Make Server Version Extraction More Flexible (#778)

#250
#261
#771

Just tried to use this library through Ormar->Databases->AsyncPG against
a Yugabyte cluster and hit issue 771. Looks like this has been a problem
for a while now so going for a complete overhaul of the server version
extraction method. Using a groupdict regex against the version string
allows for much higher flexibility in extracting what we're looking for
and fixes #771 while not breaking any of the existing version patterns.

asyncpg/serverversion.py

@@ -5,53 +5,56 @@
 # the Apache 2.0 License: http://www.apache.org/licenses/LICENSE-2.0
 
 
-from . import types
+import re
+
+from .types import ServerVersion
+
+version_regex = re.compile(
+    r"(Postgre[^\s]*)?\s*"
+    r"(?P<major>[0-9]+)\.?"
+    r"((?P<minor>[0-9]+)\.?)?"
+    r"(?P<micro>[0-9]+)?"
+    r"(?P<releaselevel>[a-z]+)?"
+    r"(?P<serial>[0-9]+)?"
+)
 
 
 def split_server_version_string(version_string):
-    version_string = version_string.strip()
-    if version_string.startswith('PostgreSQL '):
-        version_string = version_string[len('PostgreSQL '):]
-    if version_string.startswith('Postgres-XL'):
-        version_string = version_string[len('Postgres-XL '):]
-
-    # Some distros (e.g Debian) like may inject their branding
-    # into the numeric version string, so make sure to only look
-    # at stuff before the first space.
-    version_string = version_string.split(' ')[0]
-    parts = version_string.strip().split('.')
-    if not parts[-1].isdigit():
-        # release level specified
-        lastitem = parts[-1]
-        levelpart = lastitem.rstrip('0123456789').lower()
-        if levelpart != lastitem:
-            serial = int(lastitem[len(levelpart):])
-        else:
-            serial = 0
-
-        level = levelpart.lstrip('0123456789')
-        if level != levelpart:
-            parts[-1] = levelpart[:-len(level)]
-        else:
-            parts[-1] = 0
-    else:
-        level = 'final'
-        serial = 0
-
-    if int(parts[0]) >= 10:
-        # Since PostgreSQL 10 the versioning scheme has changed.
-        # 10.x really means 10.0.x.  While parsing 10.1
-        # as (10, 1) may seem less confusing, in practice most
-        # version checks are written as version[:2], and we
-        # want to keep that behaviour consistent, i.e not fail
-        # a major version check due to a bugfix release.
-        parts.insert(1, 0)
-
-    versions = [int(p) for p in parts][:3]
-    if len(versions) < 3:
-        versions += [0] * (3 - len(versions))
-
-    versions.append(level)
-    versions.append(serial)
-
-    return types.ServerVersion(*versions)
+    version_match = version_regex.search(version_string)
+
+    if version_match is None:
+        raise ValueError(
+            "Unable to parse Postgres "
+            f'version from "{version_string}"'
+        )
+
+    version = version_match.groupdict()
+    for ver_key, ver_value in version.items():
+        # Cast all possible versions parts to int
+        try:
+            version[ver_key] = int(ver_value)
+        except (TypeError, ValueError):
+            pass
+
+    if version.get("major") < 10:
+        return ServerVersion(
+            version.get("major"),
+            version.get("minor") or 0,
+            version.get("micro") or 0,
+            version.get("releaselevel") or "final",
+            version.get("serial") or 0,
+        )
+
+    # Since PostgreSQL 10 the versioning scheme has changed.
+    # 10.x really means 10.0.x.  While parsing 10.1
+    # as (10, 1) may seem less confusing, in practice most
+    # version checks are written as version[:2], and we
+    # want to keep that behaviour consistent, i.e not fail
+    # a major version check due to a bugfix release.
+    return ServerVersion(
+        version.get("major"),
+        0,
+        version.get("minor") or 0,
+        version.get("releaselevel") or "final",
+        version.get("serial") or 0,
+    )

tests/test_connect.py

@@ -64,6 +64,9 @@ def test_server_version_02(self):
             ("10.1", (10, 0, 1, 'final', 0),),
             ("11.1.2", (11, 0, 1, 'final', 0),),
             ("PostgreSQL 10.1 (Debian 10.1-3)", (10, 0, 1, 'final', 0),),
+            ("PostgreSQL 11.2-YB-2.7.1.1-b0 on x86_64-pc-linux-gnu, "
+             "compiled by gcc (Homebrew gcc 5.5.0_4) 5.5.0, 64-bit",
+             (11, 0, 2, "final", 0),),
         ]
         for version, expected in versions:
             result = split_server_version_string(version)