Have many sources use lazy imports

CHANGELOG.md

@@ -6,6 +6,8 @@
 
 - Improve plottable data endpoint to better fetch adjacent items and annotations ([#1573](../../pull/1573), [#1574](../../pull/1574))), [#1575](../../pull/1575)))
 - Support Girder flat-mount paths ([#1576](../../pull/1576))
+- Lazily import some modules to speed up large_image import speed ([#1577](../../pull/1577))
+- Create or check large images for each item in a folder ([#1572](../../pull/1572))
 
 ## 1.29.2
 

sources/gdal/large_image_source_gdal/__init__.py

@@ -20,28 +20,11 @@
 import struct
 import tempfile
 import threading
+from importlib.metadata import PackageNotFoundError
+from importlib.metadata import version as _importlib_version
 
 import numpy as np
 import PIL.Image
-from osgeo import gdal, gdal_array, gdalconst, osr
-
-try:
-    gdal.UseExceptions()
-except Exception:
-    pass
-
-# isort: off
-
-# pyproj stopped supporting older pythons, so on those versions its database is
-# aging; as such, if on those older versions of python if it is imported before
-# gdal, there can be a database version conflict; importing after gdal avoids
-# this.
-import pyproj
-
-# isort: on
-
-from importlib.metadata import PackageNotFoundError
-from importlib.metadata import version as _importlib_version
 
 from large_image.cache_util import LruCacheMetaclass, methodcache
 from large_image.constants import (TILE_FORMAT_IMAGE, TILE_FORMAT_NUMPY,
@@ -62,6 +45,42 @@
     # package is not installed
     pass
 
+gdal = None
+gdal_array = None
+gdalconst = None
+osr = None
+pyproj = None
+
+
+def _lazyImport():
+    """
+    Import the gdal module.  This is done when needed rather than in the
+    module initialization because it is slow.
+    """
+    global gdal, gdal_array, gdalconst, osr, pyproj
+
+    if gdal is None:
+        try:
+            from osgeo import gdal, gdal_array, gdalconst, osr
+
+            try:
+                gdal.UseExceptions()
+            except Exception:
+                pass
+
+            # isort: off
+
+            # pyproj stopped supporting older pythons, so on those versions its
+            # database is aging; as such, if on those older versions of python
+            # if it is imported before gdal, there can be a database version
+            # conflict; importing after gdal avoids this.
+            import pyproj
+
+            # isort: on
+        except ImportError:
+            msg = 'gdal module not found.'
+            raise TileSourceError(msg)
+
 
 class GDALFileTileSource(GDALBaseFileTileSource, metaclass=LruCacheMetaclass):
     """
@@ -91,6 +110,8 @@ def __init__(self, path, projection=None, unitsPerPixel=None, **kwargs):
             specify unitsPerPixel.
         """
         super().__init__(path, **kwargs)
+        _lazyImport()
+
         self.addKnownExtensions()
         self._bounds = {}
         self._largeImagePath = self._getLargeImagePath()
@@ -300,6 +321,8 @@ def _proj4Proj(proj):
         :returns: a proj4 projection object.  None if the specified projection
             cannot be created.
         """
+        _lazyImport()
+
         if isinstance(proj, bytes):
             proj = proj.decode()
         if not isinstance(proj, str):
@@ -940,6 +963,8 @@ def isGeospatial(path):
         :param path: The path to the file
         :returns: True if geospatial.
         """
+        _lazyImport()
+
         try:
             ds = gdal.Open(str(path), gdalconst.GA_ReadOnly)
         except Exception:
@@ -958,6 +983,8 @@ def isGeospatial(path):
     @classmethod
     def addKnownExtensions(cls):
         if not hasattr(cls, '_addedExtensions'):
+            _lazyImport()
+
             cls._addedExtensions = True
             cls.extensions = cls.extensions.copy()
             cls.mimeTypes = cls.mimeTypes.copy()

sources/multi/large_image_source_multi/__init__.py

@@ -10,7 +10,6 @@
 from importlib.metadata import version as _importlib_version
 from pathlib import Path
 
-import jsonschema
 import numpy as np
 import yaml
 
@@ -27,6 +26,26 @@
     # package is not installed
     pass
 
+jsonschema = None
+_validator = None
+
+
+def _lazyImport():
+    """
+    Import the jsonschema module.  This is done when needed rather than in the
+    module initialization because it is slow.
+    """
+    global jsonschema, _validator
+
+    if jsonschema is None:
+        try:
+            import jsonschema
+
+            _validator = jsonschema.Draft6Validator(MultiSourceSchema)
+        except ImportError:
+            msg = 'jsonschema module not found.'
+            raise TileSourceError(msg)
+
 
 SourceEntrySchema = {
     'type': 'object',
@@ -387,8 +406,6 @@ class MultiFileTileSource(FileTileSource, metaclass=LruCacheMetaclass):
     _defaultTileSize = 256
     _maxOpenHandles = 6
 
-    _validator = jsonschema.Draft6Validator(MultiSourceSchema)
-
     def __init__(self, path, **kwargs):
         """
         Initialize the tile class.  See the base class for other available
@@ -398,6 +415,8 @@ def __init__(self, path, **kwargs):
         """
         super().__init__(path, **kwargs)
 
+        _lazyImport()
+        self._validator = _validator
         self._largeImagePath = self._getLargeImagePath()
         self._lastOpenSourceLock = threading.RLock()
         # 'c' must be first as channels are special because they can have names

sources/openjpeg/large_image_source_openjpeg/__init__.py

@@ -25,7 +25,6 @@
 from importlib.metadata import version as _importlib_version
 from xml.etree import ElementTree
 
-import glymur
 import PIL.Image
 
 import large_image
@@ -41,6 +40,23 @@
     # package is not installed
     pass
 
+glymur = None
+
+
+def _lazyImport():
+    """
+    Import the glymur module.  This is done when needed rather than in the module
+    initialization because it is slow.
+    """
+    global glymur
+
+    if glymur is None:
+        try:
+            import glymur
+        except ImportError:
+            msg = 'glymur module not found.'
+            raise TileSourceError(msg)
+
 
 warnings.filterwarnings('ignore', category=UserWarning, module='glymur')
 
@@ -88,6 +104,7 @@ def __init__(self, path, **kwargs):
         """
         super().__init__(path, **kwargs)
 
+        _lazyImport()
         self._largeImagePath = str(self._getLargeImagePath())
         self._pixelInfo = {}
         try: