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 1.153.2.2

version 1.143.2.21.2.21, 2003/04/02 10:35:07	version 1.153.2.2, 2003/09/02 07:27:03
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_hash.h"
#include "pa_globals.h"	#include "pa_globals.h"
	#include "pa_hash.h"
#define PA_HTTP	#include "pa_table.h"

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

	#ifdef WIN32
	# include <windows.h>
#else	#else
	# define closesocket close
# if defined(WIN32)
# include <windows.h>
# endif

#endif	#endif

// some maybe-undefined constants	// some maybe-undefined constants
Line 61 int PASCAL closesocket(SOCKET);	Line 43 int PASCAL closesocket(SOCKET);
#endif	#endif

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

#ifdef PA_DEBUG_NO_LOCKING

#ifdef HAVE_FLOCK	#ifdef HAVE_FLOCK

Line 108 static int unlock(int fd) { FLOCK(F_TLOC	Line 87 static int unlock(int fd) { FLOCK(F_TLOC
#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

#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"

// 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);

// defines

#define DEFAULT_USER_AGENT "parser3"	#define DEFAULT_USER_AGENT "parser3"

// functions

void fix_line_breaks(char *str, size_t& length) {	void fix_line_breaks(char* buf, size_t& size) {
	if(size==0)
	return;

//_asm int 3;	//_asm int 3;
const char* const eob=str+length;	const char* const eob=buf+size;
char* dest=str;	char* dest=buf;
// fix DOS: \r\n -> \n	// fix DOS: \r\n -> \n
// fix Macintosh: \r -> \n	// fix Macintosh: \r -> \n
char* bol=str;	char* bol=buf;
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 160 void fix_line_breaks(char *str, size_t&	Line 109 void fix_line_breaks(char *str, 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;
length--;	size--;
} else // \r, not \n = Macintosh	} else // \r, not \n = Macintosh
bol=eol+1;	bol=eol+1;
}	}
Line 169 void fix_line_breaks(char *str, size_t&	Line 118 void fix_line_breaks(char *str, size_t&
memcpy(dest, bol, eob-bol);	memcpy(dest, bol, eob-bol);
}	}

char* file_read_text(Charset& charset,	char* file_read_text(Pool& pool, const String& file_spec,
const String& file_spec,
bool fail_on_read_problem,	bool fail_on_read_problem,
HashStringValue* params/, HashStringValue * out_fields*/) {	Hash params, Hash* out_fields) {
File_read_result file=	void *result; size_t size;
file_read(charset, file_spec, true, params, fail_on_read_problem);	return file_read(pool, file_spec, result, size, true, params, out_fields, 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 196 static bool set_addr(struct sockaddr_in	Line 143 static bool set_addr(struct sockaddr_in
}	}

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

if(fail_on_status_ne_200 && result!=200)	if(fail_on_status_ne_200 && *status_code!="200")
throw Exception("http.status",	throw Exception("http.status",
&status_code,	status_code,
"invalid HTTP response status");	"invalid HTTP response status");
}	}
}	}
if(result)	if(status_code)
return result;	return status_code->as_int();
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=%lu)", 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(String& response,
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",
0,	origin_string,
"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,	origin_string,
"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",
0,	origin_string,
"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",
0,	origin_string,
"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",
0,	origin_string,
"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
rethrow;	if(sock>=0)
	closesocket(sock);
	/re/throw;
	}
	#ifdef PA_USE_ALARM
}	}
	#endif
}	}

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

#ifndef DOXYGEN	#ifndef DOXYGEN
struct Http_pass_header_info {	struct Http_pass_header_info {
Charset* charset;
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(const Hash::Key& key, Hash::Val value, void info)
Http_pass_header_info *info) {	{
*info->request <<key<<": "	Http_pass_header_info& i=static_cast<Http_pass_header_info >(info);
<< attributed_meaning_to_string(*value, String::L_HTTP_HEADER, false)	Pool& pool=i.request->pool();
<< "\n";
	*(i.request)<<key<<": "
	<< attributed_meaning_to_string(static_cast<Value >(value), String::UL_HTTP_HEADER, false)
	<< CRLF;

if(String(key, String::L_TAINTED).change_case(*info->charset, String::CC_UPPER)=="USER-AGENT")	if(key.change_case(pool, String::CC_UPPER)=="USER-AGENT")
info->user_agent_specified=true;	i.user_agent_specified=true;
}	}
#ifndef DOXYGEN	/// @todo build .cookies field. use ^file.tables.SET-COOKIES.menu{ for now
struct File_read_http_result {	static void file_read_http(Pool& pool, const String& file_spec,
char *str; size_t length;	void*& data, size_t& data_size,
HashStringValue* headers;	Hash options=0, Hash* out_fields=0) {
};
#endif
static File_read_http_result file_read_http(Charset& charset, 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=0;	const char* method="GET";
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;

