mail-search-args-simplify.c revision dea8bfa31729dbdde1b12718f1ef98fac4e99db9
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen/* Copyright (c) 2002-2015 Dovecot authors, see the included COPYING file */
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen /* arg mask => prev_arg */
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainen HASH_TABLE(struct mail_search_simplify_prev_arg *,
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen struct mail_search_simplify_prev_arg *) prev_args;
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainenmail_search_simplify_prev_arg_cmp(const struct mail_search_simplify_prev_arg *arg1,
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainen const struct mail_search_simplify_prev_arg *arg2)
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainen ret = memcmp(&arg1->bin_mask, &arg2->bin_mask, sizeof(arg1->bin_mask));
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainen ret = null_strcmp(arg1->hdr_field_name_mask, arg2->hdr_field_name_mask);
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainen ret = null_strcmp(arg1->str_mask, arg2->str_mask);
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainenstatic unsigned int
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainenmail_search_simplify_prev_arg_hash(const struct mail_search_simplify_prev_arg *arg)
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainen unsigned int hash;
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainen hash = mem_hash(&arg->bin_mask, sizeof(arg->bin_mask));
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainenstatic void mail_search_arg_get_base_mask(const struct mail_search_arg *arg,
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainen mask_r->bin_mask.search_flags = arg->value.search_flags;
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainenstatic struct mail_search_arg **
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainenmail_search_args_simplify_get_prev_argp(struct mail_search_simplify_ctx *ctx,
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainen const struct mail_search_simplify_prev_arg *mask)
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen struct mail_search_simplify_prev_arg *prev_arg;
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen prev_arg = hash_table_lookup(ctx->prev_args, mask);
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen prev_arg = p_new(ctx->pool, struct mail_search_simplify_prev_arg, 1);
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainen p_strdup(ctx->pool, mask->hdr_field_name_mask);
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainen hash_table_insert(ctx->prev_args, prev_arg, prev_arg);
7000810786f2959f02cd6d2f4151a9eb61ff5db8Timo Sirainenstatic bool mail_search_args_merge_flags(struct mail_search_simplify_ctx *ctx,
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen if (!((!args->match_not && ctx->parent_and) ||
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen prev_argp = mail_search_args_simplify_get_prev_argp(ctx, &mask);
7000810786f2959f02cd6d2f4151a9eb61ff5db8Timo Sirainen (*prev_argp)->value.flags |= args->value.flags;
7000810786f2959f02cd6d2f4151a9eb61ff5db8Timo Sirainenstatic bool mail_search_args_merge_set(struct mail_search_simplify_ctx *ctx,
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen if (!((!args->match_not && ctx->parent_and) ||
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen prev_argp = mail_search_args_simplify_get_prev_argp(ctx, &mask);
7000810786f2959f02cd6d2f4151a9eb61ff5db8Timo Sirainen seq_range_array_merge(&(*prev_argp)->value.seqset,
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainenstatic bool mail_search_args_merge_time(struct mail_search_simplify_ctx *ctx,
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen struct mail_search_arg **prev_argp, *prev_arg;
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainen mask.bin_mask.date_type = args->value.date_type;
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen prev_argp = mail_search_args_simplify_get_prev_argp(ctx, &mask);
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen if (prev_arg->value.time < args->value.time) {
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen /* prev_arg < 5 AND arg < 10 */
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen /* prev_arg < 10 AND arg < 5 */
