BNF Name PEG

bnf_name.peg

bnf_name <-  
  ("Gppe " form "_" product_name_word (" "* (ratio / quantity / modifier / product_name_word))*) / 
  ((product_name_word " "?)+ "_" (" "* (!form product_name_word))* " "* form (" "* (ratio / quantity / modifier))*)
  ;


product_name_word <- r'[^\s_]+' ;

ratio <- (quantity / number) ("/" / ";") (ratio / quantity / unit / number);

modifier <- 
  "e/c"/"E/c"/"E/C"/  //Enteric-coating
  "Spec"/             //?
  "Sach"/             //Sachet
  "Lyoph"/
  "Soln"/
  "Paed"/
  "Conc"/
  "Disper"/
  "s/f"/"S/f"/"S/F"/  //Sugar-free?
  "a/f"/"A/f"/"A/F"/  //Additive-free
  "m/r"/"M/r"/"M/R"/  //Modified-release
  "c/r"/"C/r"/"C/R"/  //Controlled-release
  "g/r"/"G/r"/"G/R"/  //Gastro-resistant
  "g/f"/"G/f"/"G/F"/  //Gluten-free
  "Ud"/               //Ut dictum?
  "Vl"/               //?
  "Dil"/              //Dilute
  "Pf"/"Pfs"/         //Pre-filled, syringe
  "Amp"/              //Ampoule
  "Cart"/             //Cartridge
  "Pen"/ 
  "Syrg"/
  "Vl"/               //Vial
  "Pfa"/              //Perforated film absorbent
  "Eff"/              //Effervescent 
  ("(" r'[^\s_\)]+' ")")/ //Flavours and the like
  "SD"/"Sd"/          //Single dose
  "@gn"/
  " In ";               //NOT (just) a concentration AFAICT

form <- (form_additional " "?)* form_base (" "? (form_base / form_additional))* ;

form_additional <- 
  "Aq" / "Viscous" / "Adh" / "Prote" / "Isot" /
  "Ethylsuc" /
  "Postcoital" /
  "Paed" / "Adult" / "Child" / "Children" /
  "Foaming" / "Eff" / "Disper" /
  "Heat" /

  "Ear/Eye" / "Ear/Eye/Nose" / "Ear/Eye/Nsl" / //Or allow "/" to seperate form words

  "Oral" / "Orodisper" / "Oramucosal" / "Chble" /
  "Ophth" / "Eye" / "Subcon" /
  "Ear" /
  "Nsl" / "N/" / "Nasule" / "I/Nsl" / "Nose" /
  "Skin" / "Cutaneous" / "Top" /
  "Transdermal" / "T/Derm" / 
  "Rectal" / "Blad" / "Vag" /
  "I/V" / "I/M" / "Epidural" / "Intrathecal" /
  "Subling" / "Sublingual" / "Buccal"/ "Sub" / "Lyophilisate" /
  "Resp" / "B/A" / "Mist" /

  "Scalp" / "Face" /

  "A/" / "P/" / //Aerosol Pump, Pneumatic Pump
  "Auto-" /
  "Dissolve" / "Absorb" /
  "Bath" / "Shower" / "Cleansing" /
  "Heavy" / "Light" / "Micro-" /
  "Forte" / "Fte" / "Strong" /
  "Soft" / "Dilute" /
  "Medic" / "Ster" / "Steri-" /
  "Retention" /
  "Co " /
  "Ud" /
  "Depot" /
  "V/" /  //Vegicap

  ("Size " r'[A-Z]+') /

  "For"
  ;

form_base <-
  "Capl" / "Cap" / "Tab" / "Loz" / "Pastil" / "Pulvule" / "Divitab" / "Pills" / "Pill" / "Pellets" / "Wafer" / "Spansule" / "Span" /
  "Liq"  / "Syr" / "Linct" / "Rinse" / "Mthwsh" / "Susp" / "Inf" / "Fluid" / "Expect" / "Tinct" / "Elixir" / "Elix" / "Gelcap" /
  "Conc" / "Soln" /
  "Spy" / "spy" / "Aero" /
  "Suppos" / "Supp" / "Enem" /
  "Pess" /
  "Gran" / "Pdrs" / "Pdr" / "Mix" / "Sach" /

