parser3/src/main/pa_common.C - diff

Return to pa_common.C CVS log

Up to [parser3project] / parser3 / src / main

Diff for /parser3/src/main/pa_common.C between versions 1.143.2.21.2.21 and 1.175

version 1.143.2.21.2.21, 2003/04/02 10:35:07	version 1.175, 2004/03/01 11:43:01
Line 1	Line 1
/** @file	/** @file
Parser: commonly functions.	Parser: commonly functions.

Copyright(c) 2001-2003 ArtLebedev Group (http://www.artlebedev.com)	Copyright(c) 2001-2004 ArtLebedev Group (http://www.artlebedev.com)
Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)	Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
*/	*/

static const char* IDENT_COMMON_C="$Date$";	static const char * const IDENT_COMMON_C="$Date$";

#include "pa_common.h"	#include "pa_common.h"
#include "pa_exception.h"	#include "pa_exception.h"
#include "pa_hash.h"	#include "pa_hash.h"
#include "pa_globals.h"	#include "pa_globals.h"
	#include "pa_request_charsets.h"
	#include "pa_charsets.h"

#define PA_HTTP	#define PA_HTTP

#ifdef PA_HTTP	#ifdef PA_HTTP
#include "pa_vstring.h"	#include "pa_vstring.h"
#include "pa_vint.h"	#include "pa_vint.h"
	#include "pa_vhash.h"
	#include "pa_vtable.h"

#ifdef CYGWIN	#ifdef CYGWIN
#define _GNU_H_WINDOWS32_SOCKETS	#define _GNU_H_WINDOWS32_SOCKETS
Line 60 int PASCAL closesocket(SOCKET);	Line 64 int PASCAL closesocket(SOCKET);
# endif	# endif
#endif	#endif

// locking constants
//#define PA_DEBUG_NO_LOCKING

#ifdef PA_DEBUG_NO_LOCKING

#ifdef HAVE_FLOCK

static int lock_shared_blocking(int fd) { return flock(fd, LOCK_SH); }
static int lock_exclusive_blocking(int fd) { return flock(fd, LOCK_EX); }
static int lock_exclusive_nonblocking(int fd) { return flock(fd, LOCK_EX \|\| LOCK_NB); }
static int unlock(int fd) { return flock(fd, LOCK_UN); }

#else
#ifdef HAVE__LOCKING

#define FLOCK(operation) lseek(fd, 0, SEEK_SET); return _locking(fd, operation, 1)
static int lock_shared_blocking(int fd) { FLOCK(_LK_LOCK); }
static int lock_exclusive_blocking(int fd) { FLOCK(_LK_LOCK); }
static int lock_exclusive_nonblocking(int fd) { FLOCK(_LK_NBLCK); }
static int unlock(int fd) { FLOCK(_LK_UNLCK); }

#else
#ifdef HAVE_FCNTL

#define FLOCK(cmd, arg) struct flock ls={arg, SEEK_SET}; return fcntl(fd, cmd, &ls)
static int lock_shared_blocking(int fd) { FLOCK(F_SETLKW, F_RDLCK); }
static int lock_exclusive_blocking(int fd) { FLOCK(F_SETLKW, F_WRLCK); }
static int lock_exclusive_nonblocking(int fd) { FLOCK(F_SETLK, F_RDLCK); }
static int unlock(int fd) { FLOCK(F_SETLK, F_UNLCK); }

#else
#ifdef HAVE_LOCKF

#define FLOCK(fd, operation) lseek(fd, 0, SEEK_SET); return lockf(fd, operation, 1)
static int lock_shared_blocking(int fd) { FLOCK(F_LOCK); } // on intel solaris man doesn't have doc on shared blocking
static int lock_exclusive_blocking(int fd) { FLOCK(F_LOCK); }
static int lock_exclusive_nonblocking(int fd) { FLOCK(F_TLOCK); }
static int unlock(int fd) { FLOCK(F_TLOCK); }

#else

#error unable to find file locking func

#endif
#endif
#endif
#endif

#else
static int lock_shared_blocking(int fd) { return 0; }
static int lock_exclusive_blocking(int fd) { return 0; }
static int lock_exclusive_nonblocking(int fd) { return 0; }
static int unlock(int fd) { return 0; }

#endif

// defines for globals	// defines for globals

#define FILE_STATUS_NAME "status"	#define FILE_STATUS_NAME "status"
Line 130 const String file_status_name(FILE_STATU	Line 78 const String file_status_name(FILE_STATU
#define HTTP_TIMEOUT_NAME "timeout"	#define HTTP_TIMEOUT_NAME "timeout"
#define HTTP_HEADERS_NAME "headers"	#define HTTP_HEADERS_NAME "headers"
#define HTTP_ANY_STATUS_NAME "any-status"	#define HTTP_ANY_STATUS_NAME "any-status"
	#define HTTP_CHARSET_NAME "charset"
	#define HTTP_TABLES_NAME "tables"

