oldhepdata parser: reformat correlation/covariance matrices

* Transform a square matrix into a format with two independent variables and one dependent variable.
* This transformation is already done in the YAML format exported from the old HepData site.

Signed-off-by: Graeme Watt <[email protected]>

hepdata_converter/parsers/oldhepdata_parser.py

@@ -380,6 +380,54 @@ def _parse_table_data(self, data):
                 energy = energy_min
             self._parse_energies(energy)
 
+        if any(word in self.current_table.description.lower() for word in ['covariance', 'correlation', 'matrix']):
+            reformatted = self._reformat_matrix()
+
+    def _reformat_matrix(self):
+        """Transform a square matrix into a format with two independent variables and one dependent variable.
+        """
+        nxax = len(self.current_table.data['independent_variables'])
+        nyax = len(self.current_table.data['dependent_variables'])
+        npts = len(self.current_table.data['independent_variables'][0]['values'])
+
+        # check if 1 x-axis, and npts (>=2) equals number of y-axes
+        if nxax != 1 or nyax != npts or npts < 2:
+            return False
+
+        # add second independent variable with each value duplicated npts times
+        xheader = self.current_table.data['independent_variables'][0]['header']
+        self.current_table.data['independent_variables'].append({'header': xheader, 'values': []})
+        for value in self.current_table.data['independent_variables'][0]['values']:
+            self.current_table.data['independent_variables'][1]['values'].extend(npts * [value])
+
+        # duplicate values of first independent variable npts times
+        self.current_table.data['independent_variables'][0]['values'] \
+            = npts * self.current_table.data['independent_variables'][0]['values']
+
+        # suppress header if different for second y-axis
+        if self.current_table.data['dependent_variables'][0]['header'] != \
+                self.current_table.data['dependent_variables'][1]['header']:
+            self.current_table.data['dependent_variables'][0]['header'] = {'name': ''}
+
+        # remove qualifier if different for second y-axis
+        iqdel = [] # list of qualifier indices to be deleted
+        for iq, qualifier in enumerate(self.current_table.data['dependent_variables'][0]['qualifiers']):
+            if qualifier != self.current_table.data['dependent_variables'][1]['qualifiers'][iq]:
+                iqdel.append(iq)
+        for iq in iqdel[::-1]: # need to delete in reverse order
+            del self.current_table.data['dependent_variables'][0]['qualifiers'][iq]
+
+        # append values of second and subsequent y-axes to first dependent variable
+        for iy in range(1, nyax):
+            for value in self.current_table.data['dependent_variables'][iy]['values']:
+                self.current_table.data['dependent_variables'][0]['values'].append(value)
+
+        # finally, delete the second and subsequent y-axes in reverse order
+        for iy in range(nyax-1, 0, -1):
+            del self.current_table.data['dependent_variables'][iy]
+
+        return True
+
     def _parse_dserror(self, data):
         """Parse dserror attribute of the old HEPData format
 
@@ -447,7 +495,7 @@ def _parse_qual(self, data):
         headers = data.split(':')
         name = headers[0].strip()
 
-        name = name.split(' IN ')
+        name = re.split(' IN ', name, flags=re.I) # ignore case
         units = None
         if len(name) > 1:
             units = name[1].strip()
@@ -463,7 +511,7 @@ def _parse_qual(self, data):
 
             # extract energy if SQRT(S) is one of the qualifiers
             if name.startswith('SQRT(S)') and lower(units) in ('gev'):
-                energies = xheader['value'].split(' TO ')
+                energies = re.split(' TO ', xheader['value'], flags=re.I)
                 for energy in energies:
                     try:
                         energy = float(energy)
@@ -486,7 +534,7 @@ def _parse_header(self, data):
         headers = data.split(':')
 
         for header in headers:
-            header = header.split(' IN ')
+            header = re.split(' IN ', header, flags=re.I) # ignore case
             xheader = {'name': header[0].strip()}
             if len(header) > 1:
                 xheader['units'] = header[1].strip()