	String& connect_string=*new(pool) String(pool); // must not be local [exception may be reported outside]
	// not in ^sql{... UL_SQL ...} spirit, but closer to ^file::load one
	connect_string.append(file_spec, String::UL_URI); // tainted pieces -> URI pieces

String& connect_string=*new String;	char* connect_string_cstr=connect_string.cstr(String::UL_UNSPECIFIED);
// not in ^sql{... L_SQL ...} spirit, but closer to ^file::load one	if(strncmp(connect_string_cstr, "http://", 7)!=0)
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,	throw Exception(0,
&connect_string,	&connect_string,
"does not start with http://"); //never	"does not start with http://"); //never
current+=7;	connect_string_cstr+=7;

strncpy(host, current, sizeof(host)-1); host[sizeof(host)-1]=0;	strncpy(host, connect_string_cstr, sizeof(host)-1); host[sizeof(host)-1]=0;
char* host_uri=lsplit(host, '/');	char* host_uri=lsplit(host, '/');
uri=host_uri?current+(host_uri-1-host):"/";	uri=host_uri?connect_string_cstr+(host_uri-1-host):"/";
char* port_cstr=lsplit(host, ':');	char* port_cstr=lsplit(host, ':');
char* error_pos=0;	char* error_pos=0;
port=port_cstr?strtol(port_cstr, &error_pos, 0):80;	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=options->get(http_method_name)) {	if(Value vmethod=static_cast<Value >(options->get(*http_method_name))) {
valid_options++;	valid_options++;
method=vmethod->as_string().cstr();	method=vmethod->as_string().cstr();
}	}
if(Value* vtimeout=options->get(http_timeout_name)) {	if(Value vtimeout=static_cast<Value >(options->get(*http_timeout_name))) {
valid_options++;	valid_options++;
timeout=vtimeout->as_int();	timeout=vtimeout->as_int();
}	}
if(vheaders=options->get(http_headers_name)) {	if(vheaders=static_cast<Value >(options->get(http_headers_name))) {
valid_options++;	valid_options++;
}	}
if(Value* vany_status=options->get(http_any_status_name)) {	if(Value vany_status=static_cast<Value >(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(valid_options!=options->count())	if(valid_options!=options->size())
throw Exception("parser.runtime",	throw Exception("parser.runtime",
0,	0,
"invalid option passed");	"invalid option passed");
}	}

//making request	//making request
String request;	String request(pool);
if(method)	request<< method <<" "<< uri <<" HTTP/1.0" CRLF "Host: "<< host<< CRLF;
request<<method;
else
request<<"GET";
request<< " "<< uri <<" HTTP/1.0\nHost: "<< host<<"\n";
bool user_agent_specified=false;	bool user_agent_specified=false;
if(vheaders && !vheaders->is_string()) { // allow empty	if(vheaders && !vheaders->is_string()) { // allow empty
if(HashStringValue *headers=vheaders->get_hash()) {	if(Hash *headers=vheaders->get_hash(&connect_string)) {
Http_pass_header_info info={&charset, &request};	Http_pass_header_info info={&request};
headers->for_each(http_pass_header, &info);	headers->for_each(http_pass_header, &info);
user_agent_specified=info.user_agent_specified;	user_agent_specified=info.user_agent_specified;
} else	} else
Line 397 static File_read_http_result file_read_h	Line 345 static File_read_http_result file_read_h
"headers param must be hash");	"headers param must be hash");
};	};
if(!user_agent_specified) // defaulting	if(!user_agent_specified) // defaulting
request << "user-agent: " DEFAULT_USER_AGENT "\n";	request << "user-agent: " DEFAULT_USER_AGENT CRLF;
request<<"\n";	request<<CRLF;

