Gather more values from reference elements if publication-type=other (#…

…1086)
stencila · Oct 23, 2024 · be30792 · be30792
1 parent df787bf
commit be30792
Show file tree

Hide file tree

Showing 18 changed files with 149 additions and 32 deletions.
diff --git a/src/codecs/jats/__file_snapshots__/06908fc3-73df-1014-bb56-a21daa237ef0.json b/src/codecs/jats/__file_snapshots__/06908fc3-73df-1014-bb56-a21daa237ef0.json
@@ -1595,6 +1595,10 @@
         "type": "Date",
         "value": "2015"
       },
+      "isPartOf": {
+        "type": "Periodical",
+        "name": "Nature"
+      },
       "title": "Non-adaptive plasticity potentiates rapid adaptive evolution of gene expression in nature"
     },
     {
@@ -2552,6 +2556,10 @@
         "type": "Date",
         "value": "2021"
       },
+      "isPartOf": {
+        "type": "Periodical",
+        "name": "R package version 1.7.1-1"
+      },
       "title": "emmeans: Estimated Marginal Means, aka Least-Squares Means"
     },
     {
@@ -2714,6 +2722,10 @@
           "value": "XXXX"
         }
       ],
+      "isPartOf": {
+        "type": "Periodical",
+        "name": "bioRxiv"
+      },
       "title": "Variation in mutational (co)variances"
     },
     {
@@ -2780,6 +2792,10 @@
         "type": "Date",
         "value": "2019"
       },
+      "isPartOf": {
+        "type": "Periodical",
+        "name": "bioRxiv"
+      },
       "pageStart": 778282,
       "title": "Phenotypic stasis with genetic divergence"
     },
@@ -3731,6 +3747,10 @@
           "value": "10.1101/506865"
         }
       ],
+      "isPartOf": {
+        "type": "Periodical",
+        "name": "Ecology Letters, page"
+      },
       "title": "How differing modes of non-genetic inheritance affect population viability in fluctuating environments"
     },
     {
@@ -4114,6 +4134,10 @@
         "type": "Date",
         "value": "2018"
       },
+      "isPartOf": {
+        "type": "Periodical",
+        "name": "R package version 2.18.2"
+      },
       "title": "RStan: the R interface to Stan"
     },
     {

diff --git a/src/codecs/jats/__file_snapshots__/48c60452-6c66-1014-adf9-c7b61873ecd3.json b/src/codecs/jats/__file_snapshots__/48c60452-6c66-1014-adf9-c7b61873ecd3.json
@@ -255,6 +255,10 @@
         "type": "Date",
         "value": "2022"
       },
+      "isPartOf": {
+        "type": "Periodical",
+        "name": "Circular Statistics (version 0.4-94)"
+      },
       "title": "R package “circular”"
     },
     {
@@ -1934,6 +1938,10 @@
           "value": "10.1093/jxb/51.349.1403"
         }
       ],
+      "isPartOf": {
+        "type": "Periodical",
+        "name": "J Exp Bot"
+      },
       "title": "Hormonal influence on photocontrol of the protandry in the genus Helianthus"
     },
     {
@@ -2851,7 +2859,7 @@
           "value": "10.3120/0024-9637-56.3.209"
         }
       ],
-      "title": "Patterson B. 2009. Systematics, Evolution, and Biogeography of Compositae, Madroño. doi:10.3120/0024-9637-56.3.209"
+      "title": "Systematics, Evolution, and Biogeography of Compositae, Madroño"
     },
     {
       "type": "Article",

diff --git a/src/codecs/jats/__file_snapshots__/8758d58a-6c3d-1014-87a3-bbc202f7d832.json b/src/codecs/jats/__file_snapshots__/8758d58a-6c3d-1014-87a3-bbc202f7d832.json
@@ -1103,6 +1103,10 @@
         "type": "Date",
         "value": "1988"
       },
+      "isPartOf": {
+        "type": "PublicationVolume",
+        "volumeNumber": 8
+      },
       "title": "T-cell antigen receptor genes and T-cell recognition",
       "meta": {
         "label": "9."
@@ -5685,11 +5689,7 @@
           "value": "10.1146/annurev-immunol-102819-023144"
         }
       ],
-      "isPartOf": {
-        "type": "Periodical",
-        "name": "Annu. Rev. Immunol."
-      },
-      "title": "A. C. Hayday, P. Vantourout, Annu. Rev. Immunol., in press, doi:10.1146/annurev-immunol-102819-023144.",
+      "title": "Annu. Rev. Immunol.",
       "meta": {
         "label": "46."
       }
@@ -6651,6 +6651,10 @@
           ]
         }
       ],
+      "isPartOf": {
+        "type": "PublicationVolume",
+        "volumeNumber": 8
+      },
       "title": "OLGA: fast computation of generation probabilities of B- and T-cell receptor amino acid sequences and motifs",
       "meta": {
         "label": "62."

