more python example files

2 files changed, 424 insertions, 0 deletions

diff --git a/python/tests/files/crossref-works.single.json b/python/tests/files/crossref-works.single.json
new file mode 100644
index 00000000..2af2b358
--- /dev/null
+++ b/python/tests/files/crossref-works.single.json
@@ -0,0 +1,414 @@
+{
+  "_id": {
+    "$oid": "5a5513fb88a035a45bd63d2e"
+  },
+  "indexed": {
+    "date-parts": [
+      [
+        2017,
+        10,
+        23
+      ]
+    ],
+    "date-time": "2017-10-23T14:15:17Z",
+    "timestamp": {
+      "$numberLong": "1508768117199"
+    }
+  },
+  "reference-count": 25,
+  "publisher": "Wiley-Blackwell",
+  "issue": "4",
+  "license": [
+    {
+      "URL": "http://doi.wiley.com/10.1002/tdm_license_1.1",
+      "start": {
+        "date-parts": [
+          [
+            2015,
+            9,
+            1
+          ]
+        ],
+        "date-time": "2015-09-01T00:00:00Z",
+        "timestamp": {
+          "$numberLong": "1441065600000"
+        }
+      },
+      "delay-in-days": 6452,
+      "content-version": "tdm"
+    }
+  ],
+  "content-domain": {
+    "domain": [],
+    "crossmark-restriction": false
+  },
+  "short-container-title": [
+    "Int. J. Quant. Chem."
+  ],
+  "published-print": {
+    "date-parts": [
+      [
+        1998
+      ]
+    ]
+  },
+  "DOI": "10.1002/(sici)1097-461x(1998)66:4<261::aid-qua1>3.0.co;2-t",
+  "type": "journal-article",
+  "created": {
+    "date-parts": [
+      [
+        2002,
+        8,
+        25
+      ]
+    ],
+    "date-time": "2002-08-25T21:09:51Z",
+    "timestamp": {
+      "$numberLong": "1030309791000"
+    }
+  },
+  "page": "261-272",
+  "source": "Crossref",
+  "is-referenced-by-count": 5,
+  "title": [
+    "Renormalized perturbation theory by the moment method for degenerate states: Anharmonic oscillators"
+  ],
+  "prefix": "10.1002",
+  "volume": "66",
+  "author": [
+    {
+      "given": "Marcelo D.",
+      "family": "Radicioni",
+      "affiliation": []
+    },
+    {
+      "given": "Carlos G.",
+      "family": "Diaz",
+      "affiliation": ["Some University"]
+    },
+    {
+      "given": "Francisco M.",
+      "family": "Fern�ndez",
+      "affiliation": []
+    }
+  ],
+  "editor": [
+    {
+      "given": "A",
+      "family": "B",
+      "affiliation": []
+    }
+  ],
+  "translator": [
+    {
+      "given": "C",
+      "family": "D",
+      "affiliation": []
+    }
+  ],
+  "member": "311",
+  "ISBN": ["85-359-0277-5", "978-3-16-148410-0"],
+  "reference": [
+    {
+      "key": "10.1002/(SICI)1097-461X(1998)66:4<261::AID-QUA1>3.0.CO;2-T-BIB1",
+      "author": "Swenson",
+      "volume": "57",
+      "first-page": "1734",
+      "year": "1972",
+      "journal-title": "J. Chem. Phys.",
+      "DOI": "10.1063/1.1678462",
+      "doi-asserted-by": "crossref"
+    },
+    {
+      "key": "10.1002/(SICI)1097-461X(1998)66:4<261::AID-QUA1>3.0.CO;2-T-BIB2",
+      "author": "Killingbeck",
+      "volume": "65",
+      "first-page": "87",
+      "year": "1978",
+      "journal-title": "Phys. Lett. A",
+      "DOI": "10.1016/0375-9601(78)90580-7",
+      "doi-asserted-by": "crossref"
+    },
