implementing history.py file

Author: Boyuan Deng

mltrace/db/store.py

@@ -536,6 +536,35 @@ def get_history(
 
         return history
 
+    def get_component_runs_by_index(
+        self,
+        component_name: str,
+        front_idx: int,
+        last_idx: int,
+    ) -> typing.List[ComponentRun]:
+        """Gets lineage for the component, or a history of all its runs."""
+
+        if front_idx < 0 or last_idx < 0:
+            total_length = self.session.query(ComponentRun).count()
+            if front_idx < 0:
+                front_idx = front_idx + total_length
+            if last_idx < 0:
+                last_idx = last_idx + total_length
+
+        history = (
+            self.session.query(ComponentRun)
+            .filter(ComponentRun.component_name == component_name)
+            .order_by(ComponentRun.start_timestamp)
+            .offset(front_idx)
+            .limit(last_idx - front_idx)
+            .all()
+        )
+        return history
+
+    def get_component_runs_count(self, component_name: str):
+        return  self.session.query(ComponentRun).count()
+
+
     def get_components(self, tag: str = "", owner: str = ""):
         """Returns a list of all the components associated with the specified
         owner and/or tags."""

mltrace/entities/base_component.py

@@ -15,7 +15,7 @@
 import inspect
 import git
 import mlflow
-
+import history
 
 class Component(Base):
     def __init__(
@@ -36,6 +36,8 @@ def __init__(
         self._tags = tags
         self._beforeTests = beforeTests
         self._afterTests = afterTests
+        self._history = history.History(name)
+
 
     def beforeRun(self, **kwargs) -> dict:
         """Computation to execute before running a component.

mltrace/entities/history.py

@@ -0,0 +1,48 @@
+
+
+from datetime import datetime
+
+from mltrace.db import Store
+from mltrace import utils as clientUtils
+
+
+class History(object):
+    def __init__(
+        self,
+        componentName,
+    ):
+        # {component_run_id: component_run} sort by time
+        self.component_name = componentName
+
+        # should I choose a ordered orderDict / list ?
+        # self.history_component_runs = {}
+        
+    def get_runs_by_time(
+        self,
+        start_time: datetime,
+        end_time: datetime,
+    ):
+        store = Store(clientUtils.get_db_uri())
+        history_runs = store.get_history(self.component_name, None, start_time, end_time)
+        # TODO: Client.py - convert to client facing component run object
+        return str(history_runs)
+
+    def get_runs_by_index(
+        self,
+        front_idx: int,
+        last_idx: int,
+    ):
+        store = Store(clientUtils.get_db_uri())
+        history_runs = store.get_component_runs_by_index(self.component_name, front_idx, last_idx)
+        return str(history_runs)
+
+    # all lazy loading functions
+
+    def __getitem__(self, index):
+        store = Store(clientUtils.get_db_uri())
+        history_run = store.get_component_runs_by_index(self.component_name, index, index + 1)
+        return str(history_run)
+
+    def __len__(self):
+        store = Store(clientUtils.get_db_uri())
+        return store.get_component_runs_count(self.component_name)