source4/scripting/python/samba/ms_schema.py

   1 #!/usr/bin/env python
   2 #
   3 # create schema.ldif (as a string) from WSPP documentation
   4 #
   5 # based on minschema.py and minschema_wspp
   6 #
   7
   8 import re
   9 import base64
  10 import uuid
  11
  12 bitFields = {}
  13
  14 # ADTS: 2.2.9
  15 # bit positions as labeled in the docs
  16 bitFields["searchflags"] = {
  17     'fATTINDEX': 31,         # IX
  18     'fPDNTATTINDEX': 30,     # PI
  19     'fANR': 29, #AR
  20     'fPRESERVEONDELETE': 28,         # PR
  21     'fCOPY': 27,     # CP
  22     'fTUPLEINDEX': 26,       # TP
  23     'fSUBTREEATTINDEX': 25,  # ST
  24     'fCONFIDENTIAL': 24,     # CF
  25     'fNEVERVALUEAUDIT': 23,  # NV
  26     'fRODCAttribute': 22,    # RO
  27
  28
  29     # missing in ADTS but required by LDIF
  30     'fRODCFilteredAttribute': 22,    # RO ?
  31     'fCONFIDENTAIL': 24, # typo
  32     'fRODCFILTEREDATTRIBUTE': 22 # case
  33     }
  34
  35 # ADTS: 2.2.10
  36 bitFields["systemflags"] = {
  37     'FLAG_ATTR_NOT_REPLICATED': 31, 'FLAG_CR_NTDS_NC': 31,      # NR
  38     'FLAG_ATTR_REQ_PARTIAL_SET_MEMBER': 30, 'FLAG_CR_NTDS_DOMAIN': 30,  # PS
  39     'FLAG_ATTR_IS_CONSTRUCTED': 29, 'FLAG_CR_NTDS_NOT_GC_REPLICATED': 29,       # CS
  40     'FLAG_ATTR_IS_OPERATIONAL': 28,     # OP
  41     'FLAG_SCHEMA_BASE_OBJECT': 27,      # BS
  42     'FLAG_ATTR_IS_RDN': 26,     # RD
  43     'FLAG_DISALLOW_MOVE_ON_DELETE': 6,  # DE
  44     'FLAG_DOMAIN_DISALLOW_MOVE': 5,     # DM
  45     'FLAG_DOMAIN_DISALLOW_RENAME': 4,   # DR
  46     'FLAG_CONFIG_ALLOW_LIMITED_MOVE': 3,        # AL
  47     'FLAG_CONFIG_ALLOW_MOVE': 2,        # AM
  48     'FLAG_CONFIG_ALLOW_RENAME': 1,      # AR
  49     'FLAG_DISALLOW_DELETE': 0   # DD
  50     }
  51
  52 # ADTS: 2.2.11
  53 bitFields["schemaflagsex"] = {
  54     'FLAG_ATTR_IS_CRITICAL': 31
  55     }
  56
  57 # ADTS: 3.1.1.2.2.2
  58 oMObjectClassBER = {
  59     '1.3.12.2.1011.28.0.702' : base64.b64encode('\x2B\x0C\x02\x87\x73\x1C\x00\x85\x3E'),
  60     '1.2.840.113556.1.1.1.12': base64.b64encode('\x2A\x86\x48\x86\xF7\x14\x01\x01\x01\x0C'),
  61     '2.6.6.1.2.5.11.29'      : base64.b64encode('\x56\x06\x01\x02\x05\x0B\x1D'),
  62     '1.2.840.113556.1.1.1.11': base64.b64encode('\x2A\x86\x48\x86\xF7\x14\x01\x01\x01\x0B'),
  63     '1.3.12.2.1011.28.0.714' : base64.b64encode('\x2B\x0C\x02\x87\x73\x1C\x00\x85\x4A'),
  64     '1.3.12.2.1011.28.0.732' : base64.b64encode('\x2B\x0C\x02\x87\x73\x1C\x00\x85\x5C'),
  65     '1.2.840.113556.1.1.1.6' : base64.b64encode('\x2A\x86\x48\x86\xF7\x14\x01\x01\x01\x06')
  66 }
  67
  68 # separated by commas in docs, and must be broken up
  69 multivalued_attrs = set(["auxiliaryclass","maycontain","mustcontain","posssuperiors",
  70                          "systemauxiliaryclass","systemmaycontain","systemmustcontain",
  71                          "systemposssuperiors"])
  72
  73 def __read_folded_line(f, buffer):
  74     """ reads a line from an LDIF file, unfolding it"""
  75     line = buffer
  76
  77     while True:
  78         l = f.readline()
  79
  80         if l[:1] == " ":
  81             # continued line
  82
  83             # cannot fold an empty line
  84             assert(line != "" and line != "\n")
  85
  86             # preserves '\n '
  87             line = line + l
  88         else:
  89             # non-continued line
  90             if line == "":
  91                 line = l
  92
  93                 if l == "":
  94                     # eof, definitely won't be folded
  95                     break
  96             else:
  97                 # marks end of a folded line
  98                 # line contains the now unfolded line
  99                 # buffer contains the start of the next possibly folded line
 100                 buffer = l
 101                 break
 102
 103     return (line, buffer)
 104
 105
 106 def __read_raw_entries(f):
 107     """reads an LDIF entry, only unfolding lines"""
 108
 109     # will not match options after the attribute type
 110     attr_type_re = re.compile("^([A-Za-z]+[A-Za-z0-9-]*):")
 111
 112     buffer = ""
 113
 114     while True:
 115         entry = []
 116
 117         while True:
 118             (l, buffer) = __read_folded_line(f, buffer)
 119
 120             if l[:1] == "#":
 121                 continue
 122
 123             if l == "\n" or l == "":
 124                 break
 125
 126             m = attr_type_re.match(l)
 127
 128             if m:
 129                 if l[-1:] == "\n":
 130                     l = l[:-1]
 131
 132                 entry.append(l)
 133             else:
 134                 print >>sys.stderr, "Invalid line: %s" % l,
 135                 sys.exit(1)
 136
 137         if len(entry):
 138             yield entry
 139
 140         if l == "":
 141             break
 142
 143
 144 def fix_dn(dn):
 145     """fix a string DN to use ${SCHEMADN}"""
 146
 147     # folding?