// statics	// statics

Line 137 static const String http_method_name(HTT	Line 87 static const String http_method_name(HTT
static const String http_timeout_name(HTTP_TIMEOUT_NAME);	static const String http_timeout_name(HTTP_TIMEOUT_NAME);
static const String http_headers_name(HTTP_HEADERS_NAME);	static const String http_headers_name(HTTP_HEADERS_NAME);
static const String http_any_status_name(HTTP_ANY_STATUS_NAME);	static const String http_any_status_name(HTTP_ANY_STATUS_NAME);
	static const String http_charset_name(HTTP_CHARSET_NAME);
	static const String http_tables_name(HTTP_TABLES_NAME);

// defines	// defines

Line 164 void fix_line_breaks(char *str, size_t&	Line 116 void fix_line_breaks(char *str, size_t&
} else // \r, not \n = Macintosh	} else // \r, not \n = Macintosh
bol=eol+1;	bol=eol+1;
}	}
// last piece without \r, including terminating 0	// last piece without \r
if(dest!=bol)	if(dest!=bol)
memcpy(dest, bol, eob-bol);	memcpy(dest, bol, eob-bol);
	str[length]=0; // terminating
}	}

char* file_read_text(Charset& charset,	char* file_read_text(Request_charsets& charsets,
const String& file_spec,	const String& file_spec,
bool fail_on_read_problem,	bool fail_on_read_problem,
HashStringValue* params/, HashStringValue * out_fields*/) {	HashStringValue* params/, HashStringValue * out_fields*/) {
File_read_result file=	File_read_result file=
file_read(charset, file_spec, true, params, fail_on_read_problem);	file_read(charsets, file_spec, true, params, fail_on_read_problem);
return file.success?file.str:0;	return file.success?file.str:0;
}	}

#ifdef PA_HTTP	#ifdef PA_HTTP
//http request stuff	//http request stuff

	#undef CRLF
	#define CRLF "\r\n"

static bool set_addr(struct sockaddr_in addr, const char host, const short port){	static bool set_addr(struct sockaddr_in addr, const char host, const short port){
memset(addr, 0, sizeof(*addr));	memset(addr, 0, sizeof(*addr));
addr->sin_family=AF_INET;	addr->sin_family=AF_INET;
addr->sin_port=htons(port);	addr->sin_port=htons(port);
if(host) {	if(host) {
if(struct hostent *hostIP=gethostbyname(host))	ulong packed_ip=inet_addr(host);
	struct hostent *hostIP = packed_ip==INADDR_NONE?
	gethostbyname(host)
	: gethostbyaddr((char *)&packed_ip, sizeof(packed_ip), AF_INET);
	if(hostIP)
memcpy(&addr->sin_addr, hostIP->h_addr, hostIP->h_length);	memcpy(&addr->sin_addr, hostIP->h_addr, hostIP->h_length);
else	else
return false;	return false;
Line 195 static bool set_addr(struct sockaddr_in	Line 155 static bool set_addr(struct sockaddr_in
return true;	return true;
}	}

static int http_read_response(String& response, int sock, bool fail_on_status_ne_200){	static int http_read_response(char*& response, size_t& response_size, int sock, bool fail_on_status_ne_200){
	response=(char)pa_malloc_atomic(1/terminator*/); // setting memory block type
	response[(response_size=0)]=0;
int result=0;	int result=0;
size_t EOLat=0;	char* EOLat=0;
while(true) {	while(true) {
char *buf=new(PointerFreeGC) char[MAX_STRING];	char buf[MAX_STRING*10];
ssize_t size=recv(sock, buf, MAX_STRING, 0);	ssize_t received_size=recv(sock, buf, sizeof(buf), 0);
if(size<=0)	if(received_size<=0)
break;	break;
response.append_strdup(buf, size, String::L_TAINTED);	response=(char)pa_realloc(response, response_size+received_size+1/terminator*/);
if(!result && (EOLat=response.pos("\r\n", 2))!=STRING_NOT_FOUND) { // checking status in first response	memcpy(response+response_size, buf, received_size);
const String& status_line=response.mid(0, (size_t)EOLat);	response_size+=received_size;
	response[response_size]=0;

	if(!result && (EOLat=strstr(response, CRLF))) { // checking status in first response
	const String status_line(pa_strdup(response, EOLat-response));
ArrayString astatus;	ArrayString astatus;
size_t pos_after=0;	size_t pos_after=0;
status_line.split(astatus, pos_after, " ");	status_line.split(astatus, pos_after, " ");
Line 223 static int http_read_response(String& re	Line 189 static int http_read_response(String& re
else	else
throw Exception("http.response",	throw Exception("http.response",
0,	0,
"bad response from host - no status found (size=%lu)", response.length());	"bad response from host - no status found (size=%u)", response_size);
}	}

