git.notmuchmail.org Git - notmuch/blob - lib/parse-sexp.cc

   1 #include "database-private.h"
   2
   3 #if HAVE_SFSEXP
   4 #include "sexp.h"
   5 #include "unicode-util.h"
   6
   7 /* _sexp is used for file scope symbols to avoid clashing with
   8  * definitions from sexp.h */
   9
  10 typedef enum {
  11     SEXP_FLAG_NONE      = 0,
  12     SEXP_FLAG_FIELD     = 1 << 0,
  13     SEXP_FLAG_BOOLEAN   = 1 << 1,
  14     SEXP_FLAG_SINGLE    = 1 << 2,
  15     SEXP_FLAG_WILDCARD  = 1 << 3,
  16     SEXP_FLAG_REGEX     = 1 << 4,
  17     SEXP_FLAG_DO_REGEX  = 1 << 5,
  18 } _sexp_flag_t;
  19
  20 /*
  21  * define bitwise operators to hide casts */
  22
  23 inline _sexp_flag_t
  24 operator| (_sexp_flag_t a, _sexp_flag_t b)
  25 {
  26     return static_cast<_sexp_flag_t>(
  27         static_cast<unsigned>(a) | static_cast<unsigned>(b));
  28 }
  29
  30 inline _sexp_flag_t
  31 operator& (_sexp_flag_t a, _sexp_flag_t b)
  32 {
  33     return static_cast<_sexp_flag_t>(
  34         static_cast<unsigned>(a) & static_cast<unsigned>(b));
  35 }
  36
  37 typedef struct  {
  38     const char *name;
  39     Xapian::Query::op xapian_op;
  40     Xapian::Query initial;
  41     _sexp_flag_t flags;
  42 } _sexp_prefix_t;
  43
  44 static _sexp_prefix_t prefixes[] =
  45 {
  46     { "and",            Xapian::Query::OP_AND,          Xapian::Query::MatchAll,
  47       SEXP_FLAG_NONE },
  48     { "attachment",     Xapian::Query::OP_AND,          Xapian::Query::MatchAll,
  49       SEXP_FLAG_FIELD | SEXP_FLAG_WILDCARD },
  50     { "body",           Xapian::Query::OP_AND,          Xapian::Query::MatchAll,
  51       SEXP_FLAG_FIELD },
  52     { "from",           Xapian::Query::OP_AND,          Xapian::Query::MatchAll,
  53       SEXP_FLAG_FIELD | SEXP_FLAG_WILDCARD | SEXP_FLAG_REGEX },
  54     { "folder",         Xapian::Query::OP_OR,           Xapian::Query::MatchNothing,
  55       SEXP_FLAG_FIELD | SEXP_FLAG_BOOLEAN | SEXP_FLAG_WILDCARD | SEXP_FLAG_REGEX },
  56     { "id",             Xapian::Query::OP_OR,           Xapian::Query::MatchNothing,
  57       SEXP_FLAG_FIELD | SEXP_FLAG_BOOLEAN | SEXP_FLAG_WILDCARD | SEXP_FLAG_REGEX },
  58     { "is",             Xapian::Query::OP_AND,          Xapian::Query::MatchAll,
  59       SEXP_FLAG_FIELD | SEXP_FLAG_BOOLEAN | SEXP_FLAG_WILDCARD | SEXP_FLAG_REGEX },
  60     { "mid",            Xapian::Query::OP_OR,           Xapian::Query::MatchNothing,
  61       SEXP_FLAG_FIELD | SEXP_FLAG_BOOLEAN | SEXP_FLAG_WILDCARD | SEXP_FLAG_REGEX },
  62     { "mimetype",       Xapian::Query::OP_AND,          Xapian::Query::MatchAll,
  63       SEXP_FLAG_FIELD | SEXP_FLAG_WILDCARD },
  64     { "not",            Xapian::Query::OP_AND_NOT,      Xapian::Query::MatchAll,
  65       SEXP_FLAG_NONE },
  66     { "or",             Xapian::Query::OP_OR,           Xapian::Query::MatchNothing,
  67       SEXP_FLAG_NONE },
  68     { "path",           Xapian::Query::OP_OR,           Xapian::Query::MatchNothing,
  69       SEXP_FLAG_FIELD | SEXP_FLAG_BOOLEAN | SEXP_FLAG_WILDCARD | SEXP_FLAG_REGEX },
  70     { "property",       Xapian::Query::OP_AND,          Xapian::Query::MatchAll,
  71       SEXP_FLAG_FIELD | SEXP_FLAG_BOOLEAN | SEXP_FLAG_WILDCARD | SEXP_FLAG_REGEX },
  72     { "regex",          Xapian::Query::OP_INVALID,      Xapian::Query::MatchAll,
