src/frontend/scanner.l

   1 /*
   2  * SysDB - src/frontend/scanner.l
   3  * Copyright (C) 2013 Sebastian 'tokkee' Harl <sh@tokkee.org>
   4  * All rights reserved.
   5  *
   6  * Redistribution and use in source and binary forms, with or without
   7  * modification, are permitted provided that the following conditions
   8  * are met:
   9  * 1. Redistributions of source code must retain the above copyright
  10  *    notice, this list of conditions and the following disclaimer.
  11  * 2. Redistributions in binary form must reproduce the above copyright
  12  *    notice, this list of conditions and the following disclaimer in the
  13  *    documentation and/or other materials provided with the distribution.
  14  *
  15  * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
  16  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
  17  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
  18  * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDERS OR
  19  * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
  20  * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
  21  * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
  22  * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
  23  * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
  24  * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
  25  * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  26  */
  27
  28 %{
  29
  30 #if HAVE_CONFIG_H
  31 #       include "config.h"
  32 #endif /* HAVE_CONFIG_H */
  33
  34 #include "core/data.h"
  35 #include "frontend/connection.h"
  36 #include "frontend/parser.h"
  37 #include "frontend/grammar.h"
  38 #include "utils/error.h"
  39
  40 #include <assert.h>
  41 #include <errno.h>
  42
  43 #include <string.h>
  44 #include <stdlib.h>
  45
  46 #include <time.h>
  47
  48 #define YY_EXTRA_TYPE sdb_fe_yyextra_t *
  49
  50 static struct {
  51         const char *name;
  52         int id;
  53 } reserved_words[] = {
  54         { "ALL",        ALL },
  55         { "AND",        AND },
  56         { "ANY",        ANY },
  57         { "END",        END },
  58         { "FETCH",      FETCH },
  59         { "FILTER",     FILTER },
  60         { "IN",         IN },
  61         { "IS",         IS },
  62         { "LIST",       LIST },
  63         { "LOOKUP",     LOOKUP },
  64         { "MATCHING",   MATCHING },
  65         { "NOT",        NOT },
  66         { "NULL",       NULL_T },
  67         { "OR",         OR },
  68         { "START",      START },
  69         { "TIMESERIES", TIMESERIES },
  70 };
  71
  72 void
  73 sdb_fe_yyerror(YYLTYPE *lval, sdb_fe_yyscan_t scanner, const char *msg);
  74
  75 %}
  76
  77 %option never-interactive
  78 %option reentrant
  79 %option bison-bridge
  80 %option bison-locations
  81 %option 8bit
  82 %option yylineno
  83 %option nodefault
  84 %option noinput
  85 %option nounput
  86 %option noyywrap
  87 %option verbose
  88 %option warn
  89 %option prefix="sdb_fe_yy" outfile="lex.yy.c"
  90
  91 %x CSC
  92
  93 whitespace              ([ \t\n\r\f]+)
  94 simple_comment  ("--"[^\n\r]*)
  95
  96 /*
  97  * C style comments
  98  */
  99 csc_start       \/\*
 100 csc_inside      ([^*/]+|[^*]\/|\*[^/])
 101 csc_end         \*\/
 102
 103 /*
 104  * Strings and identifiers.
 105  */
 106 identifier      ([A-Za-z_][A-Za-z_0-9$]*)
 107 /* TODO: fully support SQL strings */
 108 string          ('([^']|'')*')
 109
 110 /*
 111  * Numeric constants.
 112  */
 113 dec                     ([\+\-]?[0-9]+)
 114 exp                     ([\+\-]?[0-9]+[Ee]\+?[0-9]+)
 115 integer         ({dec}|{exp})
 116 float1          ([\+\-]?[0-9]+\.[0-9]*([Ee][\+\-]?[0-9]+)?)
 117 float2          ([\+\-]?[0-9]*\.[0-9]+([Ee][\+\-]?[0-9]+)?)
 118 float3          ([\+\-]?[0-9]+[Ee]\-[0-9]+)
 119 float4          ([\+\-]?[Ii][Nn][Ff]([Ii][Nn][Ii][Tt][Yy])?)
 120 float5          ([Nn][Aa][Nn])
 121 float           ({float1}|{float2}|{float3}|{float4}|{float5})
 122
 123 /*
 124  * Time constants.
 125  */
 126 date            ([0-9]{4}-[0-9]{2}-[0-9]{2})
 127 time            ([0-9]{1,2}:[0-9]{1,2}(:[0-9]{1,2}(\.[0-9]{1,9})?)?)