diff --git a/src/codecs/jats/__file_snapshots__/ca7917ff-6cb0-1014-9b19-d05ef1e56b05.json b/src/codecs/jats/__file_snapshots__/ca7917ff-6cb0-1014-9b19-d05ef1e56b05.json
@@ -3538,13 +3538,9 @@
       },
       "isPartOf": {
         "type": "PublicationVolume",
-        "isPartOf": {
-          "type": "Periodical",
-          "name": "Internal amino acid state modulates yeast taste neurons to support protein homeostasis in Drosophila"
-        },
         "volumeNumber": 29
       },
-      "title": "Steck K, Walker SJ, Itskov PM, Baltazar C, Moreira J-M, Ribeiro C. 2018. Internal amino acid state modulates yeast taste neurons to support protein homeostasis in Drosophila 29."
+      "title": "Internal amino acid state modulates yeast taste neurons to support protein homeostasis in Drosophila"
     },
     {
       "type": "Article",

diff --git a/src/codecs/jats/__file_snapshots__/d6f14042-6db3-1014-b228-f9da7cb8e4cc.json b/src/codecs/jats/__file_snapshots__/d6f14042-6db3-1014-b228-f9da7cb8e4cc.json
@@ -1466,6 +1466,10 @@
         "type": "Date",
         "value": "2021"
       },
+      "isPartOf": {
+        "type": "Periodical",
+        "name": "MedRxiv"
+      },
       "title": "Obesity and accelerated epigenetic aging in a high-risk cohort of children",
       "url": "https://doi.org/10.1101/2021.11.03.21265865"
     },
@@ -3091,6 +3095,10 @@
         "type": "Date",
         "value": "2021"
       },
+      "isPartOf": {
+        "type": "Periodical",
+        "name": "MedRxiv"
+      },
       "title": "BMI is positively associated with accelerated epigenetic aging in twin pairs discordant for BMI",
       "url": "https://doi.org/10.1101/2021.03.11.21253271"
     },

diff --git a/src/codecs/jats/__file_snapshots__/e9291f49-6d06-1014-9500-e156797df82e.json b/src/codecs/jats/__file_snapshots__/e9291f49-6d06-1014-9500-e156797df82e.json
@@ -574,7 +574,7 @@
         "type": "Date",
         "value": "2010"
       },
-      "title": "Andrews, S., 2010. FastQC: a quality control tool for high throughput sequence data."
+      "title": "FastQC: a quality control tool for high throughput sequence data"
     },
     {
       "type": "Article",

diff --git a/src/codecs/jats/__file_snapshots__/elife-30274-v1.jats.xml b/src/codecs/jats/__file_snapshots__/elife-30274-v1.jats.xml
@@ -1920,6 +1920,7 @@
                     <article-title>R: A language and environment for statistical computing</article-title>
                     <person-group person-group-type="author"/>
                     <year iso-8601-date="2017">2017</year>
+                    <source>R Foundation for Statistical Computing</source>
                 </element-citation>
             </ref>
             <ref id="bib26">

diff --git a/src/codecs/jats/__file_snapshots__/elife-30274-v1.yaml b/src/codecs/jats/__file_snapshots__/elife-30274-v1.yaml
@@ -1837,6 +1837,9 @@ references:
     datePublished:
       type: Date
       value: '2017'
+    isPartOf:
+      type: Periodical
+      name: R Foundation for Statistical Computing
     title: 'R: A language and environment for statistical computing'
   - type: Article
     id: bib26

