17
type ReplacerTest struct {
23
var htmlEscaper = NewReplacer("&", "&", "<", "<", ">", ">", "\"", """)
14
var htmlEscaper = NewReplacer(
22
var htmlUnescaper = NewReplacer(
25
30
// The http package's old HTML escaping function.
26
func oldhtmlEscape(s string) string {
31
func oldHTMLEscape(s string) string {
27
32
s = Replace(s, "&", "&", -1)
28
33
s = Replace(s, "<", "<", -1)
29
34
s = Replace(s, ">", ">", -1)
30
s = Replace(s, "\"", """, -1)
35
s = Replace(s, `"`, """, -1)
31
36
s = Replace(s, "'", "'", -1)
35
var replacer = NewReplacer("aaa", "3[aaa]", "aa", "2[aa]", "a", "1[a]", "i", "i",
36
"longerst", "most long", "longer", "medium", "long", "short",
39
40
var capitalLetters = NewReplacer("a", "A", "b", "B")
41
var blankToXReplacer = NewReplacer("", "X", "o", "O")
43
var ReplacerTests = []ReplacerTest{
45
{htmlEscaper, "No changes", "No changes"},
46
{htmlEscaper, "I <3 escaping & stuff", "I <3 escaping & stuff"},
47
{htmlEscaper, "&&&", "&&&"},
50
{replacer, "fooaaabar", "foo3[aaa]b1[a]r"},
51
{replacer, "long, longerst, longer", "short, most long, medium"},
52
{replacer, "XiX", "YiY"},
55
{capitalLetters, "brad", "BrAd"},
56
{capitalLetters, Repeat("a", (32<<10)+123), Repeat("A", (32<<10)+123)},
58
// hitting "" special case
59
{blankToXReplacer, "oo", "XOXOX"},
42
// TestReplacer tests the replacer implementations.
62
43
func TestReplacer(t *testing.T) {
63
for i, tt := range ReplacerTests {
64
if s := tt.r.Replace(tt.in); s != tt.out {
65
t.Errorf("%d. Replace(%q) = %q, want %q", i, tt.in, s, tt.out)
44
type testCase struct {
48
var testCases []testCase
50
// str converts 0xff to "\xff". This isn't just string(b) since that converts to UTF-8.
51
str := func(b byte) string {
52
return string([]byte{b})
56
// inc maps "\x00"->"\x01", ..., "a"->"b", "b"->"c", ..., "\xff"->"\x00".
58
for i := 0; i < 256; i++ {
59
s = append(s, str(byte(i)), str(byte(i+1)))
61
inc := NewReplacer(s...)
63
// Test cases with 1-byte old strings, 1-byte new strings.
64
testCases = append(testCases,
65
testCase{capitalLetters, "brad", "BrAd"},
66
testCase{capitalLetters, Repeat("a", (32<<10)+123), Repeat("A", (32<<10)+123)},
67
testCase{capitalLetters, "", ""},
69
testCase{inc, "brad", "csbe"},
70
testCase{inc, "\x00\xff", "\x01\x00"},
71
testCase{inc, "", ""},
73
testCase{NewReplacer("a", "1", "a", "2"), "brad", "br1d"},
76
// repeat maps "a"->"a", "b"->"bb", "c"->"ccc", ...
78
for i := 0; i < 256; i++ {
83
s = append(s, str(byte(i)), Repeat(str(byte(i)), n))
85
repeat := NewReplacer(s...)
87
// Test cases with 1-byte old strings, variable length new strings.
88
testCases = append(testCases,
89
testCase{htmlEscaper, "No changes", "No changes"},
90
testCase{htmlEscaper, "I <3 escaping & stuff", "I <3 escaping & stuff"},
91
testCase{htmlEscaper, "&&&", "&&&"},
92
testCase{htmlEscaper, "", ""},
94
testCase{repeat, "brad", "bbrrrrrrrrrrrrrrrrrradddd"},
95
testCase{repeat, "abba", "abbbba"},
96
testCase{repeat, "", ""},
98
testCase{NewReplacer("a", "11", "a", "22"), "brad", "br11d"},
101
// The remaining test cases have variable length old strings.
103
testCases = append(testCases,
104
testCase{htmlUnescaper, "&amp;", "&"},
105
testCase{htmlUnescaper, "<b>HTML's neat</b>", "<b>HTML's neat</b>"},
106
testCase{htmlUnescaper, "", ""},
108
testCase{NewReplacer("a", "1", "a", "2", "xxx", "xxx"), "brad", "br1d"},
110
testCase{NewReplacer("a", "1", "aa", "2", "aaa", "3"), "aaaa", "1111"},
112
testCase{NewReplacer("aaa", "3", "aa", "2", "a", "1"), "aaaa", "31"},
115
// gen1 has multiple old strings of variable length. There is no
116
// overall non-empty common prefix, but some pairwise common prefixes.
