minimaws: load ROMs and disks, and add a romident subcommand

scripts/minimaws/lib/auxverbs.py

@@ -5,7 +5,150 @@
 
 from . import dbaccess
 
+import codecs
+import hashlib
+import os
+import os.path
+import struct
 import sys
+import zlib
+
+
+class _Identifier(object):
+    def __init__(self, dbcurs, **kwargs):
+        super(_Identifier, self).__init__(**kwargs)
+        self.dbcurs = dbcurs
+        self.pathwidth = 0
+        self.labelwidth = 0
+        self.matches = { }
+        self.unmatched = [ ]
+
+    def processRomFile(self, path, f):
+        crc, sha1 = self.digestRom(f)
+        matched = False
+        for shortname, description, label, bad in self.dbcurs.get_rom_dumps(crc, sha1):
+            matched = True
+            self.labelwidth = max(len(label), self.labelwidth)
+            romset = self.matches.get(shortname)
+            if romset is None:
+                romset = (description, [])
+                self.matches[shortname] = romset
+            romset[1].append((path, label, bad))
+        if not matched:
+            self.unmatched.append((path, crc, sha1))
+        self.pathwidth = max(len(path), self.pathwidth)
+
+    def processChd(self, path, sha1):
+        matched = False
+        for shortname, description, label, bad in self.dbcurs.get_disk_dumps(sha1):
+            matched = True
+            self.labelwidth = max(len(label), self.labelwidth)
+            romset = self.matches.get(shortname)
+            if romset is None:
+                romset = (description, [])
+                self.matches[shortname] = romset
+            romset[1].append((path, label, bad))
+        if not matched:
+            self.unmatched.append((path, None, sha1))
+        self.pathwidth = max(len(path), self.pathwidth)
+
+    def processFile(self, path):
+        if os.path.splitext(path)[1].lower() != '.chd':
+            with open(path, mode='rb', buffering=0) as f:
+                self.processRomFile(path, f)
+        else:
+            with open(path, mode='rb') as f:
+                sha1 = self.probeChd(f)
+                if sha1 is None:
+                    f.seek(0)
+                    self.processRomFile(path, f)
+                else:
+                    self.processChd(path, sha1)
+
+    def processPath(self, path, depth=0):
+        try:
+            if not os.path.isdir(path):
+                self.processFile(path)
+            elif depth > 5:
+                sys.stderr.write('Not examining \'%s\' - maximum depth exceeded\n')
+            else:
+                for name in os.listdir(path):
+                    self.processPath(os.path.join(path, name), depth + 1)
+        except BaseException as e:
+            sys.stderr.write('Error identifying \'%s\': %s\n' % (path, e))
+
+    def printResults(self):
+        pw = self.pathwidth - (self.pathwidth % 4) + 4
+        lw = self.labelwidth - (self.labelwidth % 4) + 4
+        first = True
+        for shortname, romset in sorted(self.matches.items()):
+            if first:
+                first = False
+            else:
+                sys.stdout.write('\n')
+            sys.stdout.write('%-20s%s\n' % (shortname, romset[0]))
+            for path, label, bad in romset[1]:
+                if bad:
+                    sys.stdout.write('    %-*s= %-*s(BAD)\n' % (pw, path, lw, label))
+                else:
+                    sys.stdout.write('    %-*s= %s\n' % (pw, path, label))
+        if self.unmatched:
+            if first:
+                first = False
+            else:
+                sys.stdout.write('\n')
+            sys.stdout.write('Unmatched\n')
+            for path, crc, sha1 in self.unmatched:
+                if crc is not None:
+                    sys.stdout.write('    %-*sCRC(%08x) SHA1(%s)\n' % (pw, path, crc, sha1))
+                else:
+                    sys.stdout.write('    %-*sSHA1(%s)\n' % (pw, path, sha1))
+
+    @staticmethod
+    def iterateBlocks(f, s=65536):
+        while True:
+            buf = f.read(s)
+            if buf:
+                yield buf
+            else:
+                break
+
+    @staticmethod
+    def digestRom(f):
+        crc = zlib.crc32(bytes())
+        sha = hashlib.sha1()
+        for block in _Identifier.iterateBlocks(f):
+            crc = zlib.crc32(block, crc)
+            sha.update(block)
+        return crc & 0xffffffff, sha.hexdigest()
+
+    @staticmethod
+    def probeChd(f):
+        buf = f.read(16)
+        if (len(buf) != 16) or (buf[:8] != b'MComprHD'):
+            return None
+        headerlen, version = struct.unpack('>II', buf[8:])
+        if version == 3:
+            if headerlen != 120:
+                return None
+            sha1offs = 80
+        elif version == 4:
+            if headerlen != 108:
+                return None
+            sha1offs = 48
+        elif version == 5:
+            if headerlen != 124:
+                return None
+            sha1offs = 84
+        else:
+            return None
+        f.seek(sha1offs)
+        if f.tell() != sha1offs:
+            return None
+        buf = f.read(20)
+        if len(buf) != 20:
+            return None
+        return codecs.getencoder('hex_codec')(buf)[0].decode('ascii')
 
