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 1.129 and 1.143.2.1

version 1.129, 2002/11/25 15:37:12	version 1.143.2.1, 2003/01/23 17:05:34
Line 1	Line 1
/** @file	/** @file
Parser: commonly functions.	Parser: commonly functions.

Copyright(c) 2001, 2002 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 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_globals.h"	#include "pa_globals.h"
#include "pa_hash.h"	#include "pa_hash.h"
#include "pa_vstring.h"

#ifdef WIN32	#ifdef WIN32
# include <windows.h>	# include <windows.h>
Line 28 static const char* IDENT_COMMON_C="$Date	Line 27 static const char* IDENT_COMMON_C="$Date
# define _O_BINARY 0	# define _O_BINARY 0
#endif	#endif

	#ifdef HAVE_FTRUNCATE
	# define PA_O_TRUNC 0
	#else
	# ifdef _O_TRUNC
	# define PA_O_TRUNC _O_TRUNC
	# else
	# error you must have either ftruncate function or _O_TRUNC bit declared
	# endif
	#endif

// locking constants	// locking constants

#ifdef HAVE_FLOCK	#ifdef HAVE_FLOCK
Line 76 static int unlock(int fd) { FLOCK(F_TLOC	Line 85 static int unlock(int fd) { FLOCK(F_TLOC
#define DEFAULT_USER_AGENT "parser3"	#define DEFAULT_USER_AGENT "parser3"



static char* strnchr(char* buf, size_t size, char c) {
// sanity check
if(!buf)
return 0;

for(; size-->0; buf++) {
if(*buf==c)
return buf;
}

return 0;
}

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

//_asm int 3;	//_asm int 3;
const char* const eob=buf+size;	const char* const eob=buf+size;
char* dest=buf;	char* dest=buf;
// fix DOS: \r\n -> \n	// fix DOS: \r\n -> \n
// fix Macintosh: \r -> \n	// fix Macintosh: \r -> \n
char* bol=buf;	char* bol=buf;
while(char* eol=strnchr(bol, eob -bol, '\r')) {	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)
memcpy(dest, bol, len);	memcpy(dest, bol, len);
Line 115 void fix_line_breaks(char* buf, size_t&	Line 113 void fix_line_breaks(char* buf, size_t&
memcpy(dest, bol, eob-bol);	memcpy(dest, bol, eob-bol);
}	}

char* file_read_text(Pool& pool, const String& file_spec,	char* file_read_text(Pool& pool,
	ConstStringPtr file_spec,
bool fail_on_read_problem,	bool fail_on_read_problem,
Hash params, Hash* out_fields) {	StringHash params, StringHash* out_fields) {
void *result; size_t size;	void *result; size_t size;
return file_read(pool, file_spec, result, size, true, params, out_fields, fail_on_read_problem)?(char *)result:0;	return file_read(pool, file_spec, result, size, true, params, out_fields, fail_on_read_problem)?
	(char *)result:0;
}	}

//http request stuff	//http request stuff
Line 139 static bool set_addr(struct sockaddr_in	Line 139 static bool set_addr(struct sockaddr_in
return true;	return true;
}	}

static void http_read_response(String& response, int sock){	static int http_read_response(Pool& pool, String& response, int sock, bool fail_on_status_ne_200){
	const String* status_code=0;
	ssize_t EOLat=0;
while(true) {	while(true) {
char buf=(char )response.pool().malloc(MAX_STRING);	char buf=(char )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_TAINTED(buf, size, "http response", 0);	response.APPEND_TAINTED(buf, size, "remote HTTP server response", 0);
}	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);
	Array astatus(response.pool());
	size_t pos_after_ref=0; status_line.split(astatus, &pos_after_ref, " ", 1);
	status_code=astatus.get_string(1);

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

/* ******************** request ************************* */	/* ******************** request ************************* */
Line 162 static void timeout_handler(int sig){	Line 181 static void timeout_handler(int sig){
}	}
#endif	#endif

static void http_request(String& response,	static int http_request(String& response,
const String *origin_string,	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){
if(!host)	if(!host)
throw Exception("http.host",	throw Exception("http.host",
origin_string,	origin_string,
Line 177 static void http_request(String& respons	Line 197 static void http_request(String& respons
#endif	#endif
int sock=-1;	int sock=-1;
try {	try {
	int result;
#ifdef WE_CAN_USE_ALARM	#ifdef WE_CAN_USE_ALARM
if(sigsetjmp(timeout_env, 1))	if(sigsetjmp(timeout_env, 1))
throw Exception("http.timeout",	throw Exception("http.timeout",
Line 206 static void http_request(String& respons	Line 227 static void http_request(String& respons
origin_string,	origin_string,
"error sending request: %s (%d)", strerror(errno), errno);	"error sending request: %s (%d)", strerror(errno), errno);

