src/apc/lexer.c

   1 /*!@file
   2   \brief   lexical analyzer implementation for APC
   3   \details The lexer manages two FIFO stacks.  One for maintaining tokens, the
   4            other for maintaining a list of files to be scanned.  During
   5            execution, the lexer will return a token from its token queue if any
   6            are present.  If not, the lexer will will pop an element from its
   7            file queue to 'scanner' to be tokenized.  If the file queue is empty,
   8            the lexer will instead call 'parsedir' to traverse the directory tree
   9            and tokenize the results.  If 'parsedir' does not generate any new
  10            tokens, we are done.
  11   \author  Jordan Lavatai
  12   \date    Aug 2016
  13   ----------------------------------------------------------------------------*/
  14 /* Standard */
  15 #include <stdio.h>
  16 #include <string.h>
  17 #include <errno.h>
  18 /* Posix */
  19 #include <unistd.h>
  20 #include <stdlib.h>
  21 #include <limits.h> //realpath, NAME_MAX, PATH_MAX
  22 #include <dirent.h>
  23 /* Local */
  24 #include "parser.tab.h"
  25 #ifndef DE_STACKSIZE
  26 #define DE_STACKSIZE 1024
  27 #endif
  28 #ifndef TK_STACKSIZE
  29 #define TK_STACKSIZE 1024
  30 #endif
  31 /* Public */
  32 int         lexer_init(void);
  33 int         lexer(void);
  34 int         lexer_lexfile(const char*);
  35 void        lexer_pushtok(int, YYSTYPE);
  36 char const* lexer_get_current_filepath(void);
  37 struct dirent* lexer_direntpa[DE_STACKSIZE],** lexer_direntpp,** lexer_direntpb;
  38 /* Private */
  39 extern //lexer_lex.rl
  40 int         lexer_lex(const char*);
  41 extern //scanner.c
  42 int         scanner_init(void);
  43 extern //scanner.c
  44 int         scanner(void);
  45 static inline
  46 int         dredge_current_depth(void);
  47 extern //bison
  48 YYSTYPE     yylval;
  49 static
  50 char const* current_filename;
  51 static
  52 struct tok
  53 { YYSTYPE lval;  //token val
  54   int     tok_t; //token type
  55 } token_stack[TK_STACKSIZE], *tsp, *tsx;
  56
  57 /* Directory Entity Array/Stack
  58    Simple array for keeping track of dirents yet to be processed by the scanner.
  59    If this list is empty and there are no tokens, the lexer is done.
  60    This array is populated by the scanner as an array, and popped locally by the
  61    lexer as a stack, and is popped as a FIFO stack.
  62 */
  63 #define DE_STACK    (lexer_direntpa)
  64 #define DE_STACKP   (lexer_direntpp)
  65 #define DE_STACKB   (lexer_direntpb)
  66 #define DE_LEN()    (DE_STACKP - DE_STACKB)
  67 #define DE_INIT()   (DE_STACKP = DE_STACKB = DE_STACK)
  68 #define DE_POP()    (*DE_STACKB++)
  69
  70 /* Token Stack
  71    This is a FIFO stack whose pointers are a union of either a pointer to an
  72    integer, or a pointer to two integers (a struct tok).  This way, integers may
  73    be added or removed from the stack either singularly (IPUSH/IPOP), or as a
  74    full token of two integers (PUSH/POP).
  75    An alignment error will occur if IPOP or IPUSH are used a non-even number of
  76    times in a sequence!
  77 */
  78 #define TK_STACK     (token_stack)
  79 #define TK_STACKP    (tsp)
  80 #define TK_STACKX    (tsx)
  81 #define TK_LEN()     (TK_STACKX - TK_STACKP)
  82 #define TK_INIT()    (TK_STACKP = TK_STACKX = TK_STACK)
  83 #define TK_POP()     (*TK_STACKP++)
  84 #define TK_PUSH(T,L) (*TK_STACKX++ = (struct tok){L,T})
  85
  86 /* Initializer
  87    The initializer returns boolean true if an error occurs, which may be handled
  88    with standard errno.
