polishing embeddings steps

4 files changed, 21 insertions, 10 deletions

diff --git a/.gitignore b/.gitignore
index 10dfccca..a4f53b82 100644
--- a/.gitignore
+++ b/.gitignore
@@ -136,3 +136,5 @@ notes.txt
 cached_embeddings.pkl
 .ruff_cache
 codeql
+
+**/.continue
\ No newline at end of file
diff --git a/continuedev/src/continuedev/core/policy.py b/continuedev/src/continuedev/core/policy.py
index 6f357ba0..926cc624 100644
--- a/continuedev/src/continuedev/core/policy.py
+++ b/continuedev/src/continuedev/core/policy.py
@@ -17,7 +17,7 @@ class DemoPolicy(Policy):
     def next(self, history: History) -> Step:
         # At the very start, run initial Steps spcecified in the config
         if history.get_current() is None:
-            return CreateCodebaseIndexChroma() >> MessageStep(message="Welcome to Continue!") >> StepsOnStartupStep()
+            return MessageStep(message="Welcome to Continue!") >> CreateCodebaseIndexChroma() >> StepsOnStartupStep()
 
         observation = history.get_current().observation
         if observation is not None and isinstance(observation, UserInputObservation):
diff --git a/continuedev/src/continuedev/libs/chroma/query.py b/continuedev/src/continuedev/libs/chroma/query.py
index 1cb178cc..c27329f0 100644
--- a/continuedev/src/continuedev/libs/chroma/query.py
+++ b/continuedev/src/continuedev/libs/chroma/query.py
@@ -38,21 +38,26 @@ class ChromaIndexManager:
             return None
 
     def check_index_exists(self):
-        return os.path.exists(self.index_dir)
+        return os.path.exists(os.path.join(self.index_dir, "metadata.json"))
 
     def create_codebase_index(self):
         """Create a new index for the current branch."""
         if not self.check_index_exists():
             os.makedirs(self.index_dir)
+        else:
+            return
 
-        print("ROOT DIRECTORY: ", self.git_root_dir)
-        documents = load_gpt_index_documents(self.git_root_dir)
+        documents = load_gpt_index_documents(self.workspace_dir)
 
         chunks = {}
         doc_chunks = []
         for doc in documents:
             text_splitter = TokenTextSplitter()
-            text_chunks = text_splitter.split_text(doc.text)
+            try:
+                text_chunks = text_splitter.split_text(doc.text)
+            except:
+                print("ERROR (probably found special token): ", doc.text)
+                continue
             filename = doc.extra_info["filename"]
             chunks[filename] = len(text_chunks)
             for i, text in enumerate(text_chunks):
@@ -88,9 +93,9 @@ class ChromaIndexManager:
         modified_deleted_files = [f for f in modified_deleted_files if f]
 
         deleted_files = [
-            f for f in modified_deleted_files if not os.path.exists(self.git_root_dir + "/" + f)]
+            f for f in modified_deleted_files if not os.path.exists(os.path.join(self.workspace_dir, f))]
         modified_files = [
-            f for f in modified_deleted_files if os.path.exists(self.git_root_dir + "/" + f)]
+            f for f in modified_deleted_files if os.path.exists(os.path.join(self.workspace_dir, f))]
 
         return filter_ignored_files(modified_files, self.index_dir), filter_ignored_files(deleted_files, self.index_dir)
 
diff --git a/continuedev/src/continuedev/steps/chroma.py b/continuedev/src/continuedev/steps/chroma.py
index 556f7252..7bb9389e 100644
--- a/continuedev/src/continuedev/steps/chroma.py
+++ b/continuedev/src/continuedev/steps/chroma.py
@@ -13,10 +13,12 @@ class CreateCodebaseIndexChroma(Step):
     hide: bool = True
 
     async def describe(self, llm) -> Coroutine[str, None, None]:
-        return "Creating a codebase index for the current branch."
+        return "Indexing the codebase..."
 
     async def run(self, sdk: ContinueSDK) -> Coroutine[Observation, None, None]:
         index = ChromaIndexManager(await sdk.ide.getWorkspaceDirectory())
+        if not index.check_index_exists():
+            self.hide = False
         index.create_codebase_index()
 
 
@@ -55,7 +57,9 @@ class AnswerQuestionChroma(Step):
             Here is the answer:""")
 
         answer = (await sdk.models.gpt35()).complete(prompt)
-        print(answer)
+        # Make paths relative to the workspace directory
+        answer = answer.replace(await sdk.ide.getWorkspaceDirectory(), "")
+
         self._answer = answer
 
         await sdk.ide.setFileOpen(files[0])
@@ -67,7 +71,7 @@ class EditFileChroma(Step):
 
     async def run(self, sdk: ContinueSDK) -> Coroutine[Observation, None, None]:
         index = ChromaIndexManager(await sdk.ide.getWorkspaceDirectory())
-        results = index.query_codebase_index(self.question)
+        results = index.query_codebase_index(self.request)
 
         resource_name = list(
             results.source_nodes[0].node.relationships.values())[0]