Set of commits to be tagged v0.18

Changelog

Major refactoring of method BackendExplorer.prepare_labels:

• argument allow_file_extensions to filter out input files based on file extension,
• argument single_input=True to pick a primary input file first, and then select the related files only,
• allow_file_extensions and single_input apply only when no metadata file is available,
• the filename metadata is used to identify the primary input file, which now comes first in the list of dependencies.

src/taggingbackends/explorer.py

@@ -24,6 +24,7 @@ else:
         logging.warning("Failed to set PyCall with compiled_modules=False")
     try:
         from julia import TaggingBackends
+        from julia import PlanarLarvae
     except ImportError:
         logging.warning(f"TaggingBackends not found in JULIA_PROJECT={JULIA_PROJECT}; \
 please see https://gitlab.pasteur.fr/nyx/TaggingBackends#recommended-installation")
@@ -443,11 +444,16 @@ run `poetry add {pkg}` from directory: \n
             files = dict(files_)
         return files
 
-    def prepare_labels(self, input_files):
+    def prepare_labels(self, input_files, allowed_file_extensions=None,
+        single_input=False):
         if isinstance(input_files, dict):
             input_files_and_labels = dict()
             for parent in input_files.keys():
-                input_files_and_labels[parent] = self.prepare_labels(input_files[parent])
+                input_files_and_labels[parent] = self.prepare_labels(
+                    input_files[parent],
+                    single_input=single_input,
+                    allowed_file_extensions=allowed_file_extensions,
+                )
             return input_files_and_labels
         #
         from .data.labels import Labels, labels_file_extension
@@ -456,30 +462,61 @@ run `poetry add {pkg}` from directory: \n
         # note: these *metadata* files are pushed by LarvaTagger.jl and form an
         #       undocumented mechanism to preserve metadata that are generated
         #       on the Julia side.
-        found = 0
+        metadata_found = 0
         for file in list(input_files):
             if file.name == "metadata":
                 input_files.remove(file)
                 with open(file, "r") as f:
                     labels.metadata = json.load(f)
-                if found == 1:
+                if metadata_found == 1:
                     logging.warning("multiple metadata files found")
-                found += 1
-        assert 0 < len(input_files)
-        # if any labels file is found, assume the other files are its data
-        # dependencies, and exclude the labels file from the list of input files
-        found = 0
-        for file in list(input_files):
-            if any([file.name.endswith(ext) for ext in labels_file_extension]):
-                labels.input_labels = file
-                input_files.remove(file)
-                if found == 1:
-                    logging.warning("multiple label files found")
-                found += 1
+                metadata_found += 1
+        # check whether an input data file has been designated
+        primary_input = None
+        if metadata_found and 'filename' in labels.metadata.keys():
+            primary_input = labels.metadata['filename']
+            for file in input_files:
+                if file.name == primary_input:
+                    primary_input = file
+                    break
+            if isinstance(primary_input, str):
+                logging.warning(f'file not found: {primary_input}')
+                primary_input = None
+        # identify existing labels (label files only)
+        labels_found = 0
+        if primary_input is None:
+            for file in input_files:
+                if any(file.name.endswith(ext) for ext in labels_file_extension):
+                    labels.input_labels = file
+                    if labels_found == 1:
+                        logging.warning("multiple label files found")
+                    labels_found += 1
+        elif any(primary_input.name.endswith(ext) for ext in labels_file_extension):
+            labels.input_labels = primary_input
+            labels_found = 1
+        # filter out likely unsupported files
+        if allowed_file_extensions:
+            input_files = [f for f in input_files
+                if any(f.name.endswith(ext) for ext in allowed_file_extensions)]
+        # force-identify the input data files
         assert 0 < len(input_files)
+        if single_input and primary_input is None:
+            primary_input = input_files[0]
+            logging.info(f'selecting file: {primary_input}')
+            input_files = [pathlib.Path(f.source)
+                for f in PlanarLarvae.Formats.find_associated_files(str(primary_input))]
+        # ensure the primary input file comes first in the list of input files
+        if not (primary_input is None or input_files[0] == primary_input):
+            input_files = [f for f in input_files if f != primary_input]
+            input_files.insert(0, primary_input)
+        #
         labels.tracking = input_files
         #
-        if not labels.metadata:
+        if labels.metadata:
+            if labels.metadata.get('date_time', None) == 'NA':
+                logging.debug('discarding metadata entry "date_time"')
+                del labels.metadata['date_time']
+        else:
             file = input_files[0]
             labels.metadata = {'filename': file.name}
         #