robotframework-parallelrunner 0.3.0__py3-none-any.whl → 0.4.0__py3-none-any.whl

ParallelRunner/__init__.py

@@ -13,7 +13,7 @@ Keyword documentation:
 https://cristiangarciavd.github.io/robotframework-parallelrunner/ParallelRunner.html
 """
 
-from .parallel_library import ParallelLibrary, ParallelRunner, ParallelTaskError
+from .parallel_library import ParallelLibrary, ParallelRunner, ParallelTaskError, log_to_robot
 from .version import __version__
 
-__all__ = ["ParallelRunner", "ParallelLibrary", "ParallelTaskError", "__version__"]
+__all__ = ["ParallelRunner", "ParallelLibrary", "ParallelTaskError", "log_to_robot", "__version__"]

ParallelRunner/parallel_library.py

@@ -1,13 +1,85 @@
 import os
-import concurrent.futures
 import importlib
+import inspect
+import queue
+import threading
+import time
 from typing import Iterable, Any, List, Dict, Optional, Tuple, Union
 from robot.api import logger
+from robot.errors import DataError
+try:
+    from robot.errors import TimeoutExceeded
+except ImportError:  # Robot Framework < 7.0
+    from robot.errors import TimeoutError as TimeoutExceeded
 from robot.libraries.BuiltIn import BuiltIn
+from robot.running.arguments import PythonArgumentParser
+from robot.utils import secs_to_timestr, timestr_to_secs
 
 from .version import __version__
 
 
+# Robot Framework log levels, in increasing severity. HTML and CONSOLE are
+# INFO messages (rendered as HTML / also written to the console).
+LOG_LEVELS = {"TRACE": 0, "DEBUG": 1, "INFO": 2, "HTML": 2, "CONSOLE": 2, "WARN": 3, "ERROR": 4}
+# Spellings from Python's `logging` module, mapped to the Robot level.
+LOG_LEVEL_ALIASES = {"WARNING": "WARN", "CRITICAL": "ERROR", "FATAL": "ERROR"}
+
+# How often the main thread checks whether the calls have finished. Waiting in
+# short sleeps (instead of blocking on a lock until the end) is what lets a
+# Robot `[Timeout]` interrupt the keyword: on Windows the timeout is raised
+# asynchronously and only arrives while Python code runs.
+_POLL_INTERVAL = 0.01
+
+
+def _normalize_level(level: Any) -> Optional[str]:
+    """Returns the Robot level name for ``level`` (any case, or a Python
+    ``logging`` alias), ``"IGNORE"``, or None if it's not a known level."""
+    name = str(level).strip().upper()
+    name = LOG_LEVEL_ALIASES.get(name, name)
+    return name if name in LOG_LEVELS or name == "IGNORE" else None
+
+
+def _normalize_name(name: str) -> str:
+    """Keyword name comparison the way Robot does it: ignoring case, spaces and underscores."""
+    return name.lower().replace(" ", "").replace("_", "")
+
+
+def _not_given(value: Any) -> bool:
+    """None, an empty string or "None" (any case - Robot's usual spelling of None)."""
+    return value is None or (isinstance(value, str) and value.strip().upper() in ("", "NONE"))
+
+
+# Who owns the result of a call that may time out: the worker that ran it, or
+# the main thread that timed it out. See `ParallelRunner._run_in_threads`.
+_WORKER, _MAIN = "worker", "main"
+_NOT_SET = object()
+
+
+def log_to_robot(msg: Any, level: str = "INFO") -> None:
+    """Logs ``msg`` straight to Robot Framework, accepting the same levels as
+    the ``_logger`` that `Run Parallel Scenarios` injects.
+
+    Use it as the fallback when a parallel-ready keyword is called directly
+    (without `Run Parallel Scenarios`)::
+
+        from ParallelRunner import log_to_robot
+
+        def verify_agent(self, agent_id, _logger=None):
+            log = _logger or log_to_robot
+            log("details", "DEBUG")
+
+    Unlike ``robot.api.logger.write``, it accepts ``IGNORE`` (dropped), any
+    case and Python ``logging`` names (``WARNING``), and logs a message with an
+    unknown level at INFO, prefixed with the level, instead of failing.
+    """
+    name = _normalize_level(level)
+    if name == "IGNORE":
+        return
+    if name is None:
+        name, msg = "INFO", f"[{level}] {msg}"
+    logger.write(msg, name)
+
+
 class ParallelTaskError(RuntimeError):
     """
     Raised when at least one parallel task failed: by ``run_parallel_scenarios``
@@ -47,7 +119,8 @@ class ParallelRunner:
 
     = How it works =
 
-    - Each call runs on a worker thread of a ``ThreadPoolExecutor``.
+    - Each call runs on one of a pool of worker threads (see
+      ``ROBOT_THREAD_WORKERS``), created for that keyword call.
       ``BuiltIn().run_keyword`` is *not* used inside threads (it is not
       thread-safe); the underlying Python method of the target library is
       called directly instead.
@@ -65,18 +138,33 @@ class ParallelRunner:
     index) as its first argument and an injected ``_logger`` callable that
     it should use instead of ``robot.api.logger``:
 