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen if (prev_arg->value.time < args->value.time) {
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen /* prev_arg < 5 OR arg < 10 */
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen /* prev_arg < 10 OR arg < 5 */
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen if (prev_arg->value.time < args->value.time) {
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen /* prev_arg >= 5 AND arg >= 10 */
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen /* prev_arg >= 10 AND arg >= 5 */
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen if (prev_arg->value.time < args->value.time) {
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen /* prev_arg >= 5 OR arg >= 10 */
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen /* prev_arg >= 10 OR arg >= 5 */
f5e8a76a128d4e92f0641135183c164fd5c5ce5eTimo Sirainenstatic bool mail_search_args_merge_size(struct mail_search_simplify_ctx *ctx,
f5e8a76a128d4e92f0641135183c164fd5c5ce5eTimo Sirainen struct mail_search_arg **prev_argp, *prev_arg;
f5e8a76a128d4e92f0641135183c164fd5c5ce5eTimo Sirainen prev_argp = mail_search_args_simplify_get_prev_argp(ctx, &mask);
f5e8a76a128d4e92f0641135183c164fd5c5ce5eTimo Sirainen if (prev_arg->value.size < args->value.size) {
f5e8a76a128d4e92f0641135183c164fd5c5ce5eTimo Sirainen /* prev_arg < 5 AND arg < 10 */
f5e8a76a128d4e92f0641135183c164fd5c5ce5eTimo Sirainen /* prev_arg < 10 AND arg < 5 */
f5e8a76a128d4e92f0641135183c164fd5c5ce5eTimo Sirainen if (prev_arg->value.size < args->value.size) {
f5e8a76a128d4e92f0641135183c164fd5c5ce5eTimo Sirainen /* prev_arg < 5 OR arg < 10 */
f5e8a76a128d4e92f0641135183c164fd5c5ce5eTimo Sirainen /* prev_arg < 10 OR arg < 5 */
f5e8a76a128d4e92f0641135183c164fd5c5ce5eTimo Sirainen if (prev_arg->value.size < args->value.size) {
f5e8a76a128d4e92f0641135183c164fd5c5ce5eTimo Sirainen /* prev_arg >= 5 AND arg >= 10 */
f5e8a76a128d4e92f0641135183c164fd5c5ce5eTimo Sirainen /* prev_arg >= 10 AND arg >= 5 */
f5e8a76a128d4e92f0641135183c164fd5c5ce5eTimo Sirainen if (prev_arg->value.size < args->value.size) {
f5e8a76a128d4e92f0641135183c164fd5c5ce5eTimo Sirainen /* prev_arg >= 5 OR arg >= 10 */
f5e8a76a128d4e92f0641135183c164fd5c5ce5eTimo Sirainen /* prev_arg >= 10 OR arg >= 5 */
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainenstatic bool mail_search_args_merge_text(struct mail_search_simplify_ctx *ctx,
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainen mask.hdr_field_name_mask = args->hdr_field_name;
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainen prev_argp = mail_search_args_simplify_get_prev_argp(ctx, &mask);
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainen /* duplicate search word. */
27e859cee42654bff801ba96677cfc4e4e0108c7Timo Sirainenmail_search_args_have_equal(const struct mail_search_arg *args,
27e859cee42654bff801ba96677cfc4e4e0108c7Timo Sirainen for (arg = args; arg != NULL; arg = arg->next) {
27e859cee42654bff801ba96677cfc4e4e0108c7Timo Sirainen if (mail_search_arg_one_equals(arg, wanted_arg))
071543cc13df9600d2e97aa35f28907be5a79477Timo Sirainenmail_search_args_remove_equal(struct mail_search_arg **argsp,
632018810af689442569cbb0139c55868923ccfeTimo Sirainen if (mail_search_arg_one_equals(*argp, wanted_arg)) {
071543cc13df9600d2e97aa35f28907be5a79477Timo Sirainen if (!mail_search_args_remove_equal(&(*argp)->value.subargs, wanted_arg, FALSE)) {
632018810af689442569cbb0139c55868923ccfeTimo Sirainen /* we already verified that this should have
27e859cee42654bff801ba96677cfc4e4e0108c7Timo Sirainenmail_search_args_have_all_equal(struct mail_search_arg *parent_arg,
27e859cee42654bff801ba96677cfc4e4e0108c7Timo Sirainen for (arg = wanted_args; arg != NULL; arg = arg->next) {
27e859cee42654bff801ba96677cfc4e4e0108c7Timo Sirainen if (!mail_search_args_have_equal(parent_arg->value.subargs, arg))
27e859cee42654bff801ba96677cfc4e4e0108c7Timo Sirainenstatic unsigned int
27e859cee42654bff801ba96677cfc4e4e0108c7Timo Sirainenmail_search_args_count(const struct mail_search_arg *args)
27e859cee42654bff801ba96677cfc4e4e0108c7Timo Sirainen unsigned int count;
8d3af185ae454653fad60e41c5f36edb1d45c868Timo Sirainenmail_search_args_simplify_drop_redundent_args(struct mail_search_arg **argsp,
