Implement #624: Use shorter hashes with CPM_SOURCE_CACHE (#631)

* Add ASSERT_CONTENTS_EQUAL test macro in testing.cmake

Checks if the contents of a file matches the given input

* Use shorter hashes with CPM_SOURCE_CACHE (#624)

Uses shorter hashes with CPM_SOURCE_CACHE. Falls back to a longer hash
if necessary (ie, if there's a collision with an existing hash).

See: https://github.com/cpm-cmake/CPM.cmake/issues/624

* Update integration tests to support shorter hashes

* trigger ci

* run cmake-format

* if already available, use the legacy cache hash

* create temporary file in current binary dir

* add test case for legacy hash

---------

Co-authored-by: Lars Melchior <lars.melchior@gmail.com>
Co-authored-by: Lars Melchior <TheLartians@users.noreply.github.com>

test/integration/test_source_cache.rb

@@ -75,6 +75,17 @@ class SourceCache < IntegrationTest
     assert_equal ver, package.ver
     expected_parent_dir = File.join(@cache_dir, name.downcase)
     assert package.src_dir.start_with?(expected_parent_dir), "#{package.src_dir} must be in #{expected_parent_dir}"
-    assert_equal dir_sha1, File.basename(package.src_dir)
+
+    # The hash has been shortened by cpm_get_shortest_hash. The following
+    # should hold:
+    # - The short hash should be a prefix of the input hash
+    # - There should be a file ".../${short_hash}.hash" which matches the full hash
+    short_hash = File.basename(package.src_dir)
+    assert dir_sha1.start_with?(short_hash), "short_hash should be a prefix of dir_sha1"
+
+    # Check that the full hash is stored in the .hash file
+    hash_file = "#{package.src_dir}.hash"
+    assert File.exist?(hash_file), "Hash file #{hash_file} should exist"
+    assert_equal dir_sha1, File.read(hash_file), "Hash file should contain the full original hash"
   end
 end

