refactor: introduce Source base class

.vscode/launch.json

@@ -5,7 +5,7 @@
             "type": "debugpy",
             "request": "launch",
             "module": "server",
-            "args": [],
+            "args": ["--reload"],
             "cwd": "${workspaceFolder}/src"
         }
     ]

src/gitingest/ingestion.py

@@ -7,7 +7,8 @@
 
 from gitingest.config import MAX_DIRECTORY_DEPTH, MAX_FILES, MAX_TOTAL_SIZE_BYTES
 from gitingest.output_formatter import format_node
-from gitingest.schemas import FileSystemNode, FileSystemNodeType, FileSystemStats
+from gitingest.schemas import FileSystemDirectory, FileSystemFile, FileSystemNode, FileSystemStats, FileSystemSymlink
+from gitingest.utils.compat_func import readlink
 from gitingest.utils.ingestion_utils import _should_exclude, _should_include
 from gitingest.utils.logging_config import get_logger
 
@@ -70,9 +71,8 @@ def ingest_query(query: IngestionQuery) -> tuple[str, str, str]:
 
         relative_path = path.relative_to(query.local_path)
 
-        file_node = FileSystemNode(
+        file_node = FileSystemFile(
             name=path.name,
-            type=FileSystemNodeType.FILE,
             size=path.stat().st_size,
             file_count=1,
             path_str=str(relative_path),
@@ -95,9 +95,8 @@ def ingest_query(query: IngestionQuery) -> tuple[str, str, str]:
 
     logger.info("Processing directory", extra={"directory_path": str(path)})
 
-    root_node = FileSystemNode(
+    root_node = FileSystemDirectory(
         name=path.name,
-        type=FileSystemNodeType.DIRECTORY,
         path_str=str(path.relative_to(query.local_path)),
         path=path,
     )
@@ -161,9 +160,8 @@ def _process_node(node: FileSystemNode, query: IngestionQuery, stats: FileSystem
                 continue
             _process_file(path=sub_path, parent_node=node, stats=stats, local_path=query.local_path)
         elif sub_path.is_dir():
-            child_directory_node = FileSystemNode(
+            child_directory_node = FileSystemDirectory(
                 name=sub_path.name,
-                type=FileSystemNodeType.DIRECTORY,
                 path_str=str(sub_path.relative_to(query.local_path)),
                 path=sub_path,
                 depth=node.depth + 1,
@@ -201,11 +199,11 @@ def _process_symlink(path: Path, parent_node: FileSystemNode, stats: FileSystemS
         The base path of the repository or directory being processed.
 
     """
-    child = FileSystemNode(
+    child = FileSystemSymlink(
         name=path.name,
-        type=FileSystemNodeType.SYMLINK,
         path_str=str(path.relative_to(local_path)),
         path=path,
+        target=str(readlink(path)),
         depth=parent_node.depth + 1,
     )
     stats.total_files += 1
@@ -258,9 +256,8 @@ def _process_file(path: Path, parent_node: FileSystemNode, stats: FileSystemStat
     stats.total_files += 1
     stats.total_size += file_size
 
-    child = FileSystemNode(
+    child = FileSystemFile(
         name=path.name,
-        type=FileSystemNodeType.FILE,
         size=file_size,
         file_count=1,
         path_str=str(path.relative_to(local_path)),

src/gitingest/output_formatter.py

@@ -8,8 +8,7 @@
 import requests.exceptions
 import tiktoken
 
-from gitingest.schemas import FileSystemNode, FileSystemNodeType
-from gitingest.utils.compat_func import readlink
+from gitingest.schemas import FileSystemDirectory, FileSystemFile, FileSystemNode, FileSystemSymlink
 from gitingest.utils.logging_config import get_logger
 
 if TYPE_CHECKING:
@@ -42,18 +41,19 @@ def format_node(node: FileSystemNode, query: IngestionQuery) -> tuple[str, str,
         A tuple containing the summary, directory structure, and file contents.
 
