src/parser.lex

   1 %{
   2
   3 #include <string.h>
   4 #include "q.h"
   5 #include "parser.h"
   6
   7 //RVALUE         {NUMBER}|{PERCENT}|{NAME}|\"{STRING}\"|{DIM}
   8 //<a>.                {printf("<a>%s\n", yytext);}
   9 // %x: exclusive, %s: inclusive
  10 char*type_names[] = {"twip","number","command","string","assignment","identifier","label","end"};
  11 static int line=1;
  12 static int column=1;
  13
  14 mem_t strings;
  15 mem_t tokens;
  16
  17 static void count(char*text, int len, int condition)
  18 {
  19     int t;
  20     for(t=0;t<len;t++) {
  21         if(text[t]=='\n') {
  22             line++;
  23             column=1;
  24         } else {
  25             column++;
  26         }
  27     }
  28 }
  29
  30 static char*prefix = 0;
  31
  32 static void unescapeString(string_t * tmp)
  33 {
  34     char *p, *p1;
  35
  36     for (p1=tmp->str; (p=strchr(p1, '\\')) != 0; p1 = p+1)
  37     {
  38         switch(p[1])
  39         {
  40             case '\\': p[1] = '\\'; break;
  41             case 'b': p[1] = '\b'; break;
  42             case 'f': p[1] = '\f'; break;
  43             case 'n': p[1] = '\n'; break;
  44             case 'r': p[1] = '\r'; break;
  45             case 't': p[1] = '\t'; break;
  46             default:
  47                 continue;
  48         }
  49         strcpy(p, p+1);
  50     }
  51 }
  52
  53 static void store(enum type_t type, int line, int column, char*text, int length)
  54 {
  55     struct token_t token;
  56     string_t tmp;
  57     token.type = type;
  58     token.line = line;
  59     token.column = column;
  60     //printf("->%d(%s) %s\n", type, type_names[type], text);fflush(stdout);
  61
  62     token.text = 0;
  63     switch(type) {
  64         case END:
  65             string_set2(&tmp, "", 0);
  66             token.text = (char*)mem_putstring(&strings, tmp);
  67         break;
  68         case STRING:
  69             string_set2(&tmp, text+1, length-2);
  70             unescapeString(&tmp);
  71             token.text = (char*)mem_putstring(&strings, tmp);
  72         break;
  73         case TWIP:
  74         case NUMBER:
  75         case IDENTIFIER:
  76             string_set2(&tmp, text, length);
  77             if(prefix) {
  78                 //strcat
  79                 token.text = (char*)mem_put(&strings, prefix, strlen(prefix));
  80                 mem_putstring(&strings, tmp);
  81             } else {
  82                 token.text = (char*)mem_putstring(&strings, tmp);
  83             }
  84             prefix = 0;
  85         break;
  86         case LABEL:
  87             string_set2(&tmp, text, length-1);
  88             token.text = (char*)mem_putstring(&strings, tmp);
  89         break;
  90         case COMMAND:
  91             string_set2(&tmp, text+1, length-1);
  92             token.text = (char*)mem_putstring(&strings, tmp);
  93         break;
  94         case ASSIGNMENT: {
  95             char*x = &text[length-1];
  96             if(x[-1] == '-' || x[-1] == '+')
  97                 x--;
  98             do{x--;} while(*x==32 || *x==10 || *x==13 || *x=='\t');
  99             x++; //first space
 100             string_set2(&tmp, text, x-text);
 101             token.text = (char*)mem_putstring(&strings, tmp);
 102             /*char*y,*x = strchr(text, '=');
 103             if(!x) exit(1);
 104             y=x;
 105             do{y--;} while(*y==32 || *y==10 || *y==13 || *y=='\t');
 106             do{x++;} while(*x==32 || *x==10 || *x==13 || *x=='\t');
 107             token.text1 = (char*)put(&strings, text, y-text + 1, 1);
 108             token.text2 = (char*)put(&strings, x, length-(x-text), 1);*/
 109         } break;
 110     }
 111
 112     mem_put(&tokens, &token, sizeof(struct token_t));
 113     prefix = 0;
