server.cpython-310.pyc•19.9 kB
o
PK�hk| � @ s� d Z ddlZddlZddlZddlZddlZddlZddlZddlm Z ddl
m Z m Z m
Z
mZ ddlmZmZ ddlmZ zddlZW n
eyQ ddlZY nw ddlmZmZ ddlmZ ddl Z ddl m Z ej ej!d � e�"e#�Z$G d
d � d �Z%G d d
� d
�Z&G dd� d�Z'ed�Z(eG dd� d��Z)G dd� d�Z*G dd� d�Z+e(�,� de-e j. fdd��Z/e(�0� de1de2de-e j3 fd d ��Z4d d � Z5e#d kr�e�6e5� � dS dS )!z�
Video Downloader MCP Server
Exposes video downloading and analysis capabilities as MCP tools, allowing
LLMs to intelligently orchestrate video extraction workflows.
� N)�Path)�Dict�List�Optional�Any)�urljoin�urlparse)� dataclass)�Server�NotificationOptions)�InitializationOptions)�levelc @ sh e Zd ZdZddee fdd�Zdeeef fdd�Z d ed
edefd d �Z
d
d� Z ddefdd�Z dS )�SecureConfigManagerz4Manages secure configuration loading from TOML filesN� config_pathc C sJ |d u rt �� d d }|jddd� |d | _nt |�| _| �� | _d S )Nz.configzvideo-downloader-mcpT)�parents�exist_ok� config.toml)r �home�mkdirr � _load_config�config)�selfr Z
config_dir� r �@/Users/slowbro/Workspaces/Sandbox/video-downloader-mcp/server.py�__init__& s
z SecureConfigManager.__init__�returnc
C s� ddiddg d�dd�ddd d
�ddd �d �}| j �� rdz"t| j d
��
}t�|�}W d � n1 s3w Y | �||�W S tyc } zt�d| j � d|� �� t� d� |W Y d}~S d}~ww | �
� |S )z6Load configuration from TOML file with secure defaults�defaultz~/video-downloaderT�� )Zmp4ZwebmZmkvZaviZmovZm4aZmp3ZaacZoggZwavZvttZsrtZassZssa)Z enforce_location_restrictions�max_filename_length�allowed_extensionsZblock_path_traversalzbest[height<=1080]�%(title)s.%(ext)sr )Zdefault_formatZdefault_filename_templateZmax_download_size)Zlog_security_eventsZ
log_downloads)�download_locationsZsecurityZytdlp�logging�rbNzError loading config from z: zUsing default configuration) r �exists�open�tomllib�load� _deep_merge� Exception�logger�error�info�_create_default_config)r Zdefault_config�fZ user_config�er r r r 1 s: �����
�
��z SecureConfigManager._load_config�base�updatec C s\ |� � }|�� D ]#\}}||v r't|| t�r't|t�r'| �|| |�||< q|||< q|S )zDeep merge two dictionaries)�copy�items�
isinstance�dictr( )r r0 r1 �result�key�valuer r r r( W s
z SecureConfigManager._deep_mergec
C s� z,t t�j}|d }|�� r%ddl}|�|| j� t�d| j� �� W dS t� d� W dS t
yG } zt� d|� �� W Y d}~dS d}~ww )z!Create default configuration filer r Nz!Created default configuration at z7Template config.toml not found, using built-in defaultsz!Failed to create default config: ) r �__file__�parentr$ �shutil�copy2r r* r, �warningr) r+ )r Z
script_dirZtemplate_configr; r/ r r r r- a s
��z*SecureConfigManager._create_default_config�pathc C s@ |� d�}| j}|D ]}t|t�r||v r|| }q
| S |S )z#Get config value using dot notation�.)�splitr r4 r5 )r r> r �keysr8 r7 r r r �getq s
zSecureConfigManager.get�N)
�__name__�
__module__� __qualname__�__doc__r �strr r r r r( r- rB r r r r r # s &
r c
@ sp e Zd ZdZdefdd�Zddededed eee e e e f fd
