Home | History | Annotate | Download | only in generator
      1 #!/usr/bin/python
      2 
      3 '''Generate header file for nanopb from a ProtoBuf FileDescriptorSet.'''
      4 nanopb_version = "nanopb-0.2.8-dev"
      5 
      6 import sys
      7 
      8 try:
      9     # Add some dummy imports to keep packaging tools happy.
     10     import google, distutils.util # bbfreeze seems to need these
     11     import pkg_resources # pyinstaller / protobuf 2.5 seem to need these
     12 except:
     13     # Don't care, we will error out later if it is actually important.
     14     pass
     15 
     16 try:
     17     import google.protobuf.text_format as text_format
     18     import google.protobuf.descriptor_pb2 as descriptor
     19 except:
     20     sys.stderr.write('''
     21          *************************************************************
     22          *** Could not import the Google protobuf Python libraries ***
     23          *** Try installing package 'python-protobuf' or similar.  ***
     24          *************************************************************
     25     ''' + '\n')
     26     raise
     27 
     28 try:
     29     import proto.nanopb_pb2 as nanopb_pb2
     30     import proto.plugin_pb2 as plugin_pb2
     31 except:
     32     sys.stderr.write('''
     33          ********************************************************************
     34          *** Failed to import the protocol definitions for generator.     ***
     35          *** You have to run 'make' in the nanopb/generator/proto folder. ***
     36          ********************************************************************
     37     ''' + '\n')
     38     raise
     39 
     40 # ---------------------------------------------------------------------------
     41 #                     Generation of single fields
     42 # ---------------------------------------------------------------------------
     43 
     44 import time
     45 import os.path
     46 
     47 # Values are tuple (c type, pb type, encoded size)
     48 FieldD = descriptor.FieldDescriptorProto
     49 datatypes = {
     50     FieldD.TYPE_BOOL:       ('bool',     'BOOL',        1),
     51     FieldD.TYPE_DOUBLE:     ('double',   'DOUBLE',      8),
     52     FieldD.TYPE_FIXED32:    ('uint32_t', 'FIXED32',     4),
     53     FieldD.TYPE_FIXED64:    ('uint64_t', 'FIXED64',     8),
     54     FieldD.TYPE_FLOAT:      ('float',    'FLOAT',       4),
     55     FieldD.TYPE_INT32:      ('int32_t',  'INT32',      10),
     56     FieldD.TYPE_INT64:      ('int64_t',  'INT64',      10),
     57     FieldD.TYPE_SFIXED32:   ('int32_t',  'SFIXED32',    4),
     58     FieldD.TYPE_SFIXED64:   ('int64_t',  'SFIXED64',    8),
     59     FieldD.TYPE_SINT32:     ('int32_t',  'SINT32',      5),
     60     FieldD.TYPE_SINT64:     ('int64_t',  'SINT64',     10),
     61     FieldD.TYPE_UINT32:     ('uint32_t', 'UINT32',      5),
     62     FieldD.TYPE_UINT64:     ('uint64_t', 'UINT64',     10)
     63 }
     64 
     65 class Names:
     66     '''Keeps a set of nested names and formats them to C identifier.'''
     67     def __init__(self, parts = ()):
     68         if isinstance(parts, Names):
     69             parts = parts.parts
     70         self.parts = tuple(parts)
     71     
     72     def __str__(self):
     73         return '_'.join(self.parts)
     74 
     75     def __add__(self, other):
     76         if isinstance(other, (str, unicode)):
     77             return Names(self.parts + (other,))
     78         elif isinstance(other, tuple):
     79             return Names(self.parts + other)
     80         else:
     81             raise ValueError("Name parts should be of type str")
     82     
     83     def __eq__(self, other):
     84         return isinstance(other, Names) and self.parts == other.parts
     85     
     86 def names_from_type_name(type_name):
     87     '''Parse Names() from FieldDescriptorProto type_name'''
     88     if type_name[0] != '.':
     89         raise NotImplementedError("Lookup of non-absolute type names is not supported")
     90     return Names(type_name[1:].split('.'))
     91 
     92 def varint_max_size(max_value):
     93     '''Returns the maximum number of bytes a varint can take when encoded.'''
     94     for i in range(1, 11):
     95         if (max_value >> (i * 7)) == 0:
     96             return i
     97     raise ValueError("Value too large for varint: " + str(max_value))
     98 
     99 assert varint_max_size(0) == 1
    100 assert varint_max_size(127) == 1
    101 assert varint_max_size(128) == 2
    102 
    103 class EncodedSize:
    104     '''Class used to represent the encoded size of a field or a message.
