관리-도구
편집 파일: identify.cpython-38.pyc
U *�e� � @ s� d dl mZ d dlZd dlZd dlZd dlZd dlZd dlZd dl Z d dl Z d dl m Z d dl mZ d dl mZ d dlmZ ee j�ZdZdZd Zd Zd Zd Zd ZdZeeeeef�Z eeef�Z eeef�Z e e e �Z e j ej!�"� � e j ej#�"� � e j ej$�"� � e j ej%�"� � ee �Z&ddd�dd�Z'ddd�dd�Z(ddd�dd�Z)ddd�d d �Z*ddd�d d �Z+dd d!�d"d#�Z,dd$d$d%�d&d'�Z-dd$d�d(d)�Z.dd$d�d*d+�Z/e�0d,ej1ej2B �Z3e�0d-�Z4ddd.�d/d0�Z5dd1d2�d3d4�Z6dS )5� )� annotationsN)�IO)� extensions)� interpreters)�licenses� directory�symlink�socket�file� executableznon-executable�text�binary�strzset[str])�path�returnc C s> zt �| �}W n& ttfk r4 t| � d���Y nX |j}t�|�rLthS t�|�r\t hS t� |�rlt hS t h}t � | t j�}|r�|�t� n |�t� tt j�| ��}t|�dkr�|�|� n*|r�t| �}t|�dkr�|�t|d �� t|@ �st| ��r |�t� n |�t � t|@ �s(t |��t |@ �s:t |��|S )N� does not exist.r ) �os�lstat�OSError� ValueError�st_mode�stat�S_ISDIR� DIRECTORY�S_ISLNK�SYMLINK�S_ISSOCK�SOCKET�FILE�access�X_OK�add� EXECUTABLE�NON_EXECUTABLE�tags_from_filenamer �basename�len�update�parse_shebang_from_file�tags_from_interpreter� ENCODING_TAGS� file_is_text�TEXT�BINARY�AssertionError� MODE_TAGS)r �sr�mode�tagsr �t�shebang� r5 �?/opt/hc_python/lib/python3.8/site-packages/identify/identify.py�tags_from_path( s< r7 c C s� t j�| �\}}t j�|�\}}t� }|g|�d� D ]"}|tjkr6|�tj| � qZq6t|�dkr�|dd � � � }|tj kr�|�tj | � n|tj kr�|�tj | � |S )N�.r � ) r r �split�splitext�setr �NAMESr' r& �lower� EXTENSIONS� EXTENSIONS_NEED_BINARY_CHECK)r �_�filename�ext�ret�partr5 r5 r6 r$ V s r$ )� interpreterr c C s@ | � d�\}}} | r:| tjkr(tj| S | � d�\} }}qt� S )N�/r8 )� rpartitionr � INTERPRETERSr< )rF rA r5 r5 r6 r) l s r) z IO[bytes]�bool)�bytesior c C sL t ddddddddg�t td d �� t td d �� }t| �d ��d|�� S )z�Return whether the first KB of contents seems to be binary. This is roughly based on libmagic's binary/text detection: https://github.com/file/file/blob/df74b09b9027676088c797528edcaae5a9ce9ad0/src/encoding.c#L203-L228 � � � � � � � � � � � � i N)� bytearray�rangerJ �read� translate)rK Z text_charsr5 r5 r6 �is_texty s � ��r\ c C sD t j�| �st| � d���t| d��}t|�W 5 Q R � S Q R X d S )Nr �rb)r r �lexistsr �openr\ )r �fr5 r5 r6 r+ � s r+ z list[str])�liner c C s. z t �| �W S tk r( | �� Y S X d S )N)�shlexr: r )ra r5 r5 r6 �_shebang_split� s rc ztuple[str, ...])rK �cmdr c C s� | � d�dkr�| �� }z|�d�}W n tk r< | Y S X |D ]}|tkrB| S qBtt|�� ��}t|d d� �D ] \}}|dkr�qx||d f}qxq |S )N� � #!�UTF-8���z-ir9 ) rZ �readline�decode�UnicodeDecodeError� printable�tuplerc �strip� enumerate)rK rd Z next_line_b� next_line�cZ line_tokens�i�tokenr5 r5 r6 �_parse_nix_shebang� s rt c C s� | � d�dkrdS | �� }z|�d�}W n tk r> Y dS X |D ]}|tkrD dS qDtt|�� ��}|r�|d dkr�|d dkr�|dd � }n |dd � }|d kr�t| |�S |S ) z8Parse the shebang from a file opened for reading binary.re rf r5 rg r z /usr/bin/envr9 z-SN)z nix-shell) rZ ri rj rk rl rm rc rn rt )rK Z first_line_b� first_linerq rd r5 r5 r6 � parse_shebang� s$ rv c C s� t j�| �st| � d���t �| t j�s,dS z,t| d��}t|�W 5 Q R � W S Q R X W n: tk r� } z |j t j kr�W Y � dS � W 5 d}~X Y nX dS )z$Parse the shebang given a file path.r r5 r] N) r r r^ r r r r_ rv r �errno�EINVAL)r r` �er5 r5 r6 r( � s r( z^\s*(Copyright|\(C\)) .*$z\s+)�sr c C s t �d| �} t�d| �} | �� S )N� � )� COPYRIGHT_RE�sub�WS_RErn )rz r5 r5 r6 � _norm_license� s r� z str | None)rB r c C s� ddl }t| dd��}|�� }W 5 Q R X t|�}tj}d}t�dt|� �}t j D ]l\}} t| �} || krr| S |r�t t|�t| � �t|� dkr�qR|� || |�} | |k rR| |k rR| }|}qR|r�||k r�|S dS dS )a� Return the spdx id for the license contained in `filename`. If no license is detected, returns `None`. spdx: https://spdx.org/licenses/ licenses from choosealicense.com: https://github.com/choosealicense.com Approximate algorithm: 1. strip copyright line 2. normalize whitespace (replace all whitespace with a single space) 3. check exact text match with existing licenses 4. failing that use edit distance r Nrg )�encodingr{ g�������?) �ukkonenr_ rZ r� �sys�maxsize�math�ceilr& r ZLICENSES�absZdistance) rB r� r` �contentsZnormZ min_edit_distZmin_edit_dist_spdx�cutoffZspdxr Z norm_licenseZ edit_distr5 r5 r6 � license_id� s( $ r� )7� __future__r rw r� �os.pathr �rerb r �stringr� �typingr �identifyr r Zidentify.vendorr � frozensetrl r r r r r" r# r, r- Z TYPE_TAGSr/ r* Z _ALL_TAGSr' r? �valuesr@ r= rI ZALL_TAGSr7 r$ r) r\ r+ rc rt rv r( �compile�I� MULTILINEr} r r� r� r5 r5 r5 r6 �<module> sV .