DomainExtractor object is now shared when the default domain list is used

Author: Gal Ben David

MANIFEST.in

@@ -1,4 +1,5 @@
 include README.md
 include images/logo.png
 graft tests
-recursive-include src *.h
+recursive-include src *.h *.cpp
+recursive-include pydomainextractor *.py *.pyi

pydomainextractor/__init__.py

@@ -0,0 +1,21 @@
+import typing
+import pydomainextractor.extractor
+
+
+class DomainExtractor:
+    '''
+    PyDomainExtractor is a highly optimized Domain Name Extraction library written in C++
+    '''
+    engine: typing.Optional[pydomainextractor.extractor.DomainExtractor] = None
+
+    def __new__(
+        cls,
+        suffix_list_data: typing.Optional[str] = None,
+    ):
+        if suffix_list_data is None:
+            if DomainExtractor.engine is None:
+                DomainExtractor.engine = pydomainextractor.extractor.DomainExtractor()
+
+            return DomainExtractor.engine
+        else:
+            return pydomainextractor.extractor.DomainExtractor(suffix_list_data)

pydomainextractor/pydomainextractor.pyi

@@ -0,0 +1,27 @@
+import typing
+
+
+class DomainExtractor:
+    def __init__(
+        self,
+        suffix_list_data: typing.Optional[str] = None,
+    ) -> None: ...
+
+    def extract(
+        self,
+        domain: str,
+    ) -> typing.dict[str, str]: ...
+
+    def extract_from_url(
+        self,
+        url: str,
+    ) -> typing.dict[str, str]: ...
+
+    def is_valid_domain(
+        self,
+        domain: str,
+    ) -> bool: ...
+
+    def get_tld_list(
+        self,
+    ) -> typing.List[str]: ...

setup.py

@@ -5,7 +5,7 @@
 
 setuptools.setup(
     name='PyDomainExtractor',
-    version='0.8.5',
+    version='0.9.0',
     author='Gal Ben David',
     author_email='[email protected]',
     url='https://github.com/Intsights/PyDomainExtractor',
@@ -29,13 +29,15 @@
     install_requires=[],
     package_data={},
     include_package_data=True,
+    packages=setuptools.find_packages(),
+    ext_package='pydomainextractor',
     ext_modules=[
         setuptools.Extension(
-            name='pydomainextractor',
+            name='extractor',
             sources=glob.glob(
                 pathname=os.path.join(
                     'src',
-                    'pydomainextractor.cpp',
+                    'extractor.cpp',
                 ),
             ),
             language='c++',

src/pydomainextractor.cpp src/extractor.cppsrc/pydomainextractor.cpp renamed to src/extractor.cpp

@@ -560,23 +560,23 @@ static PyTypeObject DomainExtractorType = {
 };
 
 
-static struct PyModuleDef pydomainextractor_definition = {
+static struct PyModuleDef extractor_definition = {
     PyModuleDef_HEAD_INIT,
-    "pydomainextractor",
+    "extractor",
     "Extracting domain strings into their parts",
     -1,
     NULL,
 };
 
 
 PyMODINIT_FUNC
-PyInit_pydomainextractor(void) {
+PyInit_extractor(void) {
     PyObject *m;
     if (PyType_Ready(&DomainExtractorType) < 0) {
         return NULL;
     }
 
-    m = PyModule_Create(&pydomainextractor_definition);
+    m = PyModule_Create(&extractor_definition);
     if (m == NULL) {
         return NULL;
     }