Add many options to datafilebuilder input to exclude columns, set defaultBasis (for formats that need it) or add metadata like geometry or set

tools/datafileBuilder.py

@@ -28,7 +28,7 @@ elif args.file!=None:
   else:
     texOps=dfbOptions()
   commands=[LaTeX.newCommand(cmd) for cmd in soup.find_all("newcommand")]
-  dat=LaTeX.tabularToData(soup.tabular,commands)
+  dat=LaTeX.tabularToData(soup.tabular,commands,texOps.excludeColumns)
   scriptpath=Path(sys.argv[0]).resolve()
   datapath=scriptpath.parents[1]/"static"/"data"
   if args.debug:

tools/lib/LaTeX.py

@@ -4,6 +4,7 @@ from .utils import *
 import itertools
 from enum import Enum
 from abc import ABCMeta
+
 class commandBase(metaclass=ABCMeta):
   def __init__(self,source,supportedTexType):
     if not (source.name==supportedTexType):
@@ -15,7 +16,6 @@ class commandBase(metaclass=ABCMeta):
 
   def __repr__(self):
     return self.source.__repr__()
-
 class newCommand(commandBase):
   def __init__(self,source):
     super(newCommand, self).__init__(source,"newcommand")
@@ -96,7 +96,7 @@ class multiColumn(commandBase):
   @property
   def contents(self):
     return nodify(list(self.source.args[2].contents))
-def tabularToData(table,commands=None):
+def tabularToData(table,commands=None,excludeColumn=set()):
   if table.name=="tabular":
     ctable=str(table)
     ctable=ctable.split("\n")
@@ -136,6 +136,7 @@ def tabularToData(table,commands=None):
       raise ValueError("This tabular is not supported because lines have not the same column numbers for each row the coulumns numbers are {}".format(lens))
     import numpy as np
     table=np.array(lnewtable,TexNode)
+    table=np.delete(table,list(excludeColumn),1)
     return table
   else:
     raise ValueError("Only tabular LaTeX environment is supported")

tools/lib/dfbOptions.py

@@ -1,6 +1,6 @@
 from TexSoup import TexSoup,TexCmd
 from . import formats
-from .data import dataFileBase,DataType,state
+from .data import dataFileBase,DataType, method, state, exSet
 from collections import defaultdict
 
 class dfbOptions(object):
@@ -10,6 +10,10 @@ class dfbOptions(object):
     self.suffix=None
     self.initialStates=defaultdict(lambda : state(1,1,"A_1"))
     self.isDouble=False
+    self.defaultBasis="aug-cc-pVTZ"
+    self.geometries = defaultdict(lambda :method("CC3","aug-cc-pVTZ"))
+    self.set = ''
+    self.excludeColumns=set()
   @staticmethod
   def readFromEnv(lateEnv):
     dfb_Opt=dfbOptions()
@@ -37,8 +41,8 @@ class dfbOptions(object):
         vRArgs=[arg.value for arg in initialState.args if arg.type=="required"]
         vOArgs=[arg.value for arg in initialState.args if arg.type=="optional"]
         if len(vOArgs)==0:
-          defaultstate=state.fromString("1 "+vRArgs[0])
-          dfb_Opt.initialStates.default_factory=lambda : defaultstate
+          defaultgeometry=state.fromString("1 "+vRArgs[0])
+          dfb_Opt.initialStates.default_factory=lambda : defaultgeometry
         else:
           mystate=state.fromString("1 "+vRArgs[0])
           dfb_Opt.initialStates[vOArgs[0]]=mystate
@@ -58,4 +62,41 @@ class dfbOptions(object):
             raise ValueError("\isDouble must be 'true' or 'false'.")
         else:
           raise ValueError("Arguments error on '\isDouble'. Only one optional argument is supported.")
+    dfbDefaultBasisNode=lateEnv.defaultBasis
+    if dfbDefaultBasisNode!=None:
+      dfbDefaultBasis=dfbDefaultBasisNode.expr
+      if type(dfbDefaultBasis) is TexCmd:
+        dfb_Opt.defaultBasis=dfbDefaultBasis.args[0].value
+    dfbgeometryNodes=list(lateEnv.find_all("geometry"))
+    for node in dfbgeometryNodes:
+      geometry=node.expr
+      if type(geometry) is TexCmd:
+        vRArgs=[arg.value for arg in geometry.args if arg.type=="required"]
+        vOArgs=[arg.value for arg in geometry.args if arg.type=="optional"]
+        if len(vOArgs)==0:
+          defaultgeometry=method(vRArgs[0],vRArgs[1])
+          dfb_Opt.geometries.default_factory=lambda : defaultgeometry
+        else:
+          mygeometry=method(vRArgs[0],vRArgs[1])
+          dfb_Opt.geometries[vOArgs[0]]=mygeometry
+    dfbSetNode=lateEnv.set
+    if dfbSetNode!=None:
+      dfbSet=dfbSetNode.expr
+      if type(dfbSet) is TexCmd:
+        setname=dfbSet.args[0].value
+        index=dfbSet.args[1].value
+        dfb_Opt.set=exSet(setname,index=int(index))
+    else:
+      dfbStrSetNode=lateEnv.strSet
+      if dfbStrSetNode!=None:
+        dfbStrSet=dfbStrSetNode.expr
+        if type(dfbStrSet) is TexCmd:
+          dfb_Opt.set=setname=dfbStrSet.args[0].value
+    dfbexcludeColumnsNodes=list(lateEnv.find_all("excludecolumns"))
+    for node in dfbexcludeColumnsNodes:
+      excludeColumns=node.expr
+      if type(excludeColumns) is TexCmd:
+        commas_string=excludeColumns.args[0].value
+        ints=[int(x.strip()) for x in commas_string.split(",")]
+        dfb_Opt.excludeColumns.update(ints)
     return dfb_Opt

