nbt.py now simply *reads files*, nothing more

overviewer_core/nbt.py

@@ -16,57 +16,54 @@
 import gzip, zlib
 import struct
 import StringIO
-import numpy
+import functools
-from functools import wraps
 
 # decorator that turns the first argument from a string into an open file
 # handle
 def _file_loader(func):
-    @wraps(func)
+    @functools.wraps(func)
     def wrapper(fileobj, *args):
         if isinstance(fileobj, basestring):
             # Is actually a filename
-            fileobj = open(fileobj, 'rb',4096)
+            fileobj = open(fileobj, 'rb', 4096)
         return func(fileobj, *args)
     return wrapper
 
 @_file_loader
 def load(fileobj):
     """Reads in the given file as NBT format, parses it, and returns the
-    result
+    result as a (name, data) tuple.    
-    
     """
     return NBTFileReader(fileobj).read_all()
 
-def load_from_region(filename, x, y, north_direction):
+@_file_loader
-    nbt = load_region(filename, north_direction).load_chunk(x, y)
+def load_region(fileobj):
-    if nbt is None:
+    """Reads in the given file as a MCR region, and returns an object
-        return None ## return none.  I think this is who we should indicate missing chunks
+    for accessing the chunks inside."""
-        #raise IOError("No such chunk in region: (%i, %i)" % (x, y))     
+    return MCRFileReader(fileobj)
-    return nbt.read_all()
+
-  
+class CorruptNBTError(Exception):
-def load_region(filename, north_direction):
+    """An exception raised when the NBTFileReader class encounters
-    return MCRFileReader(filename, north_direction)
+    something unexpected in an NBT file."""
-  
+    pass
-  
-# compile the unpacker's into a classes
-_byte   = struct.Struct("b")
-_short  = struct.Struct(">h")
-_int    = struct.Struct(">i")
-_long   = struct.Struct(">q")
-_float  = struct.Struct(">f")
-_double = struct.Struct(">d") 
 
-_24bit_int = struct.Struct("B B B")
-_unsigned_byte = struct.Struct("B")
-_unsigned_int = struct.Struct(">I")
-_chunk_header = struct.Struct(">I B")
- 
 class NBTFileReader(object):
     """Low level class that reads the Named Binary Tag format used by Minecraft
 
     """
+    
+    # compile the unpacker's into a classes
+    _byte   = struct.Struct("b")
+    _short  = struct.Struct(">h")
+    _int    = struct.Struct(">i")
+    _long   = struct.Struct(">q")
+    _float  = struct.Struct(">f")
+    _double = struct.Struct(">d") 
+ 
     def __init__(self, fileobj, is_gzip=True):
+        """Create a NBT parsing object with the given file-like
+        object. Setting is_gzip to False parses the file as a zlib
+        stream instead."""
         if is_gzip:
             self._file = gzip.GzipFile(fileobj=fileobj, mode='rb')
         else:
@@ -75,6 +72,21 @@ class NBTFileReader(object):
             data = zlib.decompress(fileobj.read())
             self._file = StringIO.StringIO(data)
 
+        # mapping of NBT type ids to functions to read them out
+        self._read_tagmap = {
+            0: self._read_tag_end,
+            1: self._read_tag_byte,
+            2: self._read_tag_short,
+            3: self._read_tag_int,
+            4: self._read_tag_long,
+            5: self._read_tag_float,
+            6: self._read_tag_double,
+            7: self._read_tag_byte_array,
+            8: self._read_tag_string,
+            9: self._read_tag_list,
+            10:self._read_tag_compound,
+        }
+
     # These private methods read the payload only of the following types
     def _read_tag_end(self):
         # Nothing to read
@@ -82,32 +94,27 @@ class NBTFileReader(object):
 
     def _read_tag_byte(self):
         byte = self._file.read(1)
-        return _byte.unpack(byte)[0]
+        return self._byte.unpack(byte)[0]
     
     def _read_tag_short(self):
         bytes = self._file.read(2)
-        global _short
+        return self._short.unpack(bytes)[0]
-        return _short.unpack(bytes)[0]
 
     def _read_tag_int(self):
         bytes = self._file.read(4)
-        global _int
+        return self._int.unpack(bytes)[0]
-        return _int.unpack(bytes)[0]
 
