improvement normalize-space function #32

antchfx · Jan 29, 2019 · c8489ed · c8489ed
1 parent 4bbdf6d
commit c8489ed
Show file tree

Hide file tree

Showing 3 changed files with 18 additions and 4 deletions.
diff --git a/README.md b/README.md
@@ -160,5 +160,8 @@ Supported Features
 Changelogs
 ===
 
+2019-01-29
+-  improvement `normalize-space` function. [#32](https://github.com/antchfx/xpath/issues/32)
+
 2018-12-07
 -  supports XPath 2.0 Sequence expressions. [#30](https://github.com/antchfx/xpath/pull/30) by [@minherz](https://github.com/minherz).
diff --git a/func.go b/func.go
@@ -4,6 +4,7 @@ import (
 	"errors"
 	"fmt"
 	"math"
+	"regexp"
 	"strconv"
 	"strings"
 )
@@ -308,6 +309,11 @@ func containsFunc(arg1, arg2 query) func(query, iterator) interface{} {
 	}
 }
 
+var (
+	regnewline  = regexp.MustCompile(`[\r\n\t]`)
+	regseqspace = regexp.MustCompile(`\s{2,}`)
+)
+
 // normalizespaceFunc is XPath functions normalize-space(string?)
 func normalizespaceFunc(q query, t iterator) interface{} {
 	var m string
@@ -317,11 +323,14 @@ func normalizespaceFunc(q query, t iterator) interface{} {
 	case query:
 		node := typ.Select(t)
 		if node == nil {
-			return false
+			return ""
 		}
 		m = node.Value()
 	}
-	return strings.TrimSpace(m)
+	m = strings.TrimSpace(m)
+	m = regnewline.ReplaceAllString(m, " ")
+	m = regseqspace.ReplaceAllString(m, " ")
+	return m
 }
 
 // substringFunc is XPath functions substring function returns a part of a given string.

diff --git a/xpath_test.go b/xpath_test.go
@@ -650,7 +650,9 @@ func example() *TNode {
 			   <meta name="language" content="en"/>
 		   </head>
 		   <body>
-				<h1> This is a H1 </h1>
+				<h1>
+				This is a H1
+				</h1>
 				<ul>
 					<li><a id="1" href="/">Home</a></li>
 					<li><a id="2" href="/about">about</a></li>
@@ -678,7 +680,7 @@ func example() *TNode {
 	// The HTML body section.
 	body := xhtml.createChildNode("body", ElementNode)
 	n = body.createChildNode("h1", ElementNode)
-	n = n.createChildNode(" This is a H1 ", TextNode)
+	n = n.createChildNode("\nThis is a H1\n", TextNode)
 	ul := body.createChildNode("ul", ElementNode)
 	n = ul.createChildNode("li", ElementNode)
 	n = n.createChildNode("a", ElementNode)