/**************************************************************************

	Program:	virtlogsplit
	Why:		Disperses a log file containing virtual domain names into
				several files readable by analog.
	
	OS:			Developed under Windows NT Ver. 4.0, but runs on Unix.
	
	Autor:		Thomas Belmer
	created:	2000-05-25



Invoke
------

	virtlogsplit logfile directory_for_dispersed_files


  
History
-------

who			when		what
---------------------------------------------------------------------------
Th. Belmer	2000-05-25	created.
Th. Belmer	2000-05-10	For testing purposes corrupt lines get printed.
Th. Belmer	2000-06-10	Algorythm for recognition of corrupt lines changed.
Th. Belmer	2001-06-01	Extract only one given virtual domain.

**************************************************************************/


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

#ifdef unix
#else
#include <crtdbg.h>
#endif

#ifndef FALSE
#define FALSE			0
#define TRUE			!FALSE
#endif

#define ONE_LINE		10000

#ifdef unix
#define C_PATH			'/'
#define CH_PATH			"/"
#else
#define C_PATH			'\\'
#define CH_PATH			"\\"
#endif


void GiveReturn (int iCode);
unsigned long ReadAndProcessLogFile (char *cfilename, char *cpath_to_put, char *vDomain);


// Terminate program by returning an error message.
void GiveReturn (int iCode)
{

	printf("Error %d.\n",iCode);
	exit (iCode);

}


unsigned long ReadAndProcessLogFile (char *cfilename, char *cpath_to_put, char *vDomain)
{
	
	FILE			*fstream		= NULL;
	FILE			*fOutput		= NULL;
	char			*cvDomain		= NULL;
	char			*cLineBuffer	= NULL;
	char			*cDatestart		= NULL;
	char			*cOutput		= NULL;
	unsigned long	ulLinesProcessed	= 0;
	unsigned long	ulWritten_lines		= 0;
	unsigned long	ulCorrupt_lines		= 0;
	unsigned long	ulOpening_errors	= 0;
	unsigned long	ulWriting_errors	= 0;

	size_t			st_path_to_put	= 0;
	size_t			st_Len			= 0;
	size_t			st_vDom			= 0;
	int				iWrite			= 0;

	// Line variables.
	char			*cOutput_line_start	= NULL;
	char			*cAfter_IP			= NULL;


	if ( (cLineBuffer = malloc (ONE_LINE+1)) == NULL ) GiveReturn (3);
	st_path_to_put = strlen(cpath_to_put);

	if (vDomain != NULL)
	{
		st_vDom = strlen (vDomain);
		if (st_vDom > 1) cvDomain = malloc (st_vDom +1);
		if (cvDomain)
		{
			strcpy (cvDomain, vDomain);
			st_Len = st_vDom;
			// Virtual Domain must only contain "_".
			for (st_Len; st_Len > 0; st_Len--) if ( *(cvDomain+st_Len-1) == '.' ) *(cvDomain+st_Len-1) = '_';
		}
	}

	// Read the file line by line or use stdin instead.

	if ( !strcmp (cfilename, "STDIN") ) fstream = stdin;
	else fstream = fopen(cfilename,"r");

	if (fstream == NULL) GiveReturn (1);

	while (!feof(fstream))
	{
		if ( fgets(cLineBuffer, ONE_LINE, fstream) != NULL )
		{

			// Example of a line.
			// www.bieruhr.de 194.24.201.114 - - [16/May/2000:21:06:02 +0200] "GET /bilder/rauten.gif HTTP/1.1" 304 -
			// www.bieruhr.de 194.24.201.114 - accessgroup [16/May/2000:21:06:02 +0200] "GET /bilder/rauten.gif HTTP/1.1" 304 -

			cAfter_IP = strstr(cLineBuffer," - ");
			cOutput_line_start = strchr(cLineBuffer, ' ');
			// Check whether the line is ok.
			if (	(cAfter_IP == NULL) ||
					(cOutput_line_start == NULL) ||
					(cOutput_line_start >= cAfter_IP)
													)
			{
				// Delete this line if you don't want to display corrupt lines.
				printf("Corrupt line:\n%s\n",cLineBuffer);
				
				ulCorrupt_lines++;
			} else
			{
				
				*cOutput_line_start = '\0';
				cOutput_line_start++;

