Developed a metadata harvesting object to facilitate metadata collection throught the code.

src/metadata_review_lib.py

@@ -316,6 +316,82 @@ def last_submit_metadata_review(reviewer_attrs):
 #import config_file
 #import hdf5_vis
 
+class MetadataHarvester:
+    def __init__(self, parent_files=None):
+        if parent_files is None:
+            parent_files = []
+        self.parent_files = parent_files
+        self.metadata = {
+            "samples": {},
+            "environment": {},
+            "instrument": {},
+            "data_level": {}
+        }
+
+    def add_sample_info(self, key_or_dict, value=None):
+        self._add_info("samples", key_or_dict, value)
+
+    def add_environment_info(self, key_or_dict, value=None):
+        self._add_info("environment", key_or_dict, value)
+
+    def add_instrument_info(self, key_or_dict, value=None):
+        self._add_info("instrument", key_or_dict, value)
+
+    def add_data_level_info(self, key_or_dict, value=None):
+        self._add_info("data_level", key_or_dict, value)
+
+    def _add_info(self, category, key_or_dict, value):
+        """Internal helper method to add information to a category."""
+        if isinstance(key_or_dict, dict):
+            self.metadata[category].update(key_or_dict)
+        else:
+            self.metadata[category][key_or_dict] = value
+
+    def get_metadata(self):
+        return {
+            "parent_files": self.parent_files,
+            "metadata": self.metadata
+        }
+
+    def clear_metadata(self):
+        self.metadata = {
+            "samples": {},
+            "environment": {},
+            "instrument": {},
+            "data_level": {}
+        }
+        self.parent_files = []
+
+# Example usage
+if __name__ == "__main__":
+    parent_files = ["file1.txt", "file2.txt", "file3.txt"]
+    harvester = MetadataHarvester(parent_files)
+
+    # Adding sample information
+    harvester.add_sample_info("sample_id", "S001")
+    harvester.add_sample_info({"sample_type": "blood", "sample_volume": "5ml"})
+
+    # Adding environment information
+    harvester.add_environment_info("location", "lab_1")
+    harvester.add_environment_info({"temperature": "22C", "humidity": "50%"})
+
+    # Adding instrument information
+    harvester.add_instrument_info("instrument_id", "I123")
+    harvester.add_instrument_info({"instrument_type": "microscope", "manufacturer": "XYZ Corp"})
+
+    # Adding data level information
+    harvester.add_data_level_info("processing_step", "initial")
+    harvester.add_data_level_info({"data_quality": "high", "data_format": "CSV"})
+
+    # Retrieving metadata
+    metadata = harvester.get_metadata()
+    print(metadata)
+
+    # Clear metadata if needed
+    harvester.clear_metadata()
+    print("Cleared metadata:", harvester.get_metadata())
+
+
 def main():    
 
     output_filename_path = "output_files/unified_file_smog_chamber_2024-03-19_UTC-OFST_+0100_NG.h5"