test-message-parser.c revision 7d800c558b4a40f11748d4ebfc2bf89eb743f03c
7cb128dc4cae2a03a742f63ba7afee23c78e3af0Phil Carmody/* Copyright (c) 2007-2015 Dovecot authors, see the included COPYING file */
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen#include "lib.h"
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen#include "istream.h"
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen#include "message-parser.h"
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen#include "test-common.h"
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainenstatic const char test_msg[] =
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen"Return-Path: <test@example.org>\n"
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen"Subject: Hello world\n"
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen"From: Test User <test@example.org>\n"
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen"To: Another User <test2@example.org>\n"
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen"Message-Id: <1.2.3.4@example>\n"
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen"Mime-Version: 1.0\n"
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen"Date: Sun, 23 May 2007 04:58:08 +0300\n"
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen"Content-Type: multipart/signed; micalg=pgp-sha1;\n"
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen" protocol=\"application/pgp-signature\";\n"
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen" boundary=\"=-GNQXLhuj24Pl1aCkk4/d\"\n"
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen"\n"
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen"--=-GNQXLhuj24Pl1aCkk4/d\n"
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen"Content-Type: text/plain\n"
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen"Content-Transfer-Encoding: quoted-printable\n"
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen"\n"
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen"There was a day=20\n"
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen"a happy=20day\n"
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen"\n"
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen"--=-GNQXLhuj24Pl1aCkk4/d\n"
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen"Content-Type: application/pgp-signature; name=signature.asc\n"
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen"\n"
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen"-----BEGIN PGP SIGNATURE-----\n"
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen"Version: GnuPG v1.2.4 (GNU/Linux)\n"
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen"\n"
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen"invalid\n"
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen"-----END PGP SIGNATURE-----\n"
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen"\n"
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen"--=-GNQXLhuj24Pl1aCkk4/d--\n"
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen"\n"
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen"\n";
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen#define TEST_MSG_LEN (sizeof(test_msg)-1)
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainenstatic bool msg_parts_cmp(struct message_part *p1, struct message_part *p2)
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen{
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen while (p1 != NULL || p2 != NULL) {
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen if ((p1 != NULL) != (p2 != NULL))
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen return FALSE;
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen if ((p1->children != NULL) != (p2->children != NULL))
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen return FALSE;
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen if (p1->children) {
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen if (!msg_parts_cmp(p1->children, p2->children))
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen return FALSE;
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen }
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen if (p1->physical_pos != p2->physical_pos ||
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen p1->header_size.physical_size != p2->header_size.physical_size ||
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen p1->header_size.virtual_size != p2->header_size.virtual_size ||
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen p1->header_size.lines != p2->header_size.lines ||
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen p1->body_size.physical_size != p2->body_size.physical_size ||
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen p1->body_size.virtual_size != p2->body_size.virtual_size ||
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen p1->body_size.lines != p2->body_size.lines ||
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen p1->flags != p2->flags)
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen return FALSE;
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen p1 = p1->next;
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen p2 = p2->next;
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen }
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen return TRUE;
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen}
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen
11d72d764c104b4f1a8c44ec8a5fee420517645bTimo Sirainenstatic void test_message_parser_small_blocks(void)
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen{
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen struct message_parser_ctx *parser;
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen struct istream *input;
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen struct message_part *parts, *parts2;
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen struct message_block block;
11d72d764c104b4f1a8c44ec8a5fee420517645bTimo Sirainen unsigned int i, end_of_headers_idx;
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen pool_t pool;
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen int ret;
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen
11d72d764c104b4f1a8c44ec8a5fee420517645bTimo Sirainen test_begin("message parser in small blocks");
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen pool = pool_alloconly_create("message parser", 10240);
11d72d764c104b4f1a8c44ec8a5fee420517645bTimo Sirainen input = test_istream_create(test_msg);
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen
11d72d764c104b4f1a8c44ec8a5fee420517645bTimo Sirainen /* full parsing */
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen parser = message_parser_init(pool, input, 0, 0);
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen while ((ret = message_parser_parse_next_block(parser, &block)) > 0) ;
11d72d764c104b4f1a8c44ec8a5fee420517645bTimo Sirainen test_assert(ret < 0);
11d72d764c104b4f1a8c44ec8a5fee420517645bTimo Sirainen test_assert(message_parser_deinit(&parser, &parts) == 0);
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen
11d72d764c104b4f1a8c44ec8a5fee420517645bTimo Sirainen /* parsing in small blocks */
11d72d764c104b4f1a8c44ec8a5fee420517645bTimo Sirainen i_stream_seek(input, 0);
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen test_istream_set_allow_eof(input, FALSE);
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen parser = message_parser_init(pool, input, 0, 0);
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen for (i = 1; i <= TEST_MSG_LEN*2+1; i++) {
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen test_istream_set_size(input, i/2);
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen if (i > TEST_MSG_LEN*2)
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen test_istream_set_allow_eof(input, TRUE);
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen while ((ret = message_parser_parse_next_block(parser,
