apogee/src/model/fetch.c

#define _GNU_SOURCE
#include "fetch.h"

// Model
#include "model/crypto/keys.h"
#include "model/crypto/http_sign.h"
#include "model/server.h"
#include "model/peer.h"

// Submodules
#include "http/client/client.h"
#include "http/url.h"
#include "util/format.h"
#include "ap/object.h"
#include "sha256/sha256.h"

// Standard Library
#include <sys/stat.h>
#include <time.h>
#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <ctype.h>

size_t fetch_handle_header( char* header, size_t size, size_t nitems, void* user )
{
	int bytes = size * nitems;
	int result = bytes;

	struct fetch_data* fd = user;
	if( !fd ) {
		printf( "No fetch data provided, skippig handling.\n" );
		return result;
	}

	//printf( "? Header: |%.*s|\n", bytes, header );

	if( 0 != strncmp("onion-location: http://",header,sizeof("onion-location: http://")-1 ) ) {
		return result;
	}

	header += (sizeof("onion-location: http://")-1);
	bytes  -= (sizeof("onion-location: http://")-1);

	//printf( "? Header: |%.*s|\n", bytes, header );

	for( int i = 0; i < bytes; ++i ) {
		if( header[i] == '/' ) {
			header[i] = '\0';
			break;
		}
	}

	char* onion_host = strndup( header, bytes );
	int pos = strlen(onion_host) - 1;
	while( !isalnum(onion_host[pos]) && !( onion_host[pos] == '.' ) ) {
		onion_host[pos] = '\0';
		pos -= 1;
	}

	//printf( "+ Onion Host: |%s|\n", onion_host );
	if( fd->p ) {
		printf( "Updating peer onion host to %s\n", onion_host );
		free(fd->p->tor_hidden_service);
		fd->p->tor_hidden_service = onion_host;
		onion_host = NULL;
	}

	free(onion_host);

	return result;
}
static bool do_fetch_tor( const char* uri, struct fetch_data* fd, const char* result_filename )
{
	char user_agent[512];
	snprintf( user_agent, sizeof(user_agent), "User-Agent: %s; +https://%s/owner/actor", g_server->user_agent, g_server->domain );

	char proxy[512];
	snprintf( proxy,512, "socks5h://localhost:%d", g_server->tor_socks_port );

	FILE* result = fopen(result_filename,"w");
	if( !result ) { return false; }

	long status_code;
	const void* request[] = {
		HTTP_REQ_URL, uri,
		HTTP_REQ_HEADER, "Accept: application/ld+json; profile=\"https://www.w3.org/ns/activitystreams\"",
		HTTP_REQ_HEADER, user_agent,
		HTTP_REQ_OUTFILE, result,
		HTTP_REQ_RESULT_STATUS, &status_code,
		HTTP_RES_HEADER_CALLBACK, fetch_handle_header, (void*)fd,
		HTTP_REQ_PROXY, proxy,
		HTTP_REQ_TIMEOUT, (void*)10,
		NULL,
	};
	printf( "GET %s\n", uri );
	http_client_do( request );
	printf( "GET %s -> %ld\n", uri, status_code );

	fclose(result);

	if( status_code != 200 ) {
		unlink(result_filename);
	}

	return status_code == 200;
}
static bool do_fetch_clearnet( const char* uri, struct fetch_data* fd, const char* result_filename )
{
	char user_agent[512];
	snprintf( user_agent, sizeof(user_agent), "User-Agent: %s; +https://%s/owner/actor", g_server->user_agent, g_server->domain );

	FILE* result = fopen( result_filename, "w" );
	if( !result ) { return false; }

	long status_code;
	const void* request[] = {
		HTTP_REQ_URL, uri,
		HTTP_REQ_HEADER, "Accept: application/ld+json; profile=\"https://www.w3.org/ns/activitystreams\"",
		HTTP_REQ_HEADER, user_agent,
		HTTP_REQ_OUTFILE, result,
		HTTP_REQ_RESULT_STATUS, &status_code,
		HTTP_RES_HEADER_CALLBACK, fetch_handle_header, (void*)fd,
		NULL,
	};
	printf( "GET %s\n", uri );
	http_client_do( request );
	printf( "GET %s -> %ld\n", uri, status_code );

	fclose(result);

	if( status_code != 200 ) {
		unlink( result_filename );
	}

	return status_code == 200;
}
static bool do_fetch_signed_clearnet( const char* uri, struct fetch_data* fd, const char* result_filename )
{
	FILE* result = fopen(result_filename,"w");
	if( !result ) { return false; }

	char user_agent[512];
	snprintf( user_agent, sizeof(user_agent), "User-Agent: %s; +https://%s/owner/actor", g_server->user_agent, g_server->domain );

	// Load crypto keys
	struct crypto_keys* keys = crypto_keys_new();
	if( !crypto_keys_load_private( keys, "data/owner/private.pem" ) ) {
		printf( "Failed to load private key\n" );
		return false;
	}

