Karim shoair commited on
Commit ·
77fe3eb
1
Parent(s): ee99d12
Moving parser tests to a better structure
Browse files
tests/parser/__init__.py
ADDED
|
File without changes
|
tests/parser/test_automatch.py
ADDED
|
@@ -0,0 +1,56 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import unittest
|
| 2 |
+
|
| 3 |
+
from scrapling import Adaptor
|
| 4 |
+
|
| 5 |
+
|
| 6 |
+
class TestParserAutoMatch(unittest.TestCase):
|
| 7 |
+
|
| 8 |
+
def test_element_relocation(self):
|
| 9 |
+
"""Test relocating element after structure change"""
|
| 10 |
+
original_html = '''
|
| 11 |
+
<div class="container">
|
| 12 |
+
<section class="products">
|
| 13 |
+
<article class="product" id="p1">
|
| 14 |
+
<h3>Product 1</h3>
|
| 15 |
+
<p class="description">Description 1</p>
|
| 16 |
+
</article>
|
| 17 |
+
<article class="product" id="p2">
|
| 18 |
+
<h3>Product 2</h3>
|
| 19 |
+
<p class="description">Description 2</p>
|
| 20 |
+
</article>
|
| 21 |
+
</section>
|
| 22 |
+
</div>
|
| 23 |
+
'''
|
| 24 |
+
changed_html = '''
|
| 25 |
+
<div class="new-container">
|
| 26 |
+
<div class="product-wrapper">
|
| 27 |
+
<section class="products">
|
| 28 |
+
<article class="product new-class" data-id="p1">
|
| 29 |
+
<div class="product-info">
|
| 30 |
+
<h3>Product 1</h3>
|
| 31 |
+
<p class="new-description">Description 1</p>
|
| 32 |
+
</div>
|
| 33 |
+
</article>
|
| 34 |
+
<article class="product new-class" data-id="p2">
|
| 35 |
+
<div class="product-info">
|
| 36 |
+
<h3>Product 2</h3>
|
| 37 |
+
<p class="new-description">Description 2</p>
|
| 38 |
+
</div>
|
| 39 |
+
</article>
|
| 40 |
+
</section>
|
| 41 |
+
</div>
|
| 42 |
+
</div>
|
| 43 |
+
'''
|
| 44 |
+
|
| 45 |
+
old_page = Adaptor(original_html, url='example.com', auto_match=True, debug=True)
|
| 46 |
+
new_page = Adaptor(changed_html, url='example.com', auto_match=True, debug=True)
|
| 47 |
+
|
| 48 |
+
# 'p1' was used as ID and now it's not and all the path elements have changes
|
| 49 |
+
# Also at the same time testing auto-match vs combined selectors
|
| 50 |
+
_ = old_page.css('#p1, #p2', auto_save=True)[0]
|
| 51 |
+
relocated = new_page.css('#p1', auto_match=True)
|
| 52 |
+
|
| 53 |
+
self.assertIsNotNone(relocated)
|
| 54 |
+
self.assertEqual(relocated[0].attrib['data-id'], 'p1')
|
| 55 |
+
self.assertTrue(relocated[0].has_class('new-class'))
|
| 56 |
+
self.assertEqual(relocated[0].css('.new-description')[0].text, 'Description 1')
|
tests/{test_parser_functions.py → parser/test_general.py}
RENAMED
|
@@ -112,7 +112,7 @@ class TestParser(unittest.TestCase):
|
|
| 112 |
|
| 113 |
def test_find_similar_elements(self):
|
| 114 |
"""Test Finding similar elements of an element"""
|
| 115 |
-
first_product = self.page.
|
| 116 |
similar_products = first_product.find_similar()
|
| 117 |
self.assertEqual(len(similar_products), 2)
|
| 118 |
|
|
@@ -265,56 +265,6 @@ class TestParser(unittest.TestCase):
|
|
| 265 |
self.assertEqual(attr_json, {'jsonable': 'data'})
|
| 266 |
self.assertEqual(type(self.page.css('#products')[0].attrib.json_string), bytes)
|
| 267 |
|
| 268 |
-
def test_element_relocation(self):
|
| 269 |
-
"""Test relocating element after structure change"""
|
| 270 |
-
original_html = '''
|
| 271 |
-
<div class="container">
|
| 272 |
-
<section class="products">
|
| 273 |
-
<article class="product" id="p1">
|
| 274 |
-
<h3>Product 1</h3>
|
| 275 |
-
<p class="description">Description 1</p>
|
| 276 |
-
</article>
|
| 277 |
-
<article class="product" id="p2">
|
| 278 |
-
<h3>Product 2</h3>
|
| 279 |
-
<p class="description">Description 2</p>
|
| 280 |
-
</article>
|
| 281 |
-
</section>
|
| 282 |
-
</div>
|
| 283 |
-
'''
|
| 284 |
-
changed_html = '''
|
| 285 |
-
<div class="new-container">
|
| 286 |
-
<div class="product-wrapper">
|
| 287 |
-
<section class="products">
|
| 288 |
-
<article class="product new-class" data-id="p1">
|
| 289 |
-
<div class="product-info">
|
| 290 |
-
<h3>Product 1</h3>
|
| 291 |
-
<p class="new-description">Description 1</p>
|
| 292 |
-
</div>
|
| 293 |
-
</article>
|
| 294 |
-
<article class="product new-class" data-id="p2">
|
| 295 |
-
<div class="product-info">
|
| 296 |
-
<h3>Product 2</h3>
|
| 297 |
-
<p class="new-description">Description 2</p>
|
| 298 |
-
</div>
|
| 299 |
-
</article>
|
| 300 |
-
</section>
|
| 301 |
-
</div>
|
| 302 |
-
</div>
|
| 303 |
-
'''
|
| 304 |
-
|
| 305 |
-
old_page = Adaptor(original_html, url='example.com', auto_match=True, debug=True)
|
| 306 |
-
new_page = Adaptor(changed_html, url='example.com', auto_match=True, debug=True)
|
| 307 |
-
|
| 308 |
-
# 'p1' was used as ID and now it's not and all the path elements have changes
|
| 309 |
-
# Also at the same time testing auto-match vs combined selectors
|
| 310 |
-
_ = old_page.css('#p1, #p2', auto_save=True)[0]
|
| 311 |
-
relocated = new_page.css('#p1', auto_match=True)
|
| 312 |
-
|
| 313 |
-
self.assertIsNotNone(relocated)
|
| 314 |
-
self.assertEqual(relocated[0].attrib['data-id'], 'p1')
|
| 315 |
-
self.assertTrue(relocated[0].has_class('new-class'))
|
| 316 |
-
self.assertEqual(relocated[0].css('.new-description')[0].text, 'Description 1')
|
| 317 |
-
|
| 318 |
def test_performance(self):
|
| 319 |
"""Test parsing and selecting speed"""
|
| 320 |
import time
|
|
|
|
| 112 |
|
| 113 |
def test_find_similar_elements(self):
|
| 114 |
"""Test Finding similar elements of an element"""
|
| 115 |
+
first_product = self.page.css_first('.product')
|
| 116 |
similar_products = first_product.find_similar()
|
| 117 |
self.assertEqual(len(similar_products), 2)
|
| 118 |
|
|
|
|
| 265 |
self.assertEqual(attr_json, {'jsonable': 'data'})
|
| 266 |
self.assertEqual(type(self.page.css('#products')[0].attrib.json_string), bytes)
|
| 267 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 268 |
def test_performance(self):
|
| 269 |
"""Test parsing and selecting speed"""
|
| 270 |
import time
|