fts-backend-solr-old.c revision adf8264ab1135c413bcede6af2e4248fd26a1ef9
2454dfa32c93c20a8522c6ed42fe057baaac9f9aStephan Bosch/* Copyright (c) 2006-2011 Dovecot authors, see the included COPYING file */
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen#include "lib.h"
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen#include "array.h"
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen#include "str.h"
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen#include "hash.h"
c115c742f730e312d6b6ab5064595cd0d8b4e26eTimo Sirainen#include "strescape.h"
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen#include "unichar.h"
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen#include "imap-utf7.h"
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen#include "mail-storage-private.h"
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen#include "mailbox-list-private.h"
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen#include "mail-search.h"
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen#include "fts-api.h"
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen#include "solr-connection.h"
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen#include "fts-solr-plugin.h"
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen#include <ctype.h>
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen#define SOLR_CMDBUF_SIZE (1024*64)
4ee00532a265bdfb38539d811fcd12d51210ac35Timo Sirainen#define SOLR_MAX_MULTI_ROWS 100000
a34bd633ab201f6a5ad1c00174fb8b0359031d00Timo Sirainen
a34bd633ab201f6a5ad1c00174fb8b0359031d00Timo Sirainenstruct solr_fts_backend {
0dffa25d211be541ee3c953b23566a1a990789dfTimo Sirainen struct fts_backend backend;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen char *id_username, *id_namespace;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen struct mail_namespace *default_ns;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen};
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainenstruct solr_fts_backend_update_context {
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen struct fts_backend_update_context ctx;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen struct mailbox *cur_box;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen char *id_box_name;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen struct solr_connection_post *post;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen uint32_t prev_uid, uid_validity;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen string_t *cmd, *hdr;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen
5f44975ec6c5755dd74bcd4c47a123a7242ecab3Timo Sirainen bool headers_open;
5f44975ec6c5755dd74bcd4c47a123a7242ecab3Timo Sirainen bool documents_added;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen};
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainenstatic struct solr_connection *solr_conn = NULL;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainenstatic bool is_valid_xml_char(unichar_t chr)
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen{
d22301419109ed4a38351715e6760011421dadecTimo Sirainen /* Valid characters in XML:
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen #x9 | #xA | #xD | [#x20-#xD7FF] | [#xE000-#xFFFD] |
4542c94adb8910e0174c784754e737cec16af59cTimo Sirainen [#x10000-#x10FFFF]
4542c94adb8910e0174c784754e737cec16af59cTimo Sirainen
4542c94adb8910e0174c784754e737cec16af59cTimo Sirainen This function gets called only for #x80 and higher */
4542c94adb8910e0174c784754e737cec16af59cTimo Sirainen if (chr > 0xd7ff && chr < 0xe000)
4542c94adb8910e0174c784754e737cec16af59cTimo Sirainen return FALSE;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen if (chr > 0xfffd && chr < 0x10000)
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen return FALSE;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen return chr < 0x10ffff;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen}
e3367d7b54864d2e4b1931903e3f660ae64fbe3aTimo Sirainen
e3367d7b54864d2e4b1931903e3f660ae64fbe3aTimo Sirainenstatic void
e3367d7b54864d2e4b1931903e3f660ae64fbe3aTimo Sirainenxml_encode_data(string_t *dest, const unsigned char *data, unsigned int len)
e3367d7b54864d2e4b1931903e3f660ae64fbe3aTimo Sirainen{
e3367d7b54864d2e4b1931903e3f660ae64fbe3aTimo Sirainen unichar_t chr;
e3367d7b54864d2e4b1931903e3f660ae64fbe3aTimo Sirainen unsigned int i;
e3367d7b54864d2e4b1931903e3f660ae64fbe3aTimo Sirainen
e3367d7b54864d2e4b1931903e3f660ae64fbe3aTimo Sirainen for (i = 0; i < len; i++) {
e3367d7b54864d2e4b1931903e3f660ae64fbe3aTimo Sirainen switch (data[i]) {
e3367d7b54864d2e4b1931903e3f660ae64fbe3aTimo Sirainen case '&':
e3367d7b54864d2e4b1931903e3f660ae64fbe3aTimo Sirainen str_append(dest, "&amp;");
e3367d7b54864d2e4b1931903e3f660ae64fbe3aTimo Sirainen break;
e3367d7b54864d2e4b1931903e3f660ae64fbe3aTimo Sirainen case '<':
e3367d7b54864d2e4b1931903e3f660ae64fbe3aTimo Sirainen str_append(dest, "&lt;");
e3367d7b54864d2e4b1931903e3f660ae64fbe3aTimo Sirainen break;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen case '>':
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen str_append(dest, "&gt;");
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen break;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen case '\t':
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen case '\n':
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen case '\r':
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen /* exceptions to the following control char check */
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen str_append_c(dest, data[i]);
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen break;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen default:
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen if (data[i] < 32) {
4542c94adb8910e0174c784754e737cec16af59cTimo Sirainen /* SOLR doesn't like control characters.
