| line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
|
1
|
|
|
|
|
|
|
MODULE = PDF::Make PACKAGE = PDF::Make::Extract |
|
2
|
|
|
|
|
|
|
PROTOTYPES: ENABLE |
|
3
|
|
|
|
|
|
|
|
|
4
|
|
|
|
|
|
|
SV * |
|
5
|
|
|
|
|
|
|
_extract_from_reader(class, reader_sv, page_index) |
|
6
|
|
|
|
|
|
|
char *class |
|
7
|
|
|
|
|
|
|
pdfmake_reader_xs_t *reader_sv |
|
8
|
|
|
|
|
|
|
UV page_index |
|
9
|
|
|
|
|
|
|
PREINIT: |
|
10
|
|
|
|
|
|
|
pdfmake_err_t err; |
|
11
|
|
|
|
|
|
|
pdfmake_arena_t *arena; |
|
12
|
|
|
|
|
|
|
pdfmake_interp_t *interp; |
|
13
|
|
|
|
|
|
|
pdfmake_textract_result_t *result; |
|
14
|
|
|
|
|
|
|
pdfmake_textract_options_t opts; |
|
15
|
|
|
|
|
|
|
pdfmake_reader_page_t *rpage; |
|
16
|
|
|
|
|
|
|
pdfmake_buf_t content_buf; |
|
17
|
|
|
|
|
|
|
pdfmake_obj_t *resources; |
|
18
|
|
|
|
|
|
|
char outbuf[65536]; |
|
19
|
|
|
|
|
|
|
size_t outlen; |
|
20
|
|
|
|
|
|
|
CODE: |
|
21
|
|
|
|
|
|
|
PERL_UNUSED_VAR(class); |
|
22
|
4
|
|
|
|
|
|
pdfmake_buf_init(&content_buf); |
|
23
|
|
|
|
|
|
|
|
|
24
|
4
|
50
|
|
|
|
|
if (page_index >= pdfmake_reader_page_count(reader_sv->reader)) |
|
25
|
0
|
|
|
|
|
|
croak("PDF::Make::Extract: page index %lu out of range", (unsigned long)page_index); |
|
26
|
|
|
|
|
|
|
|
|
27
|
4
|
|
|
|
|
|
rpage = pdfmake_reader_page_at(reader_sv->reader, page_index); |
|
28
|
4
|
50
|
|
|
|
|
if (!rpage) |
|
29
|
0
|
|
|
|
|
|
croak("PDF::Make::Extract: failed to get page %lu", (unsigned long)page_index); |
|
30
|
|
|
|
|
|
|
|
|
31
|
|
|
|
|
|
|
/* Get content stream bytes */ |
|
32
|
4
|
|
|
|
|
|
err = pdfmake_reader_page_content_bytes(reader_sv->reader, rpage, &content_buf); |
|
33
|
4
|
50
|
|
|
|
|
if (err != PDFMAKE_OK || content_buf.len == 0) { |
|
|
|
50
|
|
|
|
|
|
|
34
|
0
|
|
|
|
|
|
pdfmake_buf_free(&content_buf); |
|
35
|
0
|
|
|
|
|
|
RETVAL = newSVpvn("", 0); |
|
36
|
|
|
|
|
|
|
} else { |
|
37
|
|
|
|
|
|
|
/* Create interpreter with resources */ |
|
38
|
|
|
|
|
|
|
/* Use the parser's arena so name IDs match across: stream dicts, |
|
39
|
|
|
|
|
|
|
* font dicts (all from parser), merged resources (keys re-interned |
|
40
|
|
|
|
|
|
|
* here during merge), and the interpreter's own lookups. */ |
|
41
|
4
|
|
|
|
|
|
arena = reader_sv->reader->parser->doc->arena; |
|
42
|
4
|
|
|
|
|
|
interp = pdfmake_interp_new(arena); |
|
43
|
4
|
50
|
|
|
|
|
if (!interp) { |
|
44
|
0
|
|
|
|
|
|
pdfmake_buf_free(&content_buf); |
|
45
|
0
|
|
|
|
|
|
croak("PDF::Make::Extract: failed to create interpreter"); |
|
46
|
|
|
|
|
|
|
} |
|
47
|
|
|
|
|
|
|
/* Phase 7: enable Form XObject recursion */ |
|
48
|
4
|
|
|
|
|
|
pdfmake_interp_set_reader(interp, reader_sv->reader); |
|
49
