
|
|
|
pdfmake_obj_t *id0 = pdfmake_array_get(id_arr, 0); |
|
948
|
192
|
50
|
|
|
|
|
if (id0 && id0->kind == PDFMAKE_STR && p->doc_id_len == 0) { |
|
|
|
50
|
|
|
|
|
|
|
|
|
100
|
|
|
|
|
|
|
949
|
82
|
|
|
|
|
|
size_t len = id0->as.str.len; |
|
950
|
82
|
50
|
|
|
|
|
if (len > sizeof(p->doc_id)) len = sizeof(p->doc_id); |
|
951
|
82
|
|
|
|
|
|
memcpy(p->doc_id, id0->as.str.bytes, len); |
|
952
|
82
|
|
|
|
|
|
p->doc_id_len = len; |
|
953
|
|
|
|
|
|
|
} |
|
954
|
|
|
|
|
|
|
} |
|
955
|
|
|
|
|
|
|
|
|
956
|
244
|
|
|
|
|
|
size = pdfmake_dict_get(trailer, size_id); |
|
957
|
244
|
50
|
|
|
|
|
if (size && size->kind == PDFMAKE_INT) { |
|
|
|
50
|
|
|
|
|
|
|
958
|
244
|
50
|
|
|
|
|
if (!ensure_xref_cap(p, (uint32_t)size->as.i)) { |
|
959
|
0
|
|
|
|
|
|
return PDFMAKE_ENOMEM; |
|
960
|
|
|
|
|
|
|
} |
|
961
|
244
|
50
|
|
|
|
|
if ((size_t)size->as.i > p->xref_size) { |
|
962
|
0
|
|
|
|
|
|
p->xref_size = (size_t)size->as.i; |
|
963
|
|
|
|
|
|
|
} |
|
964
|
|
|
|
|
|
|
} |
|
965
|
|
|
|
|
|
|
|
|
966
|
|
|
|
|
|
|
/* Check for /Prev - incremental update chain */ |
|
967
|
244
|
|
|
|
|
|
prev = pdfmake_dict_get(trailer, prev_id); |
|
968
|
244
|
100
|
|
|
|
|
if (prev && prev->kind == PDFMAKE_INT) { |
|
|
|
50
|
|
|
|
|
|
|
969
|
48
|
|
|
|
|
|
uint64_t prev_offset = (uint64_t)prev->as.i; |
|
970
|
|
|
|
|
|
|
pdfmake_err_t err; |
|
971
|
|
|
|
|
|
|
pdfmake_obj_t prev_trailer; |
|
972
|
|
|
|
|
|
|
|
|
973
|
|
|
|
|
|
|
/* Cycle detection */ |
|
974
|
48
|
100
|
|
|
|
|
if (prev_visited(p, prev_offset)) { |
|
975
|
32
|
|
|
|
|
|
set_error(p, PDFMAKE_ECYCLE, 0, "Cycle detected in /Prev chain"); |
|
976
|
32
|
|
|
|
|
|
return PDFMAKE_ECYCLE; |
|
977
|
|
|
|
|
|
|
} |
|
978
|
|
|
|
|
|
|
|
|
979
|
16
|
50
|
|
|
|
|
if (!prev_add(p, prev_offset)) { |
|
980
|
0
|
|
|
|
|
|
return PDFMAKE_ENOMEM; |
|
981
|
|
|
|
|
|
|
} |
|
982
|
|
|
|
|
|
|
|
|
983
|
|
|
|
|
|
|
/* Parse previous xref */ |
|
984
|
16
|
|
|
|
|
|
err = pdfmake_parse_xref_table(p, prev_offset); |
|
985
|
16
|
50
|
|
|
|
|
if (err != PDFMAKE_OK) return err; |
|
986
|
|
|
|
|
|
|
|
|
987
|
|
|
|
|
|
|
/* Parse previous trailer */ |
|
988
|
16
|
|
|
|
|
|
err = pdfmake_parse_trailer(p, &prev_trailer); |
|
989
|
16
|
50
|
|
|
|
|
if (err != PDFMAKE_OK) return err; |
|
990
|
|
|
|
|
|
|
} |
|
991
|
|
|
|
|
|
|
|
|
992
|
212
|
|
|
|
|
|
return PDFMAKE_OK; |
|
993
|
|
|
|
|
|
|
} |
|
994
|
|
|
|
|
|
|
|
|
995
|
|
|
|
|
|
|
/*============================================================================ |
|
996
|
|
|
|
|
|
|
* Xref stream parser (§7.5.8) |
|
997
|
|
|
|
|
|
|
*==========================================================================*/ |
|
998
|
|
|
|
|
|
|
|
|
999
|
|
|
|
|
|
|
/* Parsed shape of the xref-stream dictionary parameters. */ |
|
1000
|
|
|
|
|
|
|
typedef struct { |
|
1001
|
|
|
|
|
|
|
int w1, w2, w3; /* /W field widths */ |
|
1002
|
|
|
|
|
|
|
int entry_size; /* sum of w1+w2+w3 */ |
|
1003
|
|
|
|
|
|
|
pdfmake_obj_t *index; /* /Index array (or NULL → default range) */ |
|
1004
|
|
|
|
|
|
|
int64_t size; /* /Size value (total object count) */ |
|
1005
|
|
|
|
|
|
|
} xref_stream_params_t; |
|
1006
|
|
|
|
|
|
|
|
|
1007
|
|
|
|
|
|
|
/* Extract /W, /Index and /Size from the xref-stream dict. Returns |
|
1008
|
|
|
|
|
|
|
* PDFMAKE_EXREF if /W is missing or malformed; /Index and /Size are |
|
1009
|
|
|
|
|
|
|
* optional (callers fall back to a default range). */ |
|
1010
|
|
|
|
|
|
|
static pdfmake_err_t |
|
1011
|
0
|
|
|
|
|
|
xref_stream_extract_params(pdfmake_parser_t *p, uint64_t offset, |
|
1012
|
|
|
|
|
|
|
pdfmake_obj_t *dict, |
|
1013
|
|
|
|
|
|
|
xref_stream_params_t *out) |
|
1014
|
|
|
|
|
|
|
{ |
|
1015
|
0
|
|
|
|
|
|
uint32_t w_id = intern_name(p, "W", 1); |
|
1016
|
0
|
|
|
|
|
|
pdfmake_obj_t *w_arr = pdfmake_dict_get(dict, w_id); |
|
1017
|
|
|
|
|
|
|
uint32_t index_id; |
|
1018
|
|
|
|
|
|
|
uint32_t size_id; |
|
1019
|
|
|
|
|
|
|
pdfmake_obj_t *size_obj; |
|
1020
|
|
|
|
|
|
|
|
|
1021
|
0
|
0
|
|
|
|
|
if (!w_arr || w_arr->kind != PDFMAKE_ARRAY || pdfmake_array_len(w_arr) < 3) { |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
1022
|
0
|
|
|
|
|
|
set_error(p, PDFMAKE_EXREF, offset, "Invalid /W array in xref stream"); |
|
1023
|
0
|
|
|
|
|
|
return PDFMAKE_EXREF; |
|
1024
|
|
|
|
|
|
|
} |
|
1025
|
|
|
|
|
|
|
|
|
1026
|
0
|
|
|
|
|
|
out->w1 = (int)pdfmake_get_int(pdfmake_array_get(w_arr, 0)); |
|
1027
|
0
|
|
|
|
|
|
out->w2 = (int)pdfmake_get_int(pdfmake_array_get(w_arr, 1)); |
|
1028
|
0
|
|
|
|
|
|
out->w3 = (int)pdfmake_get_int(pdfmake_array_get(w_arr, 2)); |
|
1029
|
0
|
|
|
|
|
|
out->entry_size = out->w1 + out->w2 + out->w3; |
