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_scandir(DIR*);
  29 /* Private */
  30 static
  31 int   scanner_scandir_r(DIR*);
  32 extern //lexer.rl
  33 int   lexer_init(void);
  34 extern //lexer.rl
  35 void  lexer_quit(void);
  36 extern //lexer.rl
  37 int   lexer_lexfile(uint8_t*);
  38 extern //lexer.rl
  39 int   lexer_lexdir(uint8_t*);
  40 extern //lexer.rl
  41 void  lexer_closedir(void);
  42
  43 /* Scan directory stream
  44    Recursively scans the provided directory, sending CLOPEN and CLCLOSE tokens
  45    to the parser when entering new directories (classes)
  46 */
  47 int scanner_scandir
  48 ( DIR* dirp )
  49 { int scandir_status;
  50   if (lexer_init())
  51     return -1;
  52   scandir_status = scanner_scandir_r(dirp);
  53   lexer_quit();
  54   return scandir_status;
  55 }
  56
  57 static
  58 int scanner_scandir_r
  59 ( DIR* dirp )
  60 { DIR* cdirp;
  61   struct dirent* direntp;
  62  scan_next_dirent:
  63   errno = 0;
  64   direntp = readdir(dirp);
  65   if (errno)
  66     goto libfail;
  67   if (direntp != NULL)
  68     { if (*(direntp->d_name) == '.') //skip hidden or relative files
  69         goto scan_next_dirent;
  70       switch (direntp->d_type)
  71         { case DT_REG:
  72             lexer_lexfile((uint8_t*)direntp->d_name);
  73             goto scan_next_dirent;
  74           case DT_DIR:
  75             lexer_lexdir((uint8_t*)direntp->d_name);  //lex the dirname
  76             if (chdir(direntp->d_name))    //change to the specified dir
  77               goto libfail;
  78             errno = 0;
  79             if ((cdirp = opendir(".")) == NULL || errno) //open it
  80               goto libfail;
  81             if(scanner_scandir_r(cdirp))  //scan the directory
  82               goto libfail;
  83             if (chdir(".."))              //return to the parent dir
  84               goto libfail;
  85             lexer_closedir();            //push "Close Directory" token
  86             goto scan_next_dirent;        //continue scan
  87           case DT_UNKNOWN:
  88             warnx("unknown file %s: ignoring", direntp->d_name);
  89           default:
  90             goto scan_next_dirent;
  91         }
  92     }
  93   return closedir(dirp);
  94  libfail:
  95   perror("scanner_scandir");
  96   return -1;
  97 }