mail-search-args-simplify.c revision 9432496e6b190b839622a391a15788c41e5d95d3
76b43e4417bab52e913da39b5f5bc2a130d3f149Timo Sirainen/* Copyright (c) 2002-2016 Dovecot authors, see the included COPYING file */
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen /* arg mask => prev_arg */
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen HASH_TABLE(struct mail_search_simplify_prev_arg *,
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen struct mail_search_simplify_prev_arg *) prev_args;
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainenmail_search_simplify_prev_arg_cmp(const struct mail_search_simplify_prev_arg *arg1,
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen const struct mail_search_simplify_prev_arg *arg2)
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen ret = memcmp(&arg1->bin_mask, &arg2->bin_mask, sizeof(arg1->bin_mask));
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen ret = null_strcmp(arg1->hdr_field_name_mask, arg2->hdr_field_name_mask);
d5cebe7f98e63d4e2822863ef2faa4971e8b3a5dTimo Sirainen ret = null_strcmp(arg1->str_mask, arg2->str_mask);
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainenstatic unsigned int
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainenmail_search_simplify_prev_arg_hash(const struct mail_search_simplify_prev_arg *arg)
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen unsigned int hash;
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen hash = mem_hash(&arg->bin_mask, sizeof(arg->bin_mask));
7569ab8537418b7fc369265f26595b0ef9e4cb35Timo Sirainenstatic void mail_search_arg_get_base_mask(const struct mail_search_arg *arg,
7569ab8537418b7fc369265f26595b0ef9e4cb35Timo Sirainen mask_r->bin_mask.search_flags = arg->value.search_flags;
7569ab8537418b7fc369265f26595b0ef9e4cb35Timo Sirainenstatic struct mail_search_arg **
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainenmail_search_args_simplify_get_prev_argp(struct mail_search_simplify_ctx *ctx,
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen const struct mail_search_simplify_prev_arg *mask)
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen struct mail_search_simplify_prev_arg *prev_arg;
4376643cd2c7110e752c09f838f2c4eee6ed8ac6Timo Sirainen prev_arg = hash_table_lookup(ctx->prev_args, mask);
4376643cd2c7110e752c09f838f2c4eee6ed8ac6Timo Sirainen prev_arg = p_new(ctx->pool, struct mail_search_simplify_prev_arg, 1);
4376643cd2c7110e752c09f838f2c4eee6ed8ac6Timo Sirainen p_strdup(ctx->pool, mask->hdr_field_name_mask);
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen hash_table_insert(ctx->prev_args, prev_arg, prev_arg);
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainenstatic bool mail_search_args_merge_flags(struct mail_search_simplify_ctx *ctx,
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen if (!((!args->match_not && ctx->parent_and) ||
fc8d5f0ac909cca77840538e8beef98a8d40c21cTimo Sirainen prev_argp = mail_search_args_simplify_get_prev_argp(ctx, &mask);
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen (*prev_argp)->value.flags |= args->value.flags;
4376643cd2c7110e752c09f838f2c4eee6ed8ac6Timo Sirainenstatic void mail_search_args_simplify_set(struct mail_search_arg *args)
fc8d5f0ac909cca77840538e8beef98a8d40c21cTimo Sirainen unsigned int count;
5d4855d7b4dcffb6975ed8e3c9c376dac74e5c8aTimo Sirainen /* invert the set to drop the NOT */
fc8d5f0ac909cca77840538e8beef98a8d40c21cTimo Sirainen seq_range_array_invert(&args->value.seqset, 1, (uint32_t)-1);
5d4855d7b4dcffb6975ed8e3c9c376dac74e5c8aTimo Sirainen seqset = array_get(&args->value.seqset, &count);
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen if (count == 1 && seqset->seq1 == 1 && seqset->seq2 == (uint32_t)-1) {
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen /* 1:* is the same as ALL. */
d5cebe7f98e63d4e2822863ef2faa4971e8b3a5dTimo Sirainen } else if (count == 0) {
