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 1.143.2.1 and 1.143.2.21.2.28

version 1.143.2.1, 2003/01/23 17:05:34	version 1.143.2.21.2.28, 2003/06/02 09:31:05
Line 1	Line 1
/** @file	/** @file
Parser: commonly functions.	Parser: commonly functions.

Copyright(c) 2001, 2003 ArtLebedev Group (http://www.artlebedev.com)	Copyright(c) 2001-2003 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)
*/	*/

Line 9 static const char* IDENT_COMMON_C="$Date	Line 9 static const char* IDENT_COMMON_C="$Date

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

#ifdef WIN32	#define PA_HTTP
# include <windows.h>
	#ifdef PA_HTTP
	#include "pa_vstring.h"
	#include "pa_vint.h"

	#ifdef CYGWIN
	#define _GNU_H_WINDOWS32_SOCKETS
	// for PASCAL
	#include <windows.h>
	// SOCKET
	typedef u_int SOCKET;
	int PASCAL closesocket(SOCKET);
#else	#else
# define closesocket close	# if defined(WIN32)
	# include <windows.h>
	# else
	# define closesocket close
	# endif
	#endif

	#else

	# if defined(WIN32)
	# include <windows.h>
	# endif

#endif	#endif

// some maybe-undefined constants	// some maybe-undefined constants
Line 38 static const char* IDENT_COMMON_C="$Date	Line 63 static const char* IDENT_COMMON_C="$Date
#endif	#endif

// locking constants	// locking constants
	//#define PA_DEBUG_NO_LOCKING

	#ifdef PA_DEBUG_NO_LOCKING

#ifdef HAVE_FLOCK	#ifdef HAVE_FLOCK

Line 82 static int unlock(int fd) { FLOCK(F_TLOC	Line 110 static int unlock(int fd) { FLOCK(F_TLOC
#endif	#endif
#endif	#endif

#define DEFAULT_USER_AGENT "parser3"	#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

void fix_line_breaks(char* buf, size_t& size) {	// defines for globals
if(size==0)
return;	#define FILE_STATUS_NAME "status"

	// globals

	const String file_status_name(FILE_STATUS_NAME);

	// defines for statics

	#define HTTP_METHOD_NAME "method"
	#define HTTP_TIMEOUT_NAME "timeout"
	#define HTTP_HEADERS_NAME "headers"
	#define HTTP_ANY_STATUS_NAME "any-status"
	#define HTTP_CHARSET_NAME "charset"

	// statics

	static const String http_method_name(HTTP_METHOD_NAME);
	static const String http_timeout_name(HTTP_TIMEOUT_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_charset_name(HTTP_CHARSET_NAME);

	// defines

	#define DEFAULT_USER_AGENT "parser3"

	// functions

	void fix_line_breaks(char *str, size_t& length) {
//_asm int 3;	//_asm int 3;
const char* const eob=buf+size;	const char* const eob=str+length;
char* dest=buf;	char* dest=str;
// fix DOS: \r\n -> \n	// fix DOS: \r\n -> \n
// fix Macintosh: \r -> \n	// fix Macintosh: \r -> \n
char* bol=buf;	char* bol=str;
while(char* eol=(char*)memchr(bol, '\r', eob -bol)) {	while(char* eol=(char*)memchr(bol, '\r', eob -bol)) {
size_t len=eol-bol;	size_t len=eol-bol;
if(dest!=bol)	if(dest!=bol)
Line 104 void fix_line_breaks(char* buf, size_t&	Line 164 void fix_line_breaks(char* buf, size_t&

if(&eol[1]<eob && eol[1]=='\n') { // \r, \n = DOS	if(&eol[1]<eob && eol[1]=='\n') { // \r, \n = DOS
bol=eol+2;	bol=eol+2;
size--;	length--;
} 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(Pool& pool,	char* file_read_text(Request_charsets& charsets,
ConstStringPtr file_spec,	const String& file_spec,
bool fail_on_read_problem,	bool fail_on_read_problem,
StringHash params, StringHash* out_fields) {	HashStringValue* params/, HashStringValue * out_fields*/) {
void *result; size_t size;	File_read_result file=
return file_read(pool, file_spec, result, size, true, params, out_fields, fail_on_read_problem)?	file_read(charsets, file_spec, true, params, fail_on_read_problem);
(char *)result:0;	return file.success?file.str:0;
}	}

