ttail/src/ttail_search_files.c

#include "ttail_search_files.h"

int _ttail_search_closest_files(ttail_t* t, const struct tm *tm)
{
	int ret;
	size_t i, prev;
	struct tm **ftm; /* t->logfile_sz len of struct tm[2] */
	ret = _ttail_search_closest_files_init(t);
	if(ret)
	{
		return ret;
	}
	/* Storing min & max of each files in ftm and checking that files are 
	 * sorted well */
	ftm = malloc(sizeof(struct tm*) * t->logfile_sz);
	if(!ftm)
	{
		perror("Unable to allocate memory");
		goto _ttail_search_closest_files_alloc_err;
	}
	for(i=0; i<t->logfile_sz; i++)
	{
		ftm[i] = malloc(sizeof(struct tm)*2);
		if(!ftm[i])
		{
			perror("Unable to allocate memory for time");
			goto _ttail_search_closest_files_alloc_loop_err;
		}
		if(!t->logfile[i])
		{
			continue;
		}
		ret = _ttail_file_minmax(t, 0, ftm[i]);
		if(ret < 0)
		{
			fprintf(stderr, "Minmax error\n");
			goto _ttail_search_closest_files_loop_err;
		}
		else if (ret == 1)
		{
			if(t->verbose)
			{
				fprintf(stderr, "Warning : unable to find \
a valid date in '%s'\n", t->logfile_name[i]);
			}
			free(ftm[i]);
			ftm[i] = NULL;
			fclose(t->logfile[i]);
			t->logfile[i] = NULL;
		}
		prev = i;
		if(i && prev &&
			ttail_tm_cmp(&(ftm[prev][1]), &(ftm[i][0])) > 0)
		{
			/* files are not sorted */
			fprintf(stderr, "Files do not seems to be sorted. \
File sorting not implemented yet\n");
			goto _ttail_search_closest_files_loop_err;
		}
	}
	/* TODO begining binary search of date_min */
	ret = _ttail_search_files_binary_search(t, tm, (const struct tm**)ftm);

	return 0;
	goto _ttail_search_closest_files_err;

	_ttail_search_closest_files_err:
	i = t->logfile_sz;
	do
	{
		_ttail_search_closest_files_alloc_loop_err:
		i--;
		_ttail_search_closest_files_loop_err:
		if(ftm[i])
		{
			free(ftm[i]);
		}
	}while(i);
	free(ftm);
	_ttail_search_closest_files_alloc_err:
	return -1;
}

int _ttail_search_files_binary_search(ttail_t* t, const struct tm* in,
	const struct tm** ftm)
{
	int cmin, cmax;
	size_t valid;
	off_t *off;
	size_t *id;
	off = &(t->session->file.off);
	id = &(t->session->file.id);
	*id = *off = 0;
	valid = 0;
	if(ttail_tm_cmp(&(ftm[0][0]), in) > 0)
	{
		return 1;
	}
	while(1)
	{
		if(!ftm[*id])
		{
			(*id)++;
			continue;
		}
		valid++;
		cmin = ttail_tm_cmp(&(ftm[*id][0]), in);
		cmax = ttail_tm_cmp(&(ftm[*id][1]), in);
		if(!cmin)
		{
			/* found at the begining of the file */
			return 0;
		}
		else if (!cmax)
		{
			/* found at EOF */
			*off = _ttail_from_search_from_end(t, *id, in);
			if(*off < 0)
			{
				*off = 0;
				return -1;
			}
			return 0;
		}
		else if(cmin > 0)
		{
			/* not found */
			*id=0;
			return 1;
		}
		else if(*id == t->logfile_sz - 1 ||
			cmax > 0)
		{
			/* somewhere in current file */
			break;
		}
		(*id)++;
	}
	if(!valid)
	{
		fprintf(stderr, "No files to scan");
	}
	/* the answer is somewhere in *id file */
	*off = _ttail_from_search_from_end(t, *id, in);
	return 0;
}

inline int _ttail_search_file_binary_search(ttail_t* t, const struct tm* in,
	const struct tm** ftm)
{
	off_t cur, sz, d, prev;
	int ret, cmpres;
	off_t *off;
	size_t id;
	off = &(t->session->file.off);
	id = t->session->file.id;
	
