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 /* Internal */
  24 #include "parser.tab.h"
  25 /* Public */
  26 int   scanner_init(void);
  27 void  scanner_quit(void);
  28 int   scanner_scanpath(char const*);
  29 int   scanner_scandir(DIR*);
  30 yypstate* apc_pstate;
  31 yycstate* apc_cstate;
  32 /* Private */
  33 extern //lexer.c
  34 int   lexer_lexfile(uint8_t const*);
  35 extern //lexer.rl
  36 int   lexer_lexstring(uint8_t const*, int);
  37 #define PUSHTOK(T,L) yypush_parse(apc_pstate, T, L, apc_cstate)
  38
  39 /* Init
  40    Establishes yy states
  41 */
  42 int scanner_init
  43 ( void )
  44 { if (apc_pstate != NULL || apc_cstate != NULL)
  45     scanner_quit();
  46   apc_pstate = yypstate_new();
  47   apc_cstate = yycstate_new();
  48   return (apc_pstate == NULL || apc_cstate == NULL);
  49 }
  50
  51 /* Quit
  52    Free initialized memory
  53 */
  54 void scanner_quit
  55 ( void )
  56 { yypstate_delete(apc_pstate);
  57   yycstate_delete(apc_cstate);
  58   apc_pstate = NULL;
  59   apc_cstate = NULL;
  60 }
  61
  62 /* Scan the provided path
  63    Changes working directory to the provided pathname and, if successful, sends
  64    a directory stream of the provided path to scanner_scandir
  65 */
  66 int scanner_scanpath
  67 ( char const* pathname )
  68 { DIR* dirp;
  69   errno = 0;
  70   if ((dirp = opendir(pathname)) == NULL || errno)
  71     { fprintf(stderr, "Path %s could not be accessed\n", pathname);
  72       return -1;
  73     }
  74   if (chdir(pathname))
  75     return -1;
  76   return scanner_scandir(dirp);
  77 }
  78
  79 /* Scan directory stream
  80    Recursively scans the provided directory, sending CLOPEN and CLCLOSE tokens
  81    to the parser when entering new directories (classes)
  82 */
  83 int scanner_scandir
  84 ( DIR* dirp )
  85 { DIR* cdirp;
  86   struct dirent* direntp;
  87  scan_next_dirent:
  88   errno = 0;
  89   direntp = readdir(dirp);
  90   if (errno)
  91     goto libfail;
  92   if (direntp != NULL)
  93     { if (*(direntp->d_name) == '.') //skip hidden or relative files
  94         goto scan_next_dirent;
  95       switch (direntp->d_type)
  96         { case DT_REG:
  97             printf("lexfile %s\n",direntp->d_name);
  98             //lexer_lexfile((uint8_t*)direntp->d_name);
  99             goto scan_next_dirent;
 100           case DT_DIR:
 101             //lexer_lexstring((uint8_t*)direntp->d_name);  //lex the dirname
 102             printf("lexdir %s\n",direntp->d_name);
 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             //PUSHTOK(CLOPEN, 0);           //push "Open Directory" token
 109             printf("Scanner entered [%s]\n",direntp->d_name);
 110             if(scanner_scandir(cdirp))    //scan the directory
 111               goto libfail;
 112             if (chdir(".."))              //return to the parent dir
 113               goto libfail;
 114             //PUSHTOK(CLCLOSE, 0);          //push "Close Directory" token
 115             printf("Scanner returned\n");
 116             goto scan_next_dirent;        //continue scan
 117           case DT_UNKNOWN:
 118             warnx("unknown file %s: ignoring", direntp->d_name);
 119           default:
 120             goto scan_next_dirent;
 121         }
 122     }
 123   return closedir(dirp);
 124  libfail:
 125   perror("scanner_scandir");
 126   return -1;
 127 }