82f53ea81671bcc7b9bf24a34b04a4ba2752efd3Timo Sirainen /* empty set is the same as NOT ALL. this is mainly coming
d5cebe7f98e63d4e2822863ef2faa4971e8b3a5dTimo Sirainen from mail_search_args_merge_set() intersection. */
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainenstatic bool mail_search_args_merge_set(struct mail_search_simplify_ctx *ctx,
e9503210d3521a6833ed62dc332fc42ffb0e7a13Timo Sirainen prev_argp = mail_search_args_simplify_get_prev_argp(ctx, &mask);
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen seq_range_array_intersect(&(*prev_argp)->value.seqset,
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen seq_range_array_merge(&(*prev_argp)->value.seqset,
d5cebe7f98e63d4e2822863ef2faa4971e8b3a5dTimo Sirainenstatic bool mail_search_args_merge_time(struct mail_search_simplify_ctx *ctx,
adc409a7ac9689d3baf811712ad5a5432cab2d87Timo Sirainen struct mail_search_arg **prev_argp, *prev_arg;
adc409a7ac9689d3baf811712ad5a5432cab2d87Timo Sirainen mask.bin_mask.date_type = args->value.date_type;
adc409a7ac9689d3baf811712ad5a5432cab2d87Timo Sirainen prev_argp = mail_search_args_simplify_get_prev_argp(ctx, &mask);
e9503210d3521a6833ed62dc332fc42ffb0e7a13Timo Sirainen if (prev_arg->value.time < args->value.time) {
e9503210d3521a6833ed62dc332fc42ffb0e7a13Timo Sirainen /* prev_arg < 5 AND arg < 10 */
adc409a7ac9689d3baf811712ad5a5432cab2d87Timo Sirainen /* prev_arg < 10 AND arg < 5 */
f968e62caa52a8924bd05ebf76ff515b5c18e17bTimo Sirainen if (prev_arg->value.time < args->value.time) {
b44650b0f48a4b5f0dc240ed836833a00b643b9fTimo Sirainen /* prev_arg < 5 OR arg < 10 */
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen /* prev_arg < 10 OR arg < 5 */
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen if (prev_arg->value.time < args->value.time) {
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen /* prev_arg >= 5 AND arg >= 10 */
b44650b0f48a4b5f0dc240ed836833a00b643b9fTimo Sirainen /* prev_arg >= 10 AND arg >= 5 */
137ea7ca34005345aa2304a940149b7f3774d727Timo Sirainen if (prev_arg->value.time < args->value.time) {
137ea7ca34005345aa2304a940149b7f3774d727Timo Sirainen /* prev_arg >= 5 OR arg >= 10 */
137ea7ca34005345aa2304a940149b7f3774d727Timo Sirainen /* prev_arg >= 10 OR arg >= 5 */
d5cebe7f98e63d4e2822863ef2faa4971e8b3a5dTimo Sirainenstatic bool mail_search_args_merge_size(struct mail_search_simplify_ctx *ctx,
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen struct mail_search_arg **prev_argp, *prev_arg;
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen prev_argp = mail_search_args_simplify_get_prev_argp(ctx, &mask);
e9503210d3521a6833ed62dc332fc42ffb0e7a13Timo Sirainen if (prev_arg->value.size < args->value.size) {
e9503210d3521a6833ed62dc332fc42ffb0e7a13Timo Sirainen /* prev_arg < 5 AND arg < 10 */
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen /* prev_arg < 10 AND arg < 5 */
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen if (prev_arg->value.size < args->value.size) {
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen /* prev_arg < 5 OR arg < 10 */
70c181da837ed85fc5b0426c010b65609bda5329Timo Sirainen /* prev_arg < 10 OR arg < 5 */
e2a700d0628e395d64cbcef4b5b4510816bf51c4Timo Sirainen if (prev_arg->value.size < args->value.size) {
e9503210d3521a6833ed62dc332fc42ffb0e7a13Timo Sirainen /* prev_arg >= 5 AND arg >= 10 */
d5cebe7f98e63d4e2822863ef2faa4971e8b3a5dTimo Sirainen /* prev_arg >= 10 AND arg >= 5 */
e9503210d3521a6833ed62dc332fc42ffb0e7a13Timo Sirainen if (prev_arg->value.size < args->value.size) {
e9503210d3521a6833ed62dc332fc42ffb0e7a13Timo Sirainen /* prev_arg >= 5 OR arg >= 10 */
43358fffb1d9f3091fd94895e0ac4643c50e2388Timo Sirainen /* prev_arg >= 10 OR arg >= 5 */
35136dd2baf8dc30e4e754294ed81ff48e8c1e64Timo Sirainenstatic bool mail_search_args_merge_text(struct mail_search_simplify_ctx *ctx,
35136dd2baf8dc30e4e754294ed81ff48e8c1e64Timo Sirainen mask.hdr_field_name_mask = args->hdr_field_name;
