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 <dirent.h>
  22 /* Local */
  23 #include "parser.tab.h"
  24 #ifndef DE_STACKSIZE
  25 #define DE_STACKSIZE 1024
  26 #endif
  27 #ifndef TK_STACKSIZE
  28 #define TK_STACKSIZE 1024
  29 #endif
  30 /* Public */
  31 int  lexer_init(void);
  32 int  lexer(void);
  33 void lexer_pushtok(int, YYSTYPE);
  34 extern //ragel
  35 int lexer_lex(const char*);
  36 struct dirent* lexer_direntpa[DE_STACKSIZE];
  37 /* Private */
  38 extern //scanner.c
  39 int scanner_init(void);
  40 extern //scanner.c
  41 int scanner(void);
  42 static inline
  43 int dredge_current_depth(void);
  44 extern //bison
  45 YYSTYPE yylval;
  46 static
  47 struct tok
  48 { union YYSTYPE val; //token val
  49   int tt; //token type
  50 } token_stack[TK_STACKSIZE];
  51 static
  52 union tokp
  53 { int*        tpt; //token pointer type
  54   struct tok* tok;
  55   union YYSTYPE* tvp; //token value pointer
  56 } tks, tkx;
  57 static
  58 struct dirent** dps;
  59
  60 /* Directory Entity Array/Stack
  61    Simple array for keeping track of dirents yet to be processed by the scanner.
  62    If this list is empty and there are no tokens, the lexer is done.
  63    This array is populated by the scanner as an array, and popped locally by the
  64    lexer as a stack.
  65 */
  66 #define DE_STACK    (lexer_direntpa)
  67 #define DE_STACKP   (dps)
  68 #define DE_LEN()    (DE_STACKP - DE_STACK)
  69 #define DE_INIT()   (DE_STACKP = DE_STACK)
  70 #define DE_POP()    (*--DE_STACKP)
  71
  72 /* Token Stack
  73    This is a FIFO stack whose pointers are a union of either a pointer to an
  74    integer, or a pointer to two integers (a struct tok).  This way, integers may
  75    be added or removed from the stack either singularly (IPUSH/IPOP), or as a
  76    full token of two integers (PUSH/POP).
  77    An alignment error will occur if IPOP or IPUSH are used a non-even number of
  78    times in a sequence!
  79 */
  80 #define TK_STACK     (token_stack)
  81 #define TK_STACKP    (tks.tok)
  82 #define TK_STACKPI   (tks.tpt)
  83 #define TK_STACKPL   (tks.tvp)
  84 #define TK_STACKX    (tkx.tok)
  85 #define TK_STACKXI   (tkx.tpt)
  86 #define TK_LEN()     (TK_STACKP - TK_STACKX)
  87 #define TK_INIT()    (TK_STACKP = TK_STACKX = TK_STACK)
  88 #define TK_POP()     (*TK_STACKP++)
  89 #define TK_POPI()    (*TK_STACKPI++);
  90 #define TK_POPL()    (*TK_STACKPL++);
  91 #define TK_PUSH(T,L) (*TK_STACKX++ = (struct tok){L,T})
  92
  93 /* Initializer
  94    The initializer returns boolean true if an error occurs, which may be handled   with standard errno.
  95 */
  96 int lexer_init
  97 ()
  98 { TK_INIT();
  99   DE_INIT();
 100   return scanner_init();
 101 }
 102
 103 /* Lexer
 104    If the token buffer is empty, 'lexer' will initialize the token buffer and
 105    call 'lexer_scandir'.  If #SCANDIR_ERROR is returned, an error is printed
 106    before sending a null return to bison.  If 0 tokens are generated, the error
 107    printing is skipped.  In all other cases, 'yylval' is set, and the token's
 108    integer representation is returned.
 109 */
 110 int lexer
 111 #define SCAN_ERROR -1
 112 #define TK_EMPTY   (TK_STACKP == TK_STACKX)
 113 ()
 114 { if (TK_EMPTY)
 115     { TK_INIT();
 116       if (scanner() == 0)
 117         { yylval.val = 0;
 118           return 0;
 119         }
 120     }
 121   yylval = TK_POPL();
 122   return TK_POPI();
 123 }
 124
 125
 126 /* Token Receiver
 127    This receiver takes a struct tok and pushes it to the FIFO stack.
 128 */
 129 void lexer_pushtok
 130 #define S(S)#S //stringifier
 131 #define ERR_TK "Fatal: Generated over " S(TK_STACKSIZE) " tokens in one pass."
 132 ( int tok, YYSTYPE lval )
 133 { if (TK_LEN() >= TK_STACKSIZE)
 134     { fprintf(stderr, ERR_TK);
 135       exit(EXIT_FAILURE);
 136     }
 137   TK_PUSH(tok, lval);
 138 }
 139 /*  init_file:
 140     if (lsp != NULL)
 141     while ((c = *lsp++) == *csp)
 142     { switch (c)
 143     { case DELIM:
 144     delimeters_skipped++;
 145     default:
 146     csp++; //delayed to ensure csp is the start of scannable text
 147     break;
 148     }
 149     }
 150     last_string = string;
 151     scan_text:
 152     return scanner_tokenize(csp);
 153 */