Return to op.C CVS log

Up to [parser3project] / parser3 / src / classes

^try{^call{}} now has better stack trace [has "call" & co there]

/** @file
	Parser: parser @b operators.

	Copyright (c) 2001, 2002 ArtLebedev Group (http://www.artlebedev.com)
	Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
*/

static const char* IDENT_OP_C="$Date: 2002/10/14 13:53:20 $";

#include "classes.h"
#include "pa_common.h"
#include "pa_request.h"
#include "pa_vint.h"
#include "pa_sql_connection.h"
#include "pa_vdate.h"
#include "pa_vmethod_frame.h"

// limits

#define MAX_LOOPS 10000

// defines

#define CASE_DEFAULT_VALUE "DEFAULT"

// class

class MOP : public Methoded {
public:
	MOP(Pool& pool);
public: // Methoded
	bool used_directly() { return true; }
	void configure_user(Request& r);

private:
	String main_sql_name;
	String main_sql_drivers_name;
};

// methods

static void _if(Request& r, const String&, MethodParams *params) {
	Value& condition_code=params->as_junction(0, "condition must be expression");

	bool condition=r.process_to_value(condition_code, 
		/*0/*no name* /,*/
		false/*don't intercept string*/).as_bool();
	if(condition)
		r.write_pass_lang(r.process(params->as_junction(1, "'then' parameter must be code")));
	else if(params->size()>2)
		r.write_pass_lang(r.process(params->as_junction(2, "'else' parameter must be code")));
}

static void _untaint(Request& r, const String& method_name, MethodParams *params) {
	Pool& pool=r.pool();

	uchar lang;
	if(params->size()==1)
		lang=String::UL_AS_IS; // mark as simply 'tainted'. useful in html from sql 
	else {
		const String& lang_name=params->as_string(0, "lang must be string");
		lang=untaint_lang_name2enum->get_int(lang_name);
		if(!lang)
			throw Exception(0,
				&lang_name,
				"invalid taint language");
	}

	{
		Value& vbody=params->as_junction(params->size()-1, "body must be code");
		
		Temp_lang temp_lang(r, lang); // set temporarily specified ^untaint[language;
		r.write_pass_lang(r.process(vbody)); // process marking tainted with that lang
	}
}

static void _taint(Request& r, const String&, MethodParams *params) {
	Pool& pool=r.pool();

	uchar lang;
	if(params->size()==1)
		lang=String::UL_TAINTED; // mark as simply 'tainted'. useful in table:set
	else {
		const String& lang_name=params->as_string(0, "lang must be string");
		lang=untaint_lang_name2enum->get_int(lang_name);
		if(!lang)
			throw Exception(0,
				&lang_name,
				"invalid taint language");
	}

	{
		Value& vbody=params->as_no_junction(params->size()-1, "body must not be code");
		
		String& result=*new(pool) String(pool);
		result.append(
			vbody.as_string(),  // process marking tainted with that lang
			lang, true);  // force result language to specified
		r.write_pass_lang(result);
	}
}

static void _process(Request& r, const String& method_name, MethodParams *params) {
	Pool& pool=r.pool();
	VStateless_class& self_class=*r.get_self()->get_class();
	const Method *main_method;
	{
		// temporary remove language change
		Temp_lang temp_lang(r, String::UL_PASS_APPENDED);
		// temporary zero @main so to maybe-replace it in processed code
		Temp_method temp_method_main(self_class, r.main_method_name, 0);
		// temporary zero @auto so it wouldn't be auto-called in Request::use_buf
		Temp_method temp_method_auto(self_class, *auto_method_name, 0);
		