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

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));
Line 139 static bool set_addr(struct sockaddr_in	Line 200 static bool set_addr(struct sockaddr_in
return true;	return true;
}	}

static int http_read_response(Pool& pool, String& response, int sock, bool fail_on_status_ne_200){	static int http_read_response(String& response, int sock, bool fail_on_status_ne_200){
const String* status_code=0;	int result=0;
ssize_t EOLat=0;	size_t EOLat=0;
while(true) {	while(true) {
char buf=(char )pool.malloc(MAX_STRING);	char *buf=new(PointerFreeGC) char[MAX_STRING];
ssize_t size=recv(sock, buf, MAX_STRING, 0);	ssize_t size=recv(sock, buf, MAX_STRING, 0);
if(size<=0)	if(size<=0)
break;	break;
response.APPEND_TAINTED(buf, size, "remote HTTP server response", 0);	response.append_strdup(buf, size, String::L_TAINTED);
if(!status_code && (EOLat=response.pos("\r\n", 2))>=0) { // checking status in first response	if(!result && (EOLat=response.pos("\r\n", 2))!=STRING_NOT_FOUND) { // checking status in first response
const String& status_line=response.mid(0, (size_t)EOLat);	const String& status_line=response.mid(0, (size_t)EOLat);
Array astatus(response.pool());	ArrayString astatus;
size_t pos_after_ref=0; status_line.split(astatus, &pos_after_ref, " ", 1);	size_t pos_after=0;
status_code=astatus.get_string(1);	status_line.split(astatus, pos_after, " ");
	const String& status_code=*astatus.get(1);
	result=status_code.as_int();

if(fail_on_status_ne_200 && *status_code!="200")	if(fail_on_status_ne_200 && result!=200)
throw Exception("http.status",	throw Exception("http.status",
status_code,	&status_code,
"invalid HTTP response status");	"invalid HTTP response status");
}	}
}	}
if(status_code)	if(result)
return status_code->as_int();	return result;
else	else
throw Exception("http.response",	throw Exception("http.response",
0,	0,
"bad response from host - no status found (size=%lu)", response.size());	"bad response from host - no status found (size=%lu)", response.length());
}	}

/* ******************** 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);
Line 182 static void timeout_handler(int sig){	Line 245 static void timeout_handler(int sig){
#endif	#endif

static int http_request(String& response,	static int http_request(String& response,
const String *origin_string,	const char* host, int port,
const char* host, int port,	const char* request,
const char* request,	int timeout,
int timeout,	bool fail_on_status_ne_200) {
bool fail_on_status_ne_200){
if(!host)	if(!host)
throw Exception("http.host",	throw Exception("http.host",
origin_string,	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]
origin_string,	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",
origin_string,	0,
"can not resolve hostname \"%s\"", host);	"can not resolve hostname \"%s\"", host);

if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/0/))<0)	if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/0/))<0)
throw Exception("http.connect",	throw Exception("http.connect",
origin_string,	0,
"can not make socket: %s (%d)", strerror(errno), errno);	"can not make socket: %s (%d)", strerror(errno), errno);
if(connect(sock, (struct sockaddr *)&dest, sizeof(dest)))	if(connect(sock, (struct sockaddr *)&dest, sizeof(dest)))
throw Exception("http.connect",	throw Exception("http.connect",
origin_string,	0,
"can not connect to host \"%s\": %s (%d)", host, strerror(errno), errno);	"can not connect to host \"%s\": %s (%d)", host, strerror(errno), errno);
size_t request_size=strlen(request);	size_t request_size=strlen(request);
if(send(sock, request, request_size, 0)!=(ssize_t)request_size)	if(send(sock, request, request_size, 0)!=(ssize_t)request_size)
throw Exception("http.connect",	throw Exception("http.connect",
origin_string,	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, 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
/re/throw;	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 {
	Request_charsets* charsets;
String* request;	String* request;
bool user_agent_specified;	bool user_agent_specified;
};	};
#endif	#endif
static void http_pass_header(const Hash::Key& key, Hash::Val value, void info)	static void http_pass_header(HashStringValue::key_type key,
{	HashStringValue::value_type value,
Http_pass_header_info& i=static_cast<Http_pass_header_info >(info);	Http_pass_header_info *info) {
Pool& pool=i.request->pool();	*info->request <<key<<": "
	<< attributed_meaning_to_string(*value, String::L_HTTP_HEADER, false)
*(i.request)<<key<<": "	<< "\n";
<< attributed_meaning_to_string(static_cast<Value >(value), String::UL_HTTP_HEADER, false)
<<"\n";