-    | from robot.api import logger
+    | from ParallelRunner import log_to_robot
     |
     | class MyLibrary:
-    |     def verify_agent(self, agent_id, _logger=None, **kwargs):
-    |         log = _logger or (lambda msg, level="INFO": logger.write(msg, level))
+    |     def verify_agent(self, agent_id: int, timeout: float = 10, _logger=None, **kwargs):
+    |         log = _logger or log_to_robot
     |         log(f"Checking agent {agent_id}")
     |         ...
     |         return result
 
-    ``_logger(msg, level)`` accepts the levels ``INFO``, ``WARN``, ``ERROR``
-    and ``IGNORE`` (dropped). Any extra named argument given to
-    `Run Parallel Scenarios` is forwarded to every call as ``**kwargs``.
+    ``_logger(msg, level)`` accepts every Robot Framework log level (see
+    `Log levels`) plus ``IGNORE`` (dropped). ``log_to_robot`` is the fallback
+    for when the keyword is called directly: it accepts the same levels
+    (``robot.api.logger.write`` fails on ``IGNORE`` or ``WARNING``).
+
+    Any extra named argument given to `Run Parallel Scenarios` is forwarded
+    to every call as ``**kwargs``. Arguments are converted by the method's
+    type hints (and default values) exactly as when the keyword is called
+    directly: ``timeout=5`` arrives as the number ``5.0`` above. Items that
+    are strings are converted the same way; other items (numbers, lists,
+    dictionaries from a variable) are passed as they are.
+
+    The keyword is found by its method name (``Verify Agent`` ->
+    ``verify_agent``) or by a custom name given with ``@keyword("...")``,
+    ignoring case, spaces and underscores, like Robot does.
+
+    To be stoppable when it times out, the method can also declare a
+    ``_cancelled`` parameter: see `Call timeouts and cancellation`.
 
     Logging is made thread-safe for you; your keyword's own side effects are
     not. Protect shared state (files, shared objects) with a lock.
@@ -90,7 +178,7 @@ class ParallelRunner:
     | item     | The item (or repeat index) the call received. |
     | logs     | List of ``(level, message)`` tuples captured during the call. |
     | result   | The return value of the call (only when ``status`` is ``PASS``). |
-    | error    | The error message (only when ``status`` is ``FAIL``). |
+    | error    | The error message (only when ``status`` is ``FAIL``), e.g. ``Timed out after 10 seconds`` with ``call_timeout``. |
 
     = Failures =
 
@@ -111,6 +199,78 @@ class ParallelRunner:
     ``return_values_only=True`` and `Get Result Values` always fail if any
     call failed, because a failed call has no return value.
 
+    = Log levels =
+
+    The injected ``_logger`` accepts the same levels as Robot Framework's
+    ``logger.write``: ``TRACE``, ``DEBUG``, ``INFO``, ``HTML`` (an ``INFO``
+    message rendered as HTML), ``CONSOLE`` (an ``INFO`` message also written
+    to the console), ``WARN`` and ``ERROR``. Levels are case-insensitive, and
+    Python ``logging`` names are accepted too: ``WARNING`` is ``WARN``,
+    ``CRITICAL`` and ``FATAL`` are ``ERROR``. ``IGNORE`` drops the message.
+
+    Replayed messages go through Robot's normal filtering, so ``TRACE`` and
+    ``DEBUG`` messages only show up in ``log.html`` when the run uses a low
+    enough ``--loglevel`` (e.g. ``--loglevel DEBUG``), exactly as they would
+    from a normal keyword. ``thread_log_level`` filters the threads' messages
+    on top of that (by default it lets every level through).
+
+    A message with an unknown level is not lost: it is logged at ``INFO``
+    with the level as a prefix (``[VERBOSE] message``), and a warning names
+    the unknown level once per keyword call.
+
+    = Timeouts =
+
+    A Robot ``[Timeout]`` (test or keyword) interrupts `Run Parallel Scenarios`
+    as soon as it expires, not when the whole batch has finished. Before the
+    timeout failure is reported, the logs collected so far are replayed:
+    finished calls as usual, and every call that did not finish with its
+    partial logs and an ``ERROR`` saying it was interrupted.
+
+    Python threads can't be killed, so calls that are already running keep
+    running in the background until they return (possibly while the next test
+    runs) - unless they cooperate, see `Call timeouts and cancellation`.
+    Calls that had not started yet are not started. The worker threads are
+    daemon threads, so they never keep the Robot process alive after the run.
+    A keyword that may hang (e.g. network I/O) should set its own timeouts
+    (e.g. ``requests.get(url, timeout=10)``).
+
+    A Robot timeout fails the whole test and can't be handled with
+    `Run Keyword And Ignore Error`. To fail only the call that hangs, use
+    ``call_timeout``.
+
+    = Call timeouts and cancellation =
+
+    ``call_timeout`` limits how long *each call* may run, e.g.
+    ``call_timeout=10s`` (Robot's time format). A call still running after
+    that becomes a failed call - error ``Timed out after 10 seconds``, with its
+    partial logs - while the other calls keep running and finish normally. It
+    is then reported like any other failed call: `Run Parallel Scenarios`
+    fails with ``ParallelTaskError`` (which, unlike a Robot timeout, can be
+    handled), or returns it with ``status`` ``FAIL`` when
+    ``fail_on_error=False``.
+
+    | ${results}= | Run Parallel Scenarios | keyword=Verify Agent | library=MyLibrary | for_loop_iterable=${agents} | call_timeout=10s | fail_on_error=False |
+
+    The timed-out call can't be killed either: it keeps running in the
+    background, and a new worker thread takes its place so the remaining
+    calls still run ``ROBOT_THREAD_WORKERS`` at a time.
+
+    A keyword that can stop early declares a ``_cancelled`` parameter.
+    `Run Parallel Scenarios` then passes it a ``threading.Event`` that is set
+    when that call times out, or when a Robot ``[Timeout]`` interrupts the
+    batch. The keyword checks it between steps, or waits on it instead of
+    sleeping:
+
+    | def poll_job(self, job_id, _logger=None, _cancelled=None):
+    |     while not job_done(job_id):
+    |         if _cancelled is not None and _cancelled.wait(1):   # sleeps 1 s, wakes up if cancelled
+    |             _logger(f"Stopped waiting for {job_id}", "WARN")
+    |             return None
+
+    ``_cancelled`` is only passed to methods that declare it by that name, so
+    existing keywords keep working unchanged. It can't interrupt a call that
+    is blocked in a single operation (a network request without a timeout).
+
     = Custom logger mappers =
 
