add decrecaption warning when dpdata throws errors while parsing cp2k

dpdata/cp2k/output.py

@@ -76,12 +76,14 @@
     def get_log_block_generator(self):
         lines = []
         delimiter_flag = False
+        yield_flag = False
         while True:
             line = self.log_file_object.readline()
             if line:
                 lines.append(line)
                 if any(p.match(line) for p in delimiter_patterns):
                     if delimiter_flag is True:
+                        yield_flag = True
                         yield lines
                         lines = []
                         delimiter_flag = False
@@ -91,17 +93,23 @@
                         if any(p.match(line) for p in avail_patterns):
                             delimiter_flag = True
             else:
+                if not yield_flag:
+                    raise StopIteration("None of the delimiter patterns are matched")
                 break
         if delimiter_flag is True:
             raise RuntimeError("This file lacks some content, please check")
 
     def get_xyz_block_generator(self):
         p3 = re.compile(r"^\s*(\d+)\s*")
+        yield_flag = False
         while True:
             line = self.xyz_file_object.readline()
             if not line:
+                if not yield_flag:
+                    raise StopIteration("None of the xyz patterns are matched")
                 break
             if p3.match(line):
+                yield_flag = True
                 atom_num = int(p3.match(line).group(1))
                 lines = []
                 lines.append(line)

dpdata/plugins/cp2k.py

@@ -4,29 +4,48 @@
 from dpdata.cp2k.output import Cp2kSystems
 from dpdata.format import Format
 
+string_warning = """
+Hi, you got an error from dpdata,
+please check if your cp2k files include full information,
+otherwise its version is not supported by dpdata.
+Try use dpdata plugin from cp2kdata package,
+for details, please refer to
+https://robinzyb.github.io/cp2kdata/
+"""
+
 
 @Format.register("cp2k/aimd_output")
 class CP2KAIMDOutputFormat(Format):
     def from_labeled_system(self, file_name, restart=False, **kwargs):
-        xyz_file = sorted(glob.glob(f"{file_name}/*pos*.xyz"))[0]
-        log_file = sorted(glob.glob(f"{file_name}/*.log"))[0]
-        return tuple(Cp2kSystems(log_file, xyz_file, restart))
+        try:
+            xyz_file = sorted(glob.glob(f"{file_name}/*pos*.xyz"))[0]
+            log_file = sorted(glob.glob(f"{file_name}/*.log"))[0]
+            return tuple(Cp2kSystems(log_file, xyz_file, restart))
+        except (StopIteration, RuntimeError):
+            # StopIteration is raised when pattern match is failed
+            print(string_warning)
 
 
 @Format.register("cp2k/output")
 class CP2KOutputFormat(Format):
     def from_labeled_system(self, file_name, restart=False, **kwargs):
-        data = {}
-        (
-            data["atom_names"],
-            data["atom_numbs"],
-            data["atom_types"],
-            data["cells"],
-            data["coords"],
-            data["energies"],
-            data["forces"],
-            tmp_virial,
-        ) = dpdata.cp2k.output.get_frames(file_name)
-        if tmp_virial is not None:
-            data["virials"] = tmp_virial
-        return data
+        try:
+            data = {}
+            (
+                data["atom_names"],
+                data["atom_numbs"],
+                data["atom_types"],
+                data["cells"],
+                data["coords"],
+                data["energies"],
+                data["forces"],
+                tmp_virial,
+            ) = dpdata.cp2k.output.get_frames(file_name)
+            if tmp_virial is not None:
+                data["virials"] = tmp_virial
+            return data
+        # TODO: in the future, we should add exact error type here
+        # TODO: when pattern match is failed
+        # TODO: For now just use RuntimeError as a placeholder.
+        except RuntimeError:
+            print(string_warning)