    105     Consists of a combination of symbolic sizes and integer sizes.'''
    106     def __init__(self, value = 0, symbols = []):
    107         if isinstance(value, (str, Names)):
    108             symbols = [str(value)]
    109             value = 0
    110         self.value = value
    111         self.symbols = symbols
    112     
    113     def __add__(self, other):
    114         if isinstance(other, (int, long)):
    115             return EncodedSize(self.value + other, self.symbols)
    116         elif isinstance(other, (str, Names)):
    117             return EncodedSize(self.value, self.symbols + [str(other)])
    118         elif isinstance(other, EncodedSize):
    119             return EncodedSize(self.value + other.value, self.symbols + other.symbols)
    120         else:
    121             raise ValueError("Cannot add size: " + repr(other))
    122 
    123     def __mul__(self, other):
    124         if isinstance(other, (int, long)):
    125             return EncodedSize(self.value * other, [str(other) + '*' + s for s in self.symbols])
    126         else:
    127             raise ValueError("Cannot multiply size: " + repr(other))
    128 
    129     def __str__(self):
    130         if not self.symbols:
    131             return str(self.value)
    132         else:
    133             return '(' + str(self.value) + ' + ' + ' + '.join(self.symbols) + ')'
    134 
    135     def upperlimit(self):
    136         if not self.symbols:
    137             return self.value
    138         else:
    139             return 2**32 - 1
    140 
    141 class Enum:
    142     def __init__(self, names, desc, enum_options):
    143         '''desc is EnumDescriptorProto'''
    144         
    145         self.options = enum_options
    146         self.names = names + desc.name
    147         
    148         if enum_options.long_names:
    149             self.values = [(self.names + x.name, x.number) for x in desc.value]            
    150         else:
    151             self.values = [(names + x.name, x.number) for x in desc.value] 
    152         
    153         self.value_longnames = [self.names + x.name for x in desc.value]
    154     
    155     def __str__(self):
    156         result = 'typedef enum _%s {\n' % self.names
    157         result += ',\n'.join(["    %s = %d" % x for x in self.values])
    158         result += '\n} %s;' % self.names
    159         return result
    160 
    161 class Field:
    162     def __init__(self, struct_name, desc, field_options):
    163         '''desc is FieldDescriptorProto'''
    164         self.tag = desc.number
    165         self.struct_name = struct_name
    166         self.name = desc.name
    167         self.default = None
    168         self.max_size = None
    169         self.max_count = None
    170         self.array_decl = ""
    171         self.enc_size = None
    172         self.ctype = None
    173         
    174         # Parse field options
    175         if field_options.HasField("max_size"):
    176             self.max_size = field_options.max_size
    177         
    178         if field_options.HasField("max_count"):
    179             self.max_count = field_options.max_count
    180         
    181         if desc.HasField('default_value'):
    182             self.default = desc.default_value
    183            
    184         # Check field rules, i.e. required/optional/repeated.
    185         can_be_static = True
    186         if desc.label == FieldD.LABEL_REQUIRED:
    187             self.rules = 'REQUIRED'
    188         elif desc.label == FieldD.LABEL_OPTIONAL:
    189             self.rules = 'OPTIONAL'
    190         elif desc.label == FieldD.LABEL_REPEATED:
    191             self.rules = 'REPEATED'
    192             if self.max_count is None:
    193                 can_be_static = False
    194             else:
    195                 self.array_decl = '[%d]' % self.max_count
    196         else:
    197             raise NotImplementedError(desc.label)
    198         
    199         # Check if the field can be implemented with static allocation
    200         # i.e. whether the data size is known.
    201         if desc.type == FieldD.TYPE_STRING and self.max_size is None:
    202             can_be_static = False
    203         
    204         if desc.type == FieldD.TYPE_BYTES and self.max_size is None:
    205             can_be_static = False
    206         
    207         # Decide how the field data will be allocated
    208         if field_options.type == nanopb_pb2.FT_DEFAULT:
    209             if can_be_static:
    210                 field_options.type = nanopb_pb2.FT_STATIC
    211             else:
    212                 field_options.type = nanopb_pb2.FT_CALLBACK
    213         
    214         if field_options.type == nanopb_pb2.FT_STATIC and not can_be_static:
    215             raise Exception("Field %s is defined as static, but max_size or "
    216                             "max_count is not given." % self.name)
    217         
    218         if field_options.type == nanopb_pb2.FT_STATIC:
    219             self.allocation = 'STATIC'
    220         elif field_options.type == nanopb_pb2.FT_POINTER:
    221             self.allocation = 'POINTER'
    222         elif field_options.type == nanopb_pb2.FT_CALLBACK:
    223             self.allocation = 'CALLBACK'
    224         else:
    225             raise NotImplementedError(field_options.type)
    226         
    227         # Decide the C data type to use in the struct.
    228         if datatypes.has_key(desc.type):
    229             self.ctype, self.pbtype, self.enc_size = datatypes[desc.type]
    230         elif desc.type == FieldD.TYPE_ENUM:
    231             self.pbtype = 'ENUM'
    232             self.ctype = names_from_type_name(desc.type_name)
    233             if self.default is not None:
    234                 self.default = self.ctype + self.default
    235             self.enc_size = 5 # protoc rejects enum values > 32 bits
    236         elif desc.type == FieldD.TYPE_STRING:
    237             self.pbtype = 'STRING'
    238             self.ctype = 'char'
    239             if self.allocation == 'STATIC':
    240                 self.ctype = 'char'
    241                 self.array_decl += '[%d]' % self.max_size
    242                 self.enc_size = varint_max_size(self.max_size) + self.max_size
    243         elif desc.type == FieldD.TYPE_BYTES:
    244             self.pbtype = 'BYTES'
    245             if self.allocation == 'STATIC':
    246                 self.ctype = self.struct_name + self.name + 't'
    247                 self.enc_size = varint_max_size(self.max_size) + self.max_size
    248             elif self.allocation == 'POINTER':
    249                 self.ctype = 'pb_bytes_array_t'
    250         elif desc.type == FieldD.TYPE_MESSAGE:
    251             self.pbtype = 'MESSAGE'
    252             self.ctype = self.submsgname = names_from_type_name(desc.type_name)
    253             self.enc_size = None # Needs to be filled in after the message type is available
    254         else:
    255             raise NotImplementedError(desc.type)
    256         
    257     def __cmp__(self, other):
    258         return cmp(self.tag, other.tag)
    259     
    260     def __str__(self):
    261         result = ''
    262         if self.allocation == 'POINTER':
    263             if self.rules == 'REPEATED':
    264                 result += '    size_t ' + self.name + '_count;\n'
    265             
    266             if self.pbtype == 'MESSAGE':
    267                 # Use struct definition, so recursive submessages are possible
    268                 result += '    struct _%s *%s;' % (self.ctype, self.name)
    269             elif self.rules == 'REPEATED' and self.pbtype in ['STRING', 'BYTES']:
    270                 # String/bytes arrays need to be defined as pointers to pointers
    271                 result += '    %s **%s;' % (self.ctype, self.name)
    272             else:
    273                 result += '    %s *%s;' % (self.ctype, self.name)
    274         elif self.allocation == 'CALLBACK':
    275             result += '    pb_callback_t %s;' % self.name
    276         else:
    277             if self.rules == 'OPTIONAL' and self.allocation == 'STATIC':
    278                 result += '    bool has_' + self.name + ';\n'
    279             elif self.rules == 'REPEATED' and self.allocation == 'STATIC':
    280                 result += '    size_t ' + self.name + '_count;\n'
    281             result += '    %s %s%s;' % (self.ctype, self.name, self.array_decl)
    282         return result
    283     
    284     def types(self):
    285         '''Return definitions for any special types this field might need.'''