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
/* Set page resources so Tf can resolve fonts */ |
|
51
|
4
|
|
|
|
|
|
resources = pdfmake_reader_page_resources(reader_sv->reader, rpage); |
|
52
|
4
|
50
|
|
|
|
|
if (resources) { |
|
53
|
4
|
|
|
|
|
|
pdfmake_interp_set_resources(interp, resources); |
|
54
|
|
|
|
|
|
|
} |
|
55
|
|
|
|
|
|
|
|
|
56
|
|
|
|
|
|
|
/* Run extraction */ |
|
57
|
4
|
|
|
|
|
|
result = pdfmake_textract_new(arena); |
|
58
|
4
|
|
|
|
|
|
pdfmake_textract_set_reader(result, reader_sv->reader); |
|
59
|
4
|
|
|
|
|
|
opts = pdfmake_textract_default_options(); |
|
60
|
4
|
|
|
|
|
|
err = pdfmake_textract_run(interp, content_buf.data, content_buf.len, &opts, result); |
|
61
|
|
|
|
|
|
|
|
|
62
|
4
|
|
|
|
|
|
pdfmake_interp_free(interp); |
|
63
|
4
|
|
|
|
|
|
pdfmake_buf_free(&content_buf); |
|
64
|
|
|
|
|
|
|
|
|
65
|
4
|
50
|
|
|
|
|
if (err != PDFMAKE_OK) { |
|
66
|
0
|
|
|
|
|
|
pdfmake_textract_free(result); |
|
67
|
0
|
|
|
|
|
|
croak("PDF::Make::Extract: extraction failed"); |
|
68
|
|
|
|
|
|
|
} |
|
69
|
|
|
|
|
|
|
|
|
70
|
4
|
|
|
|
|
|
outlen = pdfmake_textract_to_utf8(result, outbuf, sizeof(outbuf) - 1); |
|
71
|
4
|
|
|
|
|
|
outbuf[outlen] = '\0'; |
|
72
|
|
|
|
|
|
|
|
|
73
|
4
|
|
|
|
|
|
RETVAL = newSVpvn(outbuf, outlen); |
|
74
|
4
|
|
|
|
|
|
SvUTF8_on(RETVAL); |
|
75
|
|
|
|
|
|
|
|
|
76
|
4
|
|
|
|
|
|
pdfmake_textract_free(result); |
|
77
|
|
|
|
|
|
|
/* arena owned by reader; do not free here */ |
|
78
|
|
|
|
|
|
|
} |
|
79
|
|
|
|
|
|
|
OUTPUT: |
|
80
|
|
|
|
|
|
|
RETVAL |
|
81
|
|
|
|
|
|
|
|
|
82
|
|
|
|
|
|
|
SV * |
|
83
|
|
|
|
|
|
|
_extract_structured(class, reader_sv, page_index, ...) |
|
84
|
|
|
|
|
|
|
char *class |
|
85
|
|
|
|
|
|
|
pdfmake_reader_xs_t *reader_sv |
|
86
|
|
|
|
|
|
|
UV page_index |
|
87
|
|
|
|
|
|
|
PREINIT: |
|
88
|
|
|
|
|
|
|
pdfmake_err_t err; |
|
89
|
|
|
|
|
|
|
pdfmake_arena_t *arena; |
|
90
|
|
|
|
|
|
|
pdfmake_interp_t *interp; |
|
91
|
|
|
|
|
|
|
pdfmake_textract_result_t *result; |
|
92
|
|
|
|
|
|
|
pdfmake_textract_options_t opts; |
|
93
|
|
|
|
|
|
|
pdfmake_reader_page_t *rpage; |
|
94
|
|
|
|
|
|
|
pdfmake_buf_t content_buf; |
|
95
|
|
|
|
|
|
|
pdfmake_obj_t *resources; |
|
96
|
54
|
|
|
|
|
|
int include_invisible = 1; /* default: include Tr=3 text (OCR) */ |
|
97
|
|
|
|
|
|
|
CODE: |
|
98
|
|
|
|
|
|
|
PERL_UNUSED_VAR(class); |
|
99
|
|
|
|
|
|
|
/* Optional 4th arg: include_invisible (0 or 1). */ |
|
100
|
54
|
50
|
|
|
|
|
if (items > 3) include_invisible = (int)SvIV(ST(3)); |
|
101
|
54
|
|
|
|
|
|
pdfmake_buf_init(&content_buf); |
|
102
|
|
|
|
|
|
|
|
|
103
|
54
|
50
|
|
|
|
|
if (page_index >= pdfmake_reader_page_count(reader_sv->reader)) |
|
104
|
0
|
|
|
