Merge pull request #48 from hnesk/performance

Performance improvements

ocrd_browser/application.py

@@ -1,12 +1,16 @@
 from gi.repository import Gio, Gtk, GLib, Gdk
 
-import pkg_resources
 from typing import List
 
 from ocrd_browser.util.gtk import ActionRegistry
 from ocrd_browser.ui import MainWindow, AboutDialog, OpenDialog
 from ocrd_browser.view import ViewRegistry
 
+try:
+    from importlib.metadata import entry_points
+except ModuleNotFoundError:
+    from importlib_metadata import entry_points  # type: ignore
+
 
 class OcrdBrowserApplication(Gtk.Application):
     # TODO: Parse arguments (with Gtk or click) to open certain views by mets+page_id+view(view_configuration_dict) and deep filename e.g.
@@ -36,7 +40,7 @@ def do_startup(self) -> None:
         self.set_accels_for_action('view.zoom_to::width', ['<Ctrl>numbersign'])
         self.set_accels_for_action('view.zoom_to::page', ['<Ctrl><Alt>numbersign'])
 
-        for entry_point in pkg_resources.iter_entry_points('ocrd_browser_ext'):
+        for entry_point in entry_points().get('ocrd_browser_ext', []):
             (entry_point.load())(self)
 
         self.load_css()
@@ -65,7 +69,7 @@ def on_quit(self, _action: Gio.SimpleAction, _param: str = None) -> None:
         open_windows: int = 0
         window: MainWindow
         for window in self.get_windows():
-            if isinstance(window, MainWindow) and window.close_confirm():  # type: ignore[unreachable]
+            if isinstance(window, MainWindow) and window.close_confirm():
                 window.destroy()
             else:
                 open_windows += 1

ocrd_browser/main.py

@@ -1,6 +1,10 @@
 #!/usr/bin/env python3
 # -*- Mode: Python; coding: utf-8; indent-tabs-mode: t; c-basic-offset: 4; tab-width: 4 -*-
+import os
 import sys
+import pstats
+import io
+import cProfile
 
 import gi
 
@@ -13,7 +17,7 @@
 
 gi.require_version('WebKit2', '4.0')
 
-from gi.repository import Gtk, Gio  # noqa: E402
+from gi.repository import Gtk, Gio, GLib  # noqa: E402
 from pathlib import Path  # noqa: E402
 from typing import Type  # noqa: E402
 from types import TracebackType  # noqa: E402
@@ -22,6 +26,11 @@
 resources = Gio.resource_load(str(BASE_PATH / "ui.gresource"))
 Gio.resources_register(resources)
 
+PROFILER = None
+if 'STARTUP_PROFILE' in os.environ:
+    PROFILER = cProfile.Profile()
+    PROFILER.enable()
+
 
 def install_excepthook() -> None:
     """ Make sure we exit when an unhandled exception occurs. """
@@ -36,7 +45,17 @@ def new_hook(type_: Type[BaseException], value: BaseException, traceback: Traceb
     sys.excepthook = new_hook
 
 
+def startup_time() -> None:
+    PROFILER.disable()
+    s = io.StringIO()
+    ps = pstats.Stats(PROFILER, stream=s).sort_stats(pstats.SortKey.TIME)
+    ps.print_stats(20)
+    print(s.getvalue())
+
+
 def main() -> None:
+    if PROFILER:
+        GLib.idle_add(startup_time)
     from ocrd_utils import initLogging
     initLogging()
     from ocrd_browser.application import OcrdBrowserApplication

ocrd_browser/model/document.py

@@ -1,36 +1,39 @@
+from __future__ import annotations
+from typing import Optional, Tuple, List, Set, Union, cast, Callable, Any, Dict, TYPE_CHECKING
+
 import atexit
 import errno
 import os
 import shutil
 from functools import wraps
 
-from ocrd import Workspace, Resolver
+from ocrd import Resolver
 from ocrd_browser.model.page import Page
 from ocrd_browser.util.file_groups import best_file_group
 from ocrd_browser.util.image import add_dpi_to_png_buffer
 from ocrd_browser.util.streams import SilencedStreams
 from ocrd_modelfactory import page_from_file
-from ocrd_models import OcrdFile
-from ocrd_models.ocrd_page_generateds import PcGtsType
 from ocrd_models.constants import NAMESPACES as NS
+from ocrd_models import OcrdFile
 from ocrd_utils import pushd_popd
 from ocrd_utils.constants import MIME_TO_EXT
 from ocrd_utils import getLogger
 
