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_EXPAND    = 1 << 6,
  19     SEXP_FLAG_DO_EXPAND = 1 << 7,
  20 } _sexp_flag_t;
  21
  22 /*
  23  * define bitwise operators to hide casts */
  24
  25 inline _sexp_flag_t
  26 operator| (_sexp_flag_t a, _sexp_flag_t b)
  27 {
  28     return static_cast<_sexp_flag_t>(
  29         static_cast<unsigned>(a) | static_cast<unsigned>(b));
  30 }
  31
  32 inline _sexp_flag_t
  33 operator& (_sexp_flag_t a, _sexp_flag_t b)
  34 {
  35     return static_cast<_sexp_flag_t>(
  36         static_cast<unsigned>(a) & static_cast<unsigned>(b));
  37 }
  38
  39 typedef struct  {
  40     const char *name;
  41     Xapian::Query::op xapian_op;
  42     Xapian::Query initial;
  43     _sexp_flag_t flags;
  44 } _sexp_prefix_t;
  45
  46 static _sexp_prefix_t prefixes[] =
  47 {
  48     { "and",            Xapian::Query::OP_AND,          Xapian::Query::MatchAll,
  49       SEXP_FLAG_NONE },
  50     { "attachment",     Xapian::Query::OP_AND,          Xapian::Query::MatchAll,
  51       SEXP_FLAG_FIELD | SEXP_FLAG_WILDCARD | SEXP_FLAG_EXPAND },
  52     { "body",           Xapian::Query::OP_AND,          Xapian::Query::MatchAll,
  53       SEXP_FLAG_FIELD },
  54     { "from",           Xapian::Query::OP_AND,          Xapian::Query::MatchAll,
  55       SEXP_FLAG_FIELD | SEXP_FLAG_WILDCARD | SEXP_FLAG_REGEX | SEXP_FLAG_EXPAND },
  56     { "folder",         Xapian::Query::OP_OR,           Xapian::Query::MatchNothing,
  57       SEXP_FLAG_FIELD | SEXP_FLAG_BOOLEAN | SEXP_FLAG_WILDCARD | SEXP_FLAG_REGEX | SEXP_FLAG_EXPAND },
  58     { "id",             Xapian::Query::OP_OR,           Xapian::Query::MatchNothing,
  59       SEXP_FLAG_FIELD | SEXP_FLAG_BOOLEAN | SEXP_FLAG_WILDCARD | SEXP_FLAG_REGEX },
  60     { "infix",          Xapian::Query::OP_INVALID,      Xapian::Query::MatchAll,
  61       SEXP_FLAG_SINGLE },
  62     { "is",             Xapian::Query::OP_AND,          Xapian::Query::MatchAll,
  63       SEXP_FLAG_FIELD | SEXP_FLAG_BOOLEAN | SEXP_FLAG_WILDCARD | SEXP_FLAG_REGEX | SEXP_FLAG_EXPAND },
  64     { "matching",       Xapian::Query::OP_AND,          Xapian::Query::MatchAll,
  65       SEXP_FLAG_DO_EXPAND },
  66     { "mid",            Xapian::Query::OP_OR,           Xapian::Query::MatchNothing,
  67       SEXP_FLAG_FIELD | SEXP_FLAG_BOOLEAN | SEXP_FLAG_WILDCARD | SEXP_FLAG_REGEX },
  68     { "mimetype",       Xapian::Query::OP_AND,          Xapian::Query::MatchAll,
  69       SEXP_FLAG_FIELD | SEXP_FLAG_WILDCARD | SEXP_FLAG_EXPAND },
  70     { "not",            Xapian::Query::OP_AND_NOT,      Xapian::Query::MatchAll,
  71       SEXP_FLAG_NONE },
  72     { "of",             Xapian::Query::OP_AND,          Xapian::Query::MatchAll,
  73       SEXP_FLAG_DO_EXPAND },
  74     { "or",             Xapian::Query::OP_OR,           Xapian::Query::MatchNothing,
  75       SEXP_FLAG_NONE },
  76     { "path",           Xapian::Query::OP_OR,           Xapian::Query::MatchNothing,
  77       SEXP_FLAG_FIELD | SEXP_FLAG_BOOLEAN | SEXP_FLAG_WILDCARD | SEXP_FLAG_REGEX },
  78     { "property",       Xapian::Query::OP_AND,          Xapian::Query::MatchAll,
  79       SEXP_FLAG_FIELD | SEXP_FLAG_BOOLEAN | SEXP_FLAG_WILDCARD | SEXP_FLAG_REGEX | SEXP_FLAG_EXPAND },
  80     { "regex",          Xapian::Query::OP_INVALID,      Xapian::Query::MatchAll,
  81       SEXP_FLAG_SINGLE | SEXP_FLAG_DO_REGEX },
  82     { "rx",             Xapian::Query::OP_INVALID,      Xapian::Query::MatchAll,
  83       SEXP_FLAG_SINGLE | SEXP_FLAG_DO_REGEX },
  84     { "starts-with",    Xapian::Query::OP_WILDCARD,     Xapian::Query::MatchAll,
  85       SEXP_FLAG_SINGLE },
  86     { "subject",        Xapian::Query::OP_AND,          Xapian::Query::MatchAll,
  87       SEXP_FLAG_FIELD | SEXP_FLAG_WILDCARD | SEXP_FLAG_REGEX | SEXP_FLAG_EXPAND },
  88     { "tag",            Xapian::Query::OP_AND,          Xapian::Query::MatchAll,
  89       SEXP_FLAG_FIELD | SEXP_FLAG_BOOLEAN | SEXP_FLAG_WILDCARD | SEXP_FLAG_REGEX | SEXP_FLAG_EXPAND },
