v8.2

Author: BBC-Esq

src/database_interactions.py

@@ -29,20 +29,19 @@
 
 from typing import Optional, Any, Iterator, Union
 from langchain_community.document_loaders.blob_loaders import Blob
-
 from langchain_community.document_loaders.parsers import PyMuPDFParser
 import pymupdf
 
 from constants import DOCUMENT_LOADERS
 from extract_metadata import extract_document_metadata, add_pymupdf_page_metadata, compute_content_hash
 
-logging.basicConfig(
-    level=logging.ERROR,
-    format='%(asctime)s - %(levelname)s - %(message)s',
-    handlers=[
-        logging.FileHandler('document_processor.log', mode='w')
-    ]
-)
+# logging.basicConfig(
+    # level=logging.ERROR,
+    # format='%(asctime)s - %(levelname)s - %(message)s',
+    # handlers=[
+        # logging.FileHandler('document_processor.log', mode='w')
+    # ]
+# )
 
 warnings.filterwarnings("ignore", category=FutureWarning)
 warnings.filterwarnings("ignore", category=UserWarning)
@@ -55,12 +54,12 @@
 from typing import List
 
 class FixedSizeTextSplitter:
-    """Splits text into equally-sized character chunks without regex.
+    """Splits text into equally-sized character chunks.
 
     Parameters
     ----------
     chunk_size : int
-        Maximum characters per chunk.  Comes straight from config.yaml.
+        Maximum characters per chunk.  Taken from config.yaml.
     """
 
     def __init__(self, chunk_size: int):
@@ -104,7 +103,7 @@ def __init__(self, file_path: Union[str, PurePath], **kwargs: Any) -> None:
             extract_images=kwargs.get('extract_images', False)
         )
 
-# ensure all loader class names map correctly
+# map loaders
 for ext, loader_name in DOCUMENT_LOADERS.items():
     DOCUMENT_LOADERS[ext] = globals()[loader_name]
 

src/document_processor.py

@@ -1,28 +1,42 @@
 import logging
 from functools import partial
-from PySide6.QtWidgets import QVBoxLayout, QGroupBox, QPushButton, QHBoxLayout, QWidget, QMessageBox
+from PySide6.QtWidgets import (
+    QVBoxLayout,
+    QGroupBox,
+    QPushButton,
+    QHBoxLayout,
+    QWidget,
+    QMessageBox,
+)
 from gui_tabs_settings_server import ServerSettingsTab
 from gui_tabs_settings_database_create import ChunkSettingsTab
 from gui_tabs_settings_database_query import DatabaseSettingsTab
 from gui_tabs_settings_tts import TTSSettingsTab
 from gui_tabs_settings_vision import VisionSettingsTab
 
+
 def update_all_configs(configs):
     updated = False
     for config in configs.values():
         updated = config.update_config() or updated
     if updated:
         logging.info("config.yaml file updated")
-    
-    message = 'Settings Updated' if updated else 'No Updates'
-    details = 'One or more settings have been updated.' if updated else 'No new settings were entered.'
-    
+
+    message = "Settings Updated" if updated else "No Updates"
+    details = (
+        "One or more settings have been updated."
+        if updated
+        else "No new settings were entered."
+    )
+
     QMessageBox.information(None, message, details)
 
+
 def adjust_stretch(groups, layout):
     for group, factor in groups.items():
         layout.setStretchFactor(group, factor if group.isChecked() else 0)
 
+
 class GuiSettingsTab(QWidget):
     def __init__(self):
         super(GuiSettingsTab, self).__init__()
@@ -34,6 +48,7 @@ def __init__(self):
         }
         self.groups = {}
         self.configs = {}
+
         for title, (TabClass, stretch) in classes.items():
             settings = TabClass()
             group = QGroupBox(title)
@@ -47,7 +62,6 @@ def __init__(self):
             self.layout.addWidget(group, stretch)
             group.toggled.connect(partial(self.toggle_group, group))
 
-        # TTS
         ttsSettings = TTSSettingsTab()
         ttsGroup = QGroupBox("Text to Speech")
         ttsLayout = QVBoxLayout()
@@ -56,10 +70,9 @@ def __init__(self):
         ttsGroup.setCheckable(True)
         ttsGroup.setChecked(True)
         self.layout.addWidget(ttsGroup, 3)
-        self.groups[ttsGroup] = 2
+        self.groups[ttsGroup] = 3
         ttsGroup.toggled.connect(partial(self.toggle_tts_group, ttsSettings))
 
