get rid of duplicated code

Author: jswhit2

test/parse_clean_version.py

@@ -0,0 +1,16 @@
+from packaging.version import Version, InvalidVersion
+import re
+
+def parse_clean_version(version_str: str) -> Version:
+    # Normalize "-development" or any variation like "-dev" to ".dev0"
+    cleaned = re.sub(r'[-_.]?development', '.dev0', version_str, flags=re.IGNORECASE)
+    # Catch cases where it's just raw text hyphens (e.g., 1.0.0-dev -> 1.0.0.dev0)
+    cleaned = re.sub(r'-dev(\d*)', r'.dev\1', cleaned, flags=re.IGNORECASE)
+    try:
+        return Version(cleaned)
+    except InvalidVersion:
+        # Fallback regex to pull just the numerical digits if it's still breaking
+        match = re.search(r'(\d+(\.\d+)+)', version_str)
+        if match:
+            return Version(match.group(1))
+        raise

test/test_diskless.py

@@ -1,24 +1,10 @@
 import unittest, os, tempfile
-from packaging.version import Version, InvalidVersion
+from packaging.version import Version
+from parse_clean_version import parse_clean_version
 import numpy as np
 from numpy.random.mtrand import uniform
 from numpy.testing import assert_array_equal, assert_array_almost_equal
 import netCDF4
-import re
-
-def _parse_clean_version(version_str: str) -> Version:
-    # Normalize "-development" or any variation like "-dev" to ".dev0"
-    cleaned = re.sub(r'[-_.]?development', '.dev0', version_str, flags=re.IGNORECASE)
-    # Catch cases where it's just raw text hyphens (e.g., 1.0.0-dev -> 1.0.0.dev0)
-    cleaned = re.sub(r'-dev(\d*)', r'.dev\1', cleaned, flags=re.IGNORECASE)
-    try:
-        return Version(cleaned)
-    except InvalidVersion:
-        # Fallback regex to pull just the numerical digits if it's still breaking
-        match = re.search(r'(\d+(\.\d+)+)', version_str)
-        if match:
-            return Version(match.group(1))
-        raise
 
 # rudimentary test of diskless file capability.
 
@@ -33,7 +19,7 @@ def _parse_clean_version(version_str: str) -> Version:
 
 
 @unittest.skipIf(
-    _parse_clean_version(netCDF4.__netcdf4libversion__) < Version('4.2.1')
+    parse_clean_version(netCDF4.__netcdf4libversion__) < Version('4.2.1')
     or netCDF4.__has_parallel4_support__
     or netCDF4.__has_pnetcdf_support__,
     "no diskless support",

test/test_masked5.py

@@ -1,28 +1,14 @@
 import unittest
 import os
 import tempfile
-import re
-from packaging.version import Version, InvalidVersion
+from packaging.version import Version
+from parse_clean_version import parse_clean_version
 
 import numpy as np
 from numpy import ma
 from numpy.testing import assert_array_equal
 from netCDF4 import Dataset, __netcdf4libversion__
 
-def _parse_clean_version(version_str: str) -> Version:
-    # Normalize "-development" or any variation like "-dev" to ".dev0"
-    cleaned = re.sub(r'[-_.]?development', '.dev0', version_str, flags=re.IGNORECASE)
-    # Catch cases where it's just raw text hyphens (e.g., 1.0.0-dev -> 1.0.0.dev0)
-    cleaned = re.sub(r'-dev(\d*)', r'.dev\1', cleaned, flags=re.IGNORECASE)
-    try:
-        return Version(cleaned)
-    except InvalidVersion:
-        # Fallback regex to pull just the numerical digits if it's still breaking
-        match = re.search(r'(\d+(\.\d+)+)', version_str)
-        if match:
-            return Version(match.group(1))
-        raise
-
 # Test use of vector of missing values.
 
 class VectorMissingValues(unittest.TestCase):
@@ -71,7 +57,7 @@ def test_scaled(self):
         # issue 730
         # this part fails with netcdf 4.1.3
         # a bug in vlen strings?
-        if _parse_clean_version(__netcdf4libversion__) >= Version('4.4.0'):
+        if parse_clean_version(__netcdf4libversion__) >= Version('4.4.0'):
             assert v2[0] == 'first'
             assert v2[1] == '<missing>'
 

test/test_vlen.py

@@ -2,25 +2,11 @@
 import unittest
 import os
 import tempfile
-import re
-from packaging.version import Version, InvalidVersion
+from packaging.version import Version
 from netCDF4 import Dataset
 import numpy as np
 from numpy.testing import assert_array_equal
-
-def _parse_clean_version(version_str: str) -> Version:
-    # Normalize "-development" or any variation like "-dev" to ".dev0"
-    cleaned = re.sub(r'[-_.]?development', '.dev0', version_str, flags=re.IGNORECASE)
-    # Catch cases where it's just raw text hyphens (e.g., 1.0.0-dev -> 1.0.0.dev0)
-    cleaned = re.sub(r'-dev(\d*)', r'.dev\1', cleaned, flags=re.IGNORECASE)
-    try:
-        return Version(cleaned)
-    except InvalidVersion:
-        # Fallback regex to pull just the numerical digits if it's still breaking
-        match = re.search(r'(\d+(\.\d+)+)', version_str)
-        if match:
-            return Version(match.group(1))
-        raise
+from parse_clean_version import parse_clean_version
 
 FILE_NAME = tempfile.NamedTemporaryFile(suffix='.nc', delete=False).name
 VL_NAME = 'vlen_type'
@@ -179,7 +165,7 @@ class VlenAppendTestCase(unittest.TestCase):
     def setUp(self):
 
         import netCDF4
-        if _parse_clean_version(netCDF4.__netcdf4libversion__) < Version("4.4.1"):
+        if parse_clean_version(netCDF4.__netcdf4libversion__) < Version("4.4.1"):
             self.skip = True
             try:
                 self.skipTest("This test requires NetCDF 4.4.1 or later.")