 
 def do_listfull(options):
@@ -68,6 +211,7 @@ def do_listbrothers(options):
     dbcurs.close()
     dbconn.close()
 
+
 def do_listaffected(options):
     dbconn = dbaccess.QueryConnection(options.database)
     dbcurs = dbconn.cursor()
@@ -81,3 +225,14 @@ def do_listaffected(options):
         sys.stderr.write('No matching systems found for \'%s\'\n' % (options.pattern, ))
     dbcurs.close()
     dbconn.close()
+
+
+def do_romident(options):
+    dbconn = dbaccess.QueryConnection(options.database)
+    dbcurs = dbconn.cursor()
+    ident = _Identifier(dbcurs)
+    for path in options.path:
+        ident.processPath(path)
+    ident.printResults()
+    dbcurs.close()
+    dbconn.close()

scripts/minimaws/lib/dbaccess.py

@@ -141,6 +141,35 @@ class SchemaQueries(object):
             '    size            INTEGER NOT NULL,\n' \
             '    FOREIGN KEY (machine) REFERENCES machine (id),\n' \
             '    FOREIGN KEY (machine, size) REFERENCES ramoption (machine, size))'
+    CREATE_ROM = \
+            'CREATE TABLE rom (\n' \
+            '    id              INTEGER PRIMARY KEY,\n' \
+            '    crc             INTEGER NOT NULL,\n' \
+            '    sha1            TEXT    NOT NULL,\n' \
+            '    UNIQUE (crc ASC, sha1 ASC))'
+    CREATE_ROMDUMP = \
+            'CREATE TABLE romdump (\n' \
+            '    machine         INTEGER NOT NULL,\n' \
+            '    rom             INTEGER NOT NULL,\n' \
+            '    name            TEXT NOT NULL,\n' \
+            '    bad             INTEGER NOT NULL,\n' \
+            '    FOREIGN KEY (machine) REFERENCES machine (id),\n' \
+            '    FOREIGN KEY (rom) REFERENCES rom (id),\n' \
+            '    UNIQUE (machine, rom, name))'
+    CREATE_DISK = \
+            'CREATE TABLE disk (\n' \
+            '    id              INTEGER PRIMARY KEY,\n' \
+            '    sha1            TEXT    NOT NULL,\n' \
+            '    UNIQUE (sha1 ASC))'
+    CREATE_DISKDUMP = \
+            'CREATE TABLE diskdump (\n' \
+            '    machine         INTEGER NOT NULL,\n' \
+            '    disk            INTEGER NOT NULL,\n' \
+            '    name            TEXT NOT NULL,\n' \
+            '    bad             INTEGER NOT NULL,\n' \
+            '    FOREIGN KEY (machine) REFERENCES machine (id),\n' \
+            '    FOREIGN KEY (disk) REFERENCES disk (id),\n' \
+            '    UNIQUE (machine, disk, name))'
 
     CREATE_TEMPORARY_DEVICEREFERENCE = 'CREATE TEMPORARY TABLE temp_devicereference (id INTEGER PRIMARY KEY, machine INTEGER NOT NULL, device TEXT NOT NULL, UNIQUE (machine, device))'
     CREATE_TEMPORARY_SLOTOPTION = 'CREATE TEMPORARY TABLE temp_slotoption (id INTEGER PRIMARY KEY, slot INTEGER NOT NULL, device TEXT NOT NULL, name TEXT NOT NULL)'
@@ -164,6 +193,10 @@ class SchemaQueries(object):
 
     INDEX_DIPSWITCH_MACHINE_ISCONFIG = 'CREATE INDEX dipswitch_machine_isconfig ON dipswitch (machine ASC, isconfig ASC)'
 