|
1030
|
|
|
|
|
|
|
|
|
1031
|
0
|
|
|
|
|
|
index_id = intern_name(p, "Index", 5); |
|
1032
|
0
|
|
|
|
|
|
size_id = intern_name(p, "Size", 4); |
|
1033
|
0
|
|
|
|
|
|
out->index = pdfmake_dict_get(dict, index_id); |
|
1034
|
|
|
|
|
|
|
|
|
1035
|
0
|
|
|
|
|
|
size_obj = pdfmake_dict_get(dict, size_id); |
|
1036
|
0
|
0
|
|
|
|
|
out->size = (size_obj && size_obj->kind == PDFMAKE_INT) ? size_obj->as.i : 0; |
|
|
|
0
|
|
|
|
|
|
|
1037
|
|
|
|
|
|
|
|
|
1038
|
0
|
|
|
|
|
|
return PDFMAKE_OK; |
|
1039
|
|
|
|
|
|
|
} |
|
1040
|
|
|
|
|
|
|
|
|
1041
|
|
|
|
|
|
|
/* Propagate /Root, /Info and /Encrypt references from the xref-stream dict |
|
1042
|
|
|
|
|
|
|
* into the parser state. Mirrors the trailer dict handling used by classic |
|
1043
|
|
|
|
|
|
|
* xref tables. The /Prev chain is handled by the orchestrator because it |
|
1044
|
|
|
|
|
|
|
* recurses. */ |
|
1045
|
|
|
|
|
|
|
static void |
|
1046
|
0
|
|
|
|
|
|
xref_stream_apply_trailer(pdfmake_parser_t *p, pdfmake_obj_t *dict) |
|
1047
|
|
|
|
|
|
|
{ |
|
1048
|
0
|
|
|
|
|
|
uint32_t root_id = intern_name(p, "Root", 4); |
|
1049
|
0
|
|
|
|
|
|
uint32_t info_id = intern_name(p, "Info", 4); |
|
1050
|
0
|
|
|
|
|
|
uint32_t encrypt_id = intern_name(p, "Encrypt", 7); |
|
1051
|
|
|
|
|
|
|
pdfmake_obj_t *root; |
|
1052
|
|
|
|
|
|
|
pdfmake_obj_t *info; |
|
1053
|
|
|
|
|
|
|
pdfmake_obj_t *encrypt; |
|
1054
|
|
|
|
|
|
|
|
|
1055
|
0
|
|
|
|
|
|
root = pdfmake_dict_get(dict, root_id); |
|
1056
|
0
|
0
|
|
|
|
|
if (root && root->kind == PDFMAKE_REF) { |
|
|
|
0
|
|
|
|
|
|
|
1057
|
0
|
|
|
|
|
|
p->root_num = root->as.ref.num; |
|
1058
|
0
|
|
|
|
|
|
p->root_gen = root->as.ref.gen; |
|
1059
|
|
|
|
|
|
|
} |
|
1060
|
|
|
|
|
|
|
|
|
1061
|
0
|
|
|
|
|
|
info = pdfmake_dict_get(dict, info_id); |
|
1062
|
0
|
0
|
|
|
|
|
if (info && info->kind == PDFMAKE_REF) { |
|
|
|
0
|
|
|
|
|
|
|
1063
|
0
|
|
|
|
|
|
p->info_num = info->as.ref.num; |
|
1064
|
0
|
|
|
|
|
|
p->info_gen = info->as.ref.gen; |
|
1065
|
|
|
|
|
|
|
} |
|
1066
|
|
|
|
|
|
|
|
|
1067
|
0
|
|
|
|
|
|
encrypt = pdfmake_dict_get(dict, encrypt_id); |
|
1068
|
0
|
0
|
|
|
|
|
if (encrypt && encrypt->kind == PDFMAKE_REF) { |
|
|
|
0
|
|
|
|
|
|
|
1069
|
0
|
|
|
|
|
|
p->encrypt_num = encrypt->as.ref.num; |
|
1070
|
0
|
|
|
|
|
|
p->encrypt_gen = encrypt->as.ref.gen; |
|
1071
|
|
|
|
|
|
|
} |
|
1072
|
0
|
|
|
|
|
|
} |
|
1073
|
|
|
|
|
|
|
|
|
1074
|
|
|
|
|
|
|
/* Read one variable-width xref-stream entry at *pp (advancing it) and store |
|
1075
|
|
|
|
|
|
|
* it into p->xref[obj_num] unless that slot is already loaded from a later |
|
1076
|
|
|
|
|
|
|
* xref section. Called from both the /Index subsection loop and the |
|
1077
|
|
|
|
|
|
|
* default-range loop, which are otherwise identical. */ |
|
1078
|
|
|
|
|
|
|
static PDFMAKE_INLINE void |
|
1079
|
0
|
|
|
|
|
|
xref_stream_read_entry(pdfmake_parser_t *p, |
|
1080
|
|
|
|
|
|
|
const uint8_t **pp, |
|
1081
|
|
|
|
|
|
|
int w1, int w2, int w3, |
|
1082
|
|
|
|
|
|
|
uint32_t obj_num) |
|
1083
|
|
|
|
|
|
|
{ |
|
1084
|
0
|
|
|
|
|
|
const uint8_t *ptr = *pp; |
|
1085
|
0
|
|
|
|
|
|
uint64_t field1 = 0, field2 = 0, field3 = 0; |
|
1086
|
|
|
|
|
|
|
int j; |
|
1087
|
|
|
|
|
|
|
int type; |
|
1088
|
|
|
|
|
|
|
|
|
1089
|
0
|
0
|
|
|
|
|
for (j = 0; j < w1; j++) field1 = (field1 << 8) | *ptr++; |
|
1090
|
0
|
0
|
|
|
|
|
for (j = 0; j < w2; j++) field2 = (field2 << 8) | *ptr++; |
|
1091
|
0
|
0
|
|
|
|
|
for (j = 0; j < w3; j++) field3 = (field3 << 8) | *ptr++; |
|
1092
|
0
|
|
|
|
|
|
*pp = ptr; |
|
1093
|
|
|
|
|
|
|
|
|
1094
|
|
|
|
|
|
|
/* Default type is 1 when /W specifies a zero-width type field. */ |
|
1095
|
0
|
0
|
|
|
|
|
type = (w1 == 0) ? 1 : (int)field1; |
|
1096
|
|
|
|
|
|
|
|
|
1097
|
0
|
0
|
|
|
|
|
if (!p->xref[obj_num].loaded) { |
|
1098
|
0
|
|
|
|
|
|
p->xref[obj_num].num = obj_num; |
|
1099
|
0
|
|
|
|
|
|
switch (type) { |
|
1100
|
0
|
|
|
|
|
|
case 0: /* Free object */ |
|
1101
|
0
|
|
|
|
|
|
p->xref[obj_num].type = PDFMAKE_XREF_FREE; |
|
1102
|
0
|
|
|
|
|
|
p->xref[obj_num].loc.next_free = (uint32_t)field2; |
|
1103
|
0
|
|
|
|
|
|
p->xref[obj_num].gen = (uint16_t)field3; |
|
1104
|
0
|
|
|
|
|
|
break; |
|
1105
|
0
|
|
|
|
|
|
case 1: /* Uncompressed object */ |
|
1106
|
0
|
|
|
|
|
|
p->xref[obj_num].type = PDFMAKE_XREF_UNCOMPRESSED; |
|
1107
|
0
|
|
|
|
|
|
p->xref[obj_num].loc.offset = field2; |
|
1108
|
0
|
|
|
|
|
|
p->xref[obj_num].gen = (uint16_t)field3; |
|
1109
|
0
|
|
|
|
|
|
break; |
|
1110
|
0
|
|
|
|
|
|
case 2: /* Compressed object */ |
|
1111
|
0
|
|
|
|
|
|
p->xref[obj_num].type = PDFMAKE_XREF_COMPRESSED; |
