From b08c4ed9457edb22171322652322c5c7f93a7fa3 Mon Sep 17 00:00:00 2001
From: Charles Kawczynski <kawczynski.charles@gmail.com>
Date: Thu, 1 Feb 2024 09:03:42 -0500
Subject: [PATCH] Extract nc files from tar for repro tests

---
 examples/hybrid/driver.jl       |  5 +++++
 regression_tests/compute_mse.jl | 18 ++++++++++++++++++
 2 files changed, 23 insertions(+)

diff --git a/examples/hybrid/driver.jl b/examples/hybrid/driver.jl
index 70396873fb..36594cf271 100644
--- a/examples/hybrid/driver.jl
+++ b/examples/hybrid/driver.jl
@@ -226,6 +226,11 @@ if ClimaComms.iamroot(config.comms_ctx)
     @info "Plotting done"
 
     @info "Creating tarballs"
+    # These NC files are used by our reproducibility tests,
+    # and need to be found later when comparing against the
+    # main branch. If "nc_files.tar" is renamed, then please
+    # search for "nc_files.tar" globally and rename it in the
+    # reproducibility test folder.
     Tar.create(
         f -> endswith(f, ".nc"),
         simulation.output_dir,
diff --git a/regression_tests/compute_mse.jl b/regression_tests/compute_mse.jl
index 942a8f2a18..646b645bb6 100644
--- a/regression_tests/compute_mse.jl
+++ b/regression_tests/compute_mse.jl
@@ -1,5 +1,6 @@
 import NCRegressionTests
 import NCDatasets
+import Tar
 import ClimaCoreTempestRemap as CCTR
 
 include("self_reference_or_path.jl")
@@ -32,6 +33,23 @@ function regression_test(; job_id, reference_mse, ds_filename_computed, varname)
         ds_filename_reference = joinpath(path, ds_filename_computed)
         @info "`ds_filename_computed`: `$ds_filename_computed`"
         @info "`ds_filename_reference`: `$ds_filename_reference`"
+        if !isfile(ds_filename_reference)
+            # We may have converted to tarball, try to
+            # extract nc files from tarball first:
+            job_dir = dirname(ds_filename_reference)
+            mktempdir(joinpath(job_dir, tempdir())) do tdir
+                # We must extract to an empty folder, let's
+                # move it back to job_dir after.
+                Tar.extract(joinpath(job_dir, "nc_files.tar"), tdir) do hdr
+                    basename(hdr.path) == basename(ds_filename_reference)
+                end
+                mv(
+                    joinpath(tdir, basename(ds_filename_reference)),
+                    joinpath(job_dir, basename(ds_filename_reference));
+                    force = true,
+                )
+            end
+        end
         if !isfile(ds_filename_reference)
             msg = "\n\n"
             msg *= "Pull request author:\n"