+    INDEX_ROMDUMP_ROM = 'CREATE INDEX romdump_rom ON romdump (rom ASC)'
+
+    INDEX_DISKDUMP_DISK = 'CREATE INDEX diskdump_disk ON diskdump (disk ASC)'
+
     DROP_MACHINE_ISDEVICE_SHORTNAME = 'DROP INDEX IF EXISTS machine_isdevice_shortname'
     DROP_MACHINE_ISDEVICE_DESCRIPTION = 'DROP INDEX IF EXISTS machine_isdevice_description'
     DROP_MACHINE_RUNNABLE_SHORTNAME = 'DROP INDEX IF EXISTS machine_runnable_shortname'
@@ -178,6 +211,10 @@ class SchemaQueries(object):
 
     DROP_DIPSWITCH_MACHINE_ISCONFIG = 'DROP INDEX IF EXISTS dipswitch_machine_isconfig'
 
+    DROP_ROMDUMP_ROM = 'DROP INDEX IF EXISTS romdump_rom'
+
+    DROP_DISKDUMP_DISK = 'DROP INDEX IF EXISTS diskdump_disk'
+
     CREATE_TABLES = (
             CREATE_FEATURETYPE,
             CREATE_SOURCEFILE,
@@ -196,7 +233,11 @@ class SchemaQueries(object):
             CREATE_SLOTOPTION,
             CREATE_SLOTDEFAULT,
             CREATE_RAMOPTION,
-            CREATE_RAMDEFAULT)
+            CREATE_RAMDEFAULT,
+            CREATE_ROM,
+            CREATE_ROMDUMP,
+            CREATE_DISK,
+            CREATE_DISKDUMP)
 
     CREATE_TEMPORARY_TABLES = (
             CREATE_TEMPORARY_DEVICEREFERENCE,
@@ -212,7 +253,9 @@ class SchemaQueries(object):
             INDEX_SYSTEM_MANUFACTURER,
             INDEX_ROMOF_PARENT,
             INDEX_CLONEOF_PARENT,
-            INDEX_DIPSWITCH_MACHINE_ISCONFIG)
+            INDEX_DIPSWITCH_MACHINE_ISCONFIG,
+            INDEX_ROMDUMP_ROM,
+            INDEX_DISKDUMP_DISK)
 
     DROP_INDEXES = (
             DROP_MACHINE_ISDEVICE_SHORTNAME,
@@ -223,7 +266,9 @@ class SchemaQueries(object):
             DROP_SYSTEM_MANUFACTURER,
             DROP_ROMOF_PARENT,
             DROP_CLONEOF_PARENT,
-            DROP_DIPSWITCH_MACHINE_ISCONFIG)
+            DROP_DIPSWITCH_MACHINE_ISCONFIG,
+            DROP_ROMDUMP_ROM,
+            DROP_DISKDUMP_DISK)
 
 
 class UpdateQueries(object):
@@ -242,6 +287,10 @@ class UpdateQueries(object):
     ADD_SLOT = 'INSERT INTO slot (machine, name) VALUES (?, ?)'
     ADD_RAMOPTION = 'INSERT INTO ramoption (machine, size, name) VALUES (?, ?, ?)'
     ADD_RAMDEFAULT = 'INSERT INTO ramdefault (machine, size) VALUES (?, ?)'
+    ADD_ROM = 'INSERT OR IGNORE INTO rom (crc, sha1) VALUES (?, ?)'
+    ADD_ROMDUMP = 'INSERT OR IGNORE INTO romdump (machine, rom, name, bad) SELECT ?, id, ?, ? FROM rom WHERE crc = ? AND sha1 = ?'
+    ADD_DISK = 'INSERT OR IGNORE INTO disk (sha1) VALUES (?)'
+    ADD_DISKDUMP = 'INSERT OR IGNORE INTO diskdump (machine, disk, name, bad) SELECT ?, id, ?, ? FROM disk WHERE sha1 = ?'
 
     ADD_TEMPORARY_DEVICEREFERENCE = 'INSERT OR IGNORE INTO temp_devicereference (machine, device) VALUES (?, ?)'
     ADD_TEMPORARY_SLOTOPTION = 'INSERT INTO temp_slotoption (slot, device, name) VALUES (?, ?, ?)'
@@ -458,6 +507,20 @@ class QueryCursor(object):
                 'ORDER BY ramoption.size',
                 (machine, ))
 
