[input:bibjson] Update BibJSON support

Update BibJSON input support.

Add docs.

Add tests.

Update README description.

Close #32

README.md

@@ -1,6 +1,6 @@
 <p align="center"><img alt="Citation.js" src="https://citation.js.org/static/img/square_logo_medium.png" /></p>
 
-Citation.js converts formats like BibTeX, Wikidata JSON and ContentMine JSON to CSL-JSON to convert to other formats like APA, Vancouver and back to BibTeX.
+Citation.js converts formats like BibTeX, Wikidata JSON and BibJSON to CSL-JSON to convert to other formats like APA, Vancouver, RIS and back to BibTeX.
 
 ---
 

docs/input_formats.md

@@ -26,7 +26,9 @@ Below is a list of built-in input types.
 
 ## BibJSON
 
-  * `@bibjson/object`: [BibJSON](http://okfnlabs.org/bibjson/) object. Parses the data
+  * `@bibjson/record+object`: [BibJSON](http://okfnlabs.org/bibjson/) object. Parses the data
+  * `@bibjson/quickscrape+record+object`: BibJSON object with some [quickscrape](https://github.com/ContentMine/quickscrape) customisation. Parses the data
+  * `@bibjson/collection+object`: BibJSON object, with a list of records and some collection metadata. Parses the data
 
 ## CSL-JSON
 

src/parse/modules/bibjson/index.js

@@ -4,18 +4,51 @@
 
 import * as json from './json'
 
+let scraperLinks = ['fulltext_html', 'fulltext_xml', 'fulltext_pdf']
+
 export const ref = '@bibjson'
 export const parsers = {json}
 export const formats = {
-  '@bibjson/object': {
-    parse: json.parse,
+  '@bibjson/quickscrape+record+object': {
+    parse: json.quickscrapeRecord,
     parseType: {
-      dataType: 'SimpleObject',
       propertyConstraint: {
-        props: ['fulltext_html', 'fulltext_xml', 'fulltext_pdf'],
+        props: 'link',
+        value (links) {
+          return scraperLinks.some(link => links.find(({type}) => type === link))
+        }
+      },
+      extends: '@bibjson/record+object'
+    }
+  },
+  '@bibjson/record+object': {
+    parse: json.record,
+    parseType: {
+      dataType: 'SimpleObject',
+      propertyConstraint: [{
+        props: 'title'
+      }, {
+        props: ['author', 'editor'],
         match: 'some',
-        value: val => val && Array.isArray(val.value)
-      }
+        value (authors) {
+          return Array.isArray(authors) && authors[0] && 'name' in authors[0]
+        }
+      }]
+    }
+  },
+  '@bibjson/collection+object': {
+    parse (collection) {
+      return collection.records
+    },
+    parseType: {
+      dataType: 'SimpleObject',
+      propertyConstraint: [{
+        props: 'metadata',
+        value (metadata) { return 'collection' in metadata }
+      }, {
+        props: 'records',
+        value (records) { return Array.isArray(records) }
+      }]
     }
   }
 }

src/parse/modules/bibjson/json.js

@@ -5,34 +5,165 @@
 import parseDate from '../../date'
 import parseName from '../../name'
 