|
1112
|
0
|
|
|
|
|
|
p->xref[obj_num].loc.compressed.obj_stm_num = (uint32_t)field2; |
|
1113
|
0
|
|
|
|
|
|
p->xref[obj_num].loc.compressed.index = (uint32_t)field3; |
|
1114
|
0
|
|
|
|
|
|
p->xref[obj_num].gen = 0; |
|
1115
|
0
|
|
|
|
|
|
break; |
|
1116
|
|
|
|
|
|
|
} |
|
1117
|
|
|
|
|
|
|
} |
|
1118
|
|
|
|
|
|
|
|
|
1119
|
0
|
0
|
|
|
|
|
if (obj_num >= p->xref_size) { |
|
1120
|
0
|
|
|
|
|
|
p->xref_size = obj_num + 1; |
|
1121
|
|
|
|
|
|
|
} |
|
1122
|
0
|
|
|
|
|
|
} |
|
1123
|
|
|
|
|
|
|
|
|
1124
|
20
|
|
|
|
|
|
pdfmake_err_t pdfmake_parse_xref_stream(pdfmake_parser_t *p, uint64_t offset) { |
|
1125
|
|
|
|
|
|
|
pdfmake_tok_t num_tok; |
|
1126
|
|
|
|
|
|
|
pdfmake_tok_t gen_tok; |
|
1127
|
|
|
|
|
|
|
pdfmake_tok_t obj_tok; |
|
1128
|
|
|
|
|
|
|
pdfmake_obj_t stream; |
|
1129
|
|
|
|
|
|
|
pdfmake_obj_t stream_dict_obj; |
|
1130
|
|
|
|
|
|
|
xref_stream_params_t params; |
|
1131
|
|
|
|
|
|
|
pdfmake_err_t err; |
|
1132
|
|
|
|
|
|
|
uint8_t *decoded_data; |
|
1133
|
|
|
|
|
|
|
size_t decoded_len; |
|
1134
|
|
|
|
|
|
|
const uint8_t *ptr; |
|
1135
|
|
|
|
|
|
|
const uint8_t *end; |
|
1136
|
|
|
|
|
|
|
uint32_t prev_id; |
|
1137
|
|
|
|
|
|
|
pdfmake_obj_t *prev; |
|
1138
|
|
|
|
|
|
|
|
|
1139
|
20
|
50
|
|
|
|
|
if (offset >= p->buf_len) { |
|
1140
|
0
|
|
|
|
|
|
set_error(p, PDFMAKE_EXREF, (size_t)offset, "Xref stream offset beyond EOF"); |
|
1141
|
0
|
|
|
|
|
|
return PDFMAKE_EXREF; |
|
1142
|
|
|
|
|
|
|
} |
|
1143
|
|
|
|
|
|
|
|
|
1144
|
|
|
|
|
|
|
/* Position tokenizer at xref stream object */ |
|
1145
|
20
|
|
|
|
|
|
pdfmake_tokenizer_init(&p->tok, p->buf + offset, p->buf_len - offset); |
|
1146
|
|
|
|
|
|
|
|
|
1147
|
|
|
|
|
|
|
/* Parse as indirect object */ |
|
1148
|
20
|
|
|
|
|
|
num_tok = pdfmake_tok_next_significant(&p->tok); |
|
1149
|
20
|
50
|
|
|
|
|
if (num_tok.kind != PDFMAKE_TOK_INT) { |
|
1150
|
0
|
|
|
|
|
|
set_error(p, PDFMAKE_EXREF, offset, "Expected xref stream object number"); |
|
1151
|
0
|
|
|
|
|
|
return PDFMAKE_EXREF; |
|
1152
|
|
|
|
|
|
|
} |
|
1153
|
|
|
|
|
|
|
|
|
1154
|
20
|
|
|
|
|
|
gen_tok = pdfmake_tok_next_significant(&p->tok); |
|
1155
|
20
|
50
|
|
|
|
|
if (gen_tok.kind != PDFMAKE_TOK_INT) { |
|
1156
|
0
|
|
|
|
|
|
set_error(p, PDFMAKE_EXREF, offset, "Expected xref stream generation"); |
|
1157
|
0
|
|
|
|
|
|
return PDFMAKE_EXREF; |
|
1158
|
|
|
|
|
|
|
} |
|
1159
|
|
|
|
|
|
|
|
|
1160
|
20
|
|
|
|
|
|
obj_tok = pdfmake_tok_next_significant(&p->tok); |
|
1161
|
20
|
50
|
|
|
|
|
if (obj_tok.kind != PDFMAKE_TOK_KW_OBJ) { |
|
1162
|
20
|
|
|
|
|
|
set_error(p, PDFMAKE_EXREF, offset, "Expected 'obj' keyword"); |
|
1163
|
20
|
|
|
|
|
|
return PDFMAKE_EXREF; |
|
1164
|
|
|
|
|
|
|
} |
|
1165
|
|
|
|
|
|
|
|
|
1166
|
|
|
|
|
|
|
/* Parse stream object */ |
|
1167
|
0
|
|
|
|
|
|
stream = parse_object_internal(p); |
|
1168
|
0
|
0
|
|
|
|
|
if (p->last_err != PDFMAKE_OK) return p->last_err; |
|
1169
|
|
|
|
|
|
|
|
|
1170
|
0
|
0
|
|
|
|
|
if (stream.kind != PDFMAKE_STREAM) { |
|
1171
|
0
|
|
|
|
|
|
set_error(p, PDFMAKE_EXREF, offset, "Xref stream is not a stream"); |
|
1172
|
0
|
|
|
|
|
|
return PDFMAKE_EXREF; |
|
1173
|
|
|
|
|
|
|
} |
|
1174
|
|
|
|
|
|
|
|
|
1175
|
|
|
|
|
|
|
/* Wrap the stream dict so we can call pdfmake_dict_get on it. */ |
|
1176
|
0
|
|
|
|
|
|
stream_dict_obj.kind = PDFMAKE_DICT; |
|
1177
|
0
|
|
|
|
|
|
stream_dict_obj.as.dict = stream.as.stream->dict; |
|
1178
|
|
|
|
|
|
|
|
|
1179
|
0
|
|
|
|
|
|
err = xref_stream_extract_params(p, offset, &stream_dict_obj, ¶ms); |
|
1180
|
0
|
0
|
|
|
|
|
if (err != PDFMAKE_OK) return err; |
|
1181
|
|
|
|
|
|
|
|
|
1182
|
|
|
|
|
|
|
/* Decode stream data */ |
|
1183
|
0
|
|
|
|
|
|
err = pdfmake_decode_stream(p, stream.as.stream, &decoded_data, &decoded_len); |
|
1184
|
0
|
0
|
|
|
|
|
if (err != PDFMAKE_OK) return err; |
|
1185
|
|
|
|
|
|
|
|
|
1186
|
|
|
|
|
|
|
/* Parse entries */ |
|
1187
|
0
|
|
|
|
|
|
ptr = decoded_data; |
|
1188
|
0
|
|
|
|
|
|
end = decoded_data + decoded_len; |
|
1189
|
|
|
|
|
|
|
|
|
1190
|
0
|
0
|
|
|
|
|
if (params.index && params.index->kind == PDFMAKE_ARRAY) { |
|
|
|
0
|
|
|
|
|
|
|
1191
|
|
|
|
|
|
|
/* Multiple subsections */ |
|
1192
|
0
|
|
|
|
|
|
size_t num_pairs = pdfmake_array_len(params.index) / 2; |
|
1193
|
|
|
|
|
|
|
size_t pair; |
|
1194
|
0
|
0
|
|
|
|
|
for (pair = 0; pair < num_pairs; pair++) { |
|
1195
|
0
|
|
|
|
|
|
pdfmake_obj_t *first = pdfmake_array_get(params.index, pair * 2); |
|
1196
|
0
|
|
|
|
|
|
pdfmake_obj_t *count = pdfmake_array_get(params.index, pair * 2 + 1); |
|
1197
|
|
|
|
|
|
|
uint32_t first_num; |
|
1198
|
|
|