-        # VisionSettingsTab - handled separately
         visionSettings = VisionSettingsTab()
         visionGroup = QGroupBox("Vision Models")
         visionLayout = QVBoxLayout()
@@ -68,7 +81,7 @@ def __init__(self):
         visionGroup.setCheckable(True)
         visionGroup.setChecked(True)
         self.layout.addWidget(visionGroup, 2)
-        self.groups[visionGroup] = 1
+        self.groups[visionGroup] = 2
         visionGroup.toggled.connect(partial(self.toggle_vision_group, visionSettings))
 
         self.update_all_button = QPushButton("Update Settings")
@@ -79,6 +92,7 @@ def __init__(self):
         center_button_layout.addWidget(self.update_all_button)
         center_button_layout.addStretch(1)
         self.layout.addLayout(center_button_layout)
+
         self.setLayout(self.layout)
         adjust_stretch(self.groups, self.layout)
 

src/gui_tabs_settings.py

@@ -4,98 +4,98 @@
 
 from constants import TOOLTIPS
 
+
 class ChunkSettingsTab(QWidget):
     def __init__(self):
         super(ChunkSettingsTab, self).__init__()
-        with open('config.yaml', 'r', encoding='utf-8') as f:
+        with open("config.yaml", "r", encoding="utf-8") as f:
             config_data = yaml.safe_load(f)
-            self.database_config = config_data['database']
-            self.compute_device_options = config_data['Compute_Device']['available']
-            self.database_creation_device = config_data['Compute_Device']['database_creation']
-        
+            self.database_config = config_data["database"]
+            self.compute_device_options = config_data["Compute_Device"]["available"]
+            self.database_creation_device = config_data["Compute_Device"]["database_creation"]
+
         grid_layout = QGridLayout()
-        
+
         # Device selection and current setting
         self.device_label = QLabel("Device:")
         self.device_label.setToolTip(TOOLTIPS["CREATE_DEVICE_DB"])
         grid_layout.addWidget(self.device_label, 0, 0)
-        
+
         self.device_combo = QComboBox()
         self.device_combo.addItems(self.compute_device_options)
         self.device_combo.setToolTip(TOOLTIPS["CREATE_DEVICE_DB"])
         if self.database_creation_device in self.compute_device_options:
-            self.device_combo.setCurrentIndex(self.compute_device_options.index(self.database_creation_device))
+            self.device_combo.setCurrentIndex(
+                self.compute_device_options.index(self.database_creation_device)
+            )
         self.device_combo.setMinimumWidth(100)
         grid_layout.addWidget(self.device_combo, 0, 2)
-        
+
         self.current_device_label = QLabel(f"{self.database_creation_device}")
         self.current_device_label.setToolTip(TOOLTIPS["CREATE_DEVICE_DB"])
         grid_layout.addWidget(self.current_device_label, 0, 1)
-        
+
         # Chunk size and current setting
         self.chunk_size_label = QLabel("Chunk Size (# characters):")
         self.chunk_size_label.setToolTip(TOOLTIPS["CHUNK_SIZE"])
         grid_layout.addWidget(self.chunk_size_label, 0, 3)
-        
+
         self.chunk_size_edit = QLineEdit()
         self.chunk_size_edit.setPlaceholderText("Enter new chunk_size...")
         self.chunk_size_edit.setValidator(QIntValidator(1, 1000000))
         self.chunk_size_edit.setToolTip(TOOLTIPS["CHUNK_SIZE"])
         grid_layout.addWidget(self.chunk_size_edit, 0, 5)
-        
-        current_size = self.database_config.get('chunk_size', '')
+
+        current_size = self.database_config.get("chunk_size", "")
         self.current_size_label = QLabel(f"{current_size}")
         self.current_size_label.setToolTip(TOOLTIPS["CHUNK_SIZE"])
         grid_layout.addWidget(self.current_size_label, 0, 4)
-        
+
         # Chunk overlap and current setting
         self.chunk_overlap_label = QLabel("Overlap (# characters):")
         self.chunk_overlap_label.setToolTip(TOOLTIPS["CHUNK_OVERLAP"])
         grid_layout.addWidget(self.chunk_overlap_label, 0, 6)
-        
+
         self.chunk_overlap_edit = QLineEdit()
         self.chunk_overlap_edit.setPlaceholderText("Enter new chunk_overlap...")
         self.chunk_overlap_edit.setValidator(QIntValidator(0, 1000000))
         self.chunk_overlap_edit.setToolTip(TOOLTIPS["CHUNK_OVERLAP"])
         grid_layout.addWidget(self.chunk_overlap_edit, 0, 8)
-        
-        current_overlap = self.database_config.get('chunk_overlap', '')
+
+        current_overlap = self.database_config.get("chunk_overlap", "")
         self.current_overlap_label = QLabel(f"{current_overlap}")
         self.current_overlap_label.setToolTip(TOOLTIPS["CHUNK_OVERLAP"])
         grid_layout.addWidget(self.current_overlap_label, 0, 7)
-        
+
         # "Half-Precision" checkbox
         self.half_precision_label = QLabel("Half-Precision (2x speedup - GPU only):")
         self.half_precision_label.setToolTip(TOOLTIPS["HALF_PRECISION"])
         grid_layout.addWidget(self.half_precision_label, 1, 0, 1, 3)
-        
+
         self.half_precision_checkbox = QCheckBox()
-        self.half_precision_checkbox.setChecked(self.database_config.get('half', False))
+        self.half_precision_checkbox.setChecked(self.database_config.get("half", False))
         self.half_precision_checkbox.setToolTip(TOOLTIPS["HALF_PRECISION"])
         grid_layout.addWidget(self.half_precision_checkbox, 1, 3)
-        
+
         self.setLayout(grid_layout)
 
