plugins/fts-solr/fts-backend-solr-old.c

	fts-backend-solr-old.c revision 683d9f20766cd13e72f7d16ce1e341a8811b2679
e8a59a1671127f87e2d22f42e84c572f28299d81Timo Sirainen/* Copyright (c) 2006-2012 Dovecot authors, see the included COPYING file */
e8a59a1671127f87e2d22f42e84c572f28299d81Timo Sirainen
e8a59a1671127f87e2d22f42e84c572f28299d81Timo Sirainen#include "lib.h"
dbb1fb1c51727e2050792f8c333b212e22a36d69Timo Sirainen#include "array.h"
6789ed17e7ca4021713507baf0dcf6979bb42e0cTimo Sirainen#include "str.h"
0cb2e8eb55e70f8ebe1e8349bdf49e4cbe5d8834Timo Sirainen#include "hash.h"
dbb1fb1c51727e2050792f8c333b212e22a36d69Timo Sirainen#include "strescape.h"
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen#include "unichar.h"
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen#include "imap-utf7.h"
e667602217af55105d44d8d9b75f09a8a9ac2f14Timo Sirainen#include "mail-storage-private.h"
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen#include "mailbox-list-private.h"
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen#include "mail-search.h"
90adcaa0a00eba29b7fbd50ca66be11c8d086d6aTimo Sirainen#include "fts-api.h"
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen#include "solr-connection.h"
d67fde1a8ebc1d85704c5986d8f93aae97eccef3Timo Sirainen#include "fts-solr-plugin.h"
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen#include <ctype.h>
55773f17bccf6361d6599ffcbe072d7c9fe205bfTimo Sirainen
55773f17bccf6361d6599ffcbe072d7c9fe205bfTimo Sirainen#define SOLR_CMDBUF_SIZE (1024*64)
55773f17bccf6361d6599ffcbe072d7c9fe205bfTimo Sirainen#define SOLR_MAX_MULTI_ROWS 100000
bb6a0eeab27569790d58a036f67dcd2a965fc539Timo Sirainen
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainenstruct solr_fts_backend {
31be5ed1551c98cddeb2295a594f010aaf4b76bcTimo Sirainen    struct fts_backend backend;
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen    char *id_username, *id_namespace;
31be5ed1551c98cddeb2295a594f010aaf4b76bcTimo Sirainen    struct mail_namespace *default_ns;
31be5ed1551c98cddeb2295a594f010aaf4b76bcTimo Sirainen};
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen
1b4441e3e6f9e78ebeae8218de971959cd55bf60Timo Sirainenstruct solr_fts_backend_update_context {
1b4441e3e6f9e78ebeae8218de971959cd55bf60Timo Sirainen    struct fts_backend_update_context ctx;
1b4441e3e6f9e78ebeae8218de971959cd55bf60Timo Sirainen
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen    struct mailbox *cur_box;
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen    char *id_box_name;
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen    struct solr_connection_post *post;
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen    uint32_t prev_uid, uid_validity;
e667602217af55105d44d8d9b75f09a8a9ac2f14Timo Sirainen    string_t *cmd, *hdr;
e667602217af55105d44d8d9b75f09a8a9ac2f14Timo Sirainen
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen    bool headers_open;
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen    bool body_open;
e667602217af55105d44d8d9b75f09a8a9ac2f14Timo Sirainen    bool documents_added;