	sz = t->session->file.file_sz[id];
	d = cur = sz / 2;
	prev = 0;
	cmpres = 0;
	while(1)
	{
		cur = _ttail_file_next_line(t->logfile[id]);
		if(cur < -1)
		{
			cur = _ttail_file_start_line(t->logfile[id]);
			if(cur < 0)
			{
				return -1;
			}
		}
		if(cur == prev)
		{
			*off = cur;
			return 0;
		}
		prev = cur;
		ret = _ttail_file_cur_cmp(t, id, in, &cmpres);
		if(ret < 0)
		{
			return -1;
		}
		else if(cmpres == 0)
		{
			*off = cur;
			break;
		}
		else if(cmpres < 0)
		{
			ret = _ttail_file_cur_cmp(t, id, in, &cmpres);
			if(cmpres >=0)
			{
				*off = cur;
				break;
			}
			d/=2;
			cur += d;
			cur %= t->session->file.file_sz[id];
		}
		else
		{
			d/=2;
			cur -= d;
		}
	}
	return 0;
}

int _ttail_search_closest_files_init(ttail_t* t)
{
	struct stat stat;
	FILE *fp;
	int fd;
	size_t i;
	off_t *file_sz;

	t->session = (ttail_search_t*)malloc(sizeof(ttail_search_file_t));
	if(!t->session)
	{
		perror("Unable to allocate memory for search session");
		goto _ttail_search_closest_files_alloc_session_err;
	}
	memset(t->session, 0, sizeof(ttail_search_file_t));

	file_sz = malloc(sizeof(off_t)*t->logfile_sz);
	if(!file_sz)
	{
		perror("Unable to allocate memory for file sizes");
		goto _ttail_search_closest_files_alloc_err;
	}
	t->session->file.file_sz = file_sz;
	#ifdef HUGEFILE
	t->session->file.sz_div = 0;
	#endif

	for(i=0;i<t->logfile_sz;i++)
	{
		fp = t->logfile[i];
		if(!fp)
		{
			file_sz[i] = 0;
			continue;
		}
		if((fd = fileno(fp)) < 0)
		{
			perror("Unable to get fp");
			goto _ttail_search_closest_files_err;
		}
		if(fstat(fileno(fp), &stat))
		{
			perror("Unable to get file size");
			goto _ttail_search_closest_files_err;
		}
		file_sz[i] = stat.st_size;
	}
	/* we got all real size, determining if we need a divisor */
	/*
	 * not implemented
	 */
	if(_ttail_search_closest_files_set_fsizes(t))
	{
		goto _ttail_search_closest_files_err;
	}

	t->session->file.buf_sz = 128;
	t->session->file.buf = malloc(t->session->file.buf_sz);
	if(!t->session->file.buf)
	{
		goto _ttail_search_closest_files_err;
	}

	return 0;

	_ttail_search_closest_files_err:
	free(file_sz);
	t->session->file.file_sz = NULL;
	_ttail_search_closest_files_alloc_err:
	free(t->session);
	_ttail_search_closest_files_alloc_session_err:
	return -1;
}

int _ttail_search_closest_files_set_fsizes(ttail_t* t)
{
	size_t i;
	off_t *vfile, *vsz;
	vfile = malloc(sizeof(off_t)*t->logfile_sz);
	if(!vfile)
	{
		perror("Unable to allocate memory for file size sum");
		return -1;
	}
	t->session->file.vfile = vfile;
	vsz = &(t->session->file.vsz);
	*vsz = 0;
	for(i=0; i<t->logfile_sz;i++)
	{
		vfile[i] = *vsz;
		#ifdef HUGEFILE
		*vsz += t->session->file.file_sz[i] >> t->session->file.sz_div;
		#else
		*vsz += t->session->file.file_sz[i];
		#endif
	}
	t->session->file.vpos = 0;
	return 0;
}

inline int _ttail_file_minmax(ttail_t* t, size_t id, struct tm tm[2])
{
	FILE *fp;
	long cur;
	memset(tm, 0, sizeof(struct tm)*2);
	fp = t->logfile[id];
	if(!fp)
	{
		return 1;
	}
	if(fseek(fp, 0, SEEK_SET) < 0)
	{
		perror("Unable to manipulate fp");
		return -1;
	}
	while(1)
	{
		if(ttail_getline(t, id) < 0)
		{
			return 1;
		}
		if(!ttail_logline2date(t, ttail_getline_buf(t), tm))
		{
			break;
		}
	}
	if(fseek(fp, -1, SEEK_END) < 0)
	{
		perror("Unable to manipulate fp");
		return -1;
	}
	while(1)
	{
		if((cur = _ttail_file_start_line(fp)) < 0)
		{
			return -1;
		}
		if(ttail_getline(t, id) < 0)
		{
			return 1;
		}
		if(!ttail_logline2date(t, ttail_getline_buf(t), tm+1))
		{
			break;
		}
		if(!cur)
		{
			return 1;
		}
		if(fseek(fp, cur-1, SEEK_SET) < 0)
		{
			perror("Unable to manipulate fp");
			return -1;
		}
	}
	return 0;
}

