Skip to content

Commit 4e3cd44

Browse files
committed
Add tests for InputStream (all failures are due to our use to iconv).
1 parent 232fc09 commit 4e3cd44

File tree

1 file changed

+174
-0
lines changed

1 file changed

+174
-0
lines changed

tests/HTML5/InputStreamTest.php

Lines changed: 174 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,174 @@
1+
<?php
2+
3+
require_once dirname(__FILE__) . '/../autorun.php';
4+
5+
class HTML5_InputStreamTest extends UnitTestCase
6+
{
7+
public function invalidReplaceTestHandler($input, $name) {
8+
$stream = new HTML5_InputStream($input);
9+
$this->assertIdentical("\xEF\xBF\xBD", $stream->remainingChars(), $name);
10+
}
11+
12+
public function testInvalidReplace() {
13+
// Above U+10FFFF
14+
$this->invalidReplaceTestHandler("\xF5\x90\x80\x80", 'U+110000');
15+
16+
// Incomplete
17+
$this->invalidReplaceTestHandler("\xDF", 'Incomplete two byte sequence (missing final byte)');
18+
$this->invalidReplaceTestHandler("\xEF\xBF", 'Incomplete three byte sequence (missing final byte)');
19+
$this->invalidReplaceTestHandler("\xF4\xBF\xBF", 'Incomplete four byte sequence (missing final byte)');
20+
21+
// Min/max continuation bytes
22+
$this->invalidReplaceTestHandler("\x80", 'Lone 80 continuation byte');
23+
$this->invalidReplaceTestHandler("\xBF", 'Lone BF continuation byte');
24+
25+
// Invalid bytes (these can never occur)
26+
$this->invalidReplaceTestHandler("\xFE", 'Invalid FE byte');
27+
$this->invalidReplaceTestHandler("\xFF", 'Invalid FF byte');
28+
29+
// Min/max overlong
30+
$this->invalidReplaceTestHandler("\xC0\x80", 'Overlong representation of U+0000');
31+
$this->invalidReplaceTestHandler("\xE0\x80\x80", 'Overlong representation of U+0000');
32+
$this->invalidReplaceTestHandler("\xF0\x80\x80\x80", 'Overlong representation of U+0000');
33+
$this->invalidReplaceTestHandler("\xF8\x80\x80\x80\x80", 'Overlong representation of U+0000');
34+
$this->invalidReplaceTestHandler("\xFC\x80\x80\x80\x80\x80", 'Overlong representation of U+0000');
35+
$this->invalidReplaceTestHandler("\xC1\xBF", 'Overlong representation of U+007F');
36+
$this->invalidReplaceTestHandler("\xE0\x9F\xBF", 'Overlong representation of U+07FF');
37+
$this->invalidReplaceTestHandler("\xF0\x8F\xBF\xBF", 'Overlong representation of U+FFFF');
38+
}
39+
40+
public function testStripLeadingBOM() {
41+
$leading = new HTML5_InputStream("\xEF\xBB\xBFa");
42+
$this->assertIdentical('a', $leading->char(), 'BOM should be stripped');
43+
}
44+
45+
public function testZWNBSP() {
46+
$stream = new HTML5_InputStream("a\xEF\xBB\xBF");
47+
$this->assertIdentical("a\xEF\xBB\xBF", $stream->remainingChars(), 'A non-leading U+FEFF (BOM/ZWNBSP) should remain');
48+
}
49+
50+
public function testNull() {
51+
$stream = new HTML5_InputStream("\0\0\0");
52+
$this->assertIdentical("\xEF\xBF\xBD\xEF\xBF\xBD\xEF\xBF\xBD", $stream->remainingChars(), 'Null character should be replaced by U+FFFD');
53+
$this->assertIdentical(3, count($stream->errors), 'Null character should be throw parse error');
54+
}
55+
