Add support for :lang() selector

Author: wooorm

lib/any.js

@@ -8,6 +8,7 @@ var svg = require('property-information/svg')
 var needsIndex = require('./pseudo').needsIndex
 var test = require('./test')
 var nest = require('./nest')
+var enter = require('./enter-state')
 
 var type = zwitch('type')
 var handlers = type.handlers
@@ -43,6 +44,7 @@ function rule(query, tree, state) {
   var collect = collector(state.one)
   var opts = {
     schema: state.space === 'svg' ? svg : html,
+    language: undefined,
     iterator: match,
     one: state.one,
     shallow: state.shallow
@@ -57,6 +59,8 @@ function rule(query, tree, state) {
   return collect.result
 
   function match(query, node, index, parent, state) {
+    var exit = enter(state, node)
+
     if (test(query, node, index, parent, state)) {
       if (query.rule) {
         nest(query.rule, node, index, parent, configure(query.rule, state))
@@ -65,6 +69,8 @@ function rule(query, tree, state) {
         state.found = true
       }
     }
+
+    exit()
   }
 
   function configure(query, state) {

lib/enter-state.js

@@ -0,0 +1,35 @@
+'use strict'
+
+var svg = require('property-information/svg')
+
+module.exports = config
+
+function config(state, node) {
+  var schema = state.schema
+  var language = state.language
+  var found
+  var lang
+
+  if (node.type === 'element') {
+    if (schema.space === 'html' && node.tagName === 'svg') {
+      state.schema = svg
+      found = true
+    }
+
+    lang = node.properties.xmlLang || node.properties.lang
+
+    if (lang !== undefined && lang !== null) {
+      state.language = lang
+      found = true
+    }
+  }
+
+  return found ? reset : noop
+
+  function reset() {
+    state.schema = schema
+    state.language = language
+  }
+}
+
+function noop() {}

lib/nest.js

@@ -1,7 +1,7 @@
 'use strict'
 
 var zwitch = require('zwitch')
-var svg = require('property-information/svg')
+var enter = require('./enter-state')
 
 module.exports = zwitch('nestingOperator')
 
@@ -98,15 +98,6 @@ function walkIterator(query, parent, state) {
   var nodes = parent.children
   var typeIndex = state.index ? createTypeIndex() : null
   var delayed = []
-  var parentSchema = state.schema
-
-  if (
-    parentSchema.space === 'html' &&
-    parent.type === 'element' &&
-    parent.tagName === 'svg'
-  ) {
-    state.schema = svg
-  }
 
   return {
     prefillTypeIndex: rangeDefaults(prefillTypeIndex),
@@ -118,8 +109,6 @@ function walkIterator(query, parent, state) {
     var length = delayed.length
     var index = -1
 
-    state.schema = parentSchema
-
     while (++index < length) {
       delayed[index]()
 
@@ -175,7 +164,9 @@ function walkIterator(query, parent, state) {
     }
 
     function pushNode() {
+      var exit = enter(state, child)
       state.iterator(query, child, start, parent, state)
+      exit()
     }
   }
 

lib/pseudo.js

@@ -1,5 +1,8 @@
 'use strict'
 
+var commaSeparated = require('comma-separated-tokens').parse
+var filter = require('bcp-47-match').extendedFilter
+
 module.exports = match
 
 match.selectorPseudoSupport = ['any', 'matches', 'not']
@@ -52,6 +55,7 @@ handlers.empty = empty
 handlers.enabled = not(disabled)
 handlers['first-child'] = firstChild
 handlers['first-of-type'] = firstOfType
+handlers.lang = lang
 handlers['last-child'] = lastChild
 handlers['last-of-type'] = lastOfType
 handlers.matches = matches
@@ -145,6 +149,14 @@ function firstChild(query, node, index, parent, state) {
   return state.elementIndex === 0
 }
 