     """
-    is_single_file = node.type == FileSystemNodeType.FILE
-    summary = _create_summary_prefix(query, single_file=is_single_file)
-
-    if node.type == FileSystemNodeType.DIRECTORY:
+    # Use polymorphic properties - much cleaner!
+    summary = _create_summary_prefix(query, single_file=node.is_single_file)
+
+    # Add type-specific summary info
+    if isinstance(node, FileSystemDirectory):
         summary += f"Files analyzed: {node.file_count}\n"
-    elif node.type == FileSystemNodeType.FILE:
-        summary += f"File: {node.name}\n"
-        summary += f"Lines: {len(node.content.splitlines()):,}\n"
+    elif isinstance(node, FileSystemFile):
+        summary += f"File: {node.name or ''}\nLines: {len(node.content.splitlines()):,}\n"
 
     tree = "Directory structure:\n" + _create_tree_structure(query, node=node)
-
-    content = _gather_file_contents(node)
+
+    # Use polymorphic content gathering
+    content = node.gather_contents()
 
     token_estimate = _format_token_count(tree + content)
     if token_estimate:
@@ -102,30 +102,6 @@ def _create_summary_prefix(query: IngestionQuery, *, single_file: bool = False)
     return "\n".join(parts) + "\n"
 
 
-def _gather_file_contents(node: FileSystemNode) -> str:
-    """Recursively gather contents of all files under the given node.
-
-    This function recursively processes a directory node and gathers the contents of all files
-    under that node. It returns the concatenated content of all files as a single string.
-
-    Parameters
-    ----------
-    node : FileSystemNode
-        The current directory or file node being processed.
-
-    Returns
-    -------
-    str
-        The concatenated content of all files under the given node.
-
-    """
-    if node.type != FileSystemNodeType.DIRECTORY:
-        return node.content_string
-
-    # Recursively gather contents of all files under the current directory
-    return "\n".join(_gather_file_contents(child) for child in node.children)
-
-
 def _create_tree_structure(
     query: IngestionQuery,
     *,
@@ -162,16 +138,10 @@ def _create_tree_structure(
     tree_str = ""
     current_prefix = "└── " if is_last else "├── "
 
-    # Indicate directories with a trailing slash
-    display_name = node.name
-    if node.type == FileSystemNodeType.DIRECTORY:
-        display_name += "/"
-    elif node.type == FileSystemNodeType.SYMLINK:
-        display_name += " -> " + readlink(node.path).name
-
-    tree_str += f"{prefix}{current_prefix}{display_name}\n"
+    # Use polymorphic display name - handles files, dirs, symlinks automatically!
+    tree_str += f"{prefix}{current_prefix}{node.display_name}\n"
 
-    if node.type == FileSystemNodeType.DIRECTORY and node.children:
+    if node.children:
         prefix += "    " if is_last else "│   "
         for i, child in enumerate(node.children):
             tree_str += _create_tree_structure(query, node=child, prefix=prefix, is_last=i == len(node.children) - 1)

src/gitingest/schemas/__init__.py

@@ -1,7 +1,25 @@
 """Module containing the schemas for the Gitingest package."""
 
 from gitingest.schemas.cloning import CloneConfig
-from gitingest.schemas.filesystem import FileSystemNode, FileSystemNodeType, FileSystemStats
+from gitingest.schemas.filesystem import (
+    FileSystemDirectory,
+    FileSystemFile,
+    FileSystemNode,
+    FileSystemStats,
+    FileSystemSymlink,
+    GitRepository,
+)
 from gitingest.schemas.ingestion import IngestionQuery
+from gitingest.schemas.source import Source
 
-__all__ = ["CloneConfig", "FileSystemNode", "FileSystemNodeType", "FileSystemStats", "IngestionQuery"]
+__all__ = [
+    "CloneConfig",
+    "FileSystemDirectory",
+    "FileSystemFile",
+    "FileSystemNode",
+    "FileSystemStats",
+    "FileSystemSymlink",
+    "GitRepository",
+    "IngestionQuery",
+    "Source",
+]