4542c94adb8910e0174c784754e737cec16af59cTimo Sirainen replace them with spaces. */
4542c94adb8910e0174c784754e737cec16af59cTimo Sirainen str_append_c(dest, ' ');
5f44975ec6c5755dd74bcd4c47a123a7242ecab3Timo Sirainen } else if (data[i] >= 0x80) {
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen /* make sure the character is valid for XML
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen so we don't get XML parser errors */
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen unsigned int char_len =
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen uni_utf8_char_bytes(data[i]);
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen if (i + char_len <= len &&
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen uni_utf8_get_char_n(data + i, char_len, &chr) == 1 &&
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen is_valid_xml_char(chr))
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen str_append_n(dest, data + i, char_len);
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen else {
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen str_append_n(dest, utf8_replacement_char,
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen UTF8_REPLACEMENT_CHAR_LEN);
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen }
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen i += char_len - 1;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen } else {
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen str_append_c(dest, data[i]);
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen }
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen break;
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen }
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen }
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen}
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainenstatic void xml_encode(string_t *dest, const char *str)
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen{
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen xml_encode_data(dest, (const unsigned char *)str, strlen(str));
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen}
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainenstatic const char *solr_escape_id_str(const char *str)
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen{
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen string_t *tmp;
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen const char *p;
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen for (p = str; *p != '\0'; p++) {
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen if (*p == '/' || *p == '!')
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen break;
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen }
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen if (*p == '\0')
d85a1a9d9af4a36ded4d30cb277905c807de2ec5Timo Sirainen return str;
d85a1a9d9af4a36ded4d30cb277905c807de2ec5Timo Sirainen
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen tmp = t_str_new(64);
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen for (p = str; *p != '\0'; p++) {
d979c1179d55ad86e40f869e48ef3e4db9c817b5Timo Sirainen switch (*p) {
d979c1179d55ad86e40f869e48ef3e4db9c817b5Timo Sirainen case '/':
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen str_append(tmp, "!\\");
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen break;
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen case '!':
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen str_append(tmp, "!!");
abb404575a238f27ea03b6049880f30f1656ddc4Timo Sirainen break;
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen default:
d979c1179d55ad86e40f869e48ef3e4db9c817b5Timo Sirainen str_append_c(tmp, *p);
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen break;
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen }
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen }
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen return str_c(tmp);
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen}
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainenstatic void solr_quote(string_t *dest, const char *str)
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen{
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen str_append_c(dest, '"');
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen str_append(dest, str_escape(str));
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen str_append_c(dest, '"');
c115c742f730e312d6b6ab5064595cd0d8b4e26eTimo Sirainen}
c115c742f730e312d6b6ab5064595cd0d8b4e26eTimo Sirainen
c115c742f730e312d6b6ab5064595cd0d8b4e26eTimo Sirainenstatic void solr_quote_http(string_t *dest, const char *str)
c115c742f730e312d6b6ab5064595cd0d8b4e26eTimo Sirainen{
c115c742f730e312d6b6ab5064595cd0d8b4e26eTimo Sirainen str_append(dest, "%22");
13e130c3af3032982de6b1d13c6dcddda9164848Timo Sirainen solr_connection_http_escape(solr_conn, dest, str);
c115c742f730e312d6b6ab5064595cd0d8b4e26eTimo Sirainen str_append(dest, "%22");
c115c742f730e312d6b6ab5064595cd0d8b4e26eTimo Sirainen}
c115c742f730e312d6b6ab5064595cd0d8b4e26eTimo Sirainen
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainenstatic void fts_solr_set_default_ns(struct solr_fts_backend *backend)
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen{
c115c742f730e312d6b6ab5064595cd0d8b4e26eTimo Sirainen struct mail_namespace *ns = backend->backend.ns;
c115c742f730e312d6b6ab5064595cd0d8b4e26eTimo Sirainen struct fts_solr_user *fuser = FTS_SOLR_USER_CONTEXT(ns->user);
c115c742f730e312d6b6ab5064595cd0d8b4e26eTimo Sirainen const struct fts_solr_settings *set = &fuser->set;
c115c742f730e312d6b6ab5064595cd0d8b4e26eTimo Sirainen const char *str;
a4f09749814b93e8ad3ec8a0dc18885b874d6f8cTimo Sirainen
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen if (backend->default_ns != NULL)
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen return;
c115c742f730e312d6b6ab5064595cd0d8b4e26eTimo Sirainen
c115c742f730e312d6b6ab5064595cd0d8b4e26eTimo Sirainen if (set->default_ns_prefix != NULL) {
13e130c3af3032982de6b1d13c6dcddda9164848Timo Sirainen backend->default_ns =
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen mail_namespace_find_prefix(ns->user->namespaces,
13e130c3af3032982de6b1d13c6dcddda9164848Timo Sirainen set->default_ns_prefix);
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen if (backend->default_ns == NULL) {
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen i_error("fts_solr: default_ns setting points to "
