v2.4.0

.gitmodules

@@ -1,6 +1,3 @@
-[submodule "src/submodules/pyedt"]
-	path = src/submodules/pyedt
-	url = git@github.com:ltracegeo/pyedt.git
 [submodule "src/submodules/porespy"]
 	path = src/submodules/porespy
 	url = git@github.com:ltracegeo/porespy.git

pyproject.toml

@@ -55,17 +55,26 @@ authorized_licenses = [
     "BSD-3-clause",
     "BSD 3-Clause",
     "BSD License",
+    "CC-BY",
     "GNU Library or Lesser General Public License (LGPL)",
+    "GNU Lesser General Public License v2 or later (LGPLv2+)",
+    "ISC License (ISCL)",
     "LGPL",
     "LGPL-3.0",
     "MIT",
+    "MIT License",
     "Mozilla Public License 2.0 (MPL 2.0)",
     "new BSD",
     "OSI Approved",
     "Python Software Foundation",
     "Historical Permission Notice and Disclaimer (HPND)",
-    "Other/Proprietary"
 ]
 unauthorized_licenses = [
     "GNU General Public License v3 (GPLv3)"
 ]
+[tool.liccheck.authorized_packages]
+# The packages below are listed as "Other/Proprietary"
+# They use Intel licenses that allow distribution
+mkl = ">=2022.2.1"
+tbb = ">=2021.12.0"
+intel-openmp = ">=2022.2.1"

src/ltrace/ltrace/algorithms/detect_cups.py

@@ -163,9 +163,18 @@ def detect_rock_cylinder(array):
 PADDING = 15
 
 
+def find_suitable_null_value(array):
+    if np.issubdtype(array.dtype, np.floating):
+        return int(array.min()) - 1
+    if np.issubdtype(array.dtype, np.signedinteger):
+        return array.min() - 1
+    if np.issubdtype(array.dtype, np.unsignedinteger):
+        return 0
+
+
 def crop_cylinder(array, cylinder):
     x, y, r, z_min, z_max = cylinder
-    min_ = array.min()
+    null_value = np.array(find_suitable_null_value(array)).astype(array.dtype)
     x_min, x_max = round(x - r), round(x + r)
     y_min, y_max = round(y - r), round(y + r)
     z_min, z_max = round(z_min), round(z_max)
@@ -178,10 +187,10 @@ def crop_cylinder(array, cylinder):
     dist = np.sqrt((xx - r) ** 2 + (yy - r) ** 2)
     mask = dist <= r
     mask = mask[np.newaxis, ...]
-    array = (array * mask) + (~mask * min_)
+    array = (array * mask) + (~mask * null_value)
 
-    array = np.pad(array, ((0, 0), (PADDING, PADDING), (PADDING, PADDING)), mode="constant", constant_values=min_)
-    return array
+    array = np.pad(array, ((0, 0), (PADDING, PADDING), (PADDING, PADDING)), mode="constant", constant_values=null_value)
+    return array, null_value
 
 
 def get_origin_offset(cylinder):
@@ -314,10 +323,10 @@ def full_detect(array, callback=lambda *args: None):
     logging.debug(f"x, y, r, z0, z1 = {cylinder}")
 
     callback(50, "Cropping cylinder")
-    rock = crop_cylinder(array, cylinder)
+    rock, null_value = crop_cylinder(array, cylinder)
     callback(70, "Detecting cups")
     cup_left, cup_right = isolated_cups_slice(array, cylinder)
     refs = detect_cups_from_sides(cup_left, cup_right)
     if refs:
         logging.debug(f"(Q - T) / (A - T) = {quartz_ratio(refs)}")
-    return rock, refs, cylinder
+    return rock, null_value, refs, cylinder

src/ltrace/ltrace/algorithms/supervised/mrcnn/utils.py

@@ -849,11 +849,8 @@ def download_trained_weights(coco_model_path, verbose=1):
     """
     if verbose > 0:
         print("Downloading pretrained model to " + coco_model_path + " ...")
-    with open(coco_model_path, "wb") as out:
-        resp = urllib.request.urlopen(COCO_MODEL_URL)
+    with urllib.request.urlopen(COCO_MODEL_URL) as resp, open(coco_model_path, "wb") as out:
         shutil.copyfileobj(resp, out)
-        resp.close()
-
     if verbose > 0:
         print("... done downloading pretrained model!")
 

src/ltrace/ltrace/assets_utils.py

@@ -2,57 +2,65 @@
 import json
 
 
+ROOT = Path(__file__).resolve().absolute().parent / "assets"
+PUBLIC = ROOT / "public"
+PRIVATE = ROOT / "private"
+
+
 def get_asset(asset_name):
-    """Returns the asset's absolute Path given a relative path from this dir."""
+    """Returns the asset's absolute Path given a relative path from either public or private dir.
+    Do not include 'public' or 'private' in the path.
+    """
     if isinstance(asset_name, str):
         asset_name = Path(asset_name)
 