  73       SEXP_FLAG_SINGLE | SEXP_FLAG_DO_REGEX },
  74     { "rx",             Xapian::Query::OP_INVALID,      Xapian::Query::MatchAll,
  75       SEXP_FLAG_SINGLE | SEXP_FLAG_DO_REGEX },
  76     { "starts-with",    Xapian::Query::OP_WILDCARD,     Xapian::Query::MatchAll,
  77       SEXP_FLAG_SINGLE },
  78     { "subject",        Xapian::Query::OP_AND,          Xapian::Query::MatchAll,
  79       SEXP_FLAG_FIELD | SEXP_FLAG_WILDCARD | SEXP_FLAG_REGEX },
  80     { "tag",            Xapian::Query::OP_AND,          Xapian::Query::MatchAll,
  81       SEXP_FLAG_FIELD | SEXP_FLAG_BOOLEAN | SEXP_FLAG_WILDCARD | SEXP_FLAG_REGEX },
  82     { "thread",         Xapian::Query::OP_OR,           Xapian::Query::MatchNothing,
  83       SEXP_FLAG_FIELD | SEXP_FLAG_BOOLEAN | SEXP_FLAG_WILDCARD | SEXP_FLAG_REGEX },
  84     { "to",             Xapian::Query::OP_AND,          Xapian::Query::MatchAll,
  85       SEXP_FLAG_FIELD | SEXP_FLAG_WILDCARD },
  86     { }
  87 };
  88
  89 static notmuch_status_t _sexp_to_xapian_query (notmuch_database_t *notmuch,
  90                                                const _sexp_prefix_t *parent,
  91                                                const sexp_t *sx,
  92                                                Xapian::Query &output);
  93
  94 static notmuch_status_t
  95 _sexp_combine_query (notmuch_database_t *notmuch,
  96                      const _sexp_prefix_t *parent,
  97                      Xapian::Query::op operation,
  98                      Xapian::Query left,
  99                      const sexp_t *sx,
 100                      Xapian::Query &output)
 101 {
 102     Xapian::Query subquery;
 103
 104     notmuch_status_t status;
 105
 106     /* if we run out elements, return accumulator */
 107
 108     if (! sx) {
 109         output = left;
 110         return NOTMUCH_STATUS_SUCCESS;
 111     }
 112
 113     status = _sexp_to_xapian_query (notmuch, parent, sx, subquery);
 114     if (status)
 115         return status;
 116
 117     return _sexp_combine_query (notmuch,
 118                                 parent,
 119                                 operation,
 120                                 Xapian::Query (operation, left, subquery),
 121                                 sx->next, output);
 122 }
 123
 124 static notmuch_status_t
 125 _sexp_parse_phrase (std::string term_prefix, const char *phrase, Xapian::Query &output)
 126 {
 127     Xapian::Utf8Iterator p (phrase);
 128     Xapian::Utf8Iterator end;
 129     std::vector<std::string> terms;
 130
 131     while (p != end) {
 132         Xapian::Utf8Iterator start;
 133         while (p != end && ! Xapian::Unicode::is_wordchar (*p))
 134             p++;
 135
 136         if (p == end)
 137             break;
 138
 139         start = p;
 140
 141         while (p != end && Xapian::Unicode::is_wordchar (*p))
 142             p++;
 143
 144         if (p != start) {
 145             std::string word (start, p);
 146             word = Xapian::Unicode::tolower (word);
 147             terms.push_back (term_prefix + word);
 148         }
 149     }
 150     output = Xapian::Query (Xapian::Query::OP_PHRASE, terms.begin (), terms.end ());
 151     return NOTMUCH_STATUS_SUCCESS;
 152 }
 153
 154 static notmuch_status_t
 155 _sexp_parse_wildcard (notmuch_database_t *notmuch,