tools/lib/formats/default/TBEHandler.py

@@ -4,14 +4,14 @@ from ...data import dataFileBase,DataType,method,excitationValue,datafileSelecto
 from ...utils import getValFromCell, checkFloat
 @formatName("TBE")
 class TBEHandler(formatHandlerBase):
-  def readFromTable(self,table):
+  def _readFromTableCore(self,table):
     datalist=list()
     subtablesRange=getSubtablesRange(table)
     for myrange in subtablesRange:
       datacls=dict()
       mymolecule=str(table[myrange[0],0])
       initialState=self.TexOps.initialStates[mymolecule]
-      mymethod=(method("TBE","aug-cc-pVTZ"),method("TBE(Full)","CBS"))
+      mymethod=(method("TBE",self.TexOps.defaultBasis),method("TBE(Full)","CBS"))
       finsts=dataFileBase.convertState(table[myrange,1],initialState,default=self.TexOps.defaultType,commands=self.Commands)
       for index,row in enumerate(table[myrange,]):
         oscilatorForces=checkFloat(str(row[2]))

tools/lib/formats/default/columnHandler.py

@@ -5,7 +5,7 @@ from ...utils import getValFromCell
 import numpy as np
 @formatName("column")
 class columnHandler(formatHandlerBase):
-  def readFromTable(self,table):
+  def _readFromTableCore(self,table):
     datalist=list()
     subtablesRange=getSubtablesRange(table)
     for myrange in subtablesRange:

tools/lib/formats/default/doubleColumnHandler.py

@@ -8,7 +8,7 @@ import numpy as np
 from ...utils import getValFromCell
 @formatName("doubleColumn")
 class doubleColumnHandler(formatHandlerBase):
-  def readFromTable(self,table):
+  def _readFromTableCore(self,table):
     datalist=list()
     datacls=dict()
     subtablesMol=getSubtablesRange(table)

tools/lib/formats/default/doubleTBEHandler.py

@@ -7,7 +7,7 @@ from TexSoup import TexSoup
 import re
 @formatName("doubleTBE")
 class doubleTBEHandler(formatHandlerBase):
-  def readFromTable(self,table):
+  def _readFromTableCore(self,table):
     datalist=list()
     subtablesMol=getSubtablesRange(table)
     for rangeMol in subtablesMol:

tools/lib/formats/default/exoticColumnHandler.py

@@ -8,7 +8,7 @@ import numpy as np
 import json
 @formatName("exoticColumn")
 class exoticColumnHandler(formatHandlerBase):
-  def readFromTable(self,table):
+  def _readFromTableCore(self,table):
     datalist=list()
     subtablesRange=getSubtablesRange(table)
     for myrange in subtablesRange:

tools/lib/formats/default/fromXLSToLaTeXHandler.py

@@ -56,7 +56,7 @@ class fromXLSToLaTeXHandler(formatHandlerBase):
     else:
       return resultstr
 
-  def readFromTable(self,table):
+  def _readFromTableCore(self,table):
     datalist=list()
     subtablesRange=getSubtablesRange(table,firstindex=1,column=1)
     for myrange in subtablesRange:
@@ -65,7 +65,7 @@ class fromXLSToLaTeXHandler(formatHandlerBase):
       initialState=self.TexOps.initialStates[mymolecule]
       for col in itertools.chain(range(8,11), range(14,np.size(table,1))):
         col=table[:,col]
-        basis="aug-cc-pVTZ"
+        basis=self.TexOps.defaultBasis
         mymethcell=list(col[0])
         if len(mymethcell)==0:
           continue

tools/lib/formats/default/lineHandler.py

@@ -5,7 +5,7 @@ from ...utils import getValFromCell
 import numpy as np
 @formatName("line")
 class lineHandler(formatHandlerBase):
-  def readFromTable(self,table):
+  def _readFromTableCore(self,table):
     datalist=list()
     for col in range(1,np.size(table,1)):
       col=table[:,col]

tools/lib/formats/formatHandlerBase.py

@@ -4,5 +4,15 @@ class formatHandlerBase(object, metaclass=ABCMeta):
     self.TexOps=TexOps
     self.Commands=commands
   @abstractmethod
+  def _readFromTableCore(self,table):
+    raise NotImplementedError()
+
   def readFromTable(self,table):
-    raise NotImplementedError()
+    dataFiles=self._readFromTableCore(table)
+    for file in dataFiles:
+      self.__applyMetadataFromOptions(file)
+    return dataFiles
+  def __applyMetadataFromOptions(self,file):
+    if hasattr(file, "geometry"):
+      file.geometry = self.TexOps.geometries[file.molecule]
+    file.set = self.TexOps.set