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 1.140.2.1 and 1.143.2.2

version 1.140.2.1, 2003/02/24 12:38:54	version 1.143.2.2, 2003/01/24 08:19:09
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"
#include "pa_vdate.h"

#ifdef WIN32	#ifdef WIN32
# include <windows.h>	# include <windows.h>
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){
bool check_status=true;	int result=0;
ssize_t EOLat=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, "remote HTTP server response", 0);	response.APPEND_TAINTED(buf, size, "remote HTTP server response", 0);
if(check_status && (EOLat=response.pos("\r\n", 2))>=0) { // checking status in first response	if(!result && (EOLat=response.pos("\r\n", 2))>=0) { // checking status in first response
check_status=false;

const String& status_line=response.mid(0, (size_t)EOLat);	const String& status_line=response.mid(0, (size_t)EOLat);
Array astatus(response.pool());	Array astatus(response.pool());
size_t pos_after_ref=0; status_line.split(astatus, &pos_after_ref, " ", 1);	size_t pos_after_ref=0; status_line.split(astatus, &pos_after_ref, " ", 1);
const String* status_code=astatus.get_string(1);	result=astatus.get_string(1)->as_int();

if(!status_code \|\| *status_code!="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(check_status)	if(result)
	return result;
	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.size());
Line 171 static void http_read_response(String& r	Line 171 static void http_read_response(String& r
/* ******************** request ************************* */	/* ******************** request ************************* */

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

#ifdef PA_USE_ALARM	#ifdef WE_CAN_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 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,
"zero hostname"); //never	"zero hostname"); //never

#ifdef PA_USE_ALARM	#ifdef WE_CAN_USE_ALARM
signal(SIGALRM, timeout_handler);	signal(SIGALRM, timeout_handler);
#endif	#endif
int sock=-1;	int sock=-1;
#ifdef PA_USE_ALARM
if(sigsetjmp(timeout_env, 1)) {
// stupid gcc [2.95.4] generated bad code
// which failed to handle sigsetjmp+throw: crashed inside of pre-throw code.
// rewritten simplier [though duplicating closesocket code]
if(sock>=0)
closesocket(sock);
throw Exception("http.timeout",
origin_string,
"timeout occured while retrieving document");
} else {
alarm(timeout);
#endif
try {	try {
struct sockaddr_in dest;	int result;
	#ifdef WE_CAN_USE_ALARM
if(!set_addr(&dest, host, port))	if(sigsetjmp(timeout_env, 1))
throw Exception("http.host",	throw Exception("http.timeout",
origin_string,	origin_string,
"can not resolve hostname \"%s\"", host);	"timeout occured while retrieving document");
	else {
	alarm(timeout);
	#endif
	struct sockaddr_in dest;

if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/0/))<0)	if(!set_addr(&dest, host, port))
throw Exception("http.connect",	throw Exception("http.host",
origin_string,	origin_string,
"can not make socket: %s (%d)", strerror(errno), errno);	"can not resolve hostname \"%s\"", host);
if(connect(sock, (struct sockaddr *)&dest, sizeof(dest)))
throw Exception("http.connect",	if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/0/))<0)
origin_string,	throw Exception("http.connect",
"can not connect to host \"%s\": %s (%d)", host, strerror(errno), errno);	origin_string,
size_t request_size=strlen(request);	"can not make socket: %s (%d)", strerror(errno), errno);
if(send(sock, request, request_size, 0)!=(ssize_t)request_size)	if(connect(sock, (struct sockaddr *)&dest, sizeof(dest)))
throw Exception("http.connect",	throw Exception("http.connect",
origin_string,	origin_string,
"error sending request: %s (%d)", strerror(errno), errno);	"can not connect to host \"%s\": %s (%d)", host, strerror(errno), errno);
	size_t request_size=strlen(request);
	if(send(sock, request, request_size, 0)!=(ssize_t)request_size)
	throw Exception("http.connect",
	origin_string,
	"error sending request: %s (%d)", strerror(errno), errno);

http_read_response(response, sock);	result=http_read_response(response, sock, fail_on_status_ne_200);
closesocket(sock);	closesocket(sock);
#ifdef PA_USE_ALARM	#ifdef WE_CAN_USE_ALARM
alarm(0);	alarm(0);
	}
#endif	#endif
	return result;
} catch(...) {	} catch(...) {
if(sock>=0)	if(sock>=0)
closesocket(sock);	closesocket(sock);
#ifdef PA_USE_ALARM	#ifdef WE_CAN_USE_ALARM
alarm(0);	alarm(0);
#endif	#endif
/re/throw;	/re/throw;
Line 271 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;

String connect_string(pool);	String connect_string(pool);
Line 304 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,
Line 330 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,
&connect_string, 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);
Line 348 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",
&connect_string,
"bad response from host - no status line ");
}
//processing headers	//processing headers
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)) {
Line 375 static void file_read_http(Pool& pool, c	Line 374 static void file_read_http(Pool& pool, c

// 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) VInt(pool, status_code));
*out_fields=&headers;	*out_fields=&headers;
	}
}	}

#ifndef DOXYGEN	#ifndef DOXYGEN
Line 384 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 409 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 443 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 744 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 780 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 952 int pa_sleep(unsigned long secs, unsigne	Line 953 int pa_sleep(unsigned long secs, unsigne
}	}


// attributed meaning

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.1.140.2.1
changed lines
	Added in v.1.143.2.2