lexer.l

   1 %{
   2 /*
   3  * Copyright (C) 2013 Jo-Philipp Wich <jow@openwrt.org>
   4  *
   5  * Permission to use, copy, modify, and/or distribute this software for any
   6  * purpose with or without fee is hereby granted, provided that the above
   7  * copyright notice and this permission notice appear in all copies.
   8  *
   9  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
  10  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
  11  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
  12  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
  13  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
  14  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
  15  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
  16  */
  17
  18 #include <ctype.h>
  19
  20 #include "parser.h"
  21
  22 int yylex(struct jp_state *s);
  23
  24 #define YY_DECL int yylex(struct jp_state *s)
  25
  26 static void
  27 str_put(struct jp_state *s, char c)
  28 {
  29         if ((s->str_ptr - s->str_buf + 1) < sizeof(s->str_buf))
  30                 *s->str_ptr++ = c;
  31 }
  32
  33 static void
  34 str_decode(struct jp_state *s, const char *input, int base)
  35 {
  36         int code;
  37         char *end;
  38
  39         code = strtoul(input, &end, base);
  40
  41         if (end == input || *end)
  42                 return;
  43
  44         if (code > 0 && code <= 0x7F)
  45         {
  46                 str_put(s, code);
  47         }
  48         else if (code > 0 && code <= 0x7FF)
  49         {
  50                 str_put(s, ((code >>  6) & 0x1F) | 0xC0);
  51                 str_put(s, ( code        & 0x3F) | 0x80);
  52         }
  53         else if (code > 0 && code <= 0xFFFF)
  54         {
  55                 str_put(s, ((code >> 12) & 0x0F) | 0xE0);
  56                 str_put(s, ((code >>  6) & 0x3F) | 0x80);
  57                 str_put(s, ( code        & 0x3F) | 0x80);
  58         }
  59         else if (code > 0 && code <= 0x10FFFF)
  60         {
  61                 str_put(s, ((code >> 18) & 0x07) | 0xF0);
  62                 str_put(s, ((code >> 12) & 0x3F) | 0x80);
  63                 str_put(s, ((code >>  6) & 0x3F) | 0x80);
  64                 str_put(s, ( code        & 0x3F) | 0x80);
  65         }
  66 }
  67
  68 %}
  69
  70 %option outfile="lexer.c" header-file="lexer.h"
  71 %option noyywrap nounput noinput
  72
  73 DOT                     "."
  74 LABEL           [a-zA-Z_][a-zA-Z0-9_]*
  75
  76 BROPEN          "["
  77 BRCLOSE         "]"
  78 POPEN           "("
  79 PCLOSE          ")"
  80 COMMA           ","
  81
  82 ROOT            "$"
  83 THIS            "@"
  84
  85 LT                      "<"
  86 LE                      "<="
  87 GT                      ">"
  88 GE                      ">="
  89 NE                      "!="
  90 EQ                      "="
  91 NOT                     "!"
  92 AND                     "&&"
  93 OR                      "||"
  94
  95 NUMBER          -?[0-9]+
  96 WILDCARD        "*"
  97 BOOL            (true|false)
  98
  99 WS                      [ \t\n]*
 100
 101 %x                      STRING
 102
 103 %%
 104
 105 ["'] {
 106         s->str_ptr = s->str_buf;
 107         s->str_quote = *yytext;
 108         memset(s->str_buf, 0, sizeof(s->str_buf));
 109         BEGIN(STRING);
 110 }
 111
 112 <STRING>{
 113         ["'] {
 114                 if (*yytext == s->str_quote)
 115                 {
 116                         BEGIN(INITIAL);
 117                         yylval.op = jp_alloc_op(T_STRING, 0, s->str_buf);
 118                         return T_STRING;
 119                 }
 120
 121                 str_put(s, *yytext);
 122         }
 123
 124         \\([0-3][0-7]{1,2}|[0-7]{0,2})  { str_decode(s, yytext + 1, 8); }
 125         \\x[A-Fa-f0-9]{2}                               { str_decode(s, yytext + 2, 16); }
 126         \\u[A-Fa-f0-9]{4}                               { str_decode(s, yytext + 2, 16); }
 127         \\a                                                             { str_put(s, '\a'); }
 128         \\b                                                             { str_put(s, '\b'); }
 129         \\e                                                             { str_put(s, '\e'); }
 130         \\f                                                             { str_put(s, '\f'); }
 131         \\n                                                             { str_put(s, '\n'); }
 132         \\r                                                             { str_put(s, '\r'); }
 133         \\t                                                             { str_put(s, '\t'); }
 134         \\v                                                             { str_put(s, '\v'); }
 135         \\.                                                             { str_put(s, *yytext); }
 136         [^\\"']+                                                { while (*yytext) str_put(s, *yytext++); }
 137 }
 138
 139 {BOOL} {
 140         yylval.op = jp_alloc_op(T_BOOL, (*yytext == 't'), NULL);
 141         return T_BOOL;
 142 }
 143
 144 {NUMBER} {
 145         yylval.op = jp_alloc_op(T_NUMBER, atoi(yytext), NULL);
 146         return T_NUMBER;
 147 }
 148
 149 {LABEL} {
 150         yylval.op = jp_alloc_op(T_LABEL, 0, yytext);
 151         return T_LABEL;
 152 }
 153
 154 {WILDCARD} {
 155         yylval.op = jp_alloc_op(T_WILDCARD, 0, NULL);
 156         return T_WILDCARD;
 157 }
 158
 159 {DOT}           { return T_DOT; }
 160 {BROPEN}        { return T_BROPEN; }
 161 {BRCLOSE}       { return T_BRCLOSE; }
 162 {POPEN}         { return T_POPEN; }
 163 {PCLOSE}        { return T_PCLOSE; }
 164 {COMMA}         { return T_UNION; }
 165
 166 {ROOT}          { return T_ROOT; }
 167 {THIS}          { return T_THIS; }
 168
 169 {LT}            { return T_LT; }
 170 {LE}            { return T_LE; }
 171 {GT}            { return T_GT; }
 172 {GE}            { return T_GE; }
 173 {EQ}            { return T_EQ; }
 174 {NE}            { return T_NE; }
 175 {NOT}           { return T_NOT; }
 176 {AND}           { return T_AND; }
 177 {OR}            { return T_OR; }
 178
 179 {WS}            { }
 180
 181 %%