27e859cee42654bff801ba96677cfc4e4e0108c7Timo Sirainen struct mail_search_arg *arg, **argp, one_arg, *lowest_arg = NULL;
8d3af185ae454653fad60e41c5f36edb1d45c868Timo Sirainen child_subargs_type = and_arg ? SEARCH_OR : SEARCH_SUB;
27e859cee42654bff801ba96677cfc4e4e0108c7Timo Sirainen /* find the arg which has the lowest number of child args */
8d3af185ae454653fad60e41c5f36edb1d45c868Timo Sirainen for (arg = *argsp; arg != NULL; arg = arg->next) {
27e859cee42654bff801ba96677cfc4e4e0108c7Timo Sirainen count = mail_search_args_count(arg->value.subargs);
27e859cee42654bff801ba96677cfc4e4e0108c7Timo Sirainen /* if there are any args that include lowest_arg, drop the arg since
27e859cee42654bff801ba96677cfc4e4e0108c7Timo Sirainen it's redundant. (non-SUB duplicates are dropped elsewhere.) */
8d3af185ae454653fad60e41c5f36edb1d45c868Timo Sirainen if (*argp != lowest_arg && (*argp)->type == child_subargs_type &&
27e859cee42654bff801ba96677cfc4e4e0108c7Timo Sirainen mail_search_args_have_all_equal(*argp, lowest_arg)) {
dea8bfa31729dbdde1b12718f1ef98fac4e99db9Timo Sirainenmail_search_args_simplify_extract_common(struct mail_search_arg **argsp,
071543cc13df9600d2e97aa35f28907be5a79477Timo Sirainen /* Simple SUB example:
071543cc13df9600d2e97aa35f28907be5a79477Timo Sirainen (a AND b) OR (a AND c) -> a AND (b OR c)
071543cc13df9600d2e97aa35f28907be5a79477Timo Sirainen More complicated example:
071543cc13df9600d2e97aa35f28907be5a79477Timo Sirainen (c1 AND c2 AND u1 AND u2) OR (c1 AND c2 AND u3 AND u4) ->
071543cc13df9600d2e97aa35f28907be5a79477Timo Sirainen c1 AND c2 AND ((u1 AND u2) OR (u3 AND u4))
dea8bfa31729dbdde1b12718f1ef98fac4e99db9Timo Sirainen Similarly for ORs:
dea8bfa31729dbdde1b12718f1ef98fac4e99db9Timo Sirainen (a OR b) AND (a OR c) -> a OR (b AND c)
dea8bfa31729dbdde1b12718f1ef98fac4e99db9Timo Sirainen (c1 OR c2 OR u1 OR u2) AND (c1 OR c2 OR u3 OR u4) ->
dea8bfa31729dbdde1b12718f1ef98fac4e99db9Timo Sirainen c1 OR c2 OR ((u1 OR u2) AND (u3 OR u4))
632018810af689442569cbb0139c55868923ccfeTimo Sirainen struct mail_search_arg *arg, *sub_arg, *sub_next;
071543cc13df9600d2e97aa35f28907be5a79477Timo Sirainen struct mail_search_arg *new_arg, *child_arg, *common_args = NULL;
071543cc13df9600d2e97aa35f28907be5a79477Timo Sirainen /* single arg, nothing to extract */
dea8bfa31729dbdde1b12718f1ef98fac4e99db9Timo Sirainen child_subargs_type = and_arg ? SEARCH_OR : SEARCH_SUB;
dea8bfa31729dbdde1b12718f1ef98fac4e99db9Timo Sirainen /* find the first arg with child_subargs_type */
071543cc13df9600d2e97aa35f28907be5a79477Timo Sirainen for (arg = *argsp; arg != NULL; arg = arg->next) {
632018810af689442569cbb0139c55868923ccfeTimo Sirainen for (sub_arg = arg->value.subargs; sub_arg != NULL; sub_arg = sub_next) {
632018810af689442569cbb0139c55868923ccfeTimo Sirainen /* check if sub_arg is found from all the args */
071543cc13df9600d2e97aa35f28907be5a79477Timo Sirainen for (arg = *argsp; arg != NULL; arg = arg->next) {
632018810af689442569cbb0139c55868923ccfeTimo Sirainen if (mail_search_arg_one_equals(arg, sub_arg)) {
632018810af689442569cbb0139c55868923ccfeTimo Sirainen /* the whole arg matches */
632018810af689442569cbb0139c55868923ccfeTimo Sirainen mail_search_args_have_equal(arg->value.subargs, sub_arg)) {
632018810af689442569cbb0139c55868923ccfeTimo Sirainen /* exists as subarg */
632018810af689442569cbb0139c55868923ccfeTimo Sirainen /* extract the arg and put it to common_args */
071543cc13df9600d2e97aa35f28907be5a79477Timo Sirainen mail_search_args_remove_equal(argsp, sub_arg, TRUE);
071543cc13df9600d2e97aa35f28907be5a79477Timo Sirainen /* replace all the original args with a single new SUB/OR arg */
071543cc13df9600d2e97aa35f28907be5a79477Timo Sirainen new_arg = p_new(pool, struct mail_search_arg, 1);
632018810af689442569cbb0139c55868923ccfeTimo Sirainen /* there are only common args */
dea8bfa31729dbdde1b12718f1ef98fac4e99db9Timo Sirainen /* replace OR arg with AND(OR(non_common_args), common_args)
dea8bfa31729dbdde1b12718f1ef98fac4e99db9Timo Sirainen replace AND arg with OR(AND(non_common_args), common_args) */
071543cc13df9600d2e97aa35f28907be5a79477Timo Sirainen child_arg = p_new(pool, struct mail_search_arg, 1);
dea8bfa31729dbdde1b12718f1ef98fac4e99db9Timo Sirainen child_arg->type = and_arg ? SEARCH_SUB : SEARCH_OR;
632018810af689442569cbb0139c55868923ccfeTimo Sirainenmail_search_args_simplify_sub(struct mailbox *box, pool_t pool,
