]> git.rkrishnan.org Git - tahoe-lafs/tahoe-lafs.git/blobdiff - src/allmydata/scripts/debug.py
bin/tahoe: clean up global-vs-subcommand arguments like --node-directory
[tahoe-lafs/tahoe-lafs.git] / src / allmydata / scripts / debug.py
index 8073fe783a6e74e23237640e2a1b2d4034d87d6d..cb8e986a7136552469450f51e131d7c436aeb1e1 100644 (file)
@@ -5,9 +5,11 @@ import struct, time, os, sys
 from twisted.python import usage, failure
 from twisted.internet import defer
 from twisted.scripts import trial as twisted_trial
+from foolscap.logging import cli as foolscap_cli
+from allmydata.scripts.common import BaseOptions
 
 
-class DumpOptions(usage.Options):
+class DumpOptions(BaseOptions):
     def getSynopsis(self):
         return "Usage: tahoe debug dump-share SHARE_FILENAME"
 
@@ -17,7 +19,7 @@ class DumpOptions(usage.Options):
         ]
 
     def getUsage(self, width=None):
-        t = usage.Options.getUsage(self, width)
+        t = BaseOptions.getUsage(self, width)
         t += """
 Print lots of information about the given share, by parsing the share's
 contents. This includes share type, lease information, encoding parameters,
@@ -68,7 +70,7 @@ def dump_immutable_chk_share(f, out, options):
     from allmydata.util.encodingutil import quote_output, to_str
 
     # use a ReadBucketProxy to parse the bucket and find the uri extension
-    bp = ReadBucketProxy(None, '', '')
+    bp = ReadBucketProxy(None, None, '')
     offsets = bp._parse_offsets(f.read_share_data(0, 0x44))
     print >>out, "%20s: %d" % ("version", bp._version)
     seek = offsets['uri_extension']
@@ -181,9 +183,12 @@ def dump_mutable_share(options):
 
     share_type = "unknown"
     f.seek(m.DATA_OFFSET)
-    if f.read(1) == "\x00":
+    version = f.read(1)
+    if version == "\x00":
         # this slot contains an SMDF share
         share_type = "SDMF"
+    elif version == "\x01":
+        share_type = "MDMF"
     f.close()
 
     print >>out
@@ -210,6 +215,8 @@ def dump_mutable_share(options):
 
     if share_type == "SDMF":
         dump_SDMF_share(m, data_length, options)
+    elif share_type == "MDMF":
+        dump_MDMF_share(m, data_length, options)
 
     return 0
 
@@ -300,9 +307,106 @@ def dump_SDMF_share(m, length, options):
 
