Optimised SubtitleListModel (#67)

* Cache first and last line numbers

Saves nearly 20 seconds on a 2000 line translated file!

* Cache size hints

Bucket lines according to a height approximation and cache the expected size. Saves 3+ seconds on a 2000 line translated file.

* Use map to look up row number

Not measured, but an obvious choice

GUI/GuiHelpers.py

@@ -21,3 +21,17 @@ def GetInstructionFiles():
     instruction_path = GetResourcePath("")
     files = os.listdir(instruction_path)
     return [ file for file in files if file.lower().startswith("instructions") ]
+
+def GetLineHeight(text: str, wrap_length: int = 60) -> int:
+    """
+    Calculate the number of lines for a given text with wrapping and newline characters.
+
+    :param text: The input text.
+    :param wrap_length: The maximum number of characters per line.
+    :return: The total number of lines.
+    """
+    if not text:
+        return 0
+
+    wraps = -(-len(text) // wrap_length) if wrap_length else None  # Ceiling division
+    return text.count('\n') + wraps

GUI/ProjectViewModel.py

@@ -3,6 +3,7 @@
 from PySide6.QtCore import Qt, QModelIndex
 
 from PySide6.QtGui import QStandardItemModel, QStandardItem
+from GUI.GuiHelpers import GetLineHeight
 from PySubtitleGPT import SubtitleLine
 
 from PySubtitleGPT.Helpers import FormatMessages, Linearise, UpdateFields
@@ -85,10 +86,9 @@ def GetLineItem(self, line_number, get_translated):
 
                     if batch_item.last_line_number >= line_number:
                         lines = batch_item.translated if get_translated else batch_item.originals
-                        if lines:
-                            for line_item in lines.values():
-                                if line_item.number == line_number:
-                                    return line_item
+                        line = lines.get(line_number, None) if lines else None
+                        if line:
+                            return line
 
     def GetBatchNumbers(self):
         """
@@ -421,6 +421,7 @@ def __init__(self, is_translation, line_number, model):
         self.is_translation = is_translation
         self.number = line_number
         self.line_model = model
+        self.height = GetLineHeight(model.get('text'))
 
         self.setData(self.line_model, Qt.ItemDataRole.UserRole)
 
@@ -433,6 +434,8 @@ def Update(self, line_update):
         if line_update.get('number'):
             self.number = line_update['number']
 
+        self.height = GetLineHeight(self.line_model.get('text'))
+
         self.setData(self.line_model, Qt.ItemDataRole.UserRole)
 
     def __str__(self) -> str:
@@ -477,6 +480,10 @@ def __init__(self, scene_number, batch : SubtitleBatch):
             'errors': self._get_errors(batch.errors)
         }
 
+        # cache on demand
+        self._first_line_num = None
+        self._last_line_num = None
+
         if batch.translation and os.environ.get("DEBUG_MODE") == "1":
             self.batch_model.update({
                 'response': batch.translation.text,
@@ -485,7 +492,6 @@ def __init__(self, scene_number, batch : SubtitleBatch):
             if batch.translation.prompt:
                 self.batch_model['messages'] = FormatMessages(batch.translation.prompt.messages)
 
-
         self.setData(self.batch_model, Qt.ItemDataRole.UserRole)
 
     def AddLineItem(self, is_translation : bool, line_number : int, model : dict):
@@ -497,6 +503,8 @@ def AddLineItem(self, is_translation : bool, line_number : int, model : dict):
         else:
             self.originals[line_number] = line_item
 
+        self._invalidate_first_and_last()
+
     @property
     def original_count(self):
         return len(self.originals)
@@ -531,11 +539,15 @@ def response(self):
 
     @property
     def first_line_number(self):
-        return min(num for num in self.originals.keys() if num) if self.originals else None
+        if not self._first_line_num:
+            self._update_first_and_last()
+        return self._first_line_num
 
     @property
     def last_line_number(self):
-        return max(num for num in self.originals.keys() if num) if self.originals else None    
+        if not self._last_line_num:
+            self._update_first_and_last()
+        return self._last_line_num    
 
     @property
     def has_errors(self):
@@ -575,7 +587,16 @@ def GetContent(self):
                 'errors' : self.has_errors
             }
         }
-
+
+    def _update_first_and_last(self):
+        line_numbers = [ num for num in self.originals.keys() if num ] if self.originals else None
+        self._first_line_num = min(line_numbers) if line_numbers else None
+        self._last_line_num = max(line_numbers)
+
+    def _invalidate_first_and_last(self):
+        self._first_line_num = None
+        self._last_line_num = None
+
     def _get_errors(self, errors):
         if errors:
             if all(isinstance(e, Exception) for e in errors):

