aboutsummaryrefslogtreecommitdiffhomepage
path: root/tests
diff options
context:
space:
mode:
authorArthurHoaro <arthur@hoa.ro>2017-09-30 11:04:13 +0200
committerArthurHoaro <arthur@hoa.ro>2017-10-28 14:35:49 +0200
commitd65342e304f92643ba922200953cfebc51e1e482 (patch)
tree3097c77bb4dd0590c4644422b5dc4369a4186eb7 /tests
parenta59bbf50d7530d7e82a91896a210b9da49cb1568 (diff)
downloadShaarli-d65342e304f92643ba922200953cfebc51e1e482.tar.gz
Shaarli-d65342e304f92643ba922200953cfebc51e1e482.tar.zst
Shaarli-d65342e304f92643ba922200953cfebc51e1e482.zip
Extract the title/charset during page download, and check content type
Use CURLOPT_WRITEFUNCTION to check the response code and content type (only allow HTML). Also extract the title and charset during downloading chunk of data, and stop it when everything has been extracted. Closes #579
Diffstat (limited to 'tests')
-rw-r--r--tests/LinkUtilsTest.php244
1 files changed, 224 insertions, 20 deletions
diff --git a/tests/LinkUtilsTest.php b/tests/LinkUtilsTest.php
index 7c0d4b0b..ef650f44 100644
--- a/tests/LinkUtilsTest.php
+++ b/tests/LinkUtilsTest.php
@@ -29,27 +29,13 @@ class LinkUtilsTest extends PHPUnit_Framework_TestCase
29 } 29 }
30 30
31 /** 31 /**
32 * Test get_charset() with all priorities.
33 */
34 public function testGetCharset()
35 {
36 $headers = array('Content-Type' => 'text/html; charset=Headers');
37 $html = '<html><meta>stuff</meta><meta charset="Html"/></html>';
38 $default = 'default';
39 $this->assertEquals('headers', get_charset($headers, $html, $default));
40 $this->assertEquals('html', get_charset(array(), $html, $default));
41 $this->assertEquals($default, get_charset(array(), '', $default));
42 $this->assertEquals('utf-8', get_charset(array(), ''));
43 }
44
45 /**
46 * Test headers_extract_charset() when the charset is found. 32 * Test headers_extract_charset() when the charset is found.
47 */ 33 */
48 public function testHeadersExtractExistentCharset() 34 public function testHeadersExtractExistentCharset()
49 { 35 {
50 $charset = 'x-MacCroatian'; 36 $charset = 'x-MacCroatian';
51 $headers = array('Content-Type' => 'text/html; charset='. $charset); 37 $headers = 'text/html; charset='. $charset;
52 $this->assertEquals(strtolower($charset), headers_extract_charset($headers)); 38 $this->assertEquals(strtolower($charset), header_extract_charset($headers));
53 } 39 }
54 40
55 /** 41 /**
@@ -57,11 +43,11 @@ class LinkUtilsTest extends PHPUnit_Framework_TestCase
57 */ 43 */
58 public function testHeadersExtractNonExistentCharset() 44 public function testHeadersExtractNonExistentCharset()
59 { 45 {
60 $headers = array(); 46 $headers = '';
61 $this->assertFalse(headers_extract_charset($headers)); 47 $this->assertFalse(header_extract_charset($headers));
62 48
63 $headers = array('Content-Type' => 'text/html'); 49 $headers = 'text/html';
64 $this->assertFalse(headers_extract_charset($headers)); 50 $this->assertFalse(header_extract_charset($headers));
65 } 51 }
66 52
67 /** 53 /**
@@ -86,6 +72,131 @@ class LinkUtilsTest extends PHPUnit_Framework_TestCase
86 } 72 }
87 73
88 /** 74 /**
75 * Test the download callback with valid value
76 */
77 public function testCurlDownloadCallbackOk()
78 {
79 $callback = get_curl_download_callback($charset, $title, 'ut_curl_getinfo_ok');
80 $data = [
81 'HTTP/1.1 200 OK',
82 'Server: GitHub.com',
83 'Date: Sat, 28 Oct 2017 12:01:33 GMT',
84 'Content-Type: text/html; charset=utf-8',
85 'Status: 200 OK',
86 'end' => 'th=device-width"><title>Refactoring · GitHub</title><link rel="search" type="application/opensea',
87 '<title>ignored</title>',
88 ];
89 foreach ($data as $key => $line) {
90 $ignore = null;
91 $expected = $key !== 'end' ? strlen($line) : false;
92 $this->assertEquals($expected, $callback($ignore, $line));
93 if ($expected === false) {
94 break;
95 }
96 }
97 $this->assertEquals('utf-8', $charset);
98 $this->assertEquals('Refactoring · GitHub', $title);
99 }
100
101 /**
102 * Test the download callback with valid values and no charset
103 */
104 public function testCurlDownloadCallbackOkNoCharset()
105 {
106 $callback = get_curl_download_callback($charset, $title, 'ut_curl_getinfo_no_charset');
107 $data = [
108 'HTTP/1.1 200 OK',
109 'end' => 'th=device-width"><title>Refactoring · GitHub</title><link rel="search" type="application/opensea',
110 '<title>ignored</title>',
111 ];
112 foreach ($data as $key => $line) {
113 $ignore = null;
114 $this->assertEquals(strlen($line), $callback($ignore, $line));
115 }
116 $this->assertEmpty($charset);
117 $this->assertEquals('Refactoring · GitHub', $title);
118 }
119
120 /**
121 * Test the download callback with valid values and no charset
122 */
123 public function testCurlDownloadCallbackOkHtmlCharset()
124 {
125 $callback = get_curl_download_callback($charset, $title, 'ut_curl_getinfo_no_charset');
126 $data = [
127 'HTTP/1.1 200 OK',
128 '<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />',
129 'end' => 'th=device-width"><title>Refactoring · GitHub</title><link rel="search" type="application/opensea',
130 '<title>ignored</title>',
131 ];
132 foreach ($data as $key => $line) {
133 $ignore = null;
134 $expected = $key !== 'end' ? strlen($line) : false;
135 $this->assertEquals($expected, $callback($ignore, $line));
136 if ($expected === false) {
137 break;
138 }
139 }
140 $this->assertEquals('utf-8', $charset);
141 $this->assertEquals('Refactoring · GitHub', $title);
142 }
143
144 /**
145 * Test the download callback with valid values and no title
146 */
147 public function testCurlDownloadCallbackOkNoTitle()
148 {
149 $callback = get_curl_download_callback($charset, $title, 'ut_curl_getinfo_ok');
150 $data = [
151 'HTTP/1.1 200 OK',
152 'end' => 'th=device-width">Refactoring · GitHub<link rel="search" type="application/opensea',
153 'ignored',
154 ];
155 foreach ($data as $key => $line) {
156 $ignore = null;
157 $this->assertEquals(strlen($line), $callback($ignore, $line));
158 }
159 $this->assertEquals('utf-8', $charset);
160 $this->assertEmpty($title);
161 }
162
163 /**
164 * Test the download callback with an invalid content type.