+function nameProps (person) {
+  let {
+    firstname,
+    lastname,
+    firstName: given = firstname,
+    lastName: family = lastname
+  } = person
+
+  if (given && family) {
+    return {given, family}
+  } else if (person.name) {
+    return parseName(person.name)
+  }
+}
+
+let identifiers = [
+  'PMID',
+  'PMCID',
+  'DOI',
+  'ISBN'
+  // 'URL' is actually the URL of the record collection, if I understand it correctly,
+  // and not of the record. Otherwise, it should be included.
+]
+
+let journalIdentifiers = [
+  'ISSN'
+]
+
+function idProps (input, identifiers) {
+  let output = {}
+
+  for (let prop in input) {
+    let upperCaseProp = prop.toUpperCase()
+
+    if (identifiers.includes(upperCaseProp)) {
+      output[upperCaseProp] = input[prop]
+    }
+  }
+
+  if (input.identifier) {
+    for (let {id, type = ''} of input.identifier) {
+      type = type.toUpperCase()
+      if (identifiers.includes(type)) {
+        output[type] = id
+      }
+    }
+  }
+
+  return output
+}
+
+// copied from BibTeX, as BibJSON is based on BibTeX
+let typeMap = {
+  article: 'article',
+  book: 'book',
+  booklet: 'book',
+  proceedings: 'book',
+  mastersthesis: 'thesis',
+  inbook: 'chapter',
+  incollection: 'chapter',
+  conference: 'paper-conference',
+  inproceedings: 'paper-conference',
+  online: 'website',
+  patent: 'patent',
+  phdthesis: 'thesis',
+  techreport: 'report',
+  unpublished: 'manuscript',
+
+  // no mapping, fallback
+  manual: undefined,
+  misc: undefined
+}
+
+function quickscrapeSpecificProps () {
+  return { type: 'article-journal' }
+}
+
+function generalProps (input) {
+  let output = {
+    type: typeMap[input.type] || 'book'
+  }
+
+  if (input.title) { output.title = input.title }
+  if (input.author) { output.author = input.author.map(nameProps).filter(Boolean) }
+  if (input.editor) { output.editor = input.editor.map(nameProps).filter(Boolean) }
+  if (input.reviewer) {
+    if (input.author) { output['reviewed-author'] = output.author }
+    output.author = input.reviewer.map(nameProps).filter(Boolean)
+  }
+
+  if (Array.isArray(input.keywords)) {
+    output.keyword = input.keywords.join()
+  } else if (input.keywords) {
+    output.keyword = input.keywords
+  }
+
+  if (input.publisher) { output.publisher = input.publisher.name || input.publisher }
+
+  if (input.date && Object.keys(input.date).length > 0) {
+    let dates = input.date
+    if (dates.submitted) { output.submitted = parseDate(dates.submitted) }
+    if (dates.published) { output.issued = parseDate(dates.published) }
+  } else if (input.year) {
+    output.issued = {'date-parts': [[+input.year]]}
+  }
+  if (input.journal) {
+    let journal = input.journal
+    if (journal.name) { output['container-title'] = journal.name }
+    if (journal.volume) { output.volume = +journal.volume }
+    if (journal.issue) { output.issue = +journal.issue }
+
+    Object.assign(output, idProps(journal, journalIdentifiers))
+
+    if (journal.firstpage) { output['page-first'] = journal.firstpage }
+    if (journal.pages) {
+      output.page = journal.pages.replace('--', '-')
+    } else if (journal.firstpage && journal.lastpage) {
+      output.page = journal.firstpage + '-' + journal.lastpage
+    }
+  }
+
+  if (input.link && typeof input.link[0] === 'object') {
+    output.URL = input.link[0].url
+  }
+
+  Object.assign(output, idProps(input, identifiers))
+
+  if (input.cid) {
+    output.id = input.cid
+  } else if (output.DOI) {
+    output.id = output.DOI
+  }
+
+  return output
+}
+
 /**
- * Forat ContentMine data
+ * Parse ContentMine quickscrape data
  *
  * @access protected
- * @method parseContentMine
- *
  * @param {Object} data - The input data
- *
  * @return {Array<CSL>} The formatted input data
  */
 const parseContentMine = function (data) {
-  const res = {
-    type: 'article-journal'
-  }
-
-  Object.keys(data).forEach((prop) => { res[prop] = data[prop].value[0] })
-
-  if (res.hasOwnProperty('authors')) { res.author = data.authors.value.map(parseName) }
-  if (res.hasOwnProperty('firstpage')) {
-    res.page = res['page-first'] = res.firstpage
-  }
-  if (res.hasOwnProperty('date')) { res.issued = parseDate(res.date) }
-  if (res.hasOwnProperty('journal')) { res['container-title'] = res.journal }
-  if (res.hasOwnProperty('doi')) { res.id = res.DOI = res.doi }
+  return Object.assign(generalProps(data), quickscrapeSpecificProps(data))
+}
 
-  return res
+/**
+ * Parse BibJSON data
+ *
+ * @access protected
+ * @param {Object} data - The input data
+ * @return {Array<CSL>} The formatted input data
+ */
+const parseBibJson = function (data) {
+  return generalProps(data)
 }
 
 export {
-  parseContentMine as parse
+  parseContentMine as quickscrapeRecord,
+  parseBibJson as record
 }

test/input.spec.js

@@ -53,7 +53,8 @@ const configs = {
   },
 
   // @bibjson
-  '@bibjson/object': [input.bibjson.simple, output.bibjson.simple],
+  '@bibjson/quickscrape+record+object': [input.bibjson.quickscrape, output.bibjson.quickscrape],
+  '@bibjson/record+object': [input.bibjson.simple, output.bibjson.simple],
 