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen &block)) > 0) ;
5f237cf1915127fe143ac4c1d3cb4d1139f3b97dTimo Sirainen test_assert((ret == 0 && i <= TEST_MSG_LEN*2) ||
5f237cf1915127fe143ac4c1d3cb4d1139f3b97dTimo Sirainen (ret < 0 && i > TEST_MSG_LEN*2));
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen }
11d72d764c104b4f1a8c44ec8a5fee420517645bTimo Sirainen test_assert(message_parser_deinit(&parser, &parts2) == 0);
11d72d764c104b4f1a8c44ec8a5fee420517645bTimo Sirainen test_assert(msg_parts_cmp(parts, parts2));
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen
11d72d764c104b4f1a8c44ec8a5fee420517645bTimo Sirainen /* parsing in small blocks from preparsed parts */
11d72d764c104b4f1a8c44ec8a5fee420517645bTimo Sirainen i_stream_seek(input, 0);
11d72d764c104b4f1a8c44ec8a5fee420517645bTimo Sirainen test_istream_set_allow_eof(input, FALSE);
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen
5f237cf1915127fe143ac4c1d3cb4d1139f3b97dTimo Sirainen end_of_headers_idx = (strstr(test_msg, "\n-----") - test_msg);
11d72d764c104b4f1a8c44ec8a5fee420517645bTimo Sirainen parser = message_parser_init_from_parts(parts, input, 0,
11d72d764c104b4f1a8c44ec8a5fee420517645bTimo Sirainen MESSAGE_PARSER_FLAG_SKIP_BODY_BLOCK);
11d72d764c104b4f1a8c44ec8a5fee420517645bTimo Sirainen for (i = 1; i <= TEST_MSG_LEN*2+1; i++) {
11d72d764c104b4f1a8c44ec8a5fee420517645bTimo Sirainen test_istream_set_size(input, i/2);
11d72d764c104b4f1a8c44ec8a5fee420517645bTimo Sirainen if (i > TEST_MSG_LEN*2)
11d72d764c104b4f1a8c44ec8a5fee420517645bTimo Sirainen test_istream_set_allow_eof(input, TRUE);
11d72d764c104b4f1a8c44ec8a5fee420517645bTimo Sirainen while ((ret = message_parser_parse_next_block(parser,
11d72d764c104b4f1a8c44ec8a5fee420517645bTimo Sirainen &block)) > 0) ;
5f237cf1915127fe143ac4c1d3cb4d1139f3b97dTimo Sirainen test_assert((ret == 0 && i/2 <= end_of_headers_idx) ||
5f237cf1915127fe143ac4c1d3cb4d1139f3b97dTimo Sirainen (ret < 0 && i/2 > end_of_headers_idx));
11d72d764c104b4f1a8c44ec8a5fee420517645bTimo Sirainen }
11d72d764c104b4f1a8c44ec8a5fee420517645bTimo Sirainen test_assert(message_parser_deinit(&parser, &parts2) == 0);
11d72d764c104b4f1a8c44ec8a5fee420517645bTimo Sirainen test_assert(msg_parts_cmp(parts, parts2));
11d72d764c104b4f1a8c44ec8a5fee420517645bTimo Sirainen
11d72d764c104b4f1a8c44ec8a5fee420517645bTimo Sirainen i_stream_unref(&input);
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen pool_unref(&pool);
11d72d764c104b4f1a8c44ec8a5fee420517645bTimo Sirainen test_end();
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen}
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainenstatic void test_message_parser_truncated_mime_headers(void)
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen{
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainenstatic const char input_msg[] =
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen"Content-Type: multipart/mixed; boundary=\":foo\"\n"
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen"\n"
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen"--:foo\n"
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen"--:foo\n"
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen"Content-Type: text/plain\n"
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen"--:foo\n"
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen"Content-Type: text/plain\r\n"
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen"--:foo\n"
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen"Content-Type: text/html\n"
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen"--:foo--\n";
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen struct message_parser_ctx *parser;
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen struct istream *input;
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen struct message_part *parts, *part;
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen struct message_block block;
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen pool_t pool;
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen int ret;
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen test_begin("message parser truncated mime headers");
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen pool = pool_alloconly_create("message parser", 10240);
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen input = test_istream_create(input_msg);
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen parser = message_parser_init(pool, input, 0, 0);
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen while ((ret = message_parser_parse_next_block(parser, &block)) > 0) ;
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen test_assert(ret < 0);
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen test_assert(message_parser_deinit(&parser, &parts) == 0);
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen test_assert((parts->flags & MESSAGE_PART_FLAG_MULTIPART) != 0);
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen test_assert(parts->children->header_size.physical_size == 0);
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen test_assert(parts->children->body_size.physical_size == 0);
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen test_assert(parts->children->body_size.lines == 0);
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen test_assert(parts->children->next->header_size.physical_size == 24);
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen test_assert(parts->children->next->header_size.virtual_size == 24);
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen test_assert(parts->children->next->header_size.lines == 0);
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen test_assert(parts->children->next->next->header_size.physical_size == 24);
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen test_assert(parts->children->next->next->header_size.virtual_size == 24);
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen test_assert(parts->children->next->next->header_size.lines == 0);
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen test_assert(parts->children->next->next->next->header_size.physical_size == 23);
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen test_assert(parts->children->next->next->next->header_size.virtual_size == 23);
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen test_assert(parts->children->next->next->next->header_size.lines == 0);
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen for (part = parts->children; part != NULL; part = part->next) {
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen test_assert(part->body_size.physical_size == 0);
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen test_assert(part->body_size.virtual_size == 0);
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen }
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen test_assert(parts->children->next->next->next->next == NULL);
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen i_stream_unref(&input);
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen pool_unref(&pool);
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen test_end();
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen}
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainenint main(void)
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen{
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen static void (*test_functions[])(void) = {
11d72d764c104b4f1a8c44ec8a5fee420517645bTimo Sirainen test_message_parser_small_blocks,
7d800c558b4a40f11748d4ebfc2bf89eb743f03cTimo Sirainen test_message_parser_truncated_mime_headers,
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen NULL
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen };
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen return test_run(test_functions);
39025a2dabfcfaeee3790988b9ea00d19887a3d3Timo Sirainen}