test/unit/get_shortest_hash.cmake

@@ -0,0 +1,93 @@
+cmake_minimum_required(VERSION 3.14 FATAL_ERROR)
+
+include(${CPM_PATH}/CPM.cmake)
+include(${CPM_PATH}/testing.cmake)
+
+# Random suffix
+string(
+  RANDOM
+  LENGTH 6
+  ALPHABET "0123456789abcdef" tmpdir_suffix
+)
+
+# Seconds since epoch
+string(TIMESTAMP tmpdir_base "%s" UTC)
+
+set(tmp "${CMAKE_CURRENT_BINARY_DIR}/get_shortest_hash-${tmpdir_base}-${tmpdir_suffix}")
+
+if(IS_DIRECTORY ${tmp})
+  message(FATAL_ERROR "Test directory ${tmp} already exists")
+endif()
+
+file(MAKE_DIRECTORY "${tmp}")
+
+# 1. Sanity check: none of these directories should exist yet
+
+assert_not_exists(${tmp}/cccb.hash)
+assert_not_exists(${tmp}/cccb77ae.hash)
+assert_not_exists(${tmp}/cccb77ae9609.hash)
+assert_not_exists(${tmp}/cccb77ae9608.hash)
+assert_not_exists(${tmp}/cccb77be.hash)
+
+# 1. The directory is empty, so it should get a 4-character hash
+cpm_get_shortest_hash(${tmp} "cccb77ae9609d2768ed80dd42cec54f77b1f1455" hash)
+assert_equal(${hash} "cccb")
+assert_contents_equal(${tmp}/cccb.hash cccb77ae9609d2768ed80dd42cec54f77b1f1455)
+
+# 1. Calling the function with a new hash that differs subtly should result in more characters being
+#   used, enough to uniquely identify the hash
+
+cpm_get_shortest_hash(${tmp} "cccb77ae9609d2768ed80dd42cec54f77b1f1456" hash)
+assert_equal(${hash} "cccb77ae")
+assert_contents_equal(${tmp}/cccb77ae.hash cccb77ae9609d2768ed80dd42cec54f77b1f1456)
+
+cpm_get_shortest_hash(${tmp} "cccb77ae9609d2768ed80dd42cec54f77b1f1457" hash)
+assert_equal(${hash} "cccb77ae9609")
+assert_contents_equal(${tmp}/cccb77ae9609.hash cccb77ae9609d2768ed80dd42cec54f77b1f1457)
+
+cpm_get_shortest_hash(${tmp} "cccb77ae9608d2768ed80dd42cec54f77b1f1455" hash)
+assert_equal(${hash} "cccb77ae9608")
+assert_contents_equal(${tmp}/cccb77ae9608.hash cccb77ae9608d2768ed80dd42cec54f77b1f1455)
+
+cpm_get_shortest_hash(${tmp} "cccb77be9609d2768ed80dd42cec54f77b1f1456" hash)
+assert_equal(${hash} "cccb77be")
+assert_contents_equal(${tmp}/cccb77be.hash cccb77be9609d2768ed80dd42cec54f77b1f1456)
+
+# check that legacy hashs are recognized
+file(MAKE_DIRECTORY "${tmp}/cccb77be9609d2768ed80dd42cec54f77b1f1457")
+cpm_get_shortest_hash(${tmp} "cccb77be9609d2768ed80dd42cec54f77b1f1457" hash)
+assert_equal(${hash} "cccb77be9609d2768ed80dd42cec54f77b1f1457")
+
+# 1. The old file should still exist, and have the same content
+assert_contents_equal(${tmp}/cccb.hash cccb77ae9609d2768ed80dd42cec54f77b1f1455)
+assert_contents_equal(${tmp}/cccb77ae.hash cccb77ae9609d2768ed80dd42cec54f77b1f1456)
+assert_contents_equal(${tmp}/cccb77ae9609.hash cccb77ae9609d2768ed80dd42cec54f77b1f1457)
+assert_contents_equal(${tmp}/cccb77ae9608.hash cccb77ae9608d2768ed80dd42cec54f77b1f1455)
+assert_contents_equal(${tmp}/cccb77be.hash cccb77be9609d2768ed80dd42cec54f77b1f1456)
+
+# 1. Confirm idempotence: calling any of these function should produce the same hash as before (hash
+#   lookups work correctly once the .hash files are created)
+
+cpm_get_shortest_hash(${tmp} "cccb77ae9609d2768ed80dd42cec54f77b1f1455" hash)
+assert_equal(${hash} "cccb")
+assert_contents_equal(${tmp}/cccb.hash cccb77ae9609d2768ed80dd42cec54f77b1f1455)
+
+cpm_get_shortest_hash(${tmp} "cccb77ae9609d2768ed80dd42cec54f77b1f1456" hash)
+assert_equal(${hash} "cccb77ae")
+assert_contents_equal(${tmp}/cccb77ae.hash cccb77ae9609d2768ed80dd42cec54f77b1f1456)
+
+cpm_get_shortest_hash(${tmp} "cccb77ae9609d2768ed80dd42cec54f77b1f1457" hash)
+assert_equal(${hash} "cccb77ae9609")
+assert_contents_equal(${tmp}/cccb77ae9609.hash cccb77ae9609d2768ed80dd42cec54f77b1f1457)
+
+cpm_get_shortest_hash(${tmp} "cccb77ae9608d2768ed80dd42cec54f77b1f1455" hash)
+assert_equal(${hash} "cccb77ae9608")
+assert_contents_equal(${tmp}/cccb77ae9608.hash cccb77ae9608d2768ed80dd42cec54f77b1f1455)
+
+cpm_get_shortest_hash(${tmp} "cccb77be9609d2768ed80dd42cec54f77b1f1456" hash)
+assert_equal(${hash} "cccb77be")
+assert_contents_equal(${tmp}/cccb77be.hash cccb77be9609d2768ed80dd42cec54f77b1f1456)
+
+# 1. Cleanup - remove the temporary directory that we created
+
+file(REMOVE_RECURSE ${tmp})