d �Z
ded efd d
�Z ded efdd�Z dS )�
PathValidatorz/Validates and sanitizes file paths for securityr c C s8 || _ t|�dg ��| _|�dd�| _|�dd�| _d S )Nzsecurity.allowed_extensionsz security.max_filename_lengthr z security.block_path_traversalT)r �setrB r r �block_traversal�r r r r r r s zPathValidator.__init__Fr> �base_dir�skip_extension_checkr c
C sB z�| j r | �|�r W dS tj�|�}tj�|�s tj�||�}nW dS tj�|�}tj�|�}|�|tj �sD||krDddd|� �fW S tj�
|�}t |�| j kr\ddd| j � d�fW S | j
r|stj�|�d �� �d �}|r|| j
vrddd
|� d �fW S d |dfW S ty� } zddd
t| �� �fW Y d} ~ S d} ~ ww )zz
Validate a path against security constraints
Returns: (is_valid, normalized_path, error_message)
)FNz!Path traversal sequences detected)FNzAbsolute paths not allowedFNz Path escapes allowed directory: zFilename too long (max z chars)� r? zFile extension .z not allowedTzPath validation error: )rK �_contains_traversal�osr> �
expanduser�isabs�join�realpath�
startswith�sep�basename�lenr r �splitext�lower�lstripr) rH )
r r> rM rN � full_path�
normalizedZ base_real�filename�extr/ r r r �
validate_path� s.
��zPathValidator.validate_pathc s& g d�}|� � � t� fdd�|D ��S )z*Check if path contains traversal sequences)z../z..\z..\\z..//c 3 s � | ]}|� v V qd S rC r )�.0�pattern�Z
path_lowerr r � <genexpr>� � � z4PathValidator._contains_traversal.<locals>.<genexpr>)r[ �any)r r> Zdangerous_patternsr rd r rP � s z!PathValidator._contains_traversal�templatec C s& g d�}|}|D ]}|� |d�}q|S )z7Sanitize yt-dlp template variables to prevent injection)�|�&�;�$�`�>�<�_)�replace)r rh Zdangerous_chars� sanitized�charr r r �sanitize_template_vars� s
z$PathValidator.sanitize_template_varsN�F)
rD rE rF rG r r rH �bool�tupler ra rP rt r r r r rI | s .+rI c
@ s� e Zd ZdZdefdd�Zdeeeeef f fdd�Zdeeeeef f fdd �Z d
ede
e e e e e f fd d �Z
dd
ede e de e de
e e e e e f fdd�Zd
S )�LocationManagerz%Manages configured download locationsr c C s || _ t|�| _d | _d S rC )r rI � validator�_locations_cacherL r r r r � s
zLocationManager.__init__r c C s | j du r
| �� | _ | j S )z3Get all configured download locations with metadataN)rz �_build_locations)r r r r �
get_locations� s
z LocationManager.get_locationsc C s i }| j �di �}|�� D ]s\}}tj�|�}z7tj|ddd� t�|tj�r4||dd|� �d�||< nt �
d|� d|� d �� ||d
d|� d �d�||< W q
t y� } z&t � d |� d|� d
|� �� ||d
d|� dt
|�� d�d�||< W Y d}~q
d}~ww |S )z*Build locations dictionary with validationr! i� T)�moder zDownload location: )r> �original�writable� descriptionz Location z (z) is not writableFz (not writable)z!Failed to create/access location z): z (error: �)N)r rB r3 rQ r> rR �makedirs�access�W_OKr* r= r) r+ rH )r � locationsZconfigured_locations� location_idr> Z
expanded_pathr/ r r r r{ � s: �
�� ��� z LocationManager._build_locationsr� c C sb | � � }||vrt|�� �}ddd|� d|� �fS || }|d s*ddd|� d�fS d|d dfS )
zY
