get id string and xml

1 files changed, 132 insertions, 62 deletions

diff --git a/source/pdf/pdf-crypt.c b/source/pdf/pdf-crypt.c
index 53dc73b5..a8a1bdb5 100644
--- a/source/pdf/pdf-crypt.c
+++ b/source/pdf/pdf-crypt.c
@@ -68,18 +68,8 @@ pdf_new_crypt(fz_context *ctx, pdf_obj *dict, pdf_obj *id)
 	crypt->xjc = 0;
 	if (!strcmp("ProfXJC", pdf_to_name(ctx, obj)))
 	{
-		unsigned char akey[] = "\xCE\x69\x16\xAC\xD3\xA5\x00\x85\xF2\xD0\x22\x2A\x9C\xCE\x64\x4F\x65\xDF\x47\xE4\xE3\x6C\x24\x2B\x1B\x31\x6A\x3D\x3B\xE2\x20\xCC";
-
-		crypt->encrypt_metadata = 0;
 		crypt->xjc = 1;
-		crypt->length = 256;
-		memcpy(crypt->key, akey, 32);
-		crypt->stmf.method = PDF_CRYPT_AESV3;
-		crypt->stmf.length = crypt->length;
-		crypt->strf.method = PDF_CRYPT_AESV3;
-		crypt->strf.length = crypt->length;
-		return crypt;
-	}
+	} else
 	if (!pdf_name_eq(ctx, PDF_NAME(Standard), obj))
 	{
 		pdf_drop_crypt(ctx, crypt);
@@ -90,7 +80,7 @@ pdf_new_crypt(fz_context *ctx, pdf_obj *dict, pdf_obj *id)
 	obj = pdf_dict_get(ctx, dict, PDF_NAME(V));
 	if (pdf_is_int(ctx, obj))
 		crypt->v = pdf_to_int(ctx, obj);
-	if (crypt->v != 1 && crypt->v != 2 && crypt->v != 4 && crypt->v != 5)
+	if (!crypt->xjc && crypt->v != 1 && crypt->v != 2 && crypt->v != 4 && crypt->v != 5)
 	{
 		pdf_drop_crypt(ctx, crypt);
 		fz_throw(ctx, FZ_ERROR_GENERIC, "unknown encryption version");
@@ -98,58 +88,60 @@ pdf_new_crypt(fz_context *ctx, pdf_obj *dict, pdf_obj *id)
 
 	/* Standard security handler (PDF 1.7 table 3.19) */
 
-	obj = pdf_dict_get(ctx, dict, PDF_NAME(R));
-	if (pdf_is_int(ctx, obj))
-		crypt->r = pdf_to_int(ctx, obj);
-	else if (crypt->v <= 4)
-	{
-		fz_warn(ctx, "encryption dictionary missing revision value, guessing...");
-		if (crypt->v < 2)
-			crypt->r = 2;
-		else if (crypt->v == 2)
-			crypt->r = 3;
-		else if (crypt->v == 4)
-			crypt->r = 4;
-	}
-	else
-	{
-		pdf_drop_crypt(ctx, crypt);
-		fz_throw(ctx, FZ_ERROR_GENERIC, "encryption dictionary missing version and revision value");
-	}
-	if (crypt->r < 1 || crypt->r > 6)
-	{
-		int r = crypt->r;
-		pdf_drop_crypt(ctx, crypt);
-		fz_throw(ctx, FZ_ERROR_GENERIC, "unknown crypt revision %d", r);
-	}
+	if (!crypt->xjc) {
+		obj = pdf_dict_get(ctx, dict, PDF_NAME(R));
+		if (pdf_is_int(ctx, obj))
+			crypt->r = pdf_to_int(ctx, obj);
+		else if (crypt->v <= 4)
+		{
+			fz_warn(ctx, "encryption dictionary missing revision value, guessing...");
+			if (crypt->v < 2)
+				crypt->r = 2;
+			else if (crypt->v == 2)
+				crypt->r = 3;
+			else if (crypt->v == 4)
+				crypt->r = 4;
+		}
+		else
+		{
+			pdf_drop_crypt(ctx, crypt);
+			fz_throw(ctx, FZ_ERROR_GENERIC, "encryption dictionary missing version and revision value");
+		}
+		if (crypt->r < 1 || crypt->r > 6)
+		{
+			int r = crypt->r;
+			pdf_drop_crypt(ctx, crypt);
+			fz_throw(ctx, FZ_ERROR_GENERIC, "unknown crypt revision %d", r);
+		}
 
