Verify config files against schema

2 files changed, 133 insertions, 7 deletions

diff --git a/tools/compileconfig.py b/tools/compileconfig.py
index c2dfcb3..10fc3d6 100755
--- a/tools/compileconfig.py
+++ b/tools/compileconfig.py
@@ -550,6 +550,65 @@ def printnodenames(config):
 
     print " ".join(frontendnodenames|storagenodenames|signingnodenames|mergenodenames|statusservernodenames)
 
+localconfigschema = [
+    ("nodename", "string", "nodename"),
+    ("frontendaddress", "string", "ip address"),
+    ("ctapiaddress", "string", "ip address"),
+    ("storageaddress", "string", "ip address"),
+    ("signingaddress", "string", "ip address"),
+    ("mergeaddress", "string", "ip address"),
+    ("publichttpaddress", "string", "ip address"),
+    ("configurl", "string", "url"),
+    ("logadminkey", "string", "key"),
+    ("dbbackend", "string", ["permdb", "fsdb"]),
+    ("paths/configdir", "string", "path"),
+    ("paths/db", "string", "path"),
+    ("paths/https_cacertfile", "string", "path"),
+    ("paths/https_certfile", "string", "path"),
+    ("paths/https_keyfile", "string", "path"),
+    ("paths/knownroots", "string", "path"),
+    ("paths/logpublickey", "string", "path"),
+    ("paths/privatekeys", "string", "path"),
+    ("paths/public_cacertfile", "string", "path"),
+    ("paths/publickeys", "string", "path"),
+    ("paths/verifycert_bin", "string", "path"),
+    ("paths/mergedb", "string", "path"),
+    ("paths/logprivatekey", "string", "path"),
+    ("ratelimits/add_chain", "string", "rate"),
+    ("merge/min-delay", "integer", "seconds"),
+    ("merge/backup-window-size", "integer", "number of entries"),
+    ("merge/backup-sendlog-chunksize", "integer", "number of entries"),
+    ("merge/backup-sendentries-chunksize", "integer", "number of entries"),
+    ("merge/dist-window-size", "integer", "number of entries"),
+    ("merge/dist-sendlog-chunksize", "integer", "number of entries"),
+    ("merge/dist-sendentries-chunksize", "integer", "number of entries"),
+]
+
+globalconfigschema = [
+    ("frontendnodes/[]/name", "string", "nodename"),
+    ("frontendnodes/[]/address", "string", "ip address"),
+    ("frontendnodes/[]/publicaddress", "string", "ip address"),
+    ("mergenodes/[]/name", "string", "nodename"),
+    ("mergenodes/[]/address", "string", "ip address"),
+    ("signingnodes/[]/name", "string", "nodename"),
+    ("signingnodes/[]/address", "string", "ip address"),
+    ("storagenodes/[]/name", "string", "nodename"),
+    ("storagenodes/[]/address", "string", "ip address"),
+    ("statusservers/[]/name", "string", "nodename"),
+    ("statusservers/[]/address", "string", "ip address"),
+    ("statusservers/[]/publicaddress", "string", "ip address"),
+    ("apikeys/[]/nodename", "string", "nodename"),
+    ("apikeys/[]/publickey", "string", "key"),
+    ("baseurl", "string", "url"),
+    ("primarymergenode", "string", "nodename"),
+    ("storage-quorum-size", "integer", "number of nodes"),
+    ("backup-quorum-size", "integer", "number of nodes"),
+    ("logpublickey", "string", "key"),
+    ("cafingerprint", "string", "fingerprint"),
+    ("version", "integer", "version"),
+    ("mmd", "integer", "seconds"),
+]
+
 def main():
     parser = argparse.ArgumentParser(description="")
     parser.add_argument('--config', help="System configuration")
@@ -564,17 +623,17 @@ def main():
         sys.exit(1)
 