Validate a location ID
Returns: (is_valid, path, error_message)
FNzUnknown location 'z'. Available: r z
Location 'z' is not writableTr> )r| �listrA )r r� r� Z available�locationr r r �validate_location� s z!LocationManager.validate_locationN�
relative_path�filename_templatec C sz | � |�\}}}|sdd|fS |r |rtj�||�}n|}n |r#|}n| j�dd�}| j�|�}d|v }| jj|||d�S )zh
Construct a secure download path
Returns: (is_valid, full_path, error_message)
FNz ytdlp.default_filename_templater z%()rN ) r� rQ r> rT r rB ry rt ra ) r r� r� r� �valid� base_pathr+ Z
full_relativeZhas_template_varsr r r �construct_download_path s
z'LocationManager.construct_download_path�NN)rD rE rF rG r r r rH r| r{ rw rv r r� r� r r r r rx � s $(���rx �video-downloaderc @ s� e Zd ZU dZeed< ee ed< ee ed< ee ed< ee ed< ee ed< ee ed< ee ee
f ed < e eee eef f ed
< eed < d S )
� VideoInfoz Structured video information�title�duration� thumbnailr� �uploader� upload_date�
view_count�formats� subtitles� webpage_urlN) rD rE rF rG rH �__annotations__r �intr r r r r r r r� ) s
r� c @ s� e Zd ZdZedefdd��Zeddededee ee
f fdd ��Z ededee e ee
f fd
d ��Z
edded
ee dee de ee
f fdd��Zd S )�YtDlpExtractorzyt-dlp extraction wrapperr c C s6 z
t jddgddd� W dS t jtfy Y dS w )z Check if yt-dlp is available�yt-dlpz --versionT)�capture_output�checkF)�
subprocess�run�CalledProcessError�FileNotFoundErrorr r r r �check_availability: s �z!YtDlpExtractor.check_availabilityF�url� extract_flatc C sX z!g d�}|r |� d� |� | � tj|ddddd�}t�|j�W S ty+ Y dS w )z&Extract video information using yt-dlp)r� z-Jz
--no-warningsz--flat-playlistT� )r� �textr� �timeoutN)�appendr� r� �json�loads�stdoutr) )r� r� �cmdr6 r r r � extract_infoC s
�zYtDlpExtractor.extract_infoc C s" t �| �}|rd|v r|d S dS )z!Get available formats for a videor� N)r� r� )r� r, r r r � get_formatsQ s
zYtDlpExtractor.get_formatsN� format_id� output_pathc
C s� dg}|r |� d|g� |r|� d|g� |�| � ztj|dddd�}d|j|jd�W S tjyJ } zdt|�|j|jd�W Y d }~S d }~ww )
zDownload video using yt-dlpr� z-fz-oT)r� r� r� )�successr� �stderrF)r� r+ r� r� N)�extendr� r� r� r� r� r� rH )r� r� r� r� r6 r/ r r r �download_videoY s(
����z YtDlpExtractor.download_videoru r� )rD rE rF rG � staticmethodrv r� rH r r r r� r r� r� r r r r r� 7 s &
$2r� c
@ sn e Zd ZdZededee fdd��Zedededeee e f fdd ��Z
ededeeef fd
d ��Z d S )
�WebpageAnalyzerz/Fallback webpage analysis for unsupported sitesr� r c C s> zddi}t j| |dd�}|�� |jW S ty Y dS w )zFetch webpage source codez
User-AgentzoMozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/119.0.0.0 Safari/537.36r� )�headersr� N)�requestsrB Zraise_for_statusr� r) )r� r� Zresponser r r �fetch_page_sourcex s � �z!WebpageAnalyzer.fetch_page_source� html_content�base_urlc
C s ddgddgddgdgdgd �}i }|� � D ]i\}}g ||< |D ]^}t�|| tj�}|D ]Q}t|t�r=t|�d
kr=|d
n|} t| t�rP| d rL| d n| d
} | �d �r[d
| � �} n| �d�rft|| �} n