c115c742f730e312d6b6ab5064595cd0d8b4e26eTimo Sirainen "nonexistent namespace");
c115c742f730e312d6b6ab5064595cd0d8b4e26eTimo Sirainen }
8eba883232f80178b60fa416f73292bf5f990fecTimo Sirainen }
8eba883232f80178b60fa416f73292bf5f990fecTimo Sirainen if (backend->default_ns == NULL) {
8eba883232f80178b60fa416f73292bf5f990fecTimo Sirainen backend->default_ns =
8eba883232f80178b60fa416f73292bf5f990fecTimo Sirainen mail_namespace_find_inbox(ns->user->namespaces);
8eba883232f80178b60fa416f73292bf5f990fecTimo Sirainen }
8eba883232f80178b60fa416f73292bf5f990fecTimo Sirainen while (backend->default_ns->alias_for != NULL)
8eba883232f80178b60fa416f73292bf5f990fecTimo Sirainen backend->default_ns = backend->default_ns->alias_for;
8eba883232f80178b60fa416f73292bf5f990fecTimo Sirainen
8eba883232f80178b60fa416f73292bf5f990fecTimo Sirainen if (ns != backend->default_ns) {
8eba883232f80178b60fa416f73292bf5f990fecTimo Sirainen str = solr_escape_id_str(ns->prefix);
8eba883232f80178b60fa416f73292bf5f990fecTimo Sirainen backend->id_namespace = i_strdup(str);
8eba883232f80178b60fa416f73292bf5f990fecTimo Sirainen }
7631f16156aca373004953fe6b01a7f343fb47e0Timo Sirainen}
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainenstatic void fts_box_name_get_root(struct mail_namespace **ns, const char **name)
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen{
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen struct mail_namespace *orig_ns = *ns;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen
7631f16156aca373004953fe6b01a7f343fb47e0Timo Sirainen while ((*ns)->alias_for != NULL)
7631f16156aca373004953fe6b01a7f343fb47e0Timo Sirainen *ns = (*ns)->alias_for;
a249dd267f05d349f1b4aa27b40a56083c8ba392Timo Sirainen
03010dbaa74ec70f062994dfe3cd39bedc99a28bTimo Sirainen if (**name == '\0' && *ns != orig_ns &&
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen ((*ns)->flags & NAMESPACE_FLAG_INBOX_USER) != 0) {
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen /* ugly workaround to allow selecting INBOX from a Maildir/
efe78d3ba24fc866af1c79b9223dc0809ba26cadStephan Bosch when it's not in the inbox=yes namespace. */
5f44975ec6c5755dd74bcd4c47a123a7242ecab3Timo Sirainen *name = "INBOX";
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen }
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen}
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen
a249dd267f05d349f1b4aa27b40a56083c8ba392Timo Sirainenstatic const char *
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainenfts_box_get_root(struct mailbox *box, struct mail_namespace **ns_r)
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen{
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen struct mail_namespace *ns = mailbox_get_namespace(box);
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen const char *name;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen if (t_imap_utf8_to_utf7(box->name, &name) < 0)
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen i_unreached();
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen
a249dd267f05d349f1b4aa27b40a56083c8ba392Timo Sirainen fts_box_name_get_root(&ns, &name);
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen *ns_r = ns;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen return name;
7631f16156aca373004953fe6b01a7f343fb47e0Timo Sirainen}
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainenstatic struct fts_backend *fts_backend_solr_alloc(void)
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen{
fe5cca45e94608d1c471990216941bf893bc8adaTimo Sirainen struct solr_fts_backend *backend;
fe5cca45e94608d1c471990216941bf893bc8adaTimo Sirainen
fe5cca45e94608d1c471990216941bf893bc8adaTimo Sirainen backend = i_new(struct solr_fts_backend, 1);
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen backend->backend = fts_backend_solr_old;
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen return &backend->backend;
fe5cca45e94608d1c471990216941bf893bc8adaTimo Sirainen}
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainenstatic int
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainenfts_backend_solr_init(struct fts_backend *_backend,
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen const char **error_r ATTR_UNUSED)
fe5cca45e94608d1c471990216941bf893bc8adaTimo Sirainen{
fe5cca45e94608d1c471990216941bf893bc8adaTimo Sirainen struct solr_fts_backend *backend = (struct solr_fts_backend *)_backend;
e15b305e90c9834734ccf35ed78f0ad29d570ee9Timo Sirainen struct fts_solr_user *fuser = FTS_SOLR_USER_CONTEXT(_backend->ns->user);
e15b305e90c9834734ccf35ed78f0ad29d570ee9Timo Sirainen const struct fts_solr_settings *set = &fuser->set;
e15b305e90c9834734ccf35ed78f0ad29d570ee9Timo Sirainen const char *str;
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen if (solr_conn == NULL)
e15b305e90c9834734ccf35ed78f0ad29d570ee9Timo Sirainen solr_conn = solr_connection_init(set->url, set->debug);
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen str = solr_escape_id_str(_backend->ns->user->username);
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen backend->id_username = i_strdup(str);
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen return 0;
e15b305e90c9834734ccf35ed78f0ad29d570ee9Timo Sirainen}
e15b305e90c9834734ccf35ed78f0ad29d570ee9Timo Sirainen
421d30619384e72a27e2a5d13ff6525aff4d17feTimo Sirainenstatic void fts_backend_solr_deinit(struct fts_backend *_backend)
421d30619384e72a27e2a5d13ff6525aff4d17feTimo Sirainen{
421d30619384e72a27e2a5d13ff6525aff4d17feTimo Sirainen struct solr_fts_backend *backend = (struct solr_fts_backend *)_backend;
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen i_free(backend->id_namespace);
421d30619384e72a27e2a5d13ff6525aff4d17feTimo Sirainen i_free(backend->id_username);
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen i_free(backend);
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen}
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainenstatic void
421d30619384e72a27e2a5d13ff6525aff4d17feTimo Sirainensolr_add_ns_query(string_t *str, struct solr_fts_backend *backend,
421d30619384e72a27e2a5d13ff6525aff4d17feTimo Sirainen struct mail_namespace *ns, bool neg)
ecd69c4e8371853667e01b0c16d436ef7f7393e2Timo Sirainen{
ecd69c4e8371853667e01b0c16d436ef7f7393e2Timo Sirainen while (ns->alias_for != NULL)
ecd69c4e8371853667e01b0c16d436ef7f7393e2Timo Sirainen ns = ns->alias_for;
ecd69c4e8371853667e01b0c16d436ef7f7393e2Timo Sirainen
