FAIRmat-NFDI · sanbrock · Jul 1, 2022 · Jun 21, 2022 · Jun 27, 2022 · sanbrock
diff --git a/nexusparser.yaml b/nexusparser.yaml
@@ -0,0 +1,7 @@
+# NexusParser Parameter File - v0.0.1
+
+dataconverter:
+  reader: mpes
+  output: output.nxs
+  nxdl: NXmpes
+  input-file: ['config_file.json', 'ELN_metadata_example.yaml', 'xarray_saved_small_calibration.h5']
diff --git a/nexusparser/parser.py b/nexusparser/parser.py
@@ -19,9 +19,12 @@
 # limitations under the License.
 #
 
+from typing import Iterable, Union
+10000
import os
+import pathlib
 import numpy as np
 from nomad.datamodel import EntryArchive
-from nomad.parsing import MatchingParser
+from nomad.parsing import Parser
 # from . import metainfo  # pylint: disable=unused-import
 from nexusparser.tools import nexus as read_nexus
 from nexusparser.metainfo import nexus
@@ -172,19 +175,28 @@ def add_log(params, logstr):
     return logstr
 
 
-class NexusParser(MatchingParser):
+class NexusParser(Parser):
     """NesusParser doc
 
 """
     def __init__(self):
-        super().__init__(
-            name='parsers/nexus', code_name='NEXUS', code_homepage='https://www.nexus.eu/',
-            mainfile_mime_re=r'(application/.*)|(text/.*)',
-            mainfile_name_re=(r'.*\.nxs'),
-            supported_compressions=['gz', 'bz2', 'xz']
-        )
+        super().__init__()
+        self.name = "parsers/nexus"
         self.archive = None
         self.nxroot = None
+        self.domain = 'ems'
+
+    def is_mainfile(  # pylint: disable=too-many-arguments
+            self, filename: str, mime: str, buffer: bytes, decoded_buffer: str,
+            compression: str = None) -> Union[bool, Iterable[str]]:
+        accepted_extensions = (".nxs", ".yaml", ".yml")
+        extension = pathlib.Path(filename).suffix
+        if extension in accepted_extensions:
+            if buffer[0:8] == b'\x89HDF\r\n\x1a\n':
+                return True
+            if buffer[0:30] == b"# NexusParser Parameter File -":
+                return True
+        return False
 
 #     def get_nomad_classname(self, xml_name, xml_type, suffix):
 #         """Get nomad classname from xml file
@@ -237,6 +249,29 @@ def parse(self, mainfile: str, archive: EntryArchive, logger=None, child_archive
         self.archive.m_create(nexus.Nexus)  # type: ignore[attr-defined] # pylint: disable=no-member
         self.nxroot = self.archive.nexus
 
+        extension = pathlib.Path(mainfile).suffix
+        if extension in (".yaml", ".yml"):
+            base_dir = os.path.dirname(mainfile)
+            from nexusparser.tools.dataconverter.convert import convert, parse_params_file
+            with open(mainfile) as file:
+                conv_params = parse_params_file(file)
+
+                def check_path(path: str):
+                    """Return true if path supplied by the user is not absolute or has a ../"""
+                    if os.path.isabs(path) or ".." in path:
+                        raise Exception("The user provided an invalid path in the parameter YAML.")
+                    return path
+
+                if isinstance(conv_params["input_file"], list):
+                    conv_params["input_file"] = [f"{base_dir}{os.sep}{check_path(file)}"
+                                                 for file in conv_params["input_file"]]
+                else:
+                    conv_params["input_file"] = (f"{base_dir}{os.sep}"
+                                                 f"{check_path(conv_params['input_file'])}")
+                conv_params["output"] = f"{base_dir}{os.sep}{check_path(conv_params['output'])}"
+                convert(**conv_params)
+                mainfile = conv_params["output"]
+
         nexus_helper = read_nexus.HandleNexus(logger, [mainfile])
         nexus_helper.process_nexus_master_file(self.nexus_populate)
 

diff --git a/nexusparser/tools/dataconverter/convert.py b/nexusparser/tools/dataconverter/convert.py
@@ -27,6 +27,7 @@
 import xml.etree.ElementTree as ET
 
 import click
+import yaml
 
 from nexusparser.tools.dataconverter.readers.base.reader import BaseReader
 from nexusparser.tools.dataconverter import helpers
@@ -91,6 +92,8 @@ def convert(input_file: Tuple[str],  # pylint: disable=too-many-arguments
         return
 
     # Setting up all the input data
+    if isinstance(input_file, str):
+        input_file = (input_file,)
     bulletpoint = "\n\u2022 "
     logger.info("Using %s reader to convert the given files: %s ",
                 reader,
@@ -125,6 +128,14 @@ def convert(input_file: Tuple[str],  # pylint: disable=too-many-arguments
     logger.info("The output file generated: %s", output)
 
 
+def parse_params_file(params_file):
+    """Parses the parameters from a given dictionary and returns them"""
+    params = yaml.load(params_file, Loader=yaml.Loader)['dataconverter']
+    for param in list(params.keys()):
+        params[param.replace("-", "_")] = params.pop(param)
+    return params
+
+
 @click.command()
 @click.option(
     '--input-file',
@@ -141,7 +152,7 @@ def convert(input_file: Tuple[str],  # pylint: disable=too-many-arguments
 @click.option(
     '--nxdl',
     default=None,
-    required=True,
+    required=False,
     help='The name of the NXDL file to use without extension.'
 )
 @click.option(
@@ -155,20 +166,39 @@ def convert(input_file: Tuple[str],  # pylint: disable=too-many-arguments
     default=False,
     help='Just print out the template generated from given NXDL file.'
 )
-@click.option(  # pylint: disable=too-many-arguments
+@click.option(
     '--fair',
     is_flag=True,
     default=False,
     help='Let the converter know to be stricter in checking the documentation.'
+)  # pylint: disable=too-many-arguments
+@click.option(
+    '--params-file',
+    type=click.File('r'),
+    default=None,
+    help='Allows to pass a .yaml file with all the parameters the converter supports.'
 )
 def convert_cli(input_file: Tuple[str],
                 reader: str,
                 nxdl: str,
                 output: str,
                 generate_template: bool,
-                fair: bool):
+                fair: bool,
+                params_file: str):
     """The CLI entrypoint for the convert function"""
-    convert(input_file, reader, nxdl, output, generate_template, fair)
+    if params_file:
+        try:
+            convert(**parse_params_file(params_file))
+        except TypeError:
+            raise Exception(("Please make sure you have the following entries in your "
+                             "parameter file:\n\n# NexusParser Parameter File - v0.0.1"
+                             "\n\ndataconverter:\n\treader: value\n\tnxdl: value\n\tin"
+                             "put-file: value"))
+    else:
+        if nxdl is None:
+            raise Exception("\nError: Please supply an NXDL file with the option:"
+                            " --nxdl <path to NXDL>")
+        convert(input_file, reader, nxdl, output, generate_template, fair)
 
 
 if __name__ == '__main__':