From 1fa5918b66585f8485d741651246b8d256e678da Mon Sep 17 00:00:00 2001
From: Villu Ruusmann <villu.ruusmann@gmail.com>
Date: Mon, 14 Oct 2024 13:37:14 +0300
Subject: [PATCH] Added 'MatcherTransformer.re_flavour' and
 'ReplaceTransformer.re_flavour' attributes

See https://github.com/jpmml/sklearn2pmml/issues/228
---
 sklearn2pmml/preprocessing/__init__.py       | 10 +++++----
 sklearn2pmml/preprocessing/regex.py          | 23 +++++++++++++++-----
 sklearn2pmml/preprocessing/tests/__init__.py |  4 ++--
 3 files changed, 26 insertions(+), 11 deletions(-)

diff --git a/sklearn2pmml/preprocessing/__init__.py b/sklearn2pmml/preprocessing/__init__.py
index 971cc7b..50358bc 100644
--- a/sklearn2pmml/preprocessing/__init__.py
+++ b/sklearn2pmml/preprocessing/__init__.py
@@ -574,8 +574,9 @@ def transform(self, X):
 class MatchesTransformer(BaseEstimator, TransformerMixin):
 	"""Match RE pattern."""
 
-	def __init__(self, pattern):
+	def __init__(self, pattern, re_flavour = None):
 		self.pattern = pattern
+		self.re_flavour = re_flavour
 
 	def fit(self, X, y = None):
 		to_1d(X)
@@ -583,7 +584,7 @@ def fit(self, X, y = None):
 
 	def transform(self, X):
 		X1d = to_1d(X)
-		regex_engine = make_regex_engine(self.pattern)
+		regex_engine = make_regex_engine(self.pattern, self.re_flavour)
 		func = lambda x: bool(regex_engine.matches(x))
 		Xt = eval_rows(X1d, func, shape = X.shape)
 		return Xt
@@ -591,9 +592,10 @@ def transform(self, X):
 class ReplaceTransformer(BaseEstimator, TransformerMixin):
 	"""Replace all RE pattern matches."""
 
-	def __init__(self, pattern, replacement):
+	def __init__(self, pattern, replacement, re_flavour = None):
 		self.pattern = pattern
 		self.replacement = replacement
+		self.re_flavour = re_flavour
 
 	def fit(self, X, y = None):
 		to_1d(X)
@@ -601,7 +603,7 @@ def fit(self, X, y = None):
 
 	def transform(self, X):
 		X1d = to_1d(X)
-		regex_engine = make_regex_engine(self.pattern)
+		regex_engine = make_regex_engine(self.pattern, self.re_flavour)
 		func = lambda x: regex_engine.replace(self.replacement, x)
 		Xt = eval_rows(X1d, func, shape = X.shape)
 		return Xt
diff --git a/sklearn2pmml/preprocessing/regex.py b/sklearn2pmml/preprocessing/regex.py
index 74bde44..c2eb6cd 100644
--- a/sklearn2pmml/preprocessing/regex.py
+++ b/sklearn2pmml/preprocessing/regex.py
@@ -58,9 +58,22 @@ def matches(self, x):
 	def replace(self, replacement, x):
 		return self.pattern_.substitute(replacement, x)
 
-def make_regex_engine(pattern):
-	try:
+def make_regex_engine(pattern, re_flavour):
+	if re_flavour is None:
+		try:
+			import pcre2
+
+			re_flavour = "pcre2"
+		except ImportError:
+			warnings.warn("Perl Compatible Regular Expressions (PCRE) library is not available, falling back to built-in Regular Expressions (RE) library. Transformation results might not be reproducible between Python and PMML environments when using more complex patterns", Warning)
+			re_flavour = "re"
+
+	if re_flavour == "pcre":
 		return PCREEngine(pattern)
-	except ImportError:
-		warnings.warn("Perl Compatible Regular Expressions (PCRE) library is not available, falling back to built-in Regular Expressions (RE) library. Transformation results might not be reproducible between Python and PMML environments when using more complex patterns", Warning)
-		return REEngine(pattern)
\ No newline at end of file
+	elif re_flavour == "pcre2":
+		return PCRE2Engine(pattern)
+	elif re_flavour == "re":
+		return REEngine(pattern)
+	else:
+		re_flavours = ["pcre", "pcre2", "re"]
+		raise ValueError("Regular Expressions flavour {0} not in {1}".format(re_flavour, re_flavours))
diff --git a/sklearn2pmml/preprocessing/tests/__init__.py b/sklearn2pmml/preprocessing/tests/__init__.py
index c9e6fbe..a78704f 100644
--- a/sklearn2pmml/preprocessing/tests/__init__.py
+++ b/sklearn2pmml/preprocessing/tests/__init__.py
@@ -648,14 +648,14 @@ class MatchesTransformerTest(TestCase):
 
 	def test_transform(self):
 		X = numpy.asarray(["January", "February", "March", "April", "May", "June", "July", "August", "September", "October", "November", "December"])
-		transformer = MatchesTransformer("ar?y")
+		transformer = MatchesTransformer("ar?y", re_flavour = "re")
 		self.assertEqual([True, True, False, False, True, False, False, False, False, False, False, False], transformer.transform(X).tolist())
 
 class ReplaceTransformerTest(TestCase):
 
 	def test_transform(self):
 		X = numpy.asarray(["A", "B", "BA", "BB", "BAB", "ABBA", "BBBB"])
-		transformer = ReplaceTransformer("B+", "c")
+		transformer = ReplaceTransformer("B+", "c", re_flavour = "re")
 		self.assertEqual(["A", "c", "cA", "c", "cAc", "AcA", "c"], transformer.transform(X).tolist())
 		vectorizer = CountVectorizer()
 		pipeline = make_pipeline(transformer, vectorizer)