 148     if dn.find("<RootDomainDN>") != -1:
 149         dn = dn.replace("\n ", "")
 150         dn = dn.replace(" ", "")
 151         return dn.replace("CN=Schema,CN=Configuration,<RootDomainDN>", "${SCHEMADN}")
 152     else:
 153         return dn
 154
 155 def __convert_bitfield(key, value):
 156     """Evaluate the OR expression in 'value'"""
 157     assert(isinstance(value, str))
 158
 159     value = value.replace("\n ", "")
 160     value = value.replace(" ", "")
 161
 162     try:
 163         # some attributes already have numeric values
 164         o = int(value)
 165     except ValueError:
 166         o = 0
 167         flags = value.split("|")
 168         for f in flags:
 169             bitpos = bitFields[key][f]
 170             o = o | (1 << (31 - bitpos))
 171
 172     return str(o)
 173
 174 def __write_ldif_one(entry):
 175     """Write out entry as LDIF"""
 176     out = []
 177
 178     for l in entry:
 179         if isinstance(l[1], str):
 180             vl = [l[1]]
 181         else:
 182             vl = l[1]
 183
 184         if l[0].lower() == 'omobjectclass':
 185             out.append("%s:: %s" % (l[0], l[1]))
 186             continue
 187
 188         for v in vl:
 189             out.append("%s: %s" % (l[0], v))
 190
 191
 192     return "\n".join(out)
 193
 194 def __transform_entry(entry, objectClass):
 195     """Perform transformations required to convert the LDIF-like schema
 196        file entries to LDIF, including Samba-specific stuff."""
 197
 198     entry = [l.split(":", 1) for l in entry]
 199
 200     cn = ""
 201
 202     for l in entry:
 203         key = l[0].lower()
 204         l[1] = l[1].lstrip()
 205         l[1] = l[1].rstrip()
 206
 207         if not cn and key == "cn":
 208             cn = l[1]
 209
 210         if key in multivalued_attrs:
 211             # unlike LDIF, these are comma-separated
 212             l[1] = l[1].replace("\n ", "")
 213             l[1] = l[1].replace(" ", "")
 214
 215             l[1] = l[1].split(",")
 216
 217         if key in bitFields:
 218             l[1] = __convert_bitfield(key, l[1])
 219
 220         if key == "omobjectclass":
 221             l[1] = oMObjectClassBER[l[1].strip()]
 222
 223         if isinstance(l[1], str):
 224             l[1] = fix_dn(l[1])
 225
 226
 227     assert(cn)
 228     entry.insert(0, ["dn", "CN=%s,${SCHEMADN}" % cn])
 229     entry.insert(1, ["objectClass", ["top", objectClass]])
 230     entry.insert(2, ["cn", cn])
 231     entry.insert(2, ["objectGUID", str(uuid.uuid4())])
 232     entry.insert(2, ["adminDescription", cn])
 233     entry.insert(2, ["adminDisplayName", cn])
 234
 235     for l in entry:
 236         key = l[0].lower()
 237
 238         if key == "cn":
 239             entry.remove(l)
 240
 241     return entry
 242
 243 def __parse_schema_file(filename, objectClass):
 244     """Load and transform a schema file."""
 245
 246     out = []
 247
 248     f = open(filename, "rU")
 249     for entry in __read_raw_entries(f):
 250         out.append(__write_ldif_one(__transform_entry(entry, objectClass)))
 251
 252     return "\n\n".join(out)
 253
 254
 255 def read_ms_schema(attr_file, classes_file, dump_attributes = True, dump_classes = True, debug = False):
 256     """Read WSPP documentation-derived schema files."""
 257
 258     attr_ldif = ""
 259     classes_ldif = ""
 260
 261     if dump_attributes:
 262         attr_ldif =  __parse_schema_file(attr_file, "attributeSchema")
 263     if dump_classes:
 264         classes_ldif = __parse_schema_file(classes_file, "classSchema")
 265
 266     return attr_ldif + "\n\n" + classes_ldif + "\n\n"
 267
 268 if __name__ == '__main__':
 269     import sys
 270
 271     try:
 272         attr_file = sys.argv[1]
 273         classes_file = sys.argv[2]
 274     except IndexError:
 275         print >>sys.stderr, "Usage: %s attr-file.txt classes-file.txt" % (sys.argv[0])
 276         sys.exit(1)
 277
 278     print read_ms_schema(attr_file, classes_file)
 279
 280