     def _read_tag_long(self):
         bytes = self._file.read(8)
-        global _long
+        return self._long.unpack(bytes)[0]
-        return _long.unpack(bytes)[0]
 
     def _read_tag_float(self):
         bytes = self._file.read(4)
-        global _float
+        return self._float.unpack(bytes)[0]
-        return _float.unpack(bytes)[0]
 
     def _read_tag_double(self):
         bytes = self._file.read(8)
-        global _double
+        return self._double.unpack(bytes)[0]
-        return _double.unpack(bytes)[0]
 
     def _read_tag_byte_array(self):
         length = self._read_tag_int()
@@ -116,10 +123,8 @@ class NBTFileReader(object):
 
     def _read_tag_string(self):
         length = self._read_tag_short()
-
         # Read the string
         string = self._file.read(length)
-
         # decode it and return
         return string.decode("UTF-8")
 
@@ -127,21 +132,7 @@ class NBTFileReader(object):
         tagid = self._read_tag_byte()
         length = self._read_tag_int()
 
-        read_tagmap = {
+        read_method = self._read_tagmap[tagid]
-                0: self._read_tag_end,
-                1: self._read_tag_byte,
-                2: self._read_tag_short,
-                3: self._read_tag_int,
-                4: self._read_tag_long,
-                5: self._read_tag_float,
-                6: self._read_tag_double,
-                7: self._read_tag_byte_array,
-                8: self._read_tag_string,
-                9: self._read_tag_list,
-                10:self._read_tag_compound,
-                }
-
-        read_method = read_tagmap[tagid]
         l = []
         for _ in xrange(length):
             l.append(read_method())
@@ -158,27 +149,11 @@ class NBTFileReader(object):
                 break
 
             name = self._read_tag_string()
-            read_tagmap = {
+            payload = self._read_tagmap[tagtype]()
-                    0: self._read_tag_end,
-                    1: self._read_tag_byte,
-                    2: self._read_tag_short,
-                    3: self._read_tag_int,
-                    4: self._read_tag_long,
-                    5: self._read_tag_float,
-                    6: self._read_tag_double,
-                    7: self._read_tag_byte_array,
-                    8: self._read_tag_string,
-                    9: self._read_tag_list,
-                    10:self._read_tag_compound,
-                    }
-            payload = read_tagmap[tagtype]()
-            
             tags[name] = payload
 
         return tags
-
+    
-
-
     def read_all(self):
         """Reads the entire file and returns (name, payload)
         name is the name of the root tag, and payload is a dictionary mapping
@@ -186,179 +161,74 @@ class NBTFileReader(object):
 
         """
         # Read tag type
-        tagtype = ord(self._file.read(1))
+        try:
-        if tagtype != 10:
+            tagtype = ord(self._file.read(1))
-            raise Exception("Expected a tag compound")
+            if tagtype != 10:
-
+                raise Exception("Expected a tag compound")
-        # Read the tag name
+            
-        name = self._read_tag_string()
+            # Read the tag name
-
+            name = self._read_tag_string()
-        payload = self._read_tag_compound()
+            payload = self._read_tag_compound()
-
+            
-        return name, payload
+            return (name, payload)
+        except (struct.error, ValueError), e:
+            raise CorruptNBTError("could not parse nbt: %s" % (str(e),))
 
+class CorruptRegionError(Exception):
+    """An exception raised when the MCRFileReader class encounters an
+    error during region file parsing.
+    """
+    pass
 
 # For reference, the MCR format is outlined at
 # <http://www.minecraftwiki.net/wiki/Beta_Level_Format>
 class MCRFileReader(object):
     """A class for reading chunk region files, as introduced in the
     Beta 1.3 update. It provides functions for opening individual
-    chunks (as instances of NBTFileReader), getting chunk timestamps,
+    chunks (as (name, data) tuples), getting chunk timestamps, and for
-    and for listing chunks contained in the file."""
+    listing chunks contained in the file.
+    """
     
