Assign parameters per context rather than per environment (#607)

* Convert find_extension to return a bool
* Migrate parameters from environment to context
* Removed unnecessary parameters
* Fix for eer fractionation from frontend
* Ensure environments initialise with the right attributes
* Gain is better as a string in the client
* Remove dependencies for old parameters

Author: stephen-riggs

src/murfey/client/analyser.py

@@ -21,29 +21,12 @@
 from murfey.client.contexts.tomo import TomographyContext
 from murfey.client.instance_environment import MurfeyInstanceEnvironment
 from murfey.client.rsync import RSyncerUpdate, TransferResult
-from murfey.client.tui.forms import FormDependency
 from murfey.util.client import Observer, get_machine_config_client
 from murfey.util.mdoc import get_block
 from murfey.util.models import ProcessingParametersSPA, ProcessingParametersTomo
 
 logger = logging.getLogger("murfey.client.analyser")
 
-spa_form_dependencies: dict = {
-    "use_cryolo": FormDependency(
-        dependencies={"estimate_particle_diameter": False}, trigger_value=False
-    ),
-    "estimate_particle_diameter": FormDependency(
-        dependencies={
-            "use_cryolo": True,
-            "boxsize": "None",
-            "small_boxsize": "None",
-            "mask_diameter": "None",
-            "particle_diameter": "None",
-        },
-        trigger_value=True,
-    ),
-}
-
 
 class Analyser(Observer):
     def __init__(
@@ -85,7 +68,7 @@ def __init__(
     def __repr__(self) -> str:
         return f"<Analyser ({self._basepath})>"
 
-    def _find_extension(self, file_path: Path):
+    def _find_extension(self, file_path: Path) -> bool:
         """
         Identifies the file extension and stores that information in the class.
         """
@@ -101,32 +84,30 @@ def _find_extension(self, file_path: Path):
             .get(file_path.suffix)
         ):
             if not any(r in file_path.name for r in required_substrings):
-                return []
+                return False
 
-        # Checks for MRC, TIFF, TIF, and EER files if no extension has been defined
-        if (
-            file_path.suffix in (".mrc", ".tiff", ".tif", ".eer")
-            and not self._extension
-        ):
-            logger.info(f"File extension determined: {file_path.suffix}")
-            self._extension = file_path.suffix
-        # Check for TIFF, TIF, or EER if the file's already been assigned an extension
-        elif (
-            file_path.suffix in (".tiff", ".tif", ".eer")
-            and self._extension != file_path.suffix
-        ):
-            logger.info(f"File extension re-evaluated: {file_path.suffix}")
-            self._extension = file_path.suffix
+        # Checks for MRC, TIFF, TIF, and EER files
+        if file_path.suffix in (".mrc", ".tiff", ".tif", ".eer"):
+            if not self._extension:
+                logger.info(f"File extension determined: {file_path.suffix}")
+                self._extension = file_path.suffix
+            elif self._extension != file_path.suffix:
+                logger.info(f"File extension re-evaluated: {file_path.suffix}")
+                self._extension = file_path.suffix
+            return True
         # If we see an .mdoc file first, use that to determine the file extensions
         elif file_path.suffix == ".mdoc":
             with open(file_path, "r") as md:
                 md.seek(0)
                 mdoc_data_block = get_block(md)
             if subframe_path := mdoc_data_block.get("SubFramePath"):
                 self._extension = Path(subframe_path).suffix
+                return True
         # Check for LIF files separately
         elif file_path.suffix == ".lif":
             self._extension = file_path.suffix
+            return True
+        return False
 
     def _find_context(self, file_path: Path) -> bool:
         """
@@ -269,7 +250,10 @@ def _analyse(self):
                     elif transferred_file.suffix == ".mdoc":
                         mdoc_for_reading = transferred_file
                 if not self._context:
-                    self._find_extension(transferred_file)
+                    valid_extension = self._find_extension(transferred_file)
+                    if not valid_extension:
+                        logger.error(f"No extension found for {transferred_file}")
+                        break
                     found = self._find_context(transferred_file)
                     if not found:
                         logger.debug(
@@ -325,13 +309,6 @@ def _analyse(self):
                                 self.notify(
                                     {
                                         "form": dc_metadata,
-                                        "dependencies": (
-                                            spa_form_dependencies
-                                            if isinstance(
-                                                self._context, SPAModularContext
-                                            )
-                                            else {}
-                                        ),
                                     }
                                 )
 