//sending request	//sending request
String response;	String response(pool);
const char* request_cstr=request.cstr(String::L_UNSPECIFIED);
int status_code=http_request(response,	int status_code=http_request(response,
host, port, request_cstr,	&connect_string, host, port, request.cstr(String::UL_UNSPECIFIED),
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);	int pos=response.pos(CRLF CRLF, 4);
if(pos==STRING_NOT_FOUND \|\| pos<1){	if(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");
}	}
const String& header_block=response.mid(0, pos);	String header_block=response.mid(0, pos);
const String& body=response.mid(pos+4, response.length());	String body=response.mid(pos+4, response.size());

ArrayString aheaders;	Array aheaders(pool);
result.headers=new HashStringValue;	Hash& headers=*new(pool) Hash(pool);
size_t pos_after=0;	VHash* vtables=new(pool) VHash(pool);
header_block.split(aheaders, pos_after, "\r\n");	headers.put(*http_tables_name, vtables);
	Hash& tables=vtables->hash(0);

	size_t pos_after_ref=0;
	header_block.split(aheaders, &pos_after_ref, CRLF, 2);

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

	const String& sname=line->mid(0, pos).change_case(pool, String::CC_UPPER);
	const String& string=line->mid(pos+2, line->size());

	// tables
	{
	Value valready=(Value )tables.get(sname);
	bool existed=valready!=0;
	Table *table;
	if(existed) {
	// second+ appearence
	table=valready->get_table();
	} else {
	// first appearence
	Array& columns=*new(pool) Array(pool, 1);
	columns+=new(pool) String(pool, "value");
	table=new(pool) Table(pool, 0, &columns);
	}
	// this string becomes next row
	Array& row=*new(pool) Array(pool, 1);
	row+=&string;
	*table+=&row;
	// not existed before? add it
	if(!existed)
	tables.put(sname, new(pool) VTable(pool, table));
	}
	headers.put(sname, new(pool) VString(string));
	} else
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 headers \"%s\"", header_block.cstr());

result.headers->put(
line.mid(0, pos).change_case(charset, String::CC_UPPER),
new VString(line.mid(pos+2, line.length())));
}	}

// output response	// output response
result.str=body.cstrm();	data=body.cstr(); data_size=body.size();
result.length=body.length();	if(out_fields) {
result.headers->put(file_status_name, new VInt(status_code));	headers.put(*file_status_name, new(pool) VInt(pool, status_code));
return result;	*out_fields=&headers;
	}
}	}

#endif

#ifndef DOXYGEN	#ifndef DOXYGEN
struct File_read_action_info {	struct File_read_action_info {
char *data; size_t data_size;	void *data; size_t data_size;
};	};
#endif	#endif
static void file_read_action(	static void file_read_action(Pool& pool,
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=pool.malloc(to_read_size+(as_text?1:0), 3);
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 468 static void file_read_action(	Line 444 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=new(PointerFreeGC) char[1];	*info.data=pool.malloc(1);
(char)(*info.data)=0;	(char)(*info.data)=0;
} else	} else
*info.data=0;	*info.data=0;
Line 476 static void file_read_action(	Line 452 static void file_read_action(
return;	return;
}	}
}	}
File_read_result file_read(Charset& charset, const String& file_spec,	bool file_read(Pool& pool, const String& file_spec,
bool as_text, HashStringValue *params,	void*& data, size_t& data_size,
	bool as_text, Hash params, Hash* out_fields,
bool fail_on_read_problem) {	bool fail_on_read_problem) {
File_read_result result={false};	bool result;
#ifdef PA_HTTP	if(file_spec.starts_with("http://", 7)) {
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(pool, file_spec, data, data_size, params, out_fields);
result.success=true;	result=true;
result.str=http.str;
result.length=http.length;
result.headers=new HashStringValue;//http.headers;
} else {	} else {
#endif	File_read_action_info info={&data, &data_size};
File_read_action_info info={&result.str, &result.length};	result=file_read_action_under_lock(pool, file_spec,
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.success && as_text) {	if(result && as_text) {
// UTF-8 signature: EF BB BF	// UTF-8 signature: EF BB BF
if(result.length>=3) {	if(data_size>=3) {
char in=(char )result.str;	char in=(char )data;
if((in[0] == '\xEF') && (in[1] == '\xBB') &&	if((in[0] == '\xEF') && (in[1] == '\xBB') &&
(in[2] == '\xBF')) {	(in[2] == '\xBF')) {
result.str=in+3; result.length-=3;// skip prefix	data=in+3; data_size-=3;// skip prefix
}	}
}	}

