Update histogram paths and added sorting in merge

Function to split histograms by prefix

- Amended the histogram_paths.py to prepare for the distributed
directories of histograms after splitting.
- Amended two other files that previously expected a single
histogram.xml file.
- Added sorting of histogram nodes by name in merge_xml.py.

Change-Id: I60101777806584722cbeb778c778553259419122
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2337155
Commit-Queue: Wenhan (Han) Zhang <[email protected]>
Reviewed-by: Steven Holte <[email protected]>
Cr-Commit-Position: refs/heads/master@{#796655}

tools/metrics/histograms/histogram_ownership.py

@@ -9,19 +9,18 @@
 
 from __future__ import print_function
 
-import extract_histograms
+import errno
 import os
 import sys
-import xml.etree.ElementTree
+from xml.etree import ElementTree as ET
 
-sys.path.append(os.path.join(os.path.dirname(__file__), '..', 'common'))
-import path_util
+import extract_histograms
+import histogram_paths
+import merge_xml
 
-def main():
-  tree = xml.etree.ElementTree.parse(path_util.GetHistogramsFile())
-  root = tree.getroot()
-  assert root.tag == 'histogram-configuration'
 
+def PrintOwners(root):
+  assert root.tag == 'histogram-configuration'
   root_children = root.getchildren()
   histograms = None
   for node in root_children:
@@ -45,7 +44,6 @@ def main():
         continue
       if node.text == extract_histograms.OWNER_PLACEHOLDER:
         continue
-      assert '@' in node.text
       owners.append(node.text)
 
     if not obsolete:
@@ -55,5 +53,33 @@ def main():
         print(name, 'NO_OWNER')
 
 
+def main():
+  """Prints the owners of histograms in a specific file or of all histograms.
+
+  Args:
+    argv[1]: Optional argument that is the relative path to a specific
+        histograms.xml.
+
+  Example usage to print owners of histograms_xml/Accessibility/histograms.xml:
+    python histogram_ownership.py histograms_xml/Accessibility/histograms.xml
+
+  Example usage to print owners of all histograms:
+    python histogram_ownership.py
+  """
+  if len(sys.argv) == 1:
+    merged_xml_string = merge_xml.PrettyPrintMergedFiles(
+        histogram_paths.ALL_XMLS)
+    root = ET.fromstring(merged_xml_string)
+  else:
+    rel_path = sys.argv[1]
+    if not os.path.exists(rel_path):
+      raise FileNotFoundError(errno.ENOENT, os.strerror(errno.ENOENT))
+
+    tree = ET.parse(rel_path)
+    root = tree.getroot()
+
+  PrintOwners(root)
+
+
 if __name__ == '__main__':
   main()

tools/metrics/histograms/histogram_paths.py

@@ -10,17 +10,43 @@
 sys.path.append(os.path.join(os.path.dirname(__file__), '..', 'common'))
 import path_util
 
-ALL_XMLS_RELATIVE = [
-    'tools/metrics/histograms/enums.xml',
-    'tools/metrics/histograms/histograms.xml'
-]
-ALL_XMLS = [path_util.GetInputFile(f) for f in ALL_XMLS_RELATIVE]
-ENUMS_XML, HISTOGRAMS_XML = ALL_XMLS
+
+def _FindHistogramsXmlFiles(base_dir):
+  """Gets a list of all histograms.xml files under the directory tree, prefixed
+  with tools/metrics/histograms/."""
+  file_list = []
+  for dirName, _, fileList in os.walk(base_dir):
+    for filename in fileList:
+      if filename == 'histograms.xml':
+        filepath = os.path.join('tools/metrics/histograms/', dirName, filename)
+        file_list.append(filepath)
+  return file_list
+
+
+ENUMS_XML_RELATIVE = 'tools/metrics/histograms/enums.xml'
+# This file path accounts for cases where the script is executed from other
+# metrics-related directories.
+PATH_TO_HISTOGRAMS_XML_DIR = os.path.join('..', 'histograms/histograms_xml')
+# In the middle state, histogram paths include both the large histograms.xml
+# file as well as the split up files.
+# TODO: Improve on the current design to avoid calling `os.walk()` at the time
+# of module import.
+HISTOGRAMS_XMLS_RELATIVE = (['tools/metrics/histograms/histograms.xml'] +
+                            _FindHistogramsXmlFiles(PATH_TO_HISTOGRAMS_XML_DIR))
+OBSOLETE_XML_RELATIVE = ('tools/metrics/histograms/histograms_xml/'
+                         'obsolete_histograms.xml')
+ALL_XMLS_RELATIVE = [ENUMS_XML_RELATIVE, OBSOLETE_XML_RELATIVE
+                     ] + HISTOGRAMS_XMLS_RELATIVE
+
+ENUMS_XML = path_util.GetInputFile(ENUMS_XML_RELATIVE)
+HISTOGRAMS_XMLS = [path_util.GetInputFile(f) for f in HISTOGRAMS_XMLS_RELATIVE]
+OBSOLETE_XML = path_util.GetInputFile(OBSOLETE_XML_RELATIVE)
+ALL_XMLS = [ENUMS_XML, OBSOLETE_XML] + HISTOGRAMS_XMLS
 
 ALL_TEST_XMLS_RELATIVE = [
     'tools/metrics/histograms/test_data/enums.xml',
     'tools/metrics/histograms/test_data/histograms.xml',
     'tools/metrics/histograms/test_data/ukm.xml',
 ]
 ALL_TEST_XMLS = [path_util.GetInputFile(f) for f in ALL_TEST_XMLS_RELATIVE]
-TEST_ENUMS_XML, TEST_HISTOGRAMS_XML, TEST_UKM_XML = ALL_TEST_XMLS
+TEST_ENUMS_XML, TEST_HISTOGRAMS_XML, TEST_UKM_XML = ALL_TEST_XMLS