 128
 129 %%
 130
 131 {whitespace} |
 132 {simple_comment}        { /* ignore */ }
 133
 134 {csc_start}                     { BEGIN(CSC); }
 135 <CSC>{csc_inside}       { /* ignore */ }
 136 <CSC>{csc_end}          { BEGIN(INITIAL); }
 137 <CSC><<EOF>> {
 138                 sdb_fe_yyerror(yylloc, yyscanner, "unterminated C-style comment");
 139                 return SCANNER_ERROR;
 140         }
 141
 142 {identifier} {
 143                 size_t i;
 144                 for (i = 0; i < SDB_STATIC_ARRAY_LEN(reserved_words); ++i)
 145                         if (! strcasecmp(reserved_words[i].name, yytext))
 146                                 return reserved_words[i].id;
 147
 148                 yylval->str = strdup(yytext);
 149                 return IDENTIFIER;
 150         }
 151 {string} {
 152                 char *quot;
 153                 size_t len;
 154
 155                 /* remove the leading and trailing quote */
 156                 yytext[yyleng - 1] = '\0';
 157                 yylval->str = strdup(yytext + 1);
 158
 159                 quot = yylval->str;
 160                 len = yyleng - 2;
 161                 while ((quot = strstr(quot, "''")) != NULL) {
 162                         memmove(quot, quot + 1, len - (quot - yylval->str) - 1);
 163                         yylval->str[len - 1] = '\0';
 164                         --len;
 165                         ++quot;
 166                 }
 167                 return STRING;
 168         }
 169 {integer} {
 170                 yylval->data.data.integer = (int64_t)strtoll(yytext, NULL, 10);
 171                 yylval->data.type = SDB_TYPE_INTEGER;
 172                 return INTEGER;
 173         }
 174 {float} {
 175                 yylval->data.data.decimal = strtod(yytext, NULL);
 176                 yylval->data.type = SDB_TYPE_DECIMAL;
 177                 return FLOAT;
 178         }
 179
 180 {date} {
 181                 struct tm tm;
 182                 memset(&tm, 0, sizeof(tm));
 183                 if (! strptime(yytext, "%Y-%m-%d", &tm)) {
 184                         char errmsg[1024];
 185                         snprintf(errmsg, sizeof(errmsg),
 186                                 "Failed to parse '%s' as date", yytext);
 187                         sdb_fe_yyerror(yylloc, yyscanner, errmsg);
 188                         return SCANNER_ERROR;
 189                 }
 190                 yylval->datetime = SECS_TO_SDB_TIME(mktime(&tm));
 191                 return DATE;
 192         }
 193 {time} {
 194                 struct tm tm;
 195                 char time[9], ns[10];
 196                 char *tmp;
 197
 198                 memset(&tm, 0, sizeof(tm));
 199                 memset(time, '\0', sizeof(time));
 200                 memset(ns, '0', sizeof(ns));
 201                 ns[sizeof(ns) - 1] = '\0';
 202
 203                 tmp = strchr(yytext, '.');
 204                 if (tmp) {
 205                         size_t i;
 206                         *tmp = '\0';
 207                         ++tmp;
 208                         strncpy(ns, tmp, sizeof(ns));
 209                         for (i = strlen(ns); i < 9; ++i)
 210                                 ns[i] = '0';
 211                 }
 212                 strncpy(time, yytext, sizeof(time));
 213                 if (tmp) {
 214                         /* reset for better error messages */
 215                         --tmp;
 216                         *tmp = '.';
 217                 }
 218
 219                 tmp = strchr(time, ':');
 220                 assert(tmp);
 221                 tmp = strchr(tmp + 1, ':');
 222                 if (! tmp)
 223                         strncat(time, ":00", sizeof(time));
 224
 225                 if (! strptime(time, "%H:%M:%S", &tm)) {
 226                         char errmsg[1024];
 227                         snprintf(errmsg, sizeof(errmsg),
 228                                 "Failed to parse '%s' as time", yytext);
 229                         sdb_fe_yyerror(yylloc, yyscanner, errmsg);
 230                         return SCANNER_ERROR;
 231                 }
 232
 233                 yylval->datetime = SECS_TO_SDB_TIME(mktime(&tm));
 234                 yylval->datetime += (sdb_time_t)strtoll(ns, NULL, 10);
 235                 return TIME;
 236         }
 237
 238 =       { return CMP_EQUAL; }
 239 !=      { return CMP_NEQUAL; }
 240 =~      { return CMP_REGEX; }
 241 !~      { return CMP_NREGEX; }
 242 \<      { return CMP_LT; }
 243 \<=     { return CMP_LE; }
 244 \>=     { return CMP_GE; }
 245 \>      { return CMP_GT; }
 246 \|\|    { return CONCAT; }
 247
 248 .       { /* XXX: */ return yytext[0]; }
 249
 250 %%
 251
 252 sdb_fe_yyscan_t
 253 sdb_fe_scanner_init(const char *str, int len, sdb_fe_yyextra_t *yyext)
 254 {
 255         yyscan_t scanner;
 256
 257         if (! str)
 258                 return NULL;
 259
 260         if (sdb_fe_yylex_init(&scanner)) {
 261                 char errbuf[1024];
 262                 sdb_log(SDB_LOG_ERR, "frontend: yylex_init failed: %s",
 263                         sdb_strerror(errno, errbuf, sizeof(errbuf)));
 264                 return NULL;
 265         }
 266
 267         sdb_fe_yyset_extra(yyext, scanner);
 268
 269         if (len < 0)
 270                 len = strlen(str);
 271
 272         /* the newly allocated buffer state (YY_BUFFER_STATE) is stored inside the
 273          * scanner and, thus, will be freed by yylex_destroy */
 274         sdb_fe_yy_scan_bytes(str, len, scanner);
 275         return scanner;
 276 } /* sdb_fe_scanner_init */
 277
 278 void
 279 sdb_fe_scanner_destroy(sdb_fe_yyscan_t scanner)
 280 {
 281         sdb_fe_yylex_destroy(scanner);
 282 } /* sdb_fe_scanner_destroy */
 283
 284 /* vim: set tw=78 sw=4 ts=4 noexpandtab : */
 285