/* ******************** request ************************* */	/* ******************** request ************************* */

#if defined(SIGALRM) && defined(HAVE_SIGSETJMP) && defined(HAVE_SIGLONGJMP)	#if defined(SIGALRM) && defined(HAVE_SIGSETJMP) && defined(HAVE_SIGLONGJMP)
# define WE_CAN_USE_ALARM	# define PA_USE_ALARM
#endif	#endif

#ifdef WE_CAN_USE_ALARM	#ifdef PA_USE_ALARM
static sigjmp_buf timeout_env;	static sigjmp_buf timeout_env;
static void timeout_handler(int sig){	static void timeout_handler(int sig){
siglongjmp(timeout_env, 1);	siglongjmp(timeout_env, 1);
}	}
#endif	#endif

static int http_request(	static int http_request(char*& response, size_t& response_size,
String& response,
const char* host, int port,	const char* host, int port,
const char* request,	const char* request,
int timeout,	int
	#ifdef PA_USE_ALARM
	timeout
	#endif
	,
bool fail_on_status_ne_200) {	bool fail_on_status_ne_200) {
if(!host)	if(!host)
throw Exception("http.host",	throw Exception("http.host",
0,	0,
"zero hostname"); //never	"zero hostname"); //never

#ifdef WE_CAN_USE_ALARM	#ifdef PA_USE_ALARM
signal(SIGALRM, timeout_handler);	signal(SIGALRM, timeout_handler);
#endif	#endif
int sock=-1;	int sock=-1;
try {	#ifdef PA_USE_ALARM
int result;	if(sigsetjmp(timeout_env, 1)) {
#ifdef WE_CAN_USE_ALARM	// stupid gcc [2.95.4] generated bad code
if(sigsetjmp(timeout_env, 1))	// which failed to handle sigsetjmp+throw: crashed inside of pre-throw code.
throw Exception("http.timeout",	// rewritten simplier [though duplicating closesocket code]
0,	if(sock>=0)
"timeout occured while retrieving document");	closesocket(sock);
else {	throw Exception("http.timeout",
alarm(timeout);	origin_string,
	"timeout occured while retrieving document");
	return 0; // never
	} else {
	alarm(timeout);
#endif	#endif
	try {
	int result;
struct sockaddr_in dest;	struct sockaddr_in dest;

if(!set_addr(&dest, host, port))	if(!set_addr(&dest, host, port))
throw Exception("http.host",	throw Exception("http.host",
0,	0,
"can not resolve hostname \"%s\"", host);	"can not resolve hostname \"%s\"", host);
Line 285 static int http_request(	Line 260 static int http_request(
0,	0,
"error sending request: %s (%d)", strerror(errno), errno);	"error sending request: %s (%d)", strerror(errno), errno);

result=http_read_response(response, sock, fail_on_status_ne_200);	result=http_read_response(response, response_size, sock, fail_on_status_ne_200);
closesocket(sock);	closesocket(sock);
#ifdef WE_CAN_USE_ALARM	#ifdef PA_USE_ALARM
alarm(0);	alarm(0);
}
#endif	#endif
return result;	return result;
} catch(...) {	} catch(...) {
if(sock>=0)	#ifdef PA_USE_ALARM
closesocket(sock);	alarm(0);
#ifdef WE_CAN_USE_ALARM
alarm(0);
#endif	#endif
rethrow;	if(sock>=0)
	closesocket(sock);
	rethrow;
	}
	#ifdef PA_USE_ALARM
}	}
	#endif
}	}

