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     /* fixme - this routine expects the string to be
  36        null-terminated */
  37
  38     for (p1=tmp->str; (p=strchr(p1, '\\')); p1 = p+1)
  39     {
  40         switch(p[1])
  41         {
  42             case '\\': p[1] = '\\'; tmp->len--; break;
  43             case '"': p[1] = '"'; tmp->len--; break;
  44             case 'b': p[1] = '\b'; tmp->len--; break;
  45             case 'f': p[1] = '\f'; tmp->len--; break;
  46             case 'n': p[1] = '\n'; tmp->len--; break;
  47             case 'r': p[1] = '\r'; tmp->len--; break;
  48             case 't': p[1] = '\t'; tmp->len--; break;
  49             default:
  50                 continue;
  51         }
  52         strcpy(p, p+1);
  53     }
  54 }
  55
  56 static void store(enum type_t type, int line, int column, char*text, int length)
  57 {
  58     struct token_t token;
  59     string_t tmp;
  60     token.type = type;
  61     token.line = line;
  62     token.column = column;
  63     //printf("->%d(%s) %s\n", type, type_names[type], text);fflush(stdout);
  64
  65     token.text = 0;
  66     switch(type) {
  67         case END:
  68             string_set2(&tmp, "", 0);
  69             token.text = (char*)mem_putstring(&strings, tmp);
  70         break;
  71         case STRING:
  72             string_set2(&tmp, text+1, length-2);
  73             unescapeString(&tmp);
  74             token.text = (char*)mem_putstring(&strings, tmp);
  75         break;
  76         case TWIP:
  77         case NUMBER:
  78         case IDENTIFIER:
  79             string_set2(&tmp, text, length);
  80             if(prefix) {
  81                 //strcat
  82                 token.text = (char*)mem_put(&strings, prefix, strlen(prefix));
  83                 mem_putstring(&strings, tmp);
  84             } else {
  85                 token.text = (char*)mem_putstring(&strings, tmp);
  86             }
  87             prefix = 0;
  88         break;
  89         case LABEL:
  90             string_set2(&tmp, text, length-1);
  91             token.text = (char*)mem_putstring(&strings, tmp);
  92         break;
  93         case COMMAND:
  94             string_set2(&tmp, text+1, length-1);
  95             token.text = (char*)mem_putstring(&strings, tmp);
  96         break;
  97         case ASSIGNMENT: {
  98             char*x = &text[length-1];
  99             if(x[-1] == '-' || x[-1] == '+')
 100                 x--;
 101             do{x--;} while(*x==32 || *x==10 || *x==13 || *x=='\t');
 102             x++; //first space
 103             string_set2(&tmp, text, x-text);
 104             token.text = (char*)mem_putstring(&strings, tmp);
 105             /*char*y,*x = strchr(text, '=');
 106             if(!x) exit(1);
 107             y=x;
 108             do{y--;} while(*y==32 || *y==10 || *y==13 || *y=='\t');
 109             do{x++;} while(*x==32 || *x==10 || *x==13 || *x=='\t');
 110             token.text1 = (char*)put(&strings, text, y-text + 1, 1);
 111             token.text2 = (char*)put(&strings, x, length-(x-text), 1);*/
 112         } break;
 113     }
 114
 115     mem_put(&tokens, &token, sizeof(struct token_t));
 116     prefix = 0;
 117 }
 118
 119 #define MAX_INCLUDE_DEPTH 16
 120 YY_BUFFER_STATE include_stack[MAX_INCLUDE_DEPTH];
 121 int include_stack_ptr = 0;
 122
 123 void handleInclude(char*text, int len)
 124 {
 125     text+=9;len-=9;
 126     while(len >=1 && (text[0] == ' ' || text[0] == '\t')) {
 127         text++;len--;
 128     }
 129     while(len >= 1 && (text[len-1] == ' ' || text[len-1] == '\n')) {
 130         len--;
 131     }
 132     if(len >= 2 && text[0] == '"' && text[len-1] == '"') {
 133         text++; len-=2;
 134     }
 135     text[len] = 0;
 136     if(include_stack_ptr >= MAX_INCLUDE_DEPTH) {
 137         fprintf( stderr, "Includes nested too deeply" );
 138         exit( 1 );
 139     }
 140     include_stack[include_stack_ptr++] = YY_CURRENT_BUFFER;
 141     yyin = fopen(text, "rb");
 142     if (!yyin) {
 143         fprintf(stderr, "Couldn't open %s\n", text);
 144         exit(1);
 145     }
 146     yy_switch_to_buffer(
 147         yy_create_buffer( yyin, YY_BUF_SIZE ) );