|
|
|
croak("PDF::Make::Extract: page index %lu out of range", (unsigned long)page_index); |
|
105
|
|
|
|
|
|
|
|
|
106
|
54
|
|
|
|
|
|
rpage = pdfmake_reader_page_at(reader_sv->reader, page_index); |
|
107
|
54
|
50
|
|
|
|
|
if (!rpage) |
|
108
|
0
|
|
|
|
|
|
croak("PDF::Make::Extract: failed to get page %lu", (unsigned long)page_index); |
|
109
|
|
|
|
|
|
|
|
|
110
|
54
|
|
|
|
|
|
err = pdfmake_reader_page_content_bytes(reader_sv->reader, rpage, &content_buf); |
|
111
|
54
|
50
|
|
|
|
|
if (err != PDFMAKE_OK || content_buf.len == 0) { |
|
|
|
50
|
|
|
|
|
|
|
112
|
0
|
|
|
|
|
|
pdfmake_buf_free(&content_buf); |
|
113
|
0
|
|
|
|
|
|
RETVAL = newRV_noinc((SV *)newAV()); |
|
114
|
|
|
|
|
|
|
} else { |
|
115
|
|
|
|
|
|
|
/* Use the parser's arena so name IDs match across: stream dicts, |
|
116
|
|
|
|
|
|
|
* font dicts (all from parser), merged resources (keys re-interned |
|
117
|
|
|
|
|
|
|
* here during merge), and the interpreter's own lookups. */ |
|
118
|
54
|
|
|
|
|
|
arena = reader_sv->reader->parser->doc->arena; |
|
119
|
54
|
|
|
|
|
|
interp = pdfmake_interp_new(arena); |
|
120
|
54
|
50
|
|
|
|
|
if (!interp) { |
|
121
|
0
|
|
|
|
|
|
pdfmake_buf_free(&content_buf); |
|
122
|
0
|
|
|
|
|
|
croak("PDF::Make::Extract: failed to create interpreter"); |
|
123
|
|
|
|
|
|
|
} |
|
124
|
|
|
|
|
|
|
/* Phase 7: enable Form XObject recursion */ |
|
125
|
54
|
|
|
|
|
|
pdfmake_interp_set_reader(interp, reader_sv->reader); |
|
126
|
|
|
|
|
|
|
|
|
127
|
54
|
|
|
|
|
|
resources = pdfmake_reader_page_resources(reader_sv->reader, rpage); |
|
128
|
54
|
50
|
|
|
|
|
if (resources) |
|
129
|
54
|
|
|
|
|
|
pdfmake_interp_set_resources(interp, resources); |
|
130
|
|
|
|
|
|
|
|
|
131
|
54
|
|
|
|
|
|
result = pdfmake_textract_new(arena); |
|
132
|
54
|
|
|
|
|
|
pdfmake_textract_set_reader(result, reader_sv->reader); |
|
133
|
54
|
|
|
|
|
|
opts = pdfmake_textract_default_options(); |
|
134
|
54
|
|
|
|
|
|
opts.include_invisible = include_invisible; |
|
135
|
|
|
|
|
|
|
|
|
136
|
|
|
|
|
|
|
/* Phase 12: resolve /StructTreeRoot from the catalog (if the |
|
137
|
|
|
|
|
|
|
* PDF is tagged) so the extractor can look up structure roles |
|
138
|
|
|
|
|
|
|
* from MCIDs the visitor collects. */ |
|
139
|
54
|
50
|
|
|
|
|
if (reader_sv->reader->catalog) { |
|
140
|
54
|
|
|
|
|
|
uint32_t str_key = pdfmake_arena_intern_name( |
|
141
|
|
|
|
|
|
|
arena, "StructTreeRoot", 14); |
|
142
|
54
|
|
|
|
|
|
pdfmake_obj_t *str_root = pdfmake_dict_get( |
|
143
|
54
|
|
|
|
|
|
reader_sv->reader->catalog, str_key); |
|
144
|
54
|
100
|
|
|
|
|
if (str_root) { |
|
145
|
1
|
|
|
|
|
|
pdfmake_textract_resolve_struct_tree( |
|
146
|
|
|
|
|
|
|