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen};
1b4441e3e6f9e78ebeae8218de971959cd55bf60Timo Sirainen
1b4441e3e6f9e78ebeae8218de971959cd55bf60Timo Sirainenstatic bool is_valid_xml_char(unichar_t chr)
1b4441e3e6f9e78ebeae8218de971959cd55bf60Timo Sirainen{
1b4441e3e6f9e78ebeae8218de971959cd55bf60Timo Sirainen    /* Valid characters in XML:
1b4441e3e6f9e78ebeae8218de971959cd55bf60Timo Sirainen
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen       #x9 | #xA | #xD | [#x20-#xD7FF] | [#xE000-#xFFFD] |
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen       [#x10000-#x10FFFF]
55773f17bccf6361d6599ffcbe072d7c9fe205bfTimo Sirainen
55773f17bccf6361d6599ffcbe072d7c9fe205bfTimo Sirainen       This function gets called only for #x80 and higher */
55773f17bccf6361d6599ffcbe072d7c9fe205bfTimo Sirainen    if (chr > 0xd7ff && chr < 0xe000)
55773f17bccf6361d6599ffcbe072d7c9fe205bfTimo Sirainen        return FALSE;
55773f17bccf6361d6599ffcbe072d7c9fe205bfTimo Sirainen    if (chr > 0xfffd && chr < 0x10000)
55773f17bccf6361d6599ffcbe072d7c9fe205bfTimo Sirainen        return FALSE;
55773f17bccf6361d6599ffcbe072d7c9fe205bfTimo Sirainen    return chr < 0x10ffff;
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen}
e192a3b1ca8ae857e7d87298ea507d32977ba570Timo Sirainen
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainenstatic void
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainenxml_encode_data(string_t *dest, const unsigned char *data, unsigned int len)
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen{
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen    unichar_t chr;
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen    unsigned int i;
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen    for (i = 0; i < len; i++) {
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen        switch (data[i]) {
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen        case '&':
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen            str_append(dest, "&amp;");
dbb1fb1c51727e2050792f8c333b212e22a36d69Timo Sirainen            break;
dbb1fb1c51727e2050792f8c333b212e22a36d69Timo Sirainen        case '<':
e8a59a1671127f87e2d22f42e84c572f28299d81Timo Sirainen            str_append(dest, "&lt;");
16c89b1260c9d07c01c83a9219424d3727069b2eTimo Sirainen            break;
90adcaa0a00eba29b7fbd50ca66be11c8d086d6aTimo Sirainen        case '>':
d67fde1a8ebc1d85704c5986d8f93aae97eccef3Timo Sirainen            str_append(dest, "&gt;");
e8a59a1671127f87e2d22f42e84c572f28299d81Timo Sirainen            break;
6789ed17e7ca4021713507baf0dcf6979bb42e0cTimo Sirainen        case '\t':
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen        case '\n':
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen        case '\r':
6789ed17e7ca4021713507baf0dcf6979bb42e0cTimo Sirainen            /* exceptions to the following control char check */
e8a59a1671127f87e2d22f42e84c572f28299d81Timo Sirainen            str_append_c(dest, data[i]);
33d63688ed8b26dc333e3c2edbfb2fe6e412604dTimo Sirainen            break;
