/*!@file
\brief lexical analyzer implementation for APC
- \details this lexer scans a root directory given from the command line
- for subdirectories and files structured for the APC grammar.
+ \details The lexer manages two FIFO stacks. One for maintaining tokens, the
+ other for maintaining a list of files to be scanned. During
+ execution, the lexer will return a token from its token queue if any
+ are present. If not, the lexer will will pop an element from its
+ file queue to 'scanner' to be tokenized. If the file queue is empty,
+ the lexer will instead call 'parsedir' to traverse the directory tree
+ and tokenize the results. If 'parsedir' does not generate any new
+ tokens, we are done.
\author Jordan Lavatai
\date Aug 2016
----------------------------------------------------------------------------*/
/* Standard */
#include <stdio.h>
#include <string.h>
+#include <stdint.h>
#include <errno.h>
/* Posix */
#include <unistd.h>
+#include <unitypes.h>
+#include <unistr.h>
+#include <uniconv.h>
+#include <uniname.h>
+#include <unistdio.h>
#include <stdlib.h>
+#include <limits.h> //realpath, NAME_MAX, PATH_MAX
#include <dirent.h>
+/* Redefinitions of NAME_MAX and PATH_MAX */
+//#define NAME_MAX NAME_MAX/4
+//#define PATH_MAX PATH_MAX/4
+
/* Local */
#include "parser.tab.h"
-#define TOKEN_BUF_SIZE 1024
-#define ENTITY_BUF_SIZE 512
-#define SCANDIR_ERROR -1
-
-int lexer_init(void);
-int lexer(void);
-
-extern //main.c
-const char* cargs['Z'];
-
-static inline
-int scan(void);
-static inline
-int qcomp(const void*, const void*);
-
-/* Directory Listing Stack
- FILO Stack for keeping an open DIR* at each directory depth for treewalk.
- This stack is depth-safe, checking its depth during push operations, but not
- during pop operations, to ensure the thread doesn't open too many files at
- once (512 in c runtime), or traverse too far through symbolic links.
- A directory listing includes a DIR* and all DIR-typed entity in the directory
- as recognized by dirent, populated externally (and optionally).
- This stack behaves abnormally by incrementing its PUSH operation prior to
- evaluation, and the POP operations after evaluation. This behavior allows
- the 'DL_CURDEPTH' operation to map to the current element in the 'dl_stack'
- array, and it is always treated as the "current depth". This also allows us
- to init the root directory to 'directory_list_stack'[0] and pop it in a safe
- and explicit manner.
-*/
-#define S(S)#S
-#ifndef DL_DEPTHMAX
-#define DL_DEPTHMAX 64
+#ifndef DE_STACKSIZE
+#define DE_STACKSIZE 1024
#endif
-#ifndef DL_CHILDMAX
-#define DL_CHILDMAX DL_DEPTHMAX //square tree
+#ifndef TK_STACKSIZE
+#define TK_STACKSIZE 1024
+#endif
+#ifndef MAX_SETNAME_LEN //max setname length
+#define MAX_SETNAME_LEN 32
#endif
-#define DL_STACK (directory_list_stack)
-#define DL_STACKP (dls)
-#define DL_STACKSIZE (DL_DEPTHMAX + 1) //+1 because push increments early
-#define DL_LEN (DL_STACKP - DL_STACK)
-#define DL_CURDEPTH (DL_LEN)
-#define DL_CD_STACK ((*DL_STACKP).child_entity_stack)
-#define DL_CD_STACKP ((*DL_STACKP).cds)
-#define DL_CD_STACKSIZE (DL_DEPTHMAX) //square tree