-from typing import Optional, Tuple, List, Set, Union, cast, Callable, Any, Dict
-from collections import OrderedDict
 from pathlib import Path
 from tempfile import mkdtemp
 from datetime import datetime
 from urllib.parse import urlparse, unquote
-# noinspection PyProtectedMember
-from lxml.etree import ElementBase as Element, _ElementTree as ElementTree
-
-from numpy import array as ndarray
 from PIL import Image
 
 import cv2
 
+if TYPE_CHECKING:
+    from ocrd import Workspace
+    from ocrd_models.ocrd_page_generateds import PcGtsType
+    # noinspection PyProtectedMember
+    from lxml.etree import ElementBase as Element, _ElementTree as ElementTree
+    from numpy import array as ndarray
+
 EventCallBack = Optional[Callable[[str, Any], None]]
 
 
@@ -60,11 +63,11 @@ def __init__(self, workspace: Optional[Workspace], emitter: Optional[EventCallBa
             os.chdir(self.workspace.directory)
 
     @classmethod
-    def create(cls, emitter: EventCallBack = None) -> 'Document':
+    def create(cls, emitter: EventCallBack = None) -> Document:
         return cls(None, emitter=emitter)
 
     @classmethod
-    def load(cls, mets_url: Union[Path, str] = None, emitter: EventCallBack = None) -> 'Document':
+    def load(cls, mets_url: Union[Path, str] = None, emitter: EventCallBack = None) -> Document:
         """
         Load a project from an url as a readonly view
 
@@ -80,7 +83,7 @@ def load(cls, mets_url: Union[Path, str] = None, emitter: EventCallBack = None)
         return doc
 
     @classmethod
-    def clone(cls, mets_url: Union[Path, str], emitter: EventCallBack = None, editable: bool = True) -> 'Document':
+    def clone(cls, mets_url: Union[Path, str], emitter: EventCallBack = None, editable: bool = True) -> Document:
         """
         Clones a project (mets.xml and all used files) to a temporary directory for editing
         """
@@ -159,7 +162,7 @@ def baseurl_mets(self) -> str:
         """
         return str(self.workspace.baseurl) + '/' + self.mets_filename if self.workspace else None
 
-    def path(self, other: Union[OcrdFile, Path, str]) -> Path:
+    def path(self, other: Union[OcrdFile, Path, str]) -> Optional[Path]:
         """
         Resolves other relative to current workspace
         """
@@ -222,7 +225,7 @@ def file_groups_and_mimetypes(self) -> List[Tuple[str, str]]:
 
         @return: List[Tuple[str,str]]
         """
-        distinct_groups: OrderedDict[Tuple[str, str], None] = OrderedDict()
+        distinct_groups: Dict[Tuple[str, str], None] = {}
         for el in self.xpath('mets:fileSec/mets:fileGrp[@USE]/mets:file[@MIMETYPE]'):
             distinct_groups[(el.getparent().get('USE'), el.get('MIMETYPE'))] = None
 
@@ -407,7 +410,7 @@ def delete_page(self, page_id: str) -> None:
 
     @check_editable
     def add_image(self, image: ndarray, page_id: str, file_id: str, file_group: str = 'OCR-D-IMG', dpi: int = 300,
-                  mimetype: str = 'image/png') -> 'OcrdFile':
+                  mimetype: str = 'image/png') -> OcrdFile:
         extension = MIME_TO_EXT[mimetype]
         retval, image_array = cv2.imencode(extension, image)
         image_bytes = add_dpi_to_png_buffer(image_array.tostring(), dpi)
@@ -442,6 +445,7 @@ def editable(self, editable: bool) -> None:
             if self._original_url:
                 self.workspace = self._clone_workspace(self._original_url)
             else:
+                # noinspection PyTypeChecker
                 self.workspace = Resolver().workspace_from_nothing(directory=None, mets_basename='mets.xml')
         else:
             self.workspace = Resolver().workspace_from_url(self.baseurl_mets)

ocrd_browser/model/page.py

@@ -64,6 +64,7 @@ def file(self) -> Optional[OcrdFile]:
         elif self.image_files:
             return next(iter(self.image_files))
         else:
+            # noinspection PyTypeChecker
             any_files = self.get_files(mimetype=None)
             if any_files:
                 return next(iter(any_files))

ocrd_browser/ui/dialogs.py

@@ -1,11 +1,12 @@
 from gi.repository import Gtk, GdkPixbuf
 from typing import Any
-from pkg_resources import resource_filename
+
+from ocrd_browser.util.gtk import resource_string
 from ocrd_browser import __version__
 from ocrd_browser.model import Document
 