| �d�spt|| �} | || vr}|| � | � q,q q|S )z@Extract potential video/audio/subtitle URLs using regex patternsz.(["\'])(https?://[^"\']*\.mpd(?:\?[^"\']*)?)\1z7manifest["\']:\s*["\']([^"\']*\.mpd(?:\?[^"\']*)?)["\']z/(["\'])(https?://[^"\']*\.m3u8(?:\?[^"\']*)?)\1z8playlist["\']:\s*["\']([^"\']*\.m3u8(?:\?[^"\']*)?)["\']zC(["\'])(https?://[^"\']*\.(?:mp4|webm|mkv|avi|mov)(?:\?[^"\']*)?)\1z?src["\']:\s*["\']([^"\']*\.(?:mp4|webm|mkv)(?:\?[^"\']*)?)["\']zB(["\'])(https?://[^"\']*\.(?:mp3|m4a|aac|ogg|wav)(?:\?[^"\']*)?)\1z>(["\'])(https?://[^"\']*\.(?:vtt|srt|ass|ssa)(?:\?[^"\']*)?)\1)Z
mpd_manifestsZm3u8_playlistsZ video_filesZ audio_filesZsubtitle_filesrO r z//zhttps:�/�http)
r3 �re�findall�
IGNORECASEr4 rw rY rV r r� )
r� r� �patternsZ extracted�categoryZ pattern_listrc Zmatches�matchr� r r r �extract_video_patterns� sF ������"
���z&WebpageAnalyzer.extract_video_patternsc C s� i }g d�}|D ]}t �|| t jt jB �}|r"|�d��� |d< nqddg}|D ]}t �|| t j�}|rAt|�d��|d< |S q)|S )z Extract video metadata from HTML)z<title[^>]*>([^<]+)</title>z&["\']title["\']:\s*["\']([^"\']+)["\']zE<meta[^>]*property=["\']og:title["\'][^>]*content=["\']([^"\']+)["\']rO r� z"duration["\']:\s*["\']?(\d+)["\']?z["\']duration["\']:\s*(\d+)r� )r� �searchr� �DOTALL�group�stripr� )r� �metadataZtitle_patternsrc r� Zduration_patternsr r r �extract_metadata� s& ���z WebpageAnalyzer.extract_metadataN) rD rE rF rG r� rH r r� r r r� r� r r r r r� u s $0 r� r c � s �t jddddddd�idgd�d �t jd
d dddd d�idgd�d �t jd
dddddd�idgd�d �t jdddddd�ddd�ddd�ddd�ddd�ddd�d�dgd�d �t jddddddd�idgd�d �t jd d dddd d�idgd�d �t jd d di g d�d �gS )!z%List available video downloader tools�check_ytdlp_supportz8Check if a URL is supported by yt-dlp and get basic info�objectr� �stringzVideo URL to check)�typer� )r� Z
propertiesZrequired)�namer� Z inputSchema�get_video_infoz:Get detailed video information including available formatszVideo URL to analyze�get_video_formatsz5Get available video/audio formats and quality optionsz Video URL to get formats forr� z7Download video with specified format and output optionszVideo URL to downloadz)Specific format ID to download (optional)zPDownload location ID from configured locations (optional, defaults to 'default')z,Relative path within the location (optional)z7yt-dlp filename template (optional, defaults to config)zNDEPRECATED: Use location_id + relative_path instead. Full output path template)r� r� r� r� r� r� �analyze_webpagez=Fallback: Analyze webpage for video content when yt-dlp failszWebpage URL to analyze�extract_media_patternszAExtract video/audio URLs from webpage HTML using pattern matchingz!Webpage URL to extract media from�get_download_locationsz'Get available secure download locations)�types�Toolr r r r �handle_list_tools� s� ����������������������"�����������r� r� � argumentsc � sn �| dkrc|d }t �� stjdt�ddd��d�gS t �|�}|rUtjdt�d|�d �|�d
