Optionally pass the Beautiful Soup object to constructors

linkpreview/linkpreview.py

@@ -1,4 +1,7 @@
 from os.path import dirname
+from typing import Union
+
+from bs4 import BeautifulSoup
 
 from linkpreview.link import Link
 from linkpreview.preview import (
@@ -14,13 +17,13 @@
 
 
 class LinkPreview:
-    def __init__(self, link: Link, parser: str = PARSER):
+    def __init__(self, link: Link, parser: Union[str, None] = PARSER, soup: Union[BeautifulSoup, None] = None):
         self.link = link
-        self.generic = Generic(link, parser)
-        self.opengraph = OpenGraph(link, parser)
-        self.twitter = TwitterCard(link, parser)
-        self.microdata = Microdata(link, parser)
-        self.jsonld = JsonLd(link, parser)
+        self.generic = Generic(link, parser, soup)
+        self.opengraph = OpenGraph(link, parser, soup)
+        self.twitter = TwitterCard(link, parser, soup)
+        self.microdata = Microdata(link, parser, soup)
+        self.jsonld = JsonLd(link, parser, soup)
 
     @LazyAttribute
     def sources(self):

linkpreview/preview/base.py

@@ -1,3 +1,5 @@
+from typing import Union
+
 from bs4 import BeautifulSoup
 from linkpreview.link import Link
 
@@ -7,9 +9,16 @@ class PreviewBase(object):  # pragma: nocover
     Base for all web preview.
     """
 
-    def __init__(self, link: Link, parser: str):
+    def __init__(self, link: Link, parser: Union[str, None] = None, soup: Union[BeautifulSoup, None] = None):
+        if parser and soup:
+            raise Exception(
+                'Only one of `parser` or `soup` argument must be provided to PreviewBase')
+
         self.link = link
-        self._soup = BeautifulSoup(self.link.content, parser)
+        if soup:
+            self._soup = soup
+        else:
+            self._soup = BeautifulSoup(self.link.content, parser)
 
     @property
     def title(self):

tests/test_generic.py

@@ -1,4 +1,5 @@
 import pytest
+from bs4 import BeautifulSoup
 
 from linkpreview import Link
 from linkpreview.preview import Generic
@@ -115,6 +116,23 @@ def test_generic(tin, tout):
         assert getattr(preview, key) == tout[key]
 
 
+def test_bs4_object():
+    link = Link("http://localhost")
+    content = get_sample("generic/title.html")
+    bs_content = BeautifulSoup(content, "html.parser")
+    preview = Generic(link, soup=bs_content)
+    assert preview.title == "This title is at the title tag."
+    assert preview.description is None
+    assert preview.image is None
+
+
+def test_invalid_arguments():
+    link = Link("http://localhost")
+    bs_content = BeautifulSoup()
+    with pytest.raises(Exception):
+        Generic(link, parser="html.parser", soup=bs_content)
+
+
 @pytest.mark.parametrize(
     "tin, tout",
     (