result, str_root, rpage->page_dict); |
|
147
|
|
|
|
|
|
|
} |
|
148
|
|
|
|
|
|
|
} |
|
149
|
|
|
|
|
|
|
|
|
150
|
54
|
|
|
|
|
|
err = pdfmake_textract_run(interp, content_buf.data, content_buf.len, &opts, result); |
|
151
|
|
|
|
|
|
|
|
|
152
|
54
|
|
|
|
|
|
pdfmake_interp_free(interp); |
|
153
|
54
|
|
|
|
|
|
pdfmake_buf_free(&content_buf); |
|
154
|
|
|
|
|
|
|
|
|
155
|
54
|
50
|
|
|
|
|
if (err != PDFMAKE_OK) { |
|
156
|
0
|
|
|
|
|
|
pdfmake_textract_free(result); |
|
157
|
0
|
|
|
|
|
|
croak("PDF::Make::Extract: extraction failed"); |
|
158
|
|
|
|
|
|
|
} |
|
159
|
|
|
|
|
|
|
|
|
160
|
|
|
|
|
|
|
/* Build Perl data structure: blocks -> lines -> words -> glyphs */ |
|
161
|
54
|
|
|
|
|
|
AV *blocks_av = newAV(); |
|
162
|
292
|
100
|
|
|
|
|
for (size_t b = 0; b < result->len; b++) { |
|
163
|
238
|
|
|
|
|
|
pdfmake_text_block_t *block = &result->blocks[b]; |
|
164
|
238
|
|
|
|
|
|
HV *block_hv = newHV(); |
|
165
|
238
|
|
|
|
|
|
hv_stores(block_hv, "x0", newSVnv(block->x0)); |
|
166
|
238
|
|
|
|
|
|
hv_stores(block_hv, "y0", newSVnv(block->y0)); |
|
167
|
238
|
|
|
|
|
|
hv_stores(block_hv, "x1", newSVnv(block->x1)); |
|
168
|
238
|
|
|
|
|
|
hv_stores(block_hv, "y1", newSVnv(block->y1)); |
|
169
|
|
|
|
|
|
|
|
|
170
|
238
|
|
|
|
|
|
AV *lines_av = newAV(); |
|
171
|
910
|
100
|
|
|
|
|
for (size_t l = 0; l < block->len; l++) { |
|
172
|
672
|
|
|
|
|
|
pdfmake_text_line_t *line = &block->lines[l]; |
|
173
|
672
|
|
|
|
|
|
HV *line_hv = newHV(); |
|
174
|
672
|
|
|
|
|
|
hv_stores(line_hv, "x0", newSVnv(line->x0)); |
|
175
|
672
|
|
|
|
|
|
hv_stores(line_hv, "y0", newSVnv(line->y0)); |
|
176
|
672
|
|
|
|
|
|
hv_stores(line_hv, "x1", newSVnv(line->x1)); |
|
177
|
672
|
|
|
|
|
|
hv_stores(line_hv, "y1", newSVnv(line->y1)); |
|
178
|
672
|
|
|
|
|
|
hv_stores(line_hv, "baseline", newSVnv(line->baseline_y)); |
|
179
|
|
|
|
|
|
|
|
|
180
|
672
|
|
|
|
|
|
AV *words_av = newAV(); |
|
181
|
2763
|
100
|
|
|
|
|
for (size_t w = 0; w < line->len; w++) { |
|
182
|
2091
|
|
|
|
|
|
pdfmake_text_word_t *word = &line->words[w]; |
|
183
|
2091
|
|
|
|
|
|
HV *word_hv = newHV(); |
|
184
|
2091
|
|
|
|
|
|
hv_stores(word_hv, "x0", newSVnv(word->x0)); |
|
185
|
2091
|
|
|
|
|
|
hv_stores(word_hv, "y0", newSVnv(word->y0)); |
|
186
|
2091
|
|
|
|
|
|
hv_stores(word_hv, "x1", newSVnv(word->x1)); |
|
187
|
2091
|
|
|
|
|
|
hv_stores(word_hv, "y1", newSVnv(word->y1)); |
|
188
|
|
|
|
|
|
|
|
|
189
|
|
|
|
|
|
|
/* Build word text from glyphs */ |
|
190
|
|
|
|
|
|
|
pdfmake_buf_t wbuf; |
|
191
|
2091
|
|
|
|
|
|
pdfmake_buf_init(&wbuf); |
|
192
|
2091
|
|
|
|
|
|
double word_font_size = 0; |
|
193
|
14246
|
100
|
|
|
|
|
for (size_t g = 0; g < word->len; g++) { |
|
194