#ifndef DOXYGEN	#ifndef DOXYGEN
struct Http_pass_header_info {	struct Http_pass_header_info {
Charset* charset;	Request_charsets* charsets;
String* request;	String* request;
bool user_agent_specified;	bool user_agent_specified;
};	};
#endif	#endif
static void http_pass_header(HashStringValue::key_type key, HashStringValue::value_type value,	static void http_pass_header(HashStringValue::key_type key,
	HashStringValue::value_type value,
Http_pass_header_info *info) {	Http_pass_header_info *info) {
*info->request <<key<<": "	*info->request <<key<<": "
<< attributed_meaning_to_string(*value, String::L_HTTP_HEADER, false)	<< attributed_meaning_to_string(*value, String::L_HTTP_HEADER, false)
<< "\n";	<< CRLF;

if(String(key, String::L_TAINTED).change_case(*info->charset, String::CC_UPPER)=="USER-AGENT")	if(String(key, String::L_TAINTED).change_case(info->charsets->source(), String::CC_UPPER)=="USER-AGENT")
info->user_agent_specified=true;	info->user_agent_specified=true;
}	}


	static Charset* detect_charset(Charset& source_charset, const String& content_type_value) {
	const String::Body CONTENT_TYPE_VALUE=
	content_type_value.change_case(source_charset, String::CC_UPPER);
	// content-type: xxx/xxx; source_charset=WE-NEED-THIS
	// content-type: xxx/xxx; source_charset="WE-NEED-THIS"
	// content-type: xxx/xxx; source_charset="WE-NEED-THIS";
	size_t before_charseteq_pos=CONTENT_TYPE_VALUE.pos("CHARSET=");
	if(before_charseteq_pos!=STRING_NOT_FOUND) {
	size_t charset_begin=before_charseteq_pos+8/CHARSET="/;
	size_t open_quote_pos=CONTENT_TYPE_VALUE.pos('"', charset_begin);
	bool quoted=open_quote_pos==charset_begin;
	if(quoted)
	charset_begin++; // skip opening '"'
	size_t charset_end=CONTENT_TYPE_VALUE.length();
	if(quoted) {
	size_t close_quote_pos=CONTENT_TYPE_VALUE.pos('"', charset_begin);
	if(close_quote_pos!=STRING_NOT_FOUND)
	charset_end=close_quote_pos;
	} else {
	size_t delim_pos=CONTENT_TYPE_VALUE.pos(';', charset_begin);
	if(delim_pos!=STRING_NOT_FOUND)
	charset_end=delim_pos;
	}
	const String::Body CHARSET_NAME_BODY=
	CONTENT_TYPE_VALUE.mid(charset_begin, charset_end);

	return &charsets.get(CHARSET_NAME_BODY);
	}

	return 0;
	}


#ifndef DOXYGEN	#ifndef DOXYGEN
struct File_read_http_result {	struct File_read_http_result {
char *str; size_t length;	char *str; size_t length;
HashStringValue* headers;	HashStringValue* headers;
};	};
#endif	#endif
static File_read_http_result file_read_http(Charset& charset, const String& file_spec,	/// @todo build .cookies field. use ^file.tables.SET-COOKIES.menu{ for now
HashStringValue *options=0) {	static File_read_http_result file_read_http(Request_charsets& charsets,
	const String& file_spec,
	bool as_text,
	HashStringValue *options=0) {
File_read_http_result result;	File_read_http_result result;
char host[MAX_STRING];	char host[MAX_STRING];
const char* uri;	const char* uri;
Line 334 static File_read_http_result file_read_h	Line 350 static File_read_http_result file_read_h
int timeout=2;	int timeout=2;
bool fail_on_status_ne_200=true;	bool fail_on_status_ne_200=true;
Value* vheaders=0;	Value* vheaders=0;
	Charset *asked_remote_charset=0;
String& connect_string=*new String;
// not in ^sql{... L_SQL ...} spirit, but closer to ^file::load one
connect_string.append(file_spec, String::L_URI); // tainted pieces -> URI pieces

const char* connect_string_cstr=connect_string.cstr(String::L_UNSPECIFIED);
const char* current=connect_string_cstr;
if(strncmp(current, "http://", 7)!=0)
throw Exception(0,
&connect_string,
"does not start with http://"); //never
current+=7;

strncpy(host, current, sizeof(host)-1); host[sizeof(host)-1]=0;
char* host_uri=lsplit(host, '/');
uri=host_uri?current+(host_uri-1-host):"/";
char* port_cstr=lsplit(host, ':');
char* error_pos=0;
port=port_cstr?strtol(port_cstr, &error_pos, 0):80;

if(options) {	if(options) {
int valid_options=0;	int valid_options=0;
Line 364 static File_read_http_result file_read_h	Line 362 static File_read_http_result file_read_h
valid_options++;	valid_options++;
timeout=vtimeout->as_int();	timeout=vtimeout->as_int();
}	}
if(vheaders=options->get(http_headers_name)) {	if((vheaders=options->get(http_headers_name))) {
valid_options++;	valid_options++;
}	}
if(Value* vany_status=options->get(http_any_status_name)) {	if(Value* vany_status=options->get(http_any_status_name)) {
valid_options++;	valid_options++;
fail_on_status_ne_200=!vany_status->as_bool();	fail_on_status_ne_200=!vany_status->as_bool();
}	}
	if(Value* vcharset_name=options->get(http_charset_name)) {
	valid_options++;
	asked_remote_charset=&::charsets.get(vcharset_name->as_string().
	change_case(charsets.source(), String::CC_UPPER));
	}