56+
public function testCRLF() {
57+
$stream = new HTML5_InputStream("\r\n");
58+
$this->assertIdentical("\n", $stream->remainingChars(), 'CRLF should be replaced by LF');
59+
}
60+
61+
public function testCR() {
62+
$stream = new HTML5_InputStream("\r");
63+
$this->assertIdentical("\n", $stream->remainingChars(), 'CR should be replaced by LF');
64+
}
65+
66+
public function invalidParseErrorTestHandler($input, $numErrors, $name) {
67+
$stream = new HTML5_InputStream($input);
68+
$this->assertIdentical($input, $stream->remainingChars(), $name . ' (stream content)');
69+
$this->assertIdentical($numErrors, count($stream->errors), $name . ' (number of errors)');
70+
}
71+
72+
public function testInvalidParseError() {
73+
// C0 controls (except U+0000 and U+000D due to different handling)
74+
$this->invalidParseErrorTestHandler("\x01", 1, 'U+0001 (C0 control)');
75+
$this->invalidParseErrorTestHandler("\x02", 1, 'U+0002 (C0 control)');
76+
$this->invalidParseErrorTestHandler("\x03", 1, 'U+0003 (C0 control)');
77+
$this->invalidParseErrorTestHandler("\x04", 1, 'U+0004 (C0 control)');
78+
$this->invalidParseErrorTestHandler("\x05", 1, 'U+0005 (C0 control)');
79+
$this->invalidParseErrorTestHandler("\x06", 1, 'U+0006 (C0 control)');
80+
$this->invalidParseErrorTestHandler("\x07", 1, 'U+0007 (C0 control)');
81+
$this->invalidParseErrorTestHandler("\x08", 1, 'U+0008 (C0 control)');
82+
$this->invalidParseErrorTestHandler("\x09", 0, 'U+0009 (C0 control)');
83+
$this->invalidParseErrorTestHandler("\x0A", 0, 'U+000A (C0 control)');
84+
$this->invalidParseErrorTestHandler("\x0B", 1, 'U+000B (C0 control)');
85+
$this->invalidParseErrorTestHandler("\x0C", 0, 'U+000C (C0 control)');
86+
$this->invalidParseErrorTestHandler("\x0E", 1, 'U+000E (C0 control)');
87+
$this->invalidParseErrorTestHandler("\x0F", 1, 'U+000F (C0 control)');
88+
$this->invalidParseErrorTestHandler("\x10", 1, 'U+0010 (C0 control)');
89+
$this->invalidParseErrorTestHandler("\x11", 1, 'U+0011 (C0 control)');
90+
$this->invalidParseErrorTestHandler("\x12", 1, 'U+0012 (C0 control)');
91+
$this->invalidParseErrorTestHandler("\x13", 1, 'U+0013 (C0 control)');
92+
$this->invalidParseErrorTestHandler("\x14", 1, 'U+0014 (C0 control)');
93+
$this->invalidParseErrorTestHandler("\x15", 1, 'U+0015 (C0 control)');
94+
$this->invalidParseErrorTestHandler("\x16", 1, 'U+0016 (C0 control)');
95+
$this->invalidParseErrorTestHandler("\x17", 1, 'U+0017 (C0 control)');
96+
$this->invalidParseErrorTestHandler("\x18", 1, 'U+0018 (C0 control)');
97+
$this->invalidParseErrorTestHandler("\x19", 1, 'U+0019 (C0 control)');
98+
$this->invalidParseErrorTestHandler("\x1A", 1, 'U+001A (C0 control)');
99+
$this->invalidParseErrorTestHandler("\x1B", 1, 'U+001B (C0 control)');
100+
$this->invalidParseErrorTestHandler("\x1C", 1, 'U+001C (C0 control)');
101+
$this->invalidParseErrorTestHandler("\x1D", 1, 'U+001D (C0 control)');
102+
$this->invalidParseErrorTestHandler("\x1E", 1, 'U+001E (C0 control)');
103+
$this->invalidParseErrorTestHandler("\x1F", 1, 'U+001F (C0 control)');