   // @else
   '@else/json': {

test/input/parse.json

@@ -49,19 +49,78 @@
   },
   "bibjson": {
     "simple": {
-      "publisher": {"value": ["BioMed Central"]},
-      "journal": {"value": ["Journal of Ethnobiology and Ethnomedicine"]},
-      "title": {"value": ["Gitksan medicinal plants-cultural choice and efficacy"]},
-      "authors": {"value": ["Leslie Main Johnson"]},
-      "date": {"value": ["2006-06-21"]},
-      "doi": {"value": ["10.1186/1746-4269-2-29"]},
-      "volume": {"value": ["2"]},
-      "issue": {"value": ["1"]},
-      "firstpage": {"value": ["1"]},
-      "fulltext_html": {"value": ["http://ethnobiomed.biomedcentral.com/articles/10.1186/1746-4269-2-29"]},
-      "fulltext_pdf": {"value": ["http://ethnobiomed.biomedcentral.com/track/pdf/10.1186/1746-4269-2-29?site=http://ethnobiomed.biomedcentral.com"]},
-      "license": {"value": ["This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited."]},
-      "copyright": {"value": ["2006 Johnson; licensee BioMed Central Ltd."]}
+      "type": "article",
+      "title": "On a family of symmetric Bernoulli convolutions",
+      "author": [
+        {
+          "name": "Erdös, Paul"
+        }
+      ],
+      "journal": {
+        "name": "American Journal of Mathematics",
+        "identifier": [
+          {
+            "id": "0002-9327",
+            "type": "issn"
+          }
+        ],
+        "volume": "61",
+        "pages": "974--976"
+      },
+      "year": "1939",
+      "owner": "me",
+      "id": "ID_1",
+      "collection": "my_collection",
+      "url": "http://example.com/me/my_collection/ID_1",
+      "link":[
+        {
+          "url": "http://okfn.org",
+          "anchor": "Open Knowledge Foundation"
+        }
+      ]
+    },
+    "quickscrape": {
+      "title": "Gitksan medicinal plants-cultural choice and efficacy",
+      "link": [
+        {
+          "type": "fulltext_html",
+          "url": "https://ethnobiomed.biomedcentral.com/articles/10.1186/1746-4269-2-29"
+        },
+        {
+          "type": "fulltext_pdf",
+          "url": "https://ethnobiomed.biomedcentral.com/track/pdf/10.1186/1746-4269-2-29"
+        }
+      ],
+      "author": [
+        {
+          "name": "Leslie Main Johnson",
+          "institution": "Centre for Work and Community Studies and Centre for Integrated Studies, Athabasca University, Athabasca, Canada"
+        },
+        {
+          "institution": "Anthropology Department, University of Alberta, Edmonton, Canada"
+        }
+      ],
+      "publisher": { "name": "BioMed Central" },
+      "journal": {
+        "volume": "2",
+        "issue": "1",
+        "firstpage": "29",
+        "name": "Journal of Ethnobiology and Ethnomedicine",
+        "issn": "1746-4269"
+      },
+      "sections": {
+        "abstract": {
+          "text": "The use of plants for healing by any cultural group is integrally related to local concepts of the nature of disease, the nature of plants, and the world view of the culture. The physical and chemical properties of the plants themselves also bear on their selection by people for medicines, as does the array of plants available for people to choose from. I examine use of medicinal plants from a "
+        },
+        "description": {
+          "text": "The use of plants for healing by any cultural group is integrally related to local concepts of the nature of disease, the nature of plants, and the world view of the culture. The physical and chemical properties of the plants themselves also bear on their selection by people for medicines, as does the array of plants available for people to choose from. I examine use of medicinal plants from a "
+        }
+      },
+      "date": { "published": "2006-06-21T00:00:00+02:00" },
+      "identifier": [ { "type": "doi", "id": "10.1186/1746-4269-2-29" } ],
+      "license": [ { "raw": "This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited." } ],
+      "copyright": [ "2006 Johnson; licensee BioMed Central Ltd." ],
+      "log": [ { "date": "2018-07-12T18:35:20+02:00", "event": "scraped by quickscrape v0.4.7" } ]
     }
   },
   "csl": {