645f258ea29afaf09b673fc65d1bd788dfec8db8Timo Sirainen        default:
e8a59a1671127f87e2d22f42e84c572f28299d81Timo Sirainen            if (data[i] < 32) {
0cb2e8eb55e70f8ebe1e8349bdf49e4cbe5d8834Timo Sirainen                /* SOLR doesn't like control characters.
0cb2e8eb55e70f8ebe1e8349bdf49e4cbe5d8834Timo Sirainen                   replace them with spaces. */
e667602217af55105d44d8d9b75f09a8a9ac2f14Timo Sirainen                str_append_c(dest, ' ');
55773f17bccf6361d6599ffcbe072d7c9fe205bfTimo Sirainen            } else if (data[i] >= 0x80) {
6389aeec8c26b585e583c364b48ad12adf741898Timo Sirainen                /* make sure the character is valid for XML
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen                   so we don't get XML parser errors */
96541d31299bb40b5a6efdbf9b4cb3d4f4b4a069Timo Sirainen                unsigned int char_len =
96541d31299bb40b5a6efdbf9b4cb3d4f4b4a069Timo Sirainen                    uni_utf8_char_bytes(data[i]);
644268f7848a7c4221146d0b11feb8ed5bbed233Timo Sirainen                if (i + char_len <= len &&
8d80659e504ffb34bb0c6a633184fece35751b18Timo Sirainen                    uni_utf8_get_char_n(data + i, char_len, &chr) == 1 &&
e8a59a1671127f87e2d22f42e84c572f28299d81Timo Sirainen                    is_valid_xml_char(chr))
1bdda5c0c30463160c47151537e6bb2c6c994841Timo Sirainen                    str_append_n(dest, data + i, char_len);
55773f17bccf6361d6599ffcbe072d7c9fe205bfTimo Sirainen                else {
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen                    str_append_n(dest, utf8_replacement_char,
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen                             UTF8_REPLACEMENT_CHAR_LEN);
bb6a0eeab27569790d58a036f67dcd2a965fc539Timo Sirainen                }
a58e4aec412a30352d9d45e63726cac044aa6aa5Timo Sirainen                i += char_len - 1;
e8a59a1671127f87e2d22f42e84c572f28299d81Timo Sirainen            } else {
e8a59a1671127f87e2d22f42e84c572f28299d81Timo Sirainen                str_append_c(dest, data[i]);
1b4441e3e6f9e78ebeae8218de971959cd55bf60Timo Sirainen            }
2ebeb22b9a8a8bb7fbe2f2e2908478a220792b87Timo Sirainen            break;
e8a59a1671127f87e2d22f42e84c572f28299d81Timo Sirainen        }
e8a59a1671127f87e2d22f42e84c572f28299d81Timo Sirainen    }
e8a59a1671127f87e2d22f42e84c572f28299d81Timo Sirainen}
fdc557286bc9f92c5f3bb49096ff6e2bcec0ea79Timo Sirainen
e8a59a1671127f87e2d22f42e84c572f28299d81Timo Sirainenstatic void xml_encode(string_t *dest, const char *str)
e8a59a1671127f87e2d22f42e84c572f28299d81Timo Sirainen{
fdc557286bc9f92c5f3bb49096ff6e2bcec0ea79Timo Sirainen    xml_encode_data(dest, (const unsigned char *)str, strlen(str));
e8a59a1671127f87e2d22f42e84c572f28299d81Timo Sirainen}
0cb2e8eb55e70f8ebe1e8349bdf49e4cbe5d8834Timo Sirainen
33d63688ed8b26dc333e3c2edbfb2fe6e412604dTimo Sirainenstatic const char *solr_escape_id_str(const char *str)
e8a59a1671127f87e2d22f42e84c572f28299d81Timo Sirainen{
e8a59a1671127f87e2d22f42e84c572f28299d81Timo Sirainen    string_t *tmp;
