add more tests, make standard reply tests more relaxed

2015-12-18 18:56:41 -08:00
parent 41457d8fbd
commit 3d9ae356ea
5 changed files with 134 additions and 77 deletions
--- a/tests/utils_test.py
+++ b/tests/utils_test.py
@@ -58,3 +58,50 @@ def test_quick_detect_encoding_edge_cases(detect_encoding, cchardet_detect):
    detect_encoding.return_value = 'utf-8'
    eq_('utf-8', u.quick_detect_encoding("qwe"))
    ok_(detect_encoding.called)
+
+
+def test_html_to_text():
+    html = """<body>
+<p>Hello world!</p>
+<br>
+<ul>
+<li>One!</li>
+<li>Two</li>
+</ul>
+<p>
+Haha
+</p>
+</body>"""
+    text = u.html_to_text(html)
+    eq_("Hello world! \n\n  * One! \n  * Two \nHaha", text)
+    eq_("привет!", u.html_to_text("<b>привет!</b>"))
+
+    html = '<body><br/><br/>Hi</body>'
+    eq_ ('Hi', u.html_to_text(html))
+
+    html = """Hi
+<style type="text/css">
+
+div, p, li {
+
+font: 13px 'Lucida Grande', Arial, sans-serif;
+
+}
+</style>
+
+<style type="text/css">
+
+h1 {
+
+font: 13px 'Lucida Grande', Arial, sans-serif;
+
+}
+</style>"""
+    eq_ ('Hi', u.html_to_text(html))
+
+    html = """<div>
+<!-- COMMENT 1 -->
+<span>TEXT 1</span>
+<p>TEXT 2 <!-- COMMENT 2 --></p>
+</div>"""
+    eq_('TEXT 1 \nTEXT 2', u.html_to_text(html))