if(valid_options!=options->count())	if(valid_options!=options->count())
throw Exception("parser.runtime",	throw Exception("parser.runtime",
0,	0,
"invalid option passed");	"invalid option passed");
}	}
	if(!asked_remote_charset) // defaulting to $request:charset
	asked_remote_charset=&charsets.source();

//making request	//preparing request
String request;	String& connect_string=*new String;
if(method)	// not in ^sql{... L_SQL ...} spirit, but closer to ^file::load one
request<<method;	connect_string.append(file_spec, String::L_URI); // tainted pieces -> URI pieces
else
request<<"GET";	const char* request_cstr;
request<< " "<< uri <<" HTTP/1.0\nHost: "<< host<<"\n";	{
bool user_agent_specified=false;	// influence URLencoding of tainted pieces to String::L_URI lang
if(vheaders && !vheaders->is_string()) { // allow empty	Temp_client_charset temp(charsets, *asked_remote_charset);
if(HashStringValue *headers=vheaders->get_hash()) {
Http_pass_header_info info={&charset, &request};	const char* connect_string_cstr=connect_string.cstr(String::L_UNSPECIFIED);
headers->for_each(http_pass_header, &info);
user_agent_specified=info.user_agent_specified;	const char* current=connect_string_cstr;
} else	if(strncmp(current, "http://", 7)!=0)
throw Exception("parser.runtime",	throw Exception(0,
&connect_string,	&connect_string,
"headers param must be hash");	"does not start with http://"); //never
};	current+=7;
if(!user_agent_specified) // defaulting
request << "user-agent: " DEFAULT_USER_AGENT "\n";	strncpy(host, current, sizeof(host)-1); host[sizeof(host)-1]=0;
request<<"\n";	char* host_uri=lsplit(host, '/');
	uri=host_uri?current+(host_uri-1-host):"/";
	char* port_cstr=lsplit(host, ':');
	char* error_pos=0;
	port=port_cstr?strtol(port_cstr, &error_pos, 0):80;


	//making request
	String request;
	if(method)
	request<<method;
	else
	request<<"GET";
	request<< " "<< uri <<" HTTP/1.0" CRLF
	"host: "<< host << CRLF;
	bool user_agent_specified=false;
	if(vheaders && !vheaders->is_string()) { // allow empty
	if(HashStringValue *headers=vheaders->get_hash()) {
	Http_pass_header_info info={&charsets, &request, false};
	headers->for_each(http_pass_header, &info);
	user_agent_specified=info.user_agent_specified;
	} else
	throw Exception("parser.runtime",
	&connect_string,
	"headers param must be hash");
	};
	if(!user_agent_specified) // defaulting
	request << "user-agent: " DEFAULT_USER_AGENT CRLF;
	request << CRLF;

	request_cstr=request.cstr(String::L_UNSPECIFIED);
	}
	// recode those pieces which are not in String::L_URI lang
	// [those violating HTTP standard, but widly used]
	request_cstr=Charset::transcode(
	String::C(request_cstr, strlen(request_cstr)),
	charsets.source(),
	*asked_remote_charset);

//sending request	//sending request
String response;	char* response;
const char* request_cstr=request.cstr(String::L_UNSPECIFIED);	size_t response_size;
int status_code=http_request(response,	int status_code=http_request(response, response_size,
host, port, request_cstr,	host, port, request_cstr,
timeout, fail_on_status_ne_200);	timeout, fail_on_status_ne_200);

//processing results	//processing results
size_t pos=response.pos("\r\n\r\n", 4);	char* raw_body; size_t raw_body_size;
if(pos==STRING_NOT_FOUND \|\| pos<1){	char* headers_end_at=strstr(response, CRLF CRLF /change '4' below along!/);
	if(headers_end_at) {
	raw_body=headers_end_at+4;
	raw_body_size=response_size-(raw_body-response);
	} else
throw Exception("http.response",	throw Exception("http.response",
&connect_string,	&connect_string,
"bad response from host - no headers found");	"bad response from host - no headers found");
}
const String& header_block=response.mid(0, pos);	*headers_end_at=0;
const String& body=response.mid(pos+4, response.length());	const String header_block(response, headers_end_at-response, true);