    286         if self.pbtype == 'BYTES' and self.allocation == 'STATIC':
    287             result = 'typedef struct {\n'
    288             result += '    size_t size;\n'
    289             result += '    uint8_t bytes[%d];\n' % self.max_size
    290             result += '} %s;\n' % self.ctype
    291         else:
    292             result = None
    293         return result
    294     
    295     def default_decl(self, declaration_only = False):
    296         '''Return definition for this field's default value.'''
    297         if self.default is None:
    298             return None
    299 
    300         ctype, default = self.ctype, self.default
    301         array_decl = ''
    302         
    303         if self.pbtype == 'STRING':
    304             if self.allocation != 'STATIC':
    305                 return None # Not implemented
    306         
    307             array_decl = '[%d]' % self.max_size
    308             default = str(self.default).encode('string_escape')
    309             default = default.replace('"', '\\"')
    310             default = '"' + default + '"'
    311         elif self.pbtype == 'BYTES':
    312             if self.allocation != 'STATIC':
    313                 return None # Not implemented
    314 
    315             data = self.default.decode('string_escape')
    316             data = ['0x%02x' % ord(c) for c in data]
    317             default = '{%d, {%s}}' % (len(data), ','.join(data))
    318         elif self.pbtype in ['FIXED32', 'UINT32']:
    319             default += 'u'
    320         elif self.pbtype in ['FIXED64', 'UINT64']:
    321             default += 'ull'
    322         elif self.pbtype in ['SFIXED64', 'INT64']:
    323             default += 'll'
    324         
    325         if declaration_only:
    326             return 'extern const %s %s_default%s;' % (ctype, self.struct_name + self.name, array_decl)
    327         else:
    328             return 'const %s %s_default%s = %s;' % (ctype, self.struct_name + self.name, array_decl, default)
    329     
    330     def tags(self):
    331         '''Return the #define for the tag number of this field.'''
    332         identifier = '%s_%s_tag' % (self.struct_name, self.name)
    333         return '#define %-40s %d\n' % (identifier, self.tag)
    334     
    335     def pb_field_t(self, prev_field_name):
    336         '''Return the pb_field_t initializer to use in the constant array.
    337         prev_field_name is the name of the previous field or None.
    338         '''
    339         result  = '    PB_FIELD2(%3d, ' % self.tag
    340         result += '%-8s, ' % self.pbtype
    341         result += '%s, ' % self.rules
    342         result += '%-8s, ' % self.allocation
    343         result += '%s, ' % ("FIRST" if not prev_field_name else "OTHER")
    344         result += '%s, ' % self.struct_name
    345         result += '%s, ' % self.name
    346         result += '%s, ' % (prev_field_name or self.name)
    347         
    348         if self.pbtype == 'MESSAGE':
    349             result += '&%s_fields)' % self.submsgname
    350         elif self.default is None:
    351             result += '0)'
    352         elif self.pbtype in ['BYTES', 'STRING'] and self.allocation != 'STATIC':
    353             result += '0)' # Arbitrary size default values not implemented
    354         elif self.rules == 'OPTEXT':
    355             result += '0)' # Default value for extensions is not implemented
    356         else:
    357             result += '&%s_default)' % (self.struct_name + self.name)
    358         
    359         return result
    360     
    361     def largest_field_value(self):
    362         '''Determine if this field needs 16bit or 32bit pb_field_t structure to compile properly.
    363         Returns numeric value or a C-expression for assert.'''
    364         if self.pbtype == 'MESSAGE':
    365             if self.rules == 'REPEATED' and self.allocation == 'STATIC':
    366                 return 'pb_membersize(%s, %s[0])' % (self.struct_name, self.name)
    367             else:
    368                 return 'pb_membersize(%s, %s)' % (self.struct_name, self.name)
    369 
    370         return max(self.tag, self.max_size, self.max_count)        
    371 
    372     def encoded_size(self, allmsgs):
    373         '''Return the maximum size that this field can take when encoded,
    374         including the field tag. If the size cannot be determined, returns
    375         None.'''
    376         
    377         if self.allocation != 'STATIC':
    378             return None
    379         
    380         if self.pbtype == 'MESSAGE':
    381             for msg in allmsgs:
    382                 if msg.name == self.submsgname:
    383                     encsize = msg.encoded_size(allmsgs)
    384                     if encsize is None:
    385                         return None # Submessage size is indeterminate
    386                         
    387                     # Include submessage length prefix
    388                     encsize += varint_max_size(encsize.upperlimit())
    389                     break
    390             else:
    391                 # Submessage cannot be found, this currently occurs when
    392                 # the submessage type is defined in a different file.
    393                 # Instead of direct numeric value, reference the size that
    394                 # has been #defined in the other file.
    395                 encsize = EncodedSize(self.submsgname + 'size')
    396 
    397                 # We will have to make a conservative assumption on the length
    398                 # prefix size, though.