 156                       const _sexp_prefix_t *parent,
 157                       std::string match,
 158                       Xapian::Query &output)
 159 {
 160
 161     std::string term_prefix = parent ? _find_prefix (parent->name) : "";
 162
 163     if (parent && ! (parent->flags & SEXP_FLAG_WILDCARD)) {
 164         _notmuch_database_log (notmuch, "'%s' does not support wildcard queries\n", parent->name);
 165         return NOTMUCH_STATUS_BAD_QUERY_SYNTAX;
 166     }
 167
 168     output = Xapian::Query (Xapian::Query::OP_WILDCARD,
 169                             term_prefix + Xapian::Unicode::tolower (match));
 170     return NOTMUCH_STATUS_SUCCESS;
 171 }
 172
 173 static notmuch_status_t
 174 _sexp_parse_one_term (notmuch_database_t *notmuch, std::string term_prefix, const sexp_t *sx,
 175                       Xapian::Query &output)
 176 {
 177     Xapian::Stem stem = *(notmuch->stemmer);
 178
 179     if (sx->aty == SEXP_BASIC && unicode_word_utf8 (sx->val)) {
 180         std::string term = Xapian::Unicode::tolower (sx->val);
 181
 182         output = Xapian::Query ("Z" + term_prefix + stem (term));
 183         return NOTMUCH_STATUS_SUCCESS;
 184     } else {
 185         return _sexp_parse_phrase (term_prefix, sx->val, output);
 186     }
 187
 188 }
 189
 190 notmuch_status_t
 191 _sexp_parse_regex (notmuch_database_t *notmuch,
 192                    const _sexp_prefix_t *prefix, const _sexp_prefix_t *parent,
 193                    std::string val, Xapian::Query &output)
 194 {
 195     if (! parent) {
 196         _notmuch_database_log (notmuch, "illegal '%s' outside field\n",
 197                                prefix->name);
 198         return NOTMUCH_STATUS_BAD_QUERY_SYNTAX;
 199     }
 200
 201     if (! (parent->flags & SEXP_FLAG_REGEX)) {
 202         _notmuch_database_log (notmuch, "'%s' not supported in field '%s'\n",
 203                                prefix->name, parent->name);
 204         return NOTMUCH_STATUS_BAD_QUERY_SYNTAX;
 205     }
 206
 207     std::string msg; /* ignored */
 208
 209     return _notmuch_regexp_to_query (notmuch, Xapian::BAD_VALUENO, parent->name,
 210                                      val, output, msg);
 211 }
 212
 213 /* Here we expect the s-expression to be a proper list, with first
 214  * element defining and operation, or as a special case the empty
 215  * list */
 216
 217 static notmuch_status_t
 218 _sexp_to_xapian_query (notmuch_database_t *notmuch, const _sexp_prefix_t *parent, const sexp_t *sx,
 219                        Xapian::Query &output)
 220 {
 221     if (sx->ty == SEXP_VALUE) {
 222         std::string term_prefix = parent ? _find_prefix (parent->name) : "";
 223
 224         if (sx->aty == SEXP_BASIC && strcmp (sx->val, "*") == 0) {
 225             return _sexp_parse_wildcard (notmuch, parent, "", output);
 226         }
 227
 228         if (parent && (parent->flags & SEXP_FLAG_BOOLEAN)) {
 229             output = Xapian::Query (term_prefix + sx->val);
 230             return NOTMUCH_STATUS_SUCCESS;
 231         }
 232         if (parent) {
 233             return _sexp_parse_one_term (notmuch, term_prefix, sx, output);
 234         } else {
 235             Xapian::Query accumulator;
 236             for (_sexp_prefix_t *prefix = prefixes; prefix->name; prefix++) {