ArrayString aheaders;	ArrayString aheaders;
result.headers=new HashStringValue;	result.headers=new HashStringValue;
	VHash* vtables=new VHash;
	result.headers->put(http_tables_name, vtables);
	HashStringValue& tables=vtables->hash();

size_t pos_after=0;	size_t pos_after=0;
header_block.split(aheaders, pos_after, "\r\n");	header_block.split(aheaders, pos_after, CRLF);

//processing headers	//processing headers
size_t aheaders_count=aheaders.count();	size_t aheaders_count=aheaders.count();
	Charset* real_remote_charset=0; // undetected, yet
for(size_t i=1; i<aheaders_count; i++) {	for(size_t i=1; i<aheaders_count; i++) {
const String& line=*aheaders.get(i);	const String& line=*aheaders.get(i);
pos=line.pos(": ", 2);	size_t pos=line.pos(": ", 2);
if(pos==STRING_NOT_FOUND \|\| pos<1)	if(pos==STRING_NOT_FOUND \|\| pos<1)
throw Exception("http.response",	throw Exception("http.response",
&connect_string,	&connect_string,
"bad response from host - bad header \"%s\"", line.cstr());	"bad response from host - bad header \"%s\"", line.cstr());
	const String::Body HEADER_NAME=
result.headers->put(	line.mid(0, pos).change_case(charsets.source(), String::CC_UPPER);
line.mid(0, pos).change_case(charset, String::CC_UPPER),	const String& header_value=line.mid(pos+2, line.length());
new VString(line.mid(pos+2, line.length())));	if(HEADER_NAME=="CONTENT-TYPE")
	real_remote_charset=detect_charset(charsets.source(), header_value);

	// tables
	{
	Value valready=(Value )tables.get(HEADER_NAME);
	bool existed=valready!=0;
	Table *table;
	if(existed) {
	// second+ appearence
	table=valready->get_table();
	} else {
	// first appearence
	Table::columns_type columns =new ArrayString(1);
	*columns+=new String("value");
	table=new Table(columns);
	}
	// this string becomes next row
	ArrayString& row=*new ArrayString(1);
	row+=&header_value;
	*table+=&row;
	// not existed before? add it
	if(!existed)
	tables.put(HEADER_NAME, new VTable(table));
	}

	result.headers->put(HEADER_NAME, new VString(header_value));
}	}
	// defaulting to used-asked charset [it's never empty!]
	if(!real_remote_charset)
	real_remote_charset=asked_remote_charset;

// output response	// output response
result.str=body.cstrm();	String::C real_body=String::C(raw_body, raw_body_size);
result.length=body.length();	if(as_text && raw_body_size) // must be checked because transcode returns CONST string in case length==0, which contradicts hacking few lines below
	real_body=Charset::transcode(real_body, *real_remote_charset, charsets.source());

	result.str=const_cast<char *>(real_body.str); // hacking a little
	result.length=real_body.length;
result.headers->put(file_status_name, new VInt(status_code));	result.headers->put(file_status_name, new VInt(status_code));
return result;	return result;
}	}
Line 452 struct File_read_action_info {	Line 536 struct File_read_action_info {
};	};
#endif	#endif
static void file_read_action(	static void file_read_action(
struct stat& finfo,	struct stat& finfo,
int f,	int f,
const String& file_spec, const char* fname, bool as_text,	const String& file_spec, const char* /fname/, bool as_text,
void *context) {	void *context) {
File_read_action_info& info=static_cast<File_read_action_info >(context);	File_read_action_info& info=static_cast<File_read_action_info >(context);
if(size_t to_read_size=(size_t)finfo.st_size) {	if(size_t to_read_size=(size_t)finfo.st_size) {
*info.data=new(PointerFreeGC) char[to_read_size+(as_text?1:0)];	*info.data=new(PointerFreeGC) char[to_read_size+(as_text?1:0)];
Line 464 static void file_read_action(	Line 548 static void file_read_action(
if(ssize_t(info.data_size)<0 \|\| info.data_size>to_read_size)	if(ssize_t(info.data_size)<0 \|\| info.data_size>to_read_size)
throw Exception(0,	throw Exception(0,
&file_spec,	&file_spec,
"read failed: actually read %lu bytes count not in [0..%lu] valid range",	"read failed: actually read %u bytes count not in [0..%u] valid range",
*info.data_size, to_read_size);	*info.data_size, to_read_size);
} else { // empty file	} else { // empty file
if(as_text) {	if(as_text) {
Line 476 static void file_read_action(	Line 560 static void file_read_action(
return;	return;
}	}
}	}
File_read_result file_read(Charset& charset, const String& file_spec,	File_read_result file_read(Request_charsets& charsets, const String& file_spec,
bool as_text, HashStringValue *params,	bool as_text, HashStringValue *params,
bool fail_on_read_problem) {	bool fail_on_read_problem) {
File_read_result result={false};	File_read_result result={false, 0, 0, 0};
#ifdef PA_HTTP	#ifdef PA_HTTP
if(file_spec.starts_with("http://")) {	if(file_spec.starts_with("http://")) {
// fail on read problem	// fail on read problem
File_read_http_result http=file_read_http(charset, file_spec, params);	File_read_http_result http=file_read_http(charsets, file_spec, as_text, params);
result.success=true;	result.success=true;
result.str=http.str;	result.str=http.str;
result.length=http.length;	result.length=http.length;
result.headers=new HashStringValue;//http.headers;	result.headers=http.headers;
} else {	} else {
#endif	#endif
	if(params && params->count())
	throw Exception("parser.runtime",
	0,
	"invalid option passed");