  90     { "thread",         Xapian::Query::OP_OR,           Xapian::Query::MatchNothing,
  91       SEXP_FLAG_FIELD | SEXP_FLAG_BOOLEAN | SEXP_FLAG_WILDCARD | SEXP_FLAG_REGEX | SEXP_FLAG_EXPAND },
  92     { "to",             Xapian::Query::OP_AND,          Xapian::Query::MatchAll,
  93       SEXP_FLAG_FIELD | SEXP_FLAG_WILDCARD | SEXP_FLAG_EXPAND },
  94     { }
  95 };
  96
  97 static notmuch_status_t _sexp_to_xapian_query (notmuch_database_t *notmuch,
  98                                                const _sexp_prefix_t *parent,
  99                                                const sexp_t *sx,
 100                                                Xapian::Query &output);
 101
 102 static notmuch_status_t
 103 _sexp_combine_query (notmuch_database_t *notmuch,
 104                      const _sexp_prefix_t *parent,
 105                      Xapian::Query::op operation,
 106                      Xapian::Query left,
 107                      const sexp_t *sx,
 108                      Xapian::Query &output)
 109 {
 110     Xapian::Query subquery;
 111
 112     notmuch_status_t status;
 113
 114     /* if we run out elements, return accumulator */
 115
 116     if (! sx) {
 117         output = left;
 118         return NOTMUCH_STATUS_SUCCESS;
 119     }
 120
 121     status = _sexp_to_xapian_query (notmuch, parent, sx, subquery);
 122     if (status)
 123         return status;
 124
 125     return _sexp_combine_query (notmuch,
 126                                 parent,
 127                                 operation,
 128                                 Xapian::Query (operation, left, subquery),
 129                                 sx->next, output);
 130 }
 131
 132 static notmuch_status_t
 133 _sexp_parse_phrase (std::string term_prefix, const char *phrase, Xapian::Query &output)
 134 {
 135     Xapian::Utf8Iterator p (phrase);
 136     Xapian::Utf8Iterator end;
 137     std::vector<std::string> terms;
 138
 139     while (p != end) {
 140         Xapian::Utf8Iterator start;
 141         while (p != end && ! Xapian::Unicode::is_wordchar (*p))
 142             p++;
 143
 144         if (p == end)
 145             break;
 146
 147         start = p;
 148
 149         while (p != end && Xapian::Unicode::is_wordchar (*p))
 150             p++;
 151
 152         if (p != start) {
 153             std::string word (start, p);
 154             word = Xapian::Unicode::tolower (word);
 155             terms.push_back (term_prefix + word);
 156         }
 157     }
 158     output = Xapian::Query (Xapian::Query::OP_PHRASE, terms.begin (), terms.end ());
 159     return NOTMUCH_STATUS_SUCCESS;
 160 }
 161
 162 static notmuch_status_t
 163 _sexp_parse_wildcard (notmuch_database_t *notmuch,
 164                       const _sexp_prefix_t *parent,
 165                       std::string match,
 166                       Xapian::Query &output)
 167 {
 168
 169     std::string term_prefix = parent ? _find_prefix (parent->name) : "";
 170
 171     if (parent && ! (parent->flags & SEXP_FLAG_WILDCARD)) {
 172         _notmuch_database_log (notmuch, "'%s' does not support wildcard queries\n", parent->name);
 173         return NOTMUCH_STATUS_BAD_QUERY_SYNTAX;
 174     }
 175
 176     output = Xapian::Query (Xapian::Query::OP_WILDCARD,
 177                             term_prefix + Xapian::Unicode::tolower (match));
 178     return NOTMUCH_STATUS_SUCCESS;
 179 }
 180
 181 static notmuch_status_t
 182 _sexp_parse_one_term (notmuch_database_t *notmuch, std::string term_prefix, const sexp_t *sx,
 183                       Xapian::Query &output)
 184 {
 185     Xapian::Stem stem = *(notmuch->stemmer);
 186
 187     if (sx->aty == SEXP_BASIC && unicode_word_utf8 (sx->val)) {
 188         std::string term = Xapian::Unicode::tolower (sx->val);
 189
 190         output = Xapian::Query ("Z" + term_prefix + stem (term));
 191         return NOTMUCH_STATUS_SUCCESS;
 192     } else {
 193         return _sexp_parse_phrase (term_prefix, sx->val, output);
 194     }
 195
 196 }
 197