ecd69c4e8371853667e01b0c16d436ef7f7393e2Timo Sirainen if (ns == backend->default_ns || *ns->prefix == '\0') {
ecd69c4e8371853667e01b0c16d436ef7f7393e2Timo Sirainen if (!neg)
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen str_append(str, " -ns:[* TO *]");
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen else
a34bd633ab201f6a5ad1c00174fb8b0359031d00Timo Sirainen str_append(str, " +ns:[* TO *]");
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen } else {
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen if (!neg)
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen str_append(str, " +ns:");
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen else
a34bd633ab201f6a5ad1c00174fb8b0359031d00Timo Sirainen str_append(str, " -ns:");
a34bd633ab201f6a5ad1c00174fb8b0359031d00Timo Sirainen solr_quote(str, ns->prefix);
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen }
c0225f7f6b43d34dc58c17d3304f0fd60ab89894Timo Sirainen}
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainenstatic void
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainensolr_add_ns_query_http(string_t *str, struct solr_fts_backend *backend,
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen struct mail_namespace *ns)
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen{
a34bd633ab201f6a5ad1c00174fb8b0359031d00Timo Sirainen string_t *tmp;
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen tmp = t_str_new(64);
942302b0247403645394d848b3c620ead262a2a5Timo Sirainen solr_add_ns_query(tmp, backend, ns, FALSE);
942302b0247403645394d848b3c620ead262a2a5Timo Sirainen solr_connection_http_escape(solr_conn, str, str_c(tmp));
942302b0247403645394d848b3c620ead262a2a5Timo Sirainen}
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainenstatic int
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainenfts_backend_solr_get_last_uid_fallback(struct solr_fts_backend *backend,
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen struct mailbox *box,
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen uint32_t *last_uid_r)
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen{
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen struct mail_namespace *ns;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen struct mailbox_status status;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen struct solr_result **results;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen const struct seq_range *uidvals;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen const char *box_name;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen unsigned int count;
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen string_t *str;
a34bd633ab201f6a5ad1c00174fb8b0359031d00Timo Sirainen pool_t pool;
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen int ret = 0;
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen
942302b0247403645394d848b3c620ead262a2a5Timo Sirainen str = t_str_new(256);
942302b0247403645394d848b3c620ead262a2a5Timo Sirainen str_append(str, "fl=uid&rows=1&sort=uid+desc&q=");
942302b0247403645394d848b3c620ead262a2a5Timo Sirainen
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen box_name = fts_box_get_root(box, &ns);
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen mailbox_get_open_status(box, STATUS_UIDVALIDITY, &status);
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen str_printfa(str, "uidv:%u+box:", status.uidvalidity);
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen solr_quote_http(str, box_name);
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen solr_add_ns_query_http(str, backend, ns);
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen str_append(str, "+user:");
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen solr_quote_http(str, ns->user->username);
a34bd633ab201f6a5ad1c00174fb8b0359031d00Timo Sirainen
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen pool = pool_alloconly_create("solr last uid lookup", 1024);
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen if (solr_connection_select(solr_conn, str_c(str),
942302b0247403645394d848b3c620ead262a2a5Timo Sirainen pool, &results) < 0)
942302b0247403645394d848b3c620ead262a2a5Timo Sirainen ret = -1;
942302b0247403645394d848b3c620ead262a2a5Timo Sirainen else if (results[0] == NULL) {
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen /* no UIDs */
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen *last_uid_r = 0;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen } else {
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen uidvals = array_get(&results[0]->uids, &count);
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen i_assert(count > 0);
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen if (count == 1 && uidvals[0].seq1 == uidvals[0].seq2) {
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen *last_uid_r = uidvals[0].seq1;
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen } else {
a34bd633ab201f6a5ad1c00174fb8b0359031d00Timo Sirainen i_error("fts_solr: Last UID lookup returned multiple rows");
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen ret = -1;
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen }
942302b0247403645394d848b3c620ead262a2a5Timo Sirainen }
942302b0247403645394d848b3c620ead262a2a5Timo Sirainen pool_unref(&pool);
942302b0247403645394d848b3c620ead262a2a5Timo Sirainen return ret;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen}
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainenstatic int
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainenfts_backend_solr_get_last_uid(struct fts_backend *_backend,
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen struct mailbox *box, uint32_t *last_uid_r)
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen{
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen struct solr_fts_backend *backend =
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen (struct solr_fts_backend *)_backend;
a34bd633ab201f6a5ad1c00174fb8b0359031d00Timo Sirainen struct fts_index_header hdr;
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen if (fts_index_get_header(box, &hdr)) {
942302b0247403645394d848b3c620ead262a2a5Timo Sirainen *last_uid_r = hdr.last_indexed_uid;
942302b0247403645394d848b3c620ead262a2a5Timo Sirainen return 0;
942302b0247403645394d848b3c620ead262a2a5Timo Sirainen }
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen /* either nothing has been indexed, or the index was corrupted.