     A logger mapper replaces the default buffering logger: the target keyword
@@ -134,7 +294,7 @@ class ParallelRunner:
     = Environment variables =
 
     | =Variable=             | =Description= |
-    | ROBOT_THREAD_WORKERS   | Number of worker threads. Read when the library is imported. Default ``4``. |
+    | ROBOT_THREAD_WORKERS   | Number of worker threads. Read when the library is imported. Default ``4``. An invalid value is ignored with a warning. |
     | ROBOT_LOGGER_MAPPER    | Default ``logger_mapper`` as a ``module.function`` path, used when the argument is not given. See `Custom logger mappers`. |
 
     = When not to use it =
@@ -148,9 +308,29 @@ class ParallelRunner:
     ROBOT_LIBRARY_VERSION = __version__
     ROBOT_LIBRARY_DOC_FORMAT = 'ROBOT'
 
+    DEFAULT_WORKERS = 4
+
     def __init__(self):
-        # Workers count from Env Var or default to 4
-        self.max_workers = int(os.getenv("ROBOT_THREAD_WORKERS", "4"))
+        self.max_workers = self._workers_from_environment()
+
+    def _workers_from_environment(self) -> int:
+        """``ROBOT_THREAD_WORKERS``, or 4. An invalid value is ignored with a
+        warning (it used to make the library import fail with a bare
+        ``invalid literal for int()``)."""
+        value = os.getenv("ROBOT_THREAD_WORKERS")
+        if _not_given(value):
+            return self.DEFAULT_WORKERS
+        try:
+            workers = int(value)
+        except ValueError:
+            workers = 0
+        if workers < 1:
+            logger.warn(
+                f"Ignoring ROBOT_THREAD_WORKERS {value!r}: expected a whole number of at least 1. "
+                f"Using {self.DEFAULT_WORKERS} workers."
+            )
+            return self.DEFAULT_WORKERS
+        return workers
 
     def run_parallel_scenarios(
         self,
@@ -159,10 +339,11 @@ class ParallelRunner:
         for_loop_iterable: Optional[Iterable[Any]] = None,
         repeat: Optional[int] = None,
         remove_passing_logs: bool = False,
-        thread_log_level: str = "INFO",
+        thread_log_level: str = "TRACE",
         logger_mapper: Optional[Any] = None,
         return_values_only: bool = False,
         fail_on_error: bool = True,
+        call_timeout: Union[str, float, None] = None,
         **kwargs
     ) -> Union[List[Dict[str, Any]], Tuple[Any, ...]]:
         """Runs ``keyword`` from ``library`` concurrently, once per item or N times.
@@ -183,7 +364,9 @@ class ParallelRunner:
         - ``remove_passing_logs``: If true, logs of passing calls are not
           replayed into ``log.html``; only failed calls are shown.
         - ``thread_log_level``: Minimum level replayed from the threads:
-          ``INFO`` (default), ``WARN`` or ``ERROR``.
+          ``TRACE`` (default, every message), ``DEBUG``, ``INFO``, ``WARN``
+          or ``ERROR``. Robot's own ``--loglevel`` still applies on top of it
+          (see `Log levels`). An unknown value is ignored with a warning.
         - ``logger_mapper``: Optional callable ``mapper(msg, level)`` - or a
           ``module.function`` path to one, such as
           ``my_package.my_module.my_mapper`` - injected as ``_logger`` instead
@@ -199,7 +382,14 @@ class ParallelRunner:
           when any call failed, after all calls have finished and their logs
           have been replayed. If false, return the results anyway and leave
           checking ``status`` to the caller. See `Failures`.
-        - ``**kwargs``: Any other named argument is passed to every call.
+        - ``call_timeout``: Maximum time each call may run, in Robot's time
+          format (``10s``, ``1 minute 30 seconds``, ``2.5``). A call still
+          running after it fails with ``Timed out after ...`` while the other
+          calls keep running; calls that declare ``_cancelled`` are told to
+          stop. Not given (default): no limit. See
+          `Call timeouts and cancellation`.
+        - ``**kwargs``: Any other named argument is passed to every call,
+          converted by the method's type hints like in a direct call.
 
