Upstream spencermountain/compromise state of the touched files before 642fe4f40a

src/1-one/output/api/html.js

@@ -0,0 +1,77 @@
+const isClass = /^\../
+const isId = /^#./
+
+const escapeXml = str => {
+  str = str.replaceAll('&', '&')
+  str = str.replaceAll('<', '&lt;')
+  str = str.replaceAll('>', '&gt;')
+  str = str.replaceAll('"', '&quot;')
+  str = str.replaceAll("'", '&apos;')
+  return str
+}
+
+// interpret .class, #id, tagName
+const toTag = function (k) {
+  let start = ''
+  let end = '</span>'
+  k = escapeXml(k)
+  if (isClass.test(k)) {
+    start = `<span class="${k.replace(/^\./, '')}"`
+  } else if (isId.test(k)) {
+    start = `<span id="${k.replace(/^#/, '')}"`
+  } else {
+    start = `<${k}`
+    end = `</${k}>`
+  }
+  start += '>'
+  return { start, end }
+}
+
+const getIndex = function (doc, obj) {
+  const starts = {}
+  const ends = {}
+  Object.keys(obj).forEach(k => {
+    let res = obj[k]
+    const tag = toTag(k)
+    if (typeof res === 'string') {
+      res = doc.match(res)
+    }
+    res.docs.forEach(terms => {
+      // don't highlight implicit terms
+      if (terms.every(t => t.implicit)) {
+        return
+      }
+      const a = terms[0].id
+      starts[a] ||= []
+      starts[a].push(tag.start)
+      const b = terms.at(-1).id
+      ends[b] ||= []
+      ends[b].push(tag.end)
+    })
+  })
+  return { starts, ends }
+}
+
+const html = function (obj) {
+  // index ids to highlight
+  const { starts, ends } = getIndex(this, obj)
+  // create the text output
+  let out = ''
+  this.docs.forEach(terms => {
+    for (let i = 0; i < terms.length; i += 1) {
+      const t = terms[i]
+      // do a span tag
+      if (Object.hasOwn(starts, t.id)) {
+        out += starts[t.id].join('')
+      }
+      out += t.pre || ''
+      out += t.text || ''
+      if (Object.hasOwn(ends, t.id)) {
+        out += ends[t.id].join('')
+      }
+      out += t.post || ''
+    }
+  })
+  return out
+}
+export default { html }

tests/01-one/output/html.test.js

@@ -0,0 +1,61 @@
+import test from 'tape'
+import nlp from '../../lib/one.js'
+const here = '[one/output/html] '
+
+test('html-match', function (t) {
+  let doc = nlp(`match one two. one match two. one two match.`)
+  let html = doc.html({ i: 'match' })
+  t.equal(html, `<i>match</i> one two. one <i>match</i> two. one two <i>match</i>.`, here + 'html tag')
+
+  doc = nlp(`match one two. one match two.`)
+  html = doc.html({ '.foo': 'match' })
+  t.equal(html, `<span class="foo">match</span> one two. one <span class="foo">match</span> two.`, here + 'html class')
+
+  doc = nlp(`one match match two.`)
+  html = doc.html({ '#foo': 'match+' })
+  t.equal(html, `one <span id="foo">match match</span> two.`, here + 'html id')
+
+  doc = nlp(`one match two.`)
+  html = doc.html({ '.red': 'match+', '.blue': doc.match('two') })
+  t.equal(html, `one <span class="red">match</span> <span class="blue">two</span>.`, here + 'html two classes')
+
+  doc = nlp(`if i can recall, my grey dog loves pizza crusts (they are really good).`)
+  html = doc.html({ '.red': 'my grey dog', '.blue': doc.match('loves') })
+  t.equal(
+    html,
+    `if i can recall, <span class="red">my grey dog</span> <span class="blue">loves</span> pizza crusts (they are really good).`,
+    here + 'html pre test'
+  )
+
+  t.end()
+})
+
+test('html-nest', function (t) {
+  const doc = nlp(`one match two.`)
+  let html = doc.html({
+    i: 'match',
+    b: 'one match two',
+  })
+  t.equal(html, `<b>one <i>match</i> two</b>.`, here + 'easy nest')
+
+  html = doc.html({
+    b: 'match two',
+    i: 'match',
+  })
+  t.equal(html, `one <b><i>match</i> two</b>.`, here + 'hard nest')
+
+  // html = doc.html({
+  //   i: 'match',
+  //   b: 'match two',
+  // })
+  // t.equal(html, `one <b><i>match</i> two</b>.`, here + 'harder nest')
+
+  t.end()
+})
+
+test('html-implicit', function (t) {
+  const doc = nlp(`he's cool`)
+  const out = doc.html({ '.foo': '#Verb' })
+  t.equal(out, `he's cool`, here + 'implict')
+  t.end()
+})