def main(conf):
paperstore, papers_to_add, papers_existing, all_papers = loadEntriesAndSetUp(
conf.input, conf.cache, conf.max)
no_abstract_but_pdf = [
p for p in all_papers if not p.has_abstract and p.has_pdf
]
bulkDownload(no_abstract_but_pdf,
conf.dir,
'results_report.csv',
do_not_download_just_list=True)
successful = []
for paper in no_abstract_but_pdf:
if not os.path.exists(paper.pdf_filename):
continue
abstract = getAbstractFromPDF(paper.pdf_filename)
if abstract:
print(abstract)
paper.bib['abstract'] = abstract
paperstore.updatePapers([paper])
successful.append(paper)
print('Generated', len(successful), 'new abstracts')
writeBibtex(successful, conf.output)
def main(conf):
paperstore, papers_to_add, papers_existing, all_papers = loadEntriesAndSetUp(
conf.input, True)
paperstore.addPapers(papers_to_add)
if conf.force:
paperstore.updatePapers(papers_existing)
def main(conf):
paperstore, papers_to_add, papers_existing, all_papers = loadEntriesAndSetUp(conf.input, conf.cache, conf.max)
if conf.cache:
successful, unsuccessful = enrichAndUpdateMetadata(papers_to_add, paperstore, conf.email)
if conf.force and conf.cache:
enrichAndUpdateMetadata(papers_existing, paperstore, conf.email)
all_papers = papers_to_add + papers_existing
writeOutputBib(all_papers, conf.output)
def main(conf):
paperstore, papers_to_add, papers_existing, all_papers = loadEntriesAndSetUp(
conf.input, conf.cache)
collectStats(all_papers)
included, df = filterPapers(all_papers)
printReport(df)
df.to_csv(conf.report_path)
writeOutputBib(included, conf.output)
return df
def main(conf):
paperstore, papers_to_add, papers_existing, all_papers = loadEntriesAndSetUp(
conf.input, conf.cache)
# successful, unsuccessful = enrichAndUpdateMetadata(papers_to_add, paperstore, conf.email)
snowballed_papers, df = snowballCitations(paperstore, all_papers)
print('Number of snowballed papers:', len(snowballed_papers))
printReport(df)
successful, unsuccessful = enrichAndUpdateMetadata(snowballed_papers,
paperstore, conf.email)
# included, df = filterPapers(snowballed_papers)
# printReport(df)
writeOutputBib(snowballed_papers, conf.output)
def main(conf):
paperstore, papers_to_add, papers_existing, all_papers = loadEntriesAndSetUp(
conf.input, conf.cache)
if conf.missing_abstract:
all_bibs = []
for paper in all_papers:
if not paper.has_pdf and not paper.has_abstract:
all_bibs.append(paper.bib)
elif conf.missing_pdf:
all_bibs = []
for paper in all_papers:
if not paper.has_pdf:
all_bibs.append(paper.bib)
else:
all_bibs = [p.bib for p in all_papers]
writeBibToRISFile(all_bibs, conf.output)
def main(conf):
paperstore, papers_to_add, papers_existing, all_papers = loadEntriesAndSetUp(conf.input, conf.cache, conf.max)
bulkDownload(all_papers, conf.dir, conf.report_path, do_not_download_just_list=False)
def main(conf):
paperstore, papers_to_add, papers_existing, all_papers = loadEntriesAndSetUp(
conf.input, conf.cache)
df = dataframeFromPapers(all_papers)
df.to_csv(conf.output)