✨ Add instruction: read previous evaluation tables

README.md

@@ -104,6 +104,7 @@ The following instructions are currently supported by picometer:
   - `load` model from a cif file, given `filename` or mapping syntax:
     `{path: filename.cif, block: cif_block}`.
   - `write` table with all evaluations to a csv file.
+  - `read` table with previous evaluation (to append further).
 - **Selection instructions**
   - `select` atoms, groups, or shapes to be used; use raw element names
     or provide symmetry relation / recenter using mapping syntax, for example:

picometer/instructions.py

@@ -407,6 +407,32 @@ def handle_one(self, instruction: Instruction, ms_key: str, ms: ModelState) -> N
         logger.info(f'Evaluated dihedral {label}: {dihedral} for model state {ms_key}')
 
 
+class ReadInstructionHandler(BaseInstructionHandler):
+    name = 'read'
+    kwargs = dict(path=Path)
+
+    def handle(self, instruction: Instruction) -> None:
+        path = Path(instruction.kwargs['path'])
+        paths = [path] if path.is_file() else sorted(glob(str(path)))
+        for path in paths:
+            self._read_table(str(path))
+
+    def _read_table(self, path: str) -> None:
+        old = self.processor.evaluation_table
+        new = pd.read_csv(path, index_col=0)
+
+        if old.empty:
+            self.processor.evaluation_table = new
+        else:
+            index = old.index.union(new.index)
+            columns = old.columns.union(new.columns)
+            old = old.reindex(index=index, columns=columns)
+            old.update(new)
+            self.processor.evaluation_table = old
+
+        logger.info(f'Loaded evaluation table from {path}')
+
+
 class WriteInstructionHandler(BaseInstructionHandler):
     name = 'write'
     kwargs = dict(path=Path)

tests/test_instructions.py

@@ -490,6 +490,46 @@ def test_write(self):
         assert_frame_equal(correct, results, check_exact=False,
                            rtol=1e-13, atol=1e-12)
 
+    def test_read(self):
+        with importlib.resources.path('tests', 'test_ferrocene.yaml') as yaml_path:
+            tests_path = yaml_path.parent
+            routine_tmp = 'instructions:\n  - read: {}/ferrocene_correct.csv\n'
+            routine_text = routine_tmp.format(tests_path)
+        p = process(Routine.from_string(routine_text))
+        correct_path = tests_path / 'ferrocene_correct.csv'
+        correct = pd.read_csv(correct_path, index_col=0)
+        results = p.evaluation_table
+        results.index = correct.index  # index is env-dependent so ignore it
+        assert_frame_equal(correct, results, check_exact=False,
+                           rtol=1e-13, atol=1e-12)
+
+    def test_read_twice(self):
+        with importlib.resources.path('tests', 'test_ferrocene.yaml') as yaml_path:
+            tests_path = yaml_path.parent
+            routine_tmp = 'instructions:\n  - read: {}/ferrocene_correct.csv\n'
+            routine_tmp += '  - read: {}/ferrocene_correct.csv\n'
+            routine_text = routine_tmp.format(tests_path, tests_path)
+        p = process(Routine.from_string(routine_text))
+        correct_path = tests_path / 'ferrocene_correct.csv'
+        correct = pd.read_csv(correct_path, index_col=0)
+        results = p.evaluation_table
+        results.index = correct.index  # index is env-dependent so ignore it
+        assert_frame_equal(correct, results, check_exact=False,
+                           rtol=1e-13, atol=1e-12)
+
+    def test_read_many(self):
+        with importlib.resources.path('tests', 'test_ferrocene.yaml') as yaml_path:
+            tests_path = yaml_path.parent
+            routine_tmp = 'instructions:\n  - read: {}/ferrocene_corr*.csv\n'
+            routine_text = routine_tmp.format(tests_path)
+        p = process(Routine.from_string(routine_text))
+        correct_path = tests_path / 'ferrocene_correct.csv'
+        correct = pd.read_csv(correct_path, index_col=0)
+        results = p.evaluation_table
+        results.index = correct.index  # index is env-dependent so ignore it
+        assert_frame_equal(correct, results, check_exact=False,
+                           rtol=1e-13, atol=1e-12)
+
     def test_document_history(self):
         routine_text = get_yaml('test_ferrocene.yaml')
         original_routine = Routine.from_string(routine_text)