kausalflow/lightkone / components/bibtex-import/versions/1/examples.jsonl
· raw
{"input": {"text": "@article{Hinton2012-ab,\n title = \"{Improving neural networks by preventing co-adaptation of feature detectors}\",\n author = \"Hinton, Geoffrey E and Srivastava, Nitish and Krizhevsky, Alex\",\n journal = \"arXiv [cs.NE]\",\n year = 2012,\n month = jul,\n eprint = \"1207.0580\",\n archivePrefix = \"arXiv\",\n primaryClass = \"cs.NE\",\n keywords = \"ml;dropout\",\n file = \"All Papers/H/Hinton et al. 2012 - Improving neural networks.pdf\"\n}"}, "output": {"papers": [{"key": "Hinton2012-ab", "type": "article", "title": "Improving neural networks by preventing co-adaptation of feature detectors", "authors": ["Hinton, Geoffrey E", "Srivastava, Nitish", "Krizhevsky, Alex"], "year": 2012, "venue": "arXiv [cs.NE]", "arxiv": "1207.0580", "tags": ["ml", "dropout"], "file": "All Papers/H/Hinton et al. 2012 - Improving neural networks.pdf", "fields": {"month": "July"}}]}, "note": "Paperpile export with arXiv eprint, month macro and file path"}
{"input": {"text": "@inproceedings{vaswani2017attention,\n title={Attention is all you need},\n author={Vaswani, Ashish and Shazeer, Noam and Parmar, Niki and Uszkoreit, Jakob},\n booktitle={Advances in neural information processing systems},\n volume={30},\n year={2017}\n}"}, "output": {"papers": [{"key": "vaswani2017attention", "type": "inproceedings", "title": "Attention is all you need", "authors": ["Vaswani, Ashish", "Shazeer, Noam", "Parmar, Niki", "Uszkoreit, Jakob"], "year": 2017, "venue": "Advances in neural information processing systems", "tags": [], "fields": {"volume": "30"}}]}, "note": "Google Scholar export: booktitle becomes the venue"}
{"input": {"text": "@article{godel1931,\n\tauthor = {G{\\\"o}del, Kurt},\n\ttitle = {{\\\"U}ber formal unentscheidbare S{\\\"a}tze der {P}rincipia {M}athematica und verwandter {S}ysteme {I}},\n\tjournal = {Monatshefte f{\\\"u}r Mathematik und Physik},\n\tyear = {1931},\n\tvolume = {38},\n\tpages = {173--198},\n\tdoi = {10.1007/BF01700692},\n\turl = {https://doi.org/10.1007/BF01700692}\n}"}, "output": {"papers": [{"key": "godel1931", "type": "article", "title": "Über formal unentscheidbare Sätze der Principia Mathematica und verwandter Systeme I", "authors": ["Gödel, Kurt"], "year": 1931, "venue": "Monatshefte für Mathematik und Physik", "doi": "10.1007/BF01700692", "url": "https://doi.org/10.1007/BF01700692", "tags": [], "fields": {"volume": "38", "pages": "173--198"}}]}, "note": "LaTeX accents in braces become Unicode; protective braces go"}
{"input": {"text": "@book{Knuth1984,\n author = {Donald E. Knuth},\n title = {The \\TeX book},\n publisher = {Addison-Wesley},\n year = {1984},\n isbn = {0-201-13447-0}\n}"}, "output": {"papers": [{"key": "Knuth1984", "type": "book", "title": "The \\TeX book", "authors": ["Knuth, Donald E."], "year": 1984, "venue": "Addison-Wesley", "tags": [], "fields": {"publisher": "Addison-Wesley", "isbn": "0-201-13447-0"}}]}, "note": "publisher stands in as venue and stays in fields; unknown macros keep their letters"}
{"input": {"text": "@misc{ligo2016,\n author = {{LIGO Scientific Collaboration} and {Virgo Collaboration} and Abbott, B. P. and others},\n title = {Observation of Gravitational Waves from a Binary Black Hole Merger},\n year = {2016},\n eprint = {arXiv:1602.03837v2},\n eprinttype = {arxiv},\n doi = {https://doi.org/10.1103/PhysRevLett.116.061102}\n}"}, "output": {"papers": [{"key": "ligo2016", "type": "misc", "title": "Observation of Gravitational Waves from a Binary Black Hole Merger", "authors": ["LIGO Scientific Collaboration", "Virgo Collaboration", "Abbott, B. P.", "others"], "year": 2016, "doi": "10.1103/PhysRevLett.116.061102", "arxiv": "1602.03837", "tags": [], "fields": {}}]}, "note": "corporate authors in braces are kept; others; eprinttype; doi URL prefix removed"}
