From f6403089bfeac9c03670282b24213956f57b6202 Mon Sep 17 00:00:00 2001
From: granawkins <granthawkins88@gmail.com>
Date: Fri, 21 Jun 2024 11:02:40 +0700
Subject: [PATCH 1/2] build directory nodes from graph instead of filesys

---
 ragdaemon/annotators/hierarchy.py | 41 ++++++++++++++-----------------
 1 file changed, 18 insertions(+), 23 deletions(-)

diff --git a/ragdaemon/annotators/hierarchy.py b/ragdaemon/annotators/hierarchy.py
index 40efd56..0f464f8 100644
--- a/ragdaemon/annotators/hierarchy.py
+++ b/ragdaemon/annotators/hierarchy.py
@@ -66,37 +66,32 @@ async def annotate(
             for parent in path.parents:
                 if len(parent.parts) == 0:
                     parent = Path("ROOT")
-                directories.add(parent)
+                directories.add(parent.as_posix())
                 edges.add((parent.as_posix(), _last.as_posix()))
                 _last = parent
 
-        for dir in directories:
-            dir_str = dir.as_posix()
-            dir_path = dir if dir != Path("ROOT") else Path(".")
-            document = get_document(
-                dir_str, cwd, type="directory", ignore_patterns=self.ignore_patterns
-            )
-            checksum = hash_str(
-                "".join(
-                    checksums.get(dir_path / subpath, "")
-                    for subpath in document.split("\n")[1:]
-                )
-            )
+        for source, target in edges:
+            for id in (source, target):
+                if id not in graph and id not in directories:
+                    raise RagdaemonError(f"Node {id} not found in graph")
+            graph.add_edge(source, target, type="hierarchy")
+
+        # Fill-in directory data (same process as get_document for dirs, but more efficient)
+        for dir in sorted(
+            directories, key=lambda x: len(x) if x != "ROOT" else 0, reverse=True
+        ):
+            children = sorted(node for node in graph.successors(dir))
+            document = f"{dir}\n" + "\n".join(children)
+            checksum = hash_str("".join(checksums[Path(child)] for child in children))
             data = {
-                "id": dir_str,
+                "id": dir,
                 "type": "directory",
-                "ref": dir_str,
+                "ref": dir,
                 "document": document,
                 "checksum": checksum,
             }
-            graph.add_node(dir_str, **data)
-            checksums[dir] = checksum
-
-        for source, target in edges:
-            for id in (source, target):
-                if id not in graph:
-                    raise RagdaemonError(f"Node {id} not found in graph")
-            graph.add_edge(source, target, type="hierarchy")
+            checksums[Path(dir)] = checksum
+            graph.nodes[dir].update(data)
 
         # Sync with remote DB
         ids = list(set(checksums.values()))

From 27638c3a835c3c7544b4ce7a809bf73023dbd74d Mon Sep 17 00:00:00 2001
From: granawkins <granthawkins88@gmail.com>
Date: Fri, 21 Jun 2024 11:03:15 +0700
Subject: [PATCH 2/2] version bump

---
 pyproject.toml        | 4 ++--
 ragdaemon/__init__.py | 2 +-
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/pyproject.toml b/pyproject.toml
index d3e752a..6736684 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -7,7 +7,7 @@ packages=["ragdaemon"]
 
 [project]
 name = "ragdaemon"
-version = "0.7.7"
+version = "0.7.8"
 description = "Generate and render a call graph for a Python project."
 readme = "README.md"
 dependencies = [
@@ -47,4 +47,4 @@ dev = [
 ]
 
 [tool.pyright]
-ignore = ["tests/sample"]
\ No newline at end of file
+ignore = ["tests/sample"]
diff --git a/ragdaemon/__init__.py b/ragdaemon/__init__.py
index 4e27eed..894cebc 100644
--- a/ragdaemon/__init__.py
+++ b/ragdaemon/__init__.py
@@ -1 +1 @@
-__version__ = "0.7.7"
+__version__ = "0.7.8"