X-Git-Url: http://git.tdb.fi/?p=libs%2Fdatafile.git;a=blobdiff_plain;f=source%2Fbinaryparser.cpp;h=4ed2877a4424c2389c890473a0e71918418516ba;hp=c3d53d7241d4aaa069aeee617f98b16f6a504cb8;hb=ed78b585cfc4ecb44972e346857e887b183fd7a7;hpb=2f79370bffe0bac865dc97c5114dc87c1936fbb4

diff --git a/source/binaryparser.cpp b/source/binaryparser.cpp
index c3d53d7..4ed2877 100644
--- a/source/binaryparser.cpp
+++ b/source/binaryparser.cpp
@@ -1,14 +1,13 @@
-/* $Id$
-
-This file is part of libmspdatafile
-Copyright © 2007  Mikko Rasa, Mikkosoft Productions
-Distributed under the LGPL
-*/
-
+#include <limits>
 #include <sys/param.h>
-#include <msp/strings/formatter.h>
+#include <msp/core/maputils.h>
+#include <msp/strings/format.h>
+#include "argumentstore.h"
 #include "binaryparser.h"
+#include "binfloat.h"
+#include "except.h"
 #include "input.h"
+#include "loaderaction.h"
 
 using namespace std;
 
@@ -17,153 +16,257 @@ namespace DataFile {
 
 BinaryParser::BinaryParser(Input &i, const string &s):
 	ParserMode(i, s),
-	first(true)
+	float_precision(32),
+	cur_info(0)
 {
-	dict[1]=DictEntry("__st", "iss");
-	dict[2]=DictEntry("__enum", "is");
+	dict[-1] = StatementInfo("__kwd", "iss");
+	dict[-2] = StatementInfo("__str", "is");
+	dict[-3] = StatementInfo("__flt", "i");
 }
 
 Statement BinaryParser::parse()
 {
-	while(1)
+	const StatementKey *key;
+	if(cur_info)
 	{
-		Statement st=parse_statement();
-		if(st.keyword=="__st")
-		{
-			if(st.args.size()!=3)
-				throw TypeError(src+": Keyword definition must have three arguments");
-
-			const unsigned id=st.args[0].get<unsigned>();
-			const string &kw=st.args[1].get<const string &>();
-			const string &args=st.args[2].get<const string &>();
-			dict[id]=DictEntry(kw, args);
-		}
-		else if(st.keyword=="__enum")
-		{
-			if(st.args.size()!=2)
-				throw TypeError(src+": Enum definition must have three arguments");
-
-			const unsigned id=st.args[0].get<unsigned>();
-			enums[id]=st.args[1].get<const std::string &>();
-		}
-		else
-			return st;
+		key = &cur_info->key;
+		cur_info = 0;
 	}
-}
-
-Statement BinaryParser::parse_statement()
-{
-	while(first && in.peek()=='\n')
-		in.get();
-	first=false;
-
-	unsigned id=parse_int();
-	if(!in)
-		return Statement();
+	else
+	{
+		int id = parse_int();
+		if(!in)
+			return Statement();
 
-	Dictionary::const_iterator i=dict.find(id);
-	if(i==dict.end())
-		throw ParseError(format("%s: Unknown statement ID %d", src, id), src, 0);
-	const DictEntry &de=i->second;
+		key = &get_item(dict, id).key;
+	}
 
 	Statement result;
-	result.keyword=de.keyword;
-	result.source=src;
+	result.keyword = key->keyword;
+	result.source = src;
 
-	for(unsigned j=0; j<de.args.size(); ++j)
+	for(unsigned j=0; j<key->signature.size(); ++j)
 	{
-		switch(de.args[j])
+		switch(key->signature[j])
 		{
-		case 'i':
+		case IntType::signature:
 			result.args.push_back(parse_int());
 			break;
-		case 'f':
+		case FloatType::signature:
 			result.args.push_back(parse_float());
 			break;
-		case 's':
+		case StringType::signature:
 			result.args.push_back(parse_string());
 			break;
-		case 'b':
+		case BoolType::signature:
 			result.args.push_back(parse_bool());
 			break;
-		case 'e':
-			result.args.push_back(Value(ENUM, parse_enum()));
+		case SymbolType::signature:
+			result.args.push_back(parse_symbol());
 			break;
 		}
 	}
 
-	unsigned nsub=parse_int();
-	for(unsigned j=0; j<nsub; ++j)
+	unsigned upper_nsub = (sub_remaining.empty() ? 0 : sub_remaining.back());
+
+	unsigned nsub = parse_int();
+	for(unsigned j = 0; j<nsub; ++j)
 		result.sub.push_back(parse());
 
-	result.valid=true;
+	if(upper_nsub>0)
+		sub_remaining.back() = upper_nsub-1;
+
+	result.valid = true;
 
