morph_simplifier.py

def parse_verb_morphology(morph):
  word_pos = "Verb"
  offset = 7
  
  # Tense Consumption
  if morph[3:7] == "PRES":
    tense = "Present"
  elif morph[3:7] == "IMPF":
    tense = "Imperfect"
  elif morph[3:7] == "PLUP":
    tense = "Pluperfect"
  elif morph[3:7] == "PERF":
    tense = "Pluperfect"
  elif morph[3:7] == "FUTP" and morph[3:13] != 'FUTPASSIVE':
    tense = "FuturePerfect"
  elif morph[3:6] == 'FUT':
    offset = 6
    tense = "Future"
  elif morph == "V99XXX0X":
    # these should we just have a default?
    tense = "Undeclined"
  else:
    tense = "PROBLEM"

  # Voice consumption
  if morph[offset:offset+6] == "ACTIVE":
    voice = 'Active'
    offset += 6
  elif morph[offset:offset+7] == "PASSIVE":
    voice = 'Passive'
    offset += 7
  elif morph[offset:offset+3] == 'IND' or morph[offset:offset+3] == 'SUB' or morph[offset:offset+3] == 'INF' or morph[offset:offset+3] == 'IMP':
    # Deponent verbs - Such verbs occur in passive voice but are translated in active voice.
    voice = 'Active'
  else:
    voice = 'PROBLEM'
  
  # Mood consumption
  if morph[offset:offset+3] == "SUB":
    mood = 'Subjunctive'
  elif morph[offset:offset+3] == "IND":
    mood = 'Indicative'
  elif morph[offset:offset+3] == "IMP":
    mood = 'Imperative'
  elif morph[offset:offset+3] == "INF":
    mood = 'Infinitive'
  else:
    mood = 'PROBLEM'
  offset += 3

  # Person consumption
  if morph[offset] == '0':
    person = 'Undeclined'
  elif morph[offset] == '1':
    person = 'First'
  elif morph[offset] == '2':
    person = 'Second'
  elif morph[offset] == '3':
    person = 'Third'
  else:
    person = "PROBLEM"
  offset += 1

  # Number consumption
  if morph[offset] == 'S':
    number = 'Singular'
  elif morph[offset] == 'P':
    number = 'Plural'
  elif morph[offset] == 'X':
    number = 'Infinitive'
  else:
    number = 'PROBLEM'
  
  return {
      'pos': word_pos,
      'tense': tense,
      'voice': voice,
      'mood': mood,
      'person': person,
      'number': number
  }
  

def parse_verb_participle_morphology(morph):
  word_pos = "Participle"
  offset = 6

  # Case consumption
  if morph[offset:offset+3] == 'NOM':
    v_case = 'Nominative'
  elif morph[offset:offset+3] == 'GEN':
    v_case = 'Genitive'
  elif morph[offset:offset+3] == 'DAT':
    v_case = 'Dative'
  elif morph[offset:offset+3] == 'ACC':
    v_case = 'Accusative'
  elif morph[offset:offset+3] == 'ABL':
    v_case = 'Ablative'
  elif morph[offset:offset+3] == 'VOC':
    v_case = 'Vocative'
  elif morph[offset:offset+3] == 'LOC':
    v_case = 'Locative'
  else:
    v_case = "PROBLEM"
  offset += 3

  # Number consumption
  if morph[offset] == 'S':
    number = 'Singular'
  elif morph[offset] == 'P':
    number = 'Plural'
  elif morph[offset] == 'X':
    number = 'Infinitive'
  else:
    number = 'PROBLEM'
  offset += 1