+function lang(query, node, index, parent, state) {
+  return (
+    state.language !== '' &&
+    state.language !== undefined &&
+    filter(state.language, commaSeparated(query.value)).length !== 0
+  )
+}
+
 function lastChild(query, node, index, parent, state) {
   assertDeep(state, query)
   return state.elementIndex === state.elementCount - 1

package.json

@@ -20,6 +20,7 @@
     "Titus Wormer <tituswormer@gmail.com> (http://wooorm.com)"
   ],
   "dependencies": {
+    "bcp-47-match": "^1.0.0",
     "comma-separated-tokens": "^1.0.2",
     "css-selector-parser": "^1.3.0",
     "hast-util-has-property": "^1.0.0",

readme.md

@@ -177,6 +177,7 @@ Yields:
 *   [x] `:optional` (pseudo-class)
 *   [x] `:required` (pseudo-class)
 *   [x] `:root` (pseudo-class)
+*   [x] `:lang()` (pseudo-class)
 *   [x] `article p` (combinator: descendant selector)
 *   [x] `article > p` (combinator: child selector)
 *   [x] `h1 + p` (combinator: adjacent sibling selector)
@@ -210,7 +211,6 @@ Yields:
 *   [ ] † `:indeterminate` (pseudo-class)
 *   [ ] § `:in-range` (pseudo-class)
 *   [ ] § `:invalid` (pseudo-class)
-*   [ ] § `:lang()` (pseudo-class)
 *   [ ] † `:link` (pseudo-class)
 *   [ ] † `:local-link` (pseudo-class)
 *   [ ] † `nth-column()` (pseudo-class)

test/matches.js

@@ -900,6 +900,66 @@ test('select.matches()', function(t) {
       sst.end()
     })
 
+    st.test(':lang()', function(sst) {
+      sst.ok(
+        matches(':lang(de, en)', h('html', {xmlLang: 'en'})),
+        'true if the element has an `xml:lang` attribute'
+      )
+
+      sst.ok(
+        matches(':lang(de, en)', h('html', {lang: 'de'})),
+        'true if the element has a `lang` attribute'
+      )
+
+      sst.notOk(
+        matches(':lang(de, en)', h('html', {xmlLang: 'jp'})),
+        'false if the element has an different language set'
+      )
+
+      sst.notOk(
+        matches(':lang(de, en)', h('html', {xmlLang: 'jp', lang: 'de'})),
+        'should prefer `xmlLang` over `lang` (#1)'
+      )
+
+      sst.ok(
+        matches(':lang(de, en)', h('html', {xmlLang: 'de', lang: 'jp'})),
+        'should prefer `xmlLang` over `lang` (#2)'
+      )
+
+      sst.notOk(
+        matches(':lang(de, en)', h('html', {xmlLang: 'jp'})),
+        'false if the element has an different language set'
+      )
+
+      sst.ok(
+        matches(':lang("*")', h('html', {lang: 'en'})),
+        'should support wildcards'
+      )
+
+      sst.notOk(
+        matches(':lang(en)', h('html', {lang: ''})),
+        'false if [lang] is an empty string (means unknown language)'
+      )
+
+      sst.notOk(
+        matches(':lang(*)', h('html', {lang: ''})),
+        'false with wildcard if [lang] is an empty string (means unknown language)'
+      )
+
+      sst.ok(
+        matches(':lang("de-*-DE")', h('html', {lang: 'de-Latn-DE'})),
+        'should support non-primary wildcard subtags (#1)'
+      )
+
+      // Not supported by `css-selector-parser` yet :(
+      // sst.ok(
+      //   matches(':lang("fr-BE", "de-*-DE")', h('html', {lang: 'de-Latn-DE'})),
+      //   'should support non-primary wildcard subtags (#2)'
+      // )
+
+      sst.end()
+    })
+
     st.test(':root', function(sst) {
       sst.ok(matches(':root', h('html')), 'true if `<html>` in HTML space')
 

test/select-all.js

@@ -709,6 +709,45 @@ test('select.selectAll()', function(t) {
       sst.end()
     })
 
