Addition of cache removal classmethod

Author: ArthurDeclercq

mumble/__init__.py

@@ -1,4 +1,4 @@
 __version__ = "0.2.0"
 __all__ = ["PSMHandler"]
 
-from mumble.mumble import PSMHandler
+from mumble.mumble import PSMHandler, _ModificationCache

mumble/__main__.py

@@ -1,9 +1,10 @@
 import click
 import logging
+import sys
 
 from rich.logging import RichHandler
 
-from mumble import PSMHandler
+from mumble import PSMHandler, _ModificationCache
 
 
 # setup logging
@@ -96,11 +97,23 @@
 
 
 @click.command("cli", context_settings={"show_default": True})
-@click.argument("input_file", type=click.Path(exists=True), default=None)
-def main(**kwargs):
+@click.argument("input_file", type=click.Path(exists=True), default=None, required=False)
+@click.option(
+    "--clear-cache/--no-clear-cache",
+    is_flag=True,
+    default=False,
+    help="Remove the modification cache file and exit early.",
+)
+def main(clear_cache, **kwargs):
     """
     Finding the perfect match for your mass shift.
     """
+    # if the user just wants to clear the cache, do it and quit
+    if clear_cache:
+        _ModificationCache._remove_cache()
+        logging.info("Exiting Mumble. You will find your match another time.")
+        sys.exit(0)
+
     # Set the logging level based on the CLI option
     log_level = kwargs.get("log_level", "INFO").upper()
     logging.getLogger().setLevel(log_level)

mumble/mumble.py

@@ -417,6 +417,7 @@ def __init__(
             exclude_mutations=exclude_mutations,
             modification_file=unimod_modification_file,
         )
+        self.cache.load_cache()
 
         self.modification_df = self.cache.modification_df
         self.monoisotopic_masses = self.cache.monoisotopic_masses
@@ -780,11 +781,32 @@ def __init__(
         self.modifications_names = []
         self.modification_df = None
 
-        # Load or generate data
-        cache_file = self._get_cache_file_path()
-        self._load_or_generate_data(cache_file, force_reload=False)
+        # get cache file path
+        self.cache_file = self._get_cache_file_path()
 
-    def _get_cache_file_path(self):
+    @classmethod
+    def _remove_cache(cls):
+        """
+        Remove the cache file for modifications.
+        """
+        cache_file = cls._get_cache_file_path()
+        if os.path.exists(cache_file):
+            os.remove(cache_file)
+            logger.info("Modification cache removed.")
+        else:
+            logger.warning("Modification cache file does not exist.")
+
+    def load_cache(self, force_reload=False):
+        """
+        Load the cache or generate it if it doesn't exist.
+
+        Args:
+            force_reload (bool, optional): If True, regenerate the cache even if it exists. Defaults to False.
+        """
+        self._load_or_generate_data(self.cache_file, force_reload=force_reload)
+
+    @classmethod
+    def _get_cache_file_path(cls):
         """
         Get path to cache file for combinations of modifications.
 
@@ -828,16 +850,16 @@ def _load_or_generate_data(self, cache_file: str, force_reload: bool = False) ->
             if cache_data["metadata"] == (
                 self.combination_length,
                 self.exclude_mutations,
-                self.modification_file,
                 self.modification_file_hash,
             ):
+                logger.debug("Cache metadata matches current configuration")
                 try:
-                    logger.info("Loading cache data")
+                    logger.info("Using cached modifcation data")
                     self.modification_df = cache_data["modification_df"]
                     self.monoisotopic_masses = cache_data["monoisotopic_masses"]
                     self.modifications_names = cache_data["modifications_names"]
                 except KeyError:
-                    logger.info("Cache data missing")
+                    logger.info("Cached data invalid or incomplete, regenerating cache")
                     self._regenerate_and_save_cache(cache_file)
             else:
                 self._regenerate_and_save_cache(cache_file)
@@ -986,15 +1008,14 @@ def _regenerate_and_save_cache(self, cache_file: str) -> None:
             self._generate_modifications_combinations_lists(self.combination_length)
         )
         logger.debug(
-            f"New cache metadata: {self.combination_length}, {self.exclude_mutations}, {self.modification_file}, {self.modification_file_hash}"
+            f"New cache metadata: \ncombination length {self.combination_length}, \nexclude_mutations {self.exclude_mutations},\nmodification file hash {self.modification_file_hash}",
         )
         with open(cache_file, "wb") as f:
             pickle.dump(
                 {
                     "metadata": (
                         self.combination_length,
                         self.exclude_mutations,
-                        self.modification_file,
                         self.modification_file_hash,
                     ),
                     "modification_df": self.modification_df,