app/lib/publications/exports.mjs

Functions

Imports

It imports app/data/organisms.ts, app/lib/publications/authors.mjs, app/lib/publications/entities.mjs, app/lib/site-origin.mjs. Imported by app/lib/publications/coins.mjs, app/lib/publications/compile.mjs, app/routes/publications.$slug[.bib].ts, app/routes/publications.$slug[.ris].ts, app/routes/publications[.bib].ts, app/routes/publications[.json].ts, app/routes/publications[.ris].ts.

Source

1// Pure functions of the committed corpus (no clock, no citation count), so check:machine-readable can
2// compare bytes and two downloads match. Author names go through `canonicalAuthor` in authors.mjs.
3
4import { ORGANISM_PATTERN } from "../../data/organisms.ts";
5import { canonicalAuthor } from "./authors.mjs";
6import { decodeEntities } from "./entities.mjs";
7import { SITE_ORIGIN } from "../site-origin.mjs";
8
9export { canonicalAuthor };
10
11/**
12 * Best effort; the CSL export does not use this, it reads the structured given and family fields.
13 *
14 * @param {string} name
15 */
16function splitName(name) {
17  const parts = canonicalAuthor(name).split(" ").filter(Boolean);
18  if (parts.length < 2) return { family: parts[0] ?? "", given: "" };
19  return {
20    family: parts[parts.length - 1] ?? "",
21    given: parts.slice(0, -1).join(" "),
22  };
23}
24
25// The backslash goes first, or it escapes the escapes that follow it.
26/** @type {[string, string][]} */
27const BIBTEX_ESCAPES = [
28  ["\\", "\\textbackslash{}"],
29  ["{", "\\{"],
30  ["}", "\\}"],
31  ["&", "\\&"],
32  ["%", "\\%"],
33  ["$", "\\$"],
34  ["#", "\\#"],
35  ["_", "\\_"],
36  ["~", "\\textasciitilde{}"],
37  ["^", "\\textasciicircum{}"],
38];
39
40/** @param {string} value */
41function escapeBibtex(value) {
42  let out = decodeEntities(value);
43  for (const [from, to] of BIBTEX_ESCAPES) out = out.split(from).join(to);
44  return out;
45}
46
47/**
48 * Many BibTeX styles lowercase titles, and a lowercased genus is wrong under the nomenclature codes.
49 * Applied after escaping, or the braces would be escaped into visible punctuation.
50 *
51 * @param {string} escaped a string already through `escapeBibtex`
52 */
53function protectOrganisms(escaped) {
54  return escaped.replace(ORGANISM_PATTERN, "{$1}");
55}
56
57/**
58 * For check:machine-readable. A substring test is wrong: Mycobacterium smegmatis contains the bare
59 * genus, which the matcher correctly never braces.
60 *
61 * @param {string} text
62 * @returns {string[]} each match, in order, with duplicates kept
63 */
64export function organismsIn(text) {
65  return [...String(text ?? "").matchAll(ORGANISM_PATTERN)].map((m) => m[0]);
66}
67
68// misc, not electronic, which the standard styles render as nothing.
69const BIBTEX_TYPE = {
70  article: "article",
71  review: "article",
72  abstract: "inproceedings",
73  chapter: "incollection",
74  "teaching-resource": "misc",
75};
76
77/**
78 * The key is the record's curated id: a second key scheme would give one work two identities.
79 *
80 * @param {any} paper
81 */
82export function toBibtex(paper) {
83  const type = BIBTEX_TYPE[/** @type {keyof typeof BIBTEX_TYPE} */ (paper.type)] ?? "article";
84  const field = (/** @type {string} */ k, /** @type {string} */ v) =>
85    `  ${k} = {${v}},`;
86
87  const lines = [`@${type}{${paper.id},`];
88  lines.push(field("title", protectOrganisms(escapeBibtex(paper.title))));
89  lines.push(
90    field(
91      "author",
92      paper.authors
93        .map((/** @type {string} */ n) => escapeBibtex(canonicalAuthor(n)))
94        .join(" and "),
95    ),
96  );
97  if (paper.journal) {
98    const key = type === "incollection" ? "booktitle" : "journal";
99    lines.push(field(key, protectOrganisms(escapeBibtex(paper.journal))));
100  }
101  if (paper.year) lines.push(field("year", String(paper.year)));
102  if (paper.volume) lines.push(field("volume", escapeBibtex(String(paper.volume))));
103  if (paper.issue) lines.push(field("number", escapeBibtex(String(paper.issue))));
104  // Verbatim, not rebuilt from first and last: article-number records have no first page.
