Unittests for Pipeline

annotator/Pipeline.py

@@ -41,7 +41,10 @@ def __init__(self, syn, view=None, meta=None, activeCols=[],
             Optional. Whether to sort the columns lexicographically in
             `view` and/or `meta`. Defaults to True.
         """
-        self.syn = syn
+        if isinstance(syn, sc.Synapse):
+            self.syn = syn
+        else:
+            raise TypeError("syn must be a synapseclient.Synapse object")
         self.view = view if view is None else self._parseView(view, sortCols)
         self._entityViewSchema = (self.syn.get(view)
                                   if isinstance(view, str) else None)
@@ -62,11 +65,13 @@ def __init__(self, syn, view=None, meta=None, activeCols=[],
         self.links = links if isinstance(links, dict) else None
         self._backup = []
 
-    def backup(self, message):
+    def backup(self, message=None):
         """ Backup the state of `self` and store in `self._backup` """
         self._backup.append((Pipeline(
-            self.syn, self.view, self._meta, self._activeCols,
-            self._metaActiveCols, self.links, self._sortCols), message))
+            syn=self.syn, view=self.view, meta=self._meta,
+            activeCols=self._activeCols, metaActiveCols=self._metaActiveCols,
+            links=self.links, sortCols=self._sortCols, schema=self.schema),
+            message))
         if len(self._backup) > self.BACKUP_LENGTH:
             self._backup = self._backup[1:]
 
@@ -470,7 +475,7 @@ def _parseView(self, view, sortCols, isMeta=False):
         """
         if isinstance(view, str):
             return utils.synread(self.syn, view, sortCols=sortCols)
-        elif isinstance(view, list) and meta:
+        elif isinstance(view, list) and isMeta:
             return utils.combineSynapseTabulars(self.syn, view, axis=1)
         elif isinstance(view, pd.DataFrame):
             if sortCols:
@@ -524,7 +529,11 @@ def _getUserConfirmation(self, message="Proceed anyways? (y) or (n): "):
         Otherwise asks user to input confirmation again.
         """
         proceed = ''
+        counter = 0
         while not proceed:
+            counter += 1
+            if counter > 3:
+                return False  # after three attempts return False
             proceed = input(message)
             if len(proceed) and not proceed[0] in ['Y', 'y', 'N', 'n']:
                 proceed = ''
@@ -562,15 +571,16 @@ def _validate(self):
                                         ", ".join(self.schema.loc[k].value.values)))
         return warnings
 
-    def removeActiveCols(self, activeCols):
+    def removeActiveCols(self, activeCols, backup=True):
         """ Remove a column name from `self._activeCols`
 
         Parameters
         ----------
         activeCols : str or list-like
             Column name(s) to remove.
         """
-        self.backup("removeActiveCols")
+        if backup:
+            self.backup("removeActiveCols")
         if isinstance(activeCols, str):
             self._activeCols.remove(activeCols)
         else:  # is list-like

tests/unit/conftest.py

@@ -4,6 +4,7 @@
 import pandas
 import synapseclient
 import uuid
+import annotator
 
 
 SAMPLE_FILE = "https://raw.githubusercontent.com/Sage-Bionetworks/annotator/master/tests/sampleFile.csv"
@@ -41,17 +42,16 @@ def project(syn):
     syn.delete(project)
 
 
-def file_(syn, parent, path, annotations=None, **kwargs):
+def file_(syn, parent, path, **kwargs):
     if 'name' not in kwargs:
         name = str(uuid.uuid4())
     else:
         name = kwargs.pop('name')
-    file_ = synapseclient.File(path=path, name=name,
-                               parent=parent, **kwargs)
-    if annotations:
-        for a in annotations.items():
-            key, value = a
-            file_[key] = value
+    file_ = synapseclient.File(
+            path=path,
+            name=name,
+            parent=parent,
+            **kwargs)
     file_ = syn.store(file_)
     return file_
 
@@ -106,21 +106,29 @@ def genericTable(syn, project, sampleFile):
 def entities(syn, sampleFile, project):
     # store a folder for our entities
     sample_folder = folder(syn, project)
+    sample_folder_two = folder(syn, project)
     # store sample files
-    _file = file_(syn, sample_folder, SAMPLE_FILE, name="file1.csv")
-    _file2 = file_(syn, sample_folder, SAMPLE_FILE, name="file2.csv")
-    _file3 = file_(syn, sample_folder, SAMPLE_FILE, name="file3.csv")
+    _file = file_(syn, sample_folder, SAMPLE_FILE, name="file1.csv",
+                  synapseStore=False)
+    _file2 = file_(syn, sample_folder, SAMPLE_FILE, name="file2.csv",
+                   synapseStore=False)
+    _file3 = file_(syn, sample_folder_two, SAMPLE_FILE, name="file3.csv",
+                   synapseStore=False)
     # store a sample metadata file
-    meta = synapseclient.File(path=SAMPLE_META, name='meta',
-                              parent=sample_folder)
-    meta = syn.store(meta)
+    meta = file_(syn, project, SAMPLE_META, name='meta', synapseStore=False)
     # store a sample table (same values as sample file)
     schema = table(syn, project, sampleFile)
     # store a sample file view
-    entity_view_ = entity_view(syn, project, scopes=sample_folder)
-    ents = {'files': [_file, _file2, _file3],
-            'folder': sample_folder,
+    entity_view_ = entity_view(syn, project,
+            scopes=[sample_folder, sample_folder_two])
+    entities = {'files': [_file, _file2, _file3],
+            'folders': [sample_folder, sample_folder_two],
             'meta': meta,
             'table_schema': schema,
-            'entity_view': entity_view_}
-    return ents
+            'entity_view': entity_view_,
+            'project': project}
+    return entities
+
+@pytest.fixture(scope='session')
+def genericPipeline(syn):
+    return annotator.Pipeline(syn)