From af21d6c2b125c22ef3dbc36a0a6a67a1874549c7 Mon Sep 17 00:00:00 2001
From: Sven Twardziok <sven.twardziok@posteo.de>
Date: Thu, 12 Aug 2021 11:50:32 +0200
Subject: [PATCH] fix: Ga4gh tes bugfixes (#1127)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

* add py-tes

* check remote log and benchmark

* reduce complexity

* bugfix

* add exception for missing tes module

* add install py-tes info to doc

* add raise exception

* formatting

* remove py-tes

Co-authored-by: Johannes Köster <johannes.koester@tu-dortmund.de>
---
 docs/executing/cloud.rst         |  9 +++++++-
 snakemake/executors/ga4gh_tes.py | 39 ++++++++++++++------------------
 2 files changed, 25 insertions(+), 23 deletions(-)

diff --git a/docs/executing/cloud.rst b/docs/executing/cloud.rst
index 79352175d..ddb9d1a39 100644
--- a/docs/executing/cloud.rst
+++ b/docs/executing/cloud.rst
@@ -350,7 +350,7 @@ To install and configure Funnel follow its official `documentation <https://ohsu
 Configuration
 ~~~~~~~~~~~~~
 
-Two steps are required to make a Snakemake workflow TES ready:
+Three steps are required to make a Snakemake workflow TES ready:
 
 **Attach conda to rules:**
 Execution of Snakemake tasks via TES means, Snakemake is running in a container in the cloud and it executes a specific rule (or a group of rules) with defined input/output data.
@@ -365,6 +365,13 @@ The TES module requires using a remote file storage system for input/output file
 There are several options available in Snakemake to use remote files.
 This guide recommends to use S3 (or SWIFT) object storage.
 
+**Install py-tes module:**
+TES backend requires py-tes to be installed. Please install py-tes, e.g. via Conda or Pip.
+
+.. code-block:: console
+
+    $ pip install py-tes 
+
 Execution
 ~~~~~~~~~
 
diff --git a/snakemake/executors/ga4gh_tes.py b/snakemake/executors/ga4gh_tes.py
index e9f40453d..5675ee400 100644
--- a/snakemake/executors/ga4gh_tes.py
+++ b/snakemake/executors/ga4gh_tes.py
@@ -37,7 +37,12 @@ def __init__(
         tes_url=None,
         container_image=None,
     ):
-        import tes
+        try:
+            import tes
+        except ImportError:
+            raise WorkflowError(
+                "Unable to import Python package tes. TES backend requires py-tes to be installed. Please install py-tes, e.g. via Conda or Pip."
+            )
 
         self.container_image = container_image or get_container_image()
         self.container_workdir = "/tmp"
@@ -310,39 +315,29 @@ def _get_task_inputs(self, job, jobscript, checkdir):
 
         return inputs
 
-    def _get_task_outputs(self, job, checkdir):
-        outputs = []
-        # add output files to outputs
-        for o in job.output:
+    def _append_task_outputs(self, outputs, files):
+        for file in files:
             obj = self._prepare_file(
-                filename=o,
+                filename=file,
                 checkdir=checkdir,
                 type="Output",
             )
             if obj:
                 outputs.append(obj)
+        return outputs
+
+    def _get_task_outputs(self, job, checkdir):
+        outputs = []
+        # add output files to outputs
+        outputs = self._append_task_outputs(outputs, job.output)
 
         # add log files to outputs
         if job.log:
-            for log in job.log:
-                outputs.append(
-                    self._prepare_file(
-                        filename=log,
-                        checkdir=checkdir,
-                        type="Output",
-                    )
-                )
+            outputs = self._append_task_outputs(outputs, job.log)
 
         # add benchmark files to outputs
         if hasattr(job, "benchmark") and job.benchmark:
-            for benchmark in job.benchmark:
-                outputs.append(
-                    self._prepare_file(
-                        filename=benchmark,
-                        checkdir=checkdir,
-                        type="Output",
-                    )
-                )
+            outputs = self._append_task_outputs(outputs, job.benchmark)
 
         return outputs