 
-@Gtk.Template(filename=resource_filename(__name__, '../resources/about-dialog.ui'))
+@Gtk.Template(string=resource_string('about-dialog.ui'))
 class AboutDialog(Gtk.AboutDialog):
     __gtype_name__ = "AboutDialog"
 
@@ -16,7 +17,7 @@ def __init__(self, **kwargs: Any):
         self.set_version(__version__)
 
 
-@Gtk.Template(filename=resource_filename(__name__, '../resources/open-dialog.ui'))
+@Gtk.Template(string=resource_string('open-dialog.ui'))
 class OpenDialog(Gtk.FileChooserDialog):
     __gtype_name__ = "OpenDialog"
 
@@ -36,7 +37,7 @@ def __init__(self, **kwargs: Any):
         self.add_filter(filter_any)
 
 
-@Gtk.Template(filename=resource_filename(__name__, '../resources/save-dialog.ui'))
+@Gtk.Template(string=resource_string('save-dialog.ui'))
 class SaveDialog(Gtk.FileChooserDialog):
     __gtype_name__ = "SaveDialog"
 

ocrd_browser/ui/page_browser.py

@@ -2,12 +2,12 @@
 
 from typing import List, Callable, Optional, Any, cast
 
-from pkg_resources import resource_filename
+from ocrd_browser.util.gtk import resource_string
 from ocrd_browser.model import Document
 from .page_store import PageListStore, ChangeList
 
 
-@Gtk.Template(filename=resource_filename(__name__, '../resources/page-list.ui'))
+@Gtk.Template(string=resource_string('page-list.ui'))
 class PagePreviewList(Gtk.IconView):
     __gtype_name__ = "PagePreviewList"
 

ocrd_browser/ui/window.py

@@ -3,22 +3,20 @@
 
 from ocrd_browser.model import Document
 from ocrd_browser.view import ViewRegistry, ViewPage
-from ocrd_browser.util.gtk import ActionRegistry
+from ocrd_browser.util.gtk import ActionRegistry, resource_string
 from .dialogs import SaveDialog, SaveChangesDialog
 from .page_browser import PagePreviewList
-from pkg_resources import resource_filename
 from typing import List, cast, Any, Optional
 
 from ..view.manager import ViewManager
 
 
-@Gtk.Template(filename=resource_filename(__name__, '../resources/main-window.ui'))
+@Gtk.Template(string=resource_string('main-window.ui'))
 class MainWindow(Gtk.ApplicationWindow):
     __gtype_name__ = "MainWindow"
 
     header_bar: Gtk.HeaderBar = Gtk.Template.Child()
     page_list_scroller: Gtk.ScrolledWindow = Gtk.Template.Child()
-    panes: Gtk.Paned = Gtk.Template.Child()
     current_page_label: Gtk.Label = Gtk.Template.Child()
     view_container: Gtk.Box = Gtk.Template.Child()
     view_menu_box: Gtk.Box = Gtk.Template.Child()

ocrd_browser/util/config.py

@@ -1,5 +1,5 @@
+from __future__ import annotations
 import os
-from collections import OrderedDict
 from configparser import ConfigParser
 import shlex
 from typing import List, Optional, MutableMapping
@@ -13,7 +13,7 @@
 
 class _SubSettings:
     @classmethod
-    def from_section(cls, _name: str, section: ConfigDict) -> '_SubSettings':
+    def from_section(cls, _name: str, section: ConfigDict) -> _SubSettings:
         raise NotImplementedError('please override from_section')
 
     def validate(self) -> None:
@@ -28,7 +28,7 @@ def __init__(self, preferred_images: List[str]):
         self.preferred_images = preferred_images
 