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen do it the slow way. */
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen if (fts_backend_solr_get_last_uid_fallback(backend, box, last_uid_r) < 0)
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen return -1;
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen (void)fts_index_set_last_uid(box, *last_uid_r);
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen return 0;
a34bd633ab201f6a5ad1c00174fb8b0359031d00Timo Sirainen}
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainenstatic struct fts_backend_update_context *
942302b0247403645394d848b3c620ead262a2a5Timo Sirainenfts_backend_solr_update_init(struct fts_backend *_backend)
942302b0247403645394d848b3c620ead262a2a5Timo Sirainen{
942302b0247403645394d848b3c620ead262a2a5Timo Sirainen struct solr_fts_backend *backend =
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen (struct solr_fts_backend *)_backend;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen struct solr_fts_backend_update_context *ctx;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen ctx = i_new(struct solr_fts_backend_update_context, 1);
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen ctx->ctx.backend = _backend;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen ctx->cmd = str_new(default_pool, SOLR_CMDBUF_SIZE);
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen ctx->hdr = str_new(default_pool, 4096);
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen fts_solr_set_default_ns(backend);
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen return &ctx->ctx;
45d47bc4660fe4bbb07817c9580deef9cca63646Timo Sirainen}
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainenstatic void xml_encode_id(struct solr_fts_backend_update_context *ctx,
a34bd633ab201f6a5ad1c00174fb8b0359031d00Timo Sirainen string_t *str, uint32_t uid)
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen{
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen struct solr_fts_backend *backend =
45d47bc4660fe4bbb07817c9580deef9cca63646Timo Sirainen (struct solr_fts_backend *)ctx->ctx.backend;
45d47bc4660fe4bbb07817c9580deef9cca63646Timo Sirainen
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen if (uid != 0)
942302b0247403645394d848b3c620ead262a2a5Timo Sirainen str_printfa(str, "%u/", uid);
942302b0247403645394d848b3c620ead262a2a5Timo Sirainen else
45d47bc4660fe4bbb07817c9580deef9cca63646Timo Sirainen str_append(str, "L/");
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen if (backend->id_namespace != NULL) {
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen xml_encode(str, backend->id_namespace);
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen str_append_c(str, '/');
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen }
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen str_printfa(str, "%u/", ctx->uid_validity);
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen xml_encode(str, backend->id_username);
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen str_append_c(str, '/');
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen xml_encode(str, ctx->id_box_name);
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen}
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen
a34bd633ab201f6a5ad1c00174fb8b0359031d00Timo Sirainenstatic void
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainenfts_backend_solr_add_doc_prefix(struct solr_fts_backend_update_context *ctx,
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen uint32_t uid)
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen{
942302b0247403645394d848b3c620ead262a2a5Timo Sirainen struct solr_fts_backend *backend =
942302b0247403645394d848b3c620ead262a2a5Timo Sirainen (struct solr_fts_backend *)ctx->ctx.backend;
942302b0247403645394d848b3c620ead262a2a5Timo Sirainen struct mailbox *box = ctx->cur_box;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen struct mail_namespace *ns;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen const char *box_name;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen ctx->documents_added = TRUE;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen str_printfa(ctx->cmd, "<doc>"
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen "<field name=\"uid\">%u</field>"
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen "<field name=\"uidv\">%u</field>",
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen uid, ctx->uid_validity);
c8d1ccf9dec39e9e145d55b472eee43d95764189Timo Sirainen
c8d1ccf9dec39e9e145d55b472eee43d95764189Timo Sirainen box_name = fts_box_get_root(box, &ns);
c8d1ccf9dec39e9e145d55b472eee43d95764189Timo Sirainen
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen if (ns != backend->default_ns) {
a34bd633ab201f6a5ad1c00174fb8b0359031d00Timo Sirainen str_append(ctx->cmd, "<field name=\"ns\">");
a34bd633ab201f6a5ad1c00174fb8b0359031d00Timo Sirainen xml_encode(ctx->cmd, ns->prefix);
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen str_append(ctx->cmd, "</field>");
a4f09749814b93e8ad3ec8a0dc18885b874d6f8cTimo Sirainen }
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen str_append(ctx->cmd, "<field name=\"box\">");
c8d1ccf9dec39e9e145d55b472eee43d95764189Timo Sirainen xml_encode(ctx->cmd, box_name);
942302b0247403645394d848b3c620ead262a2a5Timo Sirainen str_append(ctx->cmd, "</field><field name=\"user\">");
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen xml_encode(ctx->cmd, ns->user->username);
942302b0247403645394d848b3c620ead262a2a5Timo Sirainen str_append(ctx->cmd, "</field>");
942302b0247403645394d848b3c620ead262a2a5Timo Sirainen}
942302b0247403645394d848b3c620ead262a2a5Timo Sirainen