+    {
+      "key": "10.1002/(SICI)1097-461X(1998)66:4<261::AID-QUA1>3.0.CO;2-T-BIB3",
+      "author": "Fernández",
+      "volume": "43",
+      "year": "1987",
+      "unstructured": "and Hypervirial Theorems, Lecture Notes in Chemistry, Vol. 43, (Springer, Berlin, 1987).",
+      "volume-title": "Hypervirial Theorems, Lecture Notes in Chemistry",
+      "DOI": "10.1007/978-3-642-93349-3",
+      "doi-asserted-by": "crossref"
+    },
+    {
+      "key": "10.1002/(SICI)1097-461X(1998)66:4<261::AID-QUA1>3.0.CO;2-T-BIB4",
+      "author": "Arteca",
+      "volume": "53",
+      "year": "1990",
+      "unstructured": "and Large Order Perturbation Theory and Summation Methods in Quantum Mechanics, Lecture Notes in Chemistry, Vol. 53, (Springer, Berlin, 1990).",
+      "volume-title": "Large Order Perturbation Theory and Summation Methods in Quantum Mechanics, Lecture Notes in Chemistry",
+      "DOI": "10.1007/978-3-642-93469-8",
+      "doi-asserted-by": "crossref"
+    },
+    {
+      "key": "10.1002/(SICI)1097-461X(1998)66:4<261::AID-QUA1>3.0.CO;2-T-BIB5",
+      "author": "Ader",
+      "volume": "97",
+      "first-page": "178",
+      "year": "1983",
+      "journal-title": "Phys. Lett. A",
+      "DOI": "10.1016/0375-9601(83)90352-3",
+      "doi-asserted-by": "crossref"
+    },
+    {
+      "key": "10.1002/(SICI)1097-461X(1998)66:4<261::AID-QUA1>3.0.CO;2-T-BIB6",
+      "author": "Blankenbecler",
+      "volume": "21",
+      "first-page": "1055",
+      "year": "1980",
+      "journal-title": "Phys. Rev. D",
+      "DOI": "10.1103/PhysRevD.21.1055",
+      "doi-asserted-by": "crossref"
+    },
+    {
+      "key": "10.1002/(SICI)1097-461X(1998)66:4<261::AID-QUA1>3.0.CO;2-T-BIB7",
+      "author": "Fernández",
+      "volume": "26",
+      "first-page": "497",
+      "year": "1984",
+      "journal-title": "Int. J. Quantum Chem.",
+      "DOI": "10.1002/qua.560260408",
+      "doi-asserted-by": "crossref"
+    },
+    {
+      "key": "10.1002/(SICI)1097-461X(1998)66:4<261::AID-QUA1>3.0.CO;2-T-BIB8",
+      "author": "Arteca",
+      "volume": "128",
+      "first-page": "253",
+      "year": "1984",
+      "journal-title": "Physica A",
+      "DOI": "10.1016/0378-4371(84)90090-6",
+      "doi-asserted-by": "crossref"
+    },
+    {
+      "key": "10.1002/(SICI)1097-461X(1998)66:4<261::AID-QUA1>3.0.CO;2-T-BIB9",
+      "author": "Fernández",
+      "volume": "28",
+      "first-page": "603",
+      "year": "1985",
+      "journal-title": "Int. J. Quantum Chem.",
+      "DOI": "10.1002/qua.560280507",
+      "doi-asserted-by": "crossref"
+    },
+    {
+      "key": "10.1002/(SICI)1097-461X(1998)66:4<261::AID-QUA1>3.0.CO;2-T-BIB10",
+      "author": "Fernández",
+      "volume": "20",
+      "first-page": "3777",
+      "year": "1987",
+      "journal-title": "J. Phys. A",
+      "DOI": "10.1088/0305-4470/20/12/027",
+      "doi-asserted-by": "crossref"
+    },
+    {
+      "key": "10.1002/(SICI)1097-461X(1998)66:4<261::AID-QUA1>3.0.CO;2-T-BIB11",
+      "author": "Austin",
+      "volume": "18",
+      "first-page": "449",
+      "year": "1984",
+      "journal-title": "Int. J. Quantum Chem. S",