{"input": {"text": "@ARTICLE{Erdos1959,\n AUTHOR = {Erd\\H{o}s, Paul and R\\'{e}nyi, Alfr\\'ed},\n TITLE = {On random graphs {I}},\n JOURNAL = {Publicationes Mathematicae},\n YEAR = {1959},\n NOTE = {Reprinted 1976}\n}"}, "output": {"papers": [{"key": "Erdos1959", "type": "article", "title": "On random graphs I", "authors": ["Erdős, Paul", "Rényi, Alfréd"], "year": 1959, "venue": "Publicationes Mathematicae", "tags": [], "notes": "Reprinted 1976", "fields": {}}]}, "note": "upper-case names; \\H and \\' accents; note becomes notes"}
{"input": {"text": "@ARTICLE{Erd\\H{o}s1959,\n AUTHOR = {Erd\\H{o}s, Paul and R\\'{e}nyi, Alfr\\'ed},\n TITLE = {On random graphs {I}},\n JOURNAL = {Publicationes Mathematicae},\n YEAR = {1959},\n NOTE = {Reprinted 1976}\n}"}, "output": {"papers": [{"key": "Erd\\H{o}s1959", "type": "article", "title": "On random graphs I", "authors": ["Erdős, Paul", "Rényi, Alfréd"], "year": 1959, "venue": "Publicationes Mathematicae", "tags": [], "notes": "Reprinted 1976", "fields": {}}]}, "note": "braces in a key are part of the key"}
{"input": {"text": "@string{prl = \"Physical Review Letters\"}\n@string{ PRD = {Physical Review D} }\n@article{a, title = {One}, journal = prl, year = 2001}\n@article{b, title = {Two}, journal = PRD # \", Part \" # {2}, year = \"circa 1999\"}"}, "output": {"papers": [{"key": "a", "type": "article", "title": "One", "authors": [], "year": 2001, "venue": "Physical Review Letters", "tags": [], "fields": {}}, {"key": "b", "type": "article", "title": "Two", "authors": [], "year": 1999, "venue": "Physical Review D, Part 2", "tags": [], "fields": {}}]}, "note": "@string macros, case-insensitive, with # concatenation; year from the first four digits"}
{"input": {"text": "@comment{jabref-meta: databaseType:bibtex;}\n@preamble{ \"\\newcommand{\\noopsort}[1]{}\" }\n% a line comment\n@misc{x1, title = {Kept}}"}, "output": {"papers": [{"key": "x1", "type": "misc", "title": "Kept", "authors": [], "tags": [], "fields": {}}]}, "note": "@comment and @preamble are not entries"}
{"input": {"text": "@comment{@article{inside, title = {Found inside a comment}}}\n@misc{after, title = {After}}"}, "output": {"papers": [{"key": "inside", "type": "article", "title": "Found inside a comment", "authors": [], "tags": [], "fields": {}}, {"key": "after", "type": "misc", "title": "After", "authors": [], "tags": [], "fields": {}}]}, "note": "scanning continues inside @comment, so an entry there is read"}
{"input": {"text": "@article(paren2020,\n title = \"Entries may use parentheses\",\n author = \"Jane Doe and John Q. Public\"\n)"}, "output": {"papers": [{"key": "paren2020", "type": "article", "title": "Entries may use parentheses", "authors": ["Doe, Jane", "Public, John Q."], "tags": [], "fields": {}}]}, "note": "parenthesised entry; First Last names flip"}
{"input": {"text": "@misc{notitle, author = {Nobody}}\n@misc{emptytitle, title = {}}\n@misc{spaces, title = { }}\n@misc{, title = {No key}}"}, "output": {"papers": [{"key": "", "type": "misc", "title": "No key", "authors": [], "tags": [], "fields": {}}]}, "note": "entries without a title are dropped, even with an empty one; a missing key is kept as empty"}
{"input": {"text": "@article{ws,\n title = {A title\n broken over\n lines},\n abstract = {First line.\n Second line with {\\em emphasis} and \\emph{more} and 50\\% off \\& \\$5.}\n}"}, "output": {"papers": [{"key": "ws", "type": "article", "title": "A title broken over lines", "authors": [], "abstract": "First line. Second line with \\em emphasis and more and 50% off & $5.", "tags": [], "fields": {}}]}, "note": "whitespace collapses; \\emph unwraps; escaped specials"}