35136dd2baf8dc30e4e754294ed81ff48e8c1e64Timo Sirainen prev_argp = mail_search_args_simplify_get_prev_argp(ctx, &mask);
35136dd2baf8dc30e4e754294ed81ff48e8c1e64Timo Sirainen /* duplicate search word. */
35136dd2baf8dc30e4e754294ed81ff48e8c1e64Timo Sirainenmail_search_args_have_equal(const struct mail_search_arg *args,
35136dd2baf8dc30e4e754294ed81ff48e8c1e64Timo Sirainen for (arg = args; arg != NULL; arg = arg->next) {
35136dd2baf8dc30e4e754294ed81ff48e8c1e64Timo Sirainen if (mail_search_arg_one_equals(arg, wanted_arg))
35136dd2baf8dc30e4e754294ed81ff48e8c1e64Timo Sirainenmail_search_args_remove_equal(struct mail_search_arg **argsp,
35136dd2baf8dc30e4e754294ed81ff48e8c1e64Timo Sirainen if (mail_search_arg_one_equals(*argp, wanted_arg)) {
35136dd2baf8dc30e4e754294ed81ff48e8c1e64Timo Sirainen if (!mail_search_args_remove_equal(&(*argp)->value.subargs, wanted_arg, FALSE)) {
35136dd2baf8dc30e4e754294ed81ff48e8c1e64Timo Sirainen /* we already verified that this should have
6fabfb7bbfd88d0c1de66981e52850f26067623bTimo Sirainenmail_search_args_have_all_equal(struct mail_search_arg *parent_arg,
0f39a57760d93cddbce3ca43096d78e0fe2f42fdTimo Sirainen for (arg = wanted_args; arg != NULL; arg = arg->next) {
0f39a57760d93cddbce3ca43096d78e0fe2f42fdTimo Sirainen if (!mail_search_args_have_equal(parent_arg->value.subargs, arg))
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainenstatic unsigned int
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainenmail_search_args_count(const struct mail_search_arg *args)
82f53ea81671bcc7b9bf24a34b04a4ba2752efd3Timo Sirainen unsigned int count;
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainenmail_search_args_simplify_drop_redundant_args(struct mail_search_arg **argsp,
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen struct mail_search_arg *arg, **argp, one_arg, *lowest_arg = NULL;
e9503210d3521a6833ed62dc332fc42ffb0e7a13Timo Sirainen child_subargs_type = and_arg ? SEARCH_OR : SEARCH_SUB;
70c181da837ed85fc5b0426c010b65609bda5329Timo Sirainen /* find the arg which has the lowest number of child args */
d5cebe7f98e63d4e2822863ef2faa4971e8b3a5dTimo Sirainen for (arg = *argsp; arg != NULL; arg = arg->next) {
e9503210d3521a6833ed62dc332fc42ffb0e7a13Timo Sirainen count = mail_search_args_count(arg->value.subargs);
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen /* if there are any args that include lowest_arg, drop the arg since
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen it's redundant. (non-SUB duplicates are dropped elsewhere.) */
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen if (*argp != lowest_arg && (*argp)->type == child_subargs_type &&
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen mail_search_args_have_all_equal(*argp, lowest_arg)) {
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainenmail_search_args_simplify_extract_common(struct mail_search_arg **argsp,
35136dd2baf8dc30e4e754294ed81ff48e8c1e64Timo Sirainen /* Simple SUB example:
35136dd2baf8dc30e4e754294ed81ff48e8c1e64Timo Sirainen (a AND b) OR (a AND c) -> a AND (b OR c)
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen More complicated example:
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen (c1 AND c2 AND u1 AND u2) OR (c1 AND c2 AND u3 AND u4) ->
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen c1 AND c2 AND ((u1 AND u2) OR (u3 AND u4))
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen Similarly for ORs:
2cfe9983ce7a6280636ee12beccc2e865111967bTimo Sirainen (a OR b) AND (a OR c) -> a OR (b AND c)
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen (c1 OR c2 OR u1 OR u2) AND (c1 OR c2 OR u3 OR u4) ->
6ef7e31619edfaa17ed044b45861d106a86191efTimo Sirainen c1 OR c2 OR ((u1 OR u2) AND (u3 OR u4))
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen struct mail_search_arg *arg, *sub_arg, *sub_next;
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen struct mail_search_arg *new_arg, *child_arg, *common_args = NULL;