         Returns a list of result dictionaries (see `Result format`), in call
         order: entry ``i`` belongs to ``for_loop_iterable[i]`` / repeat
@@ -212,6 +402,7 @@ class ParallelRunner:
         | ${results}=   | Run Parallel Scenarios | keyword=Check Health      | library=MyLibrary       | repeat=8 | agent_id=1 |
         | ${id1}    ${id2}    ${id3}= | Run Parallel Scenarios | keyword=Seed Record | library=MyLibrary | repeat=3 | return_values_only=True |
         | ${results}=   | Run Parallel Scenarios | keyword=Verify Agent Data | library=MyLibrary       | for_loop_iterable=${agents} | fail_on_error=False |
+        | ${results}=   | Run Parallel Scenarios | keyword=Verify Agent Data | library=MyLibrary       | for_loop_iterable=${agents} | call_timeout=30s |
         """
         # Determine items to process. NOTE: `repeat or 1` would be wrong here -
         # 0 is falsy in Python, so an explicit repeat=0 would silently fall
@@ -229,25 +420,20 @@ class ParallelRunner:
         if effective_mapper is None:
             effective_mapper = self._load_mapper_from_environment()
 
-        with concurrent.futures.ThreadPoolExecutor(max_workers=self.max_workers) as executor:
-            # We map the execution.
-            # Note: We must call the underlying Python method, not BuiltIn().run_keyword
-            # because run_keyword is not thread-safe.
-            lib_instance = self._get_library_instance_owning_keyword(keyword, library)
-            method = getattr(lib_instance, keyword.replace(" ", "_").lower())
-
-            # Submit every task up front so they all start running concurrently,
-            # then collect results in submission order (NOT as_completed order) -
-            # tasks[i].result() blocks only until the i-th task finishes, it
-            # doesn't force tasks to run one at a time.
-            tasks = [
-                executor.submit(self._execute_and_capture, method, item, effective_mapper, **kwargs)
-                for item in items
-            ]
-            results = [task.result() for task in tasks]
+        min_level = self._resolve_thread_log_level(thread_log_level)
+        timeout_seconds = self._resolve_call_timeout(call_timeout)
+
+        # Note: We must call the underlying Python method, not BuiltIn().run_keyword
+        # because run_keyword is not thread-safe.
+        lib_instance = self._get_library_instance_owning_keyword(keyword, library)
+        method = self._get_keyword_method(lib_instance, keyword)
+
+        results = self._run_in_threads(
+            method, list(items), effective_mapper, remove_passing_logs, min_level, kwargs, timeout_seconds
+        )
 
         # Step: Re-play logs into Robot Framework sequentially
-        self._replay_logs(results, remove_passing_logs, thread_log_level)
+        self._replay_logs(results, remove_passing_logs, min_level)
 
         if return_values_only:
             return self.get_result_values(results)
@@ -337,50 +523,331 @@ class ParallelRunner:
         return self._resolve_mapper(os.getenv("ROBOT_LOGGER_MAPPER"), source="ROBOT_LOGGER_MAPPER")
 
     def _get_library_instance_owning_keyword(self, keyword_name: str, library: str) -> Any:
-        method_name = keyword_name.replace(" ", "_").lower()
         lib_instance = BuiltIn().get_library_instance(library)
-        if hasattr(lib_instance, method_name):
+        if self._find_keyword_method(lib_instance, keyword_name) is not None:
             return lib_instance
         raise ValueError(f"Keyword '{keyword_name}' not found in library '{library}'")
 