+      "DOI": "10.1002/qua.560260841",
+      "doi-asserted-by": "crossref"
+    },
+    {
+      "key": "10.1002/(SICI)1097-461X(1998)66:4<261::AID-QUA1>3.0.CO;2-T-BIB12",
+      "author": "Killingbeck",
+      "volume": "18",
+      "first-page": "793",
+      "year": "1985",
+      "journal-title": "J. Phys. A",
+      "DOI": "10.1088/0305-4470/18/5/012",
+      "doi-asserted-by": "crossref"
+    },
+    {
+      "key": "10.1002/(SICI)1097-461X(1998)66:4<261::AID-QUA1>3.0.CO;2-T-BIB13",
+      "author": "Killingbeck",
+      "volume": "19",
+      "first-page": "705",
+      "year": "1986",
+      "journal-title": "J. Phys. A",
+      "DOI": "10.1088/0305-4470/19/5/023",
+      "doi-asserted-by": "crossref"
+    },
+    {
+      "key": "10.1002/(SICI)1097-461X(1998)66:4<261::AID-QUA1>3.0.CO;2-T-BIB14",
+      "author": "Witwit",
+      "volume": "24",
+      "first-page": "3053",
+      "year": "1991",
+      "journal-title": "J. Phys. A",
+      "DOI": "10.1088/0305-4470/24/13/020",
+      "doi-asserted-by": "crossref"
+    },
+    {
+      "key": "10.1002/(SICI)1097-461X(1998)66:4<261::AID-QUA1>3.0.CO;2-T-BIB15",
+      "author": "Witwit",
+      "volume": "24",
+      "first-page": "4535",
+      "year": "1991",
+      "journal-title": "J. Phys. A",
+      "DOI": "10.1088/0305-4470/24/19/017",
+      "doi-asserted-by": "crossref"
+    },
+    {
+      "key": "10.1002/(SICI)1097-461X(1998)66:4<261::AID-QUA1>3.0.CO;2-T-BIB16",
+      "author": "Witwit",
+      "volume": "33",
+      "first-page": "4196",
+      "year": "1992",
+      "journal-title": "J. Math. Phys.",
+      "DOI": "10.1063/1.529818",
+      "doi-asserted-by": "crossref"
+    },
+    {
+      "key": "10.1002/(SICI)1097-461X(1998)66:4<261::AID-QUA1>3.0.CO;2-T-BIB17",
+      "author": "Witwit",
+      "volume": "36",
+      "first-page": "187",
+      "year": "1995",
+      "journal-title": "J. Math. Phys.",
+      "DOI": "10.1063/1.531299",
+      "doi-asserted-by": "crossref"
+    },
+    {
+      "key": "10.1002/(SICI)1097-461X(1998)66:4<261::AID-QUA1>3.0.CO;2-T-BIB18",
+      "author": "Killingbeck",
+      "volume": "14",
+      "first-page": "1005",
+      "year": "1981",
+      "journal-title": "J. Phys. A",
+      "DOI": "10.1088/0305-4470/14/5/020",
+      "doi-asserted-by": "crossref"
+    },
+    {
+      "key": "10.1002/(SICI)1097-461X(1998)66:4<261::AID-QUA1>3.0.CO;2-T-BIB19",
+      "author": "Killingbeck",
+      "volume": "14",
+      "first-page": "l461",
+      "year": "1981",
+      "journal-title": "J. Phys. B",
+      "DOI": "10.1088/0022-3700/14/14/001",
+      "doi-asserted-by": "crossref"
+    },
+    {
+      "key": "10.1002/(SICI)1097-461X(1998)66:4<261::AID-QUA1>3.0.CO;2-T-BIB20",
+      "author": "Fernández",
+      "volume": "25",
+      "first-page": "492",
+      "year": "1992",
+      "journal-title": "J. Phys. A"
+    },
+    {
+      "key": "10.1002/(SICI)1097-461X(1998)66:4<261::AID-QUA1>3.0.CO;2-T-BIB21",
+      "author": "Fernández",
+      "volume": "165",
+      "first-page": "314",
+      "year": "1992",
+      "journal-title": "Phys. Lett. A",
+      "DOI": "10.1016/0375-9601(92)90500-L",