-#define DL_CD_LEN (DL_CD_STACKP - DL_CD_STACK)
-#define ERR_CHILD "Fatal: Maximum of " S(MAX_CHILD) \
- " child directories exceeded for directory %s at depth %i\n" \
- ,DL_STACK[DL_DEPTH()].label,DL_DEPTH()
-#define ERR_DEPTH "Fatal: Maximum directory depth of " S(DL_DEPTHMAX) \
- " exceeded during directory scan\n"
-#define DL_INIT() (DL_STACKP = DL_STACK)
-#define DL_CD_INIT() (DL_CD_STACKP = DL_CD_STACK)
-#define DL_POP() ((DIR*)(*DL_STACKP--))
-#define DL_CD_POP() (*--DL_CD_STACKP)
-#define DL_PUSH(D) (*(DIR**)(++DL_STACKP) = DIRP)
-#define DL_CD_PUSH(E) (*DL_CD_STACKP++ = E)
-#define DL_SAFEPUSH(D) \
- do { \
- if (DL_DEPTH() >= DL_DEPTHMAX) \
- { fprintf(stderr, ERR_DEPTH); \
- exit(EXIT_FAILURE); \
- } \
- if ((*(DIR**)(++DL_STACKP) = D) == NULL) \
- { perror("DL_SAFEPUSH"); \
- exit(EXIT_FAILURE); \
- } \
- } while(0)
-#define DL_CD_POPULATE() \
- do { \
- DL_CD_INIT(); \
- while ((*DL_CD_STACKP = readdir(*DL_STACKP)) != NULL) \
- { switch((*DL_CD_STACKP)->d_type) \
- { case DT_REG: \
- case DT_DIR: \
- DL_CD_STACKP++; \
- break; \
- default: \
- } \
- } \
- qsort(DL_CD_STACKP, DL_CD_LEN, sizeof *DL_CD_STACKP, qcomp); \
- } while (0)
+/* Public */
+int lexer_init(void);
+int lexer(void);
+int lexer_lexfile(const uint8_t*);
+void lexer_pushtok(int, YYSTYPE);
+uint8_t const* lexer_get_current_filepath(void);
+int lexer_lexfilename(uint8_t*);
+struct dirent* lexer_direntpa[DE_STACKSIZE],** lexer_direntpp,** lexer_direntpb;
+/* Private */
+extern //lexer_fsm.rl
+int lexer_lexstring(uint8_t*, int);
+extern //scanner.c
+int scanner_init(void);
+extern //scanner.c
+int scanner(void);
+static inline
+int dredge_current_depth(void);
+extern //bison
+YYSTYPE yylval;
static
-struct dirlist
-{ DIR* dirp;
- struct dirent** child_directory_stack[DL_CD_STACKSIZE],*** cds;
- const char* label;
-} directory_list_stack[DL_STACKSIZE],* dls;
+uint8_t const* current_filename;
static
+struct tok
+{ YYSTYPE lval; //token val
+ int tok_t; //token type
+} token_stack[TK_STACKSIZE], *tsp, *tsx;
-/* Directory Entity Stack
- Simple stack for keeping track of dirents still being processed at the
- current depth.
+/* Directory Entity Array/Stack
+ Simple array for keeping track of dirents yet to be processed by the scanner.
+ If this list is empty and there are no tokens, the lexer is done.
+ This array is populated by the scanner as an array, and popped locally by the
+ lexer as a stack, and is popped as a FIFO stack.
*/
-#define CE_STACK (child_entity_stack)
-#define CE_STACKP (ces)
-#define CE_STACKSIZE 1024
-static
-struct dirent** child_entity_stack[CE_STACKSIZE],*** ces;
-#define CE_INIT() (CE_STACKP = CE_STACK)
-#define CE_POP() (*--CE_STACK)
-#define CE_PUSH(E) (*CE_STACK++ = E)
+#define DE_STACK (lexer_direntpa)
+#define DE_STACKP (lexer_direntpp)
+#define DE_STACKB (lexer_direntpb)
+#define DE_LEN() (DE_STACKP - DE_STACKB)
+#define DE_INIT() (DE_STACKP = DE_STACKB = DE_STACK)
+#define DE_POP() (*DE_STACKB++)
/* Token Stack
This is a FIFO stack whose pointers are a union of either a pointer to an
An alignment error will occur if IPOP or IPUSH are used a non-even number of
times in a sequence!