 237                 if (prefix->flags & SEXP_FLAG_FIELD) {
 238                     notmuch_status_t status;
 239                     Xapian::Query subquery;
 240                     term_prefix = _find_prefix (prefix->name);
 241                     status = _sexp_parse_one_term (notmuch, term_prefix, sx, subquery);
 242                     if (status)
 243                         return status;
 244                     accumulator = Xapian::Query (Xapian::Query::OP_OR, accumulator, subquery);
 245                 }
 246             }
 247             output = accumulator;
 248             return NOTMUCH_STATUS_SUCCESS;
 249         }
 250     }
 251
 252     /* Empty list */
 253     if (! sx->list) {
 254         output = Xapian::Query::MatchAll;
 255         return NOTMUCH_STATUS_SUCCESS;
 256     }
 257
 258     if (sx->list->ty == SEXP_LIST) {
 259         _notmuch_database_log (notmuch, "unexpected list in field/operation position\n",
 260                                sx->list->val);
 261         return NOTMUCH_STATUS_BAD_QUERY_SYNTAX;
 262     }
 263
 264     for (_sexp_prefix_t *prefix = prefixes; prefix && prefix->name; prefix++) {
 265         if (strcmp (prefix->name, sx->list->val) == 0) {
 266             if (prefix->flags & SEXP_FLAG_FIELD) {
 267                 if (parent) {
 268                     _notmuch_database_log (notmuch, "nested field: '%s' inside '%s'\n",
 269                                            prefix->name, parent->name);
 270                     return NOTMUCH_STATUS_BAD_QUERY_SYNTAX;
 271                 }
 272                 parent = prefix;
 273             }
 274
 275             if ((prefix->flags & SEXP_FLAG_SINGLE) &&
 276                 (! sx->list->next || sx->list->next->next || sx->list->next->ty != SEXP_VALUE)) {
 277                 _notmuch_database_log (notmuch, "'%s' expects single atom as argument\n",
 278                                        prefix->name);
 279                 return NOTMUCH_STATUS_BAD_QUERY_SYNTAX;
 280             }
 281
 282             if (prefix->xapian_op == Xapian::Query::OP_WILDCARD)
 283                 return _sexp_parse_wildcard (notmuch, parent, sx->list->next->val, output);
 284
 285             if (prefix->flags & SEXP_FLAG_DO_REGEX) {
 286                 return _sexp_parse_regex (notmuch, prefix, parent, sx->list->next->val, output);
 287             }
 288
 289             return _sexp_combine_query (notmuch, parent, prefix->xapian_op, prefix->initial,
 290                                         sx->list->next, output);
 291         }
 292     }
 293
 294     _notmuch_database_log (notmuch, "unknown prefix '%s'\n", sx->list->val);
 295
 296     return NOTMUCH_STATUS_BAD_QUERY_SYNTAX;
 297 }
 298
 299 notmuch_status_t
 300 _notmuch_sexp_string_to_xapian_query (notmuch_database_t *notmuch, const char *querystr,
 301                                       Xapian::Query &output)
 302 {
 303     const sexp_t *sx = NULL;
 304     char *buf = talloc_strdup (notmuch, querystr);
 305
 306     sx = parse_sexp (buf, strlen (querystr));
 307     if (! sx) {
 308         _notmuch_database_log (notmuch, "invalid s-expression: '%s'\n", querystr);
 309         return NOTMUCH_STATUS_BAD_QUERY_SYNTAX;
 310     }
 311
 312     return _sexp_to_xapian_query (notmuch, NULL, sx, output);
 313 }
 314 #endif