    399                 encsize += 5
    400 
    401         elif self.enc_size is None:
    402             raise RuntimeError("Could not determine encoded size for %s.%s"
    403                                % (self.struct_name, self.name))
    404         else:
    405             encsize = EncodedSize(self.enc_size)
    406         
    407         encsize += varint_max_size(self.tag << 3) # Tag + wire type
    408 
    409         if self.rules == 'REPEATED':
    410             # Decoders must be always able to handle unpacked arrays.
    411             # Therefore we have to reserve space for it, even though
    412             # we emit packed arrays ourselves.
    413             encsize *= self.max_count
    414         
    415         return encsize
    416 
    417 
    418 class ExtensionRange(Field):
    419     def __init__(self, struct_name, range_start, field_options):
    420         '''Implements a special pb_extension_t* field in an extensible message
    421         structure. The range_start signifies the index at which the extensions
    422         start. Not necessarily all tags above this are extensions, it is merely
    423         a speed optimization.
    424         '''
    425         self.tag = range_start
    426         self.struct_name = struct_name
    427         self.name = 'extensions'
    428         self.pbtype = 'EXTENSION'
    429         self.rules = 'OPTIONAL'
    430         self.allocation = 'CALLBACK'
    431         self.ctype = 'pb_extension_t'
    432         self.array_decl = ''
    433         self.default = None
    434         self.max_size = 0
    435         self.max_count = 0
    436         
    437     def __str__(self):
    438         return '    pb_extension_t *extensions;'
    439     
    440     def types(self):
    441         return None
    442     
    443     def tags(self):
    444         return ''
    445         
    446     def encoded_size(self, allmsgs):
    447         # We exclude extensions from the count, because they cannot be known
    448         # until runtime. Other option would be to return None here, but this
    449         # way the value remains useful if extensions are not used.
    450         return EncodedSize(0)
    451 
    452 class ExtensionField(Field):
    453     def __init__(self, struct_name, desc, field_options):
    454         self.fullname = struct_name + desc.name
    455         self.extendee_name = names_from_type_name(desc.extendee)
    456         Field.__init__(self, self.fullname + 'struct', desc, field_options)
    457         
    458         if self.rules != 'OPTIONAL':
    459             self.skip = True
    460         else:
    461             self.skip = False
    462             self.rules = 'OPTEXT'
    463 
    464     def tags(self):
    465         '''Return the #define for the tag number of this field.'''
    466         identifier = '%s_tag' % self.fullname
    467         return '#define %-40s %d\n' % (identifier, self.tag)
    468 
    469     def extension_decl(self):
    470         '''Declaration of the extension type in the .pb.h file'''
    471         if self.skip:
    472             msg = '/* Extension field %s was skipped because only "optional"\n' % self.fullname
    473             msg +='   type of extension fields is currently supported. */\n'
    474             return msg
    475         
    476         return 'extern const pb_extension_type_t %s;\n' % self.fullname
    477 
    478     def extension_def(self):
    479         '''Definition of the extension type in the .pb.c file'''
    480 
    481         if self.skip:
    482             return ''
    483 
    484         result  = 'typedef struct {\n'
    485         result += str(self)
    486         result += '\n} %s;\n\n' % self.struct_name
    487         result += ('static const pb_field_t %s_field = \n  %s;\n\n' %
    488                     (self.fullname, self.pb_field_t(None)))
    489         result += 'const pb_extension_type_t %s = {\n' % self.fullname
    490         result += '    NULL,\n'
    491         result += '    NULL,\n'
    492         result += '    &%s_field\n' % self.fullname
    493         result += '};\n'
    494         return result
    495 
    496 
    497 # ---------------------------------------------------------------------------
    498 #                   Generation of messages (structures)
    499 # ---------------------------------------------------------------------------
    500 
    501 
    502 class Message:
    503     def __init__(self, names, desc, message_options):
    504         self.name = names
    505         self.fields = []
    506         
    507         for f in desc.field:
    508             field_options = get_nanopb_suboptions(f, message_options, self.name + f.name)
    509             if field_options.type != nanopb_pb2.FT_IGNORE:
    510                 self.fields.append(Field(self.name, f, field_options))
    511         
    512         if len(desc.extension_range) > 0:
    513             field_options = get_nanopb_suboptions(desc, message_options, self.name + 'extensions')
    514             range_start = min([r.start for r in desc.extension_range])
    515             if field_options.type != nanopb_pb2.FT_IGNORE:
    516                 self.fields.append(ExtensionRange(self.name, range_start, field_options))
    517         
    518         self.packed = message_options.packed_struct
    519         self.ordered_fields = self.fields[:]
    520         self.ordered_fields.sort()
    521 
    522     def get_dependencies(self):
    523         '''Get list of type names that this structure refers to.'''
    524         return [str(field.ctype) for field in self.fields]
    525     
    526     def __str__(self):
    527         result = 'typedef struct _%s {\n' % self.name
    528 
    529         if not self.ordered_fields:
    530             # Empty structs are not allowed in C standard.
    531             # Therefore add a dummy field if an empty message occurs.
    532             result += '    uint8_t dummy_field;'
    533 
    534         result += '\n'.join([str(f) for f in self.ordered_fields])
    535         result += '\n}'
    536         
    537         if self.packed:
    538             result += ' pb_packed'
    539         
    540         result += ' %s;' % self.name
    541         
    542         if self.packed:
    543             result = 'PB_PACKED_STRUCT_START\n' + result
    544             result += '\nPB_PACKED_STRUCT_END'
    545         
    546         return result
    547     
    548     def types(self):
    549         result = ""
    550         for field in self.fields:
    551             types = field.types()
    552             if types is not None:
    553                 result += types + '\n'
    554         return result
    555     
    556     def default_decl(self, declaration_only = False):
    557         result = ""
    558         for field in self.fields:
    559             default = field.default_decl(declaration_only)
    560             if default is not None:
    561                 result += default + '\n'
    562         return result
    563 
    564     def fields_declaration(self):
    565         result = 'extern const pb_field_t %s_fields[%d];' % (self.name, len(self.fields) + 1)
    566         return result
    567 
    568     def fields_definition(self):
    569         result = 'const pb_field_t %s_fields[%d] = {\n' % (self.name, len(self.fields) + 1)
    570         
    571         prev = None
    572         for field in self.ordered_fields:
    573             result += field.pb_field_t(prev)
    574             result += ',\n'
    575             prev = field.name
    576         
    577         result += '    PB_LAST_FIELD\n};'
    578         return result
    579 
    580     def encoded_size(self, allmsgs):
    581         '''Return the maximum size that this message can take when encoded.