*/
-#define TK_STACK (token_stack)
-#define TK_STACKP (tks.t)
-#define TK_STACKPI (tks.i)
-#define TK_STACKX (tkx.t)
-#define TK_STACKXI (tkx.i)
-#ifndef TK_STACKSIZE
-#define TK_STACKSIZE 1024
-#endif
-#define TK_EMPTY (TK_STACKP == TK_STACKX)
-#define TK_INIT() (TK_STACKP = TK_STACKX = TK_STACK)
-#define TK_POP() (*TK_STACKP++)
-#define TK_IPOP() (*TK_STACKPI++);
-#define TK_PUSH(TOK,LVAL) (*TKSTACKX++ = (struct tok){(LVAL),(TOK)})
-#define TK_IPUSH(I) (*TKSTACKXI++ = (I))
-static
-struct tok
-{ int lval;
- int tok;
-} token_stack[TK_STACKSIZE];
-static
-union tokp
-{ int* i;
- struct tok* t;
-} tks, tkx;
-
+#define TK_STACK (token_stack)
+#define TK_STACKP (tsp)
+#define TK_STACKX (tsx)
+#define TK_LEN() (TK_STACKX - TK_STACKP)
+#define TK_INIT() (TK_STACKP = TK_STACKX = TK_STACK)
+#define TK_POP() (*TK_STACKP++)
+#define TK_PUSH(T,L) (*TK_STACKX++ = (struct tok){L,T})
/* Initializer
- The initializer returns boolean true if opendir contains an error.
+ The initializer returns boolean true if an error occurs, which may be handled
+ with standard errno.
*/
-int lexer_init()
-#define ERR_UF "Fatal: Unknown file [%s] encountered in directory %s", \
- DL_CD_STACK[i]->d_name, DL_STACK[DL_DEPTH]
-{ int i = 0;
- TK_INIT();
- DL_INIT();
- DL_STACK[0].label = cargs['r'] ? cargs['r'] : "./";
- if ((DL_STACK[0].dirp = opendir(DL_STACK[0].label)) == NULL)
- return -1;
- errno = 0;
- while ((DL_CD_STACK[i] = readdir(DL_STACK[0].dirp)) != NULL)
- switch(DL_CD_STACK[i]->d_type)
- {
- default:
- fprintf(stderr, "Fatal: Unkown File %s\n", DL_CD_STACK[i]->d_name);
- }
- return (errno);
+int lexer_init
+()
+{ TK_INIT();
+ DE_INIT();
+ return scanner_init();
}
/* Lexer
If the token buffer is empty, 'lexer' will initialize the token buffer and
- call 'lexer_scandir'. If #SCANDIR_ERROR is returned, an error is printed
+ call 'lexer_scandir'. If SCAN_ERROR is returned, an error is printed
before sending a null return to bison. If 0 tokens are generated, the error
printing is skipped. In all other cases, 'yylval' is set, and the token's
integer representation is returned.
*/
-int lexer()
-{ if (TOK_BUF_EMPTY)
- { TOK_BUF_INIT();
- switch (lexer_scan)
- { case SCANDIR_ERROR:
- perror("lexer_scan");
- case 0:
- yylval = 0;
- return 0;
- default:
+int lexer
+#define $($)#$
+#define SCAN_ERROR -1
+#define TK_EMPTY (TK_STACKP == TK_STACKX)
+#define FAIL(...) \
+ do { \
+ fprintf(stderr,__VA_ARGS__); \
+ goto done; \
+ } while (0)
+()
+{ struct tok token;
+ start:
+ while (DE_LEN() > 0)//lex any directory entries in our stack
+ {
+ if (lexer_lexfile(DE_POP()->d_name) == 0)
+ FAIL("Lexer failed to tokenize [%s]\n",(*DE_STACKB)->d_name);
+ }
+ if (TK_EMPTY) //if there are no tokens,
+ { TK_INIT(); //initialize the token stack back to 0
+ switch (scanner())
+ { case SCAN_ERROR: //if an error occurred,
+ FAIL("Scanner error\n");
+ case 0: //if the the scanner finds no dirents,
+ goto done; //then we are done
+ default: //if we found some elements to scan,
+ goto start; //start over and lex them
}
}
- yylval = TOK_BUF_IPOP();
- return TOK_BUF_IPOP();
+ token = TK_POP();
+ yylval = token.lval;
+ return token.tok_t;
+ done:
+ yylval.val = 0;
+ return 0;
}
-static inline
-int lexer_scan()
-{ static DIR*
+/* Token Receiver
+ This receiver takes a struct tok and pushes it to the FIFO stack.