|
12155
|
|
|
|
|
|
pdfmake_text_glyph_t *gl = &word->glyphs[g]; |
|
195
|
12155
|
100
|
|
|
|
|
if (gl->font_size > word_font_size) |
|
196
|
2091
|
|
|
|
|
|
word_font_size = gl->font_size; |
|
197
|
|
|
|
|
|
|
/* Encode Unicode codepoint as UTF-8 */ |
|
198
|
12155
|
|
|
|
|
|
uint32_t cp = gl->unicode; |
|
199
|
12155
|
100
|
|
|
|
|
if (cp < 0x80) { |
|
200
|
12110
|
|
|
|
|
|
pdfmake_buf_append_byte(&wbuf, (uint8_t)cp); |
|
201
|
45
|
50
|
|
|
|
|
} else if (cp < 0x800) { |
|
202
|
0
|
|
|
|
|
|
pdfmake_buf_append_byte(&wbuf, 0xC0 | (cp >> 6)); |
|
203
|
0
|
|
|
|
|
|
pdfmake_buf_append_byte(&wbuf, 0x80 | (cp & 0x3F)); |
|
204
|
45
|
50
|
|
|
|
|
} else if (cp < 0x10000) { |
|
205
|
45
|
|
|
|
|
|
pdfmake_buf_append_byte(&wbuf, 0xE0 | (cp >> 12)); |
|
206
|
45
|
|
|
|
|
|
pdfmake_buf_append_byte(&wbuf, 0x80 | ((cp >> 6) & 0x3F)); |
|
207
|
45
|
|
|
|
|
|
pdfmake_buf_append_byte(&wbuf, 0x80 | (cp & 0x3F)); |
|
208
|
|
|
|
|
|
|
} else { |
|
209
|
0
|
|
|
|
|
|
pdfmake_buf_append_byte(&wbuf, 0xF0 | (cp >> 18)); |
|
210
|
0
|
|
|
|
|
|
pdfmake_buf_append_byte(&wbuf, 0x80 | ((cp >> 12) & 0x3F)); |
|
211
|
0
|
|
|
|
|
|
pdfmake_buf_append_byte(&wbuf, 0x80 | ((cp >> 6) & 0x3F)); |
|
212
|
0
|
|
|
|
|
|
pdfmake_buf_append_byte(&wbuf, 0x80 | (cp & 0x3F)); |
|
213
|
|
|
|
|
|
|
} |
|
214
|
|
|
|
|
|
|
} |
|
215
|
2091
|
|
|
|
|
|
SV *text_sv = newSVpvn((const char *)wbuf.data, wbuf.len); |
|
216
|
2091
|
|
|
|
|
|
SvUTF8_on(text_sv); |
|
217
|
2091
|
|
|
|
|
|
hv_stores(word_hv, "text", text_sv); |
|
218
|
2091
|
|
|
|
|
|
hv_stores(word_hv, "font_size", newSVnv(word_font_size)); |
|
219
|
2091
|
|
|
|
|
|
pdfmake_buf_free(&wbuf); |
|
220
|
|
|
|
|
|
|
|
|
221
|
|
|
|
|
|
|
/* Phase 12: expose marked-content tag for this word */ |
|
222
|
2091
|
100
|
|
|
|
|
if (word->mcid >= 0) { |
|
223
|
5
|
|
|
|
|
|
hv_stores(word_hv, "mcid", newSViv(word->mcid)); |
|
224
|
5
|
|
|
|
|
|
uint32_t role_id = pdfmake_textract_role_for_mcid( |
|
225
|
|
|
|
|
|
|
result, word->mcid); |
|
226
|
5
|
50
|
|
|
|
|
if (role_id) { |
|
227
|
|
|
|
|
|
|
pdfmake_obj_t role_obj; |
|
228
|
5
|
|
|
|
|
|
role_obj.kind = PDFMAKE_NAME; |
|
229
|
5
|
|
|
|
|
|
role_obj.as.name.id = role_id; |
|
230
|
|
|
|
|
|
|
const char *role_name = |
|
231
|
5
|
|
|
|
|
|
pdfmake_get_name_bytes(arena, &role_obj); |
|
232
|
5
|
50
|
|
|
|
|
if (role_name) { |
|
233
|
5
|
|
|
|
|
|
hv_stores(word_hv, "tag", |
|
234
|
|
|
|
|
|
|
newSVpv(role_name, 0)); |
|
235
|
|
|
|
|
|
|
} |
|
236
|
|
|
|
|
|
|
} |
|
237
|
|
|
|
|
|
|
} |
|
238
|
|
|
|
|
|
|
|
|
239
|
2091
|
|
|
|
|
|
av_push(words_av, newRV_noinc((SV *)word_hv)); |
|
240
|
|
|
|
|
|
|
} |
|
241
|
672
|
|
|
|
|
|
hv_stores(line_hv, "words", newRV_noinc((SV *)words_av)); |