    582         If the size cannot be determined, returns None.
    583         '''
    584         size = EncodedSize(0)
    585         for field in self.fields:
    586             fsize = field.encoded_size(allmsgs)
    587             if fsize is None:
    588                 return None
    589             size += fsize
    590         
    591         return size
    592 
    593 
    594 # ---------------------------------------------------------------------------
    595 #                    Processing of entire .proto files
    596 # ---------------------------------------------------------------------------
    597 
    598 
    599 def iterate_messages(desc, names = Names()):
    600     '''Recursively find all messages. For each, yield name, DescriptorProto.'''
    601     if hasattr(desc, 'message_type'):
    602         submsgs = desc.message_type
    603     else:
    604         submsgs = desc.nested_type
    605     
    606     for submsg in submsgs:
    607         sub_names = names + submsg.name
    608         yield sub_names, submsg
    609         
    610         for x in iterate_messages(submsg, sub_names):
    611             yield x
    612 
    613 def iterate_extensions(desc, names = Names()):
    614     '''Recursively find all extensions.
    615     For each, yield name, FieldDescriptorProto.
    616     '''
    617     for extension in desc.extension:
    618         yield names, extension
    619 
    620     for subname, subdesc in iterate_messages(desc, names):
    621         for extension in subdesc.extension:
    622             yield subname, extension
    623 
    624 def parse_file(fdesc, file_options):
    625     '''Takes a FileDescriptorProto and returns tuple (enums, messages, extensions).'''
    626     
    627     enums = []
    628     messages = []
    629     extensions = []
    630     
    631     if fdesc.package:
    632         base_name = Names(fdesc.package.split('.'))
    633     else:
    634         base_name = Names()
    635     
    636     for enum in fdesc.enum_type:
    637         enum_options = get_nanopb_suboptions(enum, file_options, base_name + enum.name)
    638         enums.append(Enum(base_name, enum, enum_options))
    639     
    640     for names, message in iterate_messages(fdesc, base_name):
    641         message_options = get_nanopb_suboptions(message, file_options, names)
    642         messages.append(Message(names, message, message_options))
    643         for enum in message.enum_type:
    644             enum_options = get_nanopb_suboptions(enum, message_options, names + enum.name)
    645             enums.append(Enum(names, enum, enum_options))
    646     
    647     for names, extension in iterate_extensions(fdesc, base_name):
    648         field_options = get_nanopb_suboptions(extension, file_options, names)
    649         if field_options.type != nanopb_pb2.FT_IGNORE:
    650             extensions.append(ExtensionField(names, extension, field_options))
    651     
    652     # Fix field default values where enum short names are used.
    653     for enum in enums:
    654         if not enum.options.long_names:
    655             for message in messages:
    656                 for field in message.fields:
    657                     if field.default in enum.value_longnames:
    658                         idx = enum.value_longnames.index(field.default)
    659                         field.default = enum.values[idx][0]
    660     
    661     return enums, messages, extensions
    662 
    663 def toposort2(data):
    664     '''Topological sort.
    665     From http://code.activestate.com/recipes/577413-topological-sort/
    666     This function is under the MIT license.
    667     '''
    668     for k, v in data.items():
    669         v.discard(k) # Ignore self dependencies
    670     extra_items_in_deps = reduce(set.union, data.values(), set()) - set(data.keys())
    671     data.update(dict([(item, set()) for item in extra_items_in_deps]))
    672     while True:
    673         ordered = set(item for item,dep in data.items() if not dep)
    674         if not ordered:
    675             break
    676         for item in sorted(ordered):
    677             yield item
    678         data = dict([(item, (dep - ordered)) for item,dep in data.items()
    679                 if item not in ordered])
    680     assert not data, "A cyclic dependency exists amongst %r" % data
    681 
    682 def sort_dependencies(messages):
    683     '''Sort a list of Messages based on dependencies.'''
    684     dependencies = {}
    685     message_by_name = {}
    686     for message in messages:
    687         dependencies[str(message.name)] = set(message.get_dependencies())
    688         message_by_name[str(message.name)] = message
    689     
    690     for msgname in toposort2(dependencies):
    691         if msgname in message_by_name:
    692             yield message_by_name[msgname]
    693 
    694 def make_identifier(headername):
    695     '''Make #ifndef identifier that contains uppercase A-Z and digits 0-9'''
    696     result = ""
    697     for c in headername.upper():
    698         if c.isalnum():
    699             result += c
    700         else:
    701             result += '_'
    702     return result
    703 
    704 def generate_header(dependencies, headername, enums, messages, extensions, options):
    705     '''Generate content for a header file.
