Add support for wildcard patterns

Related to retextjs/retext-equality#41.

index.js

@@ -18,11 +18,10 @@ function search(tree, phrases, handler, options) {
   var dashes = settings.allowDashes || false
   var literals = settings.allowLiterals
   var config = {allowApostrophes: apos, allowDashes: dashes}
-  var byWord = {}
+  var byWord = {'*': []}
   var length
   var index
   var key
-  var firstWord
 
   if (!tree || !tree.type) {
     throw new Error('Expected node')
@@ -54,14 +53,15 @@ function search(tree, phrases, handler, options) {
     var node = siblings[position]
     var count = siblings.length
     var queue = [node]
-    var expression = phrase.split(C_SPACE).slice(1)
-    var length = expression.length
+    var expressions = phrase.split(C_SPACE).slice(1)
+    var length = expressions.length
     var index = -1
+    var expression
 
     /* Move one position forward. */
     position++
 
-    /* Iterate over `expression`. */
+    /* Iterate over `expressions`. */
     while (++index < length) {
       /* Allow joining white-space. */
       while (position < count) {
@@ -76,6 +76,7 @@ function search(tree, phrases, handler, options) {
       }
 
       node = siblings[position]
+      expression = expressions[index]
 
       /* Exit if there are no nodes left, if the
        * current node is not a word, or if the
@@ -84,9 +85,10 @@ function search(tree, phrases, handler, options) {
       if (
         !node ||
         node.type !== T_WORD ||
-        normalize(expression[index], config) !== normalize(node, config)
+        (expression !== '*' &&
+          normalize(expression, config) !== normalize(node, config))
       ) {
-        return null
+        return
       }
 
       queue.push(node)
@@ -109,7 +111,7 @@ function search(tree, phrases, handler, options) {
     }
 
     word = normalize(node, config)
-    phrases = own.call(byWord, word) ? byWord[word] : []
+    phrases = byWord['*'].concat(own.call(byWord, word) ? byWord[word] : [])
     length = phrases.length
     index = -1
 
@@ -124,7 +126,7 @@ function search(tree, phrases, handler, options) {
 
   /* Handle a phrase. */
   function handlePhrase(phrase) {
-    firstWord = normalize(phrase.split(C_SPACE, 1)[0], config)
+    var firstWord = normalize(phrase.split(C_SPACE, 1)[0], config)
 
     if (own.call(byWord, firstWord)) {
       byWord[firstWord].push(phrase)

readme.md

@@ -1,6 +1,6 @@
 # nlcst-search [![Build Status][travis-badge]][travis] [![Coverage Status][codecov-badge]][codecov]
 
-Search for patterns in an [NLCST][nlcst] tree.
+Search for patterns in an [NLCST][] tree.
 
 ## Installation
 
@@ -74,10 +74,12 @@ Tree to search in ([`Node`][node]).
 ###### `patterns`
 
 Patterns to search for (`Array.<string>` or `Object`).
-If an `Object`, uses its keys.  Each pattern is a space-delimited list of
-words, where each word is [normalize][]d to remove casing, apostrophes,
-and dashes.  Spaces in a pattern mean zero or more white space nodes in
-the tree.
+If an `Object`, uses its keys as patterns.
+Each pattern is a space-delimited list of words, where each word is
+[normalize][]d to remove casing, apostrophes, and dashes.
+Spaces in a pattern mean zero or more white space nodes in the tree.
+Instead of a word, it’s also possible to use a wildcard symbol (`*`, an
+asterisk), that matches any word in a pattern (`alpha * charlie`).
 
 ###### `handler`
 

test.js

@@ -80,8 +80,6 @@ var tree = {
 }
 
 test('search(tree, patterns, handle)', function(t) {
-  t.plan(68)
-
   t.throws(
     function() {
       search()
@@ -176,15 +174,15 @@ test('search(tree, patterns, handle)', function(t) {
    * is provided  the tree contains “hell” but not “he’ll”
    * or “he'll”. */
   t.throws(function() {
-    search(tree, ['hell'], null)
+    search(tree, ['hell'])
   }, 'should find non-apostrophe words when `allowApostrophes` is absent')
 
   t.throws(function() {
-    search(tree, ['he’ll'], null)
+    search(tree, ['he’ll'])
   }, 'should find smart apostrophe words when `allowApostrophes` is absent')
 
   t.throws(function() {
-    search(tree, ["he'll"], null)
+    search(tree, ["he'll"])
   }, 'should find dumb apostrophe words when `allowApostrophes` is absent')
 
   t.throws(function() {
@@ -214,11 +212,11 @@ test('search(tree, patterns, handle)', function(t) {
   /* The tree contains “selfservice” but not “self-service” */
 
   t.throws(function() {
-    search(tree, ['selfservice'], null)
+    search(tree, ['selfservice'])
   }, 'should find non-dash words when `allowDashes` is absent and `allowApostrophes` is absent')
 
   t.throws(function() {
-    search(tree, ['self-service'], null)
+    search(tree, ['self-service'])
   }, 'should find dash words when `allowDashes` is absent and `allowApostrophes` is absent')
 
   t.throws(function() {
@@ -309,6 +307,22 @@ test('search(tree, patterns, handle)', function(t) {
     })
   }, 'should find dash words when `allowDashes` is false and `allowApostrophes` is true')
 
+  t.throws(function() {
+    search(tree, ['this * selfservice'])
+  }, 'should support wild cards (#1)')
+
+  t.doesNotThrow(function() {
+    search(tree, ['that * selfservice'])
+  }, 'should support wild cards (#2)')
+
+  t.throws(function() {
+    search(tree, ['* selfservice'])
+  }, 'should support wild cards (#3)')
+
+  t.doesNotThrow(function() {
+    search(tree, ['* zelfzervice'])
+  }, 'should support wild cards (#4)')
+
   t.doesNotThrow(function() {
     search(tree, ['mellow'])
   }, 'shouldn’t find literals by default')
@@ -321,4 +335,6 @@ test('search(tree, patterns, handle)', function(t) {
     },
     {allowLiterals: true}
   )
+
+  t.end()
 })