|
242
|
672
|
|
|
|
|
|
av_push(lines_av, newRV_noinc((SV *)line_hv)); |
|
243
|
|
|
|
|
|
|
} |
|
244
|
238
|
|
|
|
|
|
hv_stores(block_hv, "lines", newRV_noinc((SV *)lines_av)); |
|
245
|
238
|
|
|
|
|
|
av_push(blocks_av, newRV_noinc((SV *)block_hv)); |
|
246
|
|
|
|
|
|
|
} |
|
247
|
|
|
|
|
|
|
|
|
248
|
54
|
|
|
|
|
|
RETVAL = newRV_noinc((SV *)blocks_av); |
|
249
|
|
|
|
|
|
|
|
|
250
|
54
|
|
|
|
|
|
pdfmake_textract_free(result); |
|
251
|
|
|
|
|
|
|
/* arena owned by reader; do not free here */ |
|
252
|
|
|
|
|
|
|
} |
|
253
|
|
|
|
|
|
|
OUTPUT: |
|
254
|
|
|
|
|
|
|
RETVAL |
|
255
|
|
|
|
|
|
|
|
|
256
|
|
|
|
|
|
|
SV * |
|
257
|
|
|
|
|
|
|
_detect_tables(class, reader_sv, page_index) |
|
258
|
|
|
|
|
|
|
char *class |
|
259
|
|
|
|
|
|
|
pdfmake_reader_xs_t *reader_sv |
|
260
|
|
|
|
|
|
|
UV page_index |
|
261
|
|
|
|
|
|
|
PREINIT: |
|
262
|
|
|
|
|
|
|
pdfmake_err_t err; |
|
263
|
|
|
|
|
|
|
pdfmake_arena_t *arena; |
|
264
|
|
|
|
|
|
|
pdfmake_interp_t *interp; |
|
265
|
|
|
|
|
|
|
pdfmake_textract_result_t *result; |
|
266
|
|
|
|
|
|
|
pdfmake_textract_options_t opts; |
|
267
|
|
|
|
|
|
|
pdfmake_reader_page_t *rpage; |
|
268
|
|
|
|
|
|
|
pdfmake_buf_t content_buf; |
|
269
|
|
|
|
|
|
|
pdfmake_obj_t *resources; |
|
270
|
|
|
|
|
|
|
pdfmake_textract_table_list_t *tlist; |
|
271
|
|
|
|
|
|
|
CODE: |
|
272
|
|
|
|
|
|
|
PERL_UNUSED_VAR(class); |
|
273
|
2
|
|
|
|
|
|
pdfmake_buf_init(&content_buf); |
|
274
|
|
|
|
|
|
|
|
|
275
|
2
|
50
|
|
|
|
|
if (page_index >= pdfmake_reader_page_count(reader_sv->reader)) |
|
276
|
0
|
|
|
|
|
|
croak("PDF::Make::Extract: page index %lu out of range", |
|
277
|
|
|
|
|
|
|
(unsigned long)page_index); |
|
278
|
|
|
|
|
|
|
|
|
279
|
2
|
|
|
|
|
|
rpage = pdfmake_reader_page_at(reader_sv->reader, page_index); |
|
280
|
2
|
|
|
|
|
|
err = pdfmake_reader_page_content_bytes(reader_sv->reader, rpage, &content_buf); |
|
281
|
2
|
50
|
|
|
|
|
if (err != PDFMAKE_OK || content_buf.len == 0) { |
|
|
|
50
|
|
|
|
|
|
|
282
|
0
|
|
|
|
|
|
pdfmake_buf_free(&content_buf); |
|
283
|
0
|
|
|
|
|
|
RETVAL = newRV_noinc((SV *)newAV()); |
|
284
|
|
|
|
|
|
|
} else { |
|
285
|
2
|
|
|
|
|
|
arena = reader_sv->reader->parser->doc->arena; |
|
286
|
2
|
|
|
|
|
|
interp = pdfmake_interp_new(arena); |
|
287
|
2
|
50
|
|
|
|
|
if (!interp) { |
|
288
|
0
|
|
|
|
|
|
pdfmake_buf_free(&content_buf); |
|
289
|
0
|
|
|
|
|
|
croak("PDF::Make::Extract: interp alloc failed"); |
|
290
|
|
|
|
|
|
|
} |
|
291
|
2
|
|
|
|
|
|
pdfmake_interp_set_reader(interp, reader_sv->reader); |
|
292
|
2
|
|
|
|
|
|
resources = pdfmake_reader_page_resources(reader_sv->reader, rpage); |
|
293
|
2
|
50
|
|
|
|
|