|
|
|
|
uint32_t entry_count; |
|
1199
|
|
|
|
|
|
|
uint32_t i; |
|
1200
|
0
|
0
|
|
|
|
|
if (!first || !count) continue; |
|
|
|
0
|
|
|
|
|
|
|
1201
|
|
|
|
|
|
|
|
|
1202
|
0
|
|
|
|
|
|
first_num = (uint32_t)pdfmake_get_int(first); |
|
1203
|
0
|
|
|
|
|
|
entry_count = (uint32_t)pdfmake_get_int(count); |
|
1204
|
|
|
|
|
|
|
|
|
1205
|
0
|
0
|
|
|
|
|
if (!ensure_xref_cap(p, first_num + entry_count)) return PDFMAKE_ENOMEM; |
|
1206
|
|
|
|
|
|
|
|
|
1207
|
0
|
0
|
|
|
|
|
for (i = 0; i < entry_count && ptr + params.entry_size <= end; i++) { |
|
|
|
0
|
|
|
|
|
|
|
1208
|
0
|
|
|
|
|
|
xref_stream_read_entry(p, &ptr, params.w1, params.w2, params.w3, |
|
1209
|
|
|
|
|
|
|
first_num + i); |
|
1210
|
|
|
|
|
|
|
} |
|
1211
|
|
|
|
|
|
|
} |
|
1212
|
|
|
|
|
|
|
} else { |
|
1213
|
|
|
|
|
|
|
/* Default: single subsection from 0 to Size-1 */ |
|
1214
|
0
|
|
|
|
|
|
uint32_t entry_count = (uint32_t)params.size; |
|
1215
|
|
|
|
|
|
|
uint32_t obj_num; |
|
1216
|
0
|
0
|
|
|
|
|
if (!ensure_xref_cap(p, entry_count)) return PDFMAKE_ENOMEM; |
|
1217
|
|
|
|
|
|
|
|
|
1218
|
0
|
|
|
|
|
|
for (obj_num = 0; |
|
1219
|
0
|
0
|
|
|
|
|
obj_num < entry_count && ptr + params.entry_size <= end; |
|
|
|
0
|
|
|
|
|
|
|
1220
|
0
|
|
|
|
|
|
obj_num++) { |
|
1221
|
0
|
|
|
|
|
|
xref_stream_read_entry(p, &ptr, params.w1, params.w2, params.w3, obj_num); |
|
1222
|
|
|
|
|
|
|
} |
|
1223
|
|
|
|
|
|
|
} |
|
1224
|
|
|
|
|
|
|
|
|
1225
|
|
|
|
|
|
|
/* Propagate trailer-equivalent refs from the stream dict. */ |
|
1226
|
0
|
|
|
|
|
|
xref_stream_apply_trailer(p, &stream_dict_obj); |
|
1227
|
|
|
|
|
|
|
|
|
1228
|
|
|
|
|
|
|
/* Check for /Prev - incremental update chain */ |
|
1229
|
0
|
|
|
|
|
|
prev_id = intern_name(p, "Prev", 4); |
|
1230
|
0
|
|
|
|
|
|
prev = pdfmake_dict_get(&stream_dict_obj, prev_id); |
|
1231
|
0
|
0
|
|
|
|
|
if (prev && prev->kind == PDFMAKE_INT) { |
|
|
|
0
|
|
|
|
|
|
|
1232
|
0
|
|
|
|
|
|
uint64_t prev_offset = (uint64_t)prev->as.i; |
|
1233
|
|
|
|
|
|
|
pdfmake_tokenizer_t save_tok; |
|
1234
|
|
|
|
|
|
|
pdfmake_err_t prev_err; |
|
1235
|
|
|
|
|
|
|
|
|
1236
|
0
|
0
|
|
|
|
|
if (prev_visited(p, prev_offset)) { |
|
1237
|
0
|
|
|
|
|
|
set_error(p, PDFMAKE_ECYCLE, 0, "Cycle detected in /Prev chain"); |
|
1238
|
0
|
|
|
|
|
|
return PDFMAKE_ECYCLE; |
|
1239
|
|
|
|
|
|
|
} |
|
1240
|
|
|
|
|
|
|
|
|
1241
|
0
|
0
|
|
|
|
|
if (!prev_add(p, prev_offset)) { |
|
1242
|
0
|
|
|
|
|
|
return PDFMAKE_ENOMEM; |
|
1243
|
|
|
|
|
|
|
} |
|
1244
|
|
|
|
|
|
|
|
|
1245
|
|
|
|
|
|
|
/* Could be classic xref or xref stream */ |
|
1246
|
|
|
|
|
|
|
/* Try xref stream first, fall back to classic */ |
|
1247
|
0
|
|
|
|
|
|
save_tok = p->tok; |
|
1248
|
0
|
|
|
|
|
|
prev_err = pdfmake_parse_xref_stream(p, prev_offset); |
|
1249
|
0
|
0
|
|
|
|
|
if (prev_err != PDFMAKE_OK) { |
|
1250
|
0
|
|
|
|
|
|
p->tok = save_tok; |
|
1251
|
0
|
|
|
|
|
|
prev_err = pdfmake_parse_xref_table(p, prev_offset); |
|
1252
|
0
|
0
|
|
|
|
|
if (prev_err == PDFMAKE_OK) { |
|
1253
|
|
|
|
|
|
|
pdfmake_obj_t prev_trailer; |
|
1254
|
0
|
|
|
|
|
|
prev_err = pdfmake_parse_trailer(p, &prev_trailer); |
|
1255
|
|
|
|
|
|
|
} |
|
1256
|
|
|
|
|
|
|
} |
|
1257
|
0
|
0
|
|
|
|
|
if (prev_err != PDFMAKE_OK) return prev_err; |
|
1258
|
|
|
|
|
|
|
} |
|
1259
|
|
|
|
|
|
|
|
|
1260
|
0
|
|
|
|
|
|
return PDFMAKE_OK; |
|
1261
|
|
|
|
|
|
|
} |
|
1262
|
|
|
|
|
|
|
|
|
1263
|
|
|
|
|
|
|
/*============================================================================ |
|
1264
|
|
|
|
|
|
|
* Stream decoding helper |
|
1265
|
|
|
|
|
|
|
*==========================================================================*/ |
|
1266
|
|
|
|
|
|
|
|
|
1267
|
181
|
|
|
|
|
|
pdfmake_err_t pdfmake_decode_stream(pdfmake_parser_t *p, |
|
1268
|
|
|
|
|
|
|
pdfmake_stream_t *stream, |
|
1269
|
|
|
|
|
|
|
uint8_t **out_data, |
|
1270
|
|
|
|
|
|
|
size_t *out_len) { |
|
1271
|
|
|
|
|
|
|
pdfmake_obj_t dict_obj; |
|
1272
|
|
|
|
|
|
|
uint32_t filter_id; |
|
1273
|
|
|
|
|
|
|
uint32_t parms_id; |
|
1274
|
|
|
|
|
|
|
pdfmake_obj_t *filter; |
|
1275
|
|
|
|
|
|
|
pdfmake_obj_t *parms; |
|
1276
|
|
|
|
|
|
|
pdfmake_buf_t out; |
|
1277
|
|
|
|
|
|
|
pdfmake_err_t err; |
|
1278
|
|
|
|
|
|
|
size_t len; |
|
1279
|
|
|
|
|
|
|
uint8_t *data; |
|
1280
|
|
|
|
|
|
|
|
|
1281
|
181
|
50
|
|
|
|
|
if (stream->filtered) { |
|
1282
|
|
|
|
|
|
|
/* Already decoded - just return raw data */ |
|
1283
|
0
|
|
|
|
|
|
*out_data = (uint8_t *)stream->raw; |
|
1284
|
0
|
|
|
|
|
|
*out_len = stream->raw_len; |
|
1285
|
0
|
|
|
|
|
|
return PDFMAKE_OK; |
|
1286
|
|
|
|
|
|
|
} |
|
1287
|
|
|
|
|
|
|
|
|