-    absolute_path = Path(__file__).resolve().absolute().parent / "assets" / asset_name
-    if not absolute_path.exists():
-        raise RuntimeError("Invalid asset: {}".format(absolute_path))
-
-    return absolute_path
+    private_path = PRIVATE / asset_name
+    if private_path.exists():
+        return private_path
 
+    public_path = PUBLIC / asset_name
+    if public_path.exists():
+        return public_path
 
-def get_trained_model(model_name):
-    return get_asset(Path("trained_models") / Path(model_name))
+    return None
 
 
-def get_trained_models(environment=None):
-    models_path = Path(__file__).resolve().absolute().parent / "assets" / "trained_models"
-
-    if environment:
-        models_path = models_path / environment
+def get_trained_models(environment):
+    models_path = Path(__file__).resolve().absolute().parent / "assets" / "models"
 
     extensions = [".h5", ".pth"]
-    models = [file for file in models_path.glob("**/*") if file.suffix in extensions]
+    models = []
+    for root_path in (PUBLIC, PRIVATE):
+        models_path = root_path / environment
+        models += [file for file in models_path.glob("**/*") if file.suffix in extensions]
     return models
 
 
 def get_trained_models_with_metadata(environment):
-    models_path = Path(__file__).resolve().absolute().parent / "assets" / "trained_models"
-    models_path = models_path / environment
-
     models = []
-    for subdir in models_path.iterdir():
-        if not subdir.is_dir():
+    for root_path in (PUBLIC, PRIVATE):
+        models_path = root_path / environment
+        if not models_path.is_dir():
             continue
-        base = subdir.name
-        pth = subdir / f"{base}.pth"
+        for subdir in models_path.iterdir():
+            if not subdir.is_dir():
+                continue
+            base = subdir.name
+            pth = subdir / f"model.pth"
 
-        assert pth.exists(), f"Directory {base} exists but file {pth} does not exist"
+            assert pth.exists(), f"Directory {base} exists but file {pth} does not exist"
 
-        models.append(subdir)
+            models.append(subdir)
     return models
 
 
 def get_metadata(model_dir):
     model_dir = Path(model_dir)
-    with open(model_dir / f"{model_dir.name}.json") as f:
+    with open(model_dir / f"meta.json") as f:
         metadata = json.load(f)
     return metadata
 
 
 def get_pth(model_dir):
     model_dir = Path(model_dir)
-    return model_dir / f"{model_dir.name}.pth"
+    return model_dir / f"model.pth"

src/ltrace/ltrace/constants.py

@@ -17,3 +17,14 @@ class ImageLogConst:
 
     # The default layout ID value for the ImageLogData module.
     DEFAULT_LAYOUT_ID_START_VALUE = 15000
+
+
+class ImageLogInpaintConst:
+    """
+    Constants for ImageLogInpaint module
+    """
+
+    SEGMENT_ID = "Segment_1"
+    TEMP_SEGMENTATION_NAME = "Inpaint_Segmentation_ImageLog"
+    TEMP_LABEL_MAP_NAME = "Inpaint_Mask_ImageLog"
+    TEMP_VOLUME_NAME = "Inpaint_Volume_ImageLog"

src/ltrace/ltrace/file_utils.py

@@ -85,9 +85,6 @@ def detect_csv_file_delimiter(filepath, encoding, whitelist=[",", ";"]):
         # read in non binary with encoding, to detect delimiter
         with open(filepath, "r", encoding=encoding) as file:
             data = file.readlines(50)
-            if not data:
-                return None
-
             counters = {}
             for delimiter in whitelist:
                 count = number_of_delimiters_per_line(data, delimiter)

src/ltrace/ltrace/image/core_box/core_boxes_image_file.py

@@ -1,5 +1,5 @@
 from collections import namedtuple
-from ltrace.assets_utils import get_trained_model
+from ltrace.assets_utils import get_asset
 from ltrace.image.core_box.core_box import CoreBox
 from ltrace.slicer.helpers import concatenateImageArrayVertically, resizeRgbArray
 from ltrace.units import safe_atof
@@ -306,7 +306,7 @@ def __extract_cores_info(self, img):
             from ltrace.image.segmentation import TF_RGBImageArrayBinarySegmenter
 
             CoreBoxesImageFile.BinarySegmenterModel = TF_RGBImageArrayBinarySegmenter(
-                get_trained_model("unet-binary-segop.h5"), gpuEnabled=self.__gpuEnabled
+                get_asset("unet-binary-segop.h5"), gpuEnabled=self.__gpuEnabled
             )
 
         for core_cut in self.__split_cores(img, number_of_cores):