		// evaluate source to process
		const String& source=r.process_to_string(params->as_junction(0, "body must be code"));
		// calculate pseudo file name of processed chars
		// would be something like "/some/file(4) process"
		char local_place[MAX_STRING];
#ifndef NO_STRING_ORIGIN
		const Origin& source_origin=source.origin();
		const Origin& method_origin=method_name.origin();
		size_t place_size;
		if(source_origin.file==method_origin.file)
			place_size=snprintf(local_place, MAX_STRING, "%s(%d) %s",  // same file
				source_origin.file?source_origin.file:"unknown_file", 1+source_origin.line, 
				method_name.cstr())+1;
		else // different files ^process{external__file_text__or__sql}
			place_size=snprintf(local_place, MAX_STRING, "%s", 
				source_origin.file?source_origin.file:"unknown_file")+1;
#else
		strncpy(local_place, method_name.cstr(), MAX_STRING-1); place[MAX_STRING-1]=0;
		size_t place_size=strlen(local_place)+1;
#endif
		char *heap_place=(char *)r.malloc(place_size);
		memcpy(heap_place, local_place, place_size);

		// process source code, append processed methods to 'self' class
		// maybe-define new @main
		r.use_buf(
			source.cstr(String::UL_UNSPECIFIED, r.connection(0)), 
			*new(pool) String(pool, heap_place, place_size, true /*tainted*/),
			heap_place, 
			&self_class);
		
		// main_method
		main_method=self_class.get_method(r.main_method_name);
	}
	// after restoring current-request-lang
	// maybe-execute @main[]
	if(main_method) {
		// execute!	
		r.execute(*main_method->parser_code);
	}
}
	
static void _rem(Request& r, const String&, MethodParams *params) {
	params->as_junction(0, "body must be code");
}

static void _while(Request& r, const String& method_name, MethodParams *params) {
	Pool& pool=r.pool();

	Value& vcondition=params->as_junction(0, "condition must be expression");
	Value& body=params->as_junction(1, "body must be code");

	// while...
	int endless_loop_count=0;
	while(true) {
		if(++endless_loop_count>=MAX_LOOPS) // endless loop?
			throw Exception("parser.runtime",
				&method_name,
				"endless loop detected");

		bool condition=r.process_to_value(vcondition, 
				/*0/*no name* /,*/
				false/*don't intercept string*/).as_bool();
		if(!condition) // ...condition is true
			break;

		// write processed body
		r.write_pass_lang(r.process(body));
	}
}

static void _use(Request& r, const String& method_name, MethodParams *params) {
	Value& vfile=params->as_no_junction(0, "file name must not be code");
	r.use_file(vfile.as_string());
}

static void _for(Request& r, const String& method_name, MethodParams *params) {
	Pool& pool=r.pool();
	const String& var_name=params->as_string(0, "var name must be string");
	int from=params->as_int(1, "from must be int", r);
	int to=params->as_int(2, "to must be int", r);
	Value& body_code=params->as_junction(3, "body must be code");
	Value *delim_maybe_code=params->size()>4?&params->get(4):0;

	if(to-from>=MAX_LOOPS) // too long loop?
		throw Exception("parser.runtime",
			&method_name,
			"endless loop detected");

	bool need_delim=false;
	VInt *vint=new(pool) VInt(pool, 0);
	r.get_method_frame()->put_element(var_name, vint, false);
	for(int i=from; i<=to; i++) {
		vint->set_int(i);

		StringOrValue sv_processed=r.process(body_code);
		const String *s_processed=sv_processed.get_string();
		if(delim_maybe_code && s_processed && s_processed->size()) { // delimiter set and we have body
			if(need_delim) // need delim & iteration produced string?
				r.write_pass_lang(r.process(*delim_maybe_code));
			need_delim=true;
		}
		r.write_pass_lang(sv_processed);
	}
}

static void _eval(Request& r, const String& method_name, MethodParams *params) {
	Value& expr=params->as_junction(0, "need expression");
	// evaluate expresion
	Value *value_result=r.process_to_value(expr, 
		/*0/*no name YET* /,*/
		true/*don't intercept string*/).as_expr_result();
	if(params->size()>1) {
		Value& fmt=params->as_no_junction(1, "fmt must not be code");

