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 1.143.2.21.2.14 and 1.143.2.21.2.25

version 1.143.2.21.2.14, 2003/03/24 17:24:42	version 1.143.2.21.2.25, 2003/04/08 16:11:54
Line 11 static const char* IDENT_COMMON_C="$Date	Line 11 static const char* IDENT_COMMON_C="$Date
#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

Line 61 int PASCAL closesocket(SOCKET);	Line 63 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 105 static int unlock(int fd) { FLOCK(F_TLOC	Line 110 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	// defines for globals

#define FILE_STATUS_NAME "status"	#define FILE_STATUS_NAME "status"
Line 119 const String file_status_name(FILE_STATU	Line 132 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"

// statics	// statics

Line 126 static const String http_method_name(HTT	Line 140 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);

// defines	// defines

Line 133 static const String http_any_status_name	Line 148 static const String http_any_status_name

// functions	// functions

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

//_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 152 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;
}	}
Line 161 void fix_line_breaks(char* buf, size_t&	Line 173 void fix_line_breaks(char* buf, size_t&
memcpy(dest, bol, eob-bol);	memcpy(dest, bol, eob-bol);
}	}

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;
}	}

Line 195 static int http_read_response(String& re	Line 207 static int http_read_response(String& re
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(buf, size, String::L_TAINTED);	response.append_strdup(buf, size, String::L_TAINTED);
if(!result && (EOLat=response.pos("\r\n", 2))!=STRING_NOT_FOUND) { // 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);
ArrayString astatus;	ArrayString astatus;
Line 221 static int http_read_response(String& re	Line 233 static int http_read_response(String& re
/* ******************** 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 char* host, int port,	const char* host, int port,
const char* request,	const char* request,
int timeout,	int timeout,
Line 242 static int http_request(	Line 253 static int http_request(
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 279 static int http_request(	Line 296 static int http_request(

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);
	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";	<< "\n";

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 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=0;
	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	#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,	static File_read_http_result file_read_http(Request_charsets& charsets,
	const String& file_spec,
HashStringValue *options=0) {	HashStringValue *options=0) {
File_read_http_result result;	File_read_http_result result;
char host[MAX_STRING];	char host[MAX_STRING];
Line 326 static File_read_http_result file_read_h	Line 382 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;	String& connect_string=*new String;
// not in ^sql{... L_SQL ...} spirit, but closer to ^file::load one	// not in ^sql{... L_SQL ...} spirit, but closer to ^file::load one
Line 363 static File_read_http_result file_read_h	Line 420 static File_read_http_result file_read_h
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	//making request
String request;	String request;
Line 376 static File_read_http_result file_read_h	Line 440 static File_read_http_result file_read_h
request<<method;	request<<method;
else	else
request<<"GET";	request<<"GET";
request<< " "<< uri <<" HTTP/1.0\nHost: "<< host<<"\n";	request<< " "<< uri <<" HTTP/1.0\n"
	"host: "<< 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(HashStringValue *headers=vheaders->get_hash()) {
Http_pass_header_info info={&charset, &request};	Http_pass_header_info info={&charsets, &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 394 static File_read_http_result file_read_h	Line 459 static File_read_http_result file_read_h

//sending request	//sending request
String response;	String response;
const char* request_cstr=request.cstr(String::L_UNSPECIFIED);	const char* request_cstr;
	{
	// influence URLencoding of tainted pieces to String::L_URI lang
	Temp_client_charset temp(charsets, *asked_remote_charset);

	request_cstr=request.cstr(String::L_UNSPECIFIED);

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

ArrayString aheaders;	ArrayString aheaders;
result.headers=new HashStringValue;	result.headers=new HashStringValue;
Line 415 static File_read_http_result file_read_h	Line 493 static File_read_http_result file_read_h
header_block.split(aheaders, pos_after, "\r\n");	header_block.split(aheaders, pos_after, "\r\n");

//processing headers	//processing headers
for(int i=1; i<aheaders.count(); i++) {	size_t aheaders_count=aheaders.count();
	Charset* real_remote_charset=0; // undetected, yet
	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);	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 StringBody 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);
	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=Charset::transcode(
result.length=body.length();	String::C(raw_body.cstrm()/must be modifiable/, raw_body.length()),
	*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 467 static void file_read_action(	Line 557 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};
#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, params);
result.success=true;	result.success=true;
result.str=http.str;	result.str=http.str;
result.length=http.length;	result.length=http.length;
Line 493 File_read_result file_read(Charset& char	Line 583 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, "\xEB\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
}	}
}	}
Line 507 File_read_result file_read(Charset& char	Line 596 File_read_result file_read(Charset& char
return result;	return result;
}	}

	#ifdef PA_SAFE_MODE
	void check_safe_mode(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 539 bool file_read_action_under_lock(const S	Line 645 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);
Line 749 bool file_stat(const String& file_spec,	Line 850 bool file_stat(const String& file_spec,
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);	const char* fname=file_spec.cstr(String::L_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",

E-mail:

Removed from v.1.143.2.21.2.14
changed lines
	Added in v.1.143.2.21.2.25