Sign-language / src /NLP_Spacy_base_translator.py
Figea's picture
Update src/NLP_Spacy_base_translator.py
cc8a1d3 verified
import spacy
from ASL_gloss_functions import process_sentence
## custom class for translation
"""
set of rules for ASL conversion:
1. Uppercase Letters: Write each ASL sign in uppercase letters
2. Non-Manual Signals (NMS): Indicate non-manual signals such as facial expressions or body movements above the glossed sign.
3. Fingerspelling: Represent fingerspelled words with dashes between each letter.
4. Lexicalized Fingerspelling: Indicate lexicalized fingerspelling with a # symbol.
5. Repetition: Show repeated signs with a plus sign (+) after the gloss.
6. Role Shift: Indicate role shift with "rs" before the gloss.
7. Indexing/Pointing: Use "ix" followed by a subscript letter or number for indexing.
8. Directional Signs: Indicate the direction of the sign with arrows or other indicators.
9. Classifiers: Use abbreviations for classifiers.
10. Time Indicators: Place time indicators at the beginning of the sentence.
11. Topic-Comment Structure: Indicate the topic followed by the comment.
12. English Words/Concepts: Use English gloss in quotation marks for concepts without direct ASL equivalents.
"""
## reference language
nlp = spacy.load("en_core_web_sm")
class NlpSpacyBaseTranslator():
def __init__(self, sentence):
self.sentence = sentence
def translate_to_gloss(self):
"""
- doc: after nlp processing: I write a sentence for testing Today 17.05 p.m.
- gloss: TODAY
- generated_gloss: TODAY ix_1 I WRITE SENTENCE FOR TEST 17.05 P.M.
"""
print(f'self.sentence: {self.sentence}')
doc = nlp(self.sentence)
##print(f'doc after nlp processing: {doc}')
generated_gloss = process_sentence(doc) ## deterministic model = set of ASL-Gloss-rules functions
print(f'generated_gloss: {generated_gloss}')
return generated_gloss