if(key.change_case(pool, String::CC_UPPER)=="USER-AGENT")	if(String(key, String::L_TAINTED).change_case(info->charsets->source(), String::CC_UPPER)=="USER-AGENT")
i.user_agent_specified=true;	info->user_agent_specified=true;
}	}
static void file_read_http(Pool& pool, const String& file_spec,
void*& data, size_t& data_size,
Hash options=0, Hash* out_fields=0) {	static Charset* detect_charset(Charset& source_charset, const String& content_type_value) {
	const StringBody 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 StringBody CHARSET_NAME_BODY=
	CONTENT_TYPE_VALUE.mid(charset_begin, charset_end);

	return &charsets.get(CHARSET_NAME_BODY);
	}

	return 0;
	}


	#ifndef DOXYGEN
	struct File_read_http_result {
	char *str; size_t length;
	HashStringValue* headers;
	};
	#endif
	static File_read_http_result file_read_http(Request_charsets& charsets,
	const String& file_spec,
	HashStringValue *options=0) {
	File_read_http_result result;
char host[MAX_STRING];	char host[MAX_STRING];
const char* uri;	const char* uri;
int port;	int port;
const char* method="GET";	const char* method=0;
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(pool);
// not in ^sql{... UL_SQL ...} spirit, but closer to ^file::load one
connect_string.append(file_spec, String::UL_URI); // tainted pieces -> URI pieces

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

strncpy(host, connect_string_cstr, sizeof(host)-1); host[sizeof(host)-1]=0;
char* host_uri=lsplit(host, '/');
uri=host_uri?connect_string_cstr+(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;
if(Value vmethod=static_cast<Value >(options->get(*http_method_name))) {	if(Value* vmethod=options->get(http_method_name)) {
valid_options++;	valid_options++;
method=vmethod->as_string().cstr();	method=vmethod->as_string().cstr();
}	}
if(Value vtimeout=static_cast<Value >(options->get(*http_timeout_name))) {	if(Value* vtimeout=options->get(http_timeout_name)) {
valid_options++;	valid_options++;
timeout=vtimeout->as_int();	timeout=vtimeout->as_int();
}	}
if(vheaders=static_cast<Value >(options->get(http_headers_name))) {	if(vheaders=options->get(http_headers_name)) {
valid_options++;	valid_options++;
}	}
if(Value vany_status=static_cast<Value >(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->size())	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(pool);	String& connect_string=*new String;
request<< method <<" "<< uri <<" HTTP/1.0\nHost: "<< host<<"\n";	// not in ^sql{... L_SQL ...} spirit, but closer to ^file::load one
bool user_agent_specified=false;	connect_string.append(file_spec, String::L_URI); // tainted pieces -> URI pieces
if(vheaders && !vheaders->is_string()) { // allow empty
if(Hash *headers=vheaders->get_hash(&connect_string)) {	const char* request_cstr;
Http_pass_header_info info={&request};	{
headers->for_each(http_pass_header, &info);	// influence URLencoding of tainted pieces to String::L_URI lang
user_agent_specified=info.user_agent_specified;	Temp_client_charset temp(charsets, *asked_remote_charset);
} else
throw Exception("parser.runtime",	const char* connect_string_cstr=connect_string.cstr(String::L_UNSPECIFIED);
&connect_string,
"headers param must be hash");	const char* current=connect_string_cstr;
};	if(strncmp(current, "http://", 7)!=0)
if(!user_agent_specified) // defaulting	throw Exception(0,
request << "user-agent: " DEFAULT_USER_AGENT "\n";	&connect_string,
request<<"\n";	"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;


	//making request
	String request;
	if(method)
	request<<method;
	else
	request<<"GET";
	request<< " "<< uri <<" HTTP/1.0\n"
	"host: "<< host <<"\n";
	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};
	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 "\n";
	request<<"\n";

	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(pool);	String response;
int status_code=http_request(response,	int status_code=http_request(response,
&connect_string, host, port, request.cstr(String::UL_UNSPECIFIED),	host, port, request_cstr,
timeout, fail_on_status_ne_200);	timeout, fail_on_status_ne_200);