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainenstatic int
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainenfts_backed_solr_build_commit(struct solr_fts_backend_update_context *ctx)
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen{
6df0ab0c1ab91f06b6418cb30eff44405a1b8f02Timo Sirainen if (ctx->post == NULL)
6df0ab0c1ab91f06b6418cb30eff44405a1b8f02Timo Sirainen return 0;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen str_append(ctx->cmd, "</doc></add>");
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen solr_connection_post_more(ctx->post, str_data(ctx->cmd),
a08e96bb7821294656492a57482d838571d10c00Aki Tuomi str_len(ctx->cmd));
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen return solr_connection_post_end(ctx->post);
a08e96bb7821294656492a57482d838571d10c00Aki Tuomi}
a08e96bb7821294656492a57482d838571d10c00Aki Tuomi
6df0ab0c1ab91f06b6418cb30eff44405a1b8f02Timo Sirainenstatic int
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainenfts_backend_solr_update_deinit(struct fts_backend_update_context *_ctx)
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen{
a34bd633ab201f6a5ad1c00174fb8b0359031d00Timo Sirainen struct solr_fts_backend_update_context *ctx =
a08e96bb7821294656492a57482d838571d10c00Aki Tuomi (struct solr_fts_backend_update_context *)_ctx;
6df0ab0c1ab91f06b6418cb30eff44405a1b8f02Timo Sirainen const char *str;
a08e96bb7821294656492a57482d838571d10c00Aki Tuomi int ret;
a08e96bb7821294656492a57482d838571d10c00Aki Tuomi
6df0ab0c1ab91f06b6418cb30eff44405a1b8f02Timo Sirainen ret = fts_backed_solr_build_commit(ctx);
a08e96bb7821294656492a57482d838571d10c00Aki Tuomi
a08e96bb7821294656492a57482d838571d10c00Aki Tuomi /* commit and wait until the documents we just indexed are
6df0ab0c1ab91f06b6418cb30eff44405a1b8f02Timo Sirainen visible to the following search */
a08e96bb7821294656492a57482d838571d10c00Aki Tuomi str = t_strdup_printf("<commit waitFlush=\"false\" "
6df0ab0c1ab91f06b6418cb30eff44405a1b8f02Timo Sirainen "waitSearcher=\"%s\"/>",
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen ctx->documents_added ? "true" : "false");
942302b0247403645394d848b3c620ead262a2a5Timo Sirainen if (solr_connection_post(solr_conn, str) < 0)
942302b0247403645394d848b3c620ead262a2a5Timo Sirainen ret = -1;
942302b0247403645394d848b3c620ead262a2a5Timo Sirainen
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen str_free(&ctx->cmd);
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen str_free(&ctx->hdr);
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen i_free(ctx->id_box_name);
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen i_free(ctx);
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen return ret;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen}
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainenstatic void
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainenfts_backend_solr_update_set_mailbox(struct fts_backend_update_context *_ctx,
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen struct mailbox *box)
a34bd633ab201f6a5ad1c00174fb8b0359031d00Timo Sirainen{
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen struct solr_fts_backend_update_context *ctx =
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen (struct solr_fts_backend_update_context *)_ctx;
942302b0247403645394d848b3c620ead262a2a5Timo Sirainen struct mailbox_status status;
942302b0247403645394d848b3c620ead262a2a5Timo Sirainen struct mail_namespace *ns;
942302b0247403645394d848b3c620ead262a2a5Timo Sirainen
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen if (ctx->prev_uid != 0) {
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen (void)fts_index_set_last_uid(ctx->cur_box, ctx->prev_uid);
21aaa6affb9f134112b75b5db737309fc35ef1cfMartti Rannanjärvi ctx->prev_uid = 0;
21aaa6affb9f134112b75b5db737309fc35ef1cfMartti Rannanjärvi }
9af6cc9ebc9986c1275ebdfa29c39e152af1557eTimo Sirainen
9af6cc9ebc9986c1275ebdfa29c39e152af1557eTimo Sirainen ctx->cur_box = box;
21aaa6affb9f134112b75b5db737309fc35ef1cfMartti Rannanjärvi ctx->uid_validity = 0;
9af6cc9ebc9986c1275ebdfa29c39e152af1557eTimo Sirainen i_free_and_null(ctx->id_box_name);
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen
21aaa6affb9f134112b75b5db737309fc35ef1cfMartti Rannanjärvi if (box != NULL) {
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen ctx->id_box_name = i_strdup(fts_box_get_root(box, &ns));
21aaa6affb9f134112b75b5db737309fc35ef1cfMartti Rannanjärvi
21aaa6affb9f134112b75b5db737309fc35ef1cfMartti Rannanjärvi mailbox_get_open_status(box, STATUS_UIDVALIDITY, &status);
21aaa6affb9f134112b75b5db737309fc35ef1cfMartti Rannanjärvi ctx->uid_validity = status.uidvalidity;
9af6cc9ebc9986c1275ebdfa29c39e152af1557eTimo Sirainen }
9af6cc9ebc9986c1275ebdfa29c39e152af1557eTimo Sirainen}
225e82df5dd1e765f4e52b80c954558f00e5a7dfTimo Sirainen
225e82df5dd1e765f4e52b80c954558f00e5a7dfTimo Sirainenstatic void
225e82df5dd1e765f4e52b80c954558f00e5a7dfTimo Sirainenfts_backend_solr_update_expunge(struct fts_backend_update_context *_ctx,
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen uint32_t uid)
225e82df5dd1e765f4e52b80c954558f00e5a7dfTimo Sirainen{
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen struct solr_fts_backend_update_context *ctx =
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen (struct solr_fts_backend_update_context *)_ctx;
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen
225e82df5dd1e765f4e52b80c954558f00e5a7dfTimo Sirainen T_BEGIN {
225e82df5dd1e765f4e52b80c954558f00e5a7dfTimo Sirainen string_t *cmd;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen cmd = t_str_new(256);
