code_analyzer.cpython-310.pyc•13.7 kB
o
y�g�k � @ s� d Z ddlZddlZddlZddlmZmZmZmZm Z m
Z
ddl Z ddl
mZmZmZmZ ejejd� e�e�ZG dd� dej�ZG d d
� d
�ZG d d � d ej�ZdS )
z}
Code Analysis Engine
Responsible for parsing code using Python's AST and extracting entities,
relationships, and metadata.
� N)�Dict�List�Optional�Any�Tuple�Set� )�KnowledgeGraph�
ComponentType�ComponentStatus�RelationshipType)�levelc s0 e Zd ZdZdd� Z� fdd�Zdd� Z� ZS )� CodeVisitorz%AST visitor that tracks parent nodes.c C s
i | _ d S �N)�parents)�self� r �X/Users/sparshdrolia/Sparsh personal/persistent-code-mcp/persistent_code/code_analyzer.py�__init__ s
zCodeVisitor.__init__c s& t �|�D ]}|| j|< qt� �|�S r )�ast�iter_child_nodesr �super�visit)r �node�child�� __class__r r r s zCodeVisitor.visitc C s | j �|�S )z Get parent of a node safely.)r �get)r r r r r �
get_parent$ s zCodeVisitor.get_parent)�__name__�
__module__� __qualname__�__doc__r r r �
__classcell__r r r r r s
r c
@ s& e Zd ZdZdefdd�Zd dedee dee d efd
d �Zd e j
d
ed e e eef fdd�Z
d e j
d
eded eeef fdd�Zde jdeded eeef fdd�Zd e j
d
eded eeef fdd�Zd e j
d
eded eeef fdd�Zd e j
d
edeeef d eeef def
d d �ZdS ) � CodeAnalyzerz5Analyzes Python code and updates the knowledge graph.�knowledge_graphc C s
|| _ dS )zxInitialize the code analyzer.
Args:
knowledge_graph: The knowledge graph to update
N)r% )r r% r r r r + s
zCodeAnalyzer.__init__N� code_text� file_path� component_id�returnc C s� zlt �|�}t� }|�|� |rtj�|�nd}|du r4| jj|t j
|t j d|� �|r/d|ini d�}n
| jj
||t j d� | �||�}| �|||�}| �|||�} | �|||�}
| �||| ||� t�d|� �� |W S ty� } zLt�dt| �� �� |du r�| jj|r�tj�|�ndt j
|t jd | � �|r�|t| �d
�nd t| �id�}n| jj
||t jd t| �id � |W Y d} ~ S d} ~ w ty� } zt�d
t| �� �� |r�|ndW Y d} ~ S d} ~ ww )aJ Analyze code and update the knowledge graph.
Args:
code_text: The code to analyze
file_path: Path to the file (optional)
component_id: ID of an existing component to update (optional)
Returns:
component_id: ID of the analyzed component
zunnamed_file.pyNz
Python file: �path��name�component_typer& �status� description�metadata)r( r& r. z Successfully analyzed code from zSyntax error in code: z Python file with syntax errors: )r* �errorr1 )r( r& r. r0 zError analyzing code: )r �parser r �osr* �basenamer% �
add_componentr
�FILEr � IMPLEMENTED�update_component�_analyze_imports�_analyze_classes�_analyze_functions�_analyze_variables�_analyze_relationships�logger�info� SyntaxError�warning�str�PARTIAL� Exceptionr1 ) r r& r' r( �tree�visitor� file_name�imported_modulesZ classes_map�
functions_mapZ
variables_map�er r r � analyze_code3 sd
�
� �
����zCodeAnalyzer.analyze_coderE �file_idc
C s\ g }t �|�D ]�}t|t j�r>|jD ]*}| jj|jtj t
j d|j� �d|j d�d�}| jj
||tjd� |�|j|f� qqt|t j�r�|jpHd}| jj|tj t
j d|� �ddid�}|jD ]L}| jj|rm|� d|j� �n|jtjt
j d |� d
|j� �d|j |d �d�} | jj
|| tjd� | jj
|| tjd� |�|r�|� d|j� �n|j| f� q^q|S ) z�Extract and record imports from the AST.