-    def _execute_and_capture(self, method, item, logger_mapper: Optional[Any] = None, **kwargs) -> Dict[str, Any]:
-        """Worker wrapper to capture logs and results."""
-        log_buffer = []
-        
+    def _get_keyword_method(self, lib_instance: Any, keyword_name: str):
+        method = self._find_keyword_method(lib_instance, keyword_name)
+        if method is None:
+            raise ValueError(f"Keyword '{keyword_name}' not found in library '{type(lib_instance).__name__}'")
+        return method
+
+    @staticmethod
+    def _find_keyword_method(lib_instance: Any, keyword_name: str):
+        """Finds the method implementing ``keyword_name``: by method name
+        (``Verify Agent Data`` -> ``verify_agent_data``), or, like Robot
+        itself, by a custom name given with ``@keyword("...")``. Names are
+        compared ignoring case, spaces and underscores."""
+        method = getattr(lib_instance, keyword_name.replace(" ", "_").lower(), None)
+        if callable(method):
+            return method
+        wanted = _normalize_name(keyword_name)
+        for attribute in dir(lib_instance):
+            # getattr_static: don't run properties just to look at them.
+            raw = inspect.getattr_static(lib_instance, attribute, None)
+            function = raw.__func__ if isinstance(raw, (staticmethod, classmethod)) else raw
+            robot_name = getattr(function, "robot_name", None)
+            if robot_name is not None:
+                matches = _normalize_name(robot_name) == wanted
+            else:
+                matches = not attribute.startswith("_") and _normalize_name(attribute) == wanted
+            if matches and callable(getattr(lib_instance, attribute, None)):
+                return getattr(lib_instance, attribute)
+        return None
+
+    def _resolve_thread_log_level(self, thread_log_level: Any) -> int:
+        """Returns the numeric minimum level for ``thread_log_level``. Not
+        given (None, empty, "None") means every level; an unknown value is
+        ignored with a warning instead of silently filtering by INFO."""
+        if _not_given(thread_log_level):
+            return LOG_LEVELS["TRACE"]
+        level = _normalize_level(thread_log_level)
+        if level in LOG_LEVELS:
+            return LOG_LEVELS[level]
+        logger.warn(
+            f"Ignoring thread_log_level {thread_log_level!r}: expected TRACE, DEBUG, INFO, WARN or ERROR. "
+            "Replaying every level."
+        )
+        return LOG_LEVELS["TRACE"]
+
+    @staticmethod
+    def _resolve_call_timeout(call_timeout: Any) -> Optional[float]:
+        """Returns ``call_timeout`` in seconds, or None when not given. Accepts
+        Robot's time format (``10s``, ``1 minute 30 seconds``, ``1.5``)."""
+        if _not_given(call_timeout):
+            return None
+        try:
+            seconds = timestr_to_secs(call_timeout)
+        except (ValueError, DataError) as error:
+            raise ValueError(f"Invalid call_timeout {call_timeout!r}: {error}") from None
+        if seconds <= 0:
+            raise ValueError(f"call_timeout must be positive, got {call_timeout!r}.")
+        return seconds
+
+    @staticmethod
+    def _converter_for(method):
+        """Returns a function converting ``(positional, named)`` arguments by
+        the method's type hints and default values, exactly like Robot does
+        when the keyword is called directly - or None if Robot can't parse the
+        method's signature (then arguments are passed as given)."""
+        try:
+            spec = PythonArgumentParser().parse(method)
+        except Exception:
+            return None
+        return lambda positional, named: spec.convert(positional, named)
+
+    @staticmethod
+    def _accepts_cancel_event(method) -> bool:
+        """``_cancelled`` is only passed to keywords that declare it, so that
+        keywords with a fixed signature keep working."""
+        try:
+            return "_cancelled" in inspect.signature(method).parameters
+        except (TypeError, ValueError):
+            return False
+
+    def _run_in_threads(
+        self,
+        method,
+        items: List[Any],
+        logger_mapper: Optional[Any],
+        remove_passing_logs: bool,
+        min_level: int,
+        kwargs: Dict[str, Any],
+        call_timeout: Optional[float] = None,
+    ) -> List[Dict[str, Any]]:
+        """
+        Runs ``method`` once per item on up to ``max_workers`` daemon threads
+        and returns the result dicts in call order (NOT completion order).
+
+        With ``call_timeout``, a call still running after that many seconds
+        becomes a ``FAIL`` result: its thread is abandoned (it can't be
+        killed) and a new worker takes its place, so the other calls keep
+        running ``max_workers`` at a time.
+
+        If the wait is interrupted (mostly by a Robot ``[Timeout]``), the
+        workers stop starting new calls, the logs collected so far are
+        replayed and the exception is re-raised. Calls already running can't
+        be stopped: they finish in the background. The threads are daemon
+        threads (unlike the ones of ``ThreadPoolExecutor``), so a call that
+        never returns can't keep the Robot process alive at exit.
+        """
+        if self.max_workers < 1:
+            raise ValueError(f"ROBOT_THREAD_WORKERS must be at least 1, got {self.max_workers}.")
+
+        count = len(items)
+        convert = self._converter_for(method)
+        if convert is not None:
+            # Once, before any thread starts: a bad named argument fails the
+            # keyword instead of every call.
+            kwargs = dict(convert([], list(kwargs.items()))[1])
+        cancel_events = [threading.Event() for _ in items] if self._accepts_cancel_event(method) else None
+
+        # Created here, not in the workers, so the logs of calls that are still
+        # running can be replayed if the keyword is interrupted or times out.
+        log_buffers: List[list] = [[] for _ in items]
+        # Written by the workers. A worker writes `started_at[i]` when call i
+        # starts and `finished[i]` when it has returned - unless the main
+        # thread has timed the call out first. `claims` decides which of the
+        # two wins: dict.setdefault is atomic, so the first claim sticks.
+        started_at: List[Optional[float]] = [None] * count
+        finished: List[Optional[Dict[str, Any]]] = [None] * count
+        claims: Dict[int, str] = {}
+        # Written by the main thread only.
+        results: List[Optional[Dict[str, Any]]] = [None] * count
+
+        # What each call receives as its first argument. Only strings are
+        # converted (by the type hint, as when Robot calls the keyword with a
+        # value from a test file); other objects are passed as they are. The
+        # results keep the original item either way.
+        arguments = list(items)
+        pending = queue.SimpleQueue()
+        for index, item in enumerate(items):
+            if convert is not None and isinstance(item, str):
+                try:
+                    (arguments[index],), _ = convert([item], [])
+                except Exception as error:
