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
13 ----------------------------------------------------------------------------*/
15 #include <stdio.h> //print
16 #include <errno.h> //errno
18 #include <err.h> //warnx
19 #include <stdlib.h> //exit
20 #include <unistd.h> //chdir
21 #include <dirent.h> //opendir
23 #include "parser.tab.h"
25 int scanner_init(void);
26 void scanner_quit(void);
30 #define DL_STACKSIZE 64
32 #ifndef DL_CD_STACKSIZE
33 #define DL_CD_STACKSIZE DL_STACKSIZE //square tree
36 int lexer_lex(const char*);
38 void lexer_pushtok(int, int);
40 int dredge_current_depth(void);
42 struct dirent
* lexer_direntpa
[], **lexer_direntpp
;
43 extern //SRC_DIR/bin/tools/apc.c
44 const char* cargs
['Z'];
48 struct dirent
* child_directory_stack
[DL_CD_STACKSIZE
],** cds
;
49 } directory_list_stack
[DL_STACKSIZE
+ 1],* dls
; //+1 for the root dir
51 /* Directory Listing Stack
52 FILO Stack for keeping an open DIR* at each directory depth for treewalk.
53 This stack is depth-safe, checking its depth during push operations, but not
54 during pop operations, to ensure the thread doesn't open too many files at
55 once (512 in c runtime), or traverse too far through symbolic links.
56 A directory listing includes a DIR* and all DIR-typed entity in the directory
57 as recognized by dirent, populated externally (and optionally).
58 This stack behaves abnormally by incrementing its PUSH operation prior to
59 evaluation, and the POP operations after evaluation. This behavior allows
60 the 'DL_CURDEPTH' operation to map to the current element in the 'dl_stack'
61 array, and it is always treated as the "current depth". This also allows us
62 to init the root directory to 'directory_list_stack'[0] and pop it in a safe
65 #define DL_STACK (directory_list_stack)
66 #define DL_STACKP (dls)
67 #define DL_CD_STACK ((*DL_STACKP).child_directory_stack)
68 #define DL_CD_STACKP ((*DL_STACKP).cds)
69 #define DL_CURDIR() ((*DL_STACKP).dirp)
70 #define DL_LEN() (DL_STACKP - DL_STACK)
71 #define DL_CD_LEN() (DL_CD_STACKP - DL_CD_STACK)
72 #define DL_INIT() (DL_STACKP = DL_STACK)
73 #define DL_CD_INIT() (DL_CD_STACKP = DL_CD_STACK)
74 #define DL_POP() ((*DL_STACKP--).dirp)
75 #define DL_CD_POP() (*--DL_CD_STACKP)
76 #define DL_PUSH(D) ((*++DL_STACKP).dirp = D)
77 #define DL_CD_PUSH(E) (*DL_CD_STACKP++ = E)
81 Initializer expects a function pointer to its lexical analysis function.
82 Sets up stack pointers and returns boolean true if 'opendir' encounters an
83 error, or if dredge_current_depth returns boolean true.
87 #define ROOTDIR (cargs['d'] ? cargs['d'] : CWDSTR)
90 DL_STACK
[0].dirp
= opendir(ROOTDIR
);
91 printf("Root dir %s\n",ROOTDIR
);
92 return !chdir(ROOTDIR
) && (DL_STACK
[0].dirp
== NULL
|| dredge_current_depth() == -1);
99 closedir(DL_CURDIR());
103 The main driver of the scanner will advance the current treewalk state and
104 tokenize tree-based push/pop operations. It will call 'lexer_lex' to
105 tokenize directory names prior to making a push operation. safe checking for
106 all returns from the filesystem handler will exit on serious system errors.
108 after pushing a new directory to the directory list, the scanner will dredge
109 the directory and alphabetically sort all file entries into the lexer's file
110 array, while placing all subdirectory entries in the current depth's child
111 directory stack to be scanned later.
113 Returns the number of elements added to the lexer's file array, or -1 on
117 #define $($)#$ //stringifier
118 #define ERR_CHILD "Fatal: Maximum of " $(DL_CD_STACKSIZE) \
119 " child directories exceeded for directory at depth %i\n" \
121 #define ERR_DEPTH "Fatal: Maximum directory depth of " $(DL_STACKSIZE) \
122 " exceeded during directory scan\n"
123 #define ERR_DL "Fatal: Directory List Stack Corruption %x\n", DL_LEN()
125 { struct dirent
* direntp
;
128 if (DL_CD_LEN() >= DL_CD_STACKSIZE
)//fail if maxchildren exceeded
129 { fprintf(stderr
, ERR_CHILD
);
132 if (DL_CD_LEN() > 0) //There are entities to process
133 { if ((direntp
= DL_CD_POP()) == NULL
)//If the dirent is null, the library
134 goto libfail
; //function in dirent has failed
135 lexer_lex(direntp
->d_name
); //lex the directory name
136 if (DL_LEN() >= DL_STACKSIZE
) //fail if maxdepth exceeded
137 { fprintf(stderr
, ERR_DEPTH
);
140 if (chdir(direntp
->d_name
)) //move into the new directory
142 DL_PUSH(opendir(CWDSTR
));
143 if (DL_CURDIR() == NULL
) //open the cwd
145 lexer_pushtok(CLOPEN
, 0); //Push "Open Directory" token
146 return dredge_current_depth(); //Filter and sort the current depth
148 else if (DL_LEN() >= 0) //Any dirs left? (Including root)
149 { if (closedir(DL_POP())) //close the directory we just left
151 if (DL_LEN() == -1) //If we just popped root,
152 return 0; //we're done
153 lexer_pushtok(CLCLOSE
, 0); //Else push "Close Directory" token,
154 if (!chdir("..")) //move up a directory and
155 goto parse
; //start over
157 fprintf(stderr
, ERR_DL
);
164 /* Directory Entity Sort and Filter (Dredge)
165 This filter removes all unhandled file types, and places any 'DT_DIR' type
166 files in the current Directory List's directory stack. Upon finishing,
167 the 'CE_STACK' is sorted alphabetically, and the current 'DL_CD_STACK' is
168 populated. Prints warnings for unhandled files.
170 Returns -1 if 'readdir' encounters an error, otherwise returns the number of
171 directory entries sent to the external 'lexer_direntpa' array.
174 int (*qcomp
)(const void*, const void*);
176 int dredge_current_depth
177 #define READDIR_ERROR (-1)
178 #define READDIR_DONE (0)
179 #define DPS_LEN() (lexer_direntpp - lexer_direntpa)
180 #define DPS_PUSH(E) (*lexer_direntpp++ = E)
182 { struct dirent
** direntpp
= lexer_direntpa
;
183 DIR* cwd
= DL_CURDIR();
184 struct dirent
* direntp
;
187 if ((direntp
= readdir(cwd
)) != NULL
)
188 { switch (direntp
->d_type
)
193 if (*(direntp
->d_name
) == '.') //skip hidden files and relative dirs
198 warnx("unknown file %s: ignoring", direntp
->d_name
);
205 qsort(lexer_direntpa
, DPS_LEN(), sizeof direntp
, (qcomp
)alphasort
);