    706     Generates strings, which should be concatenated and stored to file.
    707     '''
    708     
    709     yield '/* Automatically generated nanopb header */\n'
    710     if options.notimestamp:
    711         yield '/* Generated by %s */\n\n' % (nanopb_version)
    712     else:
    713         yield '/* Generated by %s at %s. */\n\n' % (nanopb_version, time.asctime())
    714     
    715     symbol = make_identifier(headername)
    716     yield '#ifndef _PB_%s_\n' % symbol
    717     yield '#define _PB_%s_\n' % symbol
    718     try:
    719         yield options.libformat % ('pb.h')
    720     except TypeError:
    721         # no %s specified - use whatever was passed in as options.libformat
    722         yield options.libformat
    723     yield '\n'
    724     
    725     for dependency in dependencies:
    726         noext = os.path.splitext(dependency)[0]
    727         yield options.genformat % (noext + '.' + options.extension + '.h')
    728         yield '\n'
    729 
    730     yield '#ifdef __cplusplus\n'
    731     yield 'extern "C" {\n'
    732     yield '#endif\n\n'
    733     
    734     yield '/* Enum definitions */\n'
    735     for enum in enums:
    736         yield str(enum) + '\n\n'
    737     
    738     yield '/* Struct definitions */\n'
    739     for msg in sort_dependencies(messages):
    740         yield msg.types()
    741         yield str(msg) + '\n\n'
    742     
    743     if extensions:
    744         yield '/* Extensions */\n'
    745         for extension in extensions:
    746             yield extension.extension_decl()
    747         yield '\n'
    748         
    749     yield '/* Default values for struct fields */\n'
    750     for msg in messages:
    751         yield msg.default_decl(True)
    752     yield '\n'
    753     
    754     yield '/* Field tags (for use in manual encoding/decoding) */\n'
    755     for msg in sort_dependencies(messages):
    756         for field in msg.fields:
    757             yield field.tags()
    758     for extension in extensions:
    759         yield extension.tags()
    760     yield '\n'
    761     
    762     yield '/* Struct field encoding specification for nanopb */\n'
    763     for msg in messages:
    764         yield msg.fields_declaration() + '\n'
    765     yield '\n'
    766     
    767     yield '/* Maximum encoded size of messages (where known) */\n'
    768     for msg in messages:
    769         msize = msg.encoded_size(messages)
    770         if msize is not None:
    771             identifier = '%s_size' % msg.name
    772             yield '#define %-40s %s\n' % (identifier, msize)
    773     yield '\n'
    774     
    775     yield '#ifdef __cplusplus\n'
    776     yield '} /* extern "C" */\n'
    777     yield '#endif\n'
    778     
    779     # End of header
    780     yield '\n#endif\n'
    781 
    782 def generate_source(headername, enums, messages, extensions, options):
    783     '''Generate content for a source file.'''
    784     
    785     yield '/* Automatically generated nanopb constant definitions */\n'
    786     if options.notimestamp:
    787         yield '/* Generated by %s */\n\n' % (nanopb_version)
    788     else:
    789         yield '/* Generated by %s at %s. */\n\n' % (nanopb_version, time.asctime())
    790     yield options.genformat % (headername)
    791     yield '\n'
    792     
    793     for msg in messages:
    794         yield msg.default_decl(False)
    795     
    796     yield '\n\n'
    797     
    798     for msg in messages:
    799         yield msg.fields_definition() + '\n\n'
    800     
    801     for ext in extensions:
    802         yield ext.extension_def() + '\n'
    803         
    804     # Add checks for numeric limits
    805     if messages:
    806         count_required_fields = lambda m: len([f for f in msg.fields if f.rules == 'REQUIRED'])
    807         largest_msg = max(messages, key = count_required_fields)
    808         largest_count = count_required_fields(largest_msg)
    809         if largest_count > 64:
    810             yield '\n/* Check that missing required fields will be properly detected */\n'
    811             yield '#if PB_MAX_REQUIRED_FIELDS < %d\n' % largest_count
    812             yield '#error Properly detecting missing required fields in %s requires \\\n' % largest_msg.name
    813             yield '       setting PB_MAX_REQUIRED_FIELDS to %d or more.\n' % largest_count
    814             yield '#endif\n'
    815     
    816     worst = 0
    817     worst_field = ''
    818     checks = []
    819     checks_msgnames = []
    820     for msg in messages:
    821         checks_msgnames.append(msg.name)
    822         for field in msg.fields:
    823             status = field.largest_field_value()
    824             if isinstance(status, (str, unicode)):
    825                 checks.append(status)
    826             elif status > worst:
    827                 worst = status
    828                 worst_field = str(field.struct_name) + '.' + str(field.name)
    829 
    830     if worst > 255 or checks:
    831         yield '\n/* Check that field information fits in pb_field_t */\n'
    832         
    833         if worst > 65535 or checks:
    834             yield '#if !defined(PB_FIELD_32BIT)\n'
    835             if worst > 65535:
    836                 yield '#error Field descriptor for %s is too large. Define PB_FIELD_32BIT to fix this.\n' % worst_field
    837             else:
    838                 assertion = ' && '.join(str(c) + ' < 65536' for c in checks)
    839                 msgs = '_'.join(str(n) for n in checks_msgnames)
    840                 yield '/* If you get an error here, it means that you need to define PB_FIELD_32BIT\n'
    841                 yield ' * compile-time option. You can do that in pb.h or on compiler command line.\n'
    842                 yield ' * \n'
    843                 yield ' * The reason you need to do this is that some of your messages contain tag\n'
    844                 yield ' * numbers or field sizes that are larger than what can fit in 8 or 16 bit\n'
    845                 yield ' * field descriptors.\n'
    846                 yield ' */\n'
    847                 yield 'STATIC_ASSERT((%s), YOU_MUST_DEFINE_PB_FIELD_32BIT_FOR_MESSAGES_%s)\n'%(assertion,msgs)
    848             yield '#endif\n\n'
    849         
    850         if worst < 65536:
    851             yield '#if !defined(PB_FIELD_16BIT) && !defined(PB_FIELD_32BIT)\n'
    852             if worst > 255:
    853                 yield '#error Field descriptor for %s is too large. Define PB_FIELD_16BIT to fix this.\n' % worst_field
    854             else:
    855                 assertion = ' && '.join(str(c) + ' < 256' for c in checks)
    856                 msgs = '_'.join(str(n) for n in checks_msgnames)
    857                 yield '/* If you get an error here, it means that you need to define PB_FIELD_16BIT\n'
    858                 yield ' * compile-time option. You can do that in pb.h or on compiler command line.\n'
    859                 yield ' * \n'
    860                 yield ' * The reason you need to do this is that some of your messages contain tag\n'
    861                 yield ' * numbers or field sizes that are larger than what can fit in the default\n'
    862                 yield ' * 8 bit descriptors.\n'
    863                 yield ' */\n'
    864                 yield 'STATIC_ASSERT((%s), YOU_MUST_DEFINE_PB_FIELD_16BIT_FOR_MESSAGES_%s)\n'%(assertion,msgs)
    865             yield '#endif\n\n'
    866     
    867     # Add check for sizeof(double)
    868     has_double = False
    869     for msg in messages:
    870         for field in msg.fields:
    871             if field.ctype == 'double':
    872                 has_double = True
    873     
    874     if has_double:
    875         yield '\n'
    876         yield '/* On some platforms (such as AVR), double is really float.\n'
    877         yield ' * These are not directly supported by nanopb, but see example_avr_double.\n'
    878         yield ' * To get rid of this error, remove any double fields from your .proto.\n'
    879         yield ' */\n'
    880         yield 'STATIC_ASSERT(sizeof(double) == 8, DOUBLE_MUST_BE_8_BYTES)\n'
    881     
    882     yield '\n'
    883 
    884 # ---------------------------------------------------------------------------
    885 #                    Options parsing for the .proto files
    886 # ---------------------------------------------------------------------------
    887 
    888 from fnmatch import fnmatch
    889 
    890 def read_options_file(infile):
    891     '''Parse a separate options file to list:
    892         [(namemask, options), ...]
