//-------------------------------------------------------------------
//	grabfile.cpp
//
//	This is our initial prototype for an HTTP client which lets
//	a user obtain a file-object from an HTTP server by entering 
//	that object's URL as a string-argument on the command-line.  
//
//	     to compile:  $ g++ grabfile.cpp -o grabfile
//	     to execute:  $ ./grabfile <url>
//
//	NOTE: A test-case: $ ./grabfile www.cs.usfca.edu/index.html
//
//	programmer: ALLAN CRUSE
//	written on: 17 MAR 2009
//-------------------------------------------------------------------


#include <netdb.h>	// for gethostbyname()
#include <stdio.h>	// for printf(), perror() 
#include <fcntl.h>	// for open() 
#include <stdlib.h>	// for exit() 
#include <unistd.h>	// for read(), write(), close() 
#include <string.h>	// for strncpy()
#include <arpa/inet.h>	// socket(), connect() 


int main( int argc, char **argv )
{
	//---------------------------------------------
	// setup default-values for the URL components
	//---------------------------------------------
	char	proto[ 5 ] = "http";
	char	host[ 64 ] = "localhost";
	char	port[  8 ] = "80";
	char	path[ 80 ] = "./index.html";
	char	url[ 128 ] = { 0 };

	//------------------------------------------- 
	// now copy the entire command-line argument
	//-------------------------------------------
	if ( argc == 1 ) { fprintf( stderr, " missing URL \n" ); exit(1); }
	int	len = strlen( argv[ 1 ] );
	if ( len < sizeof( url ) ) strcpy( url, argv[ 1 ] );
	else	{ fprintf( stderr, " argument-string too long \n"); exit(1); }

	//-------------------------------------------
	// strip away the 'method' field, if present 
	//------------------------------------------- 
	char	*ptr = strstr( url, "http://" );
	if ( ptr ) strcpy( url, ptr+7 );
  	else if ( strstr( url, "://" ) )
		{
		fprintf( stderr, " protocol is not \'http\' \n" );
		exit(1);
		}

	//--------------------------------------------------
	// first let's separate away the 'path', if present 
	//-------------------------------------------------- 
	strtok_r( url, "/", &ptr );
	len = ( ptr ) ? strlen( ptr ) : 0;
	if ( len ) strncpy( path+2, ptr, 77 );

	//-------------------------------------------------------
	// next let's separate the 'host' and 'port', if present
	//-------------------------------------------------------

	// Here we use the 'strspn()' and 'strcspn()' functions 
	// from the string-library to assist in parsing various
	// degenerate cases (in which we want to keep defaults)
	// such as: no hostname preceeds the initial colon, and
	// no port-number follows the initial run of (possibly)
	// several consecutive colons, or colons do not appear. 

	// we will use these four counters to aid us in parsing 
	// of what may be a 'malformed' URL string, keeping our
	// default-strings in case host and/or port are missing 	

	int	run0 = strcspn( url, ":" );
	int	run1 = strspn( url+run0, ":" );
	int	run2 = strcspn( url+run0+run1, ":" );
	int	run3 = strspn( url+run0+run1+run2, ":" );

	// printf( " run of %d noncolons \n", run0 );
	// printf( " run of %d colons \n", run1 );
	// printf( " run of %d noncolons \n", run2 );
	// printf( " run of %d colons \n", run3 );
	
	strtok( url+run0+run1, ":" );	// to omit trailing garbage 

	if ( run0 ) 
		{ 
		// the normal case: hostname-field is present
		strtok_r( url, ":", &ptr );
		if ( ptr ) ptr += strspn( ptr, ":" );
		len = ( ptr ) ? strlen( ptr ) : 0;
		if ( len ) strncpy( port, ptr, 7 );
		strncpy( host, url, 63 );
		}
	else	{ 
		// degenerate case: the hostname-field is absent
		strtok( url, ":" );
		ptr = url + strspn( url, ":" );
		len = strcspn( ptr, ":" );
		if ( len ) strncpy( port, ptr, 7 );
		}

	//------------------------------------------------------
	// show the user how we have interpreted the URL-string 
	//------------------------------------------------------  
	printf( "\n Uniform Resource Locator is: " );
	printf( "<%s://%s:%s/%s> \n", proto, host, port, path );

	//-------------------------------------------------
	// now we get the server's hostname and IP-address
	//-------------------------------------------------
	char	peername[ 64 ] = { 0 };
	strncpy( peername, host, 63 );	
	struct hostent	*pp = gethostbyname( peername );
	if ( !pp ) { herror( "gethostbyname" ); exit(1); }
	char	peeraddr[ 16 ] = { 0 };
	strcpy( peeraddr, inet_ntoa( *(in_addr*)pp->h_addr ) );

	//---------------------------------------------------
	// we convert the port-number from ascii to a number 
	//--------------------------------------------------- 
	int	portno = atoi( port );

	//---------------------------------------------------------
	// we initialize a socket-address structure for the server
	//--------------------------------------------------------- 
	struct sockaddr_in	paddr = { 0 };
	socklen_t		palen = sizeof( paddr );
	paddr.sin_family	= AF_INET;
	paddr.sin_port		= htons( portno );
	paddr.sin_addr.s_addr	= inet_addr( peeraddr );

	//------------------------------------------------------
	// we create a socket for communicating with the server
	//------------------------------------------------------ 
	int	sock = socket( AF_INET, SOCK_STREAM, IPPROTO_TCP );
	if ( sock < 0 ) { perror( "socket" ); exit(1); }

	//-------------------------------------------
	// we establish a connection with the server
	//-------------------------------------------
	if ( connect( sock, (sockaddr*)&paddr, palen ) < 0 )
		{ perror( "connect" ); exit(1); }
	printf( "\n connected to \'%s\' on port %d \n", peername, portno );

	//------------------------------------
	// we create the HTTP Request-message 
	//------------------------------------
	char	request[ BUFSIZ ];
	len = 0;
	len += sprintf( request+len, "GET " );
	len += sprintf( request+len, "%s ", path );
	len += sprintf( request+len, "HTTP/1.0\r\n" );
	len += sprintf( request+len, "\r\n" );

	//-------------------------------------------------
	// we write the HTTP Request-message to the stream
	//-------------------------------------------------
	int	txbytes = write( sock, request, len );
	if ( txbytes < 0 ) { perror( "write" ); exit(1); }
	fflush( fdopen( sock, "rw" ) );
	printf( "\n sent %d bytes to server \n", txbytes );

	//-------------------------------------------------
	// our experiments showed some delay was essential
	//------------------------------------------------- 
	usleep( 50000 );	// 50-millisecond delay

	//---------------------------------------------------
	// we read the HTTP Response-message from the stream	
	//---------------------------------------------------
	char	response[ BUFSIZ ] = { 0 };
	int	rxbytes = read( sock, response, BUFSIZ );
	if ( rxbytes < 0 ) { perror( "read" ); exit(1); } 
	printf( " received %d bytes from the server \n", rxbytes );
	
	//--------------------------------------
	// we display the HTTP Response-message
	//--------------------------------------
	printf( "\n" );
	printf( "\e[37;40m" );
	printf( "%s", response );
	printf( "\e[0m" );
	printf( "\n" );

	//-----------------------------------------
	// we close our connection with the server
	//-----------------------------------------
	close( sock );
}