Files
recover-pdfs/recover-pdfs.c
Seán Healy 38dfc87459 More exact PDF pattern and filesize limit
Also, can now feed image in through stdin (useful for `pv image.iso |
./revocer-pdfs ...` workflow
2026-04-11 16:19:27 +01:00

134 lines
3.0 KiB
C

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <sys/stat.h>
#define CHUNK 4096
#define MAX_PDF_SIZE (100LL * 1024 * 1024) // 100 MB
int valid_pdf_header(const unsigned char *buf, size_t len) {
if (len < 9) return 0;
if (memcmp(buf, "%PDF-", 5) != 0)
return 0;
if (buf[5] < '0' || buf[5] > '9') return 0;
if (buf[6] != '.') return 0;
if (buf[7] < '0' || buf[7] > '9') return 0;
size_t i = 8;
if (buf[i] >= '0' && buf[i] <= '9') i++;
if (buf[i] == '\r' && buf[i + 1] == '\n') return 1;
if (buf[i] == '\n') return 1;
return 0;
}
size_t find_eof(const unsigned char *buf, size_t len) {
for (size_t i = 0; i + 5 <= len; i++) {
if (memcmp(buf + i, "%%EOF", 5) == 0)
return i;
}
return (size_t)-1;
}
void write_pdf(const unsigned char *data, size_t len, int idx, const char *outdir) {
char path[512];
snprintf(path, sizeof(path), "%s/pdf_%04d.pdf", outdir, idx);
FILE *f = fopen(path, "wb");
if (!f) return;
fwrite(data, 1, len, f);
fclose(f);
}
int main(int argc, char **argv) {
if (argc != 3) {
fprintf(stderr, "Usage: %s <image_file> <output_dir>\n", argv[0]);
return 1;
}
const char *infile = argv[1];
const char *outdir = argv[2];
mkdir(outdir, 0755);
if (strcmp(infile, "-") == 0) {
infile = "/dev/stdin";
}
FILE *f = fopen(infile, "rb");
if (!f) {
perror("open input");
return 1;
}
unsigned char *buf = malloc(CHUNK * 2);
size_t buf_len = 0;
unsigned char *pdf_buf = NULL;
size_t pdf_size = 0;
int collecting = 0;
int pdf_count = 0;
while (!feof(f)) {
size_t r = fread(buf + buf_len, 1, CHUNK, f);
buf_len += r;
size_t i = 0;
while (i < buf_len) {
if (!collecting) {
if (i + 10 < buf_len && valid_pdf_header(buf + i, buf_len - i)) {
collecting = 1;
pdf_buf = malloc(1024);
pdf_size = 0;
}
}
if (collecting) {
// HARD SIZE LIMIT CHECK
if (pdf_size >= MAX_PDF_SIZE) {
free(pdf_buf);
pdf_buf = NULL;
pdf_size = 0;
collecting = 0;
continue;
}
if (pdf_size % 1024 == 0) {
pdf_buf = realloc(pdf_buf, pdf_size + 1024);
}
pdf_buf[pdf_size++] = buf[i];
size_t eof_pos = find_eof(pdf_buf, pdf_size);
if (eof_pos != (size_t)-1) {
size_t end = eof_pos + 5;
write_pdf(pdf_buf, end, pdf_count++, outdir);
free(pdf_buf);
pdf_buf = NULL;
pdf_size = 0;
collecting = 0;
}
}
i++;
}
buf_len = 0;
}
free(buf);
fclose(f);
return 0;
}