     print >>out
 
+def dump_MDMF_share(m, length, options):
+    from allmydata.mutable.layout import MDMFSlotReadProxy
+    from allmydata.util import base32, hashutil
+    from allmydata.uri import MDMFVerifierURI
+    from allmydata.util.encodingutil import quote_output, to_str
+
+    offset = m.DATA_OFFSET
+    out = options.stdout
+
+    f = open(options['filename'], "rb")
+    storage_index = None; shnum = 0
+
+    class ShareDumper(MDMFSlotReadProxy):
+        def _read(self, readvs, force_remote=False, queue=False):
+            data = []
+            for (where,length) in readvs:
+                f.seek(offset+where)
+                data.append(f.read(length))
+            return defer.succeed({shnum: data})
+
+    p = ShareDumper(None, storage_index, shnum)
+    def extract(func):
+        stash = []
+        # these methods return Deferreds, but we happen to know that they run
+        # synchronously when not actually talking to a remote server
+        d = func()
+        d.addCallback(stash.append)
+        return stash[0]
+
+    verinfo = extract(p.get_verinfo)
+    encprivkey = extract(p.get_encprivkey)
+    signature = extract(p.get_signature)
+    pubkey = extract(p.get_verification_key)
+    block_hash_tree = extract(p.get_blockhashes)
+    share_hash_chain = extract(p.get_sharehashes)
+    f.close()
+
+    (seqnum, root_hash, salt_to_use, segsize, datalen, k, N, prefix,
+     offsets) = verinfo
+
+    print >>out, " MDMF contents:"
+    print >>out, "  seqnum: %d" % seqnum
+    print >>out, "  root_hash: %s" % base32.b2a(root_hash)
+    #print >>out, "  IV: %s" % base32.b2a(IV)
+    print >>out, "  required_shares: %d" % k
+    print >>out, "  total_shares: %d" % N
+    print >>out, "  segsize: %d" % segsize
+    print >>out, "  datalen: %d" % datalen
+    print >>out, "  enc_privkey: %d bytes" % len(encprivkey)
+    print >>out, "  pubkey: %d bytes" % len(pubkey)
+    print >>out, "  signature: %d bytes" % len(signature)
+    share_hash_ids = ",".join([str(hid)
+                               for hid in sorted(share_hash_chain.keys())])
+    print >>out, "  share_hash_chain: %s" % share_hash_ids
+    print >>out, "  block_hash_tree: %d nodes" % len(block_hash_tree)
+
+    # the storage index isn't stored in the share itself, so we depend upon
+    # knowing the parent directory name to get it
+    pieces = options['filename'].split(os.sep)
+    if len(pieces) >= 2:
+        piece = to_str(pieces[-2])
+        if base32.could_be_base32_encoded(piece):
+            storage_index = base32.a2b(piece)
+            fingerprint = hashutil.ssk_pubkey_fingerprint_hash(pubkey)
+            u = MDMFVerifierURI(storage_index, fingerprint)
+            verify_cap = u.to_string()
+            print >>out, "  verify-cap:", quote_output(verify_cap, quotemarks=False)
+
+    if options['offsets']:
+        # NOTE: this offset-calculation code is fragile, and needs to be
+        # merged with MutableShareFile's internals.
+
+        print >>out
+        print >>out, " Section Offsets:"
+        def printoffset(name, value, shift=0):
+            print >>out, "%s%.20s: %s   (0x%x)" % (" "*shift, name, value, value)
+        printoffset("first lease", m.HEADER_SIZE, 2)
+        printoffset("share data", m.DATA_OFFSET, 2)
+        o_seqnum = m.DATA_OFFSET + struct.calcsize(">B")
+        printoffset("seqnum", o_seqnum, 4)
+        o_root_hash = m.DATA_OFFSET + struct.calcsize(">BQ")
+        printoffset("root_hash", o_root_hash, 4)
+        for k in ["enc_privkey", "share_hash_chain", "signature",
+                  "verification_key", "verification_key_end",
+                  "share_data", "block_hash_tree", "EOF"]:
+            name = {"share_data": "block data",
+                    "verification_key": "pubkey",
+                    "verification_key_end": "end of pubkey",
+                    "EOF": "end of share data"}.get(k,k)
+            offset = m.DATA_OFFSET + offsets[k]
+            printoffset(name, offset, 4)
+        f = open(options['filename'], "rb")
+        printoffset("extra leases", m._read_extra_lease_offset(f) + 4, 2)
+        f.close()
+
+    print >>out
+
 
 
-class DumpCapOptions(usage.Options):
+class DumpCapOptions(BaseOptions):
     def getSynopsis(self):
         return "Usage: tahoe debug dump-cap [options] FILECAP"
     optParameters = [
@@ -317,7 +421,7 @@ class DumpCapOptions(usage.Options):
         self.cap = cap
 
     def getUsage(self, width=None):
-        t = usage.Options.getUsage(self, width)
+        t = BaseOptions.getUsage(self, width)
         t += """
 Print information about the given cap-string (aka: URI, file-cap, dir-cap,
 read-cap, write-cap). The URI string is parsed and unpacked. This prints the
@@ -415,9 +519,9 @@ def dump_uri_instance(u, nodeid, secret, out, show_header=True):
             print >>out, "Literal File URI:"
         print >>out, " data:", quote_output(u.data)
 
-    elif isinstance(u, uri.WriteableSSKFileURI):
+    elif isinstance(u, uri.WriteableSSKFileURI): # SDMF
         if show_header:
-            print >>out, "SSK Writeable URI:"
+            print >>out, "SDMF Writeable URI:"
         print >>out, " writekey:", base32.b2a(u.writekey)
         print >>out, " readkey:", base32.b2a(u.readkey)
         print >>out, " storage index:", si_b2a(u.get_storage_index())
@@ -428,20 +532,54 @@ def dump_uri_instance(u, nodeid, secret, out, show_header=True):
             print >>out, " write_enabler:", base32.b2a(we)
             print >>out
         _dump_secrets(u.get_storage_index(), secret, nodeid, out)
-
     elif isinstance(u, uri.ReadonlySSKFileURI):
         if show_header:
-            print >>out, "SSK Read-only URI:"
+            print >>out, "SDMF Read-only URI:"
         print >>out, " readkey:", base32.b2a(u.readkey)
         print >>out, " storage index:", si_b2a(u.get_storage_index())
         print >>out, " fingerprint:", base32.b2a(u.fingerprint)
     elif isinstance(u, uri.SSKVerifierURI):
         if show_header:
-            print >>out, "SSK Verifier URI:"
+            print >>out, "SDMF Verifier URI:"
         print >>out, " storage index:", si_b2a(u.get_storage_index())
         print >>out, " fingerprint:", base32.b2a(u.fingerprint)
 