+      "doi-asserted-by": "crossref"
+    },
+    {
+      "key": "10.1002/(SICI)1097-461X(1998)66:4<261::AID-QUA1>3.0.CO;2-T-BIB22",
+      "author": "Fernández",
+      "volume": "178",
+      "first-page": "11",
+      "year": "1993",
+      "journal-title": "Phys. Lett. A",
+      "DOI": "10.1016/0375-9601(93)90719-G",
+      "doi-asserted-by": "crossref"
+    },
+    {
+      "key": "10.1002/(SICI)1097-461X(1998)66:4<261::AID-QUA1>3.0.CO;2-T-BIB23",
+      "author": "Simon",
+      "volume": "58",
+      "first-page": "76",
+      "year": "1970",
+      "journal-title": "Ann. Phys. (N.Y.)",
+      "DOI": "10.1016/0003-4916(70)90240-X",
+      "doi-asserted-by": "crossref"
+    },
+    {
+      "key": "10.1002/(SICI)1097-461X(1998)66:4<261::AID-QUA1>3.0.CO;2-T-BIB24",
+      "author": "Press",
+      "year": "1986",
+      "unstructured": "and Numerical Recipes, (Cambridge University Press, Cambridge, 1986).",
+      "volume-title": "Numerical Recipes"
+    },
+    {
+      "key": "10.1002/(SICI)1097-461X(1998)66:4<261::AID-QUA1>3.0.CO;2-T-BIB25",
+      "author": "Spirko",
+      "volume": "102",
+      "first-page": "8916",
+      "year": "1995",
+      "journal-title": "J. Chem. Phys.",
+      "DOI": "10.1063/1.468945",
+      "doi-asserted-by": "crossref"
+    }
+  ],
+  "container-title": [
+    "International Journal of Quantum Chemistry"
+  ],
+  "link": [
+    {
+      "URL": "https://api.wiley.com/onlinelibrary/tdm/v1/articles/10.1002%2F(SICI)1097-461X(1998)66:4%3C261::AID-QUA1%3E3.0.CO;2-T",
+      "content-type": "unspecified",
+      "content-version": "vor",
+      "intended-application": "text-mining"
+    }
+  ],
+  "deposited": {
+    "date-parts": [
+      [
+        2017,
+        7,
+        23
+      ]
+    ],
+    "date-time": "2017-07-23T04:32:16Z",
+    "timestamp": {
+      "$numberLong": "1500784336000"
+    }
+  },
+  "score": 1,
+  "issued": {
+    "date-parts": [
+      [
+        1998
+      ]
+    ]
+  },
+  "references-count": 25,
+  "URL": "http://dx.doi.org/10.1002/(sici)1097-461x(1998)66:4<261::aid-qua1>3.0.co;2-t",
+  "relation": {
+    "cites": []
+  },
+  "ISSN": [
+    "0020-7608",
+    "1097-461X"
+  ],
+  "issn-type": [
+    {
+      "value": "0020-7608",
+      "type": "print"
+    },
+    {
+      "value": "1097-461X",
+      "type": "electronic"
+    }
+  ],
+  "subject": [
+    "Physical and Theoretical Chemistry",
+    "Atomic and Molecular Physics, and Optics",
+    "Condensed Matter Physics"
+  ]
+}
diff --git a/python/tests/files/matched_sample.json b/python/tests/files/matched_sample.json
new file mode 100644
index 00000000..4213f988
--- /dev/null
+++ b/python/tests/files/matched_sample.json
@@ -0,0 +1,10 @@
+{"size": 354118, "dois": ["10.1137/040604698"], "sha1": "00000088bbc15a03ab89d8da6c356bf25aea9519", "mimetype": "application/pdf", "md5": "f4308d58dc8806232c30edc56a896412", "cdx": [{"url": "http://lma.univ-pau.fr/data/pub/pub_pdf2004/0428.pdf", "dt": 2017}, {"url": "http://lma.univ-pau.fr/data/pub/pub_pdf2004/0428.pdf", "dt": 20170811223928}, {"url": "http://math.univ-lyon1.fr/homes-www/crauste/preprint04_5.pdf", "dt": 2017}, {"url": "http://math.univ-lyon1.fr/homes-www/crauste/preprint04_5.pdf", "dt": 20170809032234}, {"url": "http://math.univ-lyon1.fr/~crauste/preprint04_5.pdf", "dt": 20061118173248}, {"url": "http://math.univ-lyon1.fr/~crauste/preprint04_5.pdf", "dt": 20080221162433}, {"url": "http://math.univ-lyon1.fr/~crauste/preprint04_5.pdf", "dt": 20080419062447}]}
