findline.c

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

#include <unistd.h>
#include <fcntl.h>

#define BUFSZ 8192
static const char *const less = "/usr/bin/less";

static int unicodedatafd = -1;
static long datasize, dataseek, imaginaryseek;

static struct {
	size_t s;
	unsigned char v[BUFSZ];
	long dataoffset;
} buf;

static int ensureready(char *fn){
	unicodedatafd = open(fn, O_RDONLY);
	if(unicodedatafd < 0)
		exit(1);
	return (dataseek = datasize = lseek(unicodedatafd, 0, SEEK_END)) > 0;
}

static int fillbuf(void){
	int rs;

	if(dataseek != imaginaryseek)
		if((dataseek = lseek(unicodedatafd, imaginaryseek, SEEK_SET)) < 0)
			return 0;
	buf.s = 0;
	rs = read(unicodedatafd, buf.v, BUFSZ);
	if(rs < 1)
		return 0;
	buf.s = rs;
	buf.dataoffset = dataseek;
	return 1;
}

static int getcurchar(void){
	if(!buf.s || imaginaryseek < buf.dataoffset || imaginaryseek >= buf.dataoffset + buf.s)
		if(!fillbuf())
			return -1;
	return buf.v[imaginaryseek - buf.dataoffset];
}

static int readchar(void){
	int c = getcurchar();
	if(c < 0)
		return -1;
	imaginaryseek++;
	return c;
}

static int skiptonextline(void){
	int c;

	if(imaginaryseek == -1){
		imaginaryseek = 0;
		return 1;
	}
	while((c = readchar()) > 0)
		if(c == '\n')
			return 1;
	return 0;
}

int compare(char *key){
	for(;;){
		if(!*key)
			return 0;
		int c = readchar();
		if(c < 0)
			return -1;
		if(c != *(unsigned char *)key)
			return *(unsigned char *)key - c;
		else
			key++;
	}
}

int lookup(char *key, long *out){
	long min = 0, max;

	max = datasize;
	int ret = 0;
	while(max >= min){
		long mid = (max + min)/2;
		imaginaryseek = mid - 1;
		skiptonextline();
		long linestart = imaginaryseek;
		int cmp = compare(key);
		//fprintf(stderr, "%li > %li; mid = %li; cmp = %i\n", max, min, mid, cmp);
		ret = cmp == 0;
		if(cmp > 0)
			min = mid + 1;
		else if(cmp < 0)
			max = mid - 1;
		else if(max != mid)
			max = mid;
		else{
			min = linestart;
			break;
		}
	}
	*out = min;
	return ret;
}

int main(int argc, char **argv){
	if(argc < 1)
		return -1;
	int exact = 0;
	int prefix = 0;
	int printnotfound = 0;
	int exit = 0;
	int readstdin = 0;
	int plainargs = 0;
	int execless = 0;

	int first = 1;

	char *fn = 0;

	for(int x = 1; x < argc + readstdin; x++){
		if(plainargs || x >= argc);
		else if(!strcmp(argv[x], "-e")){
			exact = !exact;
			continue;
		}else if(!strcmp(argv[x], "-w")){
			printnotfound = !printnotfound;
			continue;
		}else if(!strcmp(argv[x], "-p")){
			prefix = !prefix;
			continue;
		}else if(!strcmp(argv[x], "-i")){
			readstdin = 1;
			continue;
		}else if(!strcmp(argv[x], "-l")){
			execless = 1;
			continue;
		}else if(!strcmp(argv[x], "--")){
			plainargs = 1;
			continue;
		}

		if(first){
			if(x > argc){
				fprintf(stderr, "too few arguments\n");
				return -1;
			}
			if(!ensureready(fn = argv[x]))
				return -1;
			first = 0;
			continue;
		}

		char *needle;
		char buf[1024];

		if(x < argc)
			needle = argv[x];
		else{
			if(!fgets(buf, sizeof buf, stdin))
				continue;
			size_t len = strlen(buf);
			if(buf[len - 1] != '\n'){
				fprintf(stderr, "line too long\n");
				int c;
				do
					c = getchar();
				while(c != EOF && c != '\n');
				if(c == '\n')
					x--;
				continue;
			}
			x--;
			buf[len - 1] = '\0';
			needle = buf;
		}

		long pos;
		int found = lookup(needle, &pos);
		exit = exit || !found;
		if(prefix){
			imaginaryseek = pos;
			for(;;){
				int shouldbreak = 0;
				long line = imaginaryseek;
				int cmp = compare(needle);
				if(cmp != 0)
					break;
				imaginaryseek = line;
				for(;;){
					int c = readchar();
					shouldbreak = c <= 0;
					if(shouldbreak || c == '\n')
						break;
					putchar(c);
				}
				if(shouldbreak)
					break;
				putchar('\n');
			}
		}else if(found || !exact){
			if(execless){
				close(unicodedatafd);
				char lesscmd[1024];
				snprintf(lesscmd, sizeof lesscmd, "++%liP", pos);
				lesscmd[sizeof lesscmd - 1] = 0;
				execv(less, (char *[]){ "less", "-n", lesscmd, fn, 0 });
				perror("execv");
				return 1;
			}
			imaginaryseek = pos;
			for(;;){
				int c = readchar();
				if(c <= 0 || c == '\n')
					break;
				putchar(c);
			}
			putchar('\n');
		}
		if(printnotfound && !found)
			fprintf(stderr, "not found: %s\n", needle);
	}

	close(unicodedatafd);

	return exit;
}