feat: add metascraper-citation

README.md

@@ -238,6 +238,7 @@ const metascraper = require('metascraper')([
 
 - [metascraper-audio](https://github.com/microlinkhq/metascraper/tree/master/packages/metascraper-audio) – Get audio property from HTML markup.
 - [metascraper-author](https://github.com/microlinkhq/metascraper/tree/master/packages/metascraper-author) – Get author property from HTML markup.
+- [metascraper-citation](https://github.com/microlinkhq/metascraper/tree/master/packages/metascraper-citation) – Get title, author, date, and publisher from Highwire Press `citation_*` and Dublin Core `dc.*` / `dcterms.*` tags.
 - [metascraper-date](https://github.com/microlinkhq/metascraper/tree/master/packages/metascraper-date) – Get date property from HTML markup.
 - [metascraper-defuddle](https://github.com/microlinkhq/metascraper/tree/master/packages/metascraper-defuddle) – A Defuddle connector for metascraper.
 - [metascraper-description](https://github.com/microlinkhq/metascraper/tree/master/packages/metascraper-description) – Get description property from HTML markup.

packages/metascraper-citation/CHANGELOG.md

@@ -0,0 +1,4 @@
+# Change Log
+
+All notable changes to this project will be documented in this file.
+See [Conventional Commits](https://conventionalcommits.org) for commit guidelines.

packages/metascraper-citation/README.md

@@ -0,0 +1,59 @@
+<div align="center">
+  <br>
+  <img style="width: 500px; margin:3rem 0 1.5rem;" src="https://metascraper.js.org/static/logo-banner.png" alt="metascraper">
+  <br>
+  <br>
+  <p align="center"><strong>metascraper-citation</strong>: Get title, author, date, and publisher from Highwire Press <code>citation_*</code> and Dublin Core <code>dc.*</code> / <code>dcterms.*</code> meta tags.</p>
+  <p align="center">See our <a href="https://metascraper.js.org" target='_blank' rel='noopener noreferrer'>website</a> for more information.</p>
+  <br>
+</div>
+
+## Install
+
+```bash
+$ npm install metascraper-citation --save
+```
+
+Load it **before** the generic property packages so these values win over Open Graph (e.g. a clean `citation_title` instead of `og:title` with a site suffix).
+
+```js
+const metascraper = require('metascraper')([
+  require('metascraper-citation')(),
+  require('metascraper-author')(),
+  require('metascraper-date')(),
+  require('metascraper-publisher')(),
+  require('metascraper-title')()
+])
+```
+
+## Supported tags
+
+The bundle is a no-op unless [`test`](#test) finds at least one Highwire or Dublin Core tag. Highwire selectors run first; Dublin Core fills gaps (and covers repositories that only emit DC). Missing fields fall through to the generic rules.
+
+Only the four properties below are set. Bibliographic extras (`citation_doi`, `dc.identifier`, ISSN, volume, issue, pages, PDF URL, language, abstract) are ignored.
+
+| Property | Tags (first match wins) |
+| --- | --- |
+| **title** | `citation_title`, `dc.title`, `dcterms.title` |
+| **author** | `citation_author`, `dc.creator`, `dcterms.creator`, `dc.contributor`, `dcterms.contributor` (first author only) |
+| **date** | `citation_publication_date`, `citation_date`, `citation_online_date`, `dcterms.issued`, `dcterms.created`, `dc.date`, `dcterms.date` |
+| **publisher** | `citation_publisher`, `citation_journal_title`, `dc.publisher`, `dcterms.publisher` |
+
+Names are matched case-insensitively (`DC.Title` and `dc.title` are the same tag).
+
+## API
+
+### test
+
+```js
+const { test } = require('metascraper-citation')
+```
+
+Returns `true` when the Cheerio document has at least one `meta[name]` starting with `citation_`, `dc.`, or `dcterms.` (case-insensitive). Used as `rules.test` so the bundle skips pages with no scholarly markup.
+
+## License
+
+**metascraper-citation** © [Microlink](https://microlink.io), released under the [MIT](https://github.com/microlinkhq/metascraper/blob/master/LICENSE.md) License.<br>
+Authored and maintained by [Microlink](https://microlink.io) with help from [contributors](https://github.com/microlinkhq/metascraper/contributors).
+
+> [microlink.io](https://microlink.io) · GitHub [microlinkhq](https://github.com/microlinkhq) · X [@microlinkhq](https://x.com/microlinkhq)