165 */
166 public function testCurlDownloadCallbackInvalidContentType()
167 {
168 $callback = get_curl_download_callback($charset, $title, 'ut_curl_getinfo_ct_ko');
169 $ignore = null;
170 $this->assertFalse($callback($ignore, ''));
171 $this->assertEmpty($charset);
172 $this->assertEmpty($title);
173 }
174
175 /**
176 * Test the download callback with an invalid response code.
177 */
178 public function testCurlDownloadCallbackInvalidResponseCode()
179 {
180 $callback = get_curl_download_callback($charset, $title, 'ut_curl_getinfo_rc_ko');
181 $ignore = null;
182 $this->assertFalse($callback($ignore, ''));
183 $this->assertEmpty($charset);
184 $this->assertEmpty($title);
185 }
186
187 /**
188 * Test the download callback with an invalid content type and response code.
189 */
190 public function testCurlDownloadCallbackInvalidContentTypeAndResponseCode()
191 {
192 $callback = get_curl_download_callback($charset, $title, 'ut_curl_getinfo_rs_ct_ko');
193 $ignore = null;
194 $this->assertFalse($callback($ignore, ''));
195 $this->assertEmpty($charset);
196 $this->assertEmpty($title);
197 }
198
199 /**
89 * Test count_private. 200 * Test count_private.
90 */ 201 */
91 public function testCountPrivateLinks() 202 public function testCountPrivateLinks()
@@ -182,3 +293,96 @@ class LinkUtilsTest extends PHPUnit_Framework_TestCase
182 return str_replace('$1', $hashtag, $hashtagLink); 293 return str_replace('$1', $hashtag, $hashtagLink);
183 } 294 }
184} 295}
296
297// old style mock: PHPUnit doesn't allow function mock
298
299/**
300 * Returns code 200 or html content type.
301 *
302 * @param resource $ch cURL resource
303 * @param int $type cURL info type
304 *
305 * @return int|string 200 or 'text/html'
306 */
307function ut_curl_getinfo_ok($ch, $type)
308{
309 switch ($type) {
310 case CURLINFO_RESPONSE_CODE:
311 return 200;
312 case CURLINFO_CONTENT_TYPE:
313 return 'text/html; charset=utf-8';
314 }
315}
316
317/**
318 * Returns code 200 or html content type without charset.
319 *
320 * @param resource $ch cURL resource
321 * @param int $type cURL info type
322 *
323 * @return int|string 200 or 'text/html'
324 */
325function ut_curl_getinfo_no_charset($ch, $type)
326{
327 switch ($type) {
328 case CURLINFO_RESPONSE_CODE:
329 return 200;
330 case CURLINFO_CONTENT_TYPE:
331 return 'text/html';
332 }
333}
334
335/**
336 * Invalid response code.
337 *
338 * @param resource $ch cURL resource
339 * @param int $type cURL info type
340 *
341 * @return int|string 404 or 'text/html'
342 */
343function ut_curl_getinfo_rc_ko($ch, $type)
344{
345 switch ($type) {
346 case CURLINFO_RESPONSE_CODE:
347 return 404;
348 case CURLINFO_CONTENT_TYPE:
349 return 'text/html; charset=utf-8';
350 }
351}
352
353/**
354 * Invalid content type.
355 *
356 * @param resource $ch cURL resource
357 * @param int $type cURL info type
358 *
359 * @return int|string 200 or 'text/plain'
360 */
361function ut_curl_getinfo_ct_ko($ch, $type)
362{
363 switch ($type) {
364 case CURLINFO_RESPONSE_CODE:
365 return 200;
366 case CURLINFO_CONTENT_TYPE:
367 return 'text/plain';
368 }
369}
370
371/**
372 * Invalid response code and content type.
373 *
374 * @param resource $ch cURL resource
375 * @param int $type cURL info type
376 *
377 * @return int|string 404 or 'text/plain'
378 */
379function ut_curl_getinfo_rs_ct_ko($ch, $type)
380{
381 switch ($type) {
382 case CURLINFO_RESPONSE_CODE:
383 return 404;
384 case CURLINFO_CONTENT_TYPE:
385 return 'text/plain';
386 }
387}
388
generated by cgit v1.2.3 (git 2.25.1) at 2024-11-22 00:46:10 +0000