+    def get_rom_dumps(self, crc, sha1):
+        return self.dbcurs.execute(
+                'SELECT machine.shortname AS shortname, machine.description AS description, romdump.name AS label, romdump.bad AS bad ' \
+                'FROM romdump LEFT JOIN machine ON romdump.machine = machine.id ' \
+                'WHERE romdump.rom = (SELECT id FROM rom WHERE crc = ? AND sha1 = ?)',
+                (crc, sha1))
+
+    def get_disk_dumps(self, sha1):
+        return self.dbcurs.execute(
+                'SELECT machine.shortname AS shortname, machine.description AS description, diskdump.name AS label, diskdump.bad AS bad ' \
+                'FROM diskdump LEFT JOIN machine ON diskdump.machine = machine.id ' \
+                'WHERE diskdump.disk = (SELECT id FROM disk WHERE sha1 = ?)',
+                (sha1, ))
+
 
 class UpdateCursor(object):
     def __init__(self, dbconn, **kwargs):
@@ -536,6 +599,22 @@ class UpdateCursor(object):
         self.dbcurs.execute(UpdateQueries.ADD_RAMDEFAULT, (machine, size))
         return self.dbcurs.lastrowid
 
+    def add_rom(self, crc, sha1):
+        self.dbcurs.execute(UpdateQueries.ADD_ROM, (crc, sha1))
+        return self.dbcurs.lastrowid
+
+    def add_romdump(self, machine, name, crc, sha1, bad):
+        self.dbcurs.execute(UpdateQueries.ADD_ROMDUMP, (machine, name, 1 if bad else 0, crc, sha1))
+        return self.dbcurs.lastrowid
+
+    def add_disk(self, sha1):
+        self.dbcurs.execute(UpdateQueries.ADD_DISK, (sha1, ))
+        return self.dbcurs.lastrowid
+
+    def add_diskdump(self, machine, name, sha1, bad):
+        self.dbcurs.execute(UpdateQueries.ADD_DISKDUMP, (machine, name, 1 if bad else 0, sha1))
+        return self.dbcurs.lastrowid
+
 
 class QueryConnection(object):
     def __init__(self, database, **kwargs):

scripts/minimaws/lib/lxparse.py

@@ -206,6 +206,22 @@ class MachineHandler(ElementHandler):
                 status = 0 if 'status' not in attrs else 2 if attrs['status'] == 'unemulated' else 1
                 overall = status if 'overall' not in attrs else 2 if attrs['overall'] == 'unemulated' else 1
                 self.dbcurs.add_feature(self.id, attrs['type'], status, overall)
+            elif name == 'rom':
+                crc = attrs.get('crc')
+                sha1 = attrs.get('sha1')
+                if (crc is not None) and (sha1 is not None):
+                    crc = int(crc, 16)
+                    sha1 = sha1.lower()
+                    self.dbcurs.add_rom(crc, sha1)
+                    status = attrs.get('status', 'good')
+                    self.dbcurs.add_romdump(self.id, attrs['name'], crc, sha1, status != 'good')
+            elif name == 'disk':
+                sha1 = attrs.get('sha1')
+                if sha1 is not None:
+                    sha1 = sha1.lower()
+                    self.dbcurs.add_disk(sha1)
+                    status = attrs.get('status', 'good')
+                    self.dbcurs.add_diskdump(self.id, attrs['name'], sha1, status != 'good')
             self.setChildHandler(name, attrs, self.IGNORE)
 
     def endChildHandler(self, name, handler):

scripts/minimaws/minimaws.py

@@ -26,6 +26,11 @@
 ## $ python minimaws.py listclones "unkch*"
 ## $ python minimaws.py listbrothers superx
 ##
+## The romident command does not support archives, but it's far faster
+## than using MAME as it has optimised indexes:
+##
+## $ python minimaws.py romident 27c64.bin dump-dir
+##
 ## One more sophisticated query command is provided that MAME has no
 ## equivalent for.  The listaffected command shows all runnable machines
 ## that reference devices defined in specified source files:
@@ -100,6 +105,9 @@ if __name__ == '__main__':
     subparser = subparsers.add_parser('listaffected', help='show drivers affected by source change(s)')
     subparser.add_argument('pattern', nargs='+', metavar='<pat>', help='source file glob pattern')
 
+    subparser = subparsers.add_parser('romident', help='identify ROM dump(s)')
+    subparser.add_argument('path', nargs='+', metavar='<path>', help='ROM dump file/directory path')
+
     subparser = subparsers.add_parser('serve', help='serve over HTTP')
     subparser.add_argument('--port', metavar='<port>', default=8080, type=int, help='server TCP port')
     subparser.add_argument('--host', metavar='<host>', default='', help='server TCP hostname')
@@ -120,6 +128,8 @@ if __name__ == '__main__':
         lib.auxverbs.do_listbrothers(options)
     elif options.command == 'listaffected':
         lib.auxverbs.do_listaffected(options)
+    elif options.command == 'romident':
+        lib.auxverbs.do_romident(options)
     elif options.command == 'serve':
         lib.wsgiserve.run_server(options)
     elif options.command == 'load':