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen    const char *p;
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen
c36ec256c1bd1abe1c12e792cf64f0b7e3b3135aTimo Sirainen    for (p = str; *p != '\0'; p++) {
c36ec256c1bd1abe1c12e792cf64f0b7e3b3135aTimo Sirainen        if (*p == '/' || *p == '!')
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen            break;
98dd8e6e81f11f1e6040ca72f4916242d246c863Timo Sirainen    }
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen    if (*p == '\0')
8d80659e504ffb34bb0c6a633184fece35751b18Timo Sirainen        return str;
8d80659e504ffb34bb0c6a633184fece35751b18Timo Sirainen
8d80659e504ffb34bb0c6a633184fece35751b18Timo Sirainen    tmp = t_str_new(64);
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen    for (p = str; *p != '\0'; p++) {
c36ec256c1bd1abe1c12e792cf64f0b7e3b3135aTimo Sirainen        switch (*p) {
c36ec256c1bd1abe1c12e792cf64f0b7e3b3135aTimo Sirainen        case '/':
e8a59a1671127f87e2d22f42e84c572f28299d81Timo Sirainen            str_append(tmp, "!\\");
e8a59a1671127f87e2d22f42e84c572f28299d81Timo Sirainen            break;
fdc557286bc9f92c5f3bb49096ff6e2bcec0ea79Timo Sirainen        case '!':
fdc557286bc9f92c5f3bb49096ff6e2bcec0ea79Timo Sirainen            str_append(tmp, "!!");
fdc557286bc9f92c5f3bb49096ff6e2bcec0ea79Timo Sirainen            break;
fdc557286bc9f92c5f3bb49096ff6e2bcec0ea79Timo Sirainen        default:
fdc557286bc9f92c5f3bb49096ff6e2bcec0ea79Timo Sirainen            str_append_c(tmp, *p);
a12399903f415a7e14c2816cffa2f7a09dcbb097Timo Sirainen            break;
fdc557286bc9f92c5f3bb49096ff6e2bcec0ea79Timo Sirainen        }
e8a59a1671127f87e2d22f42e84c572f28299d81Timo Sirainen    }
77af0bd168cf3e3ddc3ae68abc82bfad7e9b5ff4Timo Sirainen    return str_c(tmp);
6789ed17e7ca4021713507baf0dcf6979bb42e0cTimo Sirainen}
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen
d2b94d25f842cd1b7acaf4dd7de858f7c6a821c9Timo Sirainenstatic void solr_quote(string_t *dest, const char *str)
d2b94d25f842cd1b7acaf4dd7de858f7c6a821c9Timo Sirainen{
d2b94d25f842cd1b7acaf4dd7de858f7c6a821c9Timo Sirainen    str_append_c(dest, '"');
2aecf7be5834e7f6520f8deaad683a6fa1de4d61Timo Sirainen    str_append(dest, str_escape(str));
2aecf7be5834e7f6520f8deaad683a6fa1de4d61Timo Sirainen    str_append_c(dest, '"');
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen}
1bdda5c0c30463160c47151537e6bb2c6c994841Timo Sirainen
dce5a2719df4fc64a8762d2aa94ba98dcf9cd6feTimo Sirainenstatic void solr_quote_http(string_t *dest, const char *str)
dce5a2719df4fc64a8762d2aa94ba98dcf9cd6feTimo Sirainen{
87460b08cb97b31cde640d4975a6aa2c1d0e7226Timo Sirainen    str_append(dest, "%22");
dce5a2719df4fc64a8762d2aa94ba98dcf9cd6feTimo Sirainen    solr_connection_http_escape(solr_conn, dest, str);
dce5a2719df4fc64a8762d2aa94ba98dcf9cd6feTimo Sirainen    str_append(dest, "%22");
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen}
87460b08cb97b31cde640d4975a6aa2c1d0e7226Timo Sirainen
87460b08cb97b31cde640d4975a6aa2c1d0e7226Timo Sirainenstatic void fts_solr_set_default_ns(struct solr_fts_backend *backend)
