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