-	obj = pdf_dict_get(ctx, dict, PDF_NAME(O));
-	if (pdf_is_string(ctx, obj) && pdf_to_str_len(ctx, obj) == 32)
-		memcpy(crypt->o, pdf_to_str_buf(ctx, obj), 32);
-	/* /O and /U are supposed to be 48 bytes long for revision 5 and 6, they're often longer, though */
-	else if (crypt->r >= 5 && pdf_is_string(ctx, obj) && pdf_to_str_len(ctx, obj) >= 48)
-		memcpy(crypt->o, pdf_to_str_buf(ctx, obj), 48);
-	else
-	{
-		pdf_drop_crypt(ctx, crypt);
-		fz_throw(ctx, FZ_ERROR_GENERIC, "encryption dictionary missing owner password");
-	}
+		obj = pdf_dict_get(ctx, dict, PDF_NAME(O));
+		if (pdf_is_string(ctx, obj) && pdf_to_str_len(ctx, obj) == 32)
+			memcpy(crypt->o, pdf_to_str_buf(ctx, obj), 32);
+		/* /O and /U are supposed to be 48 bytes long for revision 5 and 6, they're often longer, though */
+		else if (crypt->r >= 5 && pdf_is_string(ctx, obj) && pdf_to_str_len(ctx, obj) >= 48)
+			memcpy(crypt->o, pdf_to_str_buf(ctx, obj), 48);
+		else
+		{
+			pdf_drop_crypt(ctx, crypt);
+			fz_throw(ctx, FZ_ERROR_GENERIC, "encryption dictionary missing owner password");
+		}
 
-	obj = pdf_dict_get(ctx, dict, PDF_NAME(U));
-	if (pdf_is_string(ctx, obj) && pdf_to_str_len(ctx, obj) == 32)
-		memcpy(crypt->u, pdf_to_str_buf(ctx, obj), 32);
-	/* /O and /U are supposed to be 48 bytes long for revision 5 and 6, they're often longer, though */
-	else if (crypt->r >= 5 && pdf_is_string(ctx, obj) && pdf_to_str_len(ctx, obj) >= 48)
-		memcpy(crypt->u, pdf_to_str_buf(ctx, obj), 48);
-	else if (pdf_is_string(ctx, obj) && pdf_to_str_len(ctx, obj) < 32)
-	{
-		fz_warn(ctx, "encryption password key too short (%d)", pdf_to_str_len(ctx, obj));
-		memcpy(crypt->u, pdf_to_str_buf(ctx, obj), pdf_to_str_len(ctx, obj));
-	}
-	else
-	{
-		pdf_drop_crypt(ctx, crypt);
-		fz_throw(ctx, FZ_ERROR_GENERIC, "encryption dictionary missing user password");
+		obj = pdf_dict_get(ctx, dict, PDF_NAME(U));
+		if (pdf_is_string(ctx, obj) && pdf_to_str_len(ctx, obj) == 32)
+			memcpy(crypt->u, pdf_to_str_buf(ctx, obj), 32);
+		/* /O and /U are supposed to be 48 bytes long for revision 5 and 6, they're often longer, though */
+		else if (crypt->r >= 5 && pdf_is_string(ctx, obj) && pdf_to_str_len(ctx, obj) >= 48)
+			memcpy(crypt->u, pdf_to_str_buf(ctx, obj), 48);
+		else if (pdf_is_string(ctx, obj) && pdf_to_str_len(ctx, obj) < 32)
+		{
+			fz_warn(ctx, "encryption password key too short (%d)", pdf_to_str_len(ctx, obj));
+			memcpy(crypt->u, pdf_to_str_buf(ctx, obj), pdf_to_str_len(ctx, obj));
+		}
+		else
+		{
+			pdf_drop_crypt(ctx, crypt);
+			fz_throw(ctx, FZ_ERROR_GENERIC, "encryption dictionary missing user password");
+		}
 	}
 
 	obj = pdf_dict_get(ctx, dict, PDF_NAME(P));
@@ -272,6 +264,84 @@ pdf_new_crypt(fz_context *ctx, pdf_obj *dict, pdf_obj *id)
 			crypt->length = crypt->stmf.length;
 	}
 