1288
|
|
|
|
|
|
|
/* Get /Filter and /DecodeParms from stream dict */ |
|
1289
|
181
|
|
|
|
|
|
dict_obj.kind = PDFMAKE_DICT; |
|
1290
|
181
|
|
|
|
|
|
dict_obj.as.dict = stream->dict; |
|
1291
|
|
|
|
|
|
|
|
|
1292
|
181
|
|
|
|
|
|
filter_id = intern_name(p, "Filter", 6); |
|
1293
|
181
|
|
|
|
|
|
parms_id = intern_name(p, "DecodeParms", 11); |
|
1294
|
|
|
|
|
|
|
|
|
1295
|
181
|
|
|
|
|
|
filter = pdfmake_dict_get(&dict_obj, filter_id); |
|
1296
|
181
|
|
|
|
|
|
parms = pdfmake_dict_get(&dict_obj, parms_id); |
|
1297
|
|
|
|
|
|
|
|
|
1298
|
181
|
100
|
|
|
|
|
if (!filter) { |
|
1299
|
|
|
|
|
|
|
/* No filter - raw data */ |
|
1300
|
88
|
|
|
|
|
|
*out_data = (uint8_t *)stream->raw; |
|
1301
|
88
|
|
|
|
|
|
*out_len = stream->raw_len; |
|
1302
|
88
|
|
|
|
|
|
return PDFMAKE_OK; |
|
1303
|
|
|
|
|
|
|
} |
|
1304
|
|
|
|
|
|
|
|
|
1305
|
|
|
|
|
|
|
/* Decode through filter chain */ |
|
1306
|
93
|
|
|
|
|
|
pdfmake_buf_init(&out); |
|
1307
|
|
|
|
|
|
|
|
|
1308
|
93
|
|
|
|
|
|
err = pdfmake_filter_chain_decode(p->doc->arena, filter, parms, |
|
1309
|
93
|
|
|
|
|
|
stream->raw, stream->raw_len, &out); |
|
1310
|
93
|
50
|
|
|
|
|
if (err != PDFMAKE_OK) { |
|
1311
|
0
|
|
|
|
|
|
pdfmake_buf_free(&out); |
|
1312
|
0
|
|
|
|
|
|
set_error(p, err, 0, "Stream decode failed"); |
|
1313
|
0
|
|
|
|
|
|
return err; |
|
1314
|
|
|
|
|
|
|
} |
|
1315
|
|
|
|
|
|
|
|
|
1316
|
|
|
|
|
|
|
/* Copy to arena */ |
|
1317
|
93
|
|
|
|
|
|
len = pdfmake_buf_len(&out); |
|
1318
|
93
|
|
|
|
|
|
data = pdfmake_arena_alloc(p->doc->arena, len); |
|
1319
|
93
|
50
|
|
|
|
|
if (!data) { |
|
1320
|
0
|
|
|
|
|
|
pdfmake_buf_free(&out); |
|
1321
|
0
|
|
|
|
|
|
return PDFMAKE_ENOMEM; |
|
1322
|
|
|
|
|
|
|
} |
|
1323
|
93
|
|
|
|
|
|
memcpy(data, pdfmake_buf_data(&out), len); |
|
1324
|
93
|
|
|
|
|
|
pdfmake_buf_free(&out); |
|
1325
|
|
|
|
|
|
|
|
|
1326
|
93
|
|
|
|
|
|
*out_data = data; |
|
1327
|
93
|
|
|
|
|
|
*out_len = len; |
|
1328
|
93
|
|
|
|
|
|
return PDFMAKE_OK; |
|
1329
|
|
|
|
|
|
|
} |
|
1330
|
|
|
|
|
|
|
|
|
1331
|
|
|
|
|
|
|
/*============================================================================ |
|
1332
|
|
|
|
|
|
|
* Repair mode (§7.5 fallback) |
|
1333
|
|
|
|
|
|
|
*==========================================================================*/ |
|
1334
|
|
|
|
|
|
|
|
|
1335
|
44
|
|
|
|
|
|
pdfmake_err_t pdfmake_repair_xref(pdfmake_parser_t *p) { |
|
1336
|
44
|
|
|
|
|
|
size_t trailer_offset = 0; |
|
1337
|
|
|
|
|
|
|
|
|
1338
|
|
|
|
|
|
|
/* Scan entire file for "N G obj" patterns */ |
|
1339
|
44
|
|
|
|
|
|
pdfmake_tokenizer_init(&p->tok, p->buf, p->buf_len); |
|
1340
|
|
|
|
|
|
|
|
|
1341
|
2912
|
|
|
|
|
|
while (1) { |
|
1342
|
2956
|
|
|
|
|
|
pdfmake_tok_t tok = pdfmake_tok_next_significant(&p->tok); |
|
1343
|
2956
|
100
|
|
|
|
|
if (tok.kind == PDFMAKE_TOK_EOF) break; |
|
1344
|
|
|
|
|
|
|
|
|
1345
|
2912
|
100
|
|
|
|
|
if (tok.kind == PDFMAKE_TOK_INT) { |
|
1346
|
1820
|
|
|
|
|
|
size_t save_pos = p->tok.pos; |
|
1347
|
|
|
|
|
|
|
pdfmake_tok_t gen_tok; |
|
1348
|
|
|
|
|
|
|
|
|
1349
|
1820
|
|
|
|
|
|
gen_tok = pdfmake_tok_next_significant(&p->tok); |
|
1350
|
1820
|
100
|
|
|
|
|
if (gen_tok.kind == PDFMAKE_TOK_INT) { |
|
1351
|
1046
|
|
|
|
|
|
pdfmake_tok_t obj_tok = pdfmake_tok_next_significant(&p->tok); |
|
1352
|
1046
|
100
|
|
|
|
|
if (obj_tok.kind == PDFMAKE_TOK_KW_OBJ) { |
|
1353
|
|
|
|
|
|
|
/* Found object definition */ |
|
1354
|
256
|
|
|
|
|
|
uint32_t num = (uint32_t)tok.payload.int_val; |
|
1355
|
256
|
|
|
|
|
|
uint16_t gen = (uint16_t)gen_tok.payload.int_val; |
|
1356
|
|
|
|
|
|
|
|
|
1357
|
256
|
50
|
|
|
|
|
if (ensure_xref_cap(p, num + 1)) { |
|
1358
|
256
|
|
|
|
|
|
p->xref[num].num = num; |
|
1359
|
256
|
|
|
|
|
|
p->xref[num].gen = gen; |
|
1360
|
256
|
|
|
|
|
|
p->xref[num].type = PDFMAKE_XREF_UNCOMPRESSED; |
|
1361
|
256
|
|
|
|
|
|
p->xref[num].loc.offset = tok.offset; |
|
1362
|
256
|
100
|
|
|
|
|
if (num >= p->xref_size) { |
|
1363
|
52
|
|
|
|
|
|
p->xref_size = num + 1; |
|
1364
|
|
|
|
|
|
|
} |
|
1365
|
|
|
|
|
|
|
} |
|
1366
|
|
|
|
|
|
|
|
|
1367
|
|
|
|
|
|
|
/* Skip to endobj */ |
|
1368
|
5801
|
|
|
|
|
|
while (1) { |
|
1369
|
6057
|
|
|
|
|
|
pdfmake_tok_t skip = pdfmake_tok_next_significant(&p->tok); |
|
1370
|
6057
|
100
|
|
|
|
|
if (skip.kind == PDFMAKE_TOK_KW_ENDOBJ || |
|
1371
|
5817
|
100
|
|
|
|
|
skip.kind == PDFMAKE_TOK_EOF) { |
|
1372
|
|
|
|
|
|
|
break; |
|
1373
|
|
|
|
|
|
|
} |
|
1374
|
|
|
|
|
|
|
} |
|
1375
|
256
|
|
|
|
|
|
continue; |
|
1376
|
|
|
|
|
|
|
} |
|
1377
|
|
|
|
|
|
|
} |
|
1378
|
|
|
|
|
|
|
|
|
1379
|
|
|
|
|
|
|