Args:
tree: AST of the code
file_id: ID of the file component
Returns:
imported_modules: List of (module_name, component_id) tuples
zImported module: T)�external�alias)r, r- r. r/ r0 �� source_id� target_id�relationship_type� rM �.zImported from �: )rM rN �
parent_module)r r �
isinstance�Import�namesr% r5 r, r
�MODULEr r7 �asname�add_relationshipr �IMPORTS�append�
ImportFrom�module�FUNCTION�CONTAINS)
r rE rL rH r r, � module_id� module_nameZparent_module_idZ import_idr r r r9 � s`
� ��
�
� ��&�z CodeAnalyzer._analyze_importsrF c C s� i }t �|�D ]�}t|t j�r�t �|�pd}g }|jD ]}t|t j�r)|�|j� qd} zM|j }
d} t
|d�r>|j r>|j } n|j rF|j d n|} t
| d�rT| j rT| j } n|
d } | j
�|�d �d�}
|
| krx|
t|
�krxd�|
|
d | � �} W n' ty� } zt�d |j� d
t|�� �� d |j� d �} W Y d
}~nd
}~ww | j
j|jtj| tj||dd� |j D �|j | fd�d�}| j
j ||t j d� |D ] }| j
j |tjgdd�}|r�|d d }| j
j ||t j d� q�| �!|||�}|||j< q|S )a3 Extract and record classes from the AST.
Args:
tree: AST of the code
file_id: ID of the file component
visitor: AST visitor with parent tracking
Returns:
class_map: Dictionary mapping class names to component IDs
rS r �
end_lineno������ r& �
r z Error extracting class code for rU zclass z
(...): ...Nc S s g | ] }t |tj�r|j�qS r )rW r � FunctionDefr, )�.0�mr r r �
<listcomp> s z1CodeAnalyzer._analyze_classes.<locals>.<listcomp>)�bases�methods�
line_ranger+ rO ��query�component_types�limit�id)"r r rW �ClassDef�
get_docstringrm �Namer^ rt �lineno�hasattrre �bodyr% �
get_component�split�len�joinrD r> rA r, rB r5 r
�CLASSr r7 r\ r rb � search_code�INHERITS�_analyze_methods)r rE rL rF Z class_mapr � docstringrm �base�
class_code�
start_line�end_line�
last_child�
code_linesrJ �class_idZbase_componentsZbase_id�
method_mapr r r r: � s|
�� ������ ��
�z CodeAnalyzer._analyze_classes�
class_noder� c C s i }|j D ]�}t|tj�r�t�|�pd}dd� |jjD �}|j� dd�|�� d�}d} ze| j� |�d �
d�}
|j |j } d } t |d
�rN|j
rN|j
|j } n |j rV|j d n|}
t |
d
�rg|
j
rg|
j
|j } n| d } d | kru| kr�n n| t|
�k r�d�|
| | d
� �} nd|� d�} W n* ty� } z t�d|j� d|j� dt|�� �� d|� d�} W Y d}~nd}~ww | jj|j� d|j� �tj| tj|||jtdd� |jD ��tdd� |jD ��||j t|d
d �fd�d�}| jj||t j d� |||j< q|S )aO Extract and record methods from a class definition.
Args:
class_node: AST node for the class
class_id: ID of the class component
visitor: AST visitor with parent tracking
Returns:
method_map: Dictionary mapping method names to component IDs
rS c S � g | ]}|j �qS r ��arg�rj �ar r r rl C � z1CodeAnalyzer._analyze_methods.<locals>.<listcomp>�(�, �)r& rh r re rf rg r �def �: ...z!Error extracting method code for rT rU Nc s �& � | ]}t |tj�r|jd kV qdS )� staticmethodN�rW r rw rt �rj �dr r r � <genexpr>n � �$ z0CodeAnalyzer._analyze_methods.<locals>.<genexpr>c s r� )� classmethodNr� r� r r r r� o r� )� signature�class� is_static�is_class_method�argsro r+ rO ) rz rW r ri rv r� r, r~ r% r{ r| rx ry re r} rD r> rA rB r5 r
�METHODr r7 �any�decorator_list�getattrr\ r rb )r r� r� rF r� r r� � arg_namesr� Z method_coder� r� r� r� rJ Z method_idr r r r� 0 s`
$ �$�����
�z CodeAnalyzer._analyze_methodsc C s� i }t �|�D ]�}t|t j�r�|�|�}|r t|t j�r qt �|�p#d}dd� |jjD �}|j� dd� |�� d�} d}
zM|j
} d} t |d�rM|j rM|j } n|j
rU|j
d n|}
t |
d�rc|
j rc|
j } n| d
} | j�|�d �d �}| | kr�| t|�kr�d � || d
| � �}
W n& ty� } zt�d|j� dt|�� �� d| � d�}
W Y d}~nd}~ww | jj|jtj|
tj|| ||j
| fd�d�}| jj||t j d� |||j< q|S )a; Extract and record functions from the AST.