    893     '''
    894     results = []
    895     for line in infile:
    896         line = line.strip()
    897         if not line or line.startswith('//') or line.startswith('#'):
    898             continue
    899         
    900         parts = line.split(None, 1)
    901         opts = nanopb_pb2.NanoPBOptions()
    902         text_format.Merge(parts[1], opts)
    903         results.append((parts[0], opts))
    904 
    905     return results
    906 
    907 class Globals:
    908     '''Ugly global variables, should find a good way to pass these.'''
    909     verbose_options = False
    910     separate_options = []
    911     matched_namemasks = set()
    912 
    913 def get_nanopb_suboptions(subdesc, options, name):
    914     '''Get copy of options, and merge information from subdesc.'''
    915     new_options = nanopb_pb2.NanoPBOptions()
    916     new_options.CopyFrom(options)
    917     
    918     # Handle options defined in a separate file
    919     dotname = '.'.join(name.parts)
    920     for namemask, options in Globals.separate_options:
    921         if fnmatch(dotname, namemask):
    922             Globals.matched_namemasks.add(namemask)
    923             new_options.MergeFrom(options)
    924     
    925     # Handle options defined in .proto
    926     if isinstance(subdesc.options, descriptor.FieldOptions):
    927         ext_type = nanopb_pb2.nanopb
    928     elif isinstance(subdesc.options, descriptor.FileOptions):
    929         ext_type = nanopb_pb2.nanopb_fileopt
    930     elif isinstance(subdesc.options, descriptor.MessageOptions):
    931         ext_type = nanopb_pb2.nanopb_msgopt
    932     elif isinstance(subdesc.options, descriptor.EnumOptions):
    933         ext_type = nanopb_pb2.nanopb_enumopt
    934     else:
    935         raise Exception("Unknown options type")
    936     
    937     if subdesc.options.HasExtension(ext_type):
    938         ext = subdesc.options.Extensions[ext_type]
    939         new_options.MergeFrom(ext)
    940     
    941     if Globals.verbose_options:
    942         sys.stderr.write("Options for " + dotname + ": ")
    943         sys.stderr.write(text_format.MessageToString(new_options) + "\n")
    944     
    945     return new_options
    946 
    947 
    948 # ---------------------------------------------------------------------------
    949 #                         Command line interface
    950 # ---------------------------------------------------------------------------
    951 
    952 import sys
    953 import os.path    
    954 from optparse import OptionParser
    955 
    956 optparser = OptionParser(
    957     usage = "Usage: nanopb_generator.py [options] file.pb ...",
    958     epilog = "Compile file.pb from file.proto by: 'protoc -ofile.pb file.proto'. " +
    959              "Output will be written to file.pb.h and file.pb.c.")
    960 optparser.add_option("-x", dest="exclude", metavar="FILE", action="append", default=[],
    961     help="Exclude file from generated #include list.")
    962 optparser.add_option("-e", "--extension", dest="extension", metavar="EXTENSION", default="pb",
    963     help="Set extension to use instead of 'pb' for generated files. [default: %default]")
    964 optparser.add_option("-f", "--options-file", dest="options_file", metavar="FILE", default="%s.options",
    965     help="Set name of a separate generator options file.")
    966 optparser.add_option("-Q", "--generated-include-format", dest="genformat",
    967     metavar="FORMAT", default='#include "%s"\n',
    968     help="Set format string to use for including other .pb.h files. [default: %default]")
    969 optparser.add_option("-L", "--library-include-format", dest="libformat",
    970     metavar="FORMAT", default='#include <%s>\n',
    971     help="Set format string to use for including the nanopb pb.h header. [default: %default]")
    972 optparser.add_option("-T", "--no-timestamp", dest="notimestamp", action="store_true", default=False,
    973     help="Don't add timestamp to .pb.h and .pb.c preambles")
    974 optparser.add_option("-q", "--quiet", dest="quiet", action="store_true", default=False,
    975     help="Don't print anything except errors.")
    976 optparser.add_option("-v", "--verbose", dest="verbose", action="store_true", default=False,
    977     help="Print more information.")