105  if (paper.pages) lines.push(field("pages", escapeBibtex(String(paper.pages))));
106  // As deposited: lowercasing would make the export disagree with the registry.
107  // A manuscript with no DOI yet has neither line, rather than a registry address that does not exist.
108  if (paper.doi) {
109    lines.push(field("doi", paper.doi));
110    lines.push(field("url", `https://doi.org/${paper.doi}`));
111  }
112  lines.push("}");
113  return lines.join("\n");
114}
115
116// TY first, ER last, exactly two spaces before the dash: reference managers are strict about that.
117const RIS_TYPE = {
118  article: "JOUR",
119  review: "JOUR",
120  abstract: "ABST",
121  chapter: "CHAP",
122  "teaching-resource": "ELEC",
123};
124
125/** @param {any} paper */
126export function toRis(paper) {
127  /** @type {[string, string][]} */
128  const rows = [["TY", RIS_TYPE[/** @type {keyof typeof RIS_TYPE} */ (paper.type)] ?? "JOUR"]];
129  rows.push(["TI", decodeEntities(paper.title)]);
130  for (const name of paper.authors) {
131    const { family, given } = splitName(name);
132    rows.push(["AU", given ? `${family}, ${given}` : family]);
133  }
134  if (paper.journal) rows.push(["JO", decodeEntities(paper.journal)]);
135  if (paper.year) rows.push(["PY", String(paper.year)]);
136  // RIS wants slashes and tolerates a trailing empty field, which is how a partial date is expressed.
137  if (paper.publishedDate) rows.push(["DA", paper.publishedDate.replace(/-/g, "/")]);
138  if (paper.volume) rows.push(["VL", String(paper.volume)]);
139  if (paper.issue) rows.push(["IS", String(paper.issue)]);
140  if (paper.firstPage) rows.push(["SP", paper.firstPage]);
141  if (paper.lastPage) rows.push(["EP", paper.lastPage]);
142  if (paper.abstract) rows.push(["AB", decodeEntities(paper.abstract)]);
143  if (paper.doi) {
144    rows.push(["DO", paper.doi]);
145    rows.push(["UR", `https://doi.org/${paper.doi}`]);
146  }
147  rows.push(["ER", ""]);
148  return rows.map(([tag, value]) => `${tag}  - ${value}`).join("\n");
149}
150
151/**
152 * A pass-through of the raw Crossref records the publication files carry (`csl`) that only decodes character references (stored escaped
153 * for this site's JSON-LD, not for a reference manager). Nothing is added.
154 *
155 * @param {any[]} cslRecords the `csl` record of each published paper that has one, in the index's order
156 */
157export function toCslJson(cslRecords) {
158  /**
159   * The return annotation is required: decodeDeep is recursive, and TypeScript refuses it with TS7023.
160   *
161   * @param {any} value
162   * @returns {any}
163   */
164  const decodeDeep = (value) => {
165    if (typeof value === "string") return decodeEntities(value);
166    if (Array.isArray(value)) return value.map(decodeDeep);
167    if (value && typeof value === "object") {
168      /** @type {Record<string, unknown>} */
169      const out = {};
170      for (const [k, v] of Object.entries(value)) out[k] = decodeDeep(v);
171      return out;
172    }
173    return value;
174  };
175  return `${JSON.stringify(cslRecords.map(decodeDeep), null, 2)}\n`;
176}
177
178/**
179 * No generation timestamp: two downloads of an unchanged corpus must be byte-identical.
180 *
181 * @param {string} comment the format's line-comment prefix
182 * @param {number} count
183 */
184function header(comment, count) {
185  return [
186    `${comment} ${count} works by Dustin Edwards.`,
187    `${comment} Generated from ${SITE_ORIGIN}/research/publications`,
188    `${comment} Bibliographic data from Crossref and DataCite. No generation date:`,
189    `${comment} this file is a function of the corpus, so it changes only when the`,
190    `${comment} corpus does.`,
191    "",
192  ].join("\n");
193}
194
195/** @param {any[]} papers */
196export function toBibtexAll(papers) {
197  return `${header("%", papers.length)}\n${papers.map(toBibtex).join("\n\n")}\n`;
198}
199
200/** @param {any[]} papers */
201export function toRisAll(papers) {
202  return `${header("", papers.length)}\n${papers.map(toRis).join("\n\n")}\n`;
203}