diff --git a/src/codecs/jats/__file_snapshots__/elife-43154-v2.jats.xml b/src/codecs/jats/__file_snapshots__/elife-43154-v2.jats.xml
@@ -1573,6 +1573,7 @@
                         </name>
                     </person-group>
                     <year iso-8601-date="2019">2019</year>
+                    <source>GitHub</source>
                 </element-citation>
             </ref>
             <ref id="bib12">

diff --git a/src/codecs/jats/__file_snapshots__/elife-43154-v2.yaml b/src/codecs/jats/__file_snapshots__/elife-43154-v2.yaml
@@ -1281,6 +1281,9 @@ references:
     datePublished:
       type: Date
       value: '2019'
+    isPartOf:
+      type: Periodical
+      name: GitHub
     title: Severe Malaria
     url: https://github.com/Stije/SevereMalariaAnalysis
   - type: Article

diff --git a/src/codecs/jats/__file_snapshots__/elife-46472-v3.jats.xml b/src/codecs/jats/__file_snapshots__/elife-46472-v3.jats.xml
@@ -4449,6 +4449,7 @@
                         </name>
                     </person-group>
                     <year iso-8601-date="2012">2012</year>
+                    <source>December 12, 2012; Published at US008697359B1 (Awarded)</source>
                 </element-citation>
             </ref>
         </ref-list>

diff --git a/src/codecs/jats/__file_snapshots__/elife-46472-v3.yaml b/src/codecs/jats/__file_snapshots__/elife-46472-v3.yaml
@@ -4559,6 +4559,9 @@ references:
     datePublished:
       type: Date
       value: '2012'
+    isPartOf:
+      type: Periodical
+      name: December 12, 2012; Published at US008697359B1 (Awarded)
     title: Systems methods and compositions for sequence manipulation
     url: https://patents.google.com/patent/WO2015089486A3/en
 title:

diff --git a/src/codecs/jats/__file_snapshots__/elife-46793-v1.jats.xml b/src/codecs/jats/__file_snapshots__/elife-46793-v1.jats.xml
@@ -3942,6 +3942,7 @@
             </ref>
             <ref id="bib49">
                 <element-citation>
+                    <article-title> FASTQ/A Short-Reads Pre-Processing Tools</article-title>
                     <person-group person-group-type="author">
                         <name>
                             <surname>Gordon</surname>
@@ -5121,6 +5122,7 @@
                         </name>
                     </person-group>
                     <year iso-8601-date="2019">2019</year>
+                    <source>GitHub</source>
                 </element-citation>
             </ref>
             <ref id="bib80">

diff --git a/src/codecs/jats/__file_snapshots__/elife-46793-v1.yaml b/src/codecs/jats/__file_snapshots__/elife-46793-v1.yaml
@@ -3236,6 +3236,7 @@ references:
     datePublished:
       type: Date
       value: '2010'
+    title: "\_FASTQ/A Short-Reads Pre-Processing Tools"
     url: http://hannonlab.cshl.edu/fastx_toolkit/download.html
   - type: Article
     id: bib50
@@ -4987,6 +4988,9 @@ references:
     datePublished:
       type: Date
       value: '2019'
+    isPartOf:
+      type: Periodical
+      name: GitHub
     title: GWscreen_G4sensitivity
     url: https://github.com/sblab-bioinformatics/GWscreen_G4sensitivity
   - type: Article

diff --git a/src/codecs/jats/__file_snapshots__/elife-52882-v2.jats.xml b/src/codecs/jats/__file_snapshots__/elife-52882-v2.jats.xml
@@ -11748,6 +11748,7 @@
                         </name>
                     </person-group>
                     <year iso-8601-date="2019">2019</year>
+                    <source>GitLab</source>
                 </element-citation>
             </ref>
             <ref id="bib30">
@@ -11760,6 +11761,7 @@
                         </name>
                     </person-group>
                     <year iso-8601-date="2019">2019</year>
+                    <source>GitLab</source>
                 </element-citation>
             </ref>
             <ref id="bib31">