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

return result;	return result;
}	}

bool file_read_action_under_lock(const String& file_spec,	#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(Pool& pool, 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::L_FILE_SPEC);	const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
int f;	int f;

// first open, next stat:	// first open, next stat:
Line 538 bool file_read_action_under_lock(const S	Line 525 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(finfo, f, file_spec, fname, as_text, context);	action(pool, 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)
rethrow;	/re/throw;
return false;	return false;
}	}

Line 571 bool file_read_action_under_lock(const S	Line 553 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;
size_t pos_before;	int pos_before;
while((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND) {	while((pos_before=file_spec.pos("/", 1, pos_after))>=0) {
mkdir(file_spec.mid(0, pos_before).cstr(String::L_FILE_SPEC), 0775);	mkdir(file_spec.mid(0, pos_before).cstr(String::UL_FILE_SPEC), 0775);
pos_after=pos_before+1;	pos_after=pos_before+1;
}	}
}	}
Line 592 bool file_write_action_under_lock(	Line 574 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::L_FILE_SPEC);	const char* fname=file_spec.cstr(String::UL_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 605 bool file_write_action_under_lock(	Line 587 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 620 bool file_write_action_under_lock(	Line 602 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);
rethrow;	/re/throw;
}	}

#ifdef HAVE_FTRUNCATE	#ifdef HAVE_FTRUNCATE
Line 633 bool file_write_action_under_lock(	Line 615 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 char* str; size_t length;	const void *data; size_t size;
};	};
#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.length) {	if(info.size) {
int written=write(f, info.str, info.length);	int written=write(f, info.data, info.size);
if(written<0)	if(written<0)
throw Exception(0,	throw Exception(0,
0,	0,
Line 654 static void file_write_action(int f, voi	Line 636 static void file_write_action(int f, voi
}	}
void file_write(	void file_write(
const String& file_spec,	const String& file_spec,
const char* data, size_t size,	const void *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) {
size_t pos_before;	int pos_before;
if((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND)	if((pos_before=file_spec.pos("/", 1, pos_after))>=0)
rmdir(file_spec, pos_before+1);	rmdir(file_spec, pos_before+1);

rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::L_FILE_SPEC));	rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::UL_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::L_FILE_SPEC);	const char* fname=file_spec.cstr(String::UL_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 688 bool file_delete(const String& file_spec	Line 670 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::L_FILE_SPEC);	const char* old_spec_cstr=old_spec.cstr(String::UL_FILE_SPEC);
const char* new_spec_cstr=new_spec.cstr(String::L_FILE_SPEC);	const char* new_spec_cstr=new_spec.cstr(String::UL_FILE_SPEC);

create_dir_for_file(new_spec);	create_dir_for_file(new_spec);

Line 697 void file_move(const String& old_spec, c	Line 679 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 712 bool entry_exists(const char* fname, str	Line 694 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::L_FILE_SPEC);	const char* fname=file_spec.cstr(String::UL_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.cstrm(String::L_FILE_SPEC);	char* fname=file_spec.cstr(String::UL_FILE_SPEC);
if(need_dir) {	if(need_dir) {
size_t size=strlen(fname);	size_t size=strlen(fname);
while(size) {	while(size) {
Line 741 bool file_readable(const String& file_sp	Line 723 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);
}	}
const String* file_readable(const String& path, const String& name) {	String *file_readable(const String& path, const String& name) {
String& result=*new String(path);	String *result=new(path.pool()) 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::L_FILE_SPEC), X_OK)==0;	return access(file_spec.cstr(String::UL_FILE_SPEC), X_OK)==0;
}	}

bool file_stat(const String& file_spec,	bool file_stat(const String& file_spec,
Line 757 bool file_stat(const String& file_spec,	Line 739 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) {
const char* fname=file_spec.cstr(String::L_FILE_SPEC);	Pool& pool=file_spec.pool();
struct stat finfo;	const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
	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 816 char* rsplit(char* string, char delim) {	Line 799 char* rsplit(char* string, char delim) {
}	}