				// File name.
				st_Len = strlen (cLineBuffer);
				if ( (cOutput = malloc (st_path_to_put + st_Len + 10)) == NULL )
				{
					fclose(fstream);
					GiveReturn (3);
				}
				strcpy (cOutput, cpath_to_put);
				
				// Last character must be path character.
				if ( *(cpath_to_put+st_path_to_put-1) != C_PATH ) strcat (cOutput, CH_PATH);
				
				// Replace '.' through '_' in domain name.
				for (st_Len; st_Len > 0; st_Len--) if ( *(cLineBuffer+st_Len-1) == '.' ) *(cLineBuffer+st_Len-1) = '_';

				// The complete filename.
				strcat (cOutput, cLineBuffer);

				// Unfortunately Windows wants to have a file extension.
				strcat (cOutput, ".log");

				if (cvDomain)
				{
					// printf ("%s, %s#\n", cvDomain, cLineBuffer);
					if (!strncmp (cvDomain, cLineBuffer, st_vDom)) iWrite = 1; else iWrite = 0;
				} else iWrite = 1;					

				if (iWrite)
				{
					// Write the line to the file.
					fOutput = fopen(cOutput,"a+");
					if (fOutput != NULL)
					{
						if ( fputs (cOutput_line_start, fOutput) < 0 ) ulWriting_errors++;
						else ulWritten_lines++;
						fclose (fOutput);
					} else ulOpening_errors++;
				}
				//printf ("%s\n",cOutput);

				ulLinesProcessed++;

			}


		} else break;
	}

	fclose (fstream);

	printf("Summary of run on file %s:\n", cfilename);
	printf("Total processed lines: \t%15d\n",ulLinesProcessed + ulCorrupt_lines);
	printf("Total lines ok: \t%15d\n", ulLinesProcessed);
	printf("Total written lines: \t%15d\n",ulWritten_lines);
	printf("Total corrupt lines: \t%15d\n", ulCorrupt_lines);
	printf("Total opening errors: \t%15d\n", ulOpening_errors);
	printf("Total writing errors: \t%15d\n", ulWriting_errors);
	return (ulLinesProcessed);

}


int main(int argc, char *argv[], char *envp[])
{

	char	*chlogfilename			= NULL;
	size_t	st_Len					= 0;
	unsigned long	ulReturnValue	= 0;

	double	dStartZeit;
	time_t	tStartZeit;


	// Welcome message.
	if ( (argc < 3) || (argc > 4) )
	{
		printf("virtlogsplit version 1.0 - built "__DATE__" "__TIME__"\n");
		printf("(c) 2000 Thomas Belmer, Email: Thomas.Belmer@asamnet.de\n\n");
		printf("USAGE: virtlogsplit logfile path_to_put [virtdomain]\n\n");
		printf("PARAMETERS:\n");
		printf("logfile     = logfile to process, use STDIN for standard input\n");
		printf("path_to_put = path where to put the split files.\n");
		printf("virtdomain  = extract only the given virtual domain.\n\n");
		printf("Examples: virtlogsize /usr/logs/access_log /usr/logs\n");
		printf("          cat /usr/logs/access_log | virtlogsplit STDIN /usr/logs\n\n");
		printf("FILE CONTENTS:\n");
		printf("logfile before processing: ");
		printf("www.bieruhr.de 194.24.201.114 - - [16/May/2000...\n");
		printf("files in path_to_put:      ");
		printf("194.24.201.114 - - [16/May/2000...\n\n");
		printf("RETURN CODES:\n");
		printf("0 = Success\n");
		printf("1 = Invalid logfile parameter\n");
		printf("2 = Invalid path_to_put parameter\n");
		printf("3 = Out of memory\n\n");
		return (1);
	}


	// Parameters ok?
	if ( (st_Len=strlen(argv[1])) < 1 ) GiveReturn (1);
	if ( (st_Len=strlen(argv[2])) < 1 ) GiveReturn (2);
	

	dStartZeit = (double) clock() / CLOCKS_PER_SEC;
	time (&tStartZeit);

	printf("\n");
	printf("virtlogsplit version 1.01 - built "__DATE__" "__TIME__"\n\n");
	printf("Start: %s\n", ctime(&tStartZeit) );

	ulReturnValue = ReadAndProcessLogFile (argv[1], argv[2], argv[3]);
	// printf("%d lines in %s processed.\n", ulReturnValue, argv[1]);
	printf("%lf seconds runtime.\n", ((double) clock() / CLOCKS_PER_SEC) - dStartZeit);

	return(0);

}