-    def __init__(self, filename, north_direction):
+    _table_format = struct.Struct(">1024I")
+    _chunk_header_format = struct.Struct(">I B")
+    
+    def __init__(self, fileobj):
+        """This creates a region object from the given file-like
+        object. Chances are you want to use load_region instead."""
+        self._file = fileobj
+        
+        # read in the location table
+        location_data = self._file.read(4096)
+        if not len(location_data) == 4096:
+            raise CorruptRegionError("invalid location table")
+        # read in the timestamp table
+        timestamp_data = self._file.read(4096)
+        if not len(timestamp_data) == 4096:
+            raise CorruptRegionError("invalid timestamp table")
+        
+        # turn this data into a useful list
+        self._locations = self._table_format.unpack(location_data)
+        self._timestamps = self._table_format.unpack(timestamp_data)
+    
+    def close(self):
+        """Close the region file and free any resources associated
+        with keeping it open. Using this object after closing it
+        results in undefined behaviour.
+        """
+        
+        self._file.close()
         self._file = None
-        self._filename = filename
-        self.north_direction = north_direction
-        # cache used when the entire header tables are read in get_chunks()
-        self._locations = None
-        self._timestamps = None
-        self._chunks = None
-
-    def get_north_rotations(self):
-        if self.north_direction == 'upper-left':
-            return 3
-        elif self.north_direction == 'upper-right':
-            return 2
-        elif self.north_direction == 'lower-right':
-            return 1
-        elif self.north_direction == 'lower-left':
-            return 0
-    
-    def _read_24bit_int(self):
-        """Read in a 24-bit, big-endian int, used in the chunk
-        location table."""
-        
-        ret = 0
-        bytes = self._file.read(3)
-        global _24bit_int
-        bytes = _24bit_int.unpack(bytes)
-        for i in xrange(3):
-            ret = ret << 8
-            ret += bytes[i]
-        
-        return ret
-    
-    def _read_chunk_location(self, x=None, y=None):
-        """Read and return the (offset, length) of the given chunk
-        coordinate, or None if the requested chunk doesn't exist. x
-        and y must be between 0 and 31, or None. If they are None,
-        then there will be no file seek before doing the read."""
-        
-        if x is not None and y is not None:
-            if (not x >= 0) or (not x < 32) or (not y >= 0) or (not y < 32):
-                raise ValueError("Chunk location out of range.")
-            
-            # check for a cached value
-            if self._locations:
-                return self._locations[x + y * 32]
-            
-            # go to the correct entry in the chunk location table
-            self._file.seek(4 * (x + y * 32))
-        
-        try:
-            # 3-byte offset in 4KiB sectors
-            offset_sectors = self._read_24bit_int()
-            
-            # 1-byte length in 4KiB sectors, rounded up
-            global _unsigned_byte
-            byte = self._file.read(1)
-            length_sectors = _unsigned_byte.unpack(byte)[0]
-        except (IndexError, struct.error):
-            # got a problem somewhere
-            return None
-        
-        # check for empty chunks
-        if offset_sectors == 0 or length_sectors == 0:
-            return None
-        
-        return (offset_sectors * 4096, length_sectors * 4096)
-    
-    def _read_chunk_timestamp(self, x=None, y=None):
-        """Read and return the last modification time of the given
-        chunk coordinate. x and y must be between 0 and 31, or
-        None. If they are, None, then there will be no file seek
-        before doing the read."""
-        
-        if x is not None and y is not None:
-            if (not x >= 0) or (not x < 32) or (not y >= 0) or (not y < 32):
-                raise ValueError("Chunk location out of range.")
-            
-            # check for a cached value
-            if self._timestamps:
-                return self._timestamps[x + y * 32]
-            
-            # go to the correct entry in the chunk timestamp table
-            self._file.seek(4 * (x + y * 32) + 4096)
-        
-        try:
-            bytes = self._file.read(4)
-            global _unsigned_int
-            timestamp = _unsigned_int.unpack(bytes)[0]
-        except (IndexError, struct.error):
-            return 0
-        
-        return timestamp
-    
-    def openfile(self):
-        #make sure we clean up
-        if self._file is None:
-            self._file = open(self._filename,'rb')   
-
-    def closefile(self):
-        #make sure we clean up
-        if self._file is not None:
-            self._file.close()
-            self._file =  None
 
     def get_chunks(self):    
-        """Return a list of all chunks contained in this region file,
+        """Return an iterator of all chunks contained in this region
-        as a list of (x, y) coordinate tuples. To load these chunks,
+        file, as (x, y) coordinate tuples. To load these chunks,
         provide these coordinates to load_chunk()."""
         
