Fix AarResourcesExtractor action to produce consistent zips

The `aar_native_libs_zip_creator.py` script was producing inconsistent output zips for the same input files due to changing the last modified date of the input files. This resulted in inconsistent cache keys and therefore poor build results.

Closes #11970.

PiperOrigin-RevId: 345541072

Author: nkoroste

tools/android/aar_native_libs_zip_creator.py

@@ -63,7 +63,16 @@ def CreateNativeLibsZip(aar, cpu, native_libs_zip):
       # Only replaces the first instance of jni, in case the AAR contains
       # something like /jni/x86/jni.so.
       new_filename = lib.replace("jni", "lib", 1)
-      native_libs_zip.writestr(new_filename, aar.read(lib))
+      # To guarantee reproducible zips we must specify a new zipinfo.
+      # From writestr docs: "If its a name, the date and time is set to the
+      # current date and time." which will break the HASH calculation and result
+      # in a cache miss.
+      old_zipinfo = aar.getinfo(lib)
+      new_zipinfo = zipfile.ZipInfo(filename=new_filename)
+      new_zipinfo.date_time = old_zipinfo.date_time
+      new_zipinfo.compress_type = old_zipinfo.compress_type
+
+      native_libs_zip.writestr(new_zipinfo, aar.read(lib))
 
 
 def Main(input_aar_path, output_zip_path, cpu, input_aar_path_for_error_msg):

tools/android/aar_native_libs_zip_creator_test.py

@@ -14,21 +14,38 @@
 
 """Tests for aar_native_libs_zip_creator."""
 
+import hashlib
 import io
+import tempfile
+import time
 import unittest
 import zipfile
 
 from tools.android import aar_native_libs_zip_creator
 
 
+def md5(buf):
+  hash_md5 = hashlib.md5()
+  hash_md5.update(buf)
+  return hash_md5.hexdigest()
+
+
+def md5_from_file(fname):
+  hash_md5 = hashlib.md5()
+  with open(fname, "rb") as f:
+    for chunk in iter(lambda: f.read(4096), b""):
+      hash_md5.update(chunk)
+  return hash_md5.hexdigest()
+
+
 class AarNativeLibsZipCreatorTest(unittest.TestCase):
   """Unit tests for aar_native_libs_zip_creator.py."""
 
   def testAarWithNoLibs(self):
     aar = zipfile.ZipFile(io.BytesIO(), "w")
     outzip = zipfile.ZipFile(io.BytesIO(), "w")
     aar_native_libs_zip_creator.CreateNativeLibsZip(aar, "x86", outzip)
-    self.assertEquals([], outzip.namelist())
+    self.assertEqual([], outzip.namelist())
 
   def testAarWithMissingLibs(self):
     aar = zipfile.ZipFile(io.BytesIO(), "w")
@@ -48,6 +65,50 @@ def testAarWithAllLibs(self):
     self.assertIn("lib/x86/foo.so", outzip.namelist())
     self.assertNotIn("lib/armeabi/foo.so", outzip.namelist())
 
+  def testMultipleInvocationConsistency(self):
+    input_aar = tempfile.NamedTemporaryFile(delete=False)
+    aar = zipfile.ZipFile(input_aar.name, "w")
+    aar.writestr(zipfile.ZipInfo(filename="jni/x86/foo.so"), "foo")
+    aar.writestr(zipfile.ZipInfo(filename="jni/x86/bar.so"), "bar")
+    aar.close()
+    input_aar.close()
+    # CreateNativeLibsZip expects a readonly file, this is not required but
+    # more correct
+    readonly_aar = zipfile.ZipFile(input_aar.name, "r")
+
+    outfile1 = tempfile.NamedTemporaryFile(delete=False)
+    outzip1 = zipfile.ZipFile(outfile1.name, "w")
+    aar_native_libs_zip_creator.CreateNativeLibsZip(readonly_aar, "x86",
+                                                    outzip1)
+    outfile1.close()
+
+    # Must be more than 1 second because last modified date changes on second
+    # basis
+    time.sleep(2)
+
+    outfile2 = tempfile.NamedTemporaryFile(delete=False)
+    outzip2 = zipfile.ZipFile(outfile2.name, "w")
+    aar_native_libs_zip_creator.CreateNativeLibsZip(readonly_aar, "x86",
+                                                    outzip2)
+    outfile2.close()
+
+    self.assertIn("lib/x86/foo.so", outzip1.namelist())
+    self.assertIn("lib/x86/bar.so", outzip1.namelist())
+    self.assertNotEqual(
+        md5(outzip1.read("lib/x86/foo.so")),
+        md5(outzip1.read("lib/x86/bar.so")))
+
+    self.assertIn("lib/x86/foo.so", outzip2.namelist())
+    self.assertIn("lib/x86/bar.so", outzip2.namelist())
+    self.assertNotEqual(
+        md5(outzip1.read("lib/x86/foo.so")),
+        md5(outzip1.read("lib/x86/bar.so")))
+
+    # The hash for the output zips must always match if the inputs match.
+    # Otherwise, there will be a cache miss which will produce poort build
+    # times.
+    self.assertEqual(md5_from_file(outfile1.name), md5_from_file(outfile2.name))
+
 
 if __name__ == "__main__":
   unittest.main()