  # Gender consumption
  if morph[offset] == 'M':
    gender = 'Masculine'
  elif morph[offset] == 'F':
    gender = 'Feminine'
  elif morph[offset] == 'N':
    gender = 'Neuter'
  elif morph[offset] == 'X':
    gender = 'Unknown'
  elif morph[offset] == 'C':
    gender = 'Common'
  else:
    gender = 'PROBLEM'

  offset += 1

  if morph[offset:] == 'FUTPPL' or morph[offset:] == 'FUTACTIVEPPL':
    participle_type = 'FutureActive'
  elif morph[offset:] == 'PRESPPL' or morph[offset:] == 'PRESACTIVEPPL':
    participle_type = 'Present Active'
  elif morph[offset:] == 'PERFPPL' or morph[offset:] == 'PERFACTIVEPPL':
    participle_type = 'PerfectActive'
  elif morph[offset:] == 'FUTPASSIVEPPL':
    participle_type = 'FuturePassive'
  elif morph[offset:] == 'PERFPASSIVEPPL':
    participle_type = 'PerfectPassive'
  else:
    participle_type = 'PROBLEM'

  return {
      'pos': word_pos,
      'case': v_case,
      'number': number,
      'gender': gender,
      'participle_type': participle_type
  }
  

def parse_noun_morphology(morph):
  word_pos = "Noun"
  offset = 3

  # Case consumption
  if morph[offset:offset+3] == 'NOM':
    v_case = 'Nominative'
  elif morph[offset:offset+3] == 'GEN':
    v_case = 'Genitive'
  elif morph[offset:offset+3] == 'DAT':
    v_case = 'Dative'
  elif morph[offset:offset+3] == 'ACC':
    v_case = 'Accusative'
  elif morph[offset:offset+3] == 'ABL':
    v_case = 'Ablative'
  elif morph[offset:offset+3] == 'VOC':
    v_case = 'Vocative'
  elif morph[offset:offset+3] == 'LOC':
    v_case = 'Locative'
  elif morph[offset] == 'X':
    offset -= 2
    v_case = 'Undeclined'
  else:
    v_case = "PROBLEM"
  offset += 3

  # Number consumption
  if morph[offset] == 'S':
    number = 'Singular'
  elif morph[offset] == 'P':
    number = 'Plural'
  elif morph[offset] == 'X':
    number = 'Infinitive'
  else:
    number = 'PROBLEM'
  offset += 1

  # Gender consumption
  if morph[offset] == 'M':
    gender = 'Masculine'
  elif morph[offset] == 'F':
    gender = 'Feminine'
  elif morph[offset] == 'N':
    gender = 'Neuter'
  elif morph[offset] == 'X':
    gender = 'Unknown'
  elif morph[offset] == 'C':
    gender = 'Common'
  else:
    gender = 'PROBLEM'

  return {
      'pos': word_pos,
      'case': v_case,
      'number': number,
      'gender': gender,
  }
  
def parse_adjective_morphology(morph):
  word_pos = "Adjective"
  offset = 5

  # Case consumption
  if morph[offset:offset+3] == 'NOM':
    v_case = 'Nominative'
  elif morph[offset:offset+3] == 'GEN':
    v_case = 'Genitive'
  elif morph[offset:offset+3] == 'DAT':
    v_case = 'Dative'
  elif morph[offset:offset+3] == 'ACC':
    v_case = 'Accusative'
  elif morph[offset:offset+3] == 'ABL':
    v_case = 'Ablative'
  elif morph[offset:offset+3] == 'VOC':
    v_case = 'Vocative'
  elif morph[offset:offset+3] == 'LOC':
    v_case = 'Locative'
  elif morph[offset] == 'X':
    offset -= 2
    v_case = 'Undeclined'
  else:
    v_case = "PROBLEM"
  offset += 3

  # Number consumption
  if morph[offset] == 'S':
    number = 'Singular'
  elif morph[offset] == 'P':
    number = 'Plural'
  elif morph[offset] == 'X':
    number = 'Infinitive'
  else:
    number = 'PROBLEM'
  offset += 1

  # Gender consumption
  if morph[offset] == 'M':
    gender = 'Masculine'
  elif morph[offset] == 'F':
    gender = 'Feminine'
  elif morph[offset] == 'N':
    gender = 'Neuter'
  elif morph[offset] == 'X':
    gender = 'Unknown'
  elif morph[offset] == 'C':
    gender = 'Common'
  else:
    gender = 'PROBLEM'
  offset += 1