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen struct mail_search_arg *args, bool parent_and)
7000810786f2959f02cd6d2f4151a9eb61ff5db8Timo Sirainen struct mail_search_arg *sub, *prev_arg = NULL;
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen ctx.pool = pool_alloconly_create("mail search args simplify", 1024);
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainen hash_table_create(&ctx.prev_args, ctx.pool, 0,
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen if (args->match_not && (args->type == SEARCH_SUB ||
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen /* neg(p and q and ..) == neg(p) or neg(q) or ..
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen neg(p or q or ..) == neg(p) and neg(q) and .. */
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen if ((args->type == SEARCH_SUB && parent_and) ||
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen ((args->type == SEARCH_SUB || args->type == SEARCH_OR) &&
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen /* p and (q and ..) == p and q and ..
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen p or (q or ..) == p or q or ..
dea8bfa31729dbdde1b12718f1ef98fac4e99db9Timo Sirainen if (mail_search_args_simplify_drop_redundent_args(&args->value.subargs, and_arg))
dea8bfa31729dbdde1b12718f1ef98fac4e99db9Timo Sirainen if (mail_search_args_simplify_extract_common(&args->value.subargs, pool, and_arg))
632018810af689442569cbb0139c55868923ccfeTimo Sirainen if (mail_search_args_simplify_sub(box, pool, args->value.subargs,
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen /* try to merge arguments */
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen merged = mail_search_args_merge_flags(&ctx, args);
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen merged = mail_search_args_merge_set(&ctx, args);
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen merged = mail_search_args_merge_time(&ctx, args);
f5e8a76a128d4e92f0641135183c164fd5c5ce5eTimo Sirainen merged = mail_search_args_merge_size(&ctx, args);
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainen merged = mail_search_args_merge_text(&ctx, args);
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainenmail_search_args_unnest_inthreads(struct mail_search_args *args,
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen struct mail_search_arg *arg, *thread_arg, *or_arg;
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen bool child_inthreads = FALSE, non_inthreads = FALSE;
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen for (arg = *argp; arg != NULL; arg = arg->next) {
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen /* children converted to SEARCH_INTHREADs */
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen if (!parent_inthreads || !child_inthreads || !non_inthreads)
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen /* put all non-INTHREADs under a single INTHREAD */
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen thread_arg = p_new(args->pool, struct mail_search_arg, 1);
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen /* not an INTHREAD or a SUB/OR with only INTHREADs */
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen /* We want to OR the args */
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen or_arg = p_new(args->pool, struct mail_search_arg, 1);
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen or_arg->value.subargs = thread_arg->value.subargs;
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainenvoid mail_search_args_simplify(struct mail_search_args *args)
632018810af689442569cbb0139c55868923ccfeTimo Sirainen removals = mail_search_args_simplify_sub(args->box, args->pool, args->args, TRUE);
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen if (mail_search_args_unnest_inthreads(args, &args->args,
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen /* we may have added some extra SUBs that could be dropped */
c89ceadf661bde22e1cd9dc2eac09c19202e65ecTimo Sirainen if (mail_search_args_simplify_sub(args->box, args->pool, args->args, TRUE))
8d3af185ae454653fad60e41c5f36edb1d45c868Timo Sirainen if (mail_search_args_simplify_drop_redundent_args(&args->args, TRUE))
dea8bfa31729dbdde1b12718f1ef98fac4e99db9Timo Sirainen if (mail_search_args_simplify_extract_common(&args->args, args->pool, TRUE))