map data types and spectra layouts

chem_spectra/lib/converter/jcamp/base.py

@@ -1,4 +1,5 @@
 import nmrglue as ng
+import json
 
 from chem_spectra.lib.converter.share import parse_params, parse_solvent
 
@@ -43,83 +44,21 @@ def __read(self, path):
 
     def __set_datatype(self):
         dts = self.datatypes
-        if 'NMR SPECTRUM' in dts:
-            return 'NMR SPECTRUM'
-        elif 'NMRSPECTRUM' in dts:  # MNova
-            return 'NMR SPECTRUM'
-        elif 'INFRARED SPECTRUM' in dts:
-            return 'INFRARED SPECTRUM'
-        elif 'RAMAN SPECTRUM' in dts:
-            return 'RAMAN SPECTRUM'
-        elif 'MASS SPECTRUM' in dts:
-            return 'MASS SPECTRUM'
-        elif 'HPLC UV/VIS SPECTRUM' in dts:
-            return 'HPLC UV/VIS SPECTRUM'
-        elif 'HPLC UV-VIS' in dts:
-            return 'HPLC UV/VIS SPECTRUM'
-        elif 'UV/VIS SPECTRUM' in dts:
-            return 'UV/VIS SPECTRUM'
-        elif 'UV-VIS' in dts:
-            return 'UV/VIS SPECTRUM'
-        elif 'ULTRAVIOLET SPECTRUM' in dts:
-            return 'UV/VIS SPECTRUM'
-        elif 'THERMOGRAVIMETRIC ANALYSIS' in dts:
-            return 'THERMOGRAVIMETRIC ANALYSIS'
-        elif 'X-RAY DIFFRACTION' in dts:
-            return 'X-RAY DIFFRACTION'
-        elif 'CYCLIC VOLTAMMETRY' in dts:
-            return 'CYCLIC VOLTAMMETRY'
-        elif 'SIZE EXCLUSION CHROMATOGRAPHY' in dts:
-            return 'SIZE EXCLUSION CHROMATOGRAPHY'
-        elif 'CIRCULAR DICHROISM SPECTROSCOPY' in dts:
-            return 'CIRCULAR DICHROISM SPECTROSCOPY'
-        elif 'SORPTION-DESORPTION MEASUREMENT' in dts:
-            return 'SORPTION-DESORPTION MEASUREMENT'
-        elif 'Emissions' in dts or 'EMISSIONS' in dts or 'FLUORESCENCE SPECTRUM' in dts or 'FL SPECTRUM' in dts:
-            return 'Emissions'
-        elif 'DLS ACF' in dts:
-            return 'DLS ACF'
-        elif 'DLS INTENSITY' in dts or 'DLS intensity' in dts:
-            return 'DLS intensity'
+
+        with open('data_type.json', 'r') as mapping_file:
+            data_type_mappings = json.load(mapping_file)["datatypes"]
+        for dt in dts:
+            if dt in data_type_mappings:
+                return data_type_mappings[dt]
         return ''
 
     def __typ(self):
         dt = self.datatype
-        if 'NMR SPECTRUM' == dt:
-            return 'NMR'
-        elif 'NMRSPECTRUM' == dt:  # MNova
-            return 'NMR'
-        elif 'INFRARED SPECTRUM' == dt:
-            return 'INFRARED'  # TBD
-        elif 'RAMAN SPECTRUM' == dt:
-            return 'RAMAN'  # TBD
-        elif 'MASS SPECTRUM' == dt:
-            return 'MS'
-        elif 'HPLC UV/VIS SPECTRUM' == dt:
-            return 'HPLC UVVIS'
-        elif 'HPLC UV-VIS' == dt:
-            return 'HPLC UVVIS'
-        elif 'UV/VIS SPECTRUM' == dt or 'UV-VIS' == dt or 'ULTRAVIOLET SPECTRUM' == dt:
-            return 'UVVIS'
-        elif 'THERMOGRAVIMETRIC ANALYSIS' == dt:
-            return 'THERMOGRAVIMETRIC ANALYSIS'
-        elif 'X-RAY DIFFRACTION' == dt:
-            return 'X-RAY DIFFRACTION'
-        elif 'CYCLIC VOLTAMMETRY' in dt:
-            return 'CYCLIC VOLTAMMETRY'
-        elif 'SIZE EXCLUSION CHROMATOGRAPHY' in dt:
-            return 'SIZE EXCLUSION CHROMATOGRAPHY'
-        elif 'CIRCULAR DICHROISM SPECTROSCOPY' in dt:
-            return 'CIRCULAR DICHROISM SPECTROSCOPY'
-        elif 'SORPTION-DESORPTION MEASUREMENT' in dt:
-            return 'SORPTION-DESORPTION MEASUREMENT'
-        elif 'Emissions' in dt or 'EMISSIONS' in dt or 'FLUORESCENCE SPECTRUM' in dt or 'FL SPECTRUM' in dt:
-            return 'Emissions'
-        elif 'DLS ACF' in dt:
-            return 'DLS ACF'
-        elif 'DLS INTENSITY' in dt or 'DLS intensity' in dt:
-            return 'DLS intensity'
-        return ''
+
+        with open('data_type.json', 'r') as mapping_file:
+            data_type_mappings = json.load(mapping_file)["datatypes"]
+
+        return data_type_mappings.get(dt, '')
 