 114 }
 115
 116 #define c() {count(yytext, yyleng, YY_START);}
 117 #define s(type) {store(type, line, column, yytext, yyleng);}
 118 %}
 119
 120 %s R
 121 %x BINARY
 122
 123 NAME     [a-zA-Z_./](-*[a-zA-Z0-9_./])*
 124 TWIP     ([0-9]+(\.([0-9]([05])?)?)?)
 125 NUMBER   [0-9]+(\.[0-9]*)?
 126 PERCENT  {NUMBER}%
 127 STRING   (\\.|[^\\"\n])*
 128 S        [ \n\r\t]
 129 RVALUE   \"{STRING}\"|([^ \n\r\t]+)
 130
 131 %%
 132
 133 <BINARY>\] {c();BEGIN(0);}
 134 <BINARY>.  {c();}
 135 <BINARY>\n {c();}
 136 {TWIP}/[ \n\r\t]            {s(TWIP);c();BEGIN(0);}
 137 {NUMBER}/[ \n\r\t]          {s(NUMBER);c();BEGIN(0);}
 138 ^#[^\n]*\n                  {c();}
 139 [ \t\r]#[^\n]*\n            {c();}
 140 \"{STRING}\"                {s(STRING);c();BEGIN(0);}
 141 \"{STRING}$                 {c();printf("unterminated string in line %d: %s\n", line, yytext);exit(1);yyterminate();}
 142 {NAME}{S}*\+=               {s(ASSIGNMENT);prefix="<plus>";c();BEGIN(R);}
 143 {NAME}{S}*-=                {s(ASSIGNMENT);prefix="<minus>";c();BEGIN(R);}
 144 {NAME}{S}*=                 {s(ASSIGNMENT);c();BEGIN(R);}
 145 <R>{ /* values which appear only on the right-hand side of assignments, like: x=50% */
 146     [^ \n\t\r]*             {s(IDENTIFIER);c();BEGIN(0);}
 147 }
 148 \.{NAME}                    {s(COMMAND);c();}
 149 {NAME}{S}*:                 {s(LABEL);c();}
 150 {NAME}                      {s(IDENTIFIER);c();}
 151 "["                         {c();BEGIN(BINARY);}
 152 {S}                         {c();}
 153 .                           {char c,c1=0;
 154                              printf("Syntax error in line %d, %d: %s", line, column, yytext);
 155                              while(1) {
 156                                  c=input();
 157                                  if(!c1) c1=c;
 158                                  if(c=='\n' || c==EOF)
 159                                      break;
 160                                 printf("%c", c);
 161                              }
 162                              if(c1>='0' && c1<='9')
 163                                  printf(" (identifiers must not start with a digit)");
 164                              printf("\n");
 165                              exit(1);
 166                              yyterminate();
 167                             }
 168 <<EOF>>                     {c();s(END);yyterminate();}
 169 %%
 170
 171 int yywrap()
 172 {
 173     return 1;
 174 }
 175
 176 void freeTokens(struct token_t*file)
 177 {
 178     mem_clear(&strings);
 179     mem_clear(&tokens);
 180 }
 181
 182 struct token_t* generateTokens(char*filename)
 183 {
 184     FILE*fi = fopen(filename, "rb");
 185     int t;
 186     struct token_t*result;
 187     int num;
 188     if(!fi) {
 189         printf("Couldn't find file %s\n", filename);
 190         return 0;
 191     }
 192     yyin = fi;
 193
 194     mem_init(&strings);
 195     mem_init(&tokens);
 196     mem_put(&strings, &t, 1); //hack- make all valid strings start at position >0
 197
 198     line=1;
 199     column=1;
 200
 201     yylex();
 202     yy_delete_buffer(yy_current_buffer);
 203
 204     result = (struct token_t*)tokens.buffer;
 205     num = tokens.pos/sizeof(struct token_t);
 206
 207     for(t=0;t<tokens.pos/sizeof(struct token_t);t++) {
 208         if(result[t].text)
 209             result[t].text += (int)strings.buffer;
 210     }
 211
 212     fclose(fi);
 213     return result;
 214 }
 215