[tools] Fix deprecation script for mulitple header files

Bug: v8:11165
Change-Id: Ic2c8c6e6b97f279941e8634bc6178511103edbca
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3173676Reviewed-by: Sathya Gunasekaran  <gsathya@chromium.org>
Commit-Queue: Camillo Bruni <cbruni@chromium.org>
Cr-Commit-Position: refs/heads/main@{#76988}

tools/release/list_deprecated.py

@@ -9,6 +9,8 @@ import re
 import subprocess
 import sys
 from pathlib import Path
+import logging
+from multiprocessing import Pool
 
 RE_GITHASH = re.compile(r"^[0-9a-f]{40}")
 RE_AUTHOR_TIME = re.compile(r"^author-time (\d+)$")
@@ -18,8 +20,25 @@ VERSION_CACHE = dict()
 RE_VERSION_MAJOR = re.compile(r".*V8_MAJOR_VERSION ([0-9]+)")
 RE_VERSION_MINOR = re.compile(r".*V8_MINOR_VERSION ([0-9]+)")
 
+RE_MACRO_END = re.compile(r"\);")
+RE_DEPRECATE_MACRO = re.compile(r"\(.*?,(.*)\);", re.MULTILINE)
+
+
+class HeaderFile(object):
+  def __init__(self, path):
+    self.path = path
+    self.blame_list = self.get_blame_list()
 
-def extract_version(hash):
+  @classmethod
+  def get_api_header_files(clazz, options):
+    files = subprocess.check_output(
+        ['git', 'ls-tree', '--name-only', '-r', 'HEAD', options.include_dir],
+        encoding='UTF-8')
+    files = filter(lambda l: l.endswith('.h'), files.splitlines())
+    with Pool(processes=24) as pool:
+      return pool.map(HeaderFile, files)
+
+  def extract_version(self, hash):
     if hash in VERSION_CACHE:
       return VERSION_CACHE[hash]
     if hash == '0000000000000000000000000000000000000000':
@@ -32,10 +51,11 @@ def extract_version(hash):
     VERSION_CACHE[hash] = version
     return version
 
-
-def get_blame(file_path):
+  def get_blame_list(self):
+    logging.info(f"blame list for {self.path}")
     result = subprocess.check_output(
-      ['git', 'blame', '-t', '--line-porcelain', file_path], encoding='UTF-8')
+        ['git', 'blame', '-t', '--line-porcelain', self.path],
+        encoding='UTF-8')
     line_iter = iter(result.splitlines())
     blame_list = list()
     current_blame = None
@@ -56,8 +76,8 @@ def get_blame(file_path):
         continue
       match = RE_AUTHOR_TIME.match(line)
       if match:
-      current_blame['datetime'] = datetime.fromtimestamp(int(
-          match.groups()[0]))
+        current_blame['datetime'] = datetime.fromtimestamp(
+            int(match.groups()[0]))
         continue
       match = RE_FILENAME.match(line)
       if match:
@@ -67,23 +87,21 @@ def get_blame(file_path):
     blame_list.append(current_blame)
     return blame_list
 
-
-RE_MACRO_END = re.compile(r"\);")
-RE_DEPRECATE_MACRO = re.compile(r"\(.*?,(.*)\);", re.MULTILINE)
-
-
-def filter_and_print(blame_list, macro, options):
+  def filter_and_print(self, macro, options):
     before = options.before
     index = 0
     re_macro = re.compile(macro)
     deprecated = list()
-  while index < len(blame_list):
-    blame = blame_list[index]
+    while index < len(self.blame_list):
+      blame = self.blame_list[index]
+      line = blame['content']
+      if line.startswith("#") or line.startswith("//"):
+        index += 1
+        continue
       commit_datetime = blame['datetime']
       if commit_datetime >= before:
         index += 1
         continue
-    line = blame['content']
       commit_hash = blame['hash']
       match = re_macro.search(line)
       if match:
@@ -94,7 +112,7 @@ def filter_and_print(blame_list, macro, options):
           if pos >= len(line):
             # Extend to next line
             index = index + 1
-          blame = blame_list[index]
+            blame = self.blame_list[index]
             line = line + blame['content']
           if line[pos] == '(':
             parens = parens + 1
@@ -111,39 +129,47 @@ def filter_and_print(blame_list, macro, options):
         content = line[start:pos].strip().replace('""', '')
         deprecated.append((index + 1, commit_datetime, commit_hash, content))
       index = index + 1
-  print(f"# Marked as {macro}: {len(deprecated)}")
+    if len(deprecated) == 0: return
     for linenumber, commit_datetime, commit_hash, content in deprecated:
       commit_date = commit_datetime.date()
-    file_position = (
-        f"{options.v8_header}:{linenumber}").rjust(len(options.v8_header) + 5)
+      file_position = (f"{self.path}:{linenumber}").rjust(40)
       print(f"   {file_position}\t{commit_date}\t{commit_hash[:8]}"
-          f"\t{extract_version(commit_hash)}\t{content}")
+            f"\t{self.extract_version(commit_hash)}\t{content}")
     return len(deprecated)
 
 
 def parse_options(args):
   parser = argparse.ArgumentParser(
       description="Collect deprecation statistics")
-  parser.add_argument("v8_header", nargs='?', help="Path to v8.h")
+  parser.add_argument("include_dir", nargs='?', help="Path to includes dir")
   parser.add_argument("--before", help="Filter by date")
+  parser.add_argument("--verbose",
+                      "-v",
+                      help="Verbose logging",
+                      action="store_true")
   options = parser.parse_args(args)
+  if options.verbose:
+    logging.basicConfig(level=logging.DEBUG)
   if options.before:
     options.before = datetime.strptime(options.before, '%Y-%m-%d')
   else:
     options.before = datetime.now()
-  if options.v8_header is None:
+  if options.include_dir is None:
     base_path = Path(__file__).parent.parent
-    options.v8_header = str(
-        (base_path / 'include' / 'v8.h').relative_to(base_path))
+    options.include_dir = str((base_path / 'include').relative_to(base_path))
   return options
 
 
 def main(args):
   options = parse_options(args)
-  blame_list = get_blame(options.v8_header)
-  filter_and_print(blame_list, "V8_DEPRECATE_SOON", options)
+  header_files = HeaderFile.get_api_header_files(options)
+  print("V8_DEPRECATE_SOON:")
+  for header in header_files:
+    header.filter_and_print("V8_DEPRECATE_SOON", options)
   print("\n")
-  filter_and_print(blame_list, "V8_DEPRECATED", options)
+  print("V8_DEPRECATED:")
+  for header in header_files:
+    header.filter_and_print("V8_DEPRECATED", options)
 
 
 if __name__ == "__main__":