d5cebe7f98e63d4e2822863ef2faa4971e8b3a5dTimo Sirainen if (*argsp == NULL || (*argsp)->next == NULL) {
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen /* single arg, nothing to extract */
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen child_subargs_type = and_arg ? SEARCH_OR : SEARCH_SUB;
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen /* find the first arg with child_subargs_type */
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen for (arg = *argsp; arg != NULL; arg = arg->next) {
eddd9bf1a1369aea4a2715f6be1137da6d17d293Timo Sirainen for (sub_arg = arg->value.subargs; sub_arg != NULL; sub_arg = sub_next) {
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen /* check if sub_arg is found from all the args */
d5cebe7f98e63d4e2822863ef2faa4971e8b3a5dTimo Sirainen for (arg = *argsp; arg != NULL; arg = arg->next) {
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen if (mail_search_arg_one_equals(arg, sub_arg)) {
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen /* the whole arg matches */
d5cebe7f98e63d4e2822863ef2faa4971e8b3a5dTimo Sirainen mail_search_args_have_equal(arg->value.subargs, sub_arg)) {
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen /* exists as subarg */
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen /* extract the arg and put it to common_args */
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen mail_search_args_remove_equal(argsp, sub_arg, TRUE);
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen /* replace all the original args with a single new SUB/OR arg */
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen new_arg = p_new(pool, struct mail_search_arg, 1);
c1ebcdad1b4d950eb22219704dd9d64a89d0568fTimo Sirainen /* there are only common args */
c1ebcdad1b4d950eb22219704dd9d64a89d0568fTimo Sirainen /* replace OR arg with AND(OR(non_common_args), common_args)
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen replace AND arg with OR(AND(non_common_args), common_args) */
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen child_arg = p_new(pool, struct mail_search_arg, 1);
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen child_arg->type = and_arg ? SEARCH_SUB : SEARCH_OR;
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainenmail_search_args_simplify_sub(struct mailbox *box, pool_t pool,
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen struct mail_search_arg **argsp, bool parent_and)
c1ebcdad1b4d950eb22219704dd9d64a89d0568fTimo Sirainen struct mail_search_arg *sub, **all_argsp = argsp;
d5cebe7f98e63d4e2822863ef2faa4971e8b3a5dTimo Sirainen ctx.pool = pool_alloconly_create("mail search args simplify", 1024);
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen hash_table_create(&ctx.prev_args, ctx.pool, 0,
d5cebe7f98e63d4e2822863ef2faa4971e8b3a5dTimo Sirainen if (args->match_not && (args->type == SEARCH_SUB ||
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen /* neg(p and q and ..) == neg(p) or neg(q) or ..
2e29e4797a48d78d669821722bdb54fd0a1d3b94Timo Sirainen neg(p or q or ..) == neg(p) and neg(q) and .. */
d5cebe7f98e63d4e2822863ef2faa4971e8b3a5dTimo Sirainen if ((args->type == SEARCH_SUB && parent_and) ||
d5cebe7f98e63d4e2822863ef2faa4971e8b3a5dTimo Sirainen ((args->type == SEARCH_SUB || args->type == SEARCH_OR) &&
d5cebe7f98e63d4e2822863ef2faa4971e8b3a5dTimo Sirainen /* p and (q and ..) == p and q and ..
d5cebe7f98e63d4e2822863ef2faa4971e8b3a5dTimo Sirainen p or (q or ..) == p or q or ..
case SEARCH_ALL: {
case SEARCH_FLAGS:
case SEARCH_SEQSET:
case SEARCH_UIDSET:
case SEARCH_BEFORE:
case SEARCH_ON:
case SEARCH_SINCE:
case SEARCH_SMALLER:
case SEARCH_LARGER:
case SEARCH_BODY:
case SEARCH_TEXT:
case SEARCH_HEADER:
case SEARCH_HEADER_ADDRESS:
if (merged) {
case SEARCH_SUB:
case SEARCH_OR:
case SEARCH_INTHREAD:
return FALSE;
if (!parent_and) {
return TRUE;
bool removals;
if (!removals)