 148     BEGIN(INITIAL);
 149 }
 150
 151 #define c() {count(yytext, yyleng, YY_START);}
 152 #define s(type) {store(type, line, column, yytext, yyleng);}
 153 %}
 154
 155 %s R
 156 %x BINARY
 157
 158 NAME     [a-zA-Z_./](-*[a-zA-Z0-9_./])*
 159 TWIP     ([0-9]+(\.([0-9]([05])?)?)?)
 160 NUMBER   [0-9]+(\.[0-9]*)?
 161 PERCENT  {NUMBER}%
 162 STRING   (\\.|[^\\"\n])*
 163 S        [ \n\r\t]
 164 RVALUE   \"{STRING}\"|([^ \n\r\t]+)
 165
 166 %%
 167
 168 <BINARY>\] {c();BEGIN(0);}
 169 <BINARY>.  {c();}
 170 <BINARY>\n {c();}
 171 {TWIP}/[ \n\r\t]            {s(TWIP);c();BEGIN(0);}
 172 {NUMBER}/[ \n\r\t]          {s(NUMBER);c();BEGIN(0);}
 173 ^#[^\n]*\n                  {c();}
 174 [ \t\r]#[^\n]*\n            {c();}
 175 \"{STRING}\"                {s(STRING);c();BEGIN(0);}
 176 \"{STRING}$                 {c();printf("unterminated string in line %d: %s\n", line, yytext);exit(1);yyterminate();}
 177 {NAME}{S}*\+=               {s(ASSIGNMENT);prefix="<plus>";c();BEGIN(R);}
 178 {NAME}{S}*-=                {s(ASSIGNMENT);prefix="<minus>";c();BEGIN(R);}
 179 {NAME}{S}*=                 {s(ASSIGNMENT);c();BEGIN(R);}
 180 <R>{ /* values which appear only on the right-hand side of assignments, like: x=50% */
 181     [^ \n\t\r]*             {s(IDENTIFIER);c();BEGIN(0);}
 182 }
 183 \.include{S}.*\n                    {handleInclude(yytext, yyleng);}
 184 \.{NAME}                    {s(COMMAND);c();}
 185 {NAME}{S}*:                 {s(LABEL);c();}
 186 {NAME}                      {s(IDENTIFIER);c();}
 187 "["                         {c();BEGIN(BINARY);}
 188 {S}                         {c();}
 189 .                           {char c,c1=yytext[0];
 190                              printf("Syntax error in line %d, %d: %s", line, column, yytext);
 191                              while(1) {
 192                                  c=input();
 193                                  if(c=='\n' || c==EOF)
 194                                      break;
 195                                 printf("%c", c);
 196                              }
 197                              if(c1>='0' && c1<='9')
 198                                  printf(" (identifiers must not start with a digit)");
 199                              printf("\n");
 200                              exit(1);
 201                              yyterminate();
 202                             }
 203 <<EOF>>                     {c();
 204                              if ( --include_stack_ptr < 0 ) {
 205                                 s(END);
 206                                 yyterminate();
 207                              } else {
 208                                  yy_delete_buffer( YY_CURRENT_BUFFER );
 209                                  yy_switch_to_buffer(
 210                                       include_stack[include_stack_ptr] );
 211                              }
 212                             }
 213
 214 %%
 215
 216 int yywrap()
 217 {
 218     return 1;
 219 }
 220
 221 void freeTokens(struct token_t*file)
 222 {
 223     mem_clear(&strings);
 224     mem_clear(&tokens);
 225 }
 226
 227 struct token_t* generateTokens(char*filename)
 228 {
 229     FILE*fi = fopen(filename, "rb");
 230     int t;
 231     struct token_t*result;
 232     int num;
 233     if(!fi) {
 234         printf("Couldn't find file %s\n", filename);
 235         return 0;
 236     }
 237     yyin = fi;
 238
 239     mem_init(&strings);
 240     mem_init(&tokens);
 241     mem_put(&strings, &t, 1); //hack- make all valid strings start at position >0
 242
 243     line=1;
 244     column=1;
 245
 246     yylex();
 247     yy_delete_buffer(yy_current_buffer);
 248
 249     result = (struct token_t*)tokens.buffer;
 250     num = tokens.pos/sizeof(struct token_t);
 251
 252     for(t=0;t<tokens.pos/sizeof(struct token_t);t++) {
 253         if(result[t].text)
 254             result[t].text += (int)strings.buffer;
 255     }
 256
 257     fclose(fi);
 258     return result;
 259 }
 260