1bdda5c0c30463160c47151537e6bb2c6c994841Timo Sirainen{
16c89b1260c9d07c01c83a9219424d3727069b2eTimo Sirainen    struct mail_namespace *ns = backend->backend.ns;
16c89b1260c9d07c01c83a9219424d3727069b2eTimo Sirainen    struct fts_solr_user *fuser = FTS_SOLR_USER_CONTEXT(ns->user);
0cb2e8eb55e70f8ebe1e8349bdf49e4cbe5d8834Timo Sirainen    const struct fts_solr_settings *set = &fuser->set;
8fcff4c5b52f24d9c681805fdf06b486f1d0fcbeTimo Sirainen    const char *str;
90adcaa0a00eba29b7fbd50ca66be11c8d086d6aTimo Sirainen
0cb2e8eb55e70f8ebe1e8349bdf49e4cbe5d8834Timo Sirainen    if (backend->default_ns != NULL)
d67fde1a8ebc1d85704c5986d8f93aae97eccef3Timo Sirainen        return;
d67fde1a8ebc1d85704c5986d8f93aae97eccef3Timo Sirainen
0cb2e8eb55e70f8ebe1e8349bdf49e4cbe5d8834Timo Sirainen    if (set->default_ns_prefix != NULL) {
0cb2e8eb55e70f8ebe1e8349bdf49e4cbe5d8834Timo Sirainen        backend->default_ns =
0cb2e8eb55e70f8ebe1e8349bdf49e4cbe5d8834Timo Sirainen            mail_namespace_find_prefix(ns->user->namespaces,
0cb2e8eb55e70f8ebe1e8349bdf49e4cbe5d8834Timo Sirainen                           set->default_ns_prefix);
0cb2e8eb55e70f8ebe1e8349bdf49e4cbe5d8834Timo Sirainen        if (backend->default_ns == NULL) {
0cb2e8eb55e70f8ebe1e8349bdf49e4cbe5d8834Timo Sirainen            i_error("fts_solr: default_ns setting points to "
16c89b1260c9d07c01c83a9219424d3727069b2eTimo Sirainen                "nonexistent namespace");
16c89b1260c9d07c01c83a9219424d3727069b2eTimo Sirainen        }
16c89b1260c9d07c01c83a9219424d3727069b2eTimo Sirainen    }
41e1c7380edda701719d8ce1fb4d465d2ec4c84dTimo Sirainen    if (backend->default_ns == NULL) {
0cb2e8eb55e70f8ebe1e8349bdf49e4cbe5d8834Timo Sirainen        backend->default_ns =
0cb2e8eb55e70f8ebe1e8349bdf49e4cbe5d8834Timo Sirainen            mail_namespace_find_inbox(ns->user->namespaces);
0cb2e8eb55e70f8ebe1e8349bdf49e4cbe5d8834Timo Sirainen    }
e015e2f7e7f48874495f9df8b0dd192b7ffcb5ccTimo Sirainen    while (backend->default_ns->alias_for != NULL)
0cb2e8eb55e70f8ebe1e8349bdf49e4cbe5d8834Timo Sirainen        backend->default_ns = backend->default_ns->alias_for;
0cb2e8eb55e70f8ebe1e8349bdf49e4cbe5d8834Timo Sirainen
5cda0bfea032000c4a51134c748d9efe6614870bTimo Sirainen    if (ns != backend->default_ns) {
97511ac4d7607e1ba64ce151eda3d9b5f9775519Timo Sirainen        str = solr_escape_id_str(ns->prefix);
5cda0bfea032000c4a51134c748d9efe6614870bTimo Sirainen        backend->id_namespace = i_strdup(str);
5cda0bfea032000c4a51134c748d9efe6614870bTimo Sirainen    }
97511ac4d7607e1ba64ce151eda3d9b5f9775519Timo Sirainen}
e54512a5189192fe72d1e2c53927c98c5ac920b4Timo Sirainen
e54512a5189192fe72d1e2c53927c98c5ac920b4Timo Sirainenstatic void fts_box_name_get_root(struct mail_namespace **ns, const char **name)
2ebeb22b9a8a8bb7fbe2f2e2908478a220792b87Timo Sirainen{
2ebeb22b9a8a8bb7fbe2f2e2908478a220792b87Timo Sirainen    struct mail_namespace *orig_ns = *ns;
2ebeb22b9a8a8bb7fbe2f2e2908478a220792b87Timo Sirainen
e8a59a1671127f87e2d22f42e84c572f28299d81Timo Sirainen    while ((*ns)->alias_for != NULL)
        *ns = (*ns)->alias_for;

