Code

frontend/grammar: Added support for 'ALL <obj> <cmp> <expr>'.
[sysdb.git] / src / frontend / scanner.l
index cd7ea8513109fd266437b10dfaf5a2428b55aec3..b581e3d613cea435dd22f4c759ea292850ae2675 100644 (file)
 
 %{
 
+#if HAVE_CONFIG_H
+#      include "config.h"
+#endif /* HAVE_CONFIG_H */
+
+#include "core/data.h"
+#include "frontend/connection.h"
 #include "frontend/parser.h"
 #include "frontend/grammar.h"
 #include "utils/error.h"
 
+#include <assert.h>
 #include <errno.h>
 
 #include <string.h>
+#include <stdlib.h>
+
+#include <time.h>
+
+#define YY_EXTRA_TYPE sdb_fe_yyextra_t *
+
+static struct {
+       const char *name;
+       int id;
+} reserved_words[] = {
+       { "ALL",        ALL },
+       { "AND",        AND },
+       { "ANY",        ANY },
+       { "END",        END },
+       { "FETCH",      FETCH },
+       { "FILTER",     FILTER },
+       { "IN",         IN },
+       { "IS",         IS },
+       { "LIST",       LIST },
+       { "LOOKUP",     LOOKUP },
+       { "MATCHING",   MATCHING },
+       { "NOT",        NOT },
+       { "NULL",       NULL_T },
+       { "OR",         OR },
+       { "START",      START },
+       { "TIMESERIES", TIMESERIES },
+};
 
 void
 sdb_fe_yyerror(YYLTYPE *lval, sdb_fe_yyscan_t scanner, const char *msg);
@@ -66,7 +100,31 @@ csc_start   \/\*
 csc_inside     ([^*/]+|[^*]\/|\*[^/])
 csc_end                \*\/
 
+/*
+ * Strings and identifiers.
+ */
 identifier     ([A-Za-z_][A-Za-z_0-9$]*)
+/* TODO: fully support SQL strings */
+string         ('([^']|'')*')
+
+/*
+ * Numeric constants.
+ */
+dec                    ([\+\-]?[0-9]+)
+exp                    ([\+\-]?[0-9]+[Ee]\+?[0-9]+)
+integer                ({dec}|{exp})
+float1         ([\+\-]?[0-9]+\.[0-9]*([Ee][\+\-]?[0-9]+)?)
+float2         ([\+\-]?[0-9]*\.[0-9]+([Ee][\+\-]?[0-9]+)?)
+float3         ([\+\-]?[0-9]+[Ee]\-[0-9]+)
+float4         ([\+\-]?[Ii][Nn][Ff]([Ii][Nn][Ii][Tt][Yy])?)
+float5         ([Nn][Aa][Nn])
+float          ({float1}|{float2}|{float3}|{float4}|{float5})
+
+/*
+ * Time constants.
+ */
+date           ([0-9]{4}-[0-9]{2}-[0-9]{2})
+time           ([0-9]{1,2}:[0-9]{1,2}(:[0-9]{1,2}(\.[0-9]{1,9})?)?)
 
 %%
 
@@ -77,27 +135,128 @@ identifier        ([A-Za-z_][A-Za-z_0-9$]*)
 <CSC>{csc_inside}      { /* ignore */ }
 <CSC>{csc_end}         { BEGIN(INITIAL); }
 <CSC><<EOF>> {
-               sdb_fe_yyerror(yylval, yyscanner, "unterminated C-style comment");
+               sdb_fe_yyerror(yylloc, yyscanner, "unterminated C-style comment");
                return SCANNER_ERROR;
        }
 
 {identifier} {
-               /* XXX */
-               if (! strcasecmp(yytext, "LIST"))
-                       return LIST;
+               size_t i;
+               for (i = 0; i < SDB_STATIC_ARRAY_LEN(reserved_words); ++i)
+                       if (! strcasecmp(reserved_words[i].name, yytext))
+                               return reserved_words[i].id;
 
