Merge branch 'release_24.1' into dev

lib/galaxy/datatypes/data.py

@@ -484,16 +484,17 @@ def _serve_file_download(self, headers, data, trans, to_ext, file_size, **kwd):
 
     def _serve_binary_file_contents_as_text(self, trans, data, headers, file_size, max_peek_size):
         headers["content-type"] = "text/html"
-        return (
-            trans.fill_template_mako(
-                "/dataset/binary_file.mako",
-                data=data,
-                file_contents=open(data.get_file_name(), "rb").read(max_peek_size),
-                file_size=util.nice_size(file_size),
-                truncated=file_size > max_peek_size,
-            ),
-            headers,
-        )
+        with open(data.get_file_name(), "rb") as fh:
+            return (
+                trans.fill_template_mako(
+                    "/dataset/binary_file.mako",
+                    data=data,
+                    file_contents=fh.read(max_peek_size),
+                    file_size=util.nice_size(file_size),
+                    truncated=file_size > max_peek_size,
+                ),
+                headers,
+            )
 
     def _serve_file_contents(self, trans, data, headers, preview, file_size, max_peek_size):
         from galaxy.datatypes import images
@@ -502,16 +503,17 @@ def _serve_file_contents(self, trans, data, headers, preview, file_size, max_pee
         if not preview or isinstance(data.datatype, images.Image) or file_size < max_peek_size:
             return self._yield_user_file_content(trans, data, data.get_file_name(), headers), headers
 
-        # preview large text file
-        headers["content-type"] = "text/html"
-        return (
-            trans.fill_template_mako(
-                "/dataset/large_file.mako",
-                truncated_data=open(data.get_file_name(), "rb").read(max_peek_size),
-                data=data,
-            ),
-            headers,
-        )
+        with compression_utils.get_fileobj(data.get_file_name(), "rb") as fh:
+            # preview large text file
+            headers["content-type"] = "text/html"
+            return (
+                trans.fill_template_mako(
+                    "/dataset/large_file.mako",
+                    truncated_data=fh.read(max_peek_size),
+                    data=data,
+                ),
+                headers,
+            )
 
     def display_data(
         self,

lib/galaxy/datatypes/tabular.py

@@ -197,14 +197,15 @@ def display_data(
                 return open(dataset.get_file_name(), mode="rb"), headers
             else:
                 headers["content-type"] = "text/html"
-                return (
-                    trans.fill_template_mako(
-                        "/dataset/large_file.mako",
-                        truncated_data=open(dataset.get_file_name()).read(max_peek_size),
-                        data=dataset,
-                    ),
-                    headers,
-                )
+                with compression_utils.get_fileobj(dataset.get_file_name(), "rb") as fh:
+                    return (
+                        trans.fill_template_mako(
+                            "/dataset/large_file.mako",
+                            truncated_data=fh.read(max_peek_size),
+                            data=dataset,
+                        ),
+                        headers,
+                    )
         else:
             column_names = "null"
             if dataset.metadata.column_names:

lib/galaxy/managers/hdas.py

@@ -311,7 +311,10 @@ def text_data(self, hda, preview=True):
 
         truncated = preview and os.stat(file_path).st_size > MAX_PEEK_SIZE
         with get_fileobj(file_path) as fh:
-            hda_data = fh.read(MAX_PEEK_SIZE)
+            try:
+                hda_data = fh.read(MAX_PEEK_SIZE)
+            except UnicodeDecodeError:
+                raise exceptions.RequestParameterInvalidException("Cannot generate text preview for dataset.")
         return truncated, hda_data
 
     # .... annotatable

lib/galaxy/managers/users.py

@@ -611,8 +611,9 @@ def send_reset_email(self, trans, payload, **kwd):
                 except Exception as e:
                     log.debug(body)
                     return f"Failed to submit email. Please contact the administrator: {util.unicodify(e)}"
-            else:
-                return "Failed to produce password reset token. User not found."
+        if not reset_user:
+            log.warning(f"Failed to produce password reset token. User with email '{email}' not found.")
+        return None
 
     def get_reset_token(self, trans, email):
         reset_user = get_user_by_email(trans.sa_session, email, self.app.model.User)

lib/galaxy/tool_util/biotools/source.py

@@ -55,12 +55,13 @@ def __init__(self, cache=None):
 