-        if self._chunks is not None:
-            return self._chunks
-        if self._locations is None:
-            self.get_chunk_info()       
-        self._chunks = [] 
         for x in xrange(32): 
             for y in xrange(32): 
-                if self._locations[x + y * 32] is not None:
+                if self._locations[x + y * 32] >> 8 != 0:
-                    self._chunks.append((x,y))
+                    yield (x,y)
-        return self._chunks
         
-    def get_chunk_info(self,closeFile = True):
-        """Preloads region header information."""
-        
-        if self._locations:
-            return
-        
-        self.openfile()
-
-        self._chunks = None
-        self._locations = [0]*32*32
-        self._timestamps = []
-        
-        # go to the beginning of the file
-        self._file.seek(0)        
-        
-        # read chunk location table
-        locations_index = numpy.reshape(numpy.rot90(numpy.reshape(range(32*32),
-                (32, 32)), -self.get_north_rotations()), -1)
-        for i in locations_index:
-            self._locations[i] = self._read_chunk_location()
-        
-        # read chunk timestamp table
-        timestamp_append = self._timestamps.append
-        for _ in xrange(32*32): 
-            timestamp_append(self._read_chunk_timestamp())
-        self._timestamps = numpy.reshape(numpy.rot90(numpy.reshape(
-                self._timestamps, (32,32)),self.get_north_rotations()), -1)
- 
-        if closeFile:        
-            self.closefile()
-        return
-    
     def get_chunk_timestamp(self, x, y):
         """Return the given chunk's modification time. If the given
         chunk doesn't exist, this number may be nonsense. Like
@@ -366,21 +236,16 @@ class MCRFileReader(object):
         """
         x = x % 32
         y = y % 32        
-        if self._timestamps is None:
-            self.get_chunk_info() 
         return self._timestamps[x + y * 32]   
     
-    def chunkExists(self, x, y):
+    def chunk_exists(self, x, y):
-        """Determines if a chunk exists without triggering loading of the backend data"""
+        """Determines if a chunk exists."""
         x = x % 32
         y = y % 32
-        if self._locations is None:
+        return self._locations[x + y * 32] >> 8 != 0
-            self.get_chunk_info()
-        location = self._locations[x + y * 32]
-        return location is not None        
 
-    def load_chunk(self, x, y,closeFile=True):
+    def load_chunk(self, x, y):
-        """Return a NBTFileReader instance for the given chunk, or
+        """Return a (name, data) tuple for the given chunk, or
         None if the given chunk doesn't exist in this region file. If
         you provide an x or y not between 0 and 31, it will be
         modulo'd into this range (x % 32, etc.) This is so you can
@@ -388,21 +253,21 @@ class MCRFileReader(object):
         have the chunks load out of regions properly."""
         x = x % 32
         y = y % 32
-        if self._locations is None:
-            self.get_chunk_info()   
-                    
         location = self._locations[x + y * 32]
-        if location is None:
+        offset = (location >> 8) * 4096;
+        sectors = location & 0xff;
+        
+        if offset == 0:
             return None
-
-        self.openfile()
         
         # seek to the data
-        self._file.seek(location[0])
+        self._file.seek(offset)
         
         # read in the chunk data header
-        bytes = self._file.read(5)        
+        header = self._file.read(5)
-        data_length,compression =  _chunk_header.unpack(bytes)
+        if len(header) != 5:
+            raise CorruptChunkError("chunk header is invalid")
+        data_length, compression =  self._chunk_header_format.unpack(header)
         
         # figure out the compression
         is_gzip = True
@@ -414,12 +279,13 @@ class MCRFileReader(object):
             is_gzip = False
         else:
             # unsupported!
-            raise Exception("Unsupported chunk compression type: %i" % (compression))
+            raise CorruptRegionError("unsupported chunk compression type: %i" % (compression))
+        
         # turn the rest of the data into a StringIO object
         # (using data_length - 1, as we already read 1 byte for compression)
         data = self._file.read(data_length - 1)
+        if len(data) != data_length - 1:
+            raise CorruptRegionError("chunk length is invalid")
         data = StringIO.StringIO(data)
         
-        if closeFile:        
+        return NBTFileReader(data, is_gzip=is_gzip).read_all()
-            self.closefile()        
-        return NBTFileReader(data, is_gzip=is_gzip)