htmltools

Various command line tools to transform HTML documents
git clone git://git.entf.net/htmltools
Log | Files | Refs | README | LICENSE

htmltools_test.go (4038B)

      1 package htmltools
      2 
      3 import (
      4 	"strings"
      5 	"testing"
      6 
      7 	"golang.org/x/net/html"
      8 )
      9 
     10 func getdoc(source string) *html.Node {
     11 	doc, err := html.Parse(strings.NewReader(source))
     12 	if err != nil || doc == nil || doc.FirstChild == nil {
     13 		panic(err)
     14 	}
     15 	return doc
     16 }
     17 
     18 func TestBody(t *testing.T) {
     19 	body, err := Body(getdoc(`<!DOCTYPE html><html>
     20 <head><title>test</title></head>
     21 <body>SUCCESS</body></html>`))
     22 	if err != nil {
     23 		t.Fatalf("%v", err)
     24 	}
     25 	if body == nil {
     26 		t.Fatal("body is nil")
     27 	}
     28 	if body.Type != html.ElementNode {
     29 		t.Fatal("body is not an element")
     30 	}
     31 	if body.FirstChild == nil {
     32 		t.Fatal("body has no children")
     33 	}
     34 	if body.FirstChild.Data != "SUCCESS" {
     35 		t.Fatal("could not find body text")
     36 	}
     37 
     38 	body, err = Body(body)
     39 	if err != ErrNodeIsNotADocumentNode {
     40 		t.Fatal("no error on invalid node")
     41 	}
     42 	if body != nil {
     43 		t.Fatal("body not nil on invalid node")
     44 	}
     45 }
     46 
     47 func TestFindRecursive(t *testing.T) {
     48 	doc := getdoc(`<!DOCTYPE html>
     49 <div><div><div><p></p></div><p></p></div><p></p></div><p></p>`)
     50 	ch := FindRecursive(doc, func(n *html.Node) bool {
     51 		return n.Data == "p"
     52 	})
     53 	c := 0
     54 	for _ = range ch {
     55 		c++
     56 	}
     57 	if c != 4 {
     58 		t.Fatalf("4 nodes expected, found %d", c)
     59 	}
     60 
     61 	doc = getdoc(`<!DOCTYPE html><html>
     62 <div><div><div><p>Hello</p></div><p></p></div><p></p></div><p>World</p>`)
     63 	ch = FindRecursive(doc, MatchNodeTypeFunc(html.TextNode))
     64 	c = 0
     65 	for _ = range ch {
     66 		c++
     67 	}
     68 	if c != 2 {
     69 		t.Fatalf("2 nodes expected, found %d", c)
     70 	}
     71 }
     72 
     73 func TestAttr(t *testing.T) {
     74 	body, err := Body(getdoc(`<!DOCTYPE html>
     75 <div id="1"></div>
     76 <div id="1"></div>
     77 TEST
     78 <div ID="1"></div>
     79 <div><div id="0"></div></div>`))
     80 	if err != nil {
     81 		t.Fatalf("%v", err)
     82 	}
     83 	if body == nil {
     84 		t.Fatal("body is nil")
     85 	}
     86 	topNodes := Children(body)
     87 	ids, err := Attr([]string{"id"}, topNodes...)
     88 	if err != nil {
     89 		t.Fatalf("%v", err)
     90 	}
     91 	if len(ids) != 3 {
     92 		t.Fatalf("3 attributes expected, found %d", len(ids))
     93 	}
     94 	for _, set := range ids {
     95 		if len(set) != 1 {
     96 			t.Fatalf("1 attribute in set expected, found %d", len(set))
     97 		}
     98 		if set[0] != "1" {
     99 			t.Fatal("invalid attribute value")
    100 		}
    101 	}
    102 
    103 	body, err = Body(getdoc(`<!DOCTYPE html>
    104 <div id="1" class="test"></div>
    105 <div id="1" data-test="data"></div>
    106 <div ID="1"></div>`))
    107 	if err != nil {
    108 		t.Fatalf("%v", err)
    109 	}
    110 	if body == nil {
    111 		t.Fatal("body is nil")
    112 	}
    113 	topNodes = Children(body)
    114 	ids, err = Attr([]string{"id", "data-test", "class"}, topNodes...)
    115 	if err != nil {
    116 		t.Fatalf("%v", err)
    117 	}
    118 	if len(ids) != 3 {
    119 		t.Fatalf("3 attributes expected, found %d", len(ids))
    120 	}
    121 	for _, set := range ids {
    122 		if len(set) != 3 {
    123 			t.Fatalf("3 attribute in set expected, found %d", len(set))
    124 		}
    125 		if set[0] != "1" {
    126 			t.Fatal("invalid attribute value")
    127 		}
    128 		if set[1] != "" && set[1] != "data" {
    129 			t.Fatal("invalid attribute value")
    130 		}
    131 		if set[2] != "" && set[2] != "test" {
    132 			t.Fatal("invalid attribute value")
    133 		}
    134 	}
    135 }
    136 
    137 func TestIndentHeadings(t *testing.T) {
    138 	body, err := Body(getdoc(`<!DOCTYPE html>
    139 <h1></h1>TEST<div></div><h2></h2><h6></h6>`))
    140 	if err != nil {
    141 		t.Fatalf("%v", err)
    142 	}
    143 	if body == nil {
    144 		t.Fatal("body is nil")
    145 	}
    146 	topNodes := Children(body)
    147 	err = IndentHeadings(3, topNodes...)
    148 	if err != nil {
    149 		t.Fatalf("%v", err)
    150 	}
    151 	for _, n := range topNodes {
    152 		if n.Type != html.ElementNode {
    153 			continue
    154 		}
    155 		switch n.Data {
    156 		case "div", "h4", "h5", "h6":
    157 		default:
    158 			t.Fatalf("invalid node %s", n.Data)
    159 		}
    160 	}
    161 
    162 	err = IndentHeadings(-5, topNodes...)
    163 	if err != nil {
    164 		t.Fatalf("%v", err)
    165 	}
    166 	for _, n := range topNodes {
    167 		if n.Type != html.ElementNode {
    168 			continue
    169 		}
    170 		switch n.Data {
    171 		case "div", "h1", "h2":
    172 		default:
    173 			t.Fatalf("invalid node %s", n.Data)
    174 		}
    175 	}
    176 }
    177 
    178 func TestUnwrap(t *testing.T) {
    179 	body, err := Body(getdoc(`<!DOCTYPE html>
    180 <div>TEST<!--TEST--></div>`))
    181 	if err != nil {
    182 		t.Fatalf("%v", err)
    183 	}
    184 	if body == nil {
    185 		t.Fatal("body is nil")
    186 	}
    187 	err = Unwrap(body.FirstChild)
    188 	if err != nil {
    189 		t.Fatalf("%v", err)
    190 	}
    191 	topNodes := Children(body)
    192 	if len(topNodes) != 2 {
    193 		t.Fatalf("2 nodes expected, found %d", len(topNodes))
    194 	}
    195 }