-    elif isinstance(u, uri.DirectoryURI):
+    elif isinstance(u, uri.WriteableMDMFFileURI): # MDMF
+        if show_header:
+            print >>out, "MDMF Writeable URI:"
+        print >>out, " writekey:", base32.b2a(u.writekey)
+        print >>out, " readkey:", base32.b2a(u.readkey)
+        print >>out, " storage index:", si_b2a(u.get_storage_index())
+        print >>out, " fingerprint:", base32.b2a(u.fingerprint)
+        print >>out
+        if nodeid:
+            we = hashutil.ssk_write_enabler_hash(u.writekey, nodeid)
+            print >>out, " write_enabler:", base32.b2a(we)
+            print >>out
+        _dump_secrets(u.get_storage_index(), secret, nodeid, out)
+    elif isinstance(u, uri.ReadonlyMDMFFileURI):
+        if show_header:
+            print >>out, "MDMF Read-only URI:"
+        print >>out, " readkey:", base32.b2a(u.readkey)
+        print >>out, " storage index:", si_b2a(u.get_storage_index())
+        print >>out, " fingerprint:", base32.b2a(u.fingerprint)
+    elif isinstance(u, uri.MDMFVerifierURI):
+        if show_header:
+            print >>out, "MDMF Verifier URI:"
+        print >>out, " storage index:", si_b2a(u.get_storage_index())
+        print >>out, " fingerprint:", base32.b2a(u.fingerprint)
+
+
+    elif isinstance(u, uri.ImmutableDirectoryURI): # CHK-based directory
+        if show_header:
+            print >>out, "CHK Directory URI:"
+        dump_uri_instance(u._filenode_uri, nodeid, secret, out, False)
+    elif isinstance(u, uri.ImmutableDirectoryURIVerifier):
+        if show_header:
+            print >>out, "CHK Directory Verifier URI:"
+        dump_uri_instance(u._filenode_uri, nodeid, secret, out, False)
+
+    elif isinstance(u, uri.DirectoryURI): # SDMF-based directory
         if show_header:
             print >>out, "Directory Writeable URI:"
         dump_uri_instance(u._filenode_uri, nodeid, secret, out, False)
@@ -453,10 +591,24 @@ def dump_uri_instance(u, nodeid, secret, out, show_header=True):
         if show_header:
             print >>out, "Directory Verifier URI:"
         dump_uri_instance(u._filenode_uri, nodeid, secret, out, False)
+
+    elif isinstance(u, uri.MDMFDirectoryURI): # MDMF-based directory
+        if show_header:
+            print >>out, "Directory Writeable URI:"
+        dump_uri_instance(u._filenode_uri, nodeid, secret, out, False)
+    elif isinstance(u, uri.ReadonlyMDMFDirectoryURI):
+        if show_header:
+            print >>out, "Directory Read-only URI:"
+        dump_uri_instance(u._filenode_uri, nodeid, secret, out, False)
+    elif isinstance(u, uri.MDMFDirectoryURIVerifier):
+        if show_header:
+            print >>out, "Directory Verifier URI:"
+        dump_uri_instance(u._filenode_uri, nodeid, secret, out, False)
+
     else:
         print >>out, "unknown cap type"
 
-class FindSharesOptions(usage.Options):
+class FindSharesOptions(BaseOptions):
     def getSynopsis(self):
         return "Usage: tahoe debug find-shares STORAGE_INDEX NODEDIRS.."
 
@@ -466,7 +618,7 @@ class FindSharesOptions(usage.Options):
         self.nodedirs = map(argv_to_abspath, nodedirs)
 
     def getUsage(self, width=None):
-        t = usage.Options.getUsage(self, width)
+        t = BaseOptions.getUsage(self, width)
         t += """
 Locate all shares for the given storage index. This command looks through one
 or more node directories to find the shares. It returns a list of filenames,
@@ -506,7 +658,7 @@ def find_shares(options):
     return 0
 
 
-class CatalogSharesOptions(usage.Options):
+class CatalogSharesOptions(BaseOptions):
     """
 
     """
@@ -520,7 +672,7 @@ class CatalogSharesOptions(usage.Options):
         return "Usage: tahoe debug catalog-shares NODEDIRS.."
 