//processing results	//processing results
int pos=response.pos("\r\n\r\n", 4);	size_t pos=response.pos("\r\n\r\n", 4);
if(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 - no headers found");	"bad response from host - no headers found");
}	}
String header_block=response.mid(0, pos);	const String& header_block=response.mid(0, pos);
String body=response.mid(pos+4, response.size());	const String& raw_body=response.mid(pos+4, response.length());

Array aheaders(pool);	ArrayString aheaders;
Hash& headers=*new(pool) Hash(pool);	result.headers=new HashStringValue;
size_t pos_after_ref=0;	size_t pos_after=0;
header_block.split(aheaders, &pos_after_ref, "\r\n", 2);	header_block.split(aheaders, pos_after, "\r\n");

//processing headers	//processing headers
for(int i=1;i<aheaders.size();i++) {	size_t aheaders_count=aheaders.count();
if(const String *line=aheaders.get_string(i)) {	Charset* real_remote_charset=0; // undetected, yet
pos=line->pos(": ", 2);	for(size_t i=1; i<aheaders_count; i++) {
if(pos<1)	const String& line=*aheaders.get(i);
throw Exception("http.response",	pos=line.pos(": ", 2);
&connect_string,	if(pos==STRING_NOT_FOUND \|\| pos<1)
"bad response from host - bad header \"%s\"", line->cstr());

headers.put(
line->mid(0, pos).change_case(pool, String::CC_UPPER),
new(pool) VString(line->mid(pos+2, line->size())));
} else
throw Exception("http.response",	throw Exception("http.response",
&connect_string,	&connect_string,
"bad response from host - bad headers \"%s\"", header_block.cstr());	"bad response from host - bad header \"%s\"", line.cstr());
	const StringBody HEADER_NAME=
	line.mid(0, pos).change_case(charsets.source(), String::CC_UPPER);
	const String& header_value=line.mid(pos+2, line.length());
	if(HEADER_NAME=="CONTENT-TYPE")
	real_remote_charset=detect_charset(charsets.source(), header_value);
	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
data=body.cstr(); data_size=body.size();	String::C real_body=Charset::transcode(
if(out_fields) {	String::C(raw_body.cstrm()/must be modifiable/, raw_body.length()),
headers.put(*file_status_name, new(pool) VInt(pool, status_code));	*real_remote_charset,
*out_fields=&headers;	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));
	return result;
}	}

	#endif

#ifndef DOXYGEN	#ifndef DOXYGEN
struct File_read_action_info {	struct File_read_action_info {
void *data; size_t data_size;	char *data; size_t data_size;
};	};
#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=pool.malloc(to_read_size+(as_text?1:0), 3);	*info.data=new(PointerFreeGC) char[to_read_size+(as_text?1:0)];
info.data_size=(size_t)read(f, info.data, to_read_size);	info.data_size=(size_t)read(f, info.data, to_read_size);

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)
Line 402 static void file_read_action(	Line 553 static void file_read_action(
*info.data_size, to_read_size);	*info.data_size, to_read_size);
} else { // empty file	} else { // empty file
if(as_text) {	if(as_text) {
*info.data=pool.malloc(1);	*info.data=new(PointerFreeGC) char[1];
(char)(*info.data)=0;	(char)(*info.data)=0;
} else	} else
*info.data=0;	*info.data=0;
Line 410 static void file_read_action(	Line 561 static void file_read_action(
return;	return;
}	}
}	}
bool file_read(ConstStringPtr file_spec,	File_read_result file_read(Request_charsets& charsets, const String& file_spec,
void*& data, size_t& data_size,	bool as_text, HashStringValue *params,
bool as_text, Hash params, Hash* out_fields,
bool fail_on_read_problem) {	bool fail_on_read_problem) {
bool result;	File_read_result result={false};
if(file_spec.starts_with("http://", 7)) {	#ifdef PA_HTTP
	if(file_spec.starts_with("http://")) {
// fail on read problem	// fail on read problem
file_read_http(pool, file_spec, data, data_size, params, out_fields);	File_read_http_result http=file_read_http(charsets, file_spec, params);
result=true;	result.success=true;
	result.str=http.str;
	result.length=http.length;
	result.headers=http.headers;
} else {	} else {
File_read_action_info info={&data, &data_size};	#endif
result=file_read_action_under_lock(pool, file_spec,	File_read_action_info info={&result.str, &result.length};
	result.success=file_read_action_under_lock(file_spec,
"read", file_read_action, &info,	"read", file_read_action, &info,
as_text, fail_on_read_problem);	as_text, fail_on_read_problem);
	#ifdef PA_HTTP
}	}
	#endif