  # Comparison consumption
  if morph[offset:] == 'POS':
    comparison = "Positive"
  elif morph[offset:] == 'COMP':
    comparison = "Comparative"
  elif morph[offset:] == 'SUPER':
    comparison = "Superlative"
  elif morph[offset:] == 'X':
    comparison = 'Unknown'
  else:
    comparison = "PROBLEM"
    print(morph)

  return {
      'pos': word_pos,
      'case': v_case,
      'number': number,
      'gender': gender,
      'comparison': comparison
  }
  
# PRON31NOMPM
def parse_pronoun_morphology(morph):
  word_pos = "Pronoun"
  offset = 6

  # Case consumption
  if morph[offset:offset+3] == 'NOM':
    v_case = 'Nominative'
  elif morph[offset:offset+3] == 'GEN':
    v_case = 'Genitive'
  elif morph[offset:offset+3] == 'DAT':
    v_case = 'Dative'
  elif morph[offset:offset+3] == 'ACC':
    v_case = 'Accusative'
  elif morph[offset:offset+3] == 'ABL':
    v_case = 'Ablative'
  elif morph[offset:offset+3] == 'VOC':
    v_case = 'Vocative'
  elif morph[offset:offset+3] == 'LOC':
    v_case = 'Locative'
  elif morph[offset] == 'X':
    offset -= 2
    v_case = 'Undeclined'
  else:
    v_case = "PROBLEM"
  offset += 3

  # Number consumption
  if morph[offset] == 'S':
    number = 'Singular'
  elif morph[offset] == 'P':
    number = 'Plural'
  elif morph[offset] == 'X':
    number = 'Infinitive'
  else:
    number = 'PROBLEM'
  offset += 1

  # Gender consumption
  if morph[offset] == 'M':
    gender = 'Masculine'
  elif morph[offset] == 'F':
    gender = 'Feminine'
  elif morph[offset] == 'N':
    gender = 'Neuter'
  elif morph[offset] == 'X':
    gender = 'Unknown'
  elif morph[offset] == 'C':
    gender = 'Common'
  else:
    gender = 'PROBLEM'

  return {
      'pos': word_pos,
      'case': v_case,
      'number': number,
      'gender': gender,
  }

def parse_preposition_morphology(morph):
  offset = 4
  word_pos = "Preposition"

  if morph[offset:] == 'ABL':
    v_case = 'Ablative'
  elif morph[offset:] == 'ACC':
    v_case = 'Accusative'
  else:
    print(morph[offset:])
    v_case = 'PROBLEM'
  
  return {
      'pos': word_pos,
      'case': v_case
  } 

def parse_adverb_morphology(morph):
  word_pos = "Adverb"
  offset = 3

  if len(morph[offset:]) < 1:
    comparison = "Positive"
  # Comparison consumption
  elif morph[offset:] == 'POS':
    comparison = "Positive"
  elif morph[offset:] == 'COMP':
    comparison = "Comparative"
  elif morph[offset:] == 'SUPER':
    comparison = "Superlative"
  elif morph[offset:] == 'X':
    comparison = 'Unknown'
  else:
    comparison = "PROBLEM"
    print(morph)

  return {
      'pos': word_pos,
      'comparison': comparison
  }
  
# PRON31NOMPM
def parse_supine_morphology(morph):
  word_pos = "Supine"
  offset = 8

  # Case consumption
  if morph[offset:offset+3] == 'NOM':
    v_case = 'Nominative'
  elif morph[offset:offset+3] == 'GEN':
    v_case = 'Genitive'
  elif morph[offset:offset+3] == 'DAT':
    v_case = 'Dative'
  elif morph[offset:offset+3] == 'ACC':
    v_case = 'Accusative'
  elif morph[offset:offset+3] == 'ABL':
    v_case = 'Ablative'
  elif morph[offset:offset+3] == 'VOC':
    v_case = 'Vocative'
  elif morph[offset:offset+3] == 'LOC':
    v_case = 'Locative'
  elif morph[offset] == 'X':
    offset -= 2
    v_case = 'Undeclined'
  else:
    v_case = "PROBLEM"
  offset += 3

