Add some example HTML for unit tests and test htmltreediff.

edgi-govdata-archiving · Oct 12, 2017 · 1ace639 · 1ace639
1 parent 81cd46f
commit 1ace639
Show file tree

Hide file tree

Showing 18 changed files with 83 additions and 0 deletions.
diff --git a/MANIFEST.in b/MANIFEST.in
@@ -1,3 +1,4 @@