 	return result;
 }
 
-long long BinaryParser::parse_int()
+void BinaryParser::process_control_statement(const Statement &st)
 {
-	long long result=0;
-	unsigned bits=0;
+	if(st.keyword=="__kwd")
+	{
+		int id = st.args[0].get<int>();
+		if(id<=0)
+			throw bad_definition("__kwd");
+
+		const string &kw = st.args[1].get<const string &>();
+		const string &args = st.args[2].get<const string &>();
+		for(string::const_iterator i=args.begin(); i!=args.end(); ++i)
+			for(unsigned j=0; valid_signatures[j]!=*i; ++j)
+				if(!valid_signatures[j])
+					throw bad_definition("__kwd");
+
+		dict[id] = StatementInfo(kw, args);
+	}
+	else if(st.keyword=="__str")
+	{
+		int id = st.args[0].get<int>();
+		if(id<=0)
+			throw bad_definition("__str");
+
+		strings[id] = st.args[1].get<const string &>();
+	}
+	else if(st.keyword=="__flt")
+		float_precision = st.args[0].get<unsigned>();
+}
+
+const StatementKey *BinaryParser::peek(unsigned level)
+{
+	if(level>sub_remaining.size())
+		throw nesting_error("bad level");
+	while(level<sub_remaining.size())
+	{
+		// Discard any substatements that haven't been parsed yet
+		for(unsigned i=sub_remaining.back(); i-->0; )
+			parse();
+		sub_remaining.pop_back();
+		cur_info = 0;
+	}
+
+	if(!sub_remaining.empty() && sub_remaining.back()==0)
+	{
+		// No more substatements on this level
+		cur_info = 0;
+		return 0;
+	}
+
+	if(cur_info)
+		return &cur_info->key;
+
+	int id = parse_int();
+	if(!in)
+		return 0;
+
+	cur_info = &get_item(dict, id);
+	return &cur_info->key;
+}
+
+bool BinaryParser::parse_and_load(unsigned level, Loader &ldr, const LoaderAction &act)
+{
+	if(!cur_info && !peek(level))
+		return false;
+
+	ArgumentStore args(*cur_info);
+	for(unsigned i=0; i<cur_info->key.signature.size(); ++i)
+		switch(cur_info->key.signature[i])
+		{
+		case IntType::signature:
+			args.set(i, parse_int());
+			break;
+		case FloatType::signature:
+			args.set(i, parse_float());
+			break;
+		case BoolType::signature:
+			args.set(i, parse_bool());
+			break;
+		case StringType::signature:
+			args.set(i, parse_string());
+			break;
+		case SymbolType::signature:
+			args.set(i, parse_symbol());
+			break;
+		}
+
+	if(!sub_remaining.empty())
+		--sub_remaining.back();
+	sub_remaining.push_back(parse_int());
+	cur_info = 0;
+
+	act.execute(ldr, args);
+
+	return true;
+}
+
+IntType::Store BinaryParser::parse_int()
+{
+	IntType::Store result = 0;
+	unsigned bits = 0;
 
 	while(in)
 	{
-		int c=in.get();
+		int c = in.get();
 
-		result=result<<7 | c&0x7F;
-		bits+=7;
+		result = (result<<7) | (c&0x7F);
+		bits += 7;
 
 		if(!(c&0x80))
 			break;
 	}
 
-	const long long mask=1<<(bits-1);
-	result=(result^mask)-mask;
+	const IntType::Store mask = 1LL<<(bits-1);
+	result = (result^mask)-mask;
 
 	return result;
 }
 
-float BinaryParser::parse_float()
+FloatType::Store BinaryParser::parse_float()
 {
-	union
+	UInt64 encoded = 0;
+	for(unsigned i=0; i<float_precision; i+=8)
+	{
+		int c = in.get();
+		encoded = (encoded<<8) | (c&0xFF);
+	}
+
+	BinFloat bf = BinFloat::explode(encoded, float_precision);
+
+	if(numeric_limits<FloatType::Store>::is_iec559)
+		return bf.compose_iec559<FloatType::Store>();
+	else
 	{
-		float f;
-		char d[sizeof(float)];
-	};
-
-#if BYTE_ORDER == LITTLE_ENDIAN
-	for(unsigned i=sizeof(float); i--;)
-		d[i]=in.get();
-#else
-	for(unsigned i=0; i<sizeof(float); ++i)
-		d[i]=in.get();
-#endif
-
-	return f;
+		/* Put the float together with arithmetic since we don't know its
+		internal layout */
+		FloatType::Store f = 0;
+		if(bf.infinity)
+		{
+			if(numeric_limits<FloatType::Store>::has_infinity)
+				f = numeric_limits<FloatType::Store>::infinity();
+			else
+				f = numeric_limits<FloatType::Store>::max();
+		}
+		else
+		{
+			for(unsigned i=0; i<64; ++i)
+			{
+				f /= 2;
+				if(bf.mantissa&1)
+					f += 1;
+				bf.mantissa >>= 1;
+			}
+			for(int i=0; i<bf.exponent; ++i)
+				f *= 2;
+			for(int i=0; i>bf.exponent; --i)
+				f /= 2;
+		}
+		if(bf.sign)
+			f = -f;
+		return f;
+	}
 }
 
-bool BinaryParser::parse_bool()
+BoolType::Store BinaryParser::parse_bool()
 {
 	return in.get();
 }
 
-string BinaryParser::parse_string()
+StringType::Store BinaryParser::parse_string()
 {
-	unsigned len=parse_int();
-	string result;
-	result.reserve(len);
-	for(unsigned i=0; i<len; ++i)
-		result+=in.get();
-	return result;
+	int len = parse_int();
+	if(len>=0)
+	{
+		string result;
+		result.reserve(len);
+		for(int i = 0; i<len; ++i)
+			result += in.get();
+		return result;
+	}
+	else
+		return get_item(strings, -len);
 }
 
-string BinaryParser::parse_enum()
+SymbolType::Store BinaryParser::parse_symbol()
 {
-	unsigned id=parse_int();
-	EnumMap::iterator i=enums.find(id);
-	if(i==enums.end())
-		throw KeyError("Unknown enum");
-	return i->second;
+	return get_item(strings, parse_int());
 }
 
 } // namespace DataFile