+               yylval->str = strdup(yytext);
                return IDENTIFIER;
        }
+{string} {
+               char *quot;
+               size_t len;
+
+               /* remove the leading and trailing quote */
+               yytext[yyleng - 1] = '\0';
+               yylval->str = strdup(yytext + 1);
+
+               quot = yylval->str;
+               len = yyleng - 2;
+               while ((quot = strstr(quot, "''")) != NULL) {
+                       memmove(quot, quot + 1, len - (quot - yylval->str) - 1);
+                       yylval->str[len - 1] = '\0';
+                       --len;
+                       ++quot;
+               }
+               return STRING;
+       }
+{integer} {
+               yylval->data.data.integer = (int64_t)strtoll(yytext, NULL, 10);
+               yylval->data.type = SDB_TYPE_INTEGER;
+               return INTEGER;
+       }
+{float} {
+               yylval->data.data.decimal = strtod(yytext, NULL);
+               yylval->data.type = SDB_TYPE_DECIMAL;
+               return FLOAT;
+       }
+
+{date} {
+               struct tm tm;
+               memset(&tm, 0, sizeof(tm));
+               if (! strptime(yytext, "%Y-%m-%d", &tm)) {
+                       char errmsg[1024];
+                       snprintf(errmsg, sizeof(errmsg),
+                               "Failed to parse '%s' as date", yytext);
+                       sdb_fe_yyerror(yylloc, yyscanner, errmsg);
+                       return SCANNER_ERROR;
+               }
+               yylval->datetime = SECS_TO_SDB_TIME(mktime(&tm));
+               return DATE;
+       }
+{time} {
+               struct tm tm;
+               char time[9], ns[10];
+               char *tmp;
 
-.      { /* do nothing for now */ }
+               memset(&tm, 0, sizeof(tm));
+               memset(time, '\0', sizeof(time));
+               memset(ns, '0', sizeof(ns));
+               ns[sizeof(ns) - 1] = '\0';
+
+               tmp = strchr(yytext, '.');
+               if (tmp) {
+                       size_t i;
+                       *tmp = '\0';
+                       ++tmp;
+                       strncpy(ns, tmp, sizeof(ns));
+                       for (i = strlen(ns); i < 9; ++i)
+                               ns[i] = '0';
+               }
+               strncpy(time, yytext, sizeof(time));
+               if (tmp) {
+                       /* reset for better error messages */
+                       --tmp;
+                       *tmp = '.';
+               }
+
+               tmp = strchr(time, ':');
+               assert(tmp);
+               tmp = strchr(tmp + 1, ':');
+               if (! tmp)
+                       strncat(time, ":00", sizeof(time));
+
+               if (! strptime(time, "%H:%M:%S", &tm)) {
+                       char errmsg[1024];
+                       snprintf(errmsg, sizeof(errmsg),
+                               "Failed to parse '%s' as time", yytext);
+                       sdb_fe_yyerror(yylloc, yyscanner, errmsg);
+                       return SCANNER_ERROR;
+               }
+
+               yylval->datetime = SECS_TO_SDB_TIME(mktime(&tm));
+               yylval->datetime += (sdb_time_t)strtoll(ns, NULL, 10);
+               return TIME;
+       }
+
+=      { return CMP_EQUAL; }
+!=     { return CMP_NEQUAL; }
+=~     { return CMP_REGEX; }
+!~     { return CMP_NREGEX; }
+\<     { return CMP_LT; }
+\<=    { return CMP_LE; }
+\>=    { return CMP_GE; }
+\>     { return CMP_GT; }
+\|\|   { return CONCAT; }
+
+.      { /* XXX: */ return yytext[0]; }
 
 %%
 
 sdb_fe_yyscan_t
-sdb_fe_scanner_init(const char *str)
+sdb_fe_scanner_init(const char *str, int len, sdb_fe_yyextra_t *yyext)
 {
        yyscan_t scanner;
 
+       if (! str)
+               return NULL;
+
        if (sdb_fe_yylex_init(&scanner)) {
                char errbuf[1024];
                sdb_log(SDB_LOG_ERR, "frontend: yylex_init failed: %s",
@@ -105,9 +264,14 @@ sdb_fe_scanner_init(const char *str)
                return NULL;
        }
 
+       sdb_fe_yyset_extra(yyext, scanner);
+
+       if (len < 0)
+               len = strlen(str);
+
        /* the newly allocated buffer state (YY_BUFFER_STATE) is stored inside the
         * scanner and, thus, will be freed by yylex_destroy */
-       yy_scan_string(str, scanner);
+       sdb_fe_yy_scan_bytes(str, len, scanner);
        return scanner;
 } /* sdb_fe_scanner_init */