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

Line 831 const char* format(double value, char* f	Line 814 const char* format(double value, char* f
else	else
size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value);	size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value);

return pa_strdup(local_buf, size);	char* pool_buf=(char *)pool.malloc(size+1, 4);
	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((size_t)8*0x400, size), stdout);	int chunk_written=fwrite(buf, 1, min(8*0x400, size), stdout);
if(chunk_written<=0)	if(chunk_written<=0)
break;	break;
size-=chunk_written;	size-=chunk_written;
Line 846 size_t stdout_write(const void *buf, siz	Line 831 size_t stdout_write(const void *buf, siz

return size;	return size;
#else	#else
return fwrite(buf, 1, size, stdout);	return write(1/handleof(stdout)/, buf, size);
#endif	#endif
}	}

char* unescape_chars(const char* cp, int len) {	char* unescape_chars(Pool& pool, const char* cp, int len) {
char* s=new(PointerFreeGC) char[len + 1];	char* s=(char *)pool.malloc(len + 1, 5);
enum EscapeState {	enum EscapeState {
EscapeRest,	EscapeRest,
EscapeFirst,	EscapeFirst,
Line 972 int __vsnprintf(char* b, size_t s, const	Line 957 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)
Line 1022 int pa_sleep(unsigned long secs, unsigne	Line 1007 int pa_sleep(unsigned long secs, unsigne
}	}


	// attributed meaning

	/// http://www.w3.org/Protocols/rfc2616/rfc2616-sec3.html#sec3.3
	static size_t date_attribute(const VDate& vdate, char *buf, size_t buf_size) {
	const char month_names[12][4]={
	"Jan","Feb","Mar","Apr","May","Jun","Jul","Aug","Sep","Oct","Nov","Dec"};
	const char days[7][4]={
	"Sun","Mon","Tue","Wed","Thu","Fri","Sat"};

	time_t when=vdate.get_time();
	struct tm *tms=gmtime(&when);
	if(!tms)
	throw Exception(0,
	0,
	"bad time in attribute value (seconds from epoch=%ld)", when);
	return snprintf(buf, MAX_STRING, "%s, %.2d %s %.4d %.2d:%.2d:%.2d GMT",
	days[tms->tm_wday],
	tms->tm_mday,month_names[tms->tm_mon],tms->tm_year+1900,
	tms->tm_hour,tms->tm_min,tms->tm_sec);
	}
	static void append_attribute_meaning(String& result,
	Value& value, String::Untaint_lang lang, bool forced) {
	if(const String *string=value.get_string())
	result.append(string->join_chains(result.pool(), 0), lang, forced);
	else
	if(Value *vdate=value.as(VDATE_TYPE, false)) {
	char buf=(char )result.malloc(MAX_STRING);
	size_t size=date_attribute(static_cast<VDate >(vdate),
	buf, MAX_STRING);

	result.APPEND_CLEAN(buf, size, "converted from date", 0);
	} else
	throw Exception("parser.runtime",
	&result,
	"trying to append here neither string nor date (%s)",
	value.type());
	}
	#ifndef DOXYGEN
	struct Attributed_meaning_info {
	String *header; // header line being constructed
	String::Untaint_lang lang; // language in which to append to that line
	bool forced; // do they force that lang?
	};
	#endif
	static void append_attribute_subattribute(const Hash::Key& akey, Hash::Val *avalue,
	void *info) {
	if(akey==VALUE_NAME)
	return;

	Attributed_meaning_info& ami=static_cast<Attributed_meaning_info >(info);

	// ...; charset=windows1251
	*ami.header << "; ";
	ami.header->append(akey, ami.lang, ami.forced);
	*ami.header << "=";
	append_attribute_meaning(ami.header, static_cast<Value *>(avalue), ami.lang, ami.forced);
	}
	const String& attributed_meaning_to_string(Value& meaning,
	String::Untaint_lang lang, bool forced) {
	String &result=*new(meaning.pool()) String(meaning.pool());
	if(Hash *hash=meaning.get_hash(0)) {
	// $value(value) $subattribute(subattribute value)
	if(Value value=static_cast<Value >(hash->get(*value_name)))
	append_attribute_meaning(result, *value, lang, forced);

	Attributed_meaning_info attributed_meaning_info={&result, lang};
	hash->for_each(append_attribute_subattribute, &attributed_meaning_info);
	} else // result value
	append_attribute_meaning(result, meaning, lang, forced);

	return result;
	}

E-mail:

Removed from v.1.143.2.21.2.21
changed lines
	Added in v.1.153.2.2