http_read_response(response, sock);	result=http_read_response(response, sock, fail_on_status_ne_200);
if(sock>=0)	closesocket(sock);
closesocket(sock);
#ifdef WE_CAN_USE_ALARM	#ifdef WE_CAN_USE_ALARM
	alarm(0);
}	}
#endif	#endif
	return result;
} catch(...) {	} catch(...) {
if(sock>=0)	if(sock>=0)
closesocket(sock);	closesocket(sock);
Line 233 static void http_pass_header(const Hash:	Line 255 static void http_pass_header(const Hash:
Http_pass_header_info& i=static_cast<Http_pass_header_info >(info);	Http_pass_header_info& i=static_cast<Http_pass_header_info >(info);
Pool& pool=i.request->pool();	Pool& pool=i.request->pool();

(i.request)<<key<<": "<<(static_cast<Value *>(value))->get_string()<<"\n";	*(i.request)<<key<<": "
	<< 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(key.change_case(pool, String::CC_UPPER)=="USER-AGENT")
i.user_agent_specified=true;	i.user_agent_specified=true;
}	}
Line 246 static void file_read_http(Pool& pool, c	Line 270 static void file_read_http(Pool& pool, c
int port;	int port;
const char* method="GET";	const char* method="GET";
int timeout=2;	int timeout=2;
	bool fail_on_status_ne_200=true;
Value *vheaders=0;	Value *vheaders=0;

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

strncpy(host, connect_string, 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?connect_string+(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;
Line 275 static void file_read_http(Pool& pool, c	Line 304 static void file_read_http(Pool& pool, c
if(vheaders=static_cast<Value >(options->get(http_headers_name))) {	if(vheaders=static_cast<Value >(options->get(http_headers_name))) {
valid_options++;	valid_options++;
}	}
	if(Value vany_status=static_cast<Value >(options->get(*http_any_status_name))) {
	valid_options++;
	fail_on_status_ne_200=!vany_status->as_bool();
	}

if(valid_options!=options->size())	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(pool);	String request(pool);
request<< method <<" "<< uri <<" HTTP/1.0\nHost: "<< host<<"\n";	request<< method <<" "<< 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(Hash *headers=vheaders->get_hash(&file_spec)) {	if(Hash *headers=vheaders->get_hash(&connect_string)) {
Http_pass_header_info info={&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
throw Exception("parser.runtime",	throw Exception("parser.runtime",
&file_spec,	&connect_string,
"headers param must be hash");	"headers param must be hash");
};	};
if(!user_agent_specified) // defaulting	if(!user_agent_specified) // defaulting
Line 301 static void file_read_http(Pool& pool, c	Line 335 static void file_read_http(Pool& pool, c

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

//processing results	//processing results
int pos=response.pos("\r\n\r\n", 4);	int pos=response.pos("\r\n\r\n", 4);
if(pos<1){	if(pos<1){
throw Exception("http.response",	throw Exception("http.response",
&file_spec,	&connect_string,
"bad response from host - no headers found");	"bad response from host - no headers found");
}	}
String header_block=response.mid(0, pos);	String header_block=response.mid(0, pos);
Line 319 static void file_read_http(Pool& pool, c	Line 354 static void file_read_http(Pool& pool, c
size_t pos_after_ref=0;	size_t pos_after_ref=0;
header_block.split(aheaders, &pos_after_ref, "\r\n", 2);	header_block.split(aheaders, &pos_after_ref, "\r\n", 2);

//processing status code
const String *status_line=aheaders.get_string(0);
if(!status_line){
throw Exception("http.response",
&file_spec,
"bad response from host - no status line ");
}
//processing headers	//processing headers
Array astatus(pool);
pos_after_ref=0;
status_line->split(astatus, &pos_after_ref, " ", 1);
const String * status_code=astatus.get_string(1);

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

headers.put(	headers.put(
Line 345 static void file_read_http(Pool& pool, c	Line 368 static void file_read_http(Pool& pool, c
new(pool) VString(line->mid(pos+2, line->size())));	new(pool) VString(line->mid(pos+2, line->size())));
} else	} else
throw Exception("http.response",	throw Exception("http.response",
&file_spec,	&connect_string,
"bad response from host - bad headers \"%s\"", header_block.cstr());	"bad response from host - bad headers \"%s\"", header_block.cstr());
}	}