 198 notmuch_status_t
 199 _sexp_parse_regex (notmuch_database_t *notmuch,
 200                    const _sexp_prefix_t *prefix, const _sexp_prefix_t *parent,
 201                    std::string val, Xapian::Query &output)
 202 {
 203     if (! parent) {
 204         _notmuch_database_log (notmuch, "illegal '%s' outside field\n",
 205                                prefix->name);
 206         return NOTMUCH_STATUS_BAD_QUERY_SYNTAX;
 207     }
 208
 209     if (! (parent->flags & SEXP_FLAG_REGEX)) {
 210         _notmuch_database_log (notmuch, "'%s' not supported in field '%s'\n",
 211                                prefix->name, parent->name);
 212         return NOTMUCH_STATUS_BAD_QUERY_SYNTAX;
 213     }
 214
 215     std::string msg; /* ignored */
 216
 217     return _notmuch_regexp_to_query (notmuch, Xapian::BAD_VALUENO, parent->name,
 218                                      val, output, msg);
 219 }
 220
 221
 222 static notmuch_status_t
 223 _sexp_expand_query (notmuch_database_t *notmuch,
 224                     const _sexp_prefix_t *prefix, const _sexp_prefix_t *parent,
 225                     const sexp_t *sx, Xapian::Query &output)
 226 {
 227     Xapian::Query subquery;
 228     notmuch_status_t status;
 229     std::string msg;
 230
 231     if (! (parent->flags & SEXP_FLAG_EXPAND)) {
 232         _notmuch_database_log (notmuch, "'%s' unsupported inside '%s'\n", prefix->name, parent->name);
 233         return NOTMUCH_STATUS_BAD_QUERY_SYNTAX;
 234     }
 235
 236     status = _sexp_combine_query (notmuch, NULL, prefix->xapian_op, prefix->initial, sx, subquery);
 237     if (status)
 238         return status;
 239
 240     status = _notmuch_query_expand (notmuch, parent->name, subquery, output, msg);
 241     if (status) {
 242         _notmuch_database_log (notmuch, "error expanding query %s\n", msg.c_str ());
 243     }
 244     return status;
 245 }
 246
 247 static notmuch_status_t
 248 _sexp_parse_infix (notmuch_database_t *notmuch,  const _sexp_prefix_t *parent,
 249                    const sexp_t *sx, Xapian::Query &output)
 250 {
 251     if (parent) {
 252         _notmuch_database_log (notmuch, "'infix' not supported inside '%s'\n", parent->name);
 253         return NOTMUCH_STATUS_BAD_QUERY_SYNTAX;
 254     }
 255     try {
 256         output = notmuch->query_parser->parse_query (sx->val, NOTMUCH_QUERY_PARSER_FLAGS);
 257     } catch (const Xapian::QueryParserError &error) {
 258         _notmuch_database_log (notmuch, "Syntax error in infix query: %s\n", sx->val);
 259         return NOTMUCH_STATUS_BAD_QUERY_SYNTAX;
 260     } catch (const Xapian::Error &error) {
 261         if (! notmuch->exception_reported) {
 262             _notmuch_database_log (notmuch,
 263                                    "A Xapian exception occurred parsing query: %s\n",
 264                                    error.get_msg ().c_str ());
 265             _notmuch_database_log_append (notmuch,
 266                                           "Query string was: %s\n",
 267                                           sx->val);
 268             notmuch->exception_reported = true;
 269             return NOTMUCH_STATUS_XAPIAN_EXCEPTION;
 270         }
 271     }
 272     return NOTMUCH_STATUS_SUCCESS;
 273 }
 274
 275 /* Here we expect the s-expression to be a proper list, with first
 276  * element defining and operation, or as a special case the empty
 277  * list */
 278
 279 static notmuch_status_t
 280 _sexp_to_xapian_query (notmuch_database_t *notmuch, const _sexp_prefix_t *parent, const sexp_t *sx,
 281                        Xapian::Query &output)
 282 {
 283     if (sx->ty == SEXP_VALUE) {
 284         std::string term_prefix = parent ? _find_prefix (parent->name) : "";
 285
 286         if (sx->aty == SEXP_BASIC && strcmp (sx->val, "*") == 0) {
 287             return _sexp_parse_wildcard (notmuch, parent, "", output);
 288         }
 289
 290         if (parent && (parent->flags & SEXP_FLAG_BOOLEAN)) {
 291             output = Xapian::Query (term_prefix + sx->val);
