local_pdf_to_json

View Source

 1import json
 2from intelli3text import PipelineBuilder, Intelli3Config
 3
 4cfg = Intelli3Config(export=None)  # no exporters
 5pipeline = PipelineBuilder(cfg).build()
 6res = pipeline.process("papers/sample.pdf")
 7
 8with open("result.json", "w", encoding="utf-8") as f:
 9    json.dump(res, f, ensure_ascii=False, indent=2)
10
11print("Wrote result.json with keys:", list(res.keys()))

cfg = Intelli3Config(cleaners=['ftfy', 'ocr_tilde_fix', 'pdf_breaks', 'pt_diacritics_repair', 'clean_text', 'strip_accents'], lid_primary='fasttext', lid_fallback=None, languages_supported={'es', 'pt', 'en'}, nlp_model_pref='lg', paragraph_min_chars=30, lid_min_chars=60, lid_threshold=0.65, lid_max_chars=2500, lid_use_cld3=True, lid_cld3_weight=0.4, export=None)

pipeline = <intelli3text.pipeline.Pipeline object>

res = {'language_global': 'pt', 'language_mixed': False, 'language_distribution': {}, 'raw': '', 'cleaned': '', 'normalized': '', 'paragraphs': []}