From f80b83bdea94a1f5c23a93ba9c987a496dd9f22f Mon Sep 17 00:00:00 2001
From: paulzierep <paul.zierep@googlemail.com>
Date: Tue, 5 Nov 2024 10:06:03 +0100
Subject: [PATCH] catch workflowhub error

---
 sources/README.md                       |  4 ++--
 sources/bin/extract_galaxy_workflows.py | 13 ++++++++++---
 sources/bin/shared.py                   | 14 +++++++++++---
 3 files changed, 23 insertions(+), 8 deletions(-)

diff --git a/sources/README.md b/sources/README.md
index 52fd7657..6d353ea7 100644
--- a/sources/README.md
+++ b/sources/README.md
@@ -245,13 +245,13 @@ To make a test run of the tool to check its functionalities follow [Usage](#Usag
 1. Workflow extraction
 
     ```bash
-    $ bash bin/extract_all_workflows.sh test 
+    $ bash sources/bin/extract_all_workflows.sh test 
     ```
 
 2. Workflow filtering
 
     ```bash
-    $ bash bin/get_community_workflowss.sh test
+    $ bash sources/bin/get_community_workflows.sh test
     ```
 
 # Galaxy Labs framework
diff --git a/sources/bin/extract_galaxy_workflows.py b/sources/bin/extract_galaxy_workflows.py
index 4ff03f75..cbde094f 100644
--- a/sources/bin/extract_galaxy_workflows.py
+++ b/sources/bin/extract_galaxy_workflows.py
@@ -156,9 +156,10 @@ def add_workflows_from_workflowhub(self) -> None:
                 f"https://workflowhub.eu{wf['links']['self']}",
                 header,
             )
-            wf = Workflow()
-            wf.init_from_search(wf=wfhub_wf, source="WorkflowHub", tools=self.tools)
-            self.workflows.append(wf)
+            if wfhub_wf:
+                wf = Workflow()
+                wf.init_from_search(wf=wfhub_wf, source="WorkflowHub", tools=self.tools)
+                self.workflows.append(wf)
         print(len(self.workflows))
 
     def add_workflows_from_a_server(self, server: str) -> None:
@@ -170,6 +171,12 @@ def add_workflows_from_a_server(self, server: str) -> None:
             f"{server}/api/workflows/",
             header,
         )
+
+        # test max 50 wfs
+        if self.test:
+            if len(server_wfs) > 50:
+                server_wfs = server_wfs[:50]
+
         count = 0
         for wf in server_wfs:
             if wf["published"] and wf["importable"] and not wf["deleted"] and not wf["hidden"]:
diff --git a/sources/bin/shared.py b/sources/bin/shared.py
index 0097bcd8..21c9c108 100644
--- a/sources/bin/shared.py
+++ b/sources/bin/shared.py
@@ -96,9 +96,17 @@ def get_request_json(url: str, headers: dict) -> dict:
 
     :param url: galaxy tool id
     """
-    r = requests.get(url, auth=None, headers=headers)
-    r.raise_for_status()
-    return r.json()
+
+    try:
+        r = requests.get(url, auth=None, headers=headers)
+        r.raise_for_status()  # Optional: Raises an HTTPError if the response was an unsuccessful status code
+    except ConnectionError as e:
+        raise ConnectionError("Connection aborted: Remote end closed connection without response") from e
+    except requests.exceptions.RequestException as e:
+        # This will handle other possible errors (e.g., timeout, HTTPError)
+        raise SystemExit(e)
+    else:
+        return r.json()
 
 
 def format_date(date: str) -> str: