From 623be1bc8b4b39b38405782219d2a8caae68c635 Mon Sep 17 00:00:00 2001
From: supermario94123 <supermario94123@gmail.com>
Date: Thu, 8 Aug 2019 15:54:00 +0200
Subject: [PATCH 1/2] cli interface test for -p and eval_all=True and
 eval_forever=True

---
 tests/test_edflow.py | 56 +++++++++++++++++++++++++++++++++++++++++---
 1 file changed, 53 insertions(+), 3 deletions(-)

diff --git a/tests/test_edflow.py b/tests/test_edflow.py
index 987beef..9f681d0 100644
--- a/tests/test_edflow.py
+++ b/tests/test_edflow.py
@@ -251,9 +251,11 @@ def test_3(self, tmpdir):
         assert any(list(filter(lambda x: "test_inference" in x, eval_dirs)))
 
     def test_4(self, tmpdir):
-        """
-        Tests evaluation with providing a checkpoint and using eval_all=True and eval_forever=True.
-        This should load not load any checkpoint.
+        """Tests evaluation with
+        1. providing a checkpoint
+        2. and using eval_all=True and eval_forever=True.
+
+        This should disable overwrite eval_all and eval_forever to ``False``, and then load the specified checkpoint
 
         effectively runs
             edflow -e config.yaml -b config.yaml -c logs/trained_model/train/checkpoints/model.ckpt-0
@@ -300,3 +302,51 @@ def test_4(self, tmpdir):
         # check if correct folder was created
         eval_dirs = os.listdir(os.path.join(tmpdir, "logs", "trained_model", "eval"))
         assert any(list(filter(lambda x: "test_inference" in x, eval_dirs)))
+
+    def test_5(self, tmpdir):
+        """Tests evaluation with
+        1. providing a project
+        1. using eval_all=True and eval_forever=True.
+
+        This should NOT load any checkpoint.
+
+        effectively runs
+            edflow -e config.yaml -b config.yaml -p logs/trained_model -n test_inference
+
+        and then checks if an evaluation folder "test_inference" was created in logs/trained_model/eval
+        -------
+        """
+        self.setup_tmpdir(tmpdir)
+        # command = "edflow -e eval.yaml -b train.yaml -n test"
+        config = dict()
+        config["model"] = "tests." + fullname(Model)
+        config["iterator"] = "tests." + fullname(Iterator_no_checkpoint)
+        config["dataset"] = "tests." + fullname(Dataset)
+        config["batch_size"] = 16
+        config["num_steps"] = 100
+        config["eval_all"] = True
+        config["eval_forever"] = False
+        import yaml
+
+        with open(os.path.join(tmpdir, "config.yaml"), "w") as outfile:
+            yaml.dump(config, outfile, default_flow_style=False)
+        import shutil
+
+        shutil.copytree(os.path.split(__file__)[0], os.path.join(tmpdir, "tests"))
+        command = [
+            "edflow",
+            "-e",
+            "config.yaml",
+            "-p",
+            os.path.join("logs", "trained_model"),
+            "-b",
+            "config.yaml",
+            "-n",
+            "test_inference",
+        ]
+        command = " ".join(command)
+        run_edflow_cmdline(command, cwd=tmpdir)
+
+        # check if correct folder was created
+        eval_dirs = os.listdir(os.path.join(tmpdir, "logs", "trained_model", "eval"))
+        assert any(list(filter(lambda x: "test_inference" in x, eval_dirs)))

From aa569d3745dec9bc36ae222cf70a7acbd159e3b3 Mon Sep 17 00:00:00 2001
From: supermario94123 <supermario94123@gmail.com>
Date: Thu, 8 Aug 2019 16:31:50 +0200
Subject: [PATCH 2/2] fixes -p with eval_all=True and eval_forever=True

---
 edflow/edflow | 39 +++++++++++++++++++++++++++------------
 1 file changed, 27 insertions(+), 12 deletions(-)

diff --git a/edflow/edflow b/edflow/edflow
index cb1604b..b296a24 100644
--- a/edflow/edflow
+++ b/edflow/edflow
@@ -115,28 +115,43 @@ def main(opt, additional_kwargs):
     # Evaluation
     opt.eval = opt.eval or list()
     for eval_idx, eval_config in enumerate(opt.eval):
-        if opt.checkpoint is not None:
-            checkpoint = opt.checkpoint
-        elif opt.project is not None:
-            checkpoint = get_latest_checkpoint(P.checkpoints)
-        else:
-            checkpoint = None
-
         base_config = {}
         if opt.base is not None:
             for base in opt.base:
                 with open(base) as f:
                     base_config.update(yaml.full_load(f))
-        # get path to implementation
         with open(eval_config) as f:
             config = base_config
             config.update(yaml.full_load(f))
             update_config(config, additional_kwargs)
+
+        disable_eval_all = disable_eval_forever = False
         if opt.checkpoint is not None:
-            for k in ["eval_forever", "eval_all"]:
-                if k in config:
-                    config[k] = False
-                    logger.info("{} was disabled because you specified a checkpoint.".format(k))
+            checkpoint = opt.checkpoint
+            disable_eval_all = disable_eval_forever = True
+        elif opt.project is not None:
+            if any([config.get("eval_all", False), config.get("eval_forever", False)]):
+                checkpoint = None
+            else:
+                checkpoint = get_latest_checkpoint(P.checkpoints)
+                disable_eval_all = disable_eval_forever = True
+        else:
+            checkpoint = None
+
+        if disable_eval_all:
+            config.update({"eval_all": False})
+            logger.info(
+                "{} was disabled because you specified a checkpoint.".format("eval_all")
+            )
+
+        if disable_eval_forever:
+            config.update({"eval_forever": False})
+            logger.info(
+                "{} was disabled because you specified a checkpoint.".format(
+                    "eval_forever"
+                )
+            )
+
         logger.info(
             "Evaluation config: {}\n{}".format(eval_config, yaml.dump(eval_config))
         )