// output response	// output response
data=body.cstr(); data_size=body.size();	data=body.cstr(); data_size=body.size();
if(out_fields) {	if(out_fields) {
headers.put(file_status_name, new(pool) VString(status_code));	headers.put(*file_status_name, new(pool) VInt(pool, status_code));
*out_fields=&headers;	*out_fields=&headers;
}	}
}	}
Line 362 struct File_read_action_info {	Line 385 struct File_read_action_info {
void *data; size_t data_size;	void *data; size_t data_size;
};	};
#endif	#endif
static void file_read_action(Pool& pool,	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,
Line 387 static void file_read_action(Pool& pool,	Line 410 static void file_read_action(Pool& pool,
return;	return;
}	}
}	}
bool file_read(Pool& pool, const String& file_spec,	bool file_read(ConstStringPtr file_spec,
void*& data, size_t& data_size,	void*& data, size_t& data_size,
bool as_text, Hash params, Hash* out_fields,	bool as_text, Hash params, Hash* out_fields,
bool fail_on_read_problem) {	bool fail_on_read_problem) {
Line 404 bool file_read(Pool& pool, const String&	Line 427 bool file_read(Pool& pool, const String&
}	}

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

fix_line_breaks((char *)(data), data_size);	fix_line_breaks((char *)(data), data_size);
// note: after fixing	// note: after fixing
((char*&)(data))[data_size]=0;	((char*&)(data))[data_size]=0;
Line 412 bool file_read(Pool& pool, const String&	Line 444 bool file_read(Pool& pool, const String&
return result;	return result;
}	}

bool file_read_action_under_lock(Pool& pool, 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) {
Line 443 bool file_read_action_under_lock(Pool& p	Line 475 bool file_read_action_under_lock(Pool& p
"stat failed: %s (%d), actual filename '%s'",	"stat failed: %s (%d), actual filename '%s'",
strerror(errno), errno, fname);	strerror(errno), errno, fname);

#ifdef NO_FOREIGN_GROUP_FILES	#ifdef PA_SAFE_MODE
if(finfo.st_gid/foreign?/!=getegid()) {	if(finfo.st_uid/foreign?/!=geteuid()
	&& finfo.st_gid/foreign?/!=getegid())
throw Exception("parser.runtime",	throw Exception("parser.runtime",
&file_spec,	&file_spec,
"parser reading files of foreign group disabled [recompile parser without --disable-foreign-group-files configure option], actual filename '%s'",	"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);	fname);
#endif	#endif

Line 495 bool file_write_action_under_lock(	Line 528 bool file_write_action_under_lock(
if((f=open(fname,	if((f=open(fname,
O_CREAT\|O_RDWR	O_CREAT\|O_RDWR
\|(as_text?_O_TEXT:_O_BINARY)	\|(as_text?_O_TEXT:_O_BINARY)
\|(do_append?O_APPEND:0), 0664))>=0) {	\|(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) {
if((do_block?lock_exclusive_blocking(f):lock_exclusive_nonblocking(f))!=0) {	if((do_block?lock_exclusive_blocking(f):lock_exclusive_nonblocking(f))!=0) {
Exception e("file.lock",	Exception e("file.lock",
&file_spec,	&file_spec,
Line 510 bool file_write_action_under_lock(	Line 543 bool file_write_action_under_lock(
try {	try {
action(f, context);	action(f, context);
} catch(...) {	} catch(...) {
	#ifdef HAVE_FTRUNCATE
if(!do_append)	if(!do_append)
ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower	ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower
	#endif
unlock(f);close(f);	unlock(f);close(f);
/re/throw;	/re/throw;
}	}

	#ifdef HAVE_FTRUNCATE
if(!do_append)	if(!do_append)
ftruncate(f, lseek(f, 0, SEEK_CUR)); // O_TRUNC truncates even exclusevely write-locked file [thanks to Igor Milyakov <virtan@rotabanner.com> for discovering]	ftruncate(f, lseek(f, 0, SEEK_CUR)); // O_TRUNC truncates even exclusevely write-locked file [thanks to Igor Milyakov <virtan@rotabanner.com> for discovering]
	#endif
unlock(f);close(f);	unlock(f);close(f);
return true;	return true;
} else	} else
Line 708 char* rsplit(char* string, char delim) {	Line 745 char* rsplit(char* string, char delim) {
}	}

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

Line 744 size_t stdout_write(const void *buf, siz	Line 781 size_t stdout_write(const void *buf, siz
#endif	#endif
}	}

char* unescape_chars(Pool& pool, const char* cp, int len) {	char* unescape_chars(const char* cp, int len) {
char* s=(char *)pool.malloc(len + 1, 5);	char* s=(char *)pool.malloc(len + 1, 5);
enum EscapeState {	enum EscapeState {
EscapeRest,	EscapeRest,
Line 914 int pa_sleep(unsigned long secs, unsigne	Line 951 int pa_sleep(unsigned long secs, unsigne
return (select(0, NULL, NULL, NULL, &t) == -1 ? errno : 0);	return (select(0, NULL, NULL, NULL, &t) == -1 ? errno : 0);
#endif	#endif
}	}

E-mail:

Removed from v.1.129
changed lines
	Added in v.1.143.2.1