def ocr_answer(answer_id):
answer = models.Answer.query.get(answer_id)
pages = ocr_url(answer.pdf_url)
answer.text = '\n\n'.join(pages)
models.db.session.add(answer)
models.db.session.commit()
logger.info("done OCR for %s (%d pages)", answer, len(pages))
def ocr_proposal(proposal_id, autoanalyze=False):
proposal = models.Proposal.query.get(proposal_id)
pages = ocr_url(proposal.pdf_url)
proposal.text = '\n\n'.join(pages)
models.db.session.commit()
logger.info("done OCR for %s (%d pages)", proposal, len(pages))
if autoanalyze:
sponsorships = proposal.sponsorships.all()
logger.info("scheduling analysis for %d mandates", len(sponsorships))
for sp in sponsorships:
mandate = sp.mandate
if not mandate.minority:
if (mandate.county is None
or mandate.county.geonames_code is None):
continue
analyze_sponsorship.delay(sp.id)
def ocr_proposal(proposal_id, autoanalyze=False):
proposal = models.Proposal.query.get(proposal_id)
pages = ocr_url(proposal.pdf_url)
proposal.text = '\n\n'.join(pages)
models.db.session.commit()
logger.info("done OCR for %s (%d pages)", proposal, len(pages))
if autoanalyze:
sponsorships = proposal.sponsorships.all()
logger.info("scheduling analysis for %d mandates", len(sponsorships))
for sp in sponsorships:
mandate = sp.mandate
if not mandate.minority:
if (mandate.county is None or
mandate.county.geonames_code is None):
continue
analyze_sponsorship.delay(sp.id)
def ocr_question(question_id, autoanalyze=False):
question = models.Question.query.get(question_id)
pages = ocr_url(question.pdf_url)
question.text = '\n\n'.join(pages)
models.db.session.add(question)
models.db.session.commit()
logger.info("done OCR for %s (%d pages)", question, len(pages))
if autoanalyze:
asked = question.asked.all()
logger.info("scheduling analysis for %d mandates", len(asked))
for ask in asked:
mandate = ask.mandate
if not mandate.minority:
if (mandate.county is None or
mandate.county.geonames_code is None):
continue
analyze.delay(ask.id)
def ocr_question(question_id, autoanalyze=False):
question = models.Question.query.get(question_id)
pages = ocr_url(question.pdf_url)
question.text = '\n\n'.join(pages)
models.db.session.add(question)
models.db.session.commit()
logger.info("done OCR for %s (%d pages)", question, len(pages))
if autoanalyze:
asked = question.asked.all()
logger.info("scheduling analysis for %d mandates", len(asked))
for ask in asked:
mandate = ask.mandate
if not mandate.minority:
if (mandate.county is None or
mandate.county.geonames_code is None):
continue
analyze.delay(ask.id)