+                    # Not worth starting: Robot would fail this call before running it, too.
+                    message = str(error)
+                    log_buffers[index].append(("ERROR", f"Thread failed for item {item}: {message}"))
+                    started_at[index] = time.monotonic()
+                    finished[index] = {"status": "FAIL", "item": item, "logs": log_buffers[index], "error": message}
+                    continue
+            pending.put(index)
+        stop = threading.Event()
+        workers_started = 0
+
+        def worker():
+            while not stop.is_set():
+                try:
+                    index = pending.get_nowait()
+                except queue.Empty:
+                    return
+                call_kwargs = kwargs if cancel_events is None else {**kwargs, "_cancelled": cancel_events[index]}
+                started_at[index] = time.monotonic()
+                result = self._execute_and_capture(
+                    method, items[index], logger_mapper, log_buffers[index], call_kwargs, arguments[index]
+                )
+                if claims.setdefault(index, _WORKER) != _WORKER:
+                    # The call timed out and a new worker took this thread's
+                    # place: stop here, so no more than max_workers calls run.
+                    return
+                finished[index] = result
+
+        def start_worker():
+            nonlocal workers_started
+            workers_started += 1
+            threading.Thread(target=worker, name=f"ParallelRunner-worker-{workers_started}", daemon=True).start()
+
+        try:
+            for _ in range(min(self.max_workers, count)):
+                start_worker()
+            # Deliberately no lock, Event or Future wait here: a timeout raised
+            # asynchronously while such a wait holds an internal lock could
+            # leave it locked forever. Short sleeps and plain reads are safe.
+            remaining = count
+            next_unstarted = 0
+            running: List[int] = []
+            while remaining:
+                # Calls start in index order (FIFO queue), so started calls are
+                # picked up by moving a pointer instead of scanning every slot.
+                while next_unstarted < count and started_at[next_unstarted] is not None:
+                    running.append(next_unstarted)
+                    next_unstarted += 1
+                now = time.monotonic()
+                still_running = []
+                for index in running:
+                    if finished[index] is not None:
+                        results[index] = finished[index]
+                        remaining -= 1
+                    elif (
+                        call_timeout is not None
+                        and now - started_at[index] >= call_timeout
+                        and claims.setdefault(index, _MAIN) == _MAIN
+                    ):
+                        results[index] = self._timed_out_result(items[index], log_buffers[index], call_timeout)
+                        remaining -= 1
+                        if cancel_events is not None:
+                            cancel_events[index].set()
+                        if not pending.empty():
+                            start_worker()
+                    else:
+                        still_running.append(index)
+                running = still_running
+                if remaining:
+                    time.sleep(_POLL_INTERVAL)
+        except BaseException as error:
+            stop.set()
+            if cancel_events is not None:
+                for event in cancel_events:
+                    event.set()
+            self._replay_interrupted(
+                items, results, finished, started_at, log_buffers, error, remove_passing_logs, min_level
+            )
+            raise
+        return results
+
+    @staticmethod
+    def _timed_out_result(item, log_buffer: list, call_timeout: float) -> Dict[str, Any]:
+        duration = secs_to_timestr(call_timeout)
+        logs = list(log_buffer)
+        logs.append(("ERROR", f"Call for item {item} timed out after {duration} (call_timeout); "
+                              "it keeps running in the background"))
+        return {"status": "FAIL", "item": item, "logs": logs, "error": f"Timed out after {duration}"}
+
+    def _replay_interrupted(self, items, results, finished, started_at, log_buffers, error, remove_passing_logs, min_level):
+        """Replays what the calls logged before the keyword was interrupted.
+        Calls that never started are only counted, not listed one by one."""
+        # On Windows Robot raises the timeout as a bare class, without message.
+        reason = str(error) or ("Robot timeout exceeded" if isinstance(error, TimeoutExceeded) else type(error).__name__)
+        partial = []
+        running = not_started = 0
+        for index, item in enumerate(items):
+            # Read the results before `started_at`: a call that finishes in
+            # between is then reported as still running, never as not started.
+            result = results[index] or finished[index]
+            if result is not None:
+                partial.append(result)
+            elif started_at[index] is not None:
+                running += 1
+                logs = list(log_buffers[index])
+                logs.append(("ERROR", f"Call for item {item} did not finish: still running when the keyword was interrupted ({reason})"))
+                partial.append({"status": "FAIL", "item": item, "logs": logs, "error": f"Did not finish: interrupted ({reason})"})
+            else:
+                not_started += 1
+        logger.info(
+            f"Run Parallel Scenarios was interrupted ({reason}): {len(partial) - running} call(s) finished, "
+            f"{running} still running (left running in the background), {not_started} not started."
+        )
+        self._replay_logs(partial, remove_passing_logs, min_level)
+
+    def _execute_and_capture(
+        self, method, item, logger_mapper: Optional[Any], log_buffer: list, kwargs: Dict[str, Any],
+        argument: Any = _NOT_SET,
+    ) -> Dict[str, Any]:
+        """Worker wrapper to capture logs and results.
+
+        ``argument`` is what the call receives as its first argument (``item``
+        converted by its type hint), ``item`` what the result reports. ``kwargs``
+        is a dict, not ``**kwargs``, so a keyword argument that happens to be
+        called ``method``, ``item`` or ``log_buffer`` can't clash with this
+        method's own parameters.
+        """
+        if argument is _NOT_SET:
+            argument = item
         # Injection of a custom logger into the thread
         def thread_log(msg, level="INFO"):
-            # Filter out IGNORE (sv=0) level logs
-            if level != "IGNORE":
+            # Filter out IGNORE (sv=0) level logs. Other levels are kept as
+            # given and checked when the logs are replayed.
+            if _normalize_level(level) != "IGNORE":
                 log_buffer.append((level, msg))
-        
+
         # Use mapper if provided, otherwise use thread_log
         effective_logger = logger_mapper if logger_mapper else thread_log
 