@@ -344,7 +321,10 @@ def _analyse(self):
 
                 # Handle files with tomography and SPA context differently
                 elif not self._extension or self._unseen_xml:
-                    self._find_extension(transferred_file)
+                    valid_extension = self._find_extension(transferred_file)
+                    if not valid_extension:
+                        logger.error(f"No extension found for {transferred_file}")
+                        break
                     if self._extension:
                         logger.info(
                             f"Extension found successfully for {transferred_file}"
@@ -384,13 +364,6 @@ def _analyse(self):
                                 self.notify(
                                     {
                                         "form": dc_metadata,
-                                        "dependencies": (
-                                            spa_form_dependencies
-                                            if isinstance(
-                                                self._context, SPAModularContext
-                                            )
-                                            else {}
-                                        ),
                                     }
                                 )
                 elif isinstance(

src/murfey/client/context.py

@@ -39,6 +39,7 @@ class Context:
     def __init__(self, name: str, acquisition_software: str):
         self._acquisition_software = acquisition_software
         self.name = name
+        self.data_collection_parameters: dict = {}
 
     def post_transfer(
         self,

src/murfey/client/contexts/spa.py

@@ -1,7 +1,6 @@
 from __future__ import annotations
 
 import logging
-from datetime import datetime
 from itertools import count
 from pathlib import Path
 from typing import Any, Dict, List, Optional, OrderedDict, Tuple
@@ -109,27 +108,9 @@ class SPAModularContext(Context):
             "Dose Per Frame [e- / Angstrom^2 / frame] (after EER grouping if relevant)",
             default=1,
         ),
-        ProcessingParameter(
-            "estimate_particle_diameter",
-            "Use crYOLO to Estimate Particle Diameter",
-            default=True,
-        ),
-        ProcessingParameter(
-            "particle_diameter", "Particle Diameter (Angstroms)", default=None
-        ),
-        ProcessingParameter("use_cryolo", "Use crYOLO Autopicking", default=True),
         ProcessingParameter("symmetry", "Symmetry Group", default="C1"),
         ProcessingParameter("eer_fractionation", "EER Fractionation", default=20),
-        ProcessingParameter(
-            "mask_diameter", "Mask Diameter (2D classification)", default=190
-        ),
-        ProcessingParameter("boxsize", "Box Size", default=256),
-        ProcessingParameter("downscale", "Downscale Extracted Particles", default=True),
-        ProcessingParameter(
-            "small_boxsize", "Downscaled Extracted Particle Size (pixels)", default=128
-        ),
         ProcessingParameter("gain_ref", "Gain Reference"),
-        ProcessingParameter("gain_ref_superres", "Unbinned Gain Reference"),
     ]
     metadata_params = [
         ProcessingParameter("voltage", "Voltage"),
@@ -164,7 +145,6 @@ def gather_metadata(
                 return OrderedDict({})
             data = xmltodict.parse(for_parsing)
         magnification = 0
-        num_fractions = 1
         metadata: OrderedDict = OrderedDict({})
         metadata["experiment_type"] = "SPA"
         if data.get("Acquisition"):
@@ -217,14 +197,6 @@ def gather_metadata(
                 )  # convert e / m^2 to e / A^2
             except ValueError:
                 metadata["total_exposed_dose"] = 1
-            try:
-                num_fractions = int(
-                    data["MicroscopeImage"]["microscopeData"]["acquisition"]["camera"][
-                        "CameraSpecificInput"
-                    ]["a:KeyValueOfstringanyType"][2]["a:Value"]["b:NumberOffractions"]
-                )
-            except (KeyError, IndexError):
-                pass
             c2_index = 3
             for i, el in enumerate(
                 data["MicroscopeImage"]["CustomData"]["a:KeyValueOfstringanyType"]
@@ -293,85 +265,13 @@ def gather_metadata(
         metadata["image_size_x"] = str(int(metadata["image_size_x"]) * binning_factor)
         metadata["image_size_y"] = str(int(metadata["image_size_y"]) * binning_factor)
         metadata["motion_corr_binning"] = 1 if binning_factor_xml == 2 else 2
-        if environment:
-            metadata["gain_ref"] = (
-                environment.data_collection_parameters.get("gain_ref")
-                if environment
-                and environment.data_collection_parameters.get("gain_ref")
-                not in (None, "None")
-                else f"{datetime.now().year}/{environment.visit}/processing/gain.mrc"
-            )
-            metadata["gain_ref_superres"] = (
-                environment.data_collection_parameters.get("gain_ref_superres")
-                if environment
-                and environment.data_collection_parameters.get("gain_ref_superres")
-                not in (None, "None")
-                else f"{datetime.now().year}/{environment.visit}/processing/gain_superres.mrc"
-            )
-        else:
-            metadata["gain_ref"] = None
-            metadata["gain_ref_superres"] = None
-        if metadata.get("total_exposed_dose"):
-            metadata["dose_per_frame"] = (
-                environment.data_collection_parameters.get("dose_per_frame")
-                if environment
-                and environment.data_collection_parameters.get("dose_per_frame")
-                not in (None, "None")
-                else round(metadata["total_exposed_dose"] / num_fractions, 3)
-            )
-        else:
-            metadata["dose_per_frame"] = (
-                environment.data_collection_parameters.get("dose_per_frame")
-                if environment
-                else None
-            )
-
-        metadata["use_cryolo"] = (
-            environment.data_collection_parameters.get("use_cryolo")
-            if environment
-            else None
-        ) or True
-        metadata["symmetry"] = (
-            environment.data_collection_parameters.get("symmetry")
-            if environment
-            else None
-        ) or "C1"
-        metadata["mask_diameter"] = (
-            environment.data_collection_parameters.get("mask_diameter")
-            if environment
-            else None
-        ) or 190
-        metadata["boxsize"] = (
-            environment.data_collection_parameters.get("boxsize")
-            if environment
-            else None
-        ) or 256
-        metadata["downscale"] = (
-            environment.data_collection_parameters.get("downscale")
-            if environment
-            else None
-        ) or True
-        metadata["small_boxsize"] = (
-            environment.data_collection_parameters.get("small_boxsize")
-            if environment
-            else None
-        ) or 128
+        metadata["gain_ref"] = environment.gain_ref if environment else None
+        metadata["dose_per_frame"] = environment.dose_per_frame if environment else None
+        metadata["symmetry"] = (environment.symmetry if environment else None) or "C1"
         metadata["eer_fractionation"] = (
-            environment.data_collection_parameters.get("eer_fractionation")
-            if environment
-            else None
+            environment.eer_fractionation if environment else None
         ) or 20
         metadata["source"] = str(self._basepath)
-        metadata["particle_diameter"] = (
-            environment.data_collection_parameters.get("particle_diameter")
-            if environment
-            else None
-        ) or 0
-        metadata["estimate_particle_diameter"] = (
-            environment.data_collection_parameters.get("estimate_particle_diameter")
-            if environment
-            else None
-        ) or True
         return metadata
 