     def getUsage(self, width=None):
-        t = usage.Options.getUsage(self, width)
+        t = BaseOptions.getUsage(self, width)
         t += """
 Locate all shares in the given node directories, and emit a one-line summary
 of each share. Run it like this:
@@ -577,9 +729,12 @@ def describe_share(abs_sharefile, si_s, shnum_s, now, out):
 
         share_type = "unknown"
         f.seek(m.DATA_OFFSET)
-        if f.read(1) == "\x00":
+        version = f.read(1)
+        if version == "\x00":
             # this slot contains an SMDF share
             share_type = "SDMF"
+        elif version == "\x01":
+            share_type = "MDMF"
 
         if share_type == "SDMF":
             f.seek(m.DATA_OFFSET)
@@ -601,6 +756,35 @@ def describe_share(abs_sharefile, si_s, shnum_s, now, out):
                   (si_s, k, N, datalen,
                    seqnum, base32.b2a(root_hash),
                    expiration, quote_output(abs_sharefile))
+        elif share_type == "MDMF":
+            from allmydata.mutable.layout import MDMFSlotReadProxy
+            fake_shnum = 0
+            # TODO: factor this out with dump_MDMF_share()
+            class ShareDumper(MDMFSlotReadProxy):
+                def _read(self, readvs, force_remote=False, queue=False):
+                    data = []
+                    for (where,length) in readvs:
+                        f.seek(m.DATA_OFFSET+where)
+                        data.append(f.read(length))
+                    return defer.succeed({fake_shnum: data})
+
+            p = ShareDumper(None, "fake-si", fake_shnum)
+            def extract(func):
+                stash = []
+                # these methods return Deferreds, but we happen to know that
+                # they run synchronously when not actually talking to a
+                # remote server
+                d = func()
+                d.addCallback(stash.append)
+                return stash[0]
+
+            verinfo = extract(p.get_verinfo)
+            (seqnum, root_hash, salt_to_use, segsize, datalen, k, N, prefix,
+             offsets) = verinfo
+            print >>out, "MDMF %s %d/%d %d #%d:%s %d %s" % \
+                  (si_s, k, N, datalen,
+                   seqnum, base32.b2a(root_hash),
+                   expiration, quote_output(abs_sharefile))
         else:
             print >>out, "UNKNOWN mutable %s" % quote_output(abs_sharefile)
 
@@ -610,7 +794,7 @@ def describe_share(abs_sharefile, si_s, shnum_s, now, out):
         class ImmediateReadBucketProxy(ReadBucketProxy):
             def __init__(self, sf):
                 self.sf = sf
-                ReadBucketProxy.__init__(self, "", "", "")
+                ReadBucketProxy.__init__(self, None, None, "")
             def __repr__(self):
                 return "<ImmediateReadBucketProxy>"
             def _read(self, offset, size):
@@ -696,7 +880,7 @@ def catalog_shares_one_abbrevdir(si_s, si_dir, now, out, err):
         print >>err, "Error processing %s" % quote_output(si_dir)
         failure.Failure().printTraceback(err)
 
-class CorruptShareOptions(usage.Options):
+class CorruptShareOptions(BaseOptions):
     def getSynopsis(self):
         return "Usage: tahoe debug corrupt-share SHARE_FILENAME"
 
@@ -705,7 +889,7 @@ class CorruptShareOptions(usage.Options):
         ]
 
     def getUsage(self, width=None):
-        t = usage.Options.getUsage(self, width)
+        t = BaseOptions.getUsage(self, width)
         t += """
 Corrupt the given share by flipping a bit. This will cause a
 verifying/downloading client to log an integrity-check failure incident, and
@@ -768,7 +952,7 @@ def corrupt_share(options):
     else:
         # otherwise assume it's immutable
         f = ShareFile(fn)
-        bp = ReadBucketProxy(None, '', '')
+        bp = ReadBucketProxy(None, None, '')
         offsets = bp._parse_offsets(f.read_share_data(0, 0x24))
         start = f._data_offset + offsets["data"]
         end = f._data_offset + offsets["plaintext_hash_tree"]
@@ -776,7 +960,7 @@ def corrupt_share(options):
 
 
 
-class ReplOptions(usage.Options):
+class ReplOptions(BaseOptions):
     def getSynopsis(self):
         return "Usage: tahoe debug repl"
 
@@ -816,7 +1000,50 @@ def trial(config):
     twisted_trial.run()
 
