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   return scanner_init();
  95 }
  96
  97 /* Lexer
  98    If the token buffer is empty, 'lexer' will initialize the token buffer and
  99    call 'lexer_scandir'.  If SCAN_ERROR is returned, an error is printed
 100    before sending a null return to bison.  If 0 tokens are generated, the error
 101    printing is skipped.  In all other cases, 'yylval' is set, and the token's
 102    integer representation is returned.
 103 */
 104 int lexer
 105 #define $($)#$
 106 #define SCAN_ERROR -1
 107 #define TK_EMPTY   (TK_STACKP == TK_STACKX)
 108 #define FAIL(...)                               \
 109   do {                                          \
 110     fprintf(stderr,__VA_ARGS__);                \
 111     goto done;                                  \
 112   } while (0)
 113 ()
 114 { struct tok token;
 115  start:
 116   while (DE_LEN() > 0)    //lex any directory entries in our stack
 117     if (lexer_lexfile(DE_POP()->d_name) == 0)
 118       FAIL("Lexer failed to tokenize [%s]\n",(*DE_STACKB)->d_name);
 119   if (TK_EMPTY)           //if there are no tokens,
 120     { TK_INIT();            //initialize the token stack back to 0
 121       switch (scanner())
 122         { case SCAN_ERROR:    //if an error occurred,
 123             FAIL("Scanner error\n");
 124           case 0:             //if the the scanner finds no dirents,
 125             goto done;        //then we are done
 126           default:            //if we found some elements to scan,
 127             goto start;       //start over and lex them
 128         }
 129     }
 130   token = TK_POP();
 131   yylval = token.lval;
 132   return token.tok_t;
 133  done:
 134   yylval.val = 0;
 135   return 0;
 136 }
 137
 138
 139 /* Token Receiver
 140    This receiver takes a struct tok and pushes it to the FIFO stack.
 141 */
 142 void lexer_pushtok
 143 #define $($)#$ //stringifier
 144 #define ERR_TK "Fatal: Generated over " $(TK_STACKSIZE) " tokens in one pass."
 145 ( int tok, YYSTYPE lval )
 146 { if (TK_LEN() >= TK_STACKSIZE)
 147     { fprintf(stderr, ERR_TK);
 148       exit(EXIT_FAILURE);
 149     }
 150   TK_PUSH(tok, lval);
 151 }
 152
 153 /* Lexical analysis of a file
 154    Strips a filename to its base name, then sends it to lexer_lex
 155 */
 156 int lexer_lexfile
 157 #define HIDDEN_WARNING "%s is hidden and will not be parsed!\n", filename
 158 ( const char  *filename
 159 )
 160 { static char fname[NAME_MAX];
 161   char        *last_period = NULL, *iter;
 162
 163   if (*filename == '.')
 164     { fprintf (stderr, HIDDEN_WARNING);
 165       return 0;
 166     }
 167   /* Copy the filename and remove its suffix */
 168   strncpy(fname,filename,NAME_MAX);
 169   last_period = NULL;
 170   for (iter = fname; *iter; iter++)  //find the last '.' char
 171     if (*iter == '.')
 172       last_period = iter;
 173   if (last_period)                   //if we found one,
 174     *last_period = '\0';             //truncate the string there
 175   /* Register the current_filename */
 176   current_filename = filename;
 177
 178   return lexer_lex(fname);
 179 }
 180
 181 char const* lexer_get_current_filepath
 182 ()
 183 { static char        current_path[PATH_MAX];
 184   static char const* last_filename;
 185   if ((!last_filename || last_filename != current_filename) &&
 186       (realpath(current_filename, current_path) != current_path))
 187     { perror("realpath: ");
 188       return NULL;
 189     }
 190   return (const char*)current_path;
 191 }