/* Not an object - restore position */ |
|
1380
|
1564
|
|
|
|
|
|
p->tok.pos = save_pos; |
|
1381
|
1564
|
|
|
|
|
|
p->tok.has_peeked = 0; |
|
1382
|
|
|
|
|
|
|
} |
|
1383
|
|
|
|
|
|
|
} |
|
1384
|
|
|
|
|
|
|
|
|
1385
|
|
|
|
|
|
|
/* Look for trailer dict */ |
|
1386
|
44
|
|
|
|
|
|
pdfmake_tokenizer_init(&p->tok, p->buf, p->buf_len); |
|
1387
|
|
|
|
|
|
|
|
|
1388
|
|
|
|
|
|
|
/* Find last trailer keyword */ |
|
1389
|
17848
|
|
|
|
|
|
while (1) { |
|
1390
|
17892
|
|
|
|
|
|
pdfmake_tok_t tok = pdfmake_tok_next(&p->tok); |
|
1391
|
17892
|
100
|
|
|
|
|
if (tok.kind == PDFMAKE_TOK_EOF) break; |
|
1392
|
17848
|
100
|
|
|
|
|
if (tok.kind == PDFMAKE_TOK_KW_TRAILER) { |
|
1393
|
38
|
|
|
|
|
|
trailer_offset = tok.offset; |
|
1394
|
|
|
|
|
|
|
} |
|
1395
|
|
|
|
|
|
|
} |
|
1396
|
|
|
|
|
|
|
|
|
1397
|
44
|
100
|
|
|
|
|
if (trailer_offset > 0) { |
|
1398
|
|
|
|
|
|
|
pdfmake_obj_t trailer; |
|
1399
|
|
|
|
|
|
|
pdfmake_err_t trailer_err; |
|
1400
|
38
|
|
|
|
|
|
pdfmake_tokenizer_init(&p->tok, p->buf + trailer_offset, p->buf_len - trailer_offset); |
|
1401
|
38
|
|
|
|
|
|
p->last_err = PDFMAKE_OK; /* Clear any residual error from xref scan */ |
|
1402
|
38
|
|
|
|
|
|
trailer_err = pdfmake_parse_trailer(p, &trailer); |
|
1403
|
|
|
|
|
|
|
(void)trailer_err; /* Ignore errors - best effort */ |
|
1404
|
|
|
|
|
|
|
} |
|
1405
|
|
|
|
|
|
|
|
|
1406
|
44
|
|
|
|
|
|
return PDFMAKE_OK; |
|
1407
|
|
|
|
|
|
|
} |
|
1408
|
|
|
|
|
|
|
|
|
1409
|
|
|
|
|
|
|
/*============================================================================ |
|
1410
|
|
|
|
|
|
|
* Object resolution |
|
1411
|
|
|
|
|
|
|
*==========================================================================*/ |
|
1412
|
|
|
|
|
|
|
|
|
1413
|
11365
|
|
|
|
|
|
pdfmake_obj_t *pdfmake_parser_resolve(pdfmake_parser_t *p, pdfmake_ref_t ref) { |
|
1414
|
|
|
|
|
|
|
pdfmake_xref_entry_t *entry; |
|
1415
|
|
|
|
|
|
|
|
|
1416
|
11365
|
100
|
|
|
|
|
if (ref.num >= p->xref_size) return NULL; |
|
1417
|
|
|
|
|
|
|
|
|
1418
|
11364
|
|
|
|
|
|
entry = &p->xref[ref.num]; |
|
1419
|
11364
|
50
|
|
|
|
|
if (entry->type == PDFMAKE_XREF_FREE) return NULL; |
|
1420
|
|
|
|
|
|
|
|
|
1421
|
|
|
|
|
|
|
/* Check generation */ |
|
1422
|
11364
|
50
|
|
|
|
|
if (entry->gen != ref.gen) return NULL; |
|
1423
|
|
|
|
|
|
|
|
|
1424
|
|
|
|
|
|
|
/* Already loaded? */ |
|
1425
|
11364
|
100
|
|
|
|
|
if (entry->loaded && ref.num < p->doc->obj_count) { |
|
|
|
100
|
|
|
|
|
|
|
1426
|
10453
|
|
|
|
|
|
pdfmake_obj_t *obj = pdfmake_doc_get(p->doc, ref.num); |
|
1427
|
10453
|
50
|
|
|
|
|
if (obj) return obj; |
|
1428
|
|
|
|
|
|
|
} |
|
1429
|
|
|
|
|
|
|
|
|
1430
|
|
|
|
|
|
|
/* Load object */ |
|
1431
|
911
|
50
|
|
|
|
|
if (entry->type == PDFMAKE_XREF_UNCOMPRESSED) { |
|
1432
|
|
|
|
|
|
|
/* Save tokenizer state */ |
|
1433
|
911
|
|
|
|
|
|
pdfmake_tokenizer_t save_tok = p->tok; |
|
1434
|
|
|
|
|
|
|
pdfmake_err_t err; |
|
1435
|
|
|
|
|
|
|
|
|
1436
|
|
|
|
|
|
|
/* Position at object - use full buffer but set position */ |
|
1437
|
911
|
|
|
|
|
|
pdfmake_tokenizer_init(&p->tok, p->buf, p->buf_len); |
|
1438
|
911
|
|
|
|
|
|
p->tok.pos = entry->loc.offset; |
|
1439
|
|
|
|
|
|
|
|
|
1440
|
|
|
|
|
|
|
/* Parse indirect object */ |
|
1441
|
911
|
|
|
|
|
|
err = pdfmake_parse_indirect_object(p); |
|
1442
|
|
|
|
|
|
|
|
|
1443
|
|
|
|
|
|
|
/* Restore tokenizer */ |
|
1444
|
911
|
|
|
|
|
|
p->tok = save_tok; |
|
1445
|
|
|
|
|
|
|
|
|
1446
|
911
|
50
|
|
|
|
|
if (err == PDFMAKE_OK) { |
|
1447
|
911
|
|
|
|
|
|
entry->loaded = 1; |
|
1448
|
911
|
|
|
|
|
|
return pdfmake_doc_get(p->doc, ref.num); |
|
1449
|
|
|
|
|
|
|
} |
|
1450
|
|
|
|
|
|
|
} |
|
1451
|
0
|
0
|
|
|
|
|
else if (entry->type == PDFMAKE_XREF_COMPRESSED) { |
|
1452
|
|
|
|
|
|
|
/* Object is in an object stream - need to decompress and parse */ |
|
1453
|
|
|
|
|
|
|
/* First resolve the object stream */ |
|
1454
|
|
|
|
|
|
|
pdfmake_ref_t stm_ref; |
|
1455
|
|
|
|
|
|
|
pdfmake_obj_t *stm_obj; |
|
1456
|
|
|
|
|
|
|
uint8_t *decoded; |
|
1457
|
|
|
|
|
|
|
size_t decoded_len; |
|
1458
|
|
|
|
|
|
|
pdfmake_obj_t stm_dict_obj; |
|
1459
|
|
|
|
|
|
|
uint32_t n_id; |
|
1460
|
|
|
|
|
|
|
uint32_t first_id; |
|
1461
|
|
|
|
|
|
|
pdfmake_obj_t *n_obj; |
|
1462
|
|
|
|
|
|
|
pdfmake_obj_t *first_obj; |
|
1463
|
|
|
|
|
|
|
int64_t n, first; |
|
1464
|
|
|
|
|
|
|
pdfmake_tokenizer_t stm_tok; |
|
1465
|
|
|
|
|
|
|
uint32_t target_index; |
|
1466
|
|
|
|
|
|
|
int64_t target_offset; |
|
1467
|
|
|
|
|
|
|
int64_t i; |
|
1468
|
|
|
|
|
|
|
pdfmake_tokenizer_t save_tok; |
|
1469
|
|
|
|
|
|
|
pdfmake_obj_t obj; |
|
1470
|
|
|