		Pool& pool=r.pool();
		String& string=*new(pool) String(pool);
		string.APPEND_CONST(format(pool, value_result->as_double(), fmt.as_string().cstr()));
		r.write_no_lang(string);
	} else
		r.write_no_lang(*value_result);
}

static void _connect(Request& r, const String& method_name, MethodParams *params) {
	Pool& pool=r.pool();
#ifdef RESOURCES_DEBUG
struct timeval mt[2];
#endif
	Value& url=params->as_no_junction(0, "url must not be code");
	Value& body_code=params->as_junction(1, "body must be code");

	Table *protocol2driver_and_client=
		static_cast<Table *>(r.classes_conf.get(r.OP.name()));

#ifdef RESOURCES_DEBUG
//measure:before
gettimeofday(&mt[0],NULL);
#endif
	// connect
	SQL_Connection_ptr connection=SQL_driver_manager->get_connection(
		url.as_string(), method_name, protocol2driver_and_client);

#ifdef RESOURCES_DEBUG
//measure:after connect
gettimeofday(&mt[1],NULL);

double t[2];
for(int i=0;i<2;i++)
    t[i]=mt[i].tv_sec+mt[i].tv_usec/1000000.0;

r.sql_connect_time+=t[1]-t[0];
#endif
	Temp_connection temp_connection(r, connection.get());
	// execute body
	try {
		r.write_assign_lang(r.process(body_code));
	} catch(...) { // process problem
		connection->mark_to_rollback();
		/*re*/throw; 
	}
}

#ifndef DOXYGEN
struct Switch_data {
	Request *r;
	Value *searching;
	Value *found;
	Value *_default;
};
#endif
static void _switch(Request& r, const String&, MethodParams *params) {
	Switch_data data={&r, &r.process_to_value(params->get(0))};	
	Temp_hash_value switch_data_setter(r.classes_conf, *switch_data_name, &data);

	Value& cases_code=params->as_junction(1, "switch cases must be code");
	// execution of found ^case[...]{code} must be in context of ^switch[...]{code}
	// because of stacked WWrapper used there as wcontext
	r.process(cases_code, true/*intercept_string*/);
	if(Value *selected_code=data.found ? data.found : data._default) {
		// setting code context, would execute in ^switch[...]{>>context<<}
		//selected_code->get_junction()->change_context(cases_code.get_junction());
		r.write_pass_lang(r.process(*selected_code));
	}
}

static void _case(Request& r, const String& method_name, MethodParams *params) {
	Pool& pool=r.pool();

	Switch_data *data=static_cast<Switch_data *>(r.classes_conf.get(*switch_data_name));
	if(!data)
		throw Exception("parser.runtime",
			&method_name,
			"without switch");

	int count=params->size();
	Value *code=&params->as_junction(--count, "case result must be code");
	
	// killing context for safety, would execute in ^switch[...]{>>context<<}
	// reason: context is stacked, and it would become invalid afterwards
	//code->get_junction()->change_context(0);

	for(int i=0; i<count; i++) {
		Value& value=r.process_to_value(params->get(i));

		if(value.as_string() == CASE_DEFAULT_VALUE) {
			data->_default=code;
			break;
		}

		bool matches;
		if(data->searching->is_string())
			matches=data->searching->as_string() == value.as_string();
		else
			matches=data->searching->as_double() == value.as_double();

		if(matches) {
			if(data->found)
				throw Exception("parser.runtime",
					&method_name,
					"duplicate found");

			data->found=code;
			break;
		}
	}
}

// cache--

// consts

const int DATA_STRING_SERIALIZED_VERSION=0x0002;

// helper types

#ifndef DOXYGEN
struct Data_string_serialized_prolog {
	int version;
	time_t expires;
};
#endif

void cache_delete(const String& file_spec) {
	file_delete(file_spec, false/*fail_on_read_problem*/);
}

#ifndef DOXYGEN
struct Cache_data {
	time_t expires;
};
struct Locked_process_and_cache_put_action_info {
	Request *r;
	Cache_data *data;
	Value *body_code; const String *evaluated_body;
};
#endif
static void locked_process_and_cache_put_action(int f, void *context) {
	Locked_process_and_cache_put_action_info& info=
		*static_cast<Locked_process_and_cache_put_action_info *>(context);
	
