src/scanner.c

   1 /*!@file
   2   \brief   APC Directory Scanner
   3   \details This hand-written parser/scanner traverses a directory tree and
   4            tokenizes elements of the structure which correspond to APC grammar.
   5            The parser is implemented as a 2D stack which populates a list of
   6            child directories at each depth, handling only the leaf nodes
   7            (regular files) of the directory open at the current depth to
   8            conserve memory and speed up traversal.
   9            The scanner works with the lexer to lexically analyze text, and
  10            assumes the existence of an external 'lex' function
  11   \author  Jordan Lavatai
  12   \date    Aug 2016
  13   ----------------------------------------------------------------------------*/
  14 /* Standard */
  15 #include <stdio.h>  //print
  16 #include <errno.h>  //errno
  17 /* Posix */
  18 #include <err.h>    //warnx
  19 #include <stdlib.h> //exit
  20 #include <unistd.h> //chdir
  21 #include <dirent.h> //opendir
  22 #include <unistr.h> //unicode strings
  23 #include <string.h> //strlen
  24 /* Internal */
  25 #include "parser.tab.h"
  26 /* Public */
  27 int   scanner_init(void);
  28 void  scanner_quit(void);
  29 int   scanner_scanpath(char const*);
  30 int   scanner_scandir(DIR*);
  31 yypstate* apc_pstate;
  32 yycstate* apc_cstate;
  33 /* Private */
  34 extern //lexer.rl
  35 int   lexer_lexfile(uint8_t*);
  36 extern //lexer.rl
  37 int   lexer_lexdir(uint8_t*);
  38 extern //lexer.rl
  39 void  lexer_closedir(void);
  40
  41 /* Init
  42    Establishes yy states
  43 */
  44 int scanner_init
  45 ( void )
  46 { if (apc_pstate != NULL || apc_cstate != NULL)
  47     scanner_quit();
  48   apc_pstate = yypstate_new();
  49   apc_cstate = yycstate_new();
  50   return (apc_pstate == NULL || apc_cstate == NULL);
  51 }
  52
  53 /* Quit
  54    Free initialized memory
  55 */
  56 void scanner_quit
  57 ( void )
  58 { yypstate_delete(apc_pstate);
  59   yycstate_delete(apc_cstate);
  60   apc_pstate = NULL;
  61   apc_cstate = NULL;
  62 }
  63
  64 /* Scan the provided path
  65    Changes working directory to the provided pathname and, if successful, sends
  66    a directory stream of the provided path to scanner_scandir
  67 */
  68 int scanner_scanpath
  69 ( char const* pathname )
  70 { DIR* dirp;
  71   errno = 0;
  72   if ((dirp = opendir(pathname)) == NULL || errno)
  73     { fprintf(stderr, "Path %s could not be accessed\n", pathname);
  74       return -1;
  75     }
  76   if (chdir(pathname))
  77     return -1;
  78   return scanner_scandir(dirp);
  79 }
  80
  81 /* Scan directory stream
  82    Recursively scans the provided directory, sending CLOPEN and CLCLOSE tokens
  83    to the parser when entering new directories (classes)
  84 */
  85 int scanner_scandir
  86 ( DIR* dirp )
  87 { DIR* cdirp;
  88   struct dirent* direntp;
  89  scan_next_dirent:
  90   errno = 0;
  91   direntp = readdir(dirp);
  92   if (errno)
  93     goto libfail;
  94   if (direntp != NULL)
  95     { if (*(direntp->d_name) == '.') //skip hidden or relative files
  96         goto scan_next_dirent;
  97       switch (direntp->d_type)
  98         { case DT_REG:
  99             lexer_lexfile((uint8_t*)direntp->d_name);
 100             goto scan_next_dirent;
 101           case DT_DIR:
 102             lexer_lexdir((uint8_t*)direntp->d_name);  //lex the dirname
 103             if (chdir(direntp->d_name))    //change to the specified dir
 104               goto libfail;
 105             errno = 0;
 106             if ((cdirp = opendir(".")) == NULL || errno) //open it
 107               goto libfail;
 108             if(scanner_scandir(cdirp))    //scan the directory
 109               goto libfail;
 110             if (chdir(".."))              //return to the parent dir
 111               goto libfail;
 112             lexer_closedir();            //push "Close Directory" token
 113             goto scan_next_dirent;        //continue scan
 114           case DT_UNKNOWN:
 115             warnx("unknown file %s: ignoring", direntp->d_name);
 116           default:
 117             goto scan_next_dirent;
 118         }
 119     }
 120   return closedir(dirp);
 121  libfail:
 122   perror("scanner_scandir");
 123   return -1;
 124 }