|
|
|
|
size_t idx; |
|
1471
|
|
|
|
|
|
|
|
|
1472
|
0
|
|
|
|
|
|
stm_ref.num = entry->loc.compressed.obj_stm_num; |
|
1473
|
0
|
|
|
|
|
|
stm_ref.gen = 0; |
|
1474
|
0
|
|
|
|
|
|
stm_obj = pdfmake_parser_resolve(p, stm_ref); |
|
1475
|
|
|
|
|
|
|
|
|
1476
|
0
|
0
|
|
|
|
|
if (!stm_obj || stm_obj->kind != PDFMAKE_STREAM) return NULL; |
|
|
|
0
|
|
|
|
|
|
|
1477
|
|
|
|
|
|
|
|
|
1478
|
|
|
|
|
|
|
/* Decode stream */ |
|
1479
|
0
|
0
|
|
|
|
|
if (pdfmake_decode_stream(p, stm_obj->as.stream, &decoded, &decoded_len) != PDFMAKE_OK) { |
|
1480
|
0
|
|
|
|
|
|
return NULL; |
|
1481
|
|
|
|
|
|
|
} |
|
1482
|
|
|
|
|
|
|
|
|
1483
|
|
|
|
|
|
|
/* Parse object stream header */ |
|
1484
|
|
|
|
|
|
|
/* Format: N1 offset1 N2 offset2 ... followed by objects */ |
|
1485
|
0
|
|
|
|
|
|
stm_dict_obj.kind = PDFMAKE_DICT; |
|
1486
|
0
|
|
|
|
|
|
stm_dict_obj.as.dict = stm_obj->as.stream->dict; |
|
1487
|
|
|
|
|
|
|
|
|
1488
|
0
|
|
|
|
|
|
n_id = intern_name(p, "N", 1); |
|
1489
|
0
|
|
|
|
|
|
first_id = intern_name(p, "First", 5); |
|
1490
|
|
|
|
|
|
|
|
|
1491
|
0
|
|
|
|
|
|
n_obj = pdfmake_dict_get(&stm_dict_obj, n_id); |
|
1492
|
0
|
|
|
|
|
|
first_obj = pdfmake_dict_get(&stm_dict_obj, first_id); |
|
1493
|
|
|
|
|
|
|
|
|
1494
|
0
|
0
|
|
|
|
|
if (!n_obj || !first_obj) return NULL; |
|
|
|
0
|
|
|
|
|
|
|
1495
|
|
|
|
|
|
|
|
|
1496
|
0
|
|
|
|
|
|
n = pdfmake_get_int(n_obj); |
|
1497
|
0
|
|
|
|
|
|
first = pdfmake_get_int(first_obj); |
|
1498
|
|
|
|
|
|
|
|
|
1499
|
|
|
|
|
|
|
/* Parse header (pairs of obj_num offset) */ |
|
1500
|
0
|
|
|
|
|
|
pdfmake_tokenizer_init(&stm_tok, decoded, decoded_len); |
|
1501
|
|
|
|
|
|
|
|
|
1502
|
0
|
|
|
|
|
|
target_index = entry->loc.compressed.index; |
|
1503
|
0
|
|
|
|
|
|
target_offset = -1; |
|
1504
|
|
|
|
|
|
|
|
|
1505
|
0
|
0
|
|
|
|
|
for (i = 0; i < n; i++) { |
|
1506
|
0
|
|
|
|
|
|
pdfmake_tok_t num_tok = pdfmake_tok_next_significant(&stm_tok); |
|
1507
|
0
|
|
|
|
|
|
pdfmake_tok_t off_tok = pdfmake_tok_next_significant(&stm_tok); |
|
1508
|
|
|
|
|
|
|
|
|
1509
|
0
|
0
|
|
|
|
|
if (num_tok.kind != PDFMAKE_TOK_INT || off_tok.kind != PDFMAKE_TOK_INT) { |
|
|
|
0
|
|
|
|
|
|
|
1510
|
|
|
|
|
|
|
break; |
|
1511
|
|
|
|
|
|
|
} |
|
1512
|
|
|
|
|
|
|
|
|
1513
|
0
|
0
|
|
|
|
|
if ((uint32_t)i == target_index) { |
|
1514
|
0
|
|
|
|
|
|
target_offset = first + off_tok.payload.int_val; |
|
1515
|
0
|
|
|
|
|
|
break; |
|
1516
|
|
|
|
|
|
|
} |
|
1517
|
|
|
|
|
|
|
} |
|
1518
|
|
|
|
|
|
|
|
|
1519
|
0
|
0
|
|
|
|
|
if (target_offset < 0 || (size_t)target_offset >= decoded_len) { |
|
|
|
0
|
|
|
|
|
|
|
1520
|
0
|
|
|
|
|
|
return NULL; |
|
1521
|
|
|
|
|
|
|
} |
|
1522
|
|
|
|
|
|
|
|
|
1523
|
|
|
|
|
|
|
/* Parse object at offset */ |
|
1524
|
0
|
|
|
|
|
|
save_tok = p->tok; |
|
1525
|
0
|
|
|
|
|
|
pdfmake_tokenizer_init(&p->tok, decoded + target_offset, decoded_len - target_offset); |
|
1526
|
|
|
|
|
|
|
|
|
1527
|
0
|
|
|
|
|
|
obj = parse_object_internal(p); |
|
1528
|
0
|
|
|
|
|
|
p->tok = save_tok; |
|
1529
|
|
|
|
|
|
|
|
|
1530
|
0
|
0
|
|
|
|
|
if (p->last_err != PDFMAKE_OK) return NULL; |
|
1531
|
|
|
|
|
|
|
|
|
1532
|
|
|
|
|
|
|
/* Store in document (1-based numbering: object N at index N-1) */ |
|
1533
|
0
|
0
|
|
|
|
|
while (ref.num > p->doc->obj_cap) { |
|
1534
|
0
|
0
|
|
|
|
|
size_t new_cap = p->doc->obj_cap ? p->doc->obj_cap * 2 : PDFMAKE_DOC_INIT_CAP; |
|
1535
|
|
|
|
|
|
|
pdfmake_indirect_t *new_objs; |
|
1536
|
0
|
0
|
|
|
|
|
while (new_cap < ref.num) new_cap *= 2; |
|
1537
|
0
|
|
|
|
|
|
new_objs = realloc(p->doc->objects, |
|
1538
|
|
|
|
|
|
|
new_cap * sizeof(pdfmake_indirect_t)); |
|
1539
|
0
|
0
|
|
|
|
|
if (!new_objs) return NULL; |
|
1540
|
0
|
|
|
|
|
|
memset(new_objs + p->doc->obj_cap, 0, |
|
1541
|
0
|
|
|
|
|
|
(new_cap - p->doc->obj_cap) * sizeof(pdfmake_indirect_t)); |
|
1542
|
0
|
|
|
|
|
|
p->doc->objects = new_objs; |
|
1543
|
0
|
|
|
|
|
|
p->doc->obj_cap = new_cap; |
|
1544
|
|
|
|
|
|
|
} |
|
1545
|
|
|
|
|
|
|
|
|
1546
|
0
|
|
|
|
|
|
idx = ref.num - 1; |
|
1547
|
0
|
|
|
|
|
|
p->doc->objects[idx].num = ref.num; |
|
1548
|
0
|
|
|
|
|
|
p->doc->objects[idx].gen = 0; |
|
1549
|
0
|
|
|
|
|
|
p->doc->objects[idx].obj = obj; |
|
1550
|
0
|
|
|
|
|
|
p->doc->objects[idx].in_use = 1; |
|
1551
|
|
|
|
|
|
|
|
|
1552
|
0
|
0
|
|
|
|
|
if (ref.num > p->doc->obj_count) { |
|
1553
|
0
|
|
|
|
|
|
p->doc->obj_count = ref.num; |
|
1554
|
|
|
|
|
|
|
} |
|
1555
|
|
|
|
|
|
|
|
|
1556
|
0
|
|
|
|
|
|
entry->loaded = 1; |
|
1557
|
0
|
|
|
|
|
|
return &p->doc->objects[idx].obj; |
|
1558
|
|
|
|
|
|
|
} |
|
1559
|
|
|
|
|
|
|
|
|
1560
|
0
|
|
|
|
|
|
