Allow libcdb search offline database

pwnlib/args.py

@@ -159,6 +159,10 @@ def STDERR(v):
     """Sends logging to ``stderr`` by default, instead of ``stdout``"""
     context.log_console = sys.stderr
 
+def LOCAL_LIBCDB(v):
+    """Sets local libcdb path"""
+    context.defaults['local_libcdb'] = v
+
 hooks = {
     'LOG_LEVEL': LOG_LEVEL,
     'LOG_FILE': LOG_FILE,
@@ -170,6 +174,7 @@ def STDERR(v):
     'NOASLR': NOASLR,
     'NOPTRACE': NOPTRACE,
     'STDERR': STDERR,
+    'LOCAL_LIBCDB': LOCAL_LIBCDB,
 }
 
 def initialize():

pwnlib/commandline/libcdb.py

@@ -57,6 +57,20 @@
     help = 'Do NOT attempt to unstrip the libc binary with debug symbols from a debuginfod server'
 )
 
+lookup_parser.add_argument(
+    '--no-offline',
+    action = 'store_true',
+    default = False,
+    help = 'Disable offline libcdb search mode'
+)
+
+lookup_parser.add_argument(
+    '--no-online',
+    action = 'store_true',
+    default = False,
+    help = 'Disable online libcdb search mode'
+)
+
 hash_parser = libc_commands.add_parser(
     'hash',
     help = 'Display information of a libc version given an unique hash',
@@ -100,6 +114,20 @@
     help = 'Do NOT attempt to unstrip the libc binary with debug symbols from a debuginfod server'
 )
 
+hash_parser.add_argument(
+    '--no-offline',
+    action = 'store_true',
+    default = False,
+    help = 'Disable offline libcdb search mode'
+)
+
+hash_parser.add_argument(
+    '--no-online',
+    action = 'store_true',
+    default = False,
+    help = 'Disable online libcdb search mode'
+)
+
 file_parser = libc_commands.add_parser(
     'file',
     help = 'Dump information about a libc binary',
@@ -136,18 +164,71 @@
 
 common_symbols = ['dup2', 'printf', 'puts', 'read', 'system', 'write']
 
-def find_libc(params):
+
+def find_in_online_mode(params):
     import requests
-    url    = "https://libc.rip/api/find"
+    url = "https://libc.rip/api/find"
     result = requests.post(url, json=params, timeout=20)
     log.debug('Request: %s', params)
     log.debug('Result: %s', result.json())
     if result.status_code != 200 or len(result.json()) == 0:
         log.failure("Could not find libc for %s on libc.rip", params)
         return []
-
     return result.json()
 
+
+def find_in_offline_mode(params):
+    # lookup parser
+    if params.get("symbols"):
+        matching_libcs = libcdb.find_local_libc(params)
+        return matching_libcs if matching_libcs else []
+
+    # hash parser
+    hash_type, hash_value = list(params.items())[0]
+
+    local_db = libcdb._fetch_local_database_path()
+    if not local_db:
+        log.warn_once("The environment variable `PWNLIB_LOCAL_LIBCDB` or `context.local_libcdb` is not configured.")
+        return []
+
+    db_path = Path(local_db) / "db"
+    libs_id = None
+
+    if hash_type == "id":
+        libs_id = hash_value
+    else:
+        if hash_type == "buildid":
+            hash_type = "build_id"
+
+        libc_path = libcdb.search_by_hash(hash_value, hash_type, unstrip=False, offline=True)
+        if libc_path:
+            libs_id = read(libc_path + ".id").decode()
+
+    if libs_id:
+        libc_path = db_path / ("%s.so" % libs_id)
+        symbol_path = db_path / ("%s.symbols" % libs_id)
+
+        syms = libcdb.get_libc_symbols(symbol_path)
+        return [libcdb.get_libc_info(db_path, libc_path.stem, syms)]
+
+    return []
+
+
+def find_libc(params, offline=True, online=True):
+    offline_matching = find_in_offline_mode(params) if offline else []
+    online_matching = find_in_online_mode(params) if online else []
+    log.debug("Offline result: %s, Online result: %s", offline_matching, online_matching)
+
+    matching_id = []
+    matching_libcs = []
+    for x in offline_matching + online_matching:
+        if x["id"] not in matching_id:
+            matching_id.append(x["id"])
+            matching_libcs.append(x)
+
+    return matching_libcs
+
+
 def print_libc(libc):
     log.info('%s', text.red(libc['id']))
     log.indented('\t%-20s %s', text.green('BuildID:'), libc['buildid'])
@@ -158,13 +239,10 @@ def print_libc(libc):
     for symbol in libc['symbols'].items():
         log.indented('\t%25s = %s', symbol[0], symbol[1])
 
-def handle_remote_libc(args, libc):
-    print_libc(libc)
+def fetch_libc(args, libc):
     if args.download_libc:
-        path = libcdb.search_by_build_id(libc['buildid'], args.unstrip)
+        path = libcdb.search_by_build_id(libc['buildid'], args.unstrip, not args.no_offline, not args.no_online)
         if path:
-            if args.unstrip:
-                libcdb.unstrip_libc(path)
             shutil.copy(path, './{}.so'.format(libc['id']))
 
 def translate_offset(offs, args, exe):
@@ -196,25 +274,27 @@ def main(args):
         if len(pairs) % 2 != 0:
             log.failure('Uneven number of arguments. Please provide "symbol offset" pairs')
             return
-        
+
         symbols = {pairs[i]:pairs[i+1] for i in range(0, len(pairs), 2)}
-        matched_libcs = find_libc({'symbols': symbols})
+        matched_libcs = find_libc({'symbols': symbols}, not args.no_offline, not args.no_online)
         for libc in matched_libcs:
-            handle_remote_libc(args, libc)
+            print_libc(libc)
+            fetch_libc(args, libc)
 
     elif args.libc_command == 'hash':
         for hash_value in args.hash_value:
-            matched_libcs = find_libc({args.hash_type: hash_value})
+            matched_libcs = find_libc({args.hash_type: hash_value}, args.offline)
             for libc in matched_libcs:
-                handle_remote_libc(args, libc)
+                print_libc(libc)
+                fetch_libc(args, libc)
 
     elif args.libc_command == 'file':
         from hashlib import md5, sha1, sha256
         for file in args.files:
             if not os.path.exists(file) or not os.path.isfile(file):
                 log.failure('File does not exist %s', args.file)
                 continue
-            
+
             if args.unstrip:
                 libcdb.unstrip_libc(file)
 

pwnlib/context/__init__.py

@@ -360,6 +360,7 @@ class ContextType(object):
         'endian': 'little',
         'gdbinit': "",
         'kernel': None,
+        'local_libcdb': "",
         'log_level': logging.INFO,
         'log_file': _devnull(),
         'log_console': sys.stdout,
@@ -1070,6 +1071,10 @@ def log_console(self, stream):
             stream = open(stream, 'wt')
         return stream
 
+    @_validator
+    def local_libcdb(self, path):
+        return path
+
     @property
     def mask(self):
         return (1 << self.bits) - 1