+    st.test(':lang()', function(sst) {
+      sst.deepEqual(
+        selectAll(
+          'q:lang(en)',
+          u('root', [
+            h('div', {lang: 'en'}, h('p', {lang: ''}, h('q', '0'))),
+            h('p', {lang: 'fr'}, h('q', {lang: 'fr'}, 'A')),
+            h('p', {lang: 'fr'}, h('q', {lang: 'en'}, 'B')),
+            h('p', {lang: 'fr'}, h('q', {lang: 'en-GB'}, 'C')),
+            h('p', {lang: 'fr'}, h('q', {lang: ''}, 'D')),
+            h('p', {lang: 'fr'}, h('q', 'E')),
+            h('p', {lang: 'en'}, h('q', {lang: 'fr'}, 'F')),
+            h('p', {lang: 'en'}, h('q', {lang: 'en'}, 'G')),
+            h('p', {lang: 'en'}, h('q', {lang: 'en-GB'}, 'H')),
+            h('p', {lang: 'en'}, h('q', {lang: ''}, 'I')),
+            h('p', {lang: 'en'}, h('q', 'J')),
+            h('p', {lang: 'en-GB'}, h('q', {lang: 'fr'}, 'K')),
+            h('p', {lang: 'en-GB'}, h('q', {lang: 'en'}, 'L')),
+            h('p', {lang: 'en-GB'}, h('q', {lang: 'en-GB'}, 'M')),
+            h('p', {lang: 'en-GB'}, h('q', {lang: ''}, 'N')),
+            h('p', {lang: 'en-GB'}, h('q', 'O'))
+          ])
+        ),
+        [
+          h('q', {lang: 'en'}, 'B'),
+          h('q', {lang: 'en-GB'}, 'C'),
+          h('q', {lang: 'en'}, 'G'),
+          h('q', {lang: 'en-GB'}, 'H'),
+          h('q', 'J'),
+          h('q', {lang: 'en'}, 'L'),
+          h('q', {lang: 'en-GB'}, 'M'),
+          h('q', 'O')
+        ],
+        'should return the correct matching elements'
+      )
+
+      sst.end()
+    })
+
     st.test(':root', function(sst) {
       sst.deepEqual(
         selectAll(

test/select.js

@@ -672,6 +672,25 @@ test('select.select()', function(t) {
       sst.end()
     })
 
+    st.test(':lang()', function(sst) {
+      sst.deepEqual(
+        select(
+          'q:lang(en)',
+          u('root', [
+            h('div', {lang: 'en'}, h('p', {lang: ''}, h('q', '0'))),
+            h('p', {lang: 'fr'}, h('q', {lang: 'fr'}, 'A')),
+            h('p', {lang: 'fr'}, h('q', {lang: ''}, 'B')),
+            h('p', {lang: 'fr'}, h('q', {lang: 'en-GB'}, 'C')),
+            h('p', {lang: 'fr'}, h('q', {lang: 'en'}, 'D'))
+          ])
+        ),
+        h('q', {lang: 'en-GB'}, 'C'),
+        'should return the correct matching element'
+      )
+
+      sst.end()
+    })
+
     st.test(':root', function(sst) {
       sst.deepEqual(
         select(

test/svg.js

@@ -3,17 +3,45 @@
 var test = require('tape')
 var u = require('unist-builder')
 var s = require('hastscript/svg')
+var h = require('hastscript')
 var select = require('..').select
+var selectAll = require('..').selectAll
 
 test('svg', function(t) {
   t.deepEqual(
     select(
       '[writing-mode]',
-      u('root', [s('svg', [s('text', {writingMode: 'lr-tb'}, '!')])])
+      u('root', [
+        s('svg', [s('text', {writingMode: 'lr-tb'}, '!')]),
+        s('p', [
+          h(
+            'text',
+            {writingMode: 'lr-tb'},
+            'this is a camelcased HTML attribute'
+          )
+        ])
+      ])
     ),
     s('text', {writingMode: 'lr-tb'}, '!')
   )
 
+  t.deepEqual(
+    selectAll(
+      '[writing-mode]',
+      u('root', [
+        s('svg', [s('text', {writingMode: 'lr-tb'}, '!')]),
+        s('p', [
+          h(
+            'text',
+            {writingMode: 'lr-tb'},
+            'this is a camelcased HTML attribute'
+          )
+        ])
+      ])
+    ),
+    [s('text', {writingMode: 'lr-tb'}, '!')]
+  )
+
   t.deepEqual(
     select('[writing-mode]', s('text', {writingMode: 'lr-tb'}, '!')),
     null