if(result && as_text) {	if(result.success && as_text) {
// UTF-8 signature: EF BB BF	// UTF-8 signature: EF BB BF
if(data_size>=3) {	if(result.length>=3) {
char in=(char )data;	char in=(char )result.str;
if((in[0] == '\xEF') && (in[1] == '\xBB') &&	if(strncmp(in, "\xEB\xBB\xBF", 3)==0) {
(in[2] == '\xBF')) {	result.str=in+3; result.length-=3;// skip prefix
data=in+3; data_size-=3;// skip prefix
}	}
}	}

fix_line_breaks((char *)(data), data_size);	fix_line_breaks((char *)(result.str), result.length);
// note: after fixing	// note: after fixing
((char*&)(data))[data_size]=0;	((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,
bool fail_on_read_problem) {	bool fail_on_read_problem) {
const char* fname=file_spec.cstr(String::UL_FILE_SPEC);	const char* fname=file_spec.cstr(String::L_FILE_SPEC);
int f;	int f;

// first open, next stat:	// first open, next stat:
Line 466 bool file_read_action_under_lock(const S	Line 639 bool file_read_action_under_lock(const S
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'",
strerror(errno), errno, fname);	strerror(errno), errno, fname);

struct stat finfo;	struct stat finfo;
if(stat(fname, &finfo)!=0)	if(stat(fname, &finfo)!=0)
throw Exception("file.missing", // hardly possible: we just opened it OK	throw Exception("file.missing", // hardly possible: we just opened it OK
&file_spec,	&file_spec,
"stat failed: %s (%d), actual filename '%s'",	"stat failed: %s (%d), actual filename '%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(pool, finfo, f, file_spec, fname, as_text, context);	action(finfo, f, file_spec, fname, as_text, context);
} catch(...) {	} catch(...) {
unlock(f);close(f);	unlock(f);close(f);
if(fail_on_read_problem)	if(fail_on_read_problem)
/re/throw;	rethrow;
return false;	return false;
}	}

Line 499 bool file_read_action_under_lock(const S	Line 667 bool file_read_action_under_lock(const S
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,
"%s failed: %s (%d), actual filename '%s'",	"%s failed: %s (%d), actual filename '%s'",
action_name, strerror(errno), errno, fname);	action_name, strerror(errno), errno, fname);
return false;	return false;
}	}
}	}

static void create_dir_for_file(const String& file_spec) {	static void create_dir_for_file(const String& file_spec) {
size_t pos_after=1;	size_t pos_after=1;
int pos_before;	size_t pos_before;
while((pos_before=file_spec.pos("/", 1, pos_after))>=0) {	while((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND) {
mkdir(file_spec.mid(0, pos_before).cstr(String::UL_FILE_SPEC), 0775);	mkdir(file_spec.mid(0, pos_before).cstr(String::L_FILE_SPEC), 0775);
pos_after=pos_before+1;	pos_after=pos_before+1;
}	}
}	}
Line 520 bool file_write_action_under_lock(	Line 688 bool file_write_action_under_lock(
bool do_append,	bool do_append,
bool do_block,	bool do_block,
bool fail_on_lock_problem) {	bool fail_on_lock_problem) {
const char* fname=file_spec.cstr(String::UL_FILE_SPEC);	const char* fname=file_spec.cstr(String::L_FILE_SPEC);
int f;	int f;
if(access(fname, W_OK)!=0) // no	if(access(fname, W_OK)!=0) // no
create_dir_for_file(file_spec);	create_dir_for_file(file_spec);
Line 533 bool file_write_action_under_lock(	Line 701 bool file_write_action_under_lock(
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'",
strerror(errno), errno, fname);	strerror(errno), errno, fname);
close(f);	close(f);
if(fail_on_lock_problem)	if(fail_on_lock_problem)
throw e;	throw e;
Line 548 bool file_write_action_under_lock(	Line 716 bool file_write_action_under_lock(
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);	unlock(f);close(f);
/re/throw;	rethrow;
}	}

#ifdef HAVE_FTRUNCATE	#ifdef HAVE_FTRUNCATE
Line 561 bool file_write_action_under_lock(	Line 729 bool file_write_action_under_lock(
throw Exception(errno==EACCES?"file.access":0,	throw Exception(errno==EACCES?"file.access":0,
&file_spec,	&file_spec,
"%s failed: %s (%d), actual filename '%s'",	"%s failed: %s (%d), actual filename '%s'",
action_name, strerror(errno), errno, fname);	action_name, strerror(errno), errno, fname);
// here should be nothing, see rethrow above	// here should be nothing, see rethrow above
}	}