    if (**name == '\0' && *ns != orig_ns &&
        ((*ns)->flags & NAMESPACE_FLAG_INBOX_USER) != 0) {
        /* ugly workaround to allow selecting INBOX from a Maildir/
           when it's not in the inbox=yes namespace. */
        *name = "INBOX";
    }
}

static const char *
fts_box_get_root(struct mailbox *box, struct mail_namespace **ns_r)
{
    struct mail_namespace *ns = mailbox_get_namespace(box);
    const char *name;

    if (t_imap_utf8_to_utf7(box->name, &name) < 0)
        i_unreached();

    fts_box_name_get_root(&ns, &name);
    *ns_r = ns;
    return name;
}

static struct fts_backend *fts_backend_solr_alloc(void)
{
    struct solr_fts_backend *backend;

    backend = i_new(struct solr_fts_backend, 1);
    backend->backend = fts_backend_solr_old;
    return &backend->backend;
}

static int
fts_backend_solr_init(struct fts_backend *_backend,
              const char **error_r ATTR_UNUSED)
{
    struct solr_fts_backend *backend = (struct solr_fts_backend *)_backend;
    struct fts_solr_user *fuser = FTS_SOLR_USER_CONTEXT(_backend->ns->user);
    const struct fts_solr_settings *set = &fuser->set;
    const char *str;

    if (solr_conn == NULL)
        solr_conn = solr_connection_init(set->url, set->debug);

    str = solr_escape_id_str(_backend->ns->user->username);
    backend->id_username = i_strdup(str);
    return 0;
}

static void fts_backend_solr_deinit(struct fts_backend *_backend)
{
    struct solr_fts_backend *backend = (struct solr_fts_backend *)_backend;

    i_free(backend->id_namespace);
    i_free(backend->id_username);
    i_free(backend);
}

static void
solr_add_ns_query(string_t *str, struct solr_fts_backend *backend,
          struct mail_namespace *ns, bool neg)
{
    while (ns->alias_for != NULL)
        ns = ns->alias_for;

    if (ns == backend->default_ns || *ns->prefix == '\0') {
        if (!neg)
            str_append(str, " -ns:[* TO *]");
        else
            str_append(str, " +ns:[* TO *]");
    } else {
        if (!neg)
            str_append(str, " +ns:");
        else
            str_append(str, " -ns:");
        solr_quote(str, ns->prefix);
    }
}

static void
solr_add_ns_query_http(string_t *str, struct solr_fts_backend *backend,
               struct mail_namespace *ns)
{
    string_t *tmp;

    tmp = t_str_new(64);
    solr_add_ns_query(tmp, backend, ns, FALSE);
    solr_connection_http_escape(solr_conn, str, str_c(tmp));
}

static int
fts_backend_solr_get_last_uid_fallback(struct solr_fts_backend *backend,
                       struct mailbox *box,
                       uint32_t *last_uid_r)
{
    struct mail_namespace *ns;
    struct mailbox_status status;
    struct solr_result **results;
    const struct seq_range *uidvals;
    const char *box_name;
    unsigned int count;
    string_t *str;
    pool_t pool;
    int ret = 0;

    str = t_str_new(256);
    str_append(str, "fl=uid&rows=1&sort=uid+desc&q=");

    box_name = fts_box_get_root(box, &ns);

    mailbox_get_open_status(box, STATUS_UIDVALIDITY, &status);
    str_printfa(str, "uidv:%u+box:", status.uidvalidity);
    solr_quote_http(str, box_name);
    solr_add_ns_query_http(str, backend, ns);
    str_append(str, "+user:");
    solr_quote_http(str, ns->user->username);

    pool = pool_alloconly_create("solr last uid lookup", 1024);
    if (solr_connection_select(solr_conn, str_c(str),
                   pool, &results) < 0)
        ret = -1;
    else if (results[0] == NULL) {
        /* no UIDs */
        *last_uid_r = 0;
    } else {
        uidvals = array_get(&results[0]->uids, &count);
        i_assert(count > 0);
        if (count == 1 && uidvals[0].seq1 == uidvals[0].seq2) {
            *last_uid_r = uidvals[0].seq1;
        } else {
            i_error("fts_solr: Last UID lookup returned multiple rows");
            ret = -1;
        }
    }
    pool_unref(&pool);
    return ret;
}

static int
fts_backend_solr_get_last_uid(struct fts_backend *_backend,
                  struct mailbox *box, uint32_t *last_uid_r)
{
    struct solr_fts_backend *backend =
        (struct solr_fts_backend *)_backend;
    struct fts_index_header hdr;

    if (fts_index_get_header(box, &hdr)) {
        *last_uid_r = hdr.last_indexed_uid;
        return 0;
    }

