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;
  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.
  62 */
  63 #define DE_STACK    (lexer_direntpa)
  64 #define DE_STACKP   (lexer_direntpp)
  65 #define DE_LEN()    (DE_STACKP - DE_STACK)
  66 #define DE_INIT()   (DE_STACKP = DE_STACK)
  67 #define DE_POP()    (*--DE_STACKP)
  68
  69 /* Token Stack
  70    This is a FIFO stack whose pointers are a union of either a pointer to an
  71    integer, or a pointer to two integers (a struct tok).  This way, integers may
  72    be added or removed from the stack either singularly (IPUSH/IPOP), or as a
  73    full token of two integers (PUSH/POP).
  74    An alignment error will occur if IPOP or IPUSH are used a non-even number of
  75    times in a sequence!
  76 */
  77 #define TK_STACK     (token_stack)
  78 #define TK_STACKP    (tsp)
  79 #define TK_STACKX    (tsx)
  80 #define TK_LEN()     (TK_STACKX - TK_STACKP)
  81 #define TK_INIT()    (TK_STACKP = TK_STACKX = TK_STACK)
  82 #define TK_POP()     (*TK_STACKP++)
  83 #define TK_PUSH(T,L) (*TK_STACKX++ = (struct tok){L,T})
  84
  85 /* Initializer
  86    The initializer returns boolean true if an error occurs, which may be handled
  87    with standard errno.
  88 */
  89 int lexer_init
  90 ()
  91 { TK_INIT();
  92   DE_INIT();
  93   return scanner_init();
  94 }
  95
  96 /* Lexer
  97    If the token buffer is empty, 'lexer' will initialize the token buffer and
  98    call 'lexer_scandir'.  If SCAN_ERROR is returned, an error is printed
  99    before sending a null return to bison.  If 0 tokens are generated, the error
 100    printing is skipped.  In all other cases, 'yylval' is set, and the token's
 101    integer representation is returned.
 102 */
 103 int lexer
 104 #define $($)#$
 105 #define SCAN_ERROR -1
 106 #define TK_EMPTY   (TK_STACKP == TK_STACKX)
 107 #define FAIL(...)                               \
 108   do {                                          \
 109     fprintf(stderr,__VA_ARGS__);                \
 110     goto done;                                  \
 111   } while (0)
 112 ()
 113 { struct tok token;
 114  start:
 115   while (DE_LEN() > 0)    //lex any directory entries in our stack
 116     if (lexer_lexfile(DE_POP()->d_name) == 0)
 117       FAIL("Lexer failed to tokenize [%s]\n",(*DE_STACKP)->d_name);
 118   if (TK_EMPTY)           //if there are no tokens,
 119     { TK_INIT();            //initialize the token stack back to 0
 120       switch (scanner())
 121         { case SCAN_ERROR:    //if an error occurred,
 122             FAIL("Scanner error\n");
 123           case 0:             //if the the scanner finds no dirents,
 124             goto done;        //then we are done
 125           default:            //if we found some elements to scan,
 126             goto start;       //start over and lex them
 127         }
 128     }
 129   token = TK_POP();
 130   yylval = token.lval;
 131   return token.tok_t;
 132  done:
 133   yylval.val = 0;
 134   return 0;
 135 }
 136
 137
 138 /* Token Receiver
 139    This receiver takes a struct tok and pushes it to the FIFO stack.
 140 */
 141 void lexer_pushtok
 142 #define $($)#$ //stringifier
 143 #define ERR_TK "Fatal: Generated over " $(TK_STACKSIZE) " tokens in one pass."
 144 ( int tok, YYSTYPE lval )
 145 { if (TK_LEN() >= TK_STACKSIZE)
 146     { fprintf(stderr, ERR_TK);
 147       exit(EXIT_FAILURE);
 148     }
 149   TK_PUSH(tok, lval);
 150 }
 151
 152 /* Lexical analysis of a file
 153    Strips a filename to its base name, then sends it to lexer_lex
 154 */
 155 int lexer_lexfile
 156 #define HIDDEN_WARNING "%s is hidden and will not be parsed!\n", filename
 157 ( const char  *filename
 158 )
 159 { static char fname[NAME_MAX];
 160   char        *last_period = NULL, *iter;
 161
 162   if (*filename == '.')
 163     { fprintf (stderr, HIDDEN_WARNING);
 164       return 0;
 165     }
 166   /* Copy the filename and remove its suffix */
 167   strncpy(fname,filename,NAME_MAX);
 168   last_period = NULL;
 169   for (iter = fname; *iter; iter++)  //find the last '.' char
 170     if (*iter == '.')
 171       last_period = iter;
 172   if (last_period)                   //if we found one,
 173     *last_period = '\0';             //truncate the string there
 174   /* Register the current_filename */
 175   current_filename = filename;
 176
 177   return lexer_lex(fname);
 178 }
 179
 180 char const* lexer_get_current_filepath
 181 ()
 182 { static char        current_path[PATH_MAX];
 183   static char const* last_filename;
 184   if ((!last_filename || last_filename != current_filename) &&
 185       (realpath(current_filename, current_path) != current_path))
 186     { perror("realpath: ");
 187       return NULL;
 188     }
 189   return (const char*)current_path;
 190 }