From c80b23de8b70776e37f37e4b8fc2003c553448d6 Mon Sep 17 00:00:00 2001
From: Mikko Rasa <tdb@tdb.fi>
Date: Mon, 1 Nov 2021 12:54:37 +0200
Subject: [PATCH] Add a class for loading raw bulk data

---
 source/rawdata.cpp | 92 ++++++++++++++++++++++++++++++++++++++++++++++
 source/rawdata.h   | 44 ++++++++++++++++++++++
 2 files changed, 136 insertions(+)
 create mode 100644 source/rawdata.cpp
 create mode 100644 source/rawdata.h

diff --git a/source/rawdata.cpp b/source/rawdata.cpp
new file mode 100644
index 0000000..7a90c9d
--- /dev/null
+++ b/source/rawdata.cpp
@@ -0,0 +1,92 @@
+#include <stdexcept>
+#include <msp/io/zlibcompressed.h>
+#include "collection.h"
+#include "except.h"
+#include "rawdata.h"
+
+using namespace std;
+
+namespace Msp {
+namespace DataFile {
+
+RawData::~RawData()
+{
+	delete owned_data;
+	delete compressed;
+	if(in_owned)
+		delete in;
+}
+
+void RawData::open_file(Collection &coll, const string &fn)
+{
+	if(in)
+		throw logic_error("input already exists");
+
+	RefPtr<IO::Base> opened = coll.open_raw(fn);
+	if(!opened)
+		throw IO::file_not_found(fn);
+
+	open_io(*opened, fn);
+	opened.release();
+	in_owned = true;
+}
+
+void RawData::open_io(IO::Base &i, const string &fn)
+{
+	if(in)
+		throw logic_error("input already exists");
+
+	char header[14];
+	unsigned len = i.read(header, sizeof(header));
+	if(len!=sizeof(header))
+		throw data_error(fn, 0, "Missing header");
+	if(header[0]!='M' || header[1]!='D' || header[2]!='R' || header[3]!=1)
+		throw data_error(fn, 0, "Bad header");
+
+	size = 0;
+	for(unsigned j=0; j<8; ++j)
+		size = (size<<8) | static_cast<unsigned char>(header[4+j]);
+
+	uint16_t flags = (static_cast<unsigned char>(header[12])<<8) | static_cast<unsigned char>(header[13]);
+
+	src_name = fn;
+	in = &i;
+	if(flags&COMPRESSED)
+		compressed = new IO::ZlibCompressed(*in, IO::M_READ);
+}
+
+void RawData::load()
+{
+	if(!in)
+		throw logic_error("no input");
+	if(data)
+		throw logic_error("already loaded");
+
+	owned_data = new char[size];
+	load_into(owned_data);
+}
+
+void RawData::load_into(void *buffer)
+{
+	if(!in)
+		throw logic_error("no input");
+
+	data = static_cast<char *>(buffer);
+
+	IO::Base *src = (compressed ? compressed : in);
+	size_t pos = 0;
+	while(pos<size)
+	{
+		size_t len = src->read(data+pos, size-pos);
+		if(!len)
+			throw data_error(src_name, 0, "Truncated data");
+		pos += len;
+	}
+
+	if(in_owned)
+		delete in;
+	in = nullptr;
+}
+
+} // namespace DataFile
+} // namespace Msp
diff --git a/source/rawdata.h b/source/rawdata.h
new file mode 100644
index 0000000..2db390b
--- /dev/null
+++ b/source/rawdata.h
@@ -0,0 +1,44 @@
+#ifndef MSP_DATAFILE_RAWDATA_H_
+#define MSP_DATAFILE_RAWDATA_H_
+
+#include <string>
+#include <msp/core/noncopyable.h>
+#include <msp/io/base.h>
+
+namespace Msp {
+namespace DataFile {
+
+class Collection;
+
+class RawData: public NonCopyable
+{
+private:
+	enum Flags
+	{
+		COMPRESSED = 1
+	};
+
+	std::string src_name;
+	IO::Base *in = nullptr;
+	bool in_owned = false;
+	IO::Base *compressed = nullptr;
+	std::size_t size = 0;
+	char *data = nullptr;
+	char *owned_data = nullptr;
+
+public:
+	~RawData();
+
+	void open_file(Collection &, const std::string &);
+	void open_io(IO::Base &, const std::string &);
+	void load();
+	void load_into(void *);
+
+	std::size_t get_size() const { return size; }
+	const void *get_data() const { return data; }
+};
+
+} // namespace DataFile
+} // namespace Msp
+
+#endif
-- 
2.43.0