 
-class DebugCommand(usage.Options):
+def fixOptionsClass( (subcmd, shortcut, OptionsClass, desc) ):
+    class FixedOptionsClass(OptionsClass):
+        def getSynopsis(self):
+            t = OptionsClass.getSynopsis(self)
+            i = t.find("Usage: flogtool ")
+            if i >= 0:
+                return "Usage: tahoe debug flogtool " + t[i+len("Usage: flogtool "):]
+            else:
+                return "Usage: tahoe debug flogtool %s [options]" % (subcmd,)
+    return (subcmd, shortcut, FixedOptionsClass, desc)
+
+class FlogtoolOptions(foolscap_cli.Options):
+    def __init__(self):
+        super(FlogtoolOptions, self).__init__()
+        self.subCommands = map(fixOptionsClass, self.subCommands)
+
+    def getSynopsis(self):
+        return "Usage: tahoe debug flogtool (%s) [command options]" % ("|".join([x[0] for x in self.subCommands]))
+
+    def parseOptions(self, all_subargs, *a, **kw):
+        self.flogtool_args = list(all_subargs)
+        return super(FlogtoolOptions, self).parseOptions(self.flogtool_args, *a, **kw)
+
+    def getUsage(self, width=None):
+        t = super(FlogtoolOptions, self).getUsage(width)
+        t += """
+The 'tahoe debug flogtool' command uses the correct imports for this instance
+of Tahoe-LAFS.
+
+Please run 'tahoe debug flogtool SUBCOMMAND --help' for more details on each
+subcommand.
+"""
+        return t
+
+    def opt_help(self):
+        print str(self)
+        sys.exit(0)
+
+def flogtool(config):
+    sys.argv = ['flogtool'] + config.flogtool_args
+    return foolscap_cli.run_flogtool()
+
+
+class DebugCommand(BaseOptions):
     subCommands = [
         ["dump-share", None, DumpOptions,
          "Unpack and display the contents of a share (uri_extension and leases)."],
@@ -826,15 +1053,16 @@ class DebugCommand(usage.Options):
         ["corrupt-share", None, CorruptShareOptions, "Corrupt a share by flipping a bit."],
         ["repl", None, ReplOptions, "Open a Python interpreter."],
         ["trial", None, TrialOptions, "Run tests using Twisted Trial with the right imports."],
+        ["flogtool", None, FlogtoolOptions, "Utilities to access log files."],
         ]
     def postOptions(self):
         if not hasattr(self, 'subOptions'):
             raise usage.UsageError("must specify a subcommand")
     def getSynopsis(self):
-        return "Usage: tahoe debug SUBCOMMAND"
+        return ""
     def getUsage(self, width=None):
-        #t = usage.Options.getUsage(self, width)
-        t = """
+        #t = BaseOptions.getUsage(self, width)
+        t = """Usage: tahoe debug SUBCOMMAND
 Subcommands:
     tahoe debug dump-share      Unpack and display the contents of a share.
     tahoe debug dump-cap        Unpack a read-cap or write-cap.
@@ -843,10 +1071,37 @@ Subcommands:
     tahoe debug corrupt-share   Corrupt a share by flipping a bit.
     tahoe debug repl            Open a Python interpreter.
     tahoe debug trial           Run tests using Twisted Trial with the right imports.
+    tahoe debug flogtool        Utilities to access log files.
 
 Please run e.g. 'tahoe debug dump-share --help' for more details on each
 subcommand.
 """
+        # See ticket #1441 for why we print different information when
+        # run via /usr/bin/tahoe. Note that argv[0] is the full path.
+        if sys.argv[0] == '/usr/bin/tahoe':
+            t += """
+To get branch coverage for the Tahoe test suite (on the installed copy of
+Tahoe), install the 'python-coverage' package and then use:
+
+    python-coverage run --branch /usr/bin/tahoe debug trial
+"""
+        else:
+            t += """
+Another debugging feature is that bin%stahoe allows executing an arbitrary
+"runner" command (typically an installed Python script, such as 'coverage'),
+with the Tahoe libraries on the PYTHONPATH. The runner command name is
+prefixed with '@', and any occurrences of '@tahoe' in its arguments are
+replaced by the full path to the tahoe script.
+
+For example, if 'coverage' is installed and on the PATH, you can use:
+
+    bin%stahoe @coverage run --branch @tahoe debug trial
+
+to get branch coverage for the Tahoe test suite. Or, to run python with
+the -3 option that warns about Python 3 incompatibilities:
+
+    bin%stahoe @python -3 @tahoe command [options]
+""" % (os.sep, os.sep, os.sep)
         return t
 
 subDispatch = {
@@ -857,6 +1112,7 @@ subDispatch = {
     "corrupt-share": corrupt_share,
     "repl": repl,
     "trial": trial,
+    "flogtool": flogtool,
     }