+{"size": null, "dois": ["10.3844/jmssp.2010.395.408"], "sha1": "000002d15f55047daafae1abf7e81c76b64e8062", "mimetype": "application/pdf", "md5": null, "cdx": [{"url": "http://math.uci.edu/%7Edsaari/non-parametric.pdf", "dt": 20080511221813}]}
+{"size": null, "dois": ["10.1112/jlms/jdp006"], "sha1": "0000045687dad717ed6512e395b04ec9c00995b7", "mimetype": "application/pdf", "md5": null, "cdx": [{"url": "http://www.math.purdue.edu/~agabriel/GV07.pdf", "dt": 20100713121639}]}
+{"size": 94354, "dois": ["10.5296/ijhrs.v3i1.3035"], "sha1": "0000075468d9f200713063006d92df8649cf12b5", "mimetype": "application/pdf", "md5": "e68ef08bfca4af6123b7840105444371", "cdx": [{"url": "http://www.macrothink.org/journal/index.php/ijhrs/article/download/3035/2802", "dt": 20150926123746}, {"url": "http://www.macrothink.org/journal/index.php/ijhrs/article/download/3035/2802/", "dt": 2017}]}
+{"size": null, "dois": ["10.2139/ssrn.1739020"], "sha1": "000007656d39cd303103c01d2140888f0f34cb2e", "mimetype": "application/pdf", "md5": null, "cdx": [{"url": "http://aleebrahim.com:80/Conferenc%20Papers/Nader-%20R&D%20Network%20and%20value%20Creation%20in%20SMEs.pdf", "dt": 20110515235622}]}
+{"size": 5645672, "dois": ["10.1371/journal.pntd.0002944"], "sha1": "0000087606c7da34991b73d41a6de3788cafa5d2", "mimetype": "application/pdf", "md5": "c60f5b093ef5e6caad9d7b45268be409", "cdx": [{"url": "http://journals.plos.org/plosntds/article/file?id=10.1371/journal.pntd.0002944&type=printable", "dt": 2017}]}
+{"size": null, "dois": ["10.1111/1462-2920.12063"], "sha1": "00000d6607966f4e7cdbbcff320bd34c37116c72", "mimetype": "application/pdf", "md5": null, "cdx": [{"url": "http://fenske.che.psu.edu/faculty/wood/group/publications/pdf/198MqsRcontrolsGhoSTem2013.pdf", "dt": 20170811001141}]}
+{"size": null, "dois": ["10.1016/j.procs.2014.05.330"], "sha1": "00000d9ee889c1fca84b9d3902b7db3ad9aa273a", "mimetype": "application/pdf", "md5": null, "cdx": [{"url": "https://www.hse.ru/data/2014/07/07/1312079820/ITQM%202014-Dmitriev+Lomazov-2.pdf", "dt": 20170812164945}]}
+{"size": 206812, "dois": ["10.1016/j.cell.2007.04.022"], "sha1": "0000102db78329a149d3b6319f6ccf8cc90483e2", "mimetype": "application/pdf", "md5": "995e7145d09d50eadccf322780e474d3", "cdx": [{"url": "http://publisher-connector.core.ac.uk/resourcesync/data/elsevier/pdf/212/aHR0cDovL2FwaS5lbHNldmllci5jb20vY29udGVudC9hcnRpY2xlL3BpaS9zMDA5Mjg2NzQwNzAwNTI4NA%3D%3D.pdf", "dt": 2017}]}
+{"size": 155208, "dois": ["10.1016/j.jalgebra.2011.10.010"], "sha1": "000012f098031271d16c50deea46d4da1929c418", "mimetype": "application/pdf", "md5": "90d51f06a492cc33cc830da6370248e2", "cdx": [{"url": "http://publisher-connector.core.ac.uk/resourcesync/data/elsevier/pdf/191/aHR0cDovL2FwaS5lbHNldmllci5jb20vY29udGVudC9hcnRpY2xlL3BpaS9zMDAyMTg2OTMxMTAwNTcyMg%3D%3D.pdf", "dt": 2017}]}