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 int  lexer_lex(const char*);
  34 void lexer_pushtok(int, int);
  35 struct dirent* lexer_direntpa[DE_STACKSIZE];
  36 /* Private */
  37 extern //scanner.c
  38 int scanner_init(void);
  39 extern //scanner.c
  40 int scanner(void);
  41 static inline
  42 int dredge_current_depth(void);
  43 extern //bison
  44 int yylval;
  45 static
  46 struct tok
  47 { int lval;
  48   int tok;
  49 } token_stack[TK_STACKSIZE];
  50 static
  51 union tokp
  52 { int*        i;
  53   struct tok* t;
  54 } tks, tkx;
  55 static
  56 struct dirent** dps;
  57
  58 /* Directory Entity Array/Stack
  59    Simple array for keeping track of dirents yet to be processed by the scanner.
  60    If this list is empty and there are no tokens, the lexer is done.
  61    This array is populated by the scanner as an array, and popped locally by the
  62    lexer as a stack.
  63 */
  64 #define DE_STACK    (lexer_direntpa)
  65 #define DE_STACKP   (dps)
  66 #define DE_LEN()    (DE_STACKP - DE_STACK)
  67 #define DE_INIT()   (DE_STACKP = DE_STACK)
  68 #define DE_POP()    (*--DE_STACKP)
  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    (tks.t)
  80 #define TK_STACKPI   (tks.i)
  81 #define TK_STACKX    (tkx.t)
  82 #define TK_STACKXI   (tkx.i)
  83 #define TK_LEN()     (TK_STACKP - TK_STACKX)
  84 #define TK_INIT()    (TK_STACKP = TK_STACKX = TK_STACK)
  85 #define TK_POP()     (*TK_STACKP++)
  86 #define TK_POPI()    (*TK_STACKPI++);
  87 #define TK_PUSH(T,L) (*TK_STACKX++ = (struct tok){L,T})
  88
  89 /* Initializer
  90    The initializer returns boolean true if an error occurs, which may be handled   with standard errno.
  91 */
  92 int lexer_init
  93 ()
  94 { TK_INIT();
  95   DE_INIT();
  96   return scanner_init();
  97 }
  98
  99 /* Lexer
 100    If the token buffer is empty, 'lexer' will initialize the token buffer and
 101    call 'lexer_scandir'.  If #SCANDIR_ERROR is returned, an error is printed
 102    before sending a null return to bison.  If 0 tokens are generated, the error
 103    printing is skipped.  In all other cases, 'yylval' is set, and the token's
 104    integer representation is returned.
 105 */
 106 int lexer
 107 #define SCAN_ERROR -1
 108 #define TK_EMPTY   (TK_STACKP == TK_STACKX)
 109 ()
 110 { if (TK_EMPTY)
 111     { TK_INIT();
 112       if (scanner() == 0)
 113         { yylval = 0;
 114           return 0;
 115         }
 116     }
 117   yylval = TK_POPI();
 118   return TK_POPI();
 119 }
 120
 121 /* Lexical Analysis
 122    Ragel state machine for tokenizing text.
 123 */
 124 int lexer_lex
 125 (const char* str)
 126 { lexer_pushtok(1, 2);
 127   printf (str);
 128   return 1;
 129 }
 130
 131
 132 /* Token Receiver
 133    This receiver takes a struct tok and pushes it to the FIFO stack.
 134 */
 135 void lexer_pushtok
 136 #define S(S)#S //stringifier
 137 #define ERR_TK "Fatal: Generated over " S(TK_STACKSIZE) " tokens in one pass."
 138 ( int tok, int lval )
 139 { if (TK_LEN() >= TK_STACKSIZE)
 140     { fprintf(stderr, ERR_TK);
 141       exit(EXIT_FAILURE);
 142     }
 143   TK_PUSH(tok, lval);
 144 }
 145 /*  init_file:
 146     if (lsp != NULL)
 147     while ((c = *lsp++) == *csp)
 148     { switch (c)
 149     { case DELIM:
 150     delimeters_skipped++;
 151     default:
 152     csp++; //delayed to ensure csp is the start of scannable text
 153     break;
 154     }
 155     }
 156     last_string = string;
 157     scan_text:
 158     return scanner_tokenize(csp);
 159 */