     def _raw_get_metadata(self, biotools_reference) -> Optional[str]:
         api_url = f"https://bio.tools/api/tool/{biotools_reference}?format=json"
-        req = requests.get(api_url, timeout=DEFAULT_SOCKET_TIMEOUT)
-        req.encoding = req.apparent_encoding
-        if req.status_code == 404:
-            return None
-        else:
+        try:
+            req = requests.get(api_url, timeout=DEFAULT_SOCKET_TIMEOUT)
+            req.raise_for_status()
+            req.encoding = req.apparent_encoding
             return req.text
+        except Exception:
+            return None
 
     def get_biotools_metadata(self, biotools_reference: str) -> Optional[BiotoolsEntry]:
         createfunc = functools.partial(self._raw_get_metadata, biotools_reference)

lib/galaxy/tools/parameters/basic.py

@@ -2053,7 +2053,8 @@ def src_id_to_item(
         item = sa_session.get(src_to_class[value["src"]], decoded_id)
     except KeyError:
         raise ValueError(f"Unknown input source {value['src']} passed to job submission API.")
-    assert item
+    if not item:
+        raise ValueError("Invalid input id passed to job submission API.")
     item.extra_params = {k: v for k, v in value.items() if k not in ("src", "id")}
     return item
 

lib/galaxy/web/framework/middleware/statsd.py

@@ -28,8 +28,13 @@ def __call__(self, environ, start_response):
         start_time = time.time()
         req = self.application(environ, start_response)
         dt = int((time.time() - start_time) * 1000)
-        page = environ.get("controller_action_key", None) or environ.get("PATH_INFO", "NOPATH").strip("/").replace(
-            "/", "."
+        page = (
+            environ.get("controller_action_key", None)
+            or environ.get("PATH_INFO", "NOPATH")
+            .strip("/")
+            .replace("/", ".")
+            .encode("ascii", errors="replace")
+            .decode()
         )
         self.galaxy_stasd_client.timing(page, dt)
         try:

lib/galaxy/webapps/galaxy/controllers/user.py

@@ -342,7 +342,7 @@ def reset_password(self, trans, payload=None, **kwd):
         payload = payload or {}
         if message := self.user_manager.send_reset_email(trans, payload):
             return self.message_exception(trans, message)
-        return {"message": "Reset link has been sent to your email."}
+        return {"message": "If an account exists for this email address a confirmation email will be dispatched."}
 
     def __get_redirect_url(self, redirect):
         if not redirect or redirect == "None":

lib/galaxy/workflow/modules.py

@@ -746,7 +746,8 @@ def get_all_outputs(self, data_only=False):
                         # This can happen when importing workflows with missing tools.
                         # We can't raise an exception here, as that would prevent loading
                         # the workflow.
-                        log.error(
+                        # This is also listed when opening such a workflow in the workflow editor.
+                        log.warning(
                             f"Workflow output '{workflow_output['output_name']}' defined, but not listed among data outputs"
                         )
                         continue

test/integration/objectstore/_purged_handling.py

@@ -9,7 +9,7 @@ def purge_while_job_running(dataset_populator: DatasetPopulator, extra_sleep=0):
         response = dataset_populator.run_tool(
             "all_output_types",
             inputs={
-                "sleep_param": 5,
+                "sleep_param": 5 + extra_sleep,
             },
             history_id=history_id,
         )

test/integration/test_extended_metadata.py

@@ -100,7 +100,7 @@ def test_fetch_data_library(self):
     def test_purge_while_job_running(self):
         # pass extra_sleep, since templating the command line will fail if the output
         # is deleted before remote_tool_eval runs.
-        purge_while_job_running(self.dataset_populator, extra_sleep=4)
+        purge_while_job_running(self.dataset_populator, extra_sleep=10)
 
 
 class TestExtendedMetadataDeferredIntegration(integration_util.IntegrationTestCase):

test/unit/app/managers/test_UserManager.py

@@ -240,7 +240,7 @@ def test_reset_email_user_deleted(self):
         self.user_manager.delete(user)
         assert user.deleted is True
         message = self.user_manager.send_reset_email(self.trans, {"email": user_email})
-        assert message == "Failed to produce password reset token. User not found."
+        assert message is None
 
     def test_get_user_by_identity(self):
         # return None if username/email not found