Reuse existing code for hasher_from_chunks #3941 #4556

    * Move test

Signed-off-by: Jono Yang <jyang@nexb.com>

Author: JonoYang

src/commoncode/hash.py

@@ -229,15 +229,24 @@ def checksum(location, name, base64=False):
     return checksum_from_chunks(chunks=chunks, total_length=total_length, name=name, base64=base64)
 
 
+def hasher_from_chunks(chunks, name, total_length=0):
+    """
+    Return a hasher of ``name`` checksum algorithm of contains the contents of
+    the iterator of byte strings ``chunks`` of length ``total_length``.
+    """
+    hasher = get_hasher_instance_by_name(name=name, total_length=total_length)
+    for chunk in chunks:
+        hasher.update(chunk)
+    return hasher
+
+
 def checksum_from_chunks(chunks, name, total_length=0, base64=False):
     """
     Return a checksum from the content of the iterator of byte strings ``chunks`` with a
     ``total_length`` combined length using the ``name`` checksum algorithm. The returned checksum is
     a string as a hexdigest or is base64-encoded is ``base64`` is True.
     """
-    hasher = get_hasher_instance_by_name(name=name, total_length=total_length)
-    for chunk in chunks:
-        hasher.update(chunk)
+    hasher = hasher_from_chunks(chunks=chunks, name=name, total_length=total_length)
     if base64:
         return hasher.b64digest()
     return hasher.hexdigest()

src/scancode/results_cache.py

@@ -13,6 +13,7 @@
 
 from commoncode.fileutils import create_dir
 from commoncode.hash import binary_chunks
+from commoncode.hash import hasher_from_chunks
 from scancode_config import results_cache_dir
 
 
@@ -38,22 +39,13 @@
 After scanning, the cache is updated.
 """
 
-def hasher_from_chunks(chunks):
-    """
-    Return a sha256 hasher loaded with `chunks`.
-    """
-    hasher = hashlib.sha256()
-    for chunk in chunks:
-        hasher.update(chunk)
-    return hasher
-
 
 def compute_results_cache_index(location, filename):
     """
     Compute results_cache_index value for a Resource at `location`.
     """
     chunks = binary_chunks(location=location)
-    sha256_hasher = hasher_from_chunks(chunks=chunks)
+    sha256_hasher = hasher_from_chunks(chunks=chunks, name='sha256')
     sha256_hasher.update(filename.encode('utf-8', 'surrogateescape'))
     return sha256_hasher.hexdigest()
 

tests/commoncode/test_hash.py

@@ -11,9 +11,11 @@
 import os
 
 from commoncode.hash import b64sha1
+from commoncode.hash import binary_chunks
 from commoncode.hash import checksum
 from commoncode.hash import checksum_from_chunks
 from commoncode.hash import get_hasher
+from commoncode.hash import hasher_from_chunks
 from commoncode.hash import md5
 from commoncode.hash import multi_checksums
 from commoncode.hash import sha1
@@ -200,3 +202,10 @@ def test_checksum_empty_file(self):
         test_file = self.get_test_loc("hash/empty")
         checksums = multi_checksums(location=test_file, checksum_names=("sha1",))
         assert checksums == {"sha1": None}
+
+    def test_hasher_from_chunks(self):
+        test_file = self.get_test_loc("hash/dir1/a.png")
+        chunks = binary_chunks(location=test_file)
+        hasher = hasher_from_chunks(chunks=chunks, name='sha256')
+        expected = "1b598db6fee8f1ec7bb919c0adf68956f3d20af8c9934a9cf2db52e1347efd35"
+        assert hasher.hexdigest() == expected

tests/scancode/test_results_cache.py

@@ -11,7 +11,6 @@
 from os.path import dirname
 from os.path import join
 
-from commoncode.hash import binary_chunks
 from commoncode.testcase import FileDrivenTesting
 from scancode import results_cache
 
@@ -21,13 +20,6 @@ class TestResultsCache(FileDrivenTesting):
     test_results_cache = join(test_data_dir, 'results_cache/results')
     test_results_cache_index = "b0e3dd13b9b5980bb1ca7aab89e5490cb136952374489a755947ac004309b035"
 
-    def test_hasher_from_chunks(self):
-        test_file_loc = self.get_test_loc('results_cache/package.json')
-        chunks = binary_chunks(location=test_file_loc)
-        hasher = results_cache.hasher_from_chunks(chunks=chunks)
-        expected = "cb6f5a82e473620da4d1aecf82dd4d4fa9ada393a7679b28a42cac86f0a83c92"
-        assert hasher.hexdigest() == expected
-
     def test_compute_results_cache_index(self):
         test_file_loc = self.get_test_loc('results_cache/package.json')
         results_cache_index = results_cache.compute_results_cache_index(location=test_file_loc, filename='package.json')