+*/
+void lexer_pushtok
+#define $($)#$ //stringifier
+#define ERR_TK "Fatal: Generated over " $(TK_STACKSIZE) " tokens in one pass."
+( int tok, YYSTYPE lval )
+{ if (TK_LEN() >= TK_STACKSIZE)
+ { fprintf(stderr, ERR_TK);
+ exit(EXIT_FAILURE);
+ }
+ TK_PUSH(tok, lval);
}
-/* Scanner
- Scans a filename from its alphabetically ordered list of file elements
- and tokenizes the result. If the file list is empty, then the stack of
- directory elements will be popped and processed as they are encountered.
-
- Returns the number of tokens generated.
+/* Lexical analysis of a file
+ Strips a filename to its base name, then sends it to lexer_lex
*/
-#define MAX_ENTITIES 256
-static
-int lexer_scan()
-{ static struct dirent* entity;
- static struct dirent* files[MAX_ENTITIES] = {0};
- static int num_files = 0;
-#define NO_FILES (num_files == 0)
+int lexer_lexfile
+#define HIDDEN_WARNING "%s is hidden and will not be parsed!\n", filename
+( const uint8_t *filename
+)
+{ static uint8_t fname[NAME_MAX];
+ uint8_t *last_period = NULL, *iter;
- if (NO_FILES)
+ if (*filename == '.')
+ { fprintf (stderr, HIDDEN_WARNING);
+ return 0;
+ }
+ /* Copy the filename and remove its suffix */
+ u8_strncpy(fname,filename,NAME_MAX);
+ last_period = NULL;
+ for (iter = fname; *iter; iter++) //find the last '.' char
+ if (*iter == '.')
+ last_period = iter;
+ if (last_period) //if we found one,
+ *last_period = 0; //truncate the string there
+ /* Register the current_filename */
+ current_filename = filename;
+
+ return lexer_lexfilename(fname);
+}
+uint8_t const* lexer_get_current_filepath
+()
+{ static uint8_t current_path[PATH_MAX];
+ static uint8_t const* last_filename;
+ if ((!last_filename || last_filename != current_filename) &&
+ (realpath(current_filename, current_path) != (char*) current_path))
+ { perror("realpath: ");
+ return NULL;
+ }
+ return (const char*)current_path;
+}
- //sort out files and directories, grow directories from bottom up
- while ((entity = readdir(dirp)) != NULL)
- { switch (entity->d_type)
- { case DT_LNK:
- case DT_REG:
- files[num_files++] = entity;
- break;
- case DT_DIR:
- *(dirs - num_dirs++) = entity;
- break;
- case DT_UNKNOWN:
- default:
- printf("Ignoring unknown file: %s\n", entity->d_name);
- break;
- }
+/* Scan filename and push the its tokens
+ onto the stack */
+int lexer_lexfilename
+(uint8_t* str)
+{
+ int ntok, i, cmp, len, set_len, height, width;
+ char map_key[] = "_m_";
+ static uint8_t set_name[MAX_SETNAME_LEN] = {0};
+ uint8_t *first, *map_begin;
+
+ printf("Starting lexerfilename on %s\n", str);
+
+
+ if(*str == 0)
+ printf("Lexfilename:: str is NULL so fail\n");
+ printf("setname is %s\n", set_name);
+
+ /* If last file was a mapfile, then its 5th to last token should
+ be a MOPEN. If this is the case, then we only pass MOPEN, height,
+ weight and name of the current file. */
+ if( (TK_STACKX - 5)->tok_t == MOPEN )
+ { printf("The last file was a mapfile\n");
+ if( (map_begin = strstr(map_key, str)) ) //if the current file is a mapfile