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen str_append(cmd, "<delete><id>");
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen xml_encode_id(ctx, cmd, uid);
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen str_append(cmd, "</id></delete>");
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen
a34bd633ab201f6a5ad1c00174fb8b0359031d00Timo Sirainen (void)solr_connection_post(solr_conn, str_c(cmd));
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen } T_END;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen}
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainenstatic void
0206dc57f2c04da69599dea5816235cfeb2b897aMartti Rannanjärvifts_backend_solr_uid_changed(struct solr_fts_backend_update_context *ctx,
0206dc57f2c04da69599dea5816235cfeb2b897aMartti Rannanjärvi uint32_t uid)
0206dc57f2c04da69599dea5816235cfeb2b897aMartti Rannanjärvi{
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen if (ctx->post == NULL) {
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen i_assert(ctx->prev_uid == 0);
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen ctx->post = solr_connection_post_begin(solr_conn);
18a41cbd38f83429b790414c1159c097af4a59b8Timo Sirainen str_append(ctx->cmd, "<add>");
a34bd633ab201f6a5ad1c00174fb8b0359031d00Timo Sirainen } else {
0206dc57f2c04da69599dea5816235cfeb2b897aMartti Rannanjärvi ctx->headers_open = FALSE;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen str_append(ctx->cmd, "<field name=\"hdr\">");
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen str_append_str(ctx->cmd, ctx->hdr);
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen str_append(ctx->cmd, "</field>");
e3367d7b54864d2e4b1931903e3f660ae64fbe3aTimo Sirainen str_truncate(ctx->hdr, 0);
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen str_append(ctx->cmd, "</doc>");
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen }
fe5cca45e94608d1c471990216941bf893bc8adaTimo Sirainen ctx->prev_uid = uid;
e15b305e90c9834734ccf35ed78f0ad29d570ee9Timo Sirainen
421d30619384e72a27e2a5d13ff6525aff4d17feTimo Sirainen fts_backend_solr_add_doc_prefix(ctx, uid);
ecd69c4e8371853667e01b0c16d436ef7f7393e2Timo Sirainen str_printfa(ctx->cmd, "<field name=\"id\">");
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen xml_encode_id(ctx, ctx->cmd, uid);
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen str_append(ctx->cmd, "</field>");
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen}
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen
db0735f9b388c5bcfb781b1b25015e898d63d953Timo Sirainenstatic bool
f46885a5b78b15a8d2419f6e5d13b643bd85e41fTimo Sirainenfts_backend_solr_update_set_build_key(struct fts_backend_update_context *_ctx,
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen const struct fts_backend_build_key *key)
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen{
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen struct solr_fts_backend_update_context *ctx =
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen (struct solr_fts_backend_update_context *)_ctx;
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen if (key->uid != ctx->prev_uid)
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen fts_backend_solr_uid_changed(ctx, key->uid);
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen switch (key->type) {
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen case FTS_BACKEND_BUILD_KEY_HDR:
306b3f41b05da642d87e7ca7a1496efce9f5902fTimo Sirainen case FTS_BACKEND_BUILD_KEY_MIME_HDR:
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen xml_encode(ctx->hdr, key->hdr_name);
21aaa6affb9f134112b75b5db737309fc35ef1cfMartti Rannanjärvi str_append(ctx->hdr, ": ");
dee43975a70bcdb9dc83d34d6a2b177d37bb7194Timo Sirainen ctx->headers_open = TRUE;
dee43975a70bcdb9dc83d34d6a2b177d37bb7194Timo Sirainen break;
ad48319996942463675b53877092ab7e13a7a75aTimo Sirainen case FTS_BACKEND_BUILD_KEY_BODY_PART:
f46885a5b78b15a8d2419f6e5d13b643bd85e41fTimo Sirainen ctx->headers_open = FALSE;
225e82df5dd1e765f4e52b80c954558f00e5a7dfTimo Sirainen str_append(ctx->cmd, "<field name=\"body\">");
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen break;
6564208826b0f46a00f010d1b5711d85944c3c88Timo Sirainen case FTS_BACKEND_BUILD_KEY_BODY_PART_BINARY:
6de6ec228a41275ddda972d4a554699ea75cd06dTimo Sirainen i_unreached();
24e5e4526d8f5cbc056ab97fd0d154d0936d7a5eTimo Sirainen }
return TRUE;
}
static void
fts_backend_solr_update_unset_build_key(struct fts_backend_update_context *_ctx)
{
struct solr_fts_backend_update_context *ctx =
(struct solr_fts_backend_update_context *)_ctx;
if (!ctx->headers_open)
str_append(ctx->cmd, "</field>");
else
str_append_c(ctx->hdr, '\n');
}
static int
fts_backend_solr_update_build_more(struct fts_backend_update_context *_ctx,
const unsigned char *data, size_t size)
{
struct solr_fts_backend_update_context *ctx =
(struct solr_fts_backend_update_context *)_ctx;
xml_encode_data(ctx->cmd, data, size);
if (str_len(ctx->cmd) > SOLR_CMDBUF_SIZE-128) {
solr_connection_post_more(ctx->post, str_data(ctx->cmd),
str_len(ctx->cmd));
str_truncate(ctx->cmd, 0);
}
return 0;
}
static int fts_backend_solr_refresh(struct fts_backend *backend ATTR_UNUSED)
{
return 0;
}
static int fts_backend_solr_optimize(struct fts_backend *backend ATTR_UNUSED)
{
return 0;
}
static bool
solr_add_definite_query(string_t *str, struct mail_search_arg *arg)
{
switch (arg->type) {
case SEARCH_TEXT: {
if (arg->match_not)
str_append_c(str, '-');
str_append(str, "(hdr:");
solr_quote_http(str, arg->value.str);
str_append(str, "+OR+body:");
solr_quote_http(str, arg->value.str);
str_append(str, ")");
break;
}
case SEARCH_BODY:
if (arg->match_not)
str_append_c(str, '-');
