add_handler_remove_hashcode

Currently, there are hash codes at the end of some dictionary names, making it difficult for other handlers to get the correct path. This handler "HashRemoveHandler" recursively obtain all the directories, remove hash code and save a new file. Hash code is judged by un underline followed by 8 random characters at the end of the directory's name, e.g."_(\w{8})$", and for directories with same name but only different hash code, they are distinguished by adding _variantX where X ∈ 1, 2, 3 ... N.

handlers/HashRemoveHandler.py

+import ROOT
+import re
+import tempfile
+import os
+import shutil
+
+class HashRemoveHandler:
+
+    """This class is used to remove the hash code at the end of the directory names of a rootfile. 
+    Hash code is judged by an underline and 8 random characters e.g."_(\w{8})$". 
+    If both input_file_path and output_file_path are provided, modified file will be saved in output_file_path;
+    If only input_file_path is provided, modified file will replace the old file"""
+    # Example usage
+    # modifier = HashRemoveHandler("old_file.root", "new_file.root") or modifier = HashRemoveHandler("old_file.root")
+    # modifier.modify_root_file()
+        
+    def __init__(self, input_file_path, output_file_path=None):
+        self.input_file_path = input_file_path
+        self.output_file_path = output_file_path
+
+    def process_directory(self, old_dir, new_parent_dir, subdir_counts=None, parent_path=""):
+        if subdir_counts is None:
+            subdir_counts = {}
+        directory_path = "/".join(filter(None, [parent_path, old_dir.GetName()]))
+        new_dir_name = self.remove_hashcode(old_dir.GetName())
+        complete_dir_name = self.remove_hashcode(directory_path)
+        subdir_counts[complete_dir_name] = subdir_counts.get(complete_dir_name, 0) + 1
+        variant = subdir_counts[complete_dir_name]
+        if variant > 1:
+            new_dir_name += "_variant" + str(variant)
+        if new_dir_name != self.input_file_path:
+            new_dir = new_parent_dir.mkdir(new_dir_name)
+        if new_dir_name == self.input_file_path:
+            new_dir = new_parent_dir
+
+        for key in old_dir.GetListOfKeys():
+            obj = key.ReadObj()
+            if isinstance(obj, ROOT.TDirectory):
+                self.process_directory(obj, new_dir, subdir_counts, parent_path=directory_path)
+            elif isinstance(obj, ROOT.TH1):
+                new_dir.cd()
+                histogram = obj.Clone()
+                histogram.Write()
+
+    def remove_hashcode(self, directory_name):
+        # Remove hash code at the end of the directory name
+        pattern = r"_(\w{8})$"
+        match = re.search(pattern, directory_name)
+        if match:
+            return directory_name[:match.start()]
+        return directory_name
+
+    def modify_root_file(self):
+        ROOT.gROOT.SetBatch(True)  # Enable batch mode to suppress graphics
+
+        # Open the original ROOT file
+        input_file = ROOT.TFile.Open(self.input_file_path)
+        if input_file is None or input_file.IsZombie():
+            print("Error: Failed to open the original file '{}'".format(self.input_file_path))
+            return
+
+        if self.output_file_path is None:
+            # If the output_file_path is not provided, create a temporary file
+            tmp_output_file_path = os.path.join(tempfile.gettempdir(), "tmp_modified_file.root")
+            output_file = ROOT.TFile(tmp_output_file_path, "RECREATE")
+        else:
+            # If the output_file_path is provided, create a new ROOT file
+            output_file = ROOT.TFile(self.output_file_path, "RECREATE")
+
+        if output_file.IsZombie():
+            print("Error: Failed to create the new file '{}'".format(self.output_file_path))
+            return
+
+        # Process the directories recursively
+        root_dir = input_file.GetDirectory("/")
+        self.process_directory(root_dir, output_file)
+
+        # Write and close the new ROOT file
+        output_file.Write()
+        output_file.Close()
+
+        # Close the original ROOT file
+        input_file.Close()
+
+        if self.output_file_path is None:
+            # If the output_file_path is not provided, replace the original file with the temporary one
+            shutil.move(tmp_output_file_path, self.input_file_path)
+
+        ROOT.gROOT.SetBatch(False)  # Disable batch mode