From e94338909b1f142479b83ac7d8be50e93b6f695e Mon Sep 17 00:00:00 2001
From: Fabian Moertter <fabian.moertter@gmx.net>
Date: Wed, 2 Aug 2023 13:05:14 +0200
Subject: [PATCH 1/2] Caching: cache embedded_pdf

---
 pipelines.py | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/pipelines.py b/pipelines.py
index 1927cff..fe5dbda 100644
--- a/pipelines.py
+++ b/pipelines.py
@@ -1,9 +1,11 @@
+from functools import lru_cache
 from filepreprocessing import pdf_get_text_chunks
 from tqdm import tqdm
 from embedding import get_embedding_sentence_transformer
 from similarity import model_qa, cosine_similarity
 
 
+@lru_cache(maxsize=64)
 def embedding_loaded_pdf(file_path, chunk_size, overlap):
 
     # FIRST WE LOAD PDF

From f7c27bca4c8f1ad2bb6745ecf44fc83c04225508 Mon Sep 17 00:00:00 2001
From: Fabian Moertter <fabian.moertter@gmx.net>
Date: Wed, 2 Aug 2023 13:28:16 +0200
Subject: [PATCH 2/2] Features: add/remove/search multiple docs

fixes and adding comments
---
 app.py | 72 ++++++++++++++++++++++++++++++++++++++++++++++++++++++----
 1 file changed, 67 insertions(+), 5 deletions(-)

diff --git a/app.py b/app.py
index 31debb5..55e6057 100644
--- a/app.py
+++ b/app.py
@@ -33,10 +33,13 @@
         ui.nav('Ask File',
                 ui.layout_sidebar(
                    ui.panel_sidebar(
-                       ui.input_file('document_input_file',
-                                     'Select a PDF file you wish to ask a question about',
-                                     multiple=False, accept='.pdf', button_label='Select',
+                       ui.input_file('document_input_files',
+                                     'Select one or more PDF file(s) you wish to ask a question about',
+                                     multiple=True, accept='.pdf', button_label='Select',
                                      placeholder='Your PDF here..'),
+                       ui.input_checkbox_group('selected_files', '', []),
+                       # TODO: Hide remove button behind panel_conditional when no file is uploaded yet
+                       ui.input_action_button('remove_selected', label="Remove selected PDF(s)"),
                        ui.input_text_area('question_input_file', 'What wisdom do you seek from this file?', rows=4),
                        ui.input_slider('n_chunks_file', 'Number of chunks', min=1, max=5, value=3),
                        ui.input_action_button(id="run_process_file", label="Do Magic", class_='btn-success'),
@@ -45,7 +48,7 @@
                    ),
                    ui.panel_main(
                        ui.panel_conditional(
-                           """input.run_process_file > 0 && input.question_input_file != ''""",  # && input.document_input_file != null
+                           """input.run_process_file > 0 && input.question_input_file != ''""",  # && input.document_input_files != null
                            ui.output_text('get_answer_file'),
                         ),
                    width=8,
@@ -60,6 +63,19 @@
 
 def server(input, output, session):
 
+    documents = reactive.Value([])
+
+    @reactive.Effect
+    @reactive.event(input.document_input_files)
+    def _():
+        docs = documents.get()
+        document_names = [ file['name'] for file in docs ]
+        
+        for file in input.document_input_files():
+            if file['name'] not in document_names:
+                docs.append(file)
+        documents.set(docs)
+
     val = reactive.Value(3)
     @reactive.Effect
     @reactive.event(input.n_chunks_db)
@@ -83,11 +99,57 @@ async def get_answer_db():
         answer = re.sub('\s*</s>$', '', answer)
         return answer
 
+    @reactive.Effect
+    @reactive.event(input.document_input_files)
+    def _():
+        """
+        Update checkbox_group after uploading file(s).
+        """
+        docs = documents.get()
+        choices = [ file['name'] for file in docs ]
+        ui.update_checkbox_group('selected_files', label="Selected file(s):", choices=choices, selected=choices)
+
+    @reactive.Effect
+    @reactive.event(documents.get)
+    def _():
+        """
+        Update checkbox_group after deleting file(s).
+        """
+        docs = documents.get()
+        choices = [ file['name'] for file in docs ]
+        if choices:
+            label = "Selected file(s):"
+        else:
+            label = ""
+        ui.update_checkbox_group('selected_files', label=label, choices=choices, selected=[])
+        # Use this to auto select remaining documents after deletion
+        # ui.update_checkbox_group('selected_files', label=label, choices=choices, selected=choices)
+
+    @reactive.Effect
+    @reactive.event(input.remove_selected)
+    def _():
+        """
+        Remove selected file(s).
+        """
+        docs = documents.get()
+        docs_to_keep = [ file for file in docs if file['name'] not in input.selected_files() ]
+        # Use this to keep not selected files
+        # docs_to_keep = [ file for file in docs if file['name'] in input.selected_files() ]
+        documents.set(docs_to_keep)
+
     @output()
     @render.text
     @reactive.event(input.run_process_file)
     async def get_answer_file():
-        db_items = embedding_loaded_pdf(file_path=input.document_input_file()[0]['datapath'], chunk_size=200, overlap=10)
+        db_items = []
+
+        docs = documents.get()
+
+        for file in docs:
+            if file['name'] in input.selected_files():
+                file['db_items'] = embedding_loaded_pdf(file_path=input.document_input_files()[0]['datapath'], chunk_size=200, overlap=10)
+                db_items.extend(file['db_items'])
+
         answer = pipeline_return_question_and_answer(query=input.question_input_file(),
                                                      db_items=db_items,
                                                      n_chunks=input.n_chunks_file())