int _ttail_file_reopen(ttail_t* t, size_t id)
{
	if(t->logfile[id])
	{
		fclose(t->logfile[id]);
	}
	t->logfile[id] = fopen(t->logfile_name[id], "r");
	if(!t->logfile[id] && t->verbose > 2)
	{
		fprintf(stderr, "Unable to reopen '%s'\n",
			t->logfile_name[id]);
	}
	return t->logfile[id]?0:-1;
}

inline long _ttail_file_next_line(FILE* f)
{
	ssize_t s;
	size_t r;
	char *buff;
	long res;
	int c;
	
	r=0;
	buff = NULL;
	s = getline(&buff, &r, f);
	if(s == -1)
	{
		goto _ttail_file_next_line_err;
	}
	while(1)
	{
		c = getc(f);
		if(c == EOF)
		{
			return 0;
		}
		else if(c!='\n')
		{
			if(fseek(f, -1, SEEK_CUR)<0)
			{
				goto _ttail_file_next_line_err;
			}
			break;
		}
	}

	res = ftell(f);
	free(buff);
	return res;

	_ttail_file_next_line_err:
	free(buff);
	return -1;
}

inline long _ttail_file_start_line(FILE* f)
{
	#define _STARTLN_BUFFLEN 32
	long res; /* function result */
	long read_beg, cur, last, start;
	int read_sz;
	int c;

	if((start = ftell(f)) < 0)
	{
		return -1;
	}
	res = 0;
	read_beg = start;
	while(!res && start)
	{
		if(fseek(f, read_beg, SEEK_SET) < 0)
		{
			return -1;
		}
		start = read_beg;
		read_sz = start <= _STARTLN_BUFFLEN?start:_STARTLN_BUFFLEN;
		read_beg = start - read_sz;
		if(fseek(f, read_beg, SEEK_SET) < 0)
		{
			return -1;
		}
		last = -1; /* last pos we saw a '\n' */
		cur = read_beg;
		while(cur <= start)
		{
			c = getc(f);
			if(c == EOF)
			{
				if(!res)
				{
					return 0;
				}
				break;
			}
			else if (c =='\n')
			{
				last = cur;
			}
			else if(last >= 0)
			{
				res = cur;
				last = -1;
			}
			cur++;
		}
		if(!read_beg)
		{
			break;
		}
	}
	if(fseek(f, res, SEEK_SET) < 0)
	{
		return -1;
	}
	return res;
}

inline off_t _ttail_from_search_from_end(ttail_t* t , size_t id , 
	const struct tm* tm)
{
	FILE *f;
	struct tm curtm;
	off_t last;
	int ret;

	f = t->logfile[id];
	if(fseek(f, -1, SEEK_END) < 0)
	{
		return -1;
	}
	while(1)
	{
		last = _ttail_file_start_line(f);
		if(last < 0)
		{
			goto _ttail_from_search_from_end_err;
		}
		if(ttail_getline(t, id) < 0)
		{
			goto _ttail_from_search_from_end_err;
		}
		ret = ttail_logline2date(t, ttail_getline_buf(t), &curtm);
		if(ret < 0)
		{
			goto _ttail_from_search_from_end_err;
		}
		if(!ret && !ttail_tm_cmp(&curtm, tm))
		{
			/* found */
			break;
		}
		if(last == 0)
		{
			/* considere the begining of the file as the answer */
			return 0;
		}
		if(fseek(f, last-1, SEEK_CUR))
		{
			goto _ttail_from_search_from_end_err;
		}
	}
	return last;
	
	_ttail_from_search_from_end_err:
	return -1;
}

inline int _ttail_file_off_cmp(ttail_t* t, size_t id, off_t off,
	const struct tm* tm, int *res)
{
	if(fseek(t->logfile[id], off, SEEK_CUR))
	{
		return -1;
	}
	if(ttail_getline(t, id) < 0)
	{
		return -1;
	}
	return _ttail_file_cur_cmp(t, id, tm, res);
}
inline int _ttail_file_cur_cmp(ttail_t* t, size_t id, const struct tm* tm ,
	int* res)
{
	int ret;
	struct tm ctm;
	ret = ttail_logline2date(t, ttail_getline_buf(t), &ctm);
	if(ret < 0)
	{
		return -1;
	}
	else if(ret == 1)
	{
		return 1;
	}
	*res = ttail_tm_cmp(&ctm, tm);
	return 0;

}

void _ttail_search_file_free(ttail_t* t)
{
	if(!t->session)
	{
		return;
	}
	if(t->session->file.buf)
	{
		free(t->session->file.buf);
	}
	if(t->session->file.file_sz)
	{
		free(t->session->file.file_sz);
	}
	if(t->session->file.vfile)
	{
		free(t->session->file.vfile);
	}
	free(t->session);
}