基于 pdfextact.c 的解决方案看起来像蛮力,但它有效:
- 遍历所有 pdf 对象 (
pdf_load_object
)
- 确定对象是否为嵌入文件 (
isembed
)
- 如果是 - 访问它的流并保存文件(
saveembed
)
在大多数测试用例中,嵌入文件存储在文件末尾,因此,reverce 迭代是有意义的。
static int isembed(pdf_obj *obj) {
pdf_obj *type = pdf_dict_gets(obj, "Type");
return pdf_is_name(type) && !strcmp(pdf_to_name(type), "Filespec");
}
static void saveembed(pdf_obj *dict) {
char *filename;
pdf_obj *obj = pdf_dict_gets(dict, "F");
if (obj) filename = pdf_to_str_buf(obj);
obj = pdf_dict_gets(dict, "EF");
if (!obj) return;
pdf_obj *stream = pdf_dict_gets(obj, "F");
if (!stream) return;
FILE *f;
fz_buffer *buf;
int n, len;
unsigned char *data;
buf = pdf_load_stream(doc, pdf_to_num(stream), pdf_to_gen(stream));
printf("extracting embedded file %s\n", filename);
f = fopen(filename, "wb");
len = fz_buffer_storage(ctx, buf, &data);
n = fwrite(data, 1, len, f);
fclose(f);
fz_drop_buffer(ctx, buf);
}