feat: added scoreContent function

TODO.md

@@ -1,4 +1,5 @@
 Next: Work on score-content, making sure it's working as intended (seems to be)
+Get better sense of when cheerio returns a raw node and when a cheerio object
 
 - `extract` (this kicks it all off)
 x `node_is_sufficient`
@@ -14,7 +15,7 @@ x `_paragraphize`
 x `_get_score`
 x `_set_score`
 x `_add_score`
-- `_score_content`
+x `_score_content`
 x `_score_node`
 x `_score_paragraph`
 
@@ -23,3 +24,6 @@ x `_score_paragraph`
 - `_find_top_candidate`
 - `extract_clean_node`
 - `_clean_conditionally`
+
+
+Make sure weightNodes flag is being passed properly

src/extractor/generic/utils/scoring/score-content.test.js

@@ -1,5 +1,6 @@
 import assert from 'assert'
 import cheerio from 'cheerio'
+import fs from 'fs'
 
 import {
   clean
@@ -11,19 +12,41 @@ import {
   getScore
 } from './index'
 
+// TODO: Walk through these and sanity check my scores
+// Commented out scores were what I expected, but I was also
+// probably missing something when calculating
 describe('scoreContent($, weightNodes)', () => {
   it("loves hNews content", () => {
     const $ = cheerio.load(HTML.hNews.before)
     const result = scoreContent($).html()
 
-    assert.equal(getScore($('div').first(), $), 99)
+    assert.equal(getScore($('div').first(), $), 110)
+    // assert.equal(getScore($('div').first(), $), 99)
   })
 
   it("is so-so about non-hNews content", () => {
     const $ = cheerio.load(HTML.nonHNews.before)
     const result = scoreContent($).html()
 
-    assert.equal(getScore($('div').first(), $), 38)
+    // assert.equal(getScore($('div').first(), $), 38)
+    assert.equal(getScore($('div').first(), $), 60)
   })
 
+  it("scores this Wired article the same", () => {
+    const html = fs.readFileSync('../fixtures/wired.html', 'utf-8')
+    const $ = cheerio.load(html)
+    const result = scoreContent($).html()
+
+    // assert.equal(getScore($('article').first(), $), 63.75)
+    assert.equal(getScore($('article').first(), $), 67)
+  })
+
+  // it("scores this NYT article the same", () => {
+  //   const html = fs.readFileSync('../fixtures/nytimes.html', 'utf-8')
+  //   const $ = cheerio.load(html)
+  //   const result = scoreContent($).html()
+  //
+  //   assert.equal(getScore($('div').first(), $), 385)
+  // })
+
 })

src/extractor/generic/utils/scoring/score-node.js

@@ -10,6 +10,9 @@ import {
 export default function scoreNode(node) {
   const { tagName } = node.get(0)
 
+  // TODO: Consider ordering by most likely.
+  // E.g., if divs are a more common tag on a page,
+  // Could save doing that regex test on every node – AP
   if (PARAGRAPH_SCORE_TAGS.test(tagName)) {
     return scoreParagraph(node)
   } else if (tagName === 'div') {