Category : C Source Code
Archive   : LSAM23.ZIP
Filename : DLFORMAT.C

/*
dlformat.c Filter pgm: Reads captured ASCII text from CIS
Data Libraries, converting input to
fixed-length record format file suit-
able for indexing via LSAM.
*/
#include "stdio.h"
#include "string.h"
#include "dl_clf.h" /* formatted DL record structure */
/* symbolics for readability */
#define NO 0 /* boolean switch constants */
#define YES 1
#define EOS '\0' /* end of string symbolic constant */
#define TAB '\t' /* tab character symbolic constant */
#define CR 0x0d /* carriage return character symbolic constant */
#define LF 0x0a /* linefeed character symbolic constant */
#define DOS1_PRT 9 /* DOS 1.1 system print function req. # */
#define S_INIT 0 /* proc_state discrete values for input processing */
#define S_USERID 1
#define S_NAME 2
#define S_KEYWDS 3
#define S_DESCR 4

char *errmsg[] = {
"Usage requires input filename$",
"Input file not found$",
"Output file open error$",
"Output file write error$"
};
unsigned char space = 0x20;
unsigned int line_no = 0;
unsigned char inddname[80],
outddname[80],
inbuffer[256];
#define BUFFLEN sizeof(inbuffer)

CATREC catrec, /* declare record area */
catrecinit; /* and initialized area for re-init */

extern FILE *fopen();
extern char *fgets();

main(argc,argv)
int argc;
char **argv;
{
FILE *inp, *outp;
unsigned char *sub0, /* substring pointers */
*sub1,
*ret, /* fgets() return ptr */
eofinp = NO, /* input EOF switch */
got_keys = NO, /* keyword, desrciption state switches */
got_desc = NO;
unsigned bflen = BUFFLEN;
unsigned lines_in = 0;
unsigned proc_state = S_INIT;

banner();
if(argc < 2) {
bdos(DOS1_PRT,errmsg[0]);
_exit(0x0001);
}
strcpy(inddname,*++argv);
if(!(inp = fopen(inddname,"rb"))) {
bdos(DOS1_PRT,errmsg[1]);
_exit(0x0002);
}
if(sub0 = strrchr(inddname,'.')) { /* do we have a 'dot' */
*sub0 = EOS; /* terminate string at 'dot' */
}
strcpy(outddname,inddname); /* copy in to out filename */
strcat(outddname,".clf"); /* append .clf extension */
if(!(outp = fopen(outddname,"wb"))) {
bdos(DOS1_PRT,errmsg[2]);
exit(0x0003);
}
/* set up a pre-initialized record area for simple re-initialization */
memset(&catrecinit,EOS,sizeof(CATREC)); /* zero it out first */
strcpy(catrecinit.dl_no,"DL??"); /* DL is unknown so far */
catrecinit.dlcrlf[0] = catrecinit.u_crlf[0] = catrecinit.f_crlf[0] =
catrecinit.k_crlf[0] = catrecinit.d_crlf[0] = CR;
catrecinit.dlcrlf[1] = catrecinit.u_crlf[1] = catrecinit.f_crlf[1] =
catrecinit.k_crlf[1] = catrecinit.d_crlf[1] = LF;
memcpy(&catrec,&catrecinit,sizeof(CATREC)); /* now initialize work recd*/
/* now loop through the text, writing formatted records */
while(eofinp != YES) {
ret = fgets(inbuffer,BUFFLEN,inp);
if(ret == NULL) {
if(lines_in == 0)
printf("DLFORMAT: EOF on first read of %s\n",inddname);
else
printf("DLFORMAT: EOF after line %d of %s\n",lines_in,inddname);
eofinp = YES;
} else {
lines_in++;
switch(proc_state) {
case S_INIT:
s_init: if(*inbuffer == 'D') { /* look for 'DL??' */
strcpy(catrec.dl_no,inbuffer);
strcpy(catrecinit.dl_no,inbuffer);
continue;
} else if(*inbuffer != '[') /*look for userid begin bracket*/
continue;
else /* when found */
proc_state = S_USERID; /* then fall through */
case S_USERID:
if(*inbuffer == CR || *inbuffer == LF)
continue;
sub0 = inbuffer;
sub1 = catrec.userid;
while(*sub0 != CR) /* copy up to CR */
*sub1++ = *sub0++;
*sub1 = EOS; /* terminate string */
proc_state = S_NAME;
break;
case S_NAME:
if(*inbuffer == CR || *inbuffer == LF)
continue;
sub0 = strstr(inbuffer,"Accesses: ");
if(sub0 != NULL) /* captured ? */
sscanf(inbuffer,"%20s %9s %7s Accesses: %5s %9s",
catrec.filename,
catrec.upl_date,
catrec.filesize,
catrec.acc_count,
catrec.mod_date);
else /* no, assume TAPCIS */
sscanf(inbuffer,"%20s %9s %7s %5s %9s",
catrec.filename,
catrec.upl_date,
catrec.filesize,
catrec.acc_count,
catrec.mod_date);
proc_state = S_KEYWDS;
break;
case S_KEYWDS:
if((*inbuffer == CR || *inbuffer == LF) && got_keys == NO)
continue;
else if(*inbuffer == space) {
int sl, i = 0;
sub0 = (got_keys == NO) ? inbuffer + 14 : inbuffer + 4;
strcat(catrec.keywords,sub0);
sub1 = catrec.keywords; /* point to keywords */
sl = strlen(catrec.keywords); /* get length of string */
sl--; /* back off as max offset */
while(i <= sl) { /* search for CR and */
if(*(sub1 + i) == CR) { /* if CR is found */
*(sub1 + i) = space; /* substitute a space */
*(sub1 + i + 1) = EOS; /* and follow w/EOS */
break;
}
i++;
}
got_keys = YES;
} else if(*inbuffer == CR || *inbuffer == LF) /* newlines */
proc_state = S_DESCR;
break;
case S_DESCR:
if((*inbuffer == CR || *inbuffer == LF) && got_desc == NO)
continue;
else if(*inbuffer == space) {
strcat(catrec.descript,inbuffer + 4);
got_desc = YES;
} else if(*inbuffer == 'P' /* 'Press for more !' */
|| *inbuffer == CR || *inbuffer == LF) /* newlines */
;
else {
proc_state = S_INIT; /* return to initial state */
got_desc = got_keys = NO;
fwrite(&catrec,sizeof(CATREC),1,outp);
/* initialize record area from pre-initialized record area */
memcpy(&catrec,&catrecinit,sizeof(CATREC));
goto s_init;
}
break;
}
}
}
fclose(outp);
fprintf(stderr,"DLFORMAT: DLSCAN-compatible file [%s] generated\n",
outddname);
exit(0);
}
banner()
{
fprintf(stderr,"%s%s","DLFORMAT (C) Copyright 1988, Lydian Software\n",
"CIS '.CAT' file generator for LSAM\n");
}