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 void  scanner_quit(void);
  28 int   scanner_scanpath(char const*);
  29 int   scanner_scandir(DIR*);
  30 /* Private */
  31 static
  32 int   scanner_scandir_r(DIR*);
  33 extern //lexer.rl
  34 int   lexer_init(void);
  35 extern //lexer.rl
  36 void  lexer_quit(void);
  37 extern //lexer.rl
  38 int   lexer_lexfile(uint8_t*);
  39 extern //lexer.rl
  40 int   lexer_lexdir(uint8_t*);
  41 extern //lexer.rl
  42 void  lexer_closedir(void);
  43 /* Scan the provided path
  44    Changes working directory to the provided pathname and, if successful, sends
  45    a directory stream of the provided path to scanner_scandir
  46 */
  47 int scanner_scanpath
  48 ( char const* pathname )
  49 { DIR* dirp;
  50   errno = 0;
  51   if ((dirp = opendir(pathname)) == NULL || errno)
  52     { fprintf(stderr, "Path %s could not be accessed\n", pathname);
  53       return -1;
  54     }
  55   if (chdir(pathname))
  56     return -1;
  57   return scanner_scandir(dirp);
  58 }
  59
  60 /* Scan directory stream
  61    Recursively scans the provided directory, sending CLOPEN and CLCLOSE tokens
  62    to the parser when entering new directories (classes)
  63 */
  64 int scanner_scandir
  65 ( DIR* dirp )
  66 { int scandir_status;
  67   if (lexer_init())
  68     return -1;
  69   scandir_status = scanner_scandir_r(dirp);
  70   lexer_quit();
  71   return scandir_status;
  72 }
  73
  74 static
  75 int scanner_scandir_r
  76 ( DIR* dirp )
  77 { DIR* cdirp;
  78   struct dirent* direntp;
  79  scan_next_dirent:
  80   errno = 0;
  81   direntp = readdir(dirp);
  82   if (errno)
  83     goto libfail;
  84   if (direntp != NULL)
  85     { if (*(direntp->d_name) == '.') //skip hidden or relative files
  86         goto scan_next_dirent;
  87       switch (direntp->d_type)
  88         { case DT_REG:
  89             lexer_lexfile((uint8_t*)direntp->d_name);
  90             goto scan_next_dirent;
  91           case DT_DIR:
  92             lexer_lexdir((uint8_t*)direntp->d_name);  //lex the dirname
  93             if (chdir(direntp->d_name))    //change to the specified dir
  94               goto libfail;
  95             errno = 0;
  96             if ((cdirp = opendir(".")) == NULL || errno) //open it
  97               goto libfail;
  98             if(scanner_scandir_r(cdirp))  //scan the directory
  99               goto libfail;
 100             if (chdir(".."))              //return to the parent dir
 101               goto libfail;
 102             lexer_closedir();            //push "Close Directory" token
 103             goto scan_next_dirent;        //continue scan
 104           case DT_UNKNOWN:
 105             warnx("unknown file %s: ignoring", direntp->d_name);
 106           default:
 107             goto scan_next_dirent;
 108         }
 109     }
 110   return closedir(dirp);
 111  libfail:
 112   perror("scanner_scandir");
 113   return -1;
 114 }