Moved LIF metadata function to another module to avoid unnecessary client-side dependency installation (#303)

Author: tieneupin

pyproject.toml

@@ -32,7 +32,6 @@ classifiers = [
 dependencies = [
     "defusedxml", # For safely parsing XML files
     "pydantic<2", # Pip hops between installing v2.7 or v1.10 depending on which of the additional dependencies are requested
-    "readlif", # Specific to cryo-CLEM workflow
     "requests",
     "rich",
     "werkzeug",
@@ -65,6 +64,7 @@ server = [
     "pillow",
     "prometheus_client",
     "python-jose[cryptography]",
+    "readlif", # Specific to cryo-CLEM workflow
     "sqlmodel",
     "stomp-py<=8.1.0", # 8.1.1 (released 2024-04-06) doesn't work with our project
     "tifffile", # Specific to cryo-CLEM workflow

src/murfey/client/contexts/clem.py

@@ -17,7 +17,7 @@
 from murfey.client.context import Context
 from murfey.client.instance_environment import MurfeyInstanceEnvironment
 from murfey.util import capture_post, get_machine_config, sanitise
-from murfey.util.clem import xml
+from murfey.util.clem.xml import get_image_elements
 
 # Create logger object
 logger = logging.getLogger("murfey.client.contexts.clem")
@@ -59,13 +59,13 @@ def __init__(self, acquisition_software: str, basepath: Path):
         super().__init__("CLEM", acquisition_software)
         self._basepath = basepath
         # CLEM contexts for "auto-save" acquisition mode
-        self._tiff_series: Dict[str, List[str]] = {}  # Series name : List of TIFF paths
-        self._tiff_timestamps: Dict[str, List[float]] = {}  # Series name: Timestamps
-        self._tiff_sizes: Dict[str, List[int]] = {}  # Series name: File sizes
-        self._series_metadata: Dict[str, str] = {}  # Series name: Metadata file path
-        self._metadata_timestamp: Dict[str, float] = {}  # Series name: Timestamp
-        self._metadata_size: Dict[str, int] = {}  # Series name: File size
-        self._files_in_series: Dict[str, int] = {}  # Series name : Total TIFFs
+        self._tiff_series: Dict[str, List[str]] = {}  # {Series name : TIFF path list}
+        self._tiff_timestamps: Dict[str, List[float]] = {}  # {Series name : Timestamps}
+        self._tiff_sizes: Dict[str, List[int]] = {}  # {Series name : File sizes}
+        self._series_metadata: Dict[str, str] = {}  # {Series name : Metadata file path}
+        self._metadata_timestamp: Dict[str, float] = {}  # {Series name : Timestamp}
+        self._metadata_size: Dict[str, int] = {}  # {Series name : File size}
+        self._files_in_series: Dict[str, int] = {}  # {Series name : Total TIFFs}
 
     def post_transfer(
         self,
@@ -144,7 +144,7 @@ def post_transfer(
 
                 # Extract metadata to get the expected size of the series
                 metadata = parse(file_path).getroot()
-                metadata = xml.get_image_elements(metadata)[0]
+                metadata = get_image_elements(metadata)[0]
 
                 # Get channel and dimension information
                 channels = metadata.findall(

src/murfey/util/clem/lif.py

@@ -8,19 +8,43 @@
 import logging
 import multiprocessing as mp
 from pathlib import Path
+from typing import Optional
 from xml.etree import ElementTree as ET
 
 import numpy as np
 from readlif.reader import LifFile
 
 from murfey.util import sanitise
 from murfey.util.clem.images import process_img_stk, write_to_tiff
-from murfey.util.clem.xml import get_image_elements, get_lif_xml_metadata
+from murfey.util.clem.xml import get_image_elements
 
 # Create logger object to output messages with
 logger = logging.getLogger("murfey.util.clem.lif")
 
 
+def get_lif_xml_metadata(
+    file: LifFile,
+    save_xml: Optional[Path] = None,
+) -> ET.Element:
+    """
+    Extracts and returns the metadata from the LIF file as a formatted XML Element.
+    It can be optionally saved as an XML file to the specified file path.
+    """
+
+    # Use readlif function to get XML metadata
+    xml_root: ET.Element = file.xml_root  # This one for navigating
+    xml_tree = ET.ElementTree(xml_root)  # This one for saving
+
+    # Skip saving the metadata if save_xml not provided
+    if save_xml:
+        xml_file = str(save_xml)  # Convert Path to string
+        ET.indent(xml_tree, "  ")  # Format with proper indentation
+        xml_tree.write(xml_file, encoding="utf-8")  # Save
+        logger.info(f"File metadata saved to {sanitise(xml_file)}")
+
+    return xml_root
+
+
 def process_lif_file(
     file: Path,
     scene_num: int,

src/murfey/util/clem/xml.py

@@ -8,41 +8,13 @@
 """
 
 import logging
-from pathlib import Path
-from typing import Generator, List, Optional
+from typing import Generator, List
 from xml.etree import ElementTree as ET
 
-from readlif.reader import LifFile
-
-from murfey.util import sanitise
-
 # Create logger object to output messages with
 logger = logging.getLogger("murfey.util.clem.xml")
 
 
-def get_lif_xml_metadata(
-    file: LifFile,
-    save_xml: Optional[Path] = None,
-) -> ET.Element:
-    """
-    Extracts and returns the metadata from the LIF file as a formatted XML Element.
-    It can be optionally saved as an XML file to the specified file path.
-    """
-
-    # Use readlif function to get XML metadata
-    xml_root: ET.Element = file.xml_root  # This one for navigating
-    xml_tree = ET.ElementTree(xml_root)  # This one for saving
-
-    # Skip saving the metadata if save_xml not provided
-    if save_xml:
-        xml_file = str(save_xml)  # Convert Path to string
-        ET.indent(xml_tree, "  ")  # Format with proper indentation
-        xml_tree.write(xml_file, encoding="utf-8")  # Save
-        logger.info(f"File metadata saved to {sanitise(xml_file)}")
-
-    return xml_root
-
-
 def get_image_elements(root: ET.Element) -> List[ET.Element]:
     """
     Searches the XML metadata recursively to find the nodes tagged as "Element" that