  "Inha" / "Inh" / "Inhalator" / "Respule" / "Cylinder" / "Neb" / "Reefer" / "Disk" / "Gas" / "Cyclocap" / "Ventodisk" / "Steripoule" / "Respule" / "R/Cap" / "Becodisk" / "Spincap" /
  "Autohaler" / "Turbohaler" / "Accuhaler" / "Evohaler" /

  "Inj" / "Org" / "Amp" / "Ins" /
  "Patch" /
  "System" / "Delivery System" / "D/System" / "D/system" / "Implant" /
  "Wsh" / "Wash" / "Cleanser" / "Soap" / "Bar" / "Shampoo" /
  "Chewing Gum" /
  "Dps" /
  "Crm" / "Gel" / "Oint" / "Lot" / "Emollient" / "Applic" / "Emuls" / "Emulsif" / "Paste" / "Oil" / "Lin" / "Rub" / "Balm" /
  "(S)" / "(s)" / //Solid
  "Vac" / //Vaccine
  "Foam" /
  "Film" /
  "Irrig" /
  "Strips" / "Strip" /
  "Cryst" /
  "C/Pk" / "S/Pack" / "T/Pk" / "D/P" / "Pack"
  ;

digit <- r'[0-9]';
point <- ".";
float <- digit* point digit+;
integer <- (digit+ ","?)+ ;
number <- float/integer;

quantity <- "("? number " "? unit ")"?;

unit <- 
  metric_unit /
  "%" /
  "hr" "s"? /
  "u" / "d" / "D";

metric_unit <-  metric_prefix ('g'/'l'/'L'/'m'/'mol'/'M');
metric_prefix <- 'mc'/'m'/'';

model_utils.py

from arpeggio import PTNodeVisitor, visit_parse_tree, NoMatch
from arpeggio.peg import ParserPEG
from decimal import *

class BNFNameVisitor(PTNodeVisitor):

    # http://www.nhsbsa.nhs.uk/PrescriptionServices/Documents/Drug_Appliance_Abbreviations_2009.pdf

    def visit_bnf_name(self, node, children):
        bnfname = {'product_name':' '.join(children.product_name_word), 'roa':children.form[0]['roa']}
  
        #
        # Mass, Volume and D's
        #
  
        mass_quantities = [c for c in children.quantity if c['unit'] == 'g']
        if len(mass_quantities) == 1:
          bnfname['mass/g'] = mass_quantities[0]['number']
        elif children.form[0]['is_solid']:
          bnfname['mass/g'] = Decimal(1) #1 g if solid
  
        vol_quantities = [c for c in children.quantity if c['unit'] == 'l']
        if len(vol_quantities) == 1:
          bnfname['vol/l'] = vol_quantities[0]['number']
        elif children.form[0]['is_liquid']:
          bnfname['vol/l'] = Decimal('0.001') #1 ml if liquid

        d_quantities = [c for c in children.quantity if c['unit'] == 'd']
        if len(d_quantities) == 1:
          bnfname['d'] = d_quantities[0]['number']
        else:
          bnfname['d'] = Decimal(1) # 1 tablet/ampoule...
  
        if not ('mass/g' in bnfname):
          if ('vol/l' in bnfname):
            liq_concentrations = [c for c in children.ratio if c['unit'] == 'g/l']
            if len(liq_concentrations) == 1:
              bnfname['mass/g'] = children.ratio[0]['number']*bnfname['vol/l']
  
        if 'mass/g' in bnfname: bnfname['mass/g'] *= bnfname['d']
        if 'vol/l' in bnfname: bnfname['vol/l'] *= bnfname['d']

        return bnfname

    def visit_form(self, node, children):
        form = {'form':''.join(children), 'roa':'U'}
  
        liq_forms = ['Soln','Syr','Susp','Linct','Liq','Mthwsh','Fluid','Elix','Dps','Mix','Gel','Oil']
        sol_forms = ['Crm','Oint','Lot','Emollient','Applic','(S)','Lin','Pdr']
  
        form['is_liquid'] = any(lf in form['form'] for lf in liq_forms)
        form['is_solid'] = any(sf in form['form'] for sf in sol_forms)
  
