77from enum import Enum , auto
88from pathlib import Path
99
10- from gitingest .exceptions import InvalidNotebookError
1110from gitingest .utils .ingestion_utils import _get_encoding_list
1211from gitingest .utils .notebook_utils import process_notebook
1312from gitingest .utils .textfile_checker_utils import is_textfile
1413
15- SEPARATOR = "=" * 48 + " \n "
14+ SEPARATOR = "=" * 48
1615
1716
1817class FileSystemNodeType (Enum ):
@@ -36,108 +35,104 @@ class FileSystemNode: # pylint: disable=too-many-instance-attributes
3635 """
3736 Class representing a node in the file system (either a file or directory).
3837
39- This class has more than the recommended number of attributes because it needs to
40- track various properties of files and directories for comprehensive analysis.
38+ Tracks properties of files/directories for comprehensive analysis.
4139 """
4240
4341 name : str
44- type : FileSystemNodeType # e.g., "directory" or "file"
42+ type : FileSystemNodeType
4543 path_str : str
4644 path : Path
4745 size : int = 0
4846 file_count : int = 0
4947 dir_count : int = 0
5048 depth : int = 0
51- children : list [FileSystemNode ] = field (default_factory = list ) # Using default_factory instead of empty list
49+ children : list [FileSystemNode ] = field (default_factory = list )
5250
5351 def sort_children (self ) -> None :
5452 """
5553 Sort the children nodes of a directory according to a specific order.
5654
5755 Order of sorting:
58- 1. README.md first
59- 2. Regular files (not starting with dot)
60- 3. Hidden files (starting with dot)
61- 4. Regular directories (not starting with dot)
62- 5. Hidden directories (starting with dot)
63- All groups are sorted alphanumerically within themselves.
64- """
65- # Separate files and directories
66- files = [child for child in self .children if child .type == FileSystemNodeType .FILE ]
67- directories = [child for child in self .children if child .type == FileSystemNodeType .DIRECTORY ]
56+ 2. Regular files (not starting with dot)
57+ 3. Hidden files (starting with dot)
58+ 4. Regular directories (not starting with dot)
59+ 5. Hidden directories (starting with dot)
6860
69- # Find README.md
70- readme_files = [f for f in files if f .name .lower () == "readme.md" ]
71- other_files = [f for f in files if f .name .lower () != "readme.md" ]
61+ All groups are sorted alphanumerically within themselves.
7262
73- # Separate hidden and regular files/directories
74- regular_files = [f for f in other_files if not f .name .startswith ("." )]
75- hidden_files = [f for f in other_files if f .name .startswith ("." )]
76- regular_dirs = [d for d in directories if not d .name .startswith ("." )]
77- hidden_dirs = [d for d in directories if d .name .startswith ("." )]
63+ Raises
64+ ------
65+ ValueError
66+ If the node is not a directory.
67+ """
68+ if self .type != FileSystemNodeType .DIRECTORY :
69+ raise ValueError ("Cannot sort children of a non-directory node" )
7870
79- # Sort each group alphanumerically
80- regular_files .sort (key = lambda x : x .name )
81- hidden_files .sort (key = lambda x : x .name )
82- regular_dirs .sort (key = lambda x : x .name )
83- hidden_dirs .sort (key = lambda x : x .name )
71+ def _sort_key (child : FileSystemNode ) -> tuple [int , str ]:
72+ # Groups: 0=README, 1=regular file, 2=hidden file, 3=regular dir, 4=hidden dir
73+ name = child .name .lower ()
74+ if child .type == FileSystemNodeType .FILE :
75+ if name == "readme.md" :
76+ return (0 , name )
77+ return (1 if not name .startswith ("." ) else 2 , name )
78+ return (3 if not name .startswith ("." ) else 4 , name )
8479
85- self .children = readme_files + regular_files + hidden_files + regular_dirs + hidden_dirs
80+ self .children . sort ( key = _sort_key )
8681
8782 @property
8883 def content_string (self ) -> str :
8984 """
90- Return the content of the node as a string.
91-
92- This property returns the content of the node as a string, including the path and content.
85+ Return the content of the node as a string, including path and content.
9386
9487 Returns
9588 -------
9689 str
9790 A string representation of the node's content.
9891 """
99- content_repr = SEPARATOR
92+ parts = [
93+ SEPARATOR ,
94+ f"File: { str (self .path_str ).replace (os .sep , '/' )} ,
95+ SEPARATOR ,
96+ f"{ self .content } ,
97+ ]
10098
101- # Use forward slashes in output paths
102- content_repr += f"File: { str (self .path_str ).replace (os .sep , '/' )} \n "
103- content_repr += SEPARATOR
104- content_repr += f"{ self .content } \n \n "
105- return content_repr
99+ return "\n " .join (parts ) + "\n \n "
106100
107101 @property
108102 def content (self ) -> str : # pylint: disable=too-many-return-statements
109103 """
110- Read the content of a file.
111-
112- This function attempts to open a file and read its contents using UTF-8 encoding.
113- If an error occurs during reading (e.g., file is not found or permission error),
114- it returns an error message.
104+ Read the content of a file if it's text (or a notebook). Return an error message otherwise.
115105
116106 Returns
117107 -------
118108 str
119109 The content of the file, or an error message if the file could not be read.
110+
111+ Raises
112+ ------
113+ ValueError
114+ If the node is a directory.
120115 """
121- if self .type == FileSystemNodeType .FILE and not is_textfile (self .path ):
116+ if self .type == FileSystemNodeType .DIRECTORY :
117+ raise ValueError ("Cannot read content of a directory node" )
118+
119+ if not is_textfile (self .path ):
122120 return "[Non-text file]"
123121
124- try :
125- if self .path .suffix == ".ipynb" :
126- try :
127- return process_notebook (self .path )
128- except Exception as exc :
129- return f"Error processing notebook: { exc }
130-
131- for encoding in _get_encoding_list ():
132- try :
133- with self .path .open (encoding = encoding ) as f :
134- return f .read ()
135- except UnicodeDecodeError :
136- continue
137- except OSError as exc :
138- return f"Error reading file: { exc }
139-
140- return "Error: Unable to decode file with available encodings"
141-
142- except (OSError , InvalidNotebookError ) as exc :
143- return f"Error reading file: { exc }
122+ if self .path .suffix == ".ipynb" :
123+ try :
124+ return process_notebook (self .path )
125+ except Exception as exc :
126+ return f"Error processing notebook: { exc }
127+
128+ # Try multiple encodings
129+ for encoding in _get_encoding_list ():
130+ try :
131+ with self .path .open (encoding = encoding ) as f :
132+ return f .read ()
133+ except UnicodeDecodeError :
134+ continue
135+ except OSError as exc :
136+ return f"Error reading file: { exc }
137+
138+ return "Error: Unable to decode file with available encodings"
0 commit comments