     def __set_dataclass(self):
         data_class = self.dataclasses
@@ -140,12 +79,11 @@ def __is_em_wave(self):
         return self.typ in ['INFRARED', 'RAMAN', 'UVVIS']
 
     def __non_nmr(self):
-        return self.typ in [
-            'INFRARED', 'RAMAN', 'UVVIS', 'HPLC UVVIS',
-            'THERMOGRAVIMETRIC ANALYSIS', 'MS', 'X-RAY DIFFRACTION',
-            'CYCLIC VOLTAMMETRY', 'SIZE EXCLUSION CHROMATOGRAPHY',
-            'CIRCULAR DICHROISM SPECTROSCOPY', 'SORPTION-DESORPTION MEASUREMENT', 'Emissions', 
-            'DLS ACF', 'DLS intensity']
+        with open('data_type.json', 'r') as mapping_file:
+            data_type_mappings = json.load(mapping_file).get("datatypes")
+        dts = set(data_type_mappings.values())
+        dts.discard('NMR')
+        return self.typ in dts
 
     def __is_ir(self):
         return self.typ in ['INFRARED']
@@ -175,13 +113,13 @@ def __is_aif(self):
         return self.typ in ['SORPTION-DESORPTION MEASUREMENT']
 
     def __is_emissions(self):
-        return self.typ in ['Emissions', 'EMISSIONS', 'FLUORESCENCE SPECTRUM', 'FL SPECTRUM']
+        return self.typ in ['Emissions']
 
     def __is_dls_acf(self):
         return self.typ in ['DLS ACF']
 
     def __is_dls_intensity(self):
-        return self.typ in ['DLS INTENSITY', 'DLS intensity']
+        return self.typ in ['DLS intensity']
 
     def __ncl(self):
         try:

chem_spectra/lib/converter/jcamp/data_type.json

@@ -0,0 +1,27 @@
+{
+    "datatypes": {
+      "NMR SPECTRUM": "NMR",
+      "NMRSPECTRUM": "NMR",
+      "INFRARED SPECTRUM": "INFRARED",
+      "RAMAN SPECTRUM": "RAMAN",
+      "MASS SPECTRUM": "MS",
+      "HPLC UV/VIS SPECTRUM": "HPLC UVVIS",
+      "HPLC UV-VIS": "HPLC UVVIS",
+      "UV/VIS SPECTRUM": "UVVIS",
+      "UV-VIS": "UVVIS",
+      "ULTRAVIOLET SPECTRUM": "UVVIS",
+      "THERMOGRAVIMETRIC ANALYSIS": "THERMOGRAVIMETRIC ANALYSIS",
+      "X-RAY DIFFRACTION": "X-RAY DIFFRACTION",
+      "CYCLIC VOLTAMMETRY": "CYCLIC VOLTAMMETRY",
+      "SIZE EXCLUSION CHROMATOGRAPHY": "SIZE EXCLUSION CHROMATOGRAPHY",
+      "CIRCULAR DICHROISM SPECTROSCOPY": "CIRCULAR DICHROISM SPECTROSCOPY",
+      "SORPTION-DESORPTION MEASUREMENT": "SORPTION-DESORPTION MEASUREMENT",
+      "Emissions": "Emissions",
+      "EMISSIONS": "Emissions",
+      "FLUORESCENCE SPECTRUM": "Emissions",
+      "FL SPECTRUM": "Emissions",
+      "DLS ACF": "DLS ACF",
+      "DLS INTENSITY": "DLS intensity",
+      "DLS intensity": "DLS intensity"
+  }
+}

chem_spectra/lib/converter/jcamp/ni.py

@@ -4,7 +4,7 @@
 from chem_spectra.lib.converter.datatable import DatatableModel
 from chem_spectra.lib.shared.calc import to_float
 from chem_spectra.lib.converter.jcamp.data_parse import make_ni_data_ys, make_ni_data_xs
-
+import json
 