        roa_dict = {
          'Oral'       : ['Oral', 'Orodisper', 'Oromucosal','Chble'],
          'Ocular'     : ['Eye'],
          'Rectal'     : ['Rectal'],
          'Parenteral' : ['I/V','Implant','I/M','Intrathecal'],
          'Transdermal': ['Transdermal','T/Derm'],
          'Topical'    : ['Top'],
          'Nasal'      : ['Nsl', 'Nasule','N/'],
          'Inhaled'    : ['Inh','Resp','Gas','Neb'],
          'Vaginal'    : ['Vag'],
          'Sublingual' : ['Sub ','Buccal','Subling']
        }

        for roa in roa_dict:
          if any(f in children for f in roa_dict[roa]): form['roa'] = roa
  
  
        if form['roa'] == 'U':
            # Then we try to guess RoA based on the form of the chemical
            roa_dict = {
                'Oral'       : ['Linct','Cap','Capl','Tab','Susp','Soln','Chewing Gum','Divitab', 'Spansule', 'Span', 'Gelcap',
                                'Pastil','Pulvule','Pill','Liq','Syr','Tinct', 'Pellets', 'Wafer','Elixir','Elix'],
                'Parenteral' : ['Inj','Org'],
                'Inhaled'    : ['Autohaler', 'Turbohaler', 'Accuhaler', 'Evohaler',
                                'Respule','Cylinder','Reefer','Disk','B/A','Cyclocap', 'Ventodisk','Steripoule','Respule','R/Cap', 'Spincap'], 
                'Transdermal': ['Patch'],
                'Rectal'     : ['Suppos','Supp','Enem'],
                'Vaginal'    : ['Pess'],
                'Topical'    : ['Crm','Gel','Oint','Lot','Lin'],
                'Sublingual' : ['Lyophilisate']
            }

            for roa in roa_dict:
              if any(f in children for f in roa_dict[roa]): form['roa'] = roa

        return form

    def visit_modifier(self, node, children):
        return ''.join(children).lower()

    def visit_ratio(self, node, children):
  
        qs = [children[0], children[-1]]
  
        def quantify(x): return {'number':x, 'unit':''} if isinstance(x, Decimal) else x
        qs = [quantify(q) for q in qs]
  
        #if qs[1]['unit'] == 'g': #Assume it's a combination product and return the total
        #  total = qs[0]['number'] + qs[1]['number']
        #  unit = 'g'
        #  return {'number':total, 'unit': unit, 'quantities':qs}
  
        #Assume it's a concentration or a rate
        quotient = qs[0]['number']/qs[1]['number']
        unit = "%s/%s" % (qs[0]['unit'], qs[1]['unit'])
        return {'number':quotient, 'unit': unit, 'quantities':qs}

    def visit_quantity(self, node, children):
        number = 1
  
        if len(children.number) > 0:
          number *= children.number[0]
  
        number *= children.unit[0]['number']
  
        return {'number':number, 'unit':children.unit[0]['unit']}

    def visit_unit(self, node, children):
        if len(children.metric_unit) > 0:
            return children.metric_unit[0]
  
        return {'number':Decimal(1), 'unit':''.join(children).lower()}

    def visit_metric_unit(self, node, children):
        return {'number':children[0], 'unit':children[-1].lower()}

    def visit_metric_prefix(self, node, children):
        lookup = {
          'mc' : Decimal('1e-6'),
          'm'  : Decimal('0.001'),
          ''   : Decimal(1)
        }
  
        return lookup.get(''.join(children))

    def visit_number(self, node, children):
        return Decimal(''.join(children))

    def visit_float(self, node, children):
        return ''.join(children)
    
    def visit_integer(self, node, children):
        return ''.join(children.digit)

    def visit_digit(self, node, children):
        return node.value

    def visit_point(self, node, children):
        return "."

bnf_name_parser = ParserPEG(open('./data/bnf_codes/bnf_name.peg','r').read(),'bnf_name',skipws=False)
bnf_name_visitor = BNFNameVisitor()

def dict_for_bnf_name(bnf_name):
    try:
        tree = bnf_name_parser.parse(bnf_name)
        return visit_parse_tree(tree, bnf_name_visitor)
    except NoMatch:
        return None