+	if (crypt->xjc) {
+#if 0
+		unsigned char akey[] = "\xCE\x69\x16\xAC\xD3\xA5\x00\x85\xF2\xD0\x22\x2A\x9C\xCE\x64\x4F\x65\xDF\x47\xE4\xE3\x6C\x24\x2B\x1B\x31\x6A\x3D\x3B\xE2\x20\xCC";
+#endif
+		fz_sha256 sha256;
+		fz_sha256 xml_hash;
+		unsigned char xml_hash_sum[32];
+		unsigned char *r_buf = NULL;
+		size_t r_size = 0;
+
+		// crypt->encrypt_metadata = 0;
+		obj = pdf_dict_get(ctx, dict, PDF_NAME(Length));
+		if (pdf_is_int(ctx, obj))
+			crypt->length = pdf_to_int(ctx, obj);
+
+		crypt->stmf.method = PDF_CRYPT_AESV3;
+		crypt->stmf.length = crypt->length;
+		crypt->strf.method = PDF_CRYPT_AESV3;
+		crypt->strf.length = crypt->length;
+
+		fz_sha256_update(&sha256, "ProfXJC.General", 15);
+		fz_warn(ctx, "crypt method: %d, crypt length: %d, crypt perm: %x",
+				crypt->stmf.method, crypt->length, crypt->p);
+		fz_sha256_update(&sha256, &crypt->stmf.method, 4);
+		fz_sha256_update(&sha256, &crypt->length, 4);
+		fz_sha256_update(&sha256, &crypt->p, 4);
+
+		obj = pdf_dict_gets(ctx, dict, "Recipients");
+		if (pdf_is_string(ctx, obj)) {
+			unsigned char idstr[32];
+			int l = pdf_to_str_len(ctx, obj);
+
+			if (pdf_is_array(ctx, id) && pdf_array_len(ctx, id) == 2)
+			{
+				crypt->id = id;
+			}
+			if (pdf_is_array(ctx, crypt->id)) {
+				pdf_obj *oid;
+				size_t osz;
+				const char *buf;
+
+				oid = pdf_array_get(ctx, crypt->id, 0);
+				buf = pdf_to_string(ctx, oid, &osz);
+				if (osz == 16) {
+					memcpy(idstr, buf, 16);
+				} else {
+					fz_warn(ctx, "length of id[0] is %ld", osz);
+				}
+				oid = pdf_array_get(ctx, crypt->id, 1);
+				buf = pdf_to_string(ctx, oid, &osz);
+				if (osz == 16) {
+					memcpy(idstr + 16, buf, 16);
+				} else {
+					fz_warn(ctx, "length of id[1] is %ld", osz);
+				}
+			}
+			r_buf = fz_malloc(ctx, l);
+			memcpy(r_buf, pdf_to_str_buf(ctx, obj), l);
+			unsigned char tmpbyte = 0xff;
+			int ididx = 0;
+			for (int i = 0; i < l; i++) {
+				tmpbyte ^= idstr[ididx];
+				ididx = (ididx + 1) % 32;
+				r_buf[i] ^= tmpbyte;
+				// printf("%c", r_buf[i]);
+			}
+		}
+		/* hash some weird thing */
+		/* hash an XML */
+		fz_sha256_init(&xml_hash);
+		/* hash an XML */
+		fz_sha256_final(&xml_hash, xml_hash_sum);
+		fz_sha256_update(&sha256, xml_hash_sum, 32);
+		fz_sha256_final(&sha256, crypt->key);
+
+		// memcpy(crypt->key, akey, 32); /* TODO: kill this */
+		return crypt;
+	}
 	return crypt;
 }