if (resources) pdfmake_interp_set_resources(interp, resources); |
|
294
|
|
|
|
|
|
|
|
|
295
|
2
|
|
|
|
|
|
result = pdfmake_textract_new(arena); |
|
296
|
2
|
|
|
|
|
|
pdfmake_textract_set_reader(result, reader_sv->reader); |
|
297
|
2
|
|
|
|
|
|
opts = pdfmake_textract_default_options(); |
|
298
|
2
|
|
|
|
|
|
err = pdfmake_textract_run(interp, content_buf.data, content_buf.len, &opts, result); |
|
299
|
2
|
|
|
|
|
|
pdfmake_interp_free(interp); |
|
300
|
2
|
|
|
|
|
|
pdfmake_buf_free(&content_buf); |
|
301
|
2
|
50
|
|
|
|
|
if (err != PDFMAKE_OK) { |
|
302
|
0
|
|
|
|
|
|
pdfmake_textract_free(result); |
|
303
|
0
|
|
|
|
|
|
croak("PDF::Make::Extract: extraction failed"); |
|
304
|
|
|
|
|
|
|
} |
|
305
|
|
|
|
|
|
|
|
|
306
|
2
|
|
|
|
|
|
tlist = pdfmake_textract_table_list_new(arena); |
|
307
|
2
|
|
|
|
|
|
pdfmake_textract_detect_tables(result, NULL, tlist); |
|
308
|
|
|
|
|
|
|
|
|
309
|
2
|
|
|
|
|
|
AV *av = newAV(); |
|
310
|
3
|
100
|
|
|
|
|
for (size_t ti = 0; ti < tlist->len; ti++) { |
|
311
|
1
|
|
|
|
|
|
pdfmake_textract_table_t *t = &tlist->items[ti]; |
|
312
|
1
|
|
|
|
|
|
HV *hv = newHV(); |
|
313
|
1
|
|
|
|
|
|
hv_stores(hv, "x0", newSVnv(t->x0)); |
|
314
|
1
|
|
|
|
|
|
hv_stores(hv, "y0", newSVnv(t->y0)); |
|
315
|
1
|
|
|
|
|
|
hv_stores(hv, "x1", newSVnv(t->x1)); |
|
316
|
1
|
|
|
|
|
|
hv_stores(hv, "y1", newSVnv(t->y1)); |
|
317
|
1
|
|
|
|
|
|
hv_stores(hv, "rows", newSVuv((UV)t->rows)); |
|
318
|
1
|
|
|
|
|
|
hv_stores(hv, "cols", newSVuv((UV)t->cols)); |
|
319
|
|
|
|
|
|
|
|
|
320
|
1
|
|
|
|
|
|
AV *rows_av = newAV(); |
|
321
|
5
|
100
|
|
|
|
|
for (size_t r = 0; r < t->rows; r++) { |
|
322
|
4
|
|
|
|
|
|
AV *row_av = newAV(); |
|
323
|
16
|
100
|
|
|
|
|
for (size_t c = 0; c < t->cols; c++) { |
|
324
|
12
|
|
|
|
|
|
size_t idx = r * t->cols + c; |
|
325
|
12
|
50
|
|
|
|
|
const char *txt = t->cells[idx] ? t->cells[idx] : ""; |
|
326
|
12
|
|
|
|
|
|
SV *sv = newSVpv(txt, 0); |
|
327
|
12
|
|
|
|
|
|
SvUTF8_on(sv); |
|
328
|
12
|
|
|
|
|
|
av_push(row_av, sv); |
|
329
|
|
|
|
|
|
|
} |
|
330
|
4
|
|
|
|
|
|
av_push(rows_av, newRV_noinc((SV *)row_av)); |
|
331
|
|
|
|
|
|
|
} |
|
332
|
1
|
|
|
|
|
|
hv_stores(hv, "cells", newRV_noinc((SV *)rows_av)); |
|
333
|
|
|
|
|
|
|
|
|
334
|
1
|
|
|
|
|
|
av_push(av, newRV_noinc((SV *)hv)); |
|
335
|
|
|
|
|
|
|
} |
|
336
|
|
|
|
|
|
|
|
|
337
|
2
|
|
|
|
|
|
pdfmake_textract_table_list_free(tlist); |
|
338
|
2
|
|
|
|
|
|
pdfmake_textract_free(result); |
|
339
|
2
|
|
|
|
|
|
RETVAL = newRV_noinc((SV *)av); |
|
340
|
|
|
|
|
|
|
} |
|
341
|
|
|
|
|
|
|
OUTPUT: |
|
342
|
|
|
|
|
|
|
RETVAL |
|
343
|
|
|
|
|
|
|
|
|
344
|
|
|
|
|
|
|
SV * |
|
345
|
|
|
|
|
|
|
_extract_annotations(class, reader_sv) |