    978 optparser.add_option("-s", dest="settings", metavar="OPTION:VALUE", action="append", default=[],
    979     help="Set generator option (max_size, max_count etc.).")
    980 
    981 def process_file(filename, fdesc, options):
    982     '''Process a single file.
    983     filename: The full path to the .proto or .pb source file, as string.
    984     fdesc: The loaded FileDescriptorSet, or None to read from the input file.
    985     options: Command line options as they come from OptionsParser.
    986     
    987     Returns a dict:
    988         {'headername': Name of header file,
    989          'headerdata': Data for the .h header file,
    990          'sourcename': Name of the source code file,
    991          'sourcedata': Data for the .c source code file
    992         }
    993     '''
    994     toplevel_options = nanopb_pb2.NanoPBOptions()
    995     for s in options.settings:
    996         text_format.Merge(s, toplevel_options)
    997     
    998     if not fdesc:
    999         data = open(filename, 'rb').read()
   1000         fdesc = descriptor.FileDescriptorSet.FromString(data).file[0]
   1001     
   1002     # Check if there is a separate .options file
   1003     try:
   1004         optfilename = options.options_file % os.path.splitext(filename)[0]
   1005     except TypeError:
   1006         # No %s specified, use the filename as-is
   1007         optfilename = options.options_file
   1008     
   1009     if os.path.isfile(optfilename):
   1010         if options.verbose:
   1011             sys.stderr.write('Reading options from ' + optfilename + '\n')
   1012 
   1013         Globals.separate_options = read_options_file(open(optfilename, "rU"))
   1014     else:
   1015         Globals.separate_options = []
   1016     Globals.matched_namemasks = set()
   1017     
   1018     # Parse the file
   1019     file_options = get_nanopb_suboptions(fdesc, toplevel_options, Names([filename]))
   1020     enums, messages, extensions = parse_file(fdesc, file_options)
   1021     
   1022     # Decide the file names
   1023     noext = os.path.splitext(filename)[0]
   1024     headername = noext + '.' + options.extension + '.h'
   1025     sourcename = noext + '.' + options.extension + '.c'
   1026     headerbasename = os.path.basename(headername)
   1027     
   1028     # List of .proto files that should not be included in the C header file
   1029     # even if they are mentioned in the source .proto.
   1030     excludes = ['nanopb.proto', 'google/protobuf/descriptor.proto'] + options.exclude
   1031     dependencies = [d for d in fdesc.dependency if d not in excludes]
   1032     
   1033     headerdata = ''.join(generate_header(dependencies, headerbasename, enums,
   1034                                          messages, extensions, options))
   1035 
   1036     sourcedata = ''.join(generate_source(headerbasename, enums,
   1037                                          messages, extensions, options))
   1038 
   1039     # Check if there were any lines in .options that did not match a member
   1040     unmatched = [n for n,o in Globals.separate_options if n not in Globals.matched_namemasks]
   1041     if unmatched and not options.quiet:
   1042         sys.stderr.write("Following patterns in " + optfilename + " did not match any fields: "
   1043                          + ', '.join(unmatched) + "\n")
   1044         if not Globals.verbose_options:
   1045             sys.stderr.write("Use  protoc --nanopb-out=-v:.   to see a list of the field names.\n")
   1046 
   1047     return {'headername': headername, 'headerdata': headerdata,
   1048             'sourcename': sourcename, 'sourcedata': sourcedata}
   1049     
   1050 def main_cli():
   1051     '''Main function when invoked directly from the command line.'''
   1052     
   1053     options, filenames = optparser.parse_args()
   1054     
   1055     if not filenames:
   1056         optparser.print_help()
   1057         sys.exit(1)
   1058     
   1059     if options.quiet:
   1060         options.verbose = False
   1061 
   1062     Globals.verbose_options = options.verbose
   1063     
   1064     for filename in filenames:
   1065         results = process_file(filename, None, options)
   1066         
   1067         if not options.quiet:
   1068             sys.stderr.write("Writing to " + results['headername'] + " and "
   1069                              + results['sourcename'] + "\n")
   1070     
   1071         open(results['headername'], 'w').write(results['headerdata'])
   1072         open(results['sourcename'], 'w').write(results['sourcedata'])        
   1073 
   1074 def main_plugin():
   1075     '''Main function when invoked as a protoc plugin.'''
   1076 
   1077     import sys
   1078     if sys.platform == "win32":
   1079         import os, msvcrt
   1080         # Set stdin and stdout to binary mode
   1081         msvcrt.setmode(sys.stdin.fileno(), os.O_BINARY)
   1082         msvcrt.setmode(sys.stdout.fileno(), os.O_BINARY)
   1083     
   1084     data = sys.stdin.read()
   1085     request = plugin_pb2.CodeGeneratorRequest.FromString(data)
   1086     
   1087     import shlex
   1088     args = shlex.split(request.parameter)
   1089     options, dummy = optparser.parse_args(args)
   1090     
   1091     Globals.verbose_options = options.verbose
   1092     
   1093     response = plugin_pb2.CodeGeneratorResponse()
   1094     
   1095     for filename in request.file_to_generate:
   1096         for fdesc in request.proto_file:
   1097             if fdesc.name == filename:
   1098                 results = process_file(filename, fdesc, options)
   1099                 
   1100                 f = response.file.add()
   1101                 f.name = results['headername']
   1102                 f.content = results['headerdata']
   1103 
   1104                 f = response.file.add()
   1105                 f.name = results['sourcename']
   1106                 f.content = results['sourcedata']    
   1107     
   1108     sys.stdout.write(response.SerializeToString())
   1109 
   1110 if __name__ == '__main__':
   1111     # Check if we are running as a plugin under protoc
   1112     if 'protoc-gen-' in sys.argv[0] or '--protoc-plugin' in sys.argv:
   1113         main_plugin()
   1114     else:
   1115         main_cli()
   1116 
   1117