packages/metascraper-citation/package.json

@@ -0,0 +1,50 @@
+{
+  "name": "metascraper-citation",
+  "description": "Metascraper rules for Highwire Press citation_* and Dublin Core dc.* / dcterms.* meta tags.",
+  "homepage": "https://github.com/microlinkhq/metascraper/packages/metascraper-citation",
+  "version": "5.55.2",
+  "types": "src/index.d.ts",
+  "main": "src/index.js",
+  "author": {
+    "email": "hello@microlink.io",
+    "name": "microlink.io",
+    "url": "https://microlink.io"
+  },
+  "repository": {
+    "directory": "packages/metascraper-citation",
+    "type": "git",
+    "url": "git+https://github.com/microlinkhq/metascraper.git"
+  },
+  "bugs": {
+    "url": "https://github.com/microlinkhq/metascraper/issues"
+  },
+  "keywords": [
+    "citation",
+    "dublin-core",
+    "extract",
+    "google-scholar",
+    "highwire",
+    "html",
+    "metadata",
+    "metascraper",
+    "scraper",
+    "scholar"
+  ],
+  "dependencies": {
+    "@metascraper/helpers": "workspace:*"
+  },
+  "devDependencies": {
+    "ava": "8",
+    "cheerio": "latest"
+  },
+  "engines": {
+    "node": ">= 22"
+  },
+  "files": [
+    "src"
+  ],
+  "scripts": {
+    "test": "NODE_PATH=.. TZ=UTC ava --timeout 15s"
+  },
+  "license": "MIT"
+}

packages/metascraper-citation/src/index.d.ts

@@ -0,0 +1,11 @@
+declare function rules(): import('metascraper').Rules;
+
+declare namespace rules {
+  /**
+   * `true` when the document has at least one Highwire (`citation_*`)
+   * or Dublin Core (`dc.*` / `dcterms.*`) meta tag.
+   */
+  function test(htmlDom: import('cheerio').CheerioAPI): boolean;
+}
+
+export = rules;

packages/metascraper-citation/src/index.js

@@ -0,0 +1,64 @@
+'use strict'
+
+const {
+  author,
+  date,
+  memoizeOne,
+  publisher,
+  title,
+  toRule
+} = require('@metascraper/helpers')
+
+const toAuthor = toRule(author)
+const toDate = toRule(date)
+const toPublisher = toRule(publisher)
+const toTitle = toRule(title)
+
+const meta = name => $ => $(`meta[name="${name}" i]`).attr('content')
+
+const MARKUP =
+  'meta[name^="citation_" i], meta[name^="dc." i], meta[name^="dcterms." i]'
+
+const test = memoizeOne(
+  $ => $(MARKUP).length > 0,
+  memoizeOne.EqualityFirstArgument
+)
+
+module.exports = () => {
+  const rules = {
+    title: [
+      toTitle(meta('citation_title')),
+      toTitle(meta('dc.title')),
+      toTitle(meta('dcterms.title'))
+    ],
+    author: [
+      toAuthor(meta('citation_author')),
+      toAuthor(meta('dc.creator')),
+      toAuthor(meta('dcterms.creator')),
+      toAuthor(meta('dc.contributor')),
+      toAuthor(meta('dcterms.contributor'))
+    ],
+    date: [
+      toDate(meta('citation_publication_date')),
+      toDate(meta('citation_date')),
+      toDate(meta('citation_online_date')),
+      toDate(meta('dcterms.issued')),
+      toDate(meta('dcterms.created')),
+      toDate(meta('dc.date')),
+      toDate(meta('dcterms.date'))
+    ],
+    publisher: [
+      toPublisher(meta('citation_publisher')),
+      toPublisher(meta('citation_journal_title')),
+      toPublisher(meta('dc.publisher')),
+      toPublisher(meta('dcterms.publisher'))
+    ]
+  }
+
+  rules.test = ({ htmlDom }) => test(htmlDom)
+  rules.pkgName = 'metascraper-citation'
+
+  return rules
+}
+
+module.exports.test = test