    /* either nothing has been indexed, or the index was corrupted.
       do it the slow way. */
    if (fts_backend_solr_get_last_uid_fallback(backend, box, last_uid_r) < 0)
        return -1;

    fts_index_set_last_uid(box, *last_uid_r);
    return 0;
}

static struct fts_backend_update_context *
fts_backend_solr_update_init(struct fts_backend *_backend)
{
    struct solr_fts_backend *backend =
        (struct solr_fts_backend *)_backend;
    struct solr_fts_backend_update_context *ctx;

    ctx = i_new(struct solr_fts_backend_update_context, 1);
    ctx->ctx.backend = _backend;
    ctx->cmd = str_new(default_pool, SOLR_CMDBUF_SIZE);
    ctx->hdr = str_new(default_pool, 4096);
    fts_solr_set_default_ns(backend);
    return &ctx->ctx;
}

static void xml_encode_id(struct solr_fts_backend_update_context *ctx,
              string_t *str, uint32_t uid)
{
    struct solr_fts_backend *backend =
        (struct solr_fts_backend *)ctx->ctx.backend;

    if (uid != 0)
        str_printfa(str, "%u/", uid);
    else
        str_append(str, "L/");

    if (backend->id_namespace != NULL) {
        xml_encode(str, backend->id_namespace);
        str_append_c(str, '/');
    }
    str_printfa(str, "%u/", ctx->uid_validity);
    xml_encode(str, backend->id_username);
    str_append_c(str, '/');
    xml_encode(str, ctx->id_box_name);
}

static void
fts_backend_solr_add_doc_prefix(struct solr_fts_backend_update_context *ctx,
                uint32_t uid)
{
    struct solr_fts_backend *backend =
        (struct solr_fts_backend *)ctx->ctx.backend;
    struct mailbox *box = ctx->cur_box;
    struct mail_namespace *ns;
    const char *box_name;

    ctx->documents_added = TRUE;

    str_printfa(ctx->cmd, "<doc>"
            "<field name=\"uid\">%u</field>"
            "<field name=\"uidv\">%u</field>",
            uid, ctx->uid_validity);

    box_name = fts_box_get_root(box, &ns);

    if (ns != backend->default_ns) {
        str_append(ctx->cmd, "<field name=\"ns\">");
        xml_encode(ctx->cmd, ns->prefix);
        str_append(ctx->cmd, "</field>");
    }
    str_append(ctx->cmd, "<field name=\"box\">");
    xml_encode(ctx->cmd, box_name);
    str_append(ctx->cmd, "</field><field name=\"user\">");
    xml_encode(ctx->cmd, ns->user->username);
    str_append(ctx->cmd, "</field>");
}

static int
fts_backed_solr_build_commit(struct solr_fts_backend_update_context *ctx)
{
    if (ctx->post == NULL)
        return 0;

    str_append(ctx->cmd, "</doc></add>");

    solr_connection_post_more(ctx->post, str_data(ctx->cmd),
                  str_len(ctx->cmd));
    return solr_connection_post_end(ctx->post);
}

static int
fts_backend_solr_update_deinit(struct fts_backend_update_context *_ctx)
{
    struct solr_fts_backend_update_context *ctx =
        (struct solr_fts_backend_update_context *)_ctx;
    const char *str;
    int ret;

    ret = fts_backed_solr_build_commit(ctx);

