Official Substack MCP Server

� �@�g ���SrS
SK
r
S
SKJr S
SKJr S
SKrS
SKJr S
SKrS
SK r S
SK Jr S
SKJ r Jr \
R"\5
r"SS \ 5rg) z- Medium scraper for Writer Context Protocol. �N)
�datetime)
�List)
�urlparse)
� BeautifulSoup)�BaseScraper�Postc�H�\r
SrS
rSrS\\4Sjr\S\ S\ 4Sj5r Srg) � MediumScraper�zScraper for Medium blogs.�returnc
��# �[
UR5
n
U
RRS
5
R S
5
nU
R S:XaU(aUSO
SnOU
R R S5
SnUR S5
(aUSSnS U3n[RS U35
[R"5
I
Shv
�NnURU5
IShv
�NnUR5 [RSUR35
[R"UR 5
n[RS[#UR$5
35
SSS5IShv
�N /n WR$SUR,H�n [/U R0SR2S5nUR5UR7SSS95
nSn [9U S5(a"[:R<"U R>S5n [EUR5U RF5
U RHUU SS9nU RKU5
M� [RS[#U 5
S35
U $G
N�G
N�G
N !,IShv
�N(d
f  G
N =f
![&a-n[R)S [+U5
35
/sSnA$SnAf
f=f
![@aY [:R<"U R>S5n N�![@a' [RCSU R>35
G
N%f=f
f=f
![&a@n[R)SU RSS5S[+U5
35
SnAG
M�SnAf
f=f
7f
)za Scrape posts from a Medium blog. Returns: A list of Post objects with content and metadata �
/z medium.comr��
.�
@�
Nzhttps://medium.com/feed/@zFetching RSS feed from: z,RSS feed fetched successfully. Status code: zNumber of entries in feed: zError fetching RSS feed: zhtml.parser�
T)� separator�strip� publishedz%a, %d %b %Y %H:%M:%S %zz%a, %d %b %Y %H:%M:%SzCould not parse date: )�title�url�content�date�subtitlezError processing Medium post �link�unknownz: zScraped z posts from Medium)&rr�pathr�split�netloc� startswith�logger�info�httpx�AsyncClient�get�raise_for_status�status_code� feedparser�parse�text�len�entries� Exception�error�str� max_postsrr�value�_clean_content�get_text�hasattrr�strptimer� ValueError�warningrrr�append)�self� parsed_url� path_parts�username�rss_url�client�response�feed�
e�posts�entry�soup�cleaned_text�pub_date�posts �M/Users/jonathanpolitzki/Desktop/Coding/mcp-writer-substack/scrapers/medium.py�scrape�MediumScraper.scrapes����d�h�h�'� ��_�_�*�*�3�/�5�5�c�:� �����,�(2�z�!�}��H�!�(�(�.�.�s�3�A�6�H����s�#�#��
��|�H�.�h�Z�8�� ��K�K�2�7�)�<�=��(�(�*�*�f�!'���G�!4�4���)�)�+����J�8�K_�K_�J`�a�b�!�'�'�� � �6�����9�#�d�l�l�:K�9L�M�N�+�*����\�\�/�4�>�>�2�E� g
�$�U�]�]�1�%5�%;�%;�]�K��#�2�2�4�=�=�3�VZ�=�3[�\�� ���5�+�.�.�W
�#+�#4�#4�U�_�_�F`�#a����-�-�e�k�k�:�� � �(�!�������T�"�33�: ���h�s�5�z�l�*<�=�>���S
+�4�+�*�*�*��� ��L�L�4�S��V�H�=�>��I�� �� &�W
�W
�'/�'8�'8����Ja�'b�H��)�
W
�"�N�N�-C�E�O�O�CT�+U�V�V�
W
�� W
���
g
����<�U�Y�Y�v�y�=Y�<Z�Z\�]`�ab�]c�\d�e�f�f��
g
�s
�BN
� 1J�
I;�J�J�*
I>�+B
J�,J�7
J
�8J�<N
�AL?�1!K�A
L?�(N
�;
J�>
J�

J�J�
J �J�J�N
�J� K�)"K�
K�N
�K�N
� L<�$!L�L?�,L8�3
L<�4L?�7
L8�8L<�<L?�? N � 4N�=N
�N � N
rc
�R�[R"S
SU5R5nU$)z+Remove extra whitespace and normalize text.z\s+r)�re�subr)
rs
rIr3�MediumScraper._clean_contentYs$���&�&���g�.�4�4�6�����N)�__name__� __module__�__qualname__�__firstlineno__�__doc__rrrJ�staticmethodr0r3�__static_attributes__rQrPrIr r s:��#�A
�d�4�j�A
�F�������rPr )rV�loggingr�typingrrM�urllib.parserr)r$�bs4r� scrapers.baserr� getLoggerrRr"r rQrPrI�<module>r_
sE�


�
�
��� �!����+� � � �8� $��K

�K�K

rP