return NULL; |
|
1561
|
|
|
|
|
|
|
} |
|
1562
|
|
|
|
|
|
|
|
|
1563
|
0
|
|
|
|
|
|
pdfmake_obj_t *pdfmake_doc_resolve(pdfmake_doc_t *doc, pdfmake_parser_t *parser, |
|
1564
|
|
|
|
|
|
|
uint32_t num, uint16_t gen) { |
|
1565
|
|
|
|
|
|
|
pdfmake_ref_t ref; |
|
1566
|
|
|
|
|
|
|
(void)doc; /* doc info is in parser->doc, used for consistency with other APIs */ |
|
1567
|
0
|
|
|
|
|
|
ref.num = num; |
|
1568
|
0
|
|
|
|
|
|
ref.gen = gen; |
|
1569
|
0
|
|
|
|
|
|
return pdfmake_parser_resolve(parser, ref); |
|
1570
|
|
|
|
|
|
|
} |
|
1571
|
|
|
|
|
|
|
|
|
1572
|
|
|
|
|
|
|
/*============================================================================ |
|
1573
|
|
|
|
|
|
|
* Main parser entry point |
|
1574
|
|
|
|
|
|
|
*==========================================================================*/ |
|
1575
|
|
|
|
|
|
|
|
|
1576
|
218
|
|
|
|
|
|
pdfmake_err_t pdfmake_parser_run(pdfmake_parser_t *p, pdfmake_doc_t **out_doc) { |
|
1577
|
|
|
|
|
|
|
pdfmake_err_t err; |
|
1578
|
|
|
|
|
|
|
int major, minor; |
|
1579
|
|
|
|
|
|
|
uint64_t xref_offset; |
|
1580
|
|
|
|
|
|
|
pdfmake_tok_t first; |
|
1581
|
|
|
|
|
|
|
pdfmake_obj_t trailer; |
|
1582
|
|
|
|
|
|
|
|
|
1583
|
|
|
|
|
|
|
/* Check header */ |
|
1584
|
218
|
|
|
|
|
|
err = pdfmake_check_header(p, &major, &minor); |
|
1585
|
218
|
50
|
|
|
|
|
if (err != PDFMAKE_OK) return err; |
|
1586
|
|
|
|
|
|
|
|
|
1587
|
|
|
|
|
|
|
/* Locate startxref */ |
|
1588
|
218
|
|
|
|
|
|
err = pdfmake_locate_startxref(p, &xref_offset); |
|
1589
|
218
|
100
|
|
|
|
|
if (err != PDFMAKE_OK) { |
|
1590
|
6
|
50
|
|
|
|
|
if (p->repair) { |
|
1591
|
6
|
|
|
|
|
|
err = pdfmake_repair_xref(p); |
|
1592
|
6
|
50
|
|
|
|
|
if (err != PDFMAKE_OK) return err; |
|
1593
|
|
|
|
|
|
|
} else { |
|
1594
|
0
|
|
|
|
|
|
return err; |
|
1595
|
|
|
|
|
|
|
} |
|
1596
|
|
|
|
|
|
|
} else { |
|
1597
|
|
|
|
|
|
|
/* Add to visited list for cycle detection */ |
|
1598
|
212
|
50
|
|
|
|
|
if (!prev_add(p, xref_offset)) { |
|
1599
|
0
|
|
|
|
|
|
return PDFMAKE_ENOMEM; |
|
1600
|
|
|
|
|
|
|
} |
|
1601
|
|
|
|
|
|
|
|
|
1602
|
|
|
|
|
|
|
/* Try xref stream first (common in modern PDFs) */ |
|
1603
|
|
|
|
|
|
|
/* Check what's at the offset */ |
|
1604
|
212
|
50
|
|
|
|
|
if (xref_offset < p->buf_len) { |
|
1605
|
212
|
|
|
|
|
|
pdfmake_tokenizer_init(&p->tok, p->buf + xref_offset, p->buf_len - xref_offset); |
|
1606
|
212
|
|
|
|
|
|
pdfmake_tok_peek_significant(&p->tok, &first); |
|
1607
|
|
|
|
|
|
|
|
|
1608
|
212
|
100
|
|
|
|
|
if (first.kind == PDFMAKE_TOK_INT) { |
|
1609
|
|
|
|
|
|
|
/* Xref stream */ |
|
1610
|
20
|
|
|
|
|
|
err = pdfmake_parse_xref_stream(p, xref_offset); |
|
1611
|
192
|
100
|
|
|
|
|
} else if (first.kind == PDFMAKE_TOK_KW_XREF) { |
|
1612
|
|
|
|
|
|
|
/* Classic xref */ |
|
1613
|
190
|
|
|
|
|
|
err = pdfmake_parse_xref_table(p, xref_offset); |
|
1614
|
190
|
50
|
|
|
|
|
if (err == PDFMAKE_OK) { |
|
1615
|
190
|
|
|
|
|
|
err = pdfmake_parse_trailer(p, &trailer); |
|
1616
|
|
|
|
|
|
|
} |
|
1617
|
|
|
|
|
|
|
} else { |
|
1618
|
2
|
|
|
|
|
|
err = PDFMAKE_EXREF; |
|
1619
|
2
|
|
|
|
|
|
set_error(p, err, (size_t)xref_offset, "Invalid xref location"); |
|
1620
|
|
|
|
|
|
|
} |
|
1621
|
|
|
|
|
|
|
|
|
1622
|
212
|
100
|
|
|
|
|
if (err != PDFMAKE_OK) { |
|
1623
|
38
|
50
|
|
|
|
|
if (p->repair) { |
|
1624
|
38
|
|
|
|
|
|
err = pdfmake_repair_xref(p); |
|
1625
|
38
|
50
|
|
|
|
|
if (err != PDFMAKE_OK) return err; |
|
1626
|
|
|
|
|
|
|
} else { |
|
1627
|
0
|
|
|
|
|
|
return err; |
|
1628
|
|
|
|
|
|
|
} |
|
1629
|
|
|
|
|
|
|
} |
|
1630
|
|
|
|
|
|
|
} else { |
|
1631
|
|
|
|
|
|
|
/* Offset beyond file - try repair if enabled */ |
|
1632
|
0
|
0
|
|
|
|
|
if (p->repair) { |
|
1633
|
0
|
|
|
|
|
|
err = pdfmake_repair_xref(p); |
|
1634
|
0
|
0
|
|
|
|
|
if (err != PDFMAKE_OK) return err; |
|
1635
|
|
|
|
|
|
|
} else { |
|
1636
|
0
|
|
|
|
|
|
set_error(p, PDFMAKE_EXREF, (size_t)xref_offset, "Xref offset beyond EOF"); |
|
1637
|
0
|
|
|
|
|
|
return PDFMAKE_EXREF; |
|
1638
|
|
|
|
|
|
|
} |
|
1639
|
|
|
|
|
|
|
} |
|
1640
|
|
|
|
|
|
|
} |
|
1641
|
|
|
|
|
|
|
|
|
1642
|
|
|
|
|
|
|
/* Set document trailer refs */ |
|
1643
|
218
|
100
|
|
|
|
|
if (p->root_num > 0) { |
|
1644
|
212
|
|
|
|
|
|
pdfmake_doc_set_root(p->doc, p->root_num, p->root_gen); |
|
1645
|
|
|
|
|
|
|
} |
|
1646
|
218
|
100
|
|
|
|
|
if (p->info_num > 0) { |
|
1647
|
156
|
|
|
|
|
|
pdfmake_doc_set_info(p->doc, p->info_num, p->info_gen); |
|
1648
|
|
|
|
|
|
|
} |
|
1649
|
|
|
|
|
|
|
|
|
1650
|
218
|
|
|
|
|
|
*out_doc = p->doc; |
|
1651
|
218
|
|
|
|
|
|
return PDFMAKE_OK; |
|
1652
|
|
|
|
|
|
|
} |