         try:
-            # We pass the custom logger as an extra kwarg if the method supports it
-            # or rely on the method returning its own logs.
-            result = method(item, _logger=effective_logger, **kwargs)
+            result = method(argument, _logger=effective_logger, **kwargs)
             return {"status": "PASS", "item": item, "logs": log_buffer, "result": result}
-        except Exception as e:
+        # BaseException, not Exception: whatever the call raises (even
+        # SystemExit), the worker must report a result, or the main thread
+        # would wait for this call forever.
+        except BaseException as e:
             thread_log(f"Thread failed for item {item}: {str(e)}", "ERROR")
             return {"status": "FAIL", "item": item, "logs": log_buffer, "error": str(e)}
 
-    def _replay_logs(self, results: List[Dict[str, Any]], remove_passing_logs: bool = False, thread_log_level: str = "INFO"):
-        """Sequential dump to the real RF Logger. Filters out IGNORE level logs."""
-        level_order = {"INFO": 1, "WARN": 2, "ERROR": 3}
-        min_level = level_order.get(thread_log_level, 1)
+    def _replay_logs(self, results: List[Dict[str, Any]], remove_passing_logs: bool = False, min_level: int = 0):
+        """Sequential dump to the real RF Logger, in call order.
+
+        Every Robot level goes through ``logger.write``, so Robot's
+        ``--loglevel`` filters it as usual. IGNORE messages and levels below
+        ``min_level`` are skipped. A message with an unknown level is kept at
+        INFO, prefixed with its level, plus one warning per unknown level.
+        (Before 0.3.1 only INFO/WARN/ERROR were written; everything else,
+        DEBUG included, was dropped silently.)
+        """
+        unknown_levels = []
         for entry in results:
             if remove_passing_logs and entry['status'] == 'PASS':
                 continue
             logger.info(f"--- Logs for Item: {entry['item']} ---")
-            for level, msg in entry['logs']:
-                # Skip IGNORE level and logs below minimum level
-                if level == "IGNORE" or level_order.get(level, 1) < min_level:
+            # A copy: after an interruption, a call still running may append to it.
+            for level, msg in list(entry['logs']):
+                name = _normalize_level(level)
+                if name == "IGNORE":
                     continue
-                if level == "INFO": logger.info(msg)
-                elif level == "WARN": logger.warn(msg)
-                elif level == "ERROR": logger.error(msg)
+                if name is None:
+                    if level not in unknown_levels:
+                        unknown_levels.append(level)
+                    name, msg = "INFO", f"[{level}] {msg}"
+                if LOG_LEVELS[name] >= min_level:
+                    logger.write(msg, name)
             logger.info(f"Status: {entry['status']}")
+        for level in unknown_levels:
+            logger.warn(
+                f"Unknown log level {level!r} used with _logger: its messages were logged as INFO, "
+                "prefixed with the level. Use TRACE, DEBUG, INFO, HTML, CONSOLE, WARN, ERROR or IGNORE."
+            )
 
 # Backwards-compatible name: the class was called ParallelLibrary before the
 # package was renamed to ParallelRunner (imported as `Library    ParallelRunner`).