     if args.testmakefile:
-        config = readconfig.read_config(args.config)
+        config = readconfig.read_config(args.config, schema=globalconfigschema)
         gen_testmakefile(config, args.testmakefile)
     elif args.testshellvars:
-        config = readconfig.read_config(args.config)
+        config = readconfig.read_config(args.config, schema=globalconfigschema)
         gen_testmakefile(config, args.testshellvars, shellvars=True)
     elif args.getnodenames:
-        config = readconfig.read_config(args.config)
+        config = readconfig.read_config(args.config, schema=globalconfigschema)
         printnodenames(config)
     elif args.localconfig:
-        localconfig = readconfig.read_config(args.localconfig)
-        config = readconfig.verify_and_read_config(args.config, localconfig["logadminkey"])
+        localconfig = readconfig.read_config(args.localconfig, schema=localconfigschema)
+        config = readconfig.verify_and_read_config(args.config, localconfig["logadminkey"], schema=globalconfigschema)
 
         gen_config(localconfig["nodename"], config, localconfig)
     else:
diff --git a/tools/readconfig.py b/tools/readconfig.py
index 028e319..15b9c61 100644
--- a/tools/readconfig.py
+++ b/tools/readconfig.py
@@ -50,14 +50,81 @@ def verify_config(rawconfig, signature, publickey_base64, filename):
 
     return errorhandlify(yaml.load(io.BytesIO(rawconfig), yaml.SafeLoader), filename)
 
-def verify_and_read_config(filename, publickey_base64):
+def verify_and_read_config(filename, publickey_base64, schema=None):
     rawconfig = open(filename).read()
     signature = open(filename + ".sig").read()
 
     verify_config(rawconfig, signature, publickey_base64, filename)
     config = yaml.load(io.BytesIO(rawconfig), yaml.SafeLoader)
+    if schema:
+        check_config_schema(config, schema)
     return errorhandlify(config, filename)
 
-def read_config(filename):
+def insert_schema_path(schema, path, datatype, highleveldatatype):
+    if len(path) == 1:
+        schema[path[0]] = (datatype, highleveldatatype)
+    else:
+        if path[0] not in schema:
+            schema[path[0]] = {}
+        insert_schema_path(schema[path[0]], path[1:], datatype, highleveldatatype)
+
+def transform_schema(in_schema):
+    schema = {}
+    for (rawpath, datatype, highleveldatatype) in in_schema:
+        path = rawpath.split("/")
+        insert_schema_path(schema, path, datatype, highleveldatatype)
+    return schema
+
+def check_config_schema(config, schema):
+    transformed_schema = transform_schema(schema)
+    error = check_config_schema_part(config, transformed_schema)
+    if error:
+        print >>sys.stderr, "error:", error
+        sys.exit(1)
+
+def check_config_schema_part(term, schema, path=[]):
+    joined_path = render_path(path)
+    if isinstance(term, basestring):
+        (schema_lowlevel, schema_highlevel) = schema
+        if schema_lowlevel != "string":
+            return "expected %s at %s, not a string" % (schema_lowlevel, joined_path,)
+        return None
+    elif isinstance(term, int):
+        (schema_lowlevel, schema_highlevel) = schema
+        if schema_lowlevel != "integer":
+            return "expected %s at %s, not an integer" % (schema_lowlevel, joined_path,)
+        return None
+    elif isinstance(term, dict):
+        if not isinstance(schema, dict):
+            return "expected %s at %s, not a key" % (schema, joined_path,)
+        for k, v in term.items():
+            schema_part = schema.get(k)
+            if schema_part == None and len(schema.keys()) == 1 and schema.keys()[0].startswith("*"):
+                schema_part = schema[schema.keys()[0]]
+            if schema_part == None:
+                return "configuration key '%s' at %s unknown" % (k, joined_path)
+            result = check_config_schema_part(v, schema_part, path + [k])
+            if result:
+                return result
+        return None
+    elif isinstance(term, list):
+        if not isinstance(schema, dict):
+            return "expected %s at %s, not a list" % (schema, joined_path,)
+        schema_part = schema.get("[]")
+        if schema_part == None:
+            return "expected dict at %s, not a list" % (joined_path,)
+        for i, e in enumerate(term, start=1):
+            result = check_config_schema_part(e, schema_part, path + ["item %d" % i])
+            if result:
+                return result
+        return None
+    else:
+        print >>sys.stderr, "unknown type", type(term)
+        sys.exit(1)
+
+
+def read_config(filename, schema=None):
     config = yaml.load(open(filename), yaml.SafeLoader)
+    if schema:
+        check_config_schema(config, schema)
     return errorhandlify(config, filename)