#ifndef DOXYGEN	#ifndef DOXYGEN
struct File_write_action_info {	struct File_write_action_info {
const void *data; size_t size;	const char* str; size_t length;
};	};
#endif	#endif
static void file_write_action(int f, void *context) {	static void file_write_action(int f, void *context) {
File_write_action_info& info=static_cast<File_write_action_info >(context);	File_write_action_info& info=static_cast<File_write_action_info >(context);
if(info.size) {	if(info.length) {
int written=write(f, info.data, info.size);	int written=write(f, info.str, info.length);
if(written<0)	if(written<0)
throw Exception(0,	throw Exception(0,
0,	0,
Line 582 static void file_write_action(int f, voi	Line 750 static void file_write_action(int f, voi
}	}
void file_write(	void file_write(
const String& file_spec,	const String& file_spec,
const void *data, size_t size,	const char* data, size_t size,
bool as_text,	bool as_text,
bool do_append) {	bool do_append) {
File_write_action_info info={data, size};	File_write_action_info info={data, size};
file_write_action_under_lock(	file_write_action_under_lock(
file_spec,	file_spec,
"write", file_write_action, &info,	"write", file_write_action, &info,
as_text,	as_text,
do_append);	do_append);
}	}

// throws nothing! [this is required in file_move & file_delete]	// throws nothing! [this is required in file_move & file_delete]
static void rmdir(const String& file_spec, size_t pos_after) {	static void rmdir(const String& file_spec, size_t pos_after) {
int pos_before;	size_t pos_before;
if((pos_before=file_spec.pos("/", 1, pos_after))>=0)	if((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND)
rmdir(file_spec, pos_before+1);	rmdir(file_spec, pos_before+1);

rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::UL_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_read_problem) {
const char* fname=file_spec.cstr(String::UL_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_read_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'",
strerror(errno), errno, fname);	strerror(errno), errno, fname);
else	else
return false;	return false;

Line 616 bool file_delete(const String& file_spec	Line 784 bool file_delete(const String& file_spec
return true;	return true;
}	}
void file_move(const String& old_spec, const String& new_spec) {	void file_move(const String& old_spec, const String& new_spec) {
const char* old_spec_cstr=old_spec.cstr(String::UL_FILE_SPEC);	const char* old_spec_cstr=old_spec.cstr(String::L_FILE_SPEC);
const char* new_spec_cstr=new_spec.cstr(String::UL_FILE_SPEC);	const char* new_spec_cstr=new_spec.cstr(String::L_FILE_SPEC);

create_dir_for_file(new_spec);	create_dir_for_file(new_spec);

Line 625 void file_move(const String& old_spec, c	Line 793 void file_move(const String& old_spec, c
throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,	throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
&old_spec,	&old_spec,
"rename failed: %s (%d), actual filename '%s' to '%s'",	"rename failed: %s (%d), actual filename '%s' to '%s'",
strerror(errno), errno, old_spec_cstr, new_spec_cstr);	strerror(errno), errno, old_spec_cstr, new_spec_cstr);

rmdir(old_spec, 1);	rmdir(old_spec, 1);
}	}
Line 640 bool entry_exists(const char* fname, str	Line 808 bool entry_exists(const char* fname, str
}	}

bool entry_exists(const String& file_spec) {	bool entry_exists(const String& file_spec) {
const char* fname=file_spec.cstr(String::UL_FILE_SPEC);	const char* fname=file_spec.cstr(String::L_FILE_SPEC);
return entry_exists(fname, 0);	return entry_exists(fname, 0);
}	}