	// TODO: do signed fetch
	struct http_signature hs;
	memset( &hs, 0, sizeof(hs) );
	hs.input.method = "get";
	hs.input.url = uri;

	if( !http_signature_make( keys, &hs ) ) {
		return false;
	}

	char date_header[512];
	snprintf( date_header, sizeof(date_header), "Date: %s", hs.date );
	printf( "date_header = %s\n", date_header );

	char sign_header[512];
	snprintf( sign_header, sizeof(sign_header), "Signature: keyId=\"https://%s/owner/actor#mainKey\",headers=\"(request-target) host date content-length digest\",signature=\"%s\"",
		g_server->domain,
		hs.signature
	);
	printf( "sign_header = %s\n", sign_header );

	char digest_header[512];
	snprintf( digest_header, sizeof(digest_header), "Digest: %s", hs.digest );
	printf( "digest_header = %s\n", digest_header );

	char content_length_header[512];
	snprintf( content_length_header, sizeof(content_length_header), "Content-Length: %d", hs.content_length );
	printf( "content_length_header = %s\n", content_length_header );

	long status_code = 0;
	const void* request[] = {
		HTTP_REQ_URL, uri,
		HTTP_REQ_HEADER, user_agent,
		HTTP_REQ_HEADER, date_header,
		HTTP_REQ_HEADER, sign_header,
		HTTP_REQ_HEADER, content_length_header,
		HTTP_REQ_HEADER, digest_header,
		HTTP_REQ_HEADER, "Accept: application/ld+json; profile=\"https://www.w3.org/ns/activitystreams\"",
		HTTP_REQ_OUTFILE, result,
		HTTP_REQ_RESULT_STATUS, &status_code,
		NULL,
	};

	if( !http_client_do( request ) ) {
		printf( "Failed to perform get, status_code = %ld\n", status_code );
	}
	printf( "GET %s -> %ld (first signed)\n", uri, status_code );
	fclose(result);

	if( status_code == 200 ) { return true; }

	result = fopen(result_filename,"w");

	const void* request2[] = {
		HTTP_REQ_URL, uri,
		HTTP_REQ_HEADER, user_agent,
		HTTP_REQ_HEADER, date_header,
		HTTP_REQ_HEADER, sign_header,
		HTTP_REQ_HEADER, content_length_header,
		HTTP_REQ_HEADER, digest_header,
		HTTP_REQ_HEADER, "Accept: application/activity+json",
		HTTP_REQ_OUTFILE, result,
		HTTP_REQ_RESULT_STATUS, &status_code,
		HTTP_REQ_TIMEOUT, (void*)2,
		NULL,
	};

	if( !http_client_do( request2 ) ) {
		printf( "Failed to perform get, status_code = %ld\n", status_code );
	}
	printf( "GET %s -> %ld (second signed)\n", uri, status_code );
	fclose(result);

	if( status_code == 200 ) { return true; }

	unlink(result_filename);
	return false;
}

static bool do_fetch( const char* uri, const char* fh )
{
	// Setup fetch data
	struct fetch_data fd;
	memset(&fd,0,sizeof(fd));
	{
		char host_domain[512];
		if( url_get_domain( uri, host_domain, sizeof(host_domain) ) ) {
			fd.p = peer_from_domain(host_domain);
		}
	}

	bool result = false;

	if( fd.p && fd.p->last_tor_delivery && do_fetch_tor( uri, &fd, fh ) ) {
		goto success;
	}
	if( do_fetch_clearnet( uri, &fd, fh ) ) {
		goto success;
	}
	if( do_fetch_signed_clearnet( uri, &fd, fh ) ) {
		goto success;
	}

	goto failed;
cleanup:
	// Save peer data
	if( fd.p ) {
		peer_save( fd.p );
		peer_free( fd.p );
	}
	return result;
failed:
	result = false;
	goto cleanup;
success:
	result = true;
	goto cleanup;
}

bool pull_remote_file( const char* filename, const char* uri )
{
	printf( "* Fetching %s\n", uri );
	char tmp_filename[512];
	snprintf( tmp_filename,512, "%s.tmp", filename );

	bool result = false;
	if( do_fetch( uri, tmp_filename ) ) {
		rename(tmp_filename,filename);
		result = true;
	}

	return result;
}
bool pull_remote_file_if_older( const char* filename, const char* uri, int seconds )
{
	struct stat s;
	char tmp_filename[512];
	snprintf( tmp_filename,512, "%s.tmp", filename );

	// Skip download if .tmp file exists from a failed fetch and is within the timeout window
	if( 0 == stat(tmp_filename, &s ) ) {
		if( time(NULL) - s.st_mtime <= seconds ) {
			printf( "Fetch of %s supressed due to failed fetch within %d seconds\n", uri, seconds );
			return false;
		}
	}

	if( 0 != stat(filename, &s ) ) { goto pull; }
	if( time(NULL) - s.st_mtime > seconds ) { goto pull; }

	return true;
pull:
	return pull_remote_file( filename, uri );
}

char* fetch_remote_file_to_cache( const char* uri, int seconds )
{
	char* filename = NULL;
	char hash[65];
	sha256_easy_hash_hex( uri, strlen(uri), hash );
	hash[64] = 0;

	mkdir( "data/cache", 0755 );

	asprintf( &filename, "data/cache/%s.dat", hash );
	printf( "Using %s for %s\n", filename, uri );

	if( !pull_remote_file_if_older(filename,uri,seconds) ) {
		free(filename);
		return NULL;
	}

	return filename;
}

enum {
	//LIFE = 60*60*24*2 // 2 days
	LIFE = 60*30 // 30 minutes
};

struct ap_object* fetch_ap_object_ref( const char* uri )
{
	char* filename = fetch_remote_file_to_cache(uri, LIFE );

	if( !filename ) { return NULL; }

	struct ap_object* res = ap_object_from_file(filename);
	free(filename);
	return res;
}