 292             return NOTMUCH_STATUS_SUCCESS;
 293         }
 294         if (parent) {
 295             return _sexp_parse_one_term (notmuch, term_prefix, sx, output);
 296         } else {
 297             Xapian::Query accumulator;
 298             for (_sexp_prefix_t *prefix = prefixes; prefix->name; prefix++) {
 299                 if (prefix->flags & SEXP_FLAG_FIELD) {
 300                     notmuch_status_t status;
 301                     Xapian::Query subquery;
 302                     term_prefix = _find_prefix (prefix->name);
 303                     status = _sexp_parse_one_term (notmuch, term_prefix, sx, subquery);
 304                     if (status)
 305                         return status;
 306                     accumulator = Xapian::Query (Xapian::Query::OP_OR, accumulator, subquery);
 307                 }
 308             }
 309             output = accumulator;
 310             return NOTMUCH_STATUS_SUCCESS;
 311         }
 312     }
 313
 314     /* Empty list */
 315     if (! sx->list) {
 316         output = Xapian::Query::MatchAll;
 317         return NOTMUCH_STATUS_SUCCESS;
 318     }
 319
 320     if (sx->list->ty == SEXP_LIST) {
 321         _notmuch_database_log (notmuch, "unexpected list in field/operation position\n",
 322                                sx->list->val);
 323         return NOTMUCH_STATUS_BAD_QUERY_SYNTAX;
 324     }
 325
 326     for (_sexp_prefix_t *prefix = prefixes; prefix && prefix->name; prefix++) {
 327         if (strcmp (prefix->name, sx->list->val) == 0) {
 328             if (prefix->flags & SEXP_FLAG_FIELD) {
 329                 if (parent) {
 330                     _notmuch_database_log (notmuch, "nested field: '%s' inside '%s'\n",
 331                                            prefix->name, parent->name);
 332                     return NOTMUCH_STATUS_BAD_QUERY_SYNTAX;
 333                 }
 334                 parent = prefix;
 335             }
 336
 337             if ((prefix->flags & SEXP_FLAG_SINGLE) &&
 338                 (! sx->list->next || sx->list->next->next || sx->list->next->ty != SEXP_VALUE)) {
 339                 _notmuch_database_log (notmuch, "'%s' expects single atom as argument\n",
 340                                        prefix->name);
 341                 return NOTMUCH_STATUS_BAD_QUERY_SYNTAX;
 342             }
 343
 344             if (strcmp (prefix->name, "infix") == 0) {
 345                 return _sexp_parse_infix (notmuch, parent, sx->list->next, output);
 346             }
 347
 348             if (prefix->xapian_op == Xapian::Query::OP_WILDCARD)
 349                 return _sexp_parse_wildcard (notmuch, parent, sx->list->next->val, output);
 350
 351             if (prefix->flags & SEXP_FLAG_DO_REGEX) {
 352                 return _sexp_parse_regex (notmuch, prefix, parent, sx->list->next->val, output);
 353             }
 354
 355             if (prefix->flags & SEXP_FLAG_DO_EXPAND) {
 356                 return _sexp_expand_query (notmuch, prefix, parent, sx->list->next, output);
 357             }
 358
 359             return _sexp_combine_query (notmuch, parent, prefix->xapian_op, prefix->initial,
 360                                         sx->list->next, output);
 361         }
 362     }
 363
 364     _notmuch_database_log (notmuch, "unknown prefix '%s'\n", sx->list->val);
 365
 366     return NOTMUCH_STATUS_BAD_QUERY_SYNTAX;
 367 }
 368
 369 notmuch_status_t
 370 _notmuch_sexp_string_to_xapian_query (notmuch_database_t *notmuch, const char *querystr,
 371                                       Xapian::Query &output)
 372 {
 373     const sexp_t *sx = NULL;
 374     char *buf = talloc_strdup (notmuch, querystr);
 375
 376     sx = parse_sexp (buf, strlen (querystr));
 377     if (! sx) {
 378         _notmuch_database_log (notmuch, "invalid s-expression: '%s'\n", querystr);
 379         return NOTMUCH_STATUS_BAD_QUERY_SYNTAX;
 380     }
 381
 382     return _sexp_to_xapian_query (notmuch, NULL, sx, output);
 383 }
 384 #endif