- Rename File class to FileType and add support for a filetype parser

- Add support for parsing Applesoft basic files
- Add a new File class that receives the contents of a DOS 3.3 file
  (also parsed, if applicable)
- Some minor bugfixes

src/apple2disk/applesoft.py

@@ -0,0 +1,156 @@
+import bitstring
+
+TOKENS = {
+    0x80: 'END',
+    0x81: 'FOR',
+    0x82: 'NEXT',
+    0x83: 'DATA',
+    0x84: 'INPUT',
+    0x85: 'DEL',
+    0x86: 'DIM',
+    0x87: 'READ',
+    0x88: 'GR',
+    0x89: 'TEXT',
+    0x8A: 'PR #',
+    0x8B: 'IN #',
+    0x8C: 'CALL',
+    0x8D: 'PLOT',
+    0x8E: 'HLIN',
+    0x8F: 'VLIN',
+    0x90: 'HGR2',
+    0x91: 'HGR',
+    0x92: 'HCOLOR=',
+    0x93: 'HPLOT',
+    0x94: 'DRAW',
+    0x95: 'XDRAW',
+    0x96: 'HTAB',
+    0x97: 'HOME',
+    0x98: 'ROT=',
+    0x99: 'SCALE=',
+    0x9A: 'SHLOAD',
+    0x9B: 'TRACE',
+    0x9C: 'NOTRACE',
+    0x9D: 'NORMAL',
+    0x9E: 'INVERSE',
+    0x9F: 'FLASH',
+    0xA0: 'COLOR=',
+    0xA1: 'POP',
+    0xA2: 'VTAB',
+    0xA3: 'HIMEM:',
+    0xA4: 'LOMEM:',
+    0xA5: 'ONERR',
+    0xA6: 'RESUME',
+    0xA7: 'RECALL',
+    0xA8: 'STORE',
+    0xA9: 'SPEED=',
+    0xAA: 'LET',
+    0xAB: 'GOTO',
+    0xAC: 'RUN',
+    0xAD: 'IF',
+    0xAE: 'RESTORE',
+    0xAF: '&',
+    0xB0: 'GOSUB',
+    0xB1: 'RETURN',
+    0xB2: 'REM',
+    0xB3: 'STOP',
+    0xB4: 'ON',
+    0xB5: 'WAIT',
+    0xB6: 'LOAD',
+    0xB7: 'SAVE',
+    0xB8: 'DEF FN',
+    0xB9: 'POKE',
+    0xBA: 'PRINT',
+    0xBB: 'CONT',
+    0xBC: 'LIST',
+    0xBD: 'CLEAR',
+    0xBE: 'GET',
+    0xBF: 'NEW',
+    0xC0: 'TAB',
+    0xC1: 'TO',
+    0xC2: 'FN',
+    0xC3: 'SPC(',
+    0xC4: 'THEN',
+    0xC5: 'AT',
+    0xC6: 'NOT',
+    0xC7: 'STEP',
+    0xC8: '+',
+    0xC9: '-',
+    0xCA: '*',
+    0xCB: '/',
+    0xCC: ';',
+    0xCD: 'AND',
+    0xCE: 'OR',
+    0xCF: '>',
+    0xD0: '=',
+    0xD1: '<',
+    0xD2: 'SGN',
+    0xD3: 'INT',
+    0xD4: 'ABS',
+    0xD5: 'USR',
+    0xD6: 'FRE',
+    0xD7: 'SCRN (',
+    0xD8: 'PDL',
+    0xD9: 'POS',
+    0xDA: 'SQR',
+    0xDB: 'RND',
+    0xDC: 'LOG',
+    0xDD: 'EXP',
+    0xDE: 'COS',
+    0xDF: 'SIN',
+    0xE0: 'TAN',
+    0xE1: 'ATN',
+    0xE2: 'PEEK',
+    0xE3: 'LEN',
+    0xE4: 'STR$',
+    0xE5: 'VAL',
+    0xE6: 'ASC',
+    0xE7: 'CHR$',
+    0xE8: 'LEFT$',
+    0xE9: 'RIGHT$',
+    0xEA: 'MID$'
+}
+
+# TODO: report anomaly if an unknown token is used
+
+class AppleSoft(object):
+    def __init__(self, data):
+        data = bitstring.ConstBitStream(data)
+
+        self.length = data.read('uintle:16')
+
+        self.lines = {}
+        last_line_number = -1
+        last_memory = 0x801
+        while data:
+            next_memory, line_number = data.readlist('uintle:16, uintle:16')
+            if not next_memory:
+                break
+
+            line = []
+            bytes_read = 4
+            while True:
+                token = data.read('uint:8')
+                bytes_read += 1
+                if token == 0:
+                    self.lines[line_number] = ''.join(line)
+                    break
+
+                if token in TOKENS:
+                    line.append(' ' + TOKENS[token] + ' ')
+                else:
+                    line.append(chr(token))
+
+            if last_memory + bytes_read != next_memory:
+                print "%x + %x == %x != %x (gap %d)" % (last_memory, bytes_read, last_memory + bytes_read, next_memory, next_memory - last_memory - bytes_read)
+
+            if line_number <= last_line_number:
+                print "%d <= %d: %s" % (line_number, last_line_number, ''.join(line))
+
+            print "%d %s" % (line_number, ''.join(line))
+            last_line_number = line_number
+            last_memory = next_memory
+
+    def __str__(self):
+        return '\n'.join('%s %s' % (num, line) for (num, line) in sorted(self.lines.items()))
+
+

src/apple2disk/dos33disk.py

@@ -1,20 +1,22 @@
+import applesoft
 import bitstring
 import disk as disklib
 import string
 