     def _position_analysis(
@@ -584,10 +484,10 @@ def post_transfer(
                                 f"{str(environment.url.geturl())}{url_path_for('file_io_instrument.router', 'write_eer_fractionation_file', visit_name=environment.visit, session_id=environment.murfey_session)}",
                                 json={
                                     "eer_path": str(file_transferred_to),
-                                    "fractionation": environment.data_collection_parameters[
+                                    "fractionation": self.data_collection_parameters[
                                         "eer_fractionation"
                                     ],
-                                    "dose_per_frame": environment.data_collection_parameters[
+                                    "dose_per_frame": self.data_collection_parameters[
                                         "dose_per_frame"
                                     ],
                                     "fractionation_file_name": "eer_fractionation_spa.txt",
@@ -619,20 +519,16 @@ def post_transfer(
                             "image_number": environment.movies[
                                 file_transferred_to
                             ].movie_number,
-                            "pixel_size": environment.data_collection_parameters.get(
+                            "pixel_size": self.data_collection_parameters.get(
                                 "pixel_size_on_image"
                             ),
                             "autoproc_program_id": None,
-                            "dose_per_frame": environment.data_collection_parameters.get(
-                                "dose_per_frame"
-                            ),
-                            "mc_binning": environment.data_collection_parameters.get(
+                            "dose_per_frame": environment.dose_per_frame,
+                            "mc_binning": self.data_collection_parameters.get(
                                 "motion_corr_binning", 1
                             ),
-                            "gain_ref": environment.data_collection_parameters.get(
-                                "gain_ref"
-                            ),
-                            "extract_downscale": environment.data_collection_parameters.get(
+                            "gain_ref": environment.gain_ref,
+                            "extract_downscale": self.data_collection_parameters.get(
                                 "downscale", True
                             ),
                             "eer_fractionation_file": eer_fractionation_file,