{robotframework_parallelrunner-0.3.0.dist-info → robotframework_parallelrunner-0.4.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: robotframework-parallelrunner
-Version: 0.3.0
+Version: 0.4.0
 Summary: Run a keyword in parallel inside a single Robot Framework test case, with one clean, ordered log.html.
 License-Expression: MIT
 License-File: LICENSE
@@ -155,7 +155,7 @@ Concretely:
   that needs to do the same kind of work many times (loop over a list of IDs,
   or repeat something N times) — how do I parallelize the body of that loop
   without corrupting the log or needing to merge anything?" It uses a
-  `ThreadPoolExecutor` inside a single process, and produces one
+  pool of threads inside a single process, and produces one
   already-merged, already-ordered log for that test case.
 - Because ParallelRunner uses **threads**, not processes, it shares memory and
   is subject to the GIL — it is a good fit for **I/O-bound** work (network
@@ -230,9 +230,11 @@ Things to keep in mind:
   made thread-safe for you; your own keyword's side effects are not. If it
   writes to a shared variable, file, or object without a lock, running it
   concurrently can race the same way any multi-threaded code can.
-- **You need per-item retries or a timeout.** Not implemented yet (see
-  [ROADMAP.md](https://github.com/cristiangarciavd/robotframework-parallelrunner/blob/main/ROADMAP.md)) — one hung call currently blocks the whole batch
-  from returning.
+- **You need per-item retries.** Not implemented yet (see
+  [ROADMAP.md](https://github.com/cristiangarciavd/robotframework-parallelrunner/blob/main/ROADMAP.md)). A per-call limit is: `call_timeout=10s` fails
+  only the call that hangs and lets the others finish. Note that a timed-out
+  call can't be killed - it keeps running in the background unless the
+  keyword declares `_cancelled` and stops when it is set.
 - **You need process-level isolation** (a crash in one call shouldn't be
   able to affect another) or cross-machine parallelism — that's pabot's
   domain, not this library's.
@@ -242,7 +244,7 @@ Things to keep in mind:
 ```
 robotframework-parallelrunner/
 ├── src/ParallelRunner/     # The installable library (do not depend on internals prefixed with `_`)
-├── examples/               # Example libraries used by the tests, plus the pabot timing demo
+├── examples/               # Example libraries used by the tests, plus the pabot and log levels demos
 ├── atest/                  # Robot Framework acceptance suites
 ├── utest/                  # pytest unit tests
 ├── docs/                   # INSTALLATION, QUICKSTART, API_REFERENCE, generated keyword docs

robotframework_parallelrunner-0.4.0.dist-info/RECORD

@@ -0,0 +1,9 @@
+ParallelRunner/__init__.py,sha256=x74aR4t2SZvD81MepoTobACqyLfX_rKaEWPJY08kvTA,671
+ParallelRunner/parallel_library.py,sha256=oXaLDC09KAJ0RzLooIw-fP1nZ-iuVZ7gzTs0bbck5kA,42180
+ParallelRunner/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+ParallelRunner/version.py,sha256=W8E8ymdTLqfKl_1qBIOrjGpO8kSbqPw30aYZpqYUNMk,632
+parallelrunner.py,sha256=2oyKOVeKlOaZUZ3f2i_l5xCAYFKVm11bSQZOtzSSJO8,1132
+robotframework_parallelrunner-0.4.0.dist-info/METADATA,sha256=XxHtM1t5mGkD_W5NDfP_K6ckulVLodz2eRnbjzHbzak,14488
+robotframework_parallelrunner-0.4.0.dist-info/WHEEL,sha256=L-WvLdSBvJWHXv5zrQxDvNRnIgoBzLUpcbuiw19KCuQ,88
+robotframework_parallelrunner-0.4.0.dist-info/licenses/LICENSE,sha256=xxQh3uvlk5jC1i0eoH5Nyo9cDz2Fzu0DqOzAZw8XxFs,1072
+robotframework_parallelrunner-0.4.0.dist-info/RECORD,,

robotframework_parallelrunner-0.3.0.dist-info/RECORD

@@ -1,9 +0,0 @@
-ParallelRunner/__init__.py,sha256=ifsoO1rXf4kL3i75NokrD_KHxnlvE3IuH69Fb-ByfNg,641
-ParallelRunner/parallel_library.py,sha256=qagok_v9Kw_OpsXy5yRb2hvwBpwadfgwhfC5bL1boVE,19123
-ParallelRunner/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-ParallelRunner/version.py,sha256=W8E8ymdTLqfKl_1qBIOrjGpO8kSbqPw30aYZpqYUNMk,632
-parallelrunner.py,sha256=2oyKOVeKlOaZUZ3f2i_l5xCAYFKVm11bSQZOtzSSJO8,1132
-robotframework_parallelrunner-0.3.0.dist-info/METADATA,sha256=6Ph0tfBiO_I6mZ4BP606SDGfJYwCUzOICu5LHMoq1kY,14313
-robotframework_parallelrunner-0.3.0.dist-info/WHEEL,sha256=L-WvLdSBvJWHXv5zrQxDvNRnIgoBzLUpcbuiw19KCuQ,88
-robotframework_parallelrunner-0.3.0.dist-info/licenses/LICENSE,sha256=xxQh3uvlk5jC1i0eoH5Nyo9cDz2Fzu0DqOzAZw8XxFs,1072
-robotframework_parallelrunner-0.3.0.dist-info/RECORD,,