  # Number consumption
  if morph[offset] == 'S':
    number = 'Singular'
  elif morph[offset] == 'P':
    number = 'Plural'
  elif morph[offset] == 'X':
    number = 'Infinitive'
  else:
    number = 'PROBLEM'
  offset += 1

  # Gender consumption
  if morph[offset] == 'M':
    gender = 'Masculine'
  elif morph[offset] == 'F':
    gender = 'Feminine'
  elif morph[offset] == 'N':
    gender = 'Neuter'
  elif morph[offset] == 'X':
    gender = 'Unknown'
  elif morph[offset] == 'C':
    gender = 'Common'
  else:
    gender = 'PROBLEM'

  return {
      'pos': word_pos,
      'case': v_case,
      'number': number,
      'gender': gender,
  }
  
def parse_morphology(morph):
  if len(morph) < 2:
    return {}
  # Participle
  if morph[0:4] == "VPAR":
    return parse_verb_participle_morphology(morph)
  
  # Adjective 
  elif morph[0:3] == 'ADJ':
    return parse_adjective_morphology(morph)

  # Adverb 
  elif morph[0:3] == 'ADV':
    return parse_adverb_morphology(morph)
    
  # Verb
  elif morph[0] == 'V' and morph[1] in [str(p) for p in range(9)]:
    return parse_verb_morphology(morph)

  # Noun
  elif morph[0] == 'N':
    return parse_noun_morphology(morph)

  # Pronoun
  elif morph[0:4] == 'PRON':
    return parse_pronoun_morphology(morph)

  # Preposition
  elif morph[0:4] == 'PREP':
    return parse_preposition_morphology(morph)

  # Supine
  elif morph[0:6] == 'SUPINE':
    return parse_supine_morphology(morph)

  elif morph[0:6] == 'INTERJ':
    return {
        'pos': 'Interjection'
    }

  elif morph in ['TACKON', 'PREFIX', 'SUFFIX']:
    return {
        'pos': 'Enclitic'
    }

  elif morph in ['OTHER', 'CONJ']:
    return {
        'pos': morph.capitalize()
    }

def morph_to_string(morph):
    if morph == {} or morph is None:
        return ""
    if morph['pos'] == 'Verb':
        return 'Verb_' + morph['tense'] + "_" + morph['voice'] + "_" + morph['mood'] + "_" + morph['person'] + "_" + morph['number']
    
    elif morph['pos'] == 'Participle':
        return "Participle_" + morph['case'] + "_" + morph['number'] + "_" + morph['gender'] + "_" + morph['participle_type']
    
    elif morph['pos'] == 'Noun':
        return "Noun_" + morph['case'] + "_" + morph['number'] + "_" + morph['gender']
    
    elif morph['pos'] == 'Adjective':
        return 'Adjective_' + morph['case'] + "_" + morph['number'] + "_" + morph['gender'] + '_' + morph['comparison']
    
    elif morph['pos'] == 'Pronoun':
        return "Pronoun_" + morph['case'] + "_" + morph['number'] + "_" + morph['gender'] 
    
    elif morph['pos'] == 'Preposition':
        return "Preposition_" + morph['case']
    
    elif morph['pos'] == 'Adverb':
        return "Adverb_" + morph['comparison']
    
    elif morph['pos'] == 'Supine':
        return "Supine_" + morph['case'] + "_" + morph['number'] + "_" + morph['gender'] 
    
    elif morph['pos'] == 'Enclitic':
        return 'Enclitic'
    
    elif morph['pos'] == 'Other':
        return 'OTHER'
    
    elif morph['pos'] == 'Conj':
        return 'Conjunction'
    
    elif morph['pos'] == 'Interjection':
        return 'Interjection'

def simplify_form(morph):
    return morph_to_string(parse_morphology(morph))