File_read_action_info info={&result.str, &result.length};	File_read_action_info info={&result.str, &result.length};
result.success=file_read_action_under_lock(file_spec,	result.success=file_read_action_under_lock(file_spec,
"read", file_read_action, &info,	"read", file_read_action, &info,
Line 502 File_read_result file_read(Charset& char	Line 591 File_read_result file_read(Charset& char
// UTF-8 signature: EF BB BF	// UTF-8 signature: EF BB BF
if(result.length>=3) {	if(result.length>=3) {
char in=(char )result.str;	char in=(char )result.str;
if((in[0] == '\xEF') && (in[1] == '\xBB') &&	if(strncmp(in, "\xEF\xBB\xBF", 3)==0) {
(in[2] == '\xBF')) {
result.str=in+3; result.length-=3;// skip prefix	result.str=in+3; result.length-=3;// skip prefix
}	}
}	}

fix_line_breaks((char *)(result.str), result.length);	fix_line_breaks((char *)(result.str), result.length);
// note: after fixing
((char*&)(result.str))[result.length]=0;
}	}

return result;	return result;
}	}

	#ifdef PA_SAFE_MODE
	void check_safe_mode(struct stat finfo, const String& file_spec, const char* fname) {
	if(finfo.st_uid/foreign?/!=geteuid()
	&& finfo.st_gid/foreign?/!=getegid())
	throw Exception("parser.runtime",
	&file_spec,
	"parser is in safe mode: "
	"reading files of foreign group and user disabled "
	"[recompile parser with --disable-safe-mode configure option], "
	"actual filename '%s', "
	"fuid(%d)!=euid(%d) or fgid(%d)!=egid(%d)",
	fname,
	finfo.st_uid, geteuid(),
	finfo.st_gid, getegid());
	}
	#endif

bool file_read_action_under_lock(const String& file_spec,	bool file_read_action_under_lock(const String& file_spec,
const char* action_name, File_read_action action, void *context,	const char* action_name, File_read_action action, void *context,
bool as_text,	bool as_text,
Line 534 bool file_read_action_under_lock(const S	Line 637 bool file_read_action_under_lock(const S
// later: it seems, even this does not help sometimes	// later: it seems, even this does not help sometimes
if((f=open(fname, O_RDONLY\|(as_text?_O_TEXT:_O_BINARY)))>=0) {	if((f=open(fname, O_RDONLY\|(as_text?_O_TEXT:_O_BINARY)))>=0) {
try {	try {
if(lock_shared_blocking(f)!=0)	if(pa_lock_shared_blocking(f)!=0)
throw Exception("file.lock",	throw Exception("file.lock",
&file_spec,	&file_spec,
"shared lock failed: %s (%d), actual filename '%s'",	"shared lock failed: %s (%d), actual filename '%s'",
Line 548 bool file_read_action_under_lock(const S	Line 651 bool file_read_action_under_lock(const S
strerror(errno), errno, fname);	strerror(errno), errno, fname);