str_append(str, "body:");
solr_quote_http(str, arg->value.str);
break;
default:
return FALSE;
}
return TRUE;
}
static bool
solr_add_definite_query_args(string_t *str, struct mail_search_arg *arg,
bool and_args)
{
unsigned int last_len;
last_len = str_len(str);
for (; arg != NULL; arg = arg->next) {
if (solr_add_definite_query(str, arg)) {
arg->match_always = TRUE;
last_len = str_len(str);
if (and_args)
str_append(str, "+AND+");
else
str_append(str, "+OR+");
}
}
if (str_len(str) == last_len)
return FALSE;
str_truncate(str, last_len);
return TRUE;
}
static int
fts_backend_solr_lookup(struct fts_backend *_backend, struct mailbox *box,
struct mail_search_arg *args, bool and_args,
struct fts_result *result)
{
struct solr_fts_backend *backend =
(struct solr_fts_backend *)_backend;
struct mail_namespace *ns;
struct mailbox_status status;
string_t *str;
const char *box_name;
pool_t pool;
struct solr_result **results;
int ret;
fts_solr_set_default_ns(backend);
mailbox_get_open_status(box, STATUS_UIDVALIDITY | STATUS_UIDNEXT,
&status);
str = t_str_new(256);
str_printfa(str, "fl=uid,score&rows=%u&sort=uid+asc&q=",
status.uidnext);
if (!solr_add_definite_query_args(str, args, and_args)) {
/* can't search this query */
return 0;
}
/* use a separate filter query for selecting the mailbox. it shouldn't
affect the score and there could be some caching benefits too. */
str_append(str, "&fq=%2Buser:");
solr_quote_http(str, box->storage->user->username);
box_name = fts_box_get_root(box, &ns);
str_printfa(str, "+%%2Buidv:%u+%%2Bbox:", status.uidvalidity);
solr_quote_http(str, box_name);
solr_add_ns_query_http(str, backend, ns);
pool = pool_alloconly_create("fts solr search", 1024);
ret = solr_connection_select(solr_conn, str_c(str), pool, &results);
if (ret == 0 && results[0] != NULL) {
array_append_array(&result->definite_uids, &results[0]->uids);
array_append_array(&result->scores, &results[0]->scores);
}
result->scores_sorted = TRUE;
pool_unref(&pool);
return ret;
}
static char *
mailbox_get_id(struct solr_fts_backend *backend, struct mail_namespace *ns,
const char *mailbox, uint32_t uidvalidity)
{
string_t *str = t_str_new(64);
str_printfa(str, "%u\001", uidvalidity);
str_append(str, mailbox);
if (ns != backend->default_ns)
str_printfa(str, "\001%s", ns->prefix);
return str_c_modifiable(str);
}
static int
solr_search_multi(struct solr_fts_backend *backend, string_t *str,
struct mailbox *const boxes[],
struct fts_multi_result *result)
{
struct solr_result **solr_results;
struct fts_result *fts_result;
ARRAY_DEFINE(fts_results, struct fts_result);
struct mail_namespace *ns;
struct mailbox_status status;
struct hash_table *mailboxes;
struct mailbox *box;
const char *box_name;
char *box_id;
unsigned int i, len;
/* use a separate filter query for selecting the mailbox. it shouldn't
affect the score and there could be some caching benefits too. */
str_append(str, "&fq=%2Buser:");
if (backend->backend.ns->owner != NULL)
solr_quote_http(str, backend->backend.ns->owner->username);
else
str_append(str, "%22%22");
mailboxes = hash_table_create(default_pool, default_pool, 0,
str_hash, (hash_cmp_callback_t *)strcmp);
str_append(str, "%2B(");
len = str_len(str);
for (i = 0; boxes[i] != NULL; i++) {
if (str_len(str) != len)
str_append(str, "+OR+");
box_name = fts_box_get_root(boxes[i], &ns);
mailbox_get_open_status(boxes[i], STATUS_UIDVALIDITY, &status);
str_printfa(str, "%%2B(%%2Buidv:%u+%%2Bbox:", status.uidvalidity);
solr_quote_http(str, box_name);
solr_add_ns_query_http(str, backend, ns);
str_append_c(str, ')');
box_id = mailbox_get_id(backend, ns, box_name, status.uidvalidity);
hash_table_insert(mailboxes, box_id, boxes[i]);
}
str_append_c(str, ')');
if (solr_connection_select(solr_conn, str_c(str),
result->pool, &solr_results) < 0) {
hash_table_destroy(&mailboxes);
return -1;
}
p_array_init(&fts_results, result->pool, 32);
for (i = 0; solr_results[i] != NULL; i++) {
box = hash_table_lookup(mailboxes, solr_results[i]->box_id);
if (box == NULL) {
i_warning("fts_solr: Lookup returned unexpected mailbox "
"with id=%s", solr_results[i]->box_id);
continue;
}
fts_result = array_append_space(&fts_results);
fts_result->box = box;
fts_result->definite_uids = solr_results[i]->uids;
fts_result->scores = solr_results[i]->scores;
fts_result->scores_sorted = TRUE;
}
(void)array_append_space(&fts_results);
result->box_results = array_idx_modifiable(&fts_results, 0);
hash_table_destroy(&mailboxes);
return 0;
}
static int
fts_backend_solr_lookup_multi(struct fts_backend *_backend,
struct mailbox *const boxes[],
struct mail_search_arg *args, bool and_args,
struct fts_multi_result *result)
{
struct solr_fts_backend *backend =
(struct solr_fts_backend *)_backend;
string_t *str;
fts_solr_set_default_ns(backend);
str = t_str_new(256);
str_printfa(str, "fl=ns,box,uidv,uid,score&rows=%u&sort=box+asc,uid+asc&q=",
SOLR_MAX_MULTI_ROWS);
if (solr_add_definite_query_args(str, args, and_args)) {
if (solr_search_multi(backend, str, boxes, result) < 0)
return -1;
}
/* FIXME: maybe_uids could be handled also with some more work.. */
return 0;
}
struct fts_backend fts_backend_solr_old = {
.name = "solr_old",
.flags = 0,
{
fts_backend_solr_alloc,
fts_backend_solr_init,
fts_backend_solr_deinit,
fts_backend_solr_get_last_uid,
fts_backend_solr_update_init,
fts_backend_solr_update_deinit,
fts_backend_solr_update_set_mailbox,
fts_backend_solr_update_expunge,
fts_backend_solr_update_set_build_key,
fts_backend_solr_update_unset_build_key,
fts_backend_solr_update_build_more,
fts_backend_solr_refresh,
NULL,
fts_backend_solr_optimize,
fts_backend_default_can_lookup,
fts_backend_solr_lookup,
fts_backend_solr_lookup_multi,
NULL
}
};