104+
105+
// DEL (U+007F)
106+
$this->invalidParseErrorTestHandler("\x7F", 1, 'U+007F');
107+
108+
// C1 Controls
109+
$this->invalidParseErrorTestHandler("\xC2\x80", 1, 'U+0080 (C1 control)');
110+
$this->invalidParseErrorTestHandler("\xC2\x9F", 1, 'U+009F (C1 control)');
111+
$this->invalidParseErrorTestHandler("\xC2\xA0", 0, 'U+00A0 (first codepoint above highest C1 control)');
112+
113+
// Single UTF-16 surrogates
114+
$this->invalidParseErrorTestHandler("\xED\xA0\x80", 1, 'U+D800 (UTF-16 surrogate character)');
115+
$this->invalidParseErrorTestHandler("\xED\xAD\xBF", 1, 'U+DB7F (UTF-16 surrogate character)');
116+
$this->invalidParseErrorTestHandler("\xED\xAE\x80", 1, 'U+DB80 (UTF-16 surrogate character)');
117+
$this->invalidParseErrorTestHandler("\xED\xAF\xBF", 1, 'U+DBFF (UTF-16 surrogate character)');
118+
$this->invalidParseErrorTestHandler("\xED\xB0\x80", 1, 'U+DC00 (UTF-16 surrogate character)');
119+
$this->invalidParseErrorTestHandler("\xED\xBE\x80", 1, 'U+DF80 (UTF-16 surrogate character)');
120+
$this->invalidParseErrorTestHandler("\xED\xBF\xBF", 1, 'U+DFFF (UTF-16 surrogate character)');
121+
122+
// Paired UTF-16 surrogates
123+
$this->invalidParseErrorTestHandler("\xED\xA0\x80\xED\xB0\x80", 2, 'U+D800 U+DC00 (paired UTF-16 surrogates)');
124+
$this->invalidParseErrorTestHandler("\xED\xA0\x80\xED\xBF\xBF", 2, 'U+D800 U+DFFF (paired UTF-16 surrogates)');
125+
$this->invalidParseErrorTestHandler("\xED\xAD\xBF\xED\xB0\x80", 2, 'U+DB7F U+DC00 (paired UTF-16 surrogates)');
126+
$this->invalidParseErrorTestHandler("\xED\xAD\xBF\xED\xBF\xBF", 2, 'U+DB7F U+DFFF (paired UTF-16 surrogates)');
127+
$this->invalidParseErrorTestHandler("\xED\xAE\x80\xED\xB0\x80", 2, 'U+DB80 U+DC00 (paired UTF-16 surrogates)');
128+
$this->invalidParseErrorTestHandler("\xED\xAE\x80\xED\xBF\xBF", 2, 'U+DB80 U+DFFF (paired UTF-16 surrogates)');
129+
$this->invalidParseErrorTestHandler("\xED\xAF\xBF\xED\xB0\x80", 2, 'U+DBFF U+DC00 (paired UTF-16 surrogates)');
130+
$this->invalidParseErrorTestHandler("\xED\xAF\xBF\xED\xBF\xBF", 2, 'U+DBFF U+DFFF (paired UTF-16 surrogates)');
131+
132+
// Charcters surrounding surrogates
133+
$this->invalidParseErrorTestHandler("\xED\x9F\xBF", 0, 'U+D7FF (one codepoint below lowest surrogate codepoint)');
134+
$this->invalidParseErrorTestHandler("\xEF\xBF\xBD", 0, 'U+DE00 (one codepoint above highest surrogate codepoint)');
135+
136+
// Permanent noncharacters
137+
$this->invalidParseErrorTestHandler("\xEF\xB7\x90", 1, 'U+FDD0 (permanent noncharacter)');
138+
$this->invalidParseErrorTestHandler("\xEF\xB7\xAF", 1, 'U+FDEF (permanent noncharacter)');
139+
$this->invalidParseErrorTestHandler("\xEF\xBF\xBE", 1, 'U+FFFE (permanent noncharacter)');
140+
$this->invalidParseErrorTestHandler("\xEF\xBF\xBF", 1, 'U+FFFF (permanent noncharacter)');
141+
$this->invalidParseErrorTestHandler("\xF0\x9F\xBF\xBE", 1, 'U+1FFFE (permanent noncharacter)');
142+
$this->invalidParseErrorTestHandler("\xF0\x9F\xBF\xBF", 1, 'U+1FFFF (permanent noncharacter)');
143+