122
"longerst", "most long",
130
testCases = append(testCases,
131
testCase{gen1, "fooaaabar", "foo3[aaa]b1[a]r"},
132
testCase{gen1, "long, longerst, longer", "short, most long, medium"},
133
testCase{gen1, "xxxxx", "xxxxX"},
134
testCase{gen1, "XiX", "YiY"},
135
testCase{gen1, "", ""},
138
// gen2 has multiple old strings with no pairwise common prefix.
144
testCases = append(testCases,
145
testCase{gen2, "roses are red, violets are blue...", "red are red, blue are blue..."},
146
testCase{gen2, "", ""},
149
// gen3 has multiple old strings with an overall common prefix.
151
"abracadabra", "poof",
152
"abracadabrakazam", "splat",
153
"abraham", "lincoln",
154
"abrasion", "scrape",
157
testCases = append(testCases,
158
testCase{gen3, "abracadabrakazam abraham", "poofkazam lincoln"},
159
testCase{gen3, "abrasion abracad", "scrape abracad"},
160
testCase{gen3, "abba abram abrasive", "abba abram abrasive"},
161
testCase{gen3, "", ""},
164
// foo{1,2,3,4} have multiple old strings with an overall common prefix
165
// and 1- or 2- byte extensions from the common prefix.
187
testCases = append(testCases,
188
testCase{foo1, "fofoofoo12foo32oo", "fofooA2C2oo"},
189
testCase{foo1, "", ""},
191
testCase{foo2, "fofoofoo12foo32oo", "fofooA2Doo"},
192
testCase{foo2, "", ""},
194
testCase{foo3, "fofoofoo12foo32oo", "fofooBDoo"},
195
testCase{foo3, "", ""},
197
testCase{foo4, "fofoofoo12foo32oo", "fofooBDoo"},
198
testCase{foo4, "", ""},
201
// genAll maps "\x00\x01\x02...\xfe\xff" to "[all]", amongst other things.
202
allBytes := make([]byte, 256)
203
for i := range allBytes {
204
allBytes[i] = byte(i)
206
allString := string(allBytes)
207
genAll := NewReplacer(
212
testCases = append(testCases,
213
testCase{genAll, allString, "[all]"},
214
testCase{genAll, "a\xff" + allString + "\x00", "a[ff][all][00]"},
215
testCase{genAll, "", ""},
218
// Test cases with empty old strings.
220
blankToX1 := NewReplacer("", "X")
221
blankToX2 := NewReplacer("", "X", "", "")
222
blankHighPriority := NewReplacer("", "X", "o", "O")
223
blankLowPriority := NewReplacer("o", "O", "", "X")
224
blankNoOp1 := NewReplacer("", "")
225
blankNoOp2 := NewReplacer("", "", "", "A")
226
blankFoo := NewReplacer("", "X", "foobar", "R", "foobaz", "Z")
227
testCases = append(testCases,
228
testCase{blankToX1, "foo", "XfXoXoX"},
229
testCase{blankToX1, "", "X"},
231
testCase{blankToX2, "foo", "XfXoXoX"},
232
testCase{blankToX2, "", "X"},
234
testCase{blankHighPriority, "oo", "XOXOX"},
235
testCase{blankHighPriority, "ii", "XiXiX"},
236
testCase{blankHighPriority, "oiio", "XOXiXiXOX"},
237
testCase{blankHighPriority, "iooi", "XiXOXOXiX"},
238
testCase{blankHighPriority, "", "X"},
240
testCase{blankLowPriority, "oo", "OOX"},
241
testCase{blankLowPriority, "ii", "XiXiX"},
242
testCase{blankLowPriority, "oiio", "OXiXiOX"},
243
testCase{blankLowPriority, "iooi", "XiOOXiX"},
244
testCase{blankLowPriority, "", "X"},
246
testCase{blankNoOp1, "foo", "foo"},
247
testCase{blankNoOp1, "", ""},
249
testCase{blankNoOp2, "foo", "foo"},
250
testCase{blankNoOp2, "", ""},
252
testCase{blankFoo, "foobarfoobaz", "XRXZX"},
253
testCase{blankFoo, "foobar-foobaz", "XRX-XZX"},
254
testCase{blankFoo, "", "X"},
257
// single string replacer
259
abcMatcher := NewReplacer("abc", "[match]")
261
testCases = append(testCases,
262
testCase{abcMatcher, "", ""},
263
testCase{abcMatcher, "ab", "ab"},
264
testCase{abcMatcher, "abcd", "[match]d"},
265
testCase{abcMatcher, "cabcabcdabca", "c[match][match]d[match]a"},
268
// No-arg test cases.