Args:
tree: AST of the code
file_id: ID of the file component
visitor: AST visitor with parent tracking
Returns:
function_map: Dictionary mapping function names to component IDs
rS c S r� r r� r� r r r rl � r� z3CodeAnalyzer._analyze_functions.<locals>.<listcomp>r� r� r� r re rf rg r& rh r z#Error extracting function code for rU r� r� N)r� r� ro r+ rO ) r r rW ri r ru rv r� r, r~ rx ry re rz r% r{ r| r} rD r> rA rB r5 r
ra r r7 r\ r rb )r rE rL rF Z function_mapr �parentr� r� r� Z
function_coder� r� r� r� rJ Z function_idr r r r; � s^
� �����
�z CodeAnalyzer._analyze_functionsc
C s� i }t �|�D ]Q}t|t j�rX|�|�}|r#t|t j�s"t|t j�r#q|jD ]1}t|t j�rW| j j
|j t j
|j � d�tjd|j � �d|jid�}| j j||tjd� |||j < q&q|S )aB Extract and record global variables from the AST.
Args:
tree: AST of the code
file_id: ID of the file component
visitor: AST visitor with parent tracking
Returns:
variable_map: Dictionary mapping variable names to component IDs
z = ...zGlobal variable: �liner+ rO )r r rW �Assignr ru ri �targetsrw r% r5 rt r
�VARIABLEr r7 rx r\ r rb ) r rE rL rF Z variable_mapr r� �targetZ variable_idr r r r< � s2
�� �
�z CodeAnalyzer._analyze_variables� functions�classesc C s t | ||||�}|�|� dS )ag Analyze function calls and other relationships from the AST.
Args:
tree: AST of the code
file_id: ID of the file component
functions: Dictionary of function names to component IDs
classes: Dictionary of class names to component IDs
visitor: AST visitor with parent tracking
N)�FunctionCallVisitorr )r rE rL r� r� rF Z call_visitorr r r r= s z#CodeAnalyzer._analyze_relationships)NN)r r r! r" r r rB r rK r �Moduler r r9 r r r: ru r� r; r< r= r r r r r$ ( s( $$Q$J$b$Q$Q/��
�
��r$ c @ s( e Zd ZdZdd� Zdd� Zdd� ZdS ) r� z:Visitor that finds function calls and other relationships.c C s4 || _ || _|| _|| _|| _d | _d | _g | _d S r )�analyzerrL r� r� �parent_visitor�current_function�current_method�calls)r r� rL r� r� r� r r r r s
z FunctionCallVisitor.__init__c C sl | j }| j}| j�|�}|r"t|tj�r"|j� d|j� �| _d | _ n|j| _ d | _| �|� || _ || _d S )NrT ) r� r� r� r rW r ru r, �
generic_visit)r r � old_function�
old_methodr� r r r �visit_FunctionDef! s
z%FunctionCallVisitor.visit_FunctionDefc C s~ d }t |jtj�rM|jj}|| jv rLd }| jr$| j| jv r$| j| j }n| jr<| jj j
| jt j gdd�}|r<|d d }|rL| jj j
|| j| tjd� nkt |jtj�r�t |jjtj�r�|jjj}|jj}|| jv r�| jj j
|� d|� �t j gdd�}|r�d }| jr�| j| jv r�| j| j }n| jr�| jj j
| jt j gdd�}|r�|d d }|r�| jj j
||d d tjd� | �|� d S )Nr rp r rt rO rT )rW �funcr rw rt r� r� r� r� r% r� r
r� r\ r �CALLS� Attribute�value�attrr� r� )r r � func_nameZ caller_idZmethod_components�obj_name� method_nameZcurrent_method_componentsr r r �
visit_Call6 sf
� ��
��
�z FunctionCallVisitor.visit_CallN)r r r! r" r r� r� r r r r r� s
r� )r" r r3 �logging�typingr r r r r r �networkx�nxr% r r
r r � basicConfig�INFO� getLoggerr r> � NodeVisitorr r$ r� r r r r �<module> s
o