	// body->process 
	info.evaluated_body=&info.r->process_to_string(*info.body_code);

	// expiration time not spoiled by ^cache(0) or something?
	if(info.data->expires > time(0)) {
		// string -serialize> buffer
		void *data; size_t data_size;
		info.evaluated_body->serialize(
			sizeof(Data_string_serialized_prolog), 
			data, data_size);
		Data_string_serialized_prolog& prolog=
			*static_cast<Data_string_serialized_prolog *>(data);
		prolog.version=DATA_STRING_SERIALIZED_VERSION;
		prolog.expires=info.data->expires;
		
		// buffer -write> file
		write(f, data, data_size);
	} else // expired!
		info.data->expires=0; // flag it so that could be easily checked by caller
}
const String *locked_process_and_cache_put(Request& r, 
									Value& body_code,
									Cache_data& data,
									const String& file_spec) {
	Locked_process_and_cache_put_action_info info={
		&r,
		&data,
		&body_code
	};

	const String *result=file_write_action_under_lock(
		file_spec, 
		"cache_put", locked_process_and_cache_put_action, &info,
		false/*as_text*/,
		false/*do_append*/,
		false/*block*/,
		false/*fail on lock problem*/) ? info.evaluated_body: 0;
	time_t now=time(0);
	if(data.expires<=now)
		cache_delete(file_spec);
	return result;
}
String *cache_get(Pool& pool, const String& file_spec, time_t now) {
	void* data; size_t data_size;
	if(file_read(pool, file_spec, 
			   data, data_size, 
			   false/*as_text*/, 
			   false/*fail_on_read_problem*/)
	    && data_size/* ignore reads which are empty due to 
			non-unary open+lockEX conflict with lockSH */) {
	
		Data_string_serialized_prolog& prolog=
			*static_cast<Data_string_serialized_prolog *>(data);

		String *result=new(pool) String(pool);
		if(
			data_size>=sizeof(Data_string_serialized_prolog)
			&& prolog.version==DATA_STRING_SERIALIZED_VERSION
			&& prolog.expires > now
			&& result->deserialize(
				sizeof(Data_string_serialized_prolog),  data, data_size, file_spec.cstr()))
			return result;
	}

	return 0;
}
static time_t as_expires(Request& r, const String& method_name, MethodParams *params, 
						int index, time_t now) {
	time_t result;
	Value& vlifespan_or_expires=params->get(index);
	if(Value *vdate=vlifespan_or_expires.as(VDATE_TYPE, false))
		result=static_cast<VDate *>(vdate)->get_time();
	else
		result=now+(time_t)params->as_double(index, "lifespan must be date or number", r);
	
	return result;
}
static const String as_file_spec(Request&r, MethodParams *params, int index) {
	return r.absolute(params->as_string(index, "filespec must be string"));
}
static void _cache(Request& r, const String& method_name, MethodParams *params) {
	Pool& pool=r.pool();
	time_t now=time(0);

	// ^cache[filename] ^cache(seconds) ^cache[expires date]
	if(params->size()==1) {
		if(params->get(0).is_string()) { // filename?
			cache_delete(as_file_spec(r, params, 0));
			return;
		}

		// secods|expires date
		Cache_data *data=static_cast<Cache_data *>(r.classes_conf.get(*cache_data_name));
		if(!data)
			throw Exception("parser.runtime",
				&method_name,
				"expire-time reducing instruction without cache");
		
		time_t expires=as_expires(r, method_name, params, 0, now);
		if(expires < data->expires)
			data->expires=expires;

		return;
	}
	
	// file_spec, expires, body code
	const String &file_spec=r.absolute(params->as_string(0, "filespec must be string"));