 PRINTABLE = set(string.letters + string.digits + string.punctuation + ' ')
 
-class File(object):
-    def __init__(self, short_type, long_type):
+class FileType(object):
+    def __init__(self, short_type, long_type, parser=None):
         self.short_type = short_type
         self.long_type = long_type
+        self.parser = parser
 
 FILE_TYPES = {
-    0x00: File('T', 'TEXT'),
-    0x01: File('I', 'INTEGER BASIC'),
+    0x00: FileType('T', 'TEXT'),
+    0x01: FileType('I', 'INTEGER BASIC'),
     # TODO: add handler for parsing file content
-    0x02: File('A', 'APPLESOFT BASIC'),
-    0x04: File('B', 'BINARY'),
+    0x02: FileType('A', 'APPLESOFT BASIC', applesoft.AppleSoft),
+    0x04: FileType('B', 'BINARY'),
     # TODO: others
 }
 
@@ -152,9 +154,19 @@ class Dos33Disk(disklib.Disk):
         # TODO: why does DOS 3.3 sometimes display e.g. volume 254 when the VTOC says 178
         self.volume = self.vtoc.volume
 
+        # List of stripped filenames in catalog order
+        self.filenames = []
+
+        # Maps stripped filenames to CatalogEntry objects
+        self.catalog = {}
+
         self.ReadCatalog()
+
+        # Maps stripped filename to File() object
+        self.files = {}
         for catalog_entry in self.catalog.itervalues():
-            self.ReadCatalogEntry(catalog_entry)
+            # TODO: last character has special meaning for deleted files and may legitimately be whitespace.  Could collide with a non-deleted file of the same stripped name
+            self.files[catalog_entry.FileName().rstrip()] = self.ReadCatalogEntry(catalog_entry)
 
     def _ReadVTOC(self):
         return VTOCSector.fromSector(self.ReadSector(0x11, 0x0))
@@ -218,14 +230,14 @@ class Dos33Disk(disklib.Disk):
             fds = FileDataSector.fromSector(self.ReadSector(t, s), entry.FileName())
             contents.append(fds.data)
 
-        return contents
+        return File(entry, contents)
 
     def __str__(self):
         catalog = ['DISK VOLUME %d\n' % self.volume]
         for filename in self.catalog:
             entry = self.files[filename]
             try:
-                file_type = FILE_TYPES[entry.file_type][0]
+                file_type = FILE_TYPES[entry.file_type].short_type
             except KeyError:
                 print "%s has unknown file type %02x" % (entry.FileName(), entry.file_type)
                 file_type = '?'
@@ -243,8 +255,8 @@ class CatalogEntry(object):
     def __init__(self, track, sector, file_type, file_name, length):
         self.track = track
         self.sector = sector
-        self.file_type = file_type & 0x7f
-        self.locked = file_type & 0x80
+        self.file_type = FILE_TYPES[file_type & 0x7f]
+        self.locked = bool(file_type & 0x80)
         self.file_name = file_name
         self.length = length
         # TODO: handle deleted files (track = 0xff, original track in file_name[0x20])
@@ -253,4 +265,19 @@ class CatalogEntry(object):
         return '%s' % ''.join([chr(ord(b) & 0x7f) for b in self.file_name])
 
     def __str__(self):
-        return "Track $%02x Sector $%02x Type %s Name: %s Length: %d" % (self.track, self.sector, FILE_TYPES[self.file_type], self.FileName(), self.length)
+        type_string = self.file_type.long_type
+        if self.locked:
+            type_string += ' (LOCKED)'
+        return "Track $%02x Sector $%02x Type %s Name: %s Length: %d" % (self.track, self.sector, type_string, self.FileName(), self.length)
+
+
+class File(object):
+    def __init__(self, catalog_entry, contents):
+        self.catalog_entry = catalog_entry
+
+        self.contents = contents
+        parser = catalog_entry.file_type.parser
+        if parser:
+            self.parsed_contents = parser(contents)
+        else:
+            self.parsed_contents = None

src/apple2disk/process.py

@@ -24,6 +24,14 @@ def main():
             try:
                 img = dos33disk.Dos33Disk.Taste(img)
                 print "%s is a DOS 3.3 disk, volume %d" % (f, img.volume)
+
+                for fn in img.filenames:
+                    f = img.files[fn]
+
+                    print f.catalog_entry
+                    if f.parsed_contents:
+                        print f.parsed_contents
+
             except IOError:
                 pass
             except AssertionError:
@@ -34,6 +42,7 @@ def main():
             for ts, data in sorted(img.sectors.iteritems()):
                 print data
 
+
     # Group disks by hash of RWTS sector
     rwts_hashes = {}
     for f, d in disks.iteritems():