$this->invalidParseErrorTestHandler("\xF0\xAF\xBF\xBE", 1, 'U+2FFFE (permanent noncharacter)');
144+
$this->invalidParseErrorTestHandler("\xF0\xAF\xBF\xBF", 1, 'U+2FFFF (permanent noncharacter)');
145+
$this->invalidParseErrorTestHandler("\xF0\xBF\xBF\xBE", 1, 'U+3FFFE (permanent noncharacter)');
146+
$this->invalidParseErrorTestHandler("\xF0\xBF\xBF\xBF", 1, 'U+3FFFF (permanent noncharacter)');
147+
$this->invalidParseErrorTestHandler("\xF1\x8F\xBF\xBE", 1, 'U+4FFFE (permanent noncharacter)');
148+
$this->invalidParseErrorTestHandler("\xF1\x8F\xBF\xBF", 1, 'U+4FFFF (permanent noncharacter)');
149+
$this->invalidParseErrorTestHandler("\xF1\x9F\xBF\xBE", 1, 'U+5FFFE (permanent noncharacter)');
150+
$this->invalidParseErrorTestHandler("\xF1\x9F\xBF\xBF", 1, 'U+5FFFF (permanent noncharacter)');
151+
$this->invalidParseErrorTestHandler("\xF1\xAF\xBF\xBE", 1, 'U+6FFFE (permanent noncharacter)');
152+
$this->invalidParseErrorTestHandler("\xF1\xAF\xBF\xBF", 1, 'U+6FFFF (permanent noncharacter)');
153+
$this->invalidParseErrorTestHandler("\xF1\xBF\xBF\xBE", 1, 'U+7FFFE (permanent noncharacter)');
154+
$this->invalidParseErrorTestHandler("\xF1\xBF\xBF\xBF", 1, 'U+7FFFF (permanent noncharacter)');
155+
$this->invalidParseErrorTestHandler("\xF2\x8F\xBF\xBE", 1, 'U+8FFFE (permanent noncharacter)');
156+
$this->invalidParseErrorTestHandler("\xF2\x8F\xBF\xBF", 1, 'U+8FFFF (permanent noncharacter)');
157+
$this->invalidParseErrorTestHandler("\xF2\x9F\xBF\xBE", 1, 'U+9FFFE (permanent noncharacter)');
158+
$this->invalidParseErrorTestHandler("\xF2\x9F\xBF\xBF", 1, 'U+9FFFF (permanent noncharacter)');
159+
$this->invalidParseErrorTestHandler("\xF2\xAF\xBF\xBE", 1, 'U+AFFFE (permanent noncharacter)');
160+
$this->invalidParseErrorTestHandler("\xF2\xAF\xBF\xBF", 1, 'U+AFFFF (permanent noncharacter)');
161+
$this->invalidParseErrorTestHandler("\xF2\xBF\xBF\xBE", 1, 'U+BFFFE (permanent noncharacter)');
162+
$this->invalidParseErrorTestHandler("\xF2\xBF\xBF\xBF", 1, 'U+BFFFF (permanent noncharacter)');
163+
$this->invalidParseErrorTestHandler("\xF3\x8F\xBF\xBE", 1, 'U+CFFFE (permanent noncharacter)');
164+
$this->invalidParseErrorTestHandler("\xF3\x8F\xBF\xBF", 1, 'U+CFFFF (permanent noncharacter)');
165+
$this->invalidParseErrorTestHandler("\xF3\x9F\xBF\xBE", 1, 'U+DFFFE (permanent noncharacter)');
166+
$this->invalidParseErrorTestHandler("\xF3\x9F\xBF\xBF", 1, 'U+DFFFF (permanent noncharacter)');
167+
$this->invalidParseErrorTestHandler("\xF3\xAF\xBF\xBE", 1, 'U+EFFFE (permanent noncharacter)');
168+
$this->invalidParseErrorTestHandler("\xF3\xAF\xBF\xBF", 1, 'U+EFFFF (permanent noncharacter)');
169+
$this->invalidParseErrorTestHandler("\xF3\xBF\xBF\xBE", 1, 'U+FFFFE (permanent noncharacter)');
170+
$this->invalidParseErrorTestHandler("\xF3\xBF\xBF\xBF", 1, 'U+FFFFF (permanent noncharacter)');
171+
$this->invalidParseErrorTestHandler("\xF4\x8F\xBF\xBE", 1, 'U+10FFFE (permanent noncharacter)');
172+
$this->invalidParseErrorTestHandler("\xF4\x8F\xBF\xBF", 1, 'U+10FFFF (permanent noncharacter)');
173+
}
174+
}

0 commit comments

Comments
 (0)