    /* commit and wait until the documents we just indexed are
       visible to the following search */
    str = t_strdup_printf("<commit waitFlush=\"false\" "
                  "waitSearcher=\"%s\"/>",
                  ctx->documents_added ? "true" : "false");
    if (solr_connection_post(solr_conn, str) < 0)
        ret = -1;

    str_free(&ctx->cmd);
    str_free(&ctx->hdr);
    i_free(ctx->id_box_name);
    i_free(ctx);
    return ret;
}

static void
fts_backend_solr_update_set_mailbox(struct fts_backend_update_context *_ctx,
                    struct mailbox *box)
{
    struct solr_fts_backend_update_context *ctx =
        (struct solr_fts_backend_update_context *)_ctx;
    struct mailbox_status status;
    struct mail_namespace *ns;

    if (ctx->prev_uid != 0) {
        fts_index_set_last_uid(ctx->cur_box, ctx->prev_uid);
        ctx->prev_uid = 0;
    }

    ctx->cur_box = box;
    ctx->uid_validity = 0;
    i_free_and_null(ctx->id_box_name);

    if (box != NULL) {
        ctx->id_box_name = i_strdup(fts_box_get_root(box, &ns));

        mailbox_get_open_status(box, STATUS_UIDVALIDITY, &status);
        ctx->uid_validity = status.uidvalidity;
    }
}

static void
fts_backend_solr_update_expunge(struct fts_backend_update_context *_ctx,
                uint32_t uid)
{
    struct solr_fts_backend_update_context *ctx =
        (struct solr_fts_backend_update_context *)_ctx;

    T_BEGIN {
        string_t *cmd;

        cmd = t_str_new(256);
        str_append(cmd, "<delete><id>");
        xml_encode_id(ctx, cmd, uid);
        str_append(cmd, "</id></delete>");

        (void)solr_connection_post(solr_conn, str_c(cmd));
    } T_END;
}

static void
fts_backend_solr_uid_changed(struct solr_fts_backend_update_context *ctx,
                 uint32_t uid)
{
    if (ctx->post == NULL) {
        i_assert(ctx->prev_uid == 0);

        ctx->post = solr_connection_post_begin(solr_conn);
        str_append(ctx->cmd, "<add>");
    } else {
        ctx->headers_open = FALSE;
        if (ctx->body_open) {
            ctx->body_open = FALSE;
            str_append(ctx->cmd, "</field>");
        }
        str_append(ctx->cmd, "<field name=\"hdr\">");
        str_append_str(ctx->cmd, ctx->hdr);
        str_append(ctx->cmd, "</field>");
        str_truncate(ctx->hdr, 0);

        str_append(ctx->cmd, "</doc>");
    }
    ctx->prev_uid = uid;

    fts_backend_solr_add_doc_prefix(ctx, uid);
    str_printfa(ctx->cmd, "<field name=\"id\">");
    xml_encode_id(ctx, ctx->cmd, uid);
    str_append(ctx->cmd, "</field>");
}

static bool
fts_backend_solr_update_set_build_key(struct fts_backend_update_context *_ctx,
                      const struct fts_backend_build_key *key)
{
    struct solr_fts_backend_update_context *ctx =
        (struct solr_fts_backend_update_context *)_ctx;

    if (key->uid != ctx->prev_uid)
        fts_backend_solr_uid_changed(ctx, key->uid);

    switch (key->type) {
    case FTS_BACKEND_BUILD_KEY_HDR:
    case FTS_BACKEND_BUILD_KEY_MIME_HDR:
        xml_encode(ctx->hdr, key->hdr_name);
        str_append(ctx->hdr, ": ");
        ctx->headers_open = TRUE;
        break;
    case FTS_BACKEND_BUILD_KEY_BODY_PART:
        ctx->headers_open = FALSE;
        if (!ctx->body_open) {
            ctx->body_open = TRUE;
            str_append(ctx->cmd, "<field name=\"body\">");
        }
        break;
    case FTS_BACKEND_BUILD_KEY_BODY_PART_BINARY:
        i_unreached();
    }
    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_c(ctx->hdr, '\n');
    else {
        i_assert(ctx->body_open);
        str_append_c(ctx->cmd, '\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(struct fts_result) fts_results;
    struct mail_namespace *ns;
    struct mailbox_status status;
    HASH_TABLE(char *, struct mailbox *) 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");

    hash_table_create(&mailboxes, default_pool, 0, str_hash, 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;
    }
    array_append_zero(&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
    }
};