  89 */
  90 int lexer_init
  91 ()
  92 { TK_INIT();
  93   DE_INIT();
  94   current_filename  = NULL;
  95   return scanner_init();
  96 }
  97
  98 /* Lexer
  99    If the token buffer is empty, 'lexer' will initialize the token buffer and
 100    call 'lexer_scandir'.  If SCAN_ERROR is returned, an error is printed
 101    before sending a null return to bison.  If 0 tokens are generated, the error
 102    printing is skipped.  In all other cases, 'yylval' is set, and the token's
 103    integer representation is returned.
 104 */
 105 int lexer
 106 #define $($)#$
 107 #define SCAN_ERROR -1
 108 #define TK_EMPTY   (TK_STACKP == TK_STACKX)
 109 #define FAIL(...)                               \
 110   do {                                          \
 111     fprintf(stderr,__VA_ARGS__);                \
 112     goto done;                                  \
 113   } while (0)
 114 ()
 115 { struct tok token;
 116  start:
 117   while (DE_LEN() > 0)    //lex any directory entries in our stack
 118     if (lexer_lexfile(DE_POP()->d_name) == 0)
 119       FAIL("Lexer failed to tokenize [%s]\n",(*DE_STACKB)->d_name);
 120   if (TK_EMPTY)           //if there are no tokens,
 121     { TK_INIT();            //initialize the token stack back to 0
 122       switch (scanner())
 123         { case SCAN_ERROR:    //if an error occurred,
 124             FAIL("Scanner error\n");
 125           case 0:             //if the the scanner finds no dirents,
 126             goto done;        //then we are done
 127           default:            //if we found some elements to scan,
 128             goto start;       //start over and lex them
 129         }
 130     }
 131   token = TK_POP();
 132   yylval = token.lval;
 133   return token.tok_t;
 134  done:
 135   yylval.val = 0;
 136   return 0;
 137 }
 138
 139
 140 /* Token Receiver
 141    This receiver takes a struct tok and pushes it to the FIFO stack.
 142 */
 143 void lexer_pushtok
 144 #define $($)#$ //stringifier
 145 #define ERR_TK "Fatal: Generated over " $(TK_STACKSIZE) " tokens in one pass."
 146 ( int tok, YYSTYPE lval )
 147 { if (TK_LEN() >= TK_STACKSIZE)
 148     { fprintf(stderr, ERR_TK);
 149       exit(EXIT_FAILURE);
 150     }
 151   TK_PUSH(tok, lval);
 152 }
 153
 154 /* Lexical analysis of a file
 155    Strips a filename to its base name, then sends it to lexer_lex
 156 */
 157 int lexer_lexfile
 158 #define HIDDEN_WARNING "%s is hidden and will not be parsed!\n", filename
 159 ( const char  *filename
 160 )
 161 { static char fname[NAME_MAX];
 162   char        *last_period = NULL, *iter;
 163
 164   if (*filename == '.')
 165     { fprintf (stderr, HIDDEN_WARNING);
 166       return 0;
 167     }
 168   /* Copy the filename and remove its suffix */
 169   strncpy(fname,filename,NAME_MAX);
 170   last_period = NULL;
 171   for (iter = fname; *iter; iter++)  //find the last '.' char
 172     if (*iter == '.')
 173       last_period = iter;
 174   if (last_period)                   //if we found one,
 175     *last_period = '\0';             //truncate the string there
 176   /* Register the current_filename */
 177   current_filename = filename;
 178
 179   return lexer_lex(fname);
 180 }
 181
 182 char const* lexer_get_current_filepath
 183 ()
 184 { static char        current_path[PATH_MAX];
 185   static char const* last_filename;
 186   if ((!last_filename || last_filename != current_filename) &&
 187       (realpath(current_filename, current_path) != current_path))
 188     { perror("realpath: ");
 189       return NULL;
 190     }
 191   return (const char*)current_path;
 192 }