|
346
|
|
|
|
|
|
|
char *class |
|
347
|
|
|
|
|
|
|
pdfmake_reader_xs_t *reader_sv |
|
348
|
|
|
|
|
|
|
PREINIT: |
|
349
|
|
|
|
|
|
|
pdfmake_annot_text_list_t *list; |
|
350
|
|
|
|
|
|
|
pdfmake_arena_t *arena; |
|
351
|
|
|
|
|
|
|
CODE: |
|
352
|
|
|
|
|
|
|
PERL_UNUSED_VAR(class); |
|
353
|
2
|
|
|
|
|
|
arena = reader_sv->reader->parser->doc->arena; |
|
354
|
2
|
|
|
|
|
|
list = pdfmake_annot_text_list_new(arena); |
|
355
|
2
|
50
|
|
|
|
|
if (!list) croak("PDF::Make::Extract: annot list alloc failed"); |
|
356
|
|
|
|
|
|
|
|
|
357
|
2
|
50
|
|
|
|
|
if (pdfmake_textract_annotations(reader_sv->reader, list) != PDFMAKE_OK) { |
|
358
|
0
|
|
|
|
|
|
pdfmake_annot_text_list_free(list); |
|
359
|
0
|
|
|
|
|
|
croak("PDF::Make::Extract: annotation extraction failed"); |
|
360
|
|
|
|
|
|
|
} |
|
361
|
|
|
|
|
|
|
|
|
362
|
2
|
|
|
|
|
|
AV *av = newAV(); |
|
363
|
5
|
100
|
|
|
|
|
for (size_t i = 0; i < list->len; i++) { |
|
364
|
3
|
|
|
|
|
|
pdfmake_annot_text_t *r = &list->items[i]; |
|
365
|
3
|
|
|
|
|
|
HV *hv = newHV(); |
|
366
|
3
|
50
|
|
|
|
|
if (r->kind) hv_stores(hv, "kind", newSVpv(r->kind, 0)); |
|
367
|
3
|
50
|
|
|
|
|
if (r->page_index != (size_t)-1) |
|
368
|
3
|
|
|
|
|
|
hv_stores(hv, "page", newSVuv((UV)r->page_index)); |
|
369
|
3
|
|
|
|
|
|
AV *rect_av = newAV(); |
|
370
|
15
|
100
|
|
|
|
|
for (int k = 0; k < 4; k++) |
|
371
|
12
|
|
|
|
|
|
av_push(rect_av, newSVnv(r->rect[k])); |
|
372
|
3
|
|
|
|
|
|
hv_stores(hv, "rect", newRV_noinc((SV *)rect_av)); |
|
373
|
|
|
|
|
|
|
|
|
374
|
3
|
50
|
|
|
|
|
SV *tsv = newSVpv(r->text ? r->text : "", 0); |
|
375
|
3
|
|
|
|
|
|
SvUTF8_on(tsv); |
|
376
|
3
|
|
|
|
|
|
hv_stores(hv, "text", tsv); |
|
377
|
|
|
|
|
|
|
|
|
378
|
3
|
100
|
|
|
|
|
if (r->author) { |
|
379
|
2
|
|
|
|
|
|
SV *sv = newSVpv(r->author, 0); |
|
380
|
2
|
|
|
|
|
|
SvUTF8_on(sv); |
|
381
|
2
|
|
|
|
|
|
hv_stores(hv, "author", sv); |
|
382
|
|
|
|
|
|
|
} |
|
383
|
3
|
100
|
|
|
|
|
if (r->subject) { |
|
384
|
2
|
|
|
|
|
|
SV *sv = newSVpv(r->subject, 0); |
|
385
|
2
|
|
|
|
|
|
SvUTF8_on(sv); |
|
386
|
2
|
|
|
|
|
|
hv_stores(hv, "subject", sv); |
|
387
|
|
|
|
|
|
|
} |
|
388
|
3
|
100
|
|
|
|
|
if (r->field_name) { |
|
389
|
1
|
|
|
|
|
|
SV *sv = newSVpv(r->field_name, 0); |
|
390
|
1
|
|
|
|
|
|
SvUTF8_on(sv); |
|
391
|
1
|
|
|
|
|
|
hv_stores(hv, "field_name", sv); |
|
392
|
|
|
|
|
|
|
} |
|
393
|
|
|
|
|
|
|
|
|
394
|
3
|
|
|
|
|
|
av_push(av, newRV_noinc((SV *)hv)); |
|
395
|
|
|
|
|
|
|
} |
|
396
|
|
|
|
|
|
|
|
|
397
|
2
|
|
|
|
|
|
pdfmake_annot_text_list_free(list); |
|
398
|
2
|
|
|
|
|
|
RETVAL = newRV_noinc((SV *)av); |
|
399
|
|
|
|
|
|
|
OUTPUT: |
|
400
|
|
|
|
|
|
|
RETVAL |