     def update_config(self):
         try:
-            with open('config.yaml', 'r', encoding='utf-8') as f:
+            with open("config.yaml", "r", encoding="utf-8") as f:
                 config_data = yaml.safe_load(f)
         except Exception as e:
             QMessageBox.critical(
-                self, 
-                "Error Loading Configuration", 
-                f"An error occurred while loading the configuration: {e}"
+                self,
+                "Error Loading Configuration",
+                f"An error occurred while loading the configuration: {e}",
             )
             return False
-        
+
         settings_changed = False
         errors = []
 
         new_device = self.device_combo.currentText()
-        if new_device != self.database_creation_device:
-            device_changed = True
-        else:
-            device_changed = False
+        device_changed = new_device != self.database_creation_device
 
         new_chunk_size_text = self.chunk_size_edit.text().strip()
         if new_chunk_size_text:
@@ -106,7 +106,7 @@ def update_config(self):
             except ValueError as ve:
                 errors.append(f"Chunk size must be a positive integer: {str(ve)}")
         else:
-            new_chunk_size = self.database_config.get('chunk_size', 0)
+            new_chunk_size = self.database_config.get("chunk_size", 0)
 
         new_chunk_overlap_text = self.chunk_overlap_edit.text().strip()
         if new_chunk_overlap_text:
@@ -115,60 +115,70 @@ def update_config(self):
                 if new_chunk_overlap < 0:
                     raise ValueError("Chunk overlap cannot be negative.")
             except ValueError as ve:
-                errors.append(f"Chunk overlap must be a non-negative integer: {str(ve)}")
+                errors.append(
+                    f"Chunk overlap must be a non-negative integer: {str(ve)}"
+                )
         else:
-            new_chunk_overlap = self.database_config.get('chunk_overlap', 0)
+            new_chunk_overlap = self.database_config.get("chunk_overlap", 0)
 
         if new_chunk_size and new_chunk_overlap >= new_chunk_size:
             errors.append("Chunk overlap must be less than chunk size.")
 
         if errors:
             error_message = "\n".join(errors)
             QMessageBox.warning(
-                self, 
-                "Invalid Input", 
-                f"The following errors occurred:\n{error_message}"
+                self, "Invalid Input", f"The following errors occurred:\n{error_message}"
             )
             return False
 
         if device_changed:
-            config_data['Compute_Device']['database_creation'] = new_device
+            config_data["Compute_Device"]["database_creation"] = new_device
             self.database_creation_device = new_device
             self.current_device_label.setText(f"{new_device}")
             settings_changed = True
 
-        if new_chunk_size_text and new_chunk_size != self.database_config.get('chunk_size', 0):
-            config_data['database']['chunk_size'] = new_chunk_size
+        if new_chunk_size_text and new_chunk_size != self.database_config.get(
+            "chunk_size", 0
+        ):
+            config_data["database"]["chunk_size"] = new_chunk_size
             self.current_size_label.setText(f"{new_chunk_size}")
             settings_changed = True
 
-        if new_chunk_overlap_text and new_chunk_overlap != self.database_config.get('chunk_overlap', 0):
-            config_data['database']['chunk_overlap'] = new_chunk_overlap
+        if new_chunk_overlap_text and new_chunk_overlap != self.database_config.get(
+            "chunk_overlap", 0
+        ):
+            config_data["database"]["chunk_overlap"] = new_chunk_overlap
             self.current_overlap_label.setText(f"{new_chunk_overlap}")
             settings_changed = True
 
         new_half_precision = self.half_precision_checkbox.isChecked()
-        if new_half_precision != self.database_config.get('half', False):
-            config_data['database']['half'] = new_half_precision
-            self.database_config['half'] = new_half_precision
+        if new_half_precision != self.database_config.get("half", False):
+            config_data["database"]["half"] = new_half_precision
             settings_changed = True
 
         if settings_changed:
             try:
-                with open('config.yaml', 'w', encoding='utf-8') as f:
+                with open("config.yaml", "w", encoding="utf-8") as f:
                     yaml.safe_dump(config_data, f)
-                
+
+                self.database_config["chunk_size"] = config_data["database"]["chunk_size"]
+                self.database_config["chunk_overlap"] = config_data["database"]["chunk_overlap"]
+                self.database_config["half"] = config_data["database"]["half"]
+
+                self.database_creation_device = config_data["Compute_Device"][
+                    "database_creation"
+                ]
+
                 self.chunk_overlap_edit.clear()
                 self.chunk_size_edit.clear()
             except Exception as e:
                 QMessageBox.critical(
-                    self, 
-                    "Error Saving Configuration", 
-                    f"An error occurred while saving the configuration: {e}"
+                    self,
+                    "Error Saving Configuration",
+                    f"An error occurred while saving the configuration: {e}",
                 )
                 return False
-
         else:
             return False
 
-        return settings_changed
+        return settings_changed