#ifdef PA_SAFE_MODE	#ifdef PA_SAFE_MODE
if(finfo.st_uid/foreign?/!=geteuid()	check_safe_mode(finfo, file_spec, fname);
&& finfo.st_gid/foreign?/!=getegid())
throw Exception("parser.runtime",
&file_spec,
"parser is in safe mode: reading files of foreign group and user disabled [recompile parser with --disable-safe-mode configure option], actual filename '%s'",
fname);
#endif	#endif

action(finfo, f, file_spec, fname, as_text, context);	action(finfo, f, file_spec, fname, as_text, context);
} catch(...) {	} catch(...) {
unlock(f);close(f);	pa_unlock(f);close(f);
if(fail_on_read_problem)	if(fail_on_read_problem)
rethrow;	rethrow;
return false;	return false;
}	}

unlock(f);close(f);	pa_unlock(f);close(f);
return true;	return true;
} else {	} else {
if(fail_on_read_problem)	if(fail_on_read_problem)
Line 601 bool file_write_action_under_lock(	Line 699 bool file_write_action_under_lock(
O_CREAT\|O_RDWR	O_CREAT\|O_RDWR
\|(as_text?_O_TEXT:_O_BINARY)	\|(as_text?_O_TEXT:_O_BINARY)
\|(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) {	\|(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) {
if((do_block?lock_exclusive_blocking(f):lock_exclusive_nonblocking(f))!=0) {	if((do_block?pa_lock_exclusive_blocking(f):pa_lock_exclusive_nonblocking(f))!=0) {
Exception e("file.lock",	Exception e("file.lock",
&file_spec,	&file_spec,
"shared lock failed: %s (%d), actual filename '%s'",	"shared lock failed: %s (%d), actual filename '%s'",
Line 619 bool file_write_action_under_lock(	Line 717 bool file_write_action_under_lock(
if(!do_append)	if(!do_append)
ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower	ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower
#endif	#endif
unlock(f);close(f);	pa_unlock(f);close(f);
rethrow;	rethrow;
}	}

Line 627 bool file_write_action_under_lock(	Line 725 bool file_write_action_under_lock(
if(!do_append)	if(!do_append)
ftruncate(f, lseek(f, 0, SEEK_CUR)); // O_TRUNC truncates even exclusevely write-locked file [thanks to Igor Milyakov <virtan@rotabanner.com> for discovering]	ftruncate(f, lseek(f, 0, SEEK_CUR)); // O_TRUNC truncates even exclusevely write-locked file [thanks to Igor Milyakov <virtan@rotabanner.com> for discovering]
#endif	#endif
unlock(f);close(f);	pa_unlock(f);close(f);
return true;	return true;
} else	} else
throw Exception(errno==EACCES?"file.access":0,	throw Exception(errno==EACCES?"file.access":0,
Line 673 static void rmdir(const String& file_spe	Line 771 static void rmdir(const String& file_spe

rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::L_FILE_SPEC));	rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::L_FILE_SPEC));
}	}
bool file_delete(const String& file_spec, bool fail_on_read_problem) {	bool file_delete(const String& file_spec, bool fail_on_problem) {
const char* fname=file_spec.cstr(String::L_FILE_SPEC);	const char* fname=file_spec.cstr(String::L_FILE_SPEC);
if(unlink(fname)!=0)	if(unlink(fname)!=0)
if(fail_on_read_problem)	if(fail_on_problem)
throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,	throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
&file_spec,	&file_spec,
"unlink failed: %s (%d), actual filename '%s'",	"unlink failed: %s (%d), actual filename '%s'",
Line 967 int __vsnprintf(char* b, size_t s, const	Line 1065 int __vsnprintf(char* b, size_t s, const
// note: on win32& maybe somewhere else	// note: on win32& maybe somewhere else
// vsnprintf do not writes terminating 0 in 'buffer full' case, reducing	// vsnprintf do not writes terminating 0 in 'buffer full' case, reducing
--s;	--s;

	// clients do not check for negative 's', feature: ignore such prints
	if((ssize_t)s<0)
	return 0;

#if _MSC_VER	#if _MSC_VER
/*	/*
win32:	win32:
Line 992 int __vsnprintf(char* b, size_t s, const	Line 1095 int __vsnprintf(char* b, size_t s, const

if(r<0)	if(r<0)
r=0;	r=0;
else if(r>s)	else if((size_t)r>s)
r=s;	r=s;
#endif	#endif
b[r]=0;	b[r]=0;
Line 1006 int __snprintf(char* b, size_t s, const	Line 1109 int __snprintf(char* b, size_t s, const
va_end(l);	va_end(l);
return r;	return r;
}	}

int pa_sleep(unsigned long secs, unsigned long usecs) {
for (; usecs >= 1000000; ++secs, usecs -= 1000000);

#ifdef WIN32
Sleep(secs * 1000 + usecs / 1000);
return 0;
#else
struct timeval t;
t.tv_sec = secs;
t.tv_usec = usecs;
return (select(0, NULL, NULL, NULL, &t) == -1 ? errno : 0);
#endif
}

E-mail:

Removed from v.1.143.2.21.2.21
changed lines
	Added in v.1.175