/* Boolean operations on file sets based on MD5 digests.
 * Jerzy Tarasiuk			22-Jul-2000 */

#include <stdio.h>
#include <string.h>
#include <stdlib.h>

const char usage[]="Boolean operations on file sets "
		"based on MD5 digests. Jerzy Tarasiuk\n\n"
		"Usage: %s [-p] {-a|-b}[number] "
		"additional_input_files < input > output\n\n"
		"-p\tprints path only (w/o MD5 digest),\n"
		"-a/-b\trequire at least or below number "
		"of additional_input_files\n"
		"\tto contain same MD5 digests, "
		"for the line to be passed.\n\n"
		"All input files (STDIN and additional) "
		"must be sorted by MD5 digest value.\n";

/* Boolean operations on file sets based on MD5 digests.
 * - works on files sorted by MD5 digest values in ascending
 *   order, containing in every line: md5_digest and file_path,
 *   separted by two blanks (such a files can be produced by
 *   'md5sum' on Linux or by MD5TREE (mine) on DOS
 * - works as filter with additional input files, passing from
 *   STDIN to STDOUT lines which match criteria defined by
 *   filtering options and additional input files
 * - names of additional input files cannot start from '-'
 * - option: -a[number] or -b[number] (default number is 1)
 *   defines filtering rule: -a means at least number, -b less
 *   than number of additional input files contains MD5 digest
 *   same as in input line for the line to pass through filter
 * - option: -p means passing file_path only to STDOUT */

#define OPT_P 1
#define OPT_A 2
#define OPT_B 4

typedef char aibf_t[300]; /* seems may need 32+2+256+1 */
int optflg=0,count=1;
FILE **aifd; aibf_t buf,*aibf;

int compare_md5(int fi) {
	int c; char *fbf=aibf[fi];
	do {	if((c=strncmp(buf,fbf,33))<=0) return c==0;
	} while(fgets(fbf,sizeof(*aibf),aifd[fi])!=NULL);
	if(ferror(aifd[fi])) { sprintf(fbf,"input %u",fi+1); perror(fbf); }
	for(fbf[c=sizeof(*aibf)-1]=0; c-->0; fbf[c]=0x7f); return 0;
}

main(int ac, char **av) {
	int ai,nf,v,c; char *cp;
	for(nf=ai=0; ++ai<ac; )
		if(*(cp=av[ai])!='-') nf++;
		else switch(c=*++cp) {
			case'p':if(*++cp) goto badopt;
				optflg^=OPT_P; break;
			case'a':
			case'b':if((optflg&~OPT_P)==0 && (*++cp==0
				 || (sscanf(cp,"%u%n",&count,&v)==1
				  && cp[v]==0))) {
				/*	fprintf(stderr,"opt=%s\n",cp);	*/
					optflg|=c=='a'?OPT_A:OPT_B; break;
				}
			default: goto badopt;
		}
	if((optflg&~OPT_P)==0) goto badopt;
	fprintf(stderr,"filter=%s %u occurrences\n",
		(optflg & OPT_A) ? "at least" : "below", count);
	if(count>nf) { cp="%s: too less input files"; goto badarg; }
	if((aifd=malloc(nf*sizeof(*aifd)))==NULL
	|| (aibf=malloc(nf*sizeof(*aibf)))==NULL)
		{ perror("malloc"); return -1; }
	for(nf=ai=0; ++ai<ac; ) if(*(cp=av[ai])!='-')
		if((aifd[nf]=fopen(cp,"rt"))==NULL)
			{ perror(cp); return -1; }
		else aibf[nf++][0]=0;
	while(fgets(buf,sizeof(buf),stdin)!=NULL) {
		for(c=v=0; v<nf; v++) c+=compare_md5(v);
		if(optflg & OPT_A) {	if(c<count) continue; }
		else			if(c>=count) continue;
		cp=buf; if((optflg & OPT_P)
			&& (cp=strchr(buf,' '))!=NULL
			&& *cp==' ' && *++cp==' ') cp++;
		if(cp!=NULL) fputs(cp,stdout);
	}
	if(ferror(stdin)) perror("STDIN");
	if(fclose(stdout)!=0) perror("STDOUT");
	return 0;
badopt:	cp=usage;
badarg:	{	char pn[10],*ap;
		for(ap=*av,v=0; (c=*ap++)!=0; ) switch(c) {
			case':':case'/':case'\\': v=0; break;
			case'.':pn[v]=0; v=9; break;
			default:if(v<9) pn[v++]=c;
		}
		pn[v]=0; fprintf(stderr,cp,pn); return -1;
}	}