 THRESHOLD_IR = 0.93
 THRESHOLD_RAMAN = 0.07
@@ -74,42 +74,32 @@ def __init__(self, base):
 
     def __thres(self):
         dt = self.datatype
-        if 'NMR SPECTRUM' == dt:
-            return THRESHOLD_NMR
-        elif 'NMRSPECTRUM' == dt:  # MNova
+        with open('data_type.json', 'r') as mapping_file:
+            data_type_mappings = json.load(mapping_file)["datatypes"]
+
+        if data_type_mappings.get(dt) == "NMR":
             return THRESHOLD_NMR
-        elif 'INFRARED SPECTRUM' == dt:
+        elif data_type_mappings.get(dt) == "INFRARED":
             return THRESHOLD_IR
-        elif 'RAMAN SPECTRUM' == dt:
+        elif data_type_mappings.get(dt) == "RAMAN":
             return THRESHOLD_RAMAN
-        elif 'MASS SPECTRUM' == dt:
+        elif data_type_mappings.get(dt) == "MS":
             return THRESHOLD_MS
-        elif 'HPLC UV/VIS SPECTRUM' == dt:
-            return THRESHOLD_UVVIS
-        elif 'HPLC UV-VIS' == dt:
-            return THRESHOLD_UVVIS
-        elif dt in ['UV/VIS SPECTRUM', 'UV-VIS', 'ULTRAVIOLET SPECTRUM']:
+        elif data_type_mappings.get(dt) in ["HPLC UVVIS", 'UVVIS']:
             return THRESHOLD_UVVIS
-        elif dt in ['THERMOGRAVIMETRIC ANALYSIS', 'DLS ACF']:
+        elif data_type_mappings.get(dt) in ['THERMOGRAVIMETRIC ANALYSIS', 'DLS ACF']:
             return THRESHOLD_TGA
-        elif dt in ['X-RAY DIFFRACTION', 'CIRCULAR DICHROISM SPECTROSCOPY', 'CYCLIC VOLTAMMETRY', 'SORPTION-DESORPTION MEASUREMENT', 
-                    'DLS INTENSITY', 'DLS intensity']:
+        elif data_type_mappings.get(dt) in ['X-RAY DIFFRACTION', 'CIRCULAR DICHROISM SPECTROSCOPY', 'CYCLIC VOLTAMMETRY', 
+                                            'SORPTION-DESORPTION MEASUREMENT', 'DLS intensity']:
             return THRESHOLD_XRD
-        elif dt in ['Emissions', 'EMISSIONS', 'FLUORESCENCE SPECTRUM', 'FL SPECTRUM']:
+        elif data_type_mappings.get(dt) in ['Emissions']:
             return THRESHOLD_EMISSION
         return 0.5
 
     def __index_target(self):
-        target_topics = [
-            'NMR SPECTRUM', 'NMRSPECTRUM',
-            'INFRARED SPECTRUM', 'RAMAN SPECTRUM',
-            'MASS SPECTRUM', 'UV/VIS SPECTRUM', 'UV-VIS', 'ULTRAVIOLET SPECTRUM',
-            'HPLC UV-VIS', 'HPLC UV/VIS SPECTRUM',
-            'THERMOGRAVIMETRIC ANALYSIS', 'X-RAY DIFFRACTION',
-            'CYCLIC VOLTAMMETRY', 'SIZE EXCLUSION CHROMATOGRAPHY',
-            'CIRCULAR DICHROISM SPECTROSCOPY', 'SORPTION-DESORPTION MEASUREMENT',
-            'Emissions', 'EMISSIONS', 'FLUORESCENCE SPECTRUM', 'FL SPECTRUM', 'DLS ACF', 'DLS INTENSITY', 'DLS intensity'
-        ]
+        with open('data_type.json', 'r') as mapping_file:
+            target_topics = json.load(mapping_file).get("datatypes")
+
         for tp in target_topics:
             if tp in self.datatypes:
                 idx = self.datatypes.index(tp)