static bool entry_readable(const String& file_spec, bool need_dir) {	static bool entry_readable(const String& file_spec, bool need_dir) {
char* fname=file_spec.cstr(String::UL_FILE_SPEC);	char* fname=file_spec.cstrm(String::L_FILE_SPEC);
if(need_dir) {	if(need_dir) {
size_t size=strlen(fname);	size_t size=strlen(fname);
while(size) {	while(size) {
Line 669 bool file_readable(const String& file_sp	Line 837 bool file_readable(const String& file_sp
bool dir_readable(const String& file_spec) {	bool dir_readable(const String& file_spec) {
return entry_readable(file_spec, true);	return entry_readable(file_spec, true);
}	}
String *file_readable(const String& path, const String& name) {	const String* file_readable(const String& path, const String& name) {
String *result=new(path.pool()) String(path);	String& result=*new String(path);
*result << "/";	result << "/";
*result << name;	result << name;
return file_readable(*result)?result:0;	return file_readable(result)?&result:0;
}	}
bool file_executable(const String& file_spec) {	bool file_executable(const String& file_spec) {
return access(file_spec.cstr(String::UL_FILE_SPEC), X_OK)==0;	return access(file_spec.cstr(String::L_FILE_SPEC), X_OK)==0;
}	}

bool file_stat(const String& file_spec,	bool file_stat(const String& file_spec,
Line 685 bool file_stat(const String& file_spec,	Line 853 bool file_stat(const String& file_spec,
time_t& rmtime,	time_t& rmtime,
time_t& rctime,	time_t& rctime,
bool fail_on_read_problem) {	bool fail_on_read_problem) {
Pool& pool=file_spec.pool();	const char* fname=file_spec.cstr(String::L_FILE_SPEC);
const char* fname=file_spec.cstr(String::UL_FILE_SPEC);	struct stat finfo;
struct stat finfo;
if(stat(fname, &finfo)!=0)	if(stat(fname, &finfo)!=0)
if(fail_on_read_problem)	if(fail_on_read_problem)
throw Exception("file.missing",	throw Exception("file.missing",
&file_spec,	&file_spec,
"getting file size failed: %s (%d), real filename '%s'",	"getting file size failed: %s (%d), real filename '%s'",
strerror(errno), errno, fname);	strerror(errno), errno, fname);
else	else
return false;	return false;
rsize=finfo.st_size;	rsize=finfo.st_size;
Line 745 char* rsplit(char* string, char delim) {	Line 912 char* rsplit(char* string, char delim) {
}	}

/// @todo less stupid type detection	/// @todo less stupid type detection
char* format(double value, char* fmt) {	const char* format(double value, char* fmt) {
char local_buf[MAX_NUMBER];	char local_buf[MAX_NUMBER];
size_t size;	size_t size;

Line 760 char* format(double value, char* fmt) {	Line 927 char* format(double value, char* fmt) {
else	else
size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value);	size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value);

char* pool_buf=(char *)pool.malloc(size+1, 4);	return pa_strdup(local_buf, size);
memcpy(pool_buf, local_buf, size+1);
return pool_buf;
}	}

size_t stdout_write(const void *buf, size_t size) {	size_t stdout_write(const void *buf, size_t size) {
#ifdef WIN32	#ifdef WIN32
do{	do{
int chunk_written=fwrite(buf, 1, min(8*0x400, size), stdout);	int chunk_written=fwrite(buf, 1, min((size_t)8*0x400, size), stdout);
if(chunk_written<=0)	if(chunk_written<=0)
break;	break;
size-=chunk_written;	size-=chunk_written;
Line 782 size_t stdout_write(const void *buf, siz	Line 947 size_t stdout_write(const void *buf, siz
}	}

char* unescape_chars(const char* cp, int len) {	char* unescape_chars(const char* cp, int len) {
char* s=(char *)pool.malloc(len + 1, 5);	char* s=new(PointerFreeGC) char[len + 1];
enum EscapeState {	enum EscapeState {
EscapeRest,	EscapeRest,
EscapeFirst,	EscapeFirst,
Line 903 int __vsnprintf(char* b, size_t s, const	Line 1068 int __vsnprintf(char* b, size_t s, const
win32:	win32:
mk:@MSITStore:C:\Program%20Files\Microsoft%20Visual%20Studio\MSDN\2001APR\1033\vccore.chm::/html/_crt__vsnprintf.2c_._vsnwprintf.htm	mk:@MSITStore:C:\Program%20Files\Microsoft%20Visual%20Studio\MSDN\2001APR\1033\vccore.chm::/html/_crt__vsnprintf.2c_._vsnwprintf.htm

if the number of bytes to write exceeds buffer, then count bytes are written and �1 is returned	if the number of bytes to write exceeds buffer, then count bytes are written and �1 is returned
*/	*/
r=_vsnprintf(b, s, f, l);	r=_vsnprintf(b, s, f, l);
if(r<0)	if(r<0)

E-mail:

Removed from v.1.143.2.1
changed lines
	Added in v.1.143.2.21.2.28