271
testCases = append(testCases,
272
testCase{nop, "abc", "abc"},
273
testCase{nop, "", ""},
276
// Run the test cases.
278
for i, tc := range testCases {
279
if s := tc.r.Replace(tc.in); s != tc.out {
280
t.Errorf("%d. Replace(%q) = %q, want %q", i, tc.in, s, tc.out)
67
282
var buf bytes.Buffer
68
n, err := tt.r.WriteString(&buf, tt.in)
283
n, err := tc.r.WriteString(&buf, tc.in)
70
285
t.Errorf("%d. WriteString: %v", i, err)
73
288
got := buf.String()
75
t.Errorf("%d. WriteString(%q) wrote %q, want %q", i, tt.in, got, tt.out)
290
t.Errorf("%d. WriteString(%q) wrote %q, want %q", i, tc.in, got, tc.out)
293
if n != len(tc.out) {
79
294
t.Errorf("%d. WriteString(%q) wrote correct string but reported %d bytes; want %d (%q)",
80
i, tt.in, n, len(tt.out), tt.out)
295
i, tc.in, n, len(tc.out), tc.out)
85
// pickAlgorithmTest is a test that verifies that given input for a
86
// Replacer that we pick the correct algorithm.
87
type pickAlgorithmTest struct {
89
want string // name of algorithm
92
var pickAlgorithmTests = []pickAlgorithmTest{
93
{capitalLetters, "*strings.byteReplacer"},
94
{NewReplacer("12", "123"), "*strings.genericReplacer"},
95
{NewReplacer("1", "12"), "*strings.byteStringReplacer"},
96
{htmlEscaper, "*strings.byteStringReplacer"},
300
// TestPickAlgorithm tests that NewReplacer picks the correct algorithm.
99
301
func TestPickAlgorithm(t *testing.T) {
100
for i, tt := range pickAlgorithmTests {
101
got := fmt.Sprintf("%T", tt.r.Replacer())
103
t.Errorf("%d. algorithm = %s, want %s", i, got, tt.want)
108
func BenchmarkGenericMatch(b *testing.B) {
302
testCases := []struct {
306
{capitalLetters, "*strings.byteReplacer"},
307
{htmlEscaper, "*strings.byteStringReplacer"},
308
{NewReplacer("12", "123"), "*strings.singleStringReplacer"},
309
{NewReplacer("1", "12"), "*strings.byteStringReplacer"},
310
{NewReplacer("", "X"), "*strings.genericReplacer"},
311
{NewReplacer("a", "1", "b", "12", "cde", "123"), "*strings.genericReplacer"},
313
for i, tc := range testCases {
314
got := fmt.Sprintf("%T", tc.r.Replacer())
316
t.Errorf("%d. algorithm = %s, want %s", i, got, tc.want)
321
// TestGenericTrieBuilding verifies the structure of the generated trie. There
322
// is one node per line, and the key ending with the current line is in the
323
// trie if it ends with a "+".
324
func TestGenericTrieBuilding(t *testing.T) {
325
testCases := []struct{ in, out string }{
326
{"abc;abdef;abdefgh;xx;xy;z", `-
338
{"abracadabra;abracadabrakazam;abraham;abrasion", `-
349
{"aaa;aa;a;i;longerst;longer;long;xx;x;X;Y", `-
370
for _, tc := range testCases {
371
keys := Split(tc.in, ";")
372
args := make([]string, len(keys)*2)
373
for i, key := range keys {
377
got := NewReplacer(args...).PrintTrie()
378
// Remove tabs from tc.out
379
wantbuf := make([]byte, 0, len(tc.out))
380
for i := 0; i < len(tc.out); i++ {
381
if tc.out[i] != '\t' {
382
wantbuf = append(wantbuf, tc.out[i])
385
want := string(wantbuf)
388
t.Errorf("PrintTrie(%q)\ngot\n%swant\n%s", tc.in, got, want)
393
func BenchmarkGenericNoMatch(b *testing.B) {
109
394
str := Repeat("A", 100) + Repeat("B", 100)
110
395
generic := NewReplacer("a", "A", "b", "B", "12", "123") // varying lengths forces generic
111
396
for i := 0; i < b.N; i++ {