	Cache_data data;
	Temp_hash_value cache_data_setter(r.classes_conf, *cache_data_name, &data);
	data.expires=as_expires(r, method_name, params, 1, now);
	Value& body_code=params->as_junction(2, "body must be code");

	if(data.expires>now) { // valid 'expires' specified? try cached copy...
		// hence we don't hope to have unary create/lockEX
		// we need some plan to live in a life like that, so... 
		// worst races plan:
		// A        B
		// open
		//          |open
		// lockSH
		//          |nonblocking-lockEX fails
		// unlockSH
		// close, cache_get returns 0
		// open
		// nonblocking-lockEX succeeds; process, write, close
		//          |retry1: open
		// ...
		//          |lockSH succeeds; ...

		for(int retry=0; retry<2; retry++) {
			if(String *cached_body=cache_get(pool, file_spec, now)) { // have cached copy?
				// write it out 
				r.write_assign_lang(*cached_body);
				// happy with it
				return;
			}

			// non-blocked lock; process; cache it
			if(const String*processed_body=
				locked_process_and_cache_put(r, body_code, data, file_spec)) {
				// write it out 
				r.write_assign_lang(*processed_body);
				// happy with it
				return;
			} else { // somebody writing result right now
				pa_sleep(0, 500000); // waiting half a second
				retry=0; // prolonging our wait, than could cache_get it, without processing body_code
			}
		}
		throw Exception(0,
			&file_spec,
			"locking problem");
	} else { 
		// instructed not to cache; forget cached copy
		cache_delete(file_spec);
		// process
		const String& processed_body=r.process_to_string(body_code);
		// write it out 
		r.write_assign_lang(processed_body);
		// happy with it
		return;
	}
	// never reached
}

// also used in pa_request.C to pass param to @unhandled_exception
VHash& exception2vhash(Pool& pool, const Exception& e) {
	VHash& result=*new(pool) VHash(pool);
	Hash& hash=result.hash(0);
	if(const char *type=e.type(true))
		hash.put(*exception_type_part_name, new(pool) VString(*new(pool) String(pool, type)));
	if(const String *asource=e.problem_source()) {
		String& source=*new(pool) String(pool); 
		source.append(*asource, String::UL_TAINTED, true/*forced*/);

		hash.put(*exception_source_part_name, new(pool) VString(source));
#ifndef NO_STRING_ORIGIN
		const Origin& origin=source.origin();
		hash.put(*new(pool) String(pool, "file", 0, true),
			new(pool) VString(*new(pool) String(pool, origin.file)));
		hash.put(*new(pool) String(pool, "lineno"),
			new(pool) VInt(pool, 1+origin.line));
#endif
	}
	if(const char *ecomment=e.comment(true)) {
		int comment_size=strlen(ecomment);
		char *pcomment=(char *)pool.malloc(comment_size);
		memcpy(pcomment, ecomment, comment_size);
		hash.put(*exception_comment_part_name, 
			new(pool) VString(*new(pool) String(pool, pcomment, comment_size, true/*tainted*/)));
	}
	hash.put(*exception_handled_part_name, 
		new(pool) VBool(pool, false));

	return result;
}

static void _try_operator(Request& r, const String& method_name, MethodParams *params) {
	Pool& pool=r.pool();

	Value& body_code=params->as_junction(0, "body_code must be code");
	Value& catch_code=params->as_junction(1, "catch_code must be code");

	StringOrValue result;
	try {
		Request_context_saver context_saved(r); // taking snapshot of request processing status, restoring it at }
		result=r.process(body_code);
		context_saved.restore(); // manually restoring context
	} catch(const Exception& e) {
		VHash& vhash=exception2vhash(pool, e);

		Junction *junction=catch_code.get_junction();
		Value *method_frame=junction->method_frame;
		Value *saved_exception_var_value=method_frame->get_element(*exception_var_name, method_frame, false);
		junction->method_frame->put_element(*exception_var_name, &vhash, false);
		result=r.process(catch_code);
		bool handled=false;
		if(Value *value=static_cast<Value *>(vhash.hash(0).get(*exception_handled_part_name)))
			handled=value->as_bool();		
		junction->method_frame->put_element(*exception_var_name, saved_exception_var_value, false);

		if(!handled)
			throw(e); // rethrow
	}
	// write it out 
	r.write_pass_lang(result);
}