diff --git a/src/codecs/jats/__file_snapshots__/elife-52882-v2.yaml b/src/codecs/jats/__file_snapshots__/elife-52882-v2.yaml
@@ -1482,6 +1482,9 @@ references:
     datePublished:
       type: Date
       value: '2019'
+    isPartOf:
+      type: Periodical
+      name: GitLab
     title: Programs closed-loop phototaxis
     url: https://gitlab.com/Phiasso/programs_closed-loop_phototaxis.git
   - type: Article
@@ -1495,6 +1498,9 @@ references:
     datePublished:
       type: Date
       value: '2019'
+    isPartOf:
+      type: Periodical
+      name: GitLab
     title: Analysis behavioral phototaxis
     url: https://gitlab.com/Phiasso/analysis.git
   - type: Article

diff --git a/src/codecs/jats/index.ts b/src/codecs/jats/index.ts
@@ -1351,42 +1351,56 @@ export function decodeReference(
 
   let title: string | undefined
   let isPartOf: stencila.CreativeWork | undefined
-  if (publicationType === 'journal' || publicationType === 'preprint') {
+  let chapterTitle: string | undefined
+
+  // Assign the title based upon the publicationType
+  if (['journal', 'preprint'].includes(publicationType ?? '')) {
     title = textOrUndefined(child(elem, 'article-title'))
+  } else if (['book', 'report'].includes(publicationType ?? '')) {
+    chapterTitle = textOrUndefined(child(elem, 'chapter-title'))
+    // Not a chapter so title is <source>
+    title = chapterTitle ?? textOrUndefined(child(elem, 'source'))
+  } else {
+    // e.g. publicationType: 'software', 'web', 'patent'
+    title = textOrUndefined(
+      child(elem, ['article-title', 'chapter-title', 'data-title']),
+    )
+  }
 
-    const periodicalName = textOrUndefined(child(elem, 'source'))
-    const volumeNumber = intOrUndefined(child(elem, 'volume'))
-    const issueNumber = intOrUndefined(child(elem, 'issue'))
-    if (periodicalName !== undefined)
-      isPartOf = stencila.periodical({ name: periodicalName })
-    if (volumeNumber !== undefined)
-      isPartOf = stencila.publicationVolume({ volumeNumber, isPartOf })
-    if (issueNumber !== undefined)
-      isPartOf = stencila.publicationIssue({ issueNumber, isPartOf })
-  } else if (publicationType === 'book' || publicationType === 'report') {
-    title = textOrUndefined(child(elem, 'chapter-title'))
-    if (title !== undefined) {
+  // Assign isPartOf based upon the publicationType
+  if (['book', 'report'].includes(publicationType ?? '')) {
+    if (chapterTitle !== undefined) {
       // Book or report chapter so try to create a `isPartOf` property
       const book = textOrUndefined(child(elem, 'source'))
       if (book !== undefined) {
         isPartOf = stencila.creativeWork({ name: book })
       }
+    }
+  } else {
+    // e.g. publicationType: 'journal', 'preprint', 'software', 'web', 'patent', 'other'
+    if (title !== undefined) {
+      const periodicalName = textOrUndefined(child(elem, 'source'))
+      if (periodicalName !== undefined)
+        isPartOf = stencila.periodical({ name: periodicalName })
     } else {
-      // Not a chapter so title is <source>
       title = textOrUndefined(child(elem, 'source'))
     }
-  } else {
-    // e.g. publicationType: 'software', 'web', 'patent'
-    title = textOrUndefined(
-      child(elem, ['article-title', 'chapter-title', 'data-title']),
-    )
+    const volumeNumber = intOrUndefined(child(elem, 'volume'))
+    const issueNumber = intOrUndefined(child(elem, 'issue'))
+    if (volumeNumber !== undefined)
+      isPartOf = stencila.publicationVolume({ volumeNumber, isPartOf })
+    if (issueNumber !== undefined)
+      isPartOf = stencila.publicationIssue({ issueNumber, isPartOf })
   }
 
   // If title is still undefined and elem is a <mixed-citation> then use
   // its text content e.g.
   //    <mixed-citation publication-type="other" xlink:type="simple">Maynard Smith J (1982) Evolution and the Theory of Games. Cambridge University Press.</mixed-citation>
   if (title === undefined && elem.name === 'mixed-citation') {
     title = textOrUndefined(elem)
+    // Removing authors and isPartOf to reduce duplication
+    authors = []
+    isPartOf = undefined
   }
 
   // Remove any space placeholders