Shuffled context code.

app.py

@@ -113,18 +113,12 @@ def moderator(message):
 
 
 # Retrieve context from the ChromaDB.
-def get_context(message):
+def get_context(message, num):
     if not collection:
         return ""
-    ctxsize = os.getenv("CTXSIZE")
-    if not ctxsize:
-        ctxsize = 3  # 9 seems to introduce rubbish...
-    else:
-        ctxsize = int(ctxsize)
-    DBG("get_context: " + str(ctxsize))
     results = collection.query(
         query_texts=[message],
-        n_results=ctxsize,
+        n_results=num,
     )
     data = rerank(message, results)
     # data = results["documents"]  # [0][0]
@@ -136,8 +130,8 @@ def get_context(message):
 
 
 # Hybrid retriever from hybrid, uses pufendorfstore.
-def get_hybrid_context(message):
-    documents = retrieve(hybrid_retrieval, message, top_k=3, scale=True)
+def get_hybrid_context(message, num):
+    documents = retrieve(hybrid_retrieval, message, top_k=num, scale=True)
     return documents
 
 
@@ -348,8 +342,20 @@ with gr.Blocks(js=js_func, theme="Monochrome") as demo_blocks:
             yield history, gr.update()
             return
 
-        context = get_context(user_message)
-        hybrid_context = get_hybrid_context(user_message)
+        ctxkeep = os.getenv("CTXKEEP")
+        if not ctxkeep:
+            ctxkeep = 3
+        else:
+            ctxkeep = int(ctxkeep)
+        hybridkeep = os.getenv("HYBRIDKEEP")
+        if not hybridkeep:
+            hybridkeep = 3
+        else:
+            hybridkeep = int(hybridkeep)
+
+        context = get_context(user_message, ctxkeep)
+        hybrid_context = get_hybrid_context(user_message, hybridkeep)
+
         for hc in hybrid_context:
             DBG(
                 str(hc.meta["file_path"])
@@ -363,28 +369,15 @@ with gr.Blocks(js=js_func, theme="Monochrome") as demo_blocks:
         for x in context:
             DBG(x)
         # Take the top-3, they have already been reranked in the get_context(...) fn.
-        ctxkeep = os.getenv("CTXKEEP")
-        if not ctxkeep:
-            ctxkeep = 3
-        else:
-            ctxkeep = int(ctxkeep)
-        DBG("CONTEXT")
         DBG("context keep: " + str(ctxkeep))
-        context = context[0:ctxkeep]
         context_str = ""
         if ctxkeep > 0:
             for i, x in enumerate(context):  # note different after reranking
                 DBG(x)
                 context_str += x + "\n\n"
         # The hc is the new haystack contents.
-        hybridkeep = os.getenv("HYBRIDKEEP")
-        if not hybridkeep:
-            hybridkeep = 3
-        else:
-            hybridkeep = int(hybridkeep)
         DBG("hybrid context keep: " + str(hybridkeep))
         if hybridkeep > 0:
-            hybrid_context = hybrid_context[0:hybridkeep]
             for i, x in enumerate(hybrid_context):
                 DBG(x)
                 context_str += x.content + "\n\n"

vector3_db/a1b2bf9f-4f30-46a6-a6c2-b6ca99effce9/data_level0.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5c9af344e120a5be64710368f5276638e6f52b1f06593604c5de84036a6f72b0
+oid sha256:33dca61881e04f1a2571b2ec84053c3c673fcb03982b223f8401454a5f0ec6c2
 size 16760000

vector3_db/a1b2bf9f-4f30-46a6-a6c2-b6ca99effce9/length.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:248727a598e2d668480e8705407c5f5c97986b7e407a2382bbf44f9120743b09
+oid sha256:ee2dfe0fac21ee43690bb2a35b3438cd9ac4872685ceecaaa55ac1641b188be0
 size 40000

vector3_db/chroma.sqlite3

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7485f035f1cb2d45b176f1cb03183a22a5522e9cb6d6f8e75e30af74cba13ef1
+oid sha256:9b9dbe7b759b87c6de968679094d0211bc13eff736a7e926a61c5a38712489a8
 size 11452416