     @classmethod
-    def from_section(cls, _name: str, section: ConfigDict) -> '_FileGroups':
+    def from_section(cls, _name: str, section: ConfigDict) -> _FileGroups:
         preferred_images = section.get('preferredImages', 'OCR-D-IMG, OCR-D-IMG.*')
         return cls(
             [grp.strip() for grp in preferred_images.split(',')]
@@ -50,7 +50,7 @@ def validate(self) -> None:
             raise ValueError('Could not locate executable "{}"'.format(executable))
 
     @classmethod
-    def from_section(cls, name: str, section: ConfigDict) -> '_Tool':
+    def from_section(cls, name: str, section: ConfigDict) -> _Tool:
         return cls(
             name[len(cls.PREFIX):],
             section['commandline'],
@@ -66,7 +66,7 @@ def __init__(self, config: ConfigParser, validate: bool = True):
         self.file_groups = _FileGroups.from_section('FileGroups', config['FileGroups'] if 'FileGroups' in config else {'': ''})
         if validate:
             self.file_groups.validate()
-        self.tools = OrderedDict()
+        self.tools = {}
         for name, section in config.items():
             if name.startswith(_Tool.PREFIX):
                 tool = _Tool.from_section(name, section)
@@ -78,13 +78,13 @@ def __repr__(self) -> str:
         return '{}({})'.format(self.__class__.__name__, repr(vars(self)))
 
     @classmethod
-    def get(cls) -> 'Settings':
+    def get(cls) -> Settings:
         if cls._settings is None:
             cls._settings = Settings.build_default()
         return cls._settings
 
     @classmethod
-    def build_default(cls, config_dirs: Optional[List[str]] = None, validate: bool = True) -> 'Settings':
+    def build_default(cls, config_dirs: Optional[List[str]] = None, validate: bool = True) -> Settings:
         if config_dirs is None:
             config_dirs = GLib.get_system_config_dirs() + [GLib.get_user_config_dir()]
             try:
@@ -97,7 +97,7 @@ def build_default(cls, config_dirs: Optional[List[str]] = None, validate: bool =
         return cls.build_from_files(config_files, validate)
 
     @classmethod
-    def build_from_files(cls, files: List[str], validate: bool = True) -> 'Settings':
+    def build_from_files(cls, files: List[str], validate: bool = True) -> Settings:
         log = getLogger('ocrd_browser.util.config._Settings.build_from_files')
         config = ConfigParser()
         setattr(config, 'optionxform', lambda option: option)

ocrd_browser/util/gtk.py

@@ -1,7 +1,14 @@
+from __future__ import annotations
 from gi.repository import Gio, GLib, Gtk
 
 from typing import Callable, Dict, Optional, Set, Any
 
+try:
+    from importlib.resources import read_text
+except ModuleNotFoundError:
+    from importlib_resources import read_text  # type: ignore
+
+
 ActionCallback = Optional[Callable[[Gio.SimpleAction, Any], None]]
 
 
@@ -70,14 +77,14 @@ class WhenIdle:
 
     Usage: see WhenIdle.call
     """
-    _instance: 'WhenIdle' = None
+    _instance: WhenIdle = None
 
     def __init__(self, runner_callback: Callable):  # type: ignore[type-arg]
         self._runner_callback = runner_callback
         self._callbacks: Set[Callback] = set()
 
     @classmethod
-    def instance(cls) -> 'WhenIdle':
+    def instance(cls) -> WhenIdle:
         if cls._instance is None:
             cls._instance = cls(GLib.idle_add)
         return cls._instance
@@ -102,3 +109,7 @@ def _run(self) -> None:
             callback()
             self._callbacks.remove(callback)
             self._runner_callback(self._run)
+
+
+def resource_string(resource: str, package: str = 'ocrd_browser.resources') -> str:
+    return read_text(package, resource)

ocrd_browser/view/__init__.py

@@ -1,12 +1,12 @@
 from .base import View
-from .registry import ViewRegistry
 from .html import ViewHtml
 from .images import ViewImages
 from .text import ViewText
 from .xml import ViewXml
 from .empty import ViewEmpty
 from .diff import ViewDiff
 from .page import ViewPage
+from .registry import ViewRegistry
 
 
 __all__ = ['View', 'ViewRegistry', 'ViewImages', 'ViewText', 'ViewXml', 'ViewHtml', 'ViewEmpty', 'ViewDiff', 'ViewPage']