static void _throw_operator(Request& r, const String& method_name, MethodParams *params) {
	Pool& pool=r.pool();

	if(params->size()==1) {
		Value& param0=params->get(0);
		if(Hash *hash=param0.get_hash(&method_name)) {
			const char *type=0;
			if(Value *value=static_cast<Value *>(hash->get(*exception_type_part_name)))
				type=value->as_string().cstr();
			const String *source=0;
			if(Value *value=static_cast<Value *>(hash->get(*exception_source_part_name)))
				source=&value->as_string();
			const char *comment=0;
			if(Value *value=
				static_cast<Value *>(hash->get(*exception_comment_part_name)))
				comment=value->as_string().cstr();

			throw Exception(type,
				source?source:&method_name,
				comment);
		} else
			throw Exception("parser.runtime",
				&method_name,
				"one-param version has hash param");
	} else {
		const char *type=params->as_string(0, "type must be string").cstr();
		const String& source=params->as_string(1, "source must be string");
		const char *comment=params->size()>2?params->as_string(2, "comment must be string").cstr():0;
		throw Exception(type, &source, "%s", comment?comment:"");
	}
}
	
// constructor

MOP::MOP(Pool& apool) : Methoded(apool, "OP"),
	main_sql_name(apool, MAIN_SQL_NAME),
	main_sql_drivers_name(apool, MAIN_SQL_DRIVERS_NAME)
{
	// ^if(condition){code-when-true}
	// ^if(condition){code-when-true}{code-when-false}
	add_native_method("if", Method::CT_ANY, _if, 2, 3);

	// ^untaint[as-is|uri|sql|js|html|html-typo]{code}
	add_native_method("untaint", Method::CT_ANY, _untaint, 1, 2);

	// ^taint[as-is|uri|sql|js|html|html-typo]{code}
	add_native_method("taint", Method::CT_ANY, _taint, 1, 2);

	// ^process[code]
	add_native_method("process", Method::CT_ANY, _process, 1, 1);

	// ^rem{code}
	add_native_method("rem", Method::CT_ANY, _rem, 1, 10000);

	// ^while(condition){code}
	add_native_method("while", Method::CT_ANY, _while, 2, 2);

	// ^use[file]
	add_native_method("use", Method::CT_ANY, _use, 1, 1);

	// ^for[i](from-number;to-number-inclusive){code}[delim]
	add_native_method("for", Method::CT_ANY, _for, 3+1, 3+1+1);

	// ^eval(expr)
	// ^eval(expr)[format]
	add_native_method("eval", Method::CT_ANY, _eval, 1, 2);

	// ^connect[protocol://user:pass@host[:port]/database]{code with ^sql-s}
	add_native_method("connect", Method::CT_ANY, _connect, 2, 2);


	// ^cache[file_spec] delete cache
	// ^cache[file_spec](time){code} time=0 no cache
	add_native_method("cache", Method::CT_ANY, _cache, 1, 3);
	
	// switch

	// ^switch[value]{cases}
	add_native_method("switch", Method::CT_ANY, _switch, 2, 2);

	// ^case[value]{code}
	add_native_method("case", Method::CT_ANY, _case, 2, 10000);

	// try-catch

	// ^try{code}{catch code}
	add_native_method("try", Method::CT_ANY, _try_operator, 2, 2);
	// ^throw[$exception hash]
	// ^throw[type;source;comment]
	add_native_method("throw", Method::CT_ANY, _throw_operator, 1, 3);

}

// constructor & configurator

Methoded *MOP_create(Pool& pool) {
	return new(pool) MOP(pool);
}

void MOP::configure_user(Request& r) {
	Pool& pool=r.pool();

	// $MAIN:SQL.drivers
	if(Value *sql=r.main_class->get_element(main_sql_name, r.main_class, false))
		if(Value *element=sql->get_element(main_sql_drivers_name, sql, false))
			if(Table *protocol2library=element->get_table())
				r.classes_conf.put(name(), protocol2library);
}