GUI/SubtitleListModel.py

@@ -1,5 +1,5 @@
 import logging
-from PySide6.QtCore import QAbstractProxyModel, QModelIndex, Qt
+from PySide6.QtCore import QAbstractProxyModel, QModelIndex, Qt, QSize
 from GUI.ProjectViewModel import BatchItem, ProjectViewModel, SceneItem, LineItem, ViewModelItem
 from GUI.ProjectSelection import ProjectSelection
 from GUI.Widgets.Widgets import LineItemView
@@ -11,6 +11,8 @@ def __init__(self, show_translated, viewmodel=None, parent=None):
         self.viewmodel : ProjectViewModel = viewmodel
         self.selected_batch_numbers = []
         self.visible = []
+        self.visible_row_map : dict(int, int) = {}
+        self.size_map : dict(int, QSize ) = {}
 
         # Connect signals to update mapping when source model changes
         if self.viewmodel:
@@ -50,16 +52,16 @@ def ShowSelectedBatches(self, batch_numbers):
                     visible.extend(visible_lines)
 
         self.visible = visible
+        self.visible_row_map = { item[2] : row for row, item in enumerate(self.visible) }
         self.layoutChanged.emit()
 
     def mapFromSource(self, source_index : QModelIndex):
         item : ViewModelItem = self.viewmodel.itemFromIndex(source_index)
 
         if isinstance(item, LineItem):
-            for row, key in enumerate(self.visible):
-                scene, batch, line = key
-                if line == item.number:
-                    return self.index(row, 0, QModelIndex())
+            row = self.visible_row_map.get(item.number, None)
+            if row is not None:
+                return self.index(row, 0, QModelIndex())
 
         return QModelIndex()
 
@@ -144,7 +146,15 @@ def data(self, index, role):
             return LineItemView(item)
 
         if role == Qt.ItemDataRole.SizeHintRole:
-            return LineItemView(item).sizeHint()
+            if not item.height:
+                return LineItemView(item).sizeHint()
+
+            if item.height in self.size_map:
+                return self.size_map[item.height]
+
+            size = LineItemView(item).sizeHint()
+            self.size_map[item.height] = size
+            return size
 
         return None
 

gui-subtrans.py

@@ -2,10 +2,13 @@
 import logging
 import os
 import sys
+import cProfile
+from pstats import Stats
 
 from PySide6.QtWidgets import QApplication
 from GUI.MainWindow import MainWindow
-from PySubtitleGPT.Options import Options, settings_path
+from PySubtitleGPT.Options import Options, settings_path, config_dir
+
 
 # This seems insane but ChatGPT told me to do it.
 project_dir = os.path.abspath(os.path.dirname(__file__))
@@ -31,6 +34,7 @@ def parse_arguments():
     parser.add_argument('--theme', type=str, default=None, help="Stylesheet to load")
     parser.add_argument('--firstrun', action='store_true', help="Show the first-run options dialog on launch")
     parser.add_argument('--includeoriginal', action='store_true', help="Include the original text in the translated subtitles")
+    parser.add_argument('--profile', action='store_true', help="Profile execution and write stats to the console")
 
     try:
         args = parser.parse_args()
@@ -50,11 +54,28 @@ def parse_arguments():
         'scene_threshold': args.scenethreshold,
         'project': args.project and args.project.lower() or 'true',
         'theme': args.theme,
-        'firstrun': args.firstrun
+        'firstrun': args.firstrun,
+        'profile': args.profile
     }
 
     return arguments, args.filepath
 
+def run_with_profiler(app):
+    profiler = cProfile.Profile()
+    profiler.enable()
+
+    app.exec()
+
+    profiler.disable()
+
+    profile_path = os.path.join(config_dir, 'profile_guisubtrans.txt')
+    with open(profile_path, 'w') as stream:
+        stats = Stats(profiler, stream=stream)
+        stats.sort_stats('tottime')
+        stats.print_stats(100)
+
+    logging.info(f"Profiling stats written to {profile_path}")
+
 if __name__ == "__main__":
     app = QApplication(sys.argv)
 
@@ -71,4 +92,7 @@ def parse_arguments():
     app.main_window = MainWindow( options=options, filepath=filepath)
     app.main_window.show()
 
-    app.exec()
+    if arguments.get('profile'):
+        run_with_profiler(app)
+    else:
+        app.exec()