{"input": {"text": "@article{latex,\n title = {Na\\\"{\\i}ve {\\ss}tra{\\ss}e, \\o{}resund, {\\AE}sir, \\l{}\\'od\\'z and \\c{c}a va},\n author = {M\\\"uller, J\\\"org and Nu\\~{n}ez, Pe\\~na},\n journal = {Caf\\'e \\& Croissant}\n}"}, "output": {"papers": [{"key": "latex", "type": "article", "title": "Naïve ßtraße, øresund, Æsir, łódź and ça va", "authors": ["Müller, Jörg", "Nuñez, Peña"], "venue": "Café & Croissant", "tags": [], "fields": {}}]}, "note": "accents without braces, on \\i, \\ss, \\o, \\AE, \\l, cedilla"}
{"input": {"text": "@article{kw,\n title = {Keywords},\n keywords = {Machine Learning, physics; doi:10.1/abc; URL: http://x.org; arXiv:2101.00001; ISBN:123; https://example.org; {Quantum} Fields}\n}"}, "output": {"papers": [{"key": "kw", "type": "article", "title": "Keywords", "authors": [], "tags": ["Machine Learning", "physics", "Quantum Fields"], "fields": {}}]}, "note": "keywords split on , and ; with identifiers dropped"}
{"input": {"text": "@article{arxurl,\n title = {From a URL},\n url = {https://arxiv.org/abs/2106.09685v3?context=cs}\n}\n@article{arxpdf,\n title = {From a PDF URL},\n url = {http://arXiv.org/pdf/hep-th/9711200v1.pdf}\n}\n@article{arxid, title = {arxivid field}, arxivid = {1312.6114}}\n@article{bare, title = {Bare new-style eprint}, eprint = {2203.02155}}\n@article{other, title = {Other eprint}, eprint = {hep-th/9711200}}"}, "output": {"papers": [{"key": "arxurl", "type": "article", "title": "From a URL", "authors": [], "arxiv": "2106.09685", "url": "https://arxiv.org/abs/2106.09685v3?context=cs", "tags": [], "fields": {}}, {"key": "arxpdf", "type": "article", "title": "From a PDF URL", "authors": [], "arxiv": "hep-th/9711200", "url": "http://arXiv.org/pdf/hep-th/9711200v1.pdf", "tags": [], "fields": {}}, {"key": "arxid", "type": "article", "title": "arxivid field", "authors": [], "arxiv": "1312.6114", "tags": [], "fields": {}}, {"key": "bare", "type": "article", "title": "Bare new-style eprint", "authors": [], "arxiv": "2203.02155", "tags": [], "fields": {}}, {"key": "other", "type": "article", "title": "Other eprint", "authors": [], "tags": [], "fields": {"eprint": "hep-th/9711200"}}]}, "note": "arXiv ids from url, arxivid and bare eprints; old-style eprint without prefix stays in fields"}
{"input": {"text": "@article{dup,\n title = {First},\n title = {Second},\n Pages = {1-2},\n pages = {3-4}\n}"}, "output": {"papers": [{"key": "dup", "type": "article", "title": "Second", "authors": [], "tags": [], "fields": {"pages": "3-4"}}]}, "note": "a later field replaces an earlier one"}
{"input": {"text": "@article{malformed,\n title = {Survives},\n this is not a field,\n = {no name},\n volume = 7,\n note\n}"}, "output": {"papers": [{"key": "malformed", "type": "article", "title": "Survives", "authors": [], "tags": [], "fields": {"volume": "7"}}]}, "note": "malformed fields are skipped, and the fields after them are read"}
{"input": {"text": "@article{editor, title = {Edited}, editor = {Ada Lovelace and Charles Babbage}}\n@book{both, title = {Both}, author = {Real Author}, editor = {An Editor}}\n@book{emptyauthor, title = {Empty author}, author = {}, editor = {Not Used}}"}, "output": {"papers": [{"key": "editor", "type": "article", "title": "Edited", "authors": ["Lovelace, Ada", "Babbage, Charles"], "tags": [], "fields": {}}, {"key": "both", "type": "book", "title": "Both", "authors": ["Author, Real"], "tags": [], "fields": {"editor": "An Editor"}}, {"key": "emptyauthor", "type": "book", "title": "Empty author", "authors": [], "tags": [], "fields": {"editor": "Not Used"}}]}, "note": "editor stands in only when there is no author field"}