�|�d �|�d �|�d
�|�d�rM|�dd�dd� d ndd��d�gS tjdt�ddd��d�gS | dkr�|d }t �|�}|s�tjdt�ddd��d�gS tjdt�d|�d �|�d
�|�d�|�d�|�d �|�d
�|�d �|�d�t|�dg ��t |�di ��
� �d � �d�gS | d k�r8|d }t � |�}|s�tjdt�dd d��d�gS g }|D ]J}|� |�d �|�d �|�d!�r�|�d!d"�� d#�nd$|�d%�|�d&�|�d'�|�d(�|�d)�|�d*�|�d��r$|�dd�dd+� d ndd,�
� q�tjdt�d|d-��d�gS | d.k�r�|d }|�d �}d}d/|v �rW|d/ }t
�d0� naz:t� } t| �}
|�d1d2�} |�d3�} |�d4�}
|
�| | |
�\}}}|�s�tjdt�dd5|� �d��d�gW S |}W n& t�y� } ztjdt�dd6t|�� �d��d�gW Y d}~S d}~ww t �|||�}|�d7��r�||d8< | �d9d��r�t
�d:|� d;|� �� tjdt�|�d�gS | d<k�r%|d }t�|�}|�stjdt�dd=d��d�gS t�|�}tjdt�d|t|�d>|�� v d?|�� v d@��d�gS | dAk�rj|d }t�|�}|�sDtjdt�dd=d��d�gS t�||�}t�|�}tdBdC� |� � D ��}tjdt�d|||dD��d�gS | dEk�r�zt� } t| �}
|
� � }tjdt�d|dF��d�gW S t�y� } ztjdt�ddGt|�� �d��d�gW Y d}~S d}~ww t dH| � ���)IzHandle tool callsr� r� r� Fz$yt-dlp is not installed or available)� supportedr+ )r� r� Tr� r� r� r� r� r� � N�� z...)r� r� r� r� r� r� r� zURL not supported by yt-dlpr� z#Could not extract video information)r� r+ r� r� r� r� ) r� r� r� r� r� r� r� r� r� Z format_countZsubtitle_languagesr� zCould not get video formatsr� r` Zheight�unknown�pz
audio-only�fps�vcodec�acodec�filesize�tbr� format_note�d )
r� r` Z
resolutionr� r� r� r� r� r� r� )r� r� r� r� z`Using deprecated output_path parameter. Consider using location_id + relative_path for security.r� r r� r� zPath validation failed: z*Failed to construct secure download path: r� Z
download_pathzlogging.log_downloadszDownload completed: z -> r� z Could not fetch webpage contentz<videoz<iframe)r� r� Zcontent_lengthZhas_video_tagsZ
has_iframer� c s s � | ]}t |�V qd S rC )rY )rb Zurlsr r r re rf z#handle_call_tool.<locals>.<genexpr>)r� Ztotal_media_urlsr� r� r� )r� r� z"Failed to get download locations: zUnknown tool: ) r� r� r� � TextContentr� �dumpsr� rB rY r� rA r� r� r* r= r rx r� r) rH r� r, r� r� r� r[ r� �sum�valuesr| �
ValueError)r� r� r� r, r� Zprocessed_formats�fmtr� r� r �location_managerr� r� r� r� Z secure_pathr+ r/ r6 r� r� r� Z total_foundr� r r r �handle_call_toolO s� ���
$��
��
����
�� &
�
��
��
� ����
�
��
��
��
��
�
� ����
r� c � sv �t jj�� 4 I d H �#\} }t�| |tddtjt� i d�d��I d H W d �I d H d S 1 I d H s4w Y d S )Nr� z0.1.0)Znotification_optionsZexperimental_capabilities)Z server_nameZserver_versionZ capabilities)�mcp�serverZstdioZ stdio_serverr� r Zget_capabilitiesr )Z read_streamZ write_streamr r r �mainC s ��� �.�r� �__main__)7rG Zasyncior� r� r� r� rQ r" �pathlibr �typingr r r r � urllib.parser r Z dataclassesr r& � ImportErrorZtomliZ
mcp.serverr
r Zmcp.server.modelsr Zmcp.server.stdior� Z mcp.typesr� Z basicConfig�INFOZ getLoggerrD r* r rI rx r� r� r� r� Z
list_toolsr� r� r� Z call_toolrH r5 r� r� r� r� r r r r �<module> sP �
YCh
>cv t