Hide pytest from code snippets (dask#8198)

distributed/client.py

@@ -3022,6 +3022,7 @@ def _get_computation_code(
         """
         if nframes <= 0:
             return ()
+
         ignore_modules = dask.config.get(
             "distributed.diagnostics.computations.ignore-modules"
         )
@@ -3030,10 +3031,26 @@ def _get_computation_code(
                 "Ignored modules must be a list. Instead got "
                 f"({type(ignore_modules)}, {ignore_modules})"
             )
-        pattern: re.Pattern | None = None
+        ignore_files = dask.config.get(
+            "distributed.diagnostics.computations.ignore-files"
+        )
+        if not isinstance(ignore_files, list):
+            raise TypeError(
+                "Ignored files must be a list. Instead got "
+                f"({type(ignore_files)}, {ignore_files})"
+            )
+
+        mod_pattern: re.Pattern | None = None
+        fname_pattern: re.Pattern | None = None
         if stacklevel is None:
             if ignore_modules:
-                pattern = re.compile("|".join([f"(?:{mod})" for mod in ignore_modules]))
+                mod_pattern = re.compile(
+                    "|".join([f"(?:{mod})" for mod in ignore_modules])
+                )
+            if ignore_files:
+                fname_pattern = re.compile(
+                    r".*[\\/](" + "|".join(mod for mod in ignore_files) + r")([\\/]|$)"
+                )
         else:
             # stacklevel 0 or less - shows dask internals which likely isn't helpful
             stacklevel = stacklevel if stacklevel > 0 else 1
@@ -3044,12 +3061,13 @@ def _get_computation_code(
         ):
             if len(code) >= nframes:
                 break
-            elif stacklevel is not None and i != stacklevel:
+            if stacklevel is not None and i != stacklevel:
                 continue
-            elif pattern is not None and (
-                pattern.match(fr.f_globals.get("__name__", ""))
-                or fr.f_code.co_name in ("<listcomp>", "<dictcomp>")
-            ):
+            if fr.f_code.co_name in ("<listcomp>", "<dictcomp>"):
+                continue
+            if mod_pattern and mod_pattern.match(fr.f_globals.get("__name__", "")):
+                continue
+            if fname_pattern and fname_pattern.match(fr.f_code.co_filename):
                 continue
 
             kwargs = dict(

distributed/distributed-schema.yaml

@@ -1020,6 +1020,12 @@ properties:
                   A list of modules which are ignored when trying to collect the
                   code context when submitting a computation. Accepts regular
                   expressions.
+              ignore-files:
+                type: array
+                description: |
+                  A list of files and directories which are ignored when trying to
+                  collect the code context when submitting a computation. Accepts
+                  regular expressions.
           erred-tasks:
             type: object
             properties:

distributed/distributed.yaml

@@ -273,15 +273,27 @@ distributed:
       ignore-modules:
         - asyncio
         - functools
+        - datashader
         - dask
+        - debugpy
         - distributed
         - coiled
         - cudf
         - cuml
+        - pluggy  # part of pytest
         - prefect
         - rechunker
         - xarray
         - xgboost
+      ignore-files:
+        - runpy\.py  # `python -m pytest` (or other module) shell command
+        - pytest  # `pytest` shell command
+        - py\.test  # `py.test` shell command
+        - pytest-script\.py  # `pytest` shell command in Windows
+        - _pytest  # pytest implementation
+        - pycharm  # Run pytest from PyCharm GUI
+        - vscode_pytest
+        - get_output_via_markers\.py
     erred-tasks:
       max-history: 100
 

distributed/tests/test_client.py

@@ -7314,7 +7314,10 @@ def nested_call():
             code = Client._get_computation_code()
 
     with dask.config.set(
-        {"distributed.diagnostics.computations.ignore-modules": ["test_client"]}
+        {
+            "distributed.diagnostics.computations.ignore-modules": ["test_client"],
+            "distributed.diagnostics.computations.ignore-files": [],
+        }
     ):
         import sys
 
@@ -7413,10 +7416,8 @@ def fetch_comp_code(dask_scheduler):
         return comp.code[0]
 
     code = client.run_on_scheduler(fetch_comp_code)
-
-    assert len(code) == 2
-    assert code[-1].code == test_function_code
-    assert code[-2].code == inspect.getsource(sys._getframe(1))
+    assert len(code) == 1
+    assert code[0].code == test_function_code
 
 
 def test_computation_object_code_dask_compute_no_frames_default(client):