{"input": {"text": "@article{annote, title = {Annote}, annote = {my notes}, note = {publisher note}}\n@article{emptyfields, title = {Empty}, abstract = {}, doi = {}, url = {}, file = {}, journal = {}}"}, "output": {"papers": [{"key": "annote", "type": "article", "title": "Annote", "authors": [], "tags": [], "notes": "my notes", "fields": {"note": "publisher note"}}, {"key": "emptyfields", "type": "article", "title": "Empty", "authors": [], "url": "", "tags": [], "file": "", "fields": {}}]}, "note": "annote wins over note; empty abstract, doi and venue are absent but empty url and file are kept"}
{"input": {"text": "@article{quoted, title = \"A {\"}quoted{\"} word and {braced \"quote\"}\", author = \"O'Neil, Cathy\"}"}, "output": {"papers": [{"key": "quoted", "type": "article", "title": "A \"quoted\" word and braced \"quote\"", "authors": ["O'Neil, Cathy"], "tags": [], "fields": {}}]}, "note": "quotes inside braces do not end a quoted value"}
{"input": {"text": "@article{nested, title = {Outer {Inner {Deepest}} back}, journal = {J. {\\\"O}sterr.}}"}, "output": {"papers": [{"key": "nested", "type": "article", "title": "Outer Inner Deepest back", "authors": [], "venue": "J. Österr.", "tags": [], "fields": {}}]}, "note": "nested braces"}
{"input": {"text": "@article{year0, title = {Year}, year = {0000}}\n@article{year5, title = {Year}, year = {19999}}\n@article{yearword, title = {Year}, year = {forthcoming}}"}, "output": {"papers": [{"key": "year0", "type": "article", "title": "Year", "authors": [], "year": 0, "tags": [], "fields": {}}, {"key": "year5", "type": "article", "title": "Year", "authors": [], "year": 1999, "tags": [], "fields": {}}, {"key": "yearword", "type": "article", "title": "Year", "authors": [], "tags": [], "fields": {}}]}, "note": "year: first four digits as a number, even 0; no digits means absent"}
{"input": {"text": "@article{unterminated, title = {This value never ends"}, "output": {"papers": [{"key": "unterminated", "type": "article", "title": "This value never ends", "authors": [], "tags": [], "fields": {}}]}, "note": "an unterminated value runs to the end"}
{"input": {"text": ""}, "output": {"papers": []}, "note": "empty text"}
{"input": {"text": "no entries at all, just @ signs @ here"}, "output": {"papers": []}, "note": "an @ not followed by a type and brace is skipped"}
{"input": {"text": "@article{mixedcase, Title = {Mixed}, AUTHOR = {de la Fontaine, Jean and Ludwig van Beethoven}, Journal = {X}, MONTH = dec}"}, "output": {"papers": [{"key": "mixedcase", "type": "article", "title": "Mixed", "authors": ["de la Fontaine, Jean", "Beethoven, Ludwig van"], "venue": "X", "tags": [], "fields": {"month": "December"}}]}, "note": "particles: First Last flips only the last word"}
{"input": {"text": "@article{ids, title = {T}, doi = {HTTP://DX.DOI.ORG/10.5555/abc}, eprint = {ARXIV:2101.00001V2}, archiveprefix = {ArXiv}}"}, "output": {"papers": [{"key": "ids", "type": "article", "title": "T", "authors": [], "doi": "10.5555/abc", "arxiv": "2101.00001V2", "tags": [], "fields": {}}]}, "note": "case-insensitive prefixes; an upper-case version suffix stays"}
{"input": {"text": "@article{k, title={Kept}, junk here, year = 1999, note={also kept}}"}, "output": {"papers": [{"key": "k", "type": "article", "title": "Kept", "authors": [], "year": 1999, "tags": [], "notes": "also kept", "fields": {}}]}, "note": "a malformed field ending at a comma: later fields survive"}
{"input": {"text": "@article{k, title={Kept}, = {x}, year = 1999}\n@book{b, title={Next entry}}"}, "output": {"papers": [{"key": "k", "type": "article", "title": "Kept", "authors": [], "year": 1999, "tags": [], "fields": {}}, {"key": "b", "type": "book", "title": "Next entry", "authors": [], "tags": [], "fields": {}}]}, "note": "a malformed field's braced value is skipped whole, so later fields survive"}