+ { printf("The current file is a variant of the last mapfile\n");
+ printf("Start lexing mapfile %s\n", str);
+ ntok += lexer_lexstring(map_begin, strlen(map_begin));
+ }
+ printf("Current file is not a variant of the last mapfile\n");
}
- if (errno)
- perror("readdir");
- qsort(&files[0], num_files, sizeof struct dirent*, qalpha);
- num_ents = scandirat(dirfd, ".", &namelist, scanfilter, scancompar);
- if (num_ents < 0)
- { perror("scandirat");
- return -1;
+ else //last file was not a mapfile
+ { printf("Last file was not a mapfile\n");
+
+ first = (uint8_t*) u8_strchr(str, '_'); //find the first '_' to find where str set_name ends
+
+ if(set_name[0] != 0) //if there is a set_name from last str
+ { printf("There is a set_name (%s) present\n", set_name);
+ set_len = first - str;
+
+ if(u8_strncmp(str, set_name, set_len) == 0) //check if it matches the current set_name
+ { str = str + set_len + 1; //if so, remove it from str
+ printf("str set_name matched last set_name, set str to %s\n", str);
+ }
+ else //update set_name to be str set_name
+ { u8_cpy(set_name, str, set_len);
+ set_name[set_len] = 0;
+
+ }
+ }
+ else //set set_name
+ { u8_cpy(set_name, str, first-str);
+ }
+ /* Call lexer_lexstring to tokenize the string */
+ printf("calling lexstring to tokenize str (%s) of len %d\n", str, u8_strlen(str));
+ ntok += lexer_lexstring(str, u8_strlen(str));
}
- //process files
+
+ /*TODO: if regfile, store full path for later */
- //recurse into directories
- return tbx - tbp;
+ printf("Ending lexer_lex on %s, %d tokens were lexed\n", str, ntok);
+ return ntok;
}
-/* Quicksort comparison function
- sort each dirent encountered first by its file type (regular files first)
- and then by their alphabetical ordering.
-*/
-static
-int qcompar
-( const void* a,
- const void* b
- )
-{
-}
+/* int lexer_lexmapfile */
+/* #define INC_X() */
+/* (int height, int width) */
+/* { */
+/* int x, y; */
+
+/* /\* Give scanner_scanpixels a buffer and a len. Iterate through */
+/* buf with buf[n]. If n == 0, do nothing. if n has a value, push x, */
+/* push y, push (z = n << 24), push (ref_id = n >> 8) *\/ */
+/* //scanner_scanpixels() */
+
+/* for(i = 0; i < len; i++) */
+/* if(buf[i] == 0) */
+/* if(x == width) */
+/* x = 0; */
+/* else */
+
+
+
+/* } */
+/* fname_bytes = (uint8_t*)(DE_POP()->d_name); */
+ /* printf("d_name is %s\n", fname_bytes); */
+ /* for (fnp = filename, i = 0; i < NAME_MAX; i += unit_size, fnp++) */
+ /* { unit_size = u8_mblen(fname_bytes + i, min(4, NAME_MAX - i)); */
+ /* if (u8_mbtouc(fnp, fname_bytes + i, unit_size) == -1) //add ucs4 char to the filename */
+ /* FAIL("Lexer failed to convert ^%s to unicode\n", (fname_bytes + i)); */
+ /* if (*fnp == 0) //added a terminating char */
+ /* break; */
+ /* } */
+ /* if(u8_mbtouc(filename, DE_POP()->d_name, NAME_MAX) == -1) */
+ /* FAIL("Lexer failed to convert d_name into uint8_t\n"); */
+ /* ulc_fprintf(stdout, "filename is %11U\n c", filename); */