Lazy statepoint loading

signac/contrib/job.py

@@ -52,12 +52,21 @@ class Job(object):
 
     KEY_DATA = 'signac_data'
 
-    def __init__(self, project, statepoint, _id=None):
+    def __init__(self, project, statepoint=None, _id=None):
         self._project = project
 
-        # Set statepoint and id
-        self._statepoint = SyncedAttrDict(statepoint, parent=_sp_save_hook(self))
-        self._id = calc_id(self._statepoint()) if _id is None else _id
+        if statepoint is None and _id is None:
+            raise ValueError('Either statepoint or _id must be provided.')
+
+        # Set statepoint if provided
+        if statepoint is not None:
+            self._statepoint = SyncedAttrDict(statepoint, parent=_sp_save_hook(self))
+        else:
+            # Statepoint will be loaded lazily
+            self._statepoint = None
+
+        # Set id
+        self._id = calc_id(self.statepoint()) if _id is None else _id
 
         # Prepare job working directory
         self._wd = os.path.join(project.workspace(), self._id)
@@ -89,7 +98,7 @@ def __str__(self):
     def __repr__(self):
         return "{}(project={}, statepoint={})".format(
             self.__class__.__name__,
-            repr(self._project), self._statepoint)
+            repr(self._project), self.statepoint)
 
     def __eq__(self, other):
         return hash(self) == hash(other)
@@ -145,7 +154,7 @@ def reset_statepoint(self, new_statepoint):
             else:
                 raise
         # Update this instance
-        self._statepoint._data = dst._statepoint._data
+        self.statepoint._data = dst.statepoint._data
         self._id = dst._id
         self._wd = dst._wd
         self._fn_doc = dst._fn_doc
@@ -203,6 +212,11 @@ def statepoint(self):
             `sp_dict = job.statepoint()` instead of `sp = job.statepoint`.
             For more information, see :class:`~signac.JSONDict`.
         """
+        if self._statepoint is None:
+            # Loads statepoint file lazily
+            statepoint = self._project.get_statepoint(self._id)
+            self._statepoint = SyncedAttrDict(statepoint, parent=_sp_save_hook(self))
+
         return self._statepoint
 
     @statepoint.setter
@@ -326,7 +340,7 @@ def _init(self, force=False):
 
         try:
             # Ensure to create the binary to write before file creation
-            blob = json.dumps(self._statepoint, indent=2)
+            blob = json.dumps(self.statepoint, indent=2)
 
             try:
                 # Open the file for writing only if it does not exist yet.
@@ -568,7 +582,7 @@ def __exit__(self, err_type, err_value, tb):
 
     def __setstate__(self, state):
         self.__dict__.update(state)
-        self._statepoint._parent.jobs.append(self)
+        self.statepoint._parent.jobs.append(self)
 
     def __deepcopy__(self, memo):
         cls = self.__class__

signac/contrib/project.py

@@ -363,7 +363,12 @@ def open_job(self, statepoint=None, id=None):
                     id = matches[0]
                 elif len(matches) > 1:
                     raise LookupError(id)
-            return self.Job(project=self, statepoint=self.get_statepoint(id), _id=id)
+                else:
+                    # By elimination, len(matches) == 0
+                    raise KeyError(id)
+            elif not self._contains_job_id(id):
+                raise KeyError(id)
+            return self.Job(project=self, statepoint=None, _id=id)
 
     def _job_dirs(self):
         try:
@@ -396,6 +401,16 @@ def num_jobs(self):
 
     __len__ = num_jobs
 
+    def _contains_job_id(self, id):
+        """Determine whether job id is in the project's data space.
+
+        :param id: The job id to test for initialization.
+        :type id: str
+        :returns: True when the job is initialized for this project.
+        :rtype: bool
+        """
+        return os.path.exists(os.path.join(self._wd, id))
+
     def __contains__(self, job):
         """Determine whether job is in the project's data space.
 
@@ -404,7 +419,7 @@ def __contains__(self, job):
         :returns: True when the job is initialized for this project.
         :rtype: bool
         """
-        return os.path.exists(os.path.join(self._wd, job.get_id()))
+        return self._contains_job_id(job.get_id())
 
     @deprecated(deprecated_in="1.3", removed_in="2.0", current_version=__version__)
     def build_job_search_index(self, index, _trust=False):
@@ -704,7 +719,7 @@ def write_statepoints(self, statepoints=None, fn=None, indent=2):
 
     def _register(self, job):
         "Register the job within the local index."
-        self._sp_cache[job._id] = job._statepoint._as_dict()
+        self._sp_cache[job._id] = job.statepoint()
 
     def _get_statepoint_from_workspace(self, jobid):
         "Attempt to read the statepoint from the workspace."

tests/test_project.py

@@ -392,7 +392,7 @@ def test_corrupted_statepoint_file(self):
         try:
             logging.disable(logging.CRITICAL)
             with self.assertRaises(JobsCorruptedError):
-                self.project.open_job(id=job.get_id())
+                self.project.open_job(id=job.get_id()).statepoint
         finally:
             logging.disable(logging.NOTSET)
 
@@ -467,7 +467,7 @@ def test_repair_corrupted_workspace(self):
             # Iterating through the jobs should now result in an error.
             with self.assertRaises(JobsCorruptedError):
                 for job in self.project:
-                    pass
+                    job.statepoint
 
             with self.assertRaises(JobsCorruptedError):
                 self.project.repair()