fix(author): be sure byline rule is not a date (#410)

* fix(author): be sure byline rule is not a date closes #408 * build: update dependencies
microlinkhq · May 21, 2021 · c8c035d · c8c035d
1 parent 68800c3
commit c8c035d
Show file tree

Hide file tree

Showing 13 changed files with 31 additions and 18 deletions.
diff --git a/packages/metascraper-amazon/package.json b/packages/metascraper-amazon/package.json
@@ -18,7 +18,7 @@
   ],
   "dependencies": {
     "@metascraper/helpers": "^5.21.7",
-    "tldts": "~5.7.29"
+    "tldts": "~5.7.37"
   },
   "devDependencies": {
     "lodash": "latest",

diff --git a/packages/metascraper-audio/package.json b/packages/metascraper-audio/package.json
@@ -18,7 +18,7 @@
   ],
   "dependencies": {
     "@metascraper/helpers": "^5.21.7",
-    "cheerio": "~1.0.0-rc.6",
+    "cheerio": "~1.0.0-rc.9",
     "got": "~11.8.2",
     "p-reflect": "~2.1.0"
   },

diff --git a/packages/metascraper-author/index.js b/packages/metascraper-author/index.js
@@ -1,6 +1,12 @@
 'use strict'
 
-const { $jsonld, $filter, toRule, author } = require('@metascraper/helpers')
+const {
+  $jsonld,
+  $filter,
+  toRule,
+  date,
+  author
+} = require('@metascraper/helpers')
 
 const REGEX_STRICT = /^\S+\s+\S+/
 
@@ -32,6 +38,13 @@ module.exports = () => ({
     strict(toAuthor($ => $filter($, $('a[href*="/author/" i]')))),
     toAuthor($ => $filter($, $('a[class*="screenname" i]'))),
     strict(toAuthor($ => $filter($, $('[class*="author" i]')))),
-    strict(toAuthor($ => $filter($, $('[class*="byline" i]'))))
+    strict(
+      toAuthor($ =>
+        $filter($, $('[class*="byline" i]'), el => {
+          const value = $filter.fn(el)
+          return !date(value) && value
+        })
+      )
+    )
   ]
 })
diff --git a/packages/metascraper-clearbit/package.json b/packages/metascraper-clearbit/package.json
@@ -23,7 +23,7 @@
     "async-memoize-one": "~1.1.0",
     "got": "~11.8.2",
     "lodash": "~4.17.21",
-    "tldts": "~5.7.29"
+    "tldts": "~5.7.37"
   },
   "devDependencies": {
     "mocha": "latest",

diff --git a/packages/metascraper-helpers/index.js b/packages/metascraper-helpers/index.js
@@ -149,13 +149,13 @@ const titleize = (src, opts = {}) => {
   return title
 }
 
-const defaultFn = el => el.text().trim()
-
-const $filter = ($, domNodes, fn = defaultFn) => {
+const $filter = ($, domNodes, fn = $filter.fn) => {
   const el = domNodes.filter((i, el) => fn($(el))).first()
   return fn(el)
 }
 
+$filter.fn = el => el.text().trim()
+
 const isAuthor = (str, opts = { relative: false }) =>
   !isUrl(str, opts) &&
   !isEmpty(str) &&

diff --git a/packages/metascraper-media-provider/package.json b/packages/metascraper-media-provider/package.json
@@ -27,8 +27,8 @@
     "p-reflect": "~2.1.0",
     "p-retry": "~4.5.0",
     "p-timeout": "~4.1.0",
-    "tldts": "~5.7.29",
-    "youtube-dl-exec": "~1.2.1"
+    "tldts": "~5.7.37",
+    "youtube-dl-exec": "~1.2.3"
   },
   "devDependencies": {
     "is-ci": "latest",

diff --git a/packages/metascraper-soundcloud/package.json b/packages/metascraper-soundcloud/package.json
@@ -19,7 +19,7 @@
   ],
   "dependencies": {
     "@metascraper/helpers": "^5.21.7",
-    "tldts": "~5.7.29"
+    "tldts": "~5.7.37"
   },
   "devDependencies": {
     "mocha": "latest",

diff --git a/packages/metascraper-spotify/package.json b/packages/metascraper-spotify/package.json
@@ -20,8 +20,8 @@
   "dependencies": {
     "@metascraper/helpers": "^5.21.7",
     "async-memoize-one": "~1.1.0",
-    "spotify-url-info": "~2.2.0",
-    "tldts": "~5.7.29"
+    "spotify-url-info": "~2.2.2",
+    "tldts": "~5.7.37"
   },
   "devDependencies": {
     "kind-of": "latest",

diff --git a/packages/metascraper-telegram/package.json b/packages/metascraper-telegram/package.json
@@ -20,7 +20,7 @@
     "@metascraper/helpers": "^5.21.7",
     "async-memoize-one": "~1.1.0",
     "css-urls": "~2.0.13",
-    "tldts": "~5.7.29"
+    "tldts": "~5.7.37"
   },
   "devDependencies": {
     "mocha": "latest",

diff --git a/packages/metascraper-uol/package.json b/packages/metascraper-uol/package.json
@@ -20,7 +20,7 @@
   ],
   "dependencies": {
     "@metascraper/helpers": "^5.21.7",
-    "tldts": "~5.7.29"
+    "tldts": "~5.7.37"
   },
   "devDependencies": {
     "mocha": "latest",

diff --git a/packages/metascraper-video/package.json b/packages/metascraper-video/package.json
@@ -18,7 +18,7 @@
   ],
   "dependencies": {
     "@metascraper/helpers": "^5.21.7",
-    "cheerio": "~1.0.0-rc.6",
+    "cheerio": "~1.0.0-rc.9",
     "got": "~11.8.2",
     "lodash": "~4.17.21",
     "p-reflect": "~2.1.0"

diff --git a/packages/metascraper/__snapshots__/index.js.snap-shot b/packages/metascraper/__snapshots__/index.js.snap-shot
@@ -988,7 +988,7 @@ exports['wired 1'] = {
 }
 
 exports['wnycstudios 1'] = {
-  "author": "August 14, 2020",
+  "author": null,
   "date": "2020-08-14T12:00:00.000Z",
   "description": "In the 1980s, as China was modernizing, they ran into a problem. There was no way to fit their 70,000 plus character language on a QWERTY keyboard. Today, the story of how they did it.",
   "image": "https://media.wnyc.org/i/1355/1600/c/80/2020/08/QWERTY.JPG",

diff --git a/packages/metascraper/package.json b/packages/metascraper/package.json
@@ -58,7 +58,7 @@
   ],
   "dependencies": {
     "@metascraper/helpers": "^5.21.7",
-    "cheerio": "~1.0.0-rc.6",
+    "cheerio": "~1.0.0-rc.9",
     "lodash": "~4.17.21",
     "whoops": "~4.1.0"
   },