{"input": {"text": "@article{k, title = {A
B C}, author = {Doe, Jane and Roe, R.}}"}, "output": {"papers": [{"key": "k", "type": "article", "title": "A B C", "authors": ["Doe, Jane", "Roe, R."], "tags": [], "fields": {}}]}, "note": "Unicode whitespace (no-break space, line separator, ideographic space, BOM) counts as whitespace"}
{"input": {"text": "@article{k, title={A
B\u001fC}}"}, "output": {"papers": [{"key": "k", "type": "article", "title": "A
B\u001fC", "authors": [], "tags": [], "fields": {}}]}, "note": "U+0085 and U+001F are not whitespace"}
{"input": {"text": "@article{k, title={From a URL}, url={HTTPS://ARXIV.ORG/ABS/2106.09685V3}}"}, "output": {"papers": [{"key": "k", "type": "article", "title": "From a URL", "authors": [], "arxiv": "2106.09685", "url": "HTTPS://ARXIV.ORG/ABS/2106.09685V3", "tags": [], "fields": {}}]}, "note": "the arXiv URL match ignores case, version suffix included"}
{"input": {"text": "@article{k, title={From a PDF URL}, url={https://arxiv.org/pdf/2106.09685v2.PDF}}"}, "output": {"papers": [{"key": "k", "type": "article", "title": "From a PDF URL", "authors": [], "arxiv": "2106.09685", "url": "https://arxiv.org/pdf/2106.09685v2.PDF", "tags": [], "fields": {}}]}, "note": "an upper-case .PDF suffix is removed too"}
{"input": {"text": "@article{k, title={Empty values}, file = {}, url = \"\"}"}, "output": {"papers": [{"key": "k", "type": "article", "title": "Empty values", "authors": [], "url": "", "tags": [], "file": "", "fields": {}}]}, "note": "empty url and file stay as empty strings"}
{"input": {"text": "@art icle{k, title={Lost}}\n@misc{m, title={Found}}"}, "output": {"papers": [{"key": "m", "type": "misc", "title": "Found", "authors": [], "tags": [], "fields": {}}]}, "note": "a no-break space ends the entry type, so the first entry is skipped"}
{"input": {"text": "@article{y, title = {Two-digit year}, year = {99}}\n@article{z, title = {Year in words}, year = {circa 1999}}"}, "output": {"papers": [{"key": "y", "type": "article", "title": "Two-digit year", "authors": [], "tags": [], "fields": {}}, {"key": "z", "type": "article", "title": "Year in words", "authors": [], "year": 1999, "tags": [], "fields": {}}]}, "note": "year needs four digits in a row: 99 gives no year; circa 1999 gives 1999"}
{"input": {"text": "@inproceedings{a, title={A}, journal={}, booktitle={NeurIPS}}\n@inproceedings{b, title={B}, journal={ {} }, booktitle={ICML}, publisher={PMLR}}"}, "output": {"papers": [{"key": "a", "type": "inproceedings", "title": "A", "authors": [], "venue": "NeurIPS", "tags": [], "fields": {}}, {"key": "b", "type": "inproceedings", "title": "B", "authors": [], "venue": "ICML", "tags": [], "fields": {"publisher": "PMLR"}}]}, "note": "an empty journal does not hide the booktitle; the booktitle becomes the venue and leaves the fields"}
{"input": {"text": "@book{c, title={C}, journal={}, booktitle={ }, publisher={MIT Press}}\n@article{d, title={D}, journal={JMLR}, booktitle={NeurIPS}}\n@misc{e, title={E}, journal={}, booktitle={}}"}, "output": {"papers": [{"key": "c", "type": "book", "title": "C", "authors": [], "venue": "MIT Press", "tags": [], "fields": {"publisher": "MIT Press"}}, {"key": "d", "type": "article", "title": "D", "authors": [], "venue": "JMLR", "tags": [], "fields": {"booktitle": "NeurIPS"}}, {"key": "e", "type": "misc", "title": "E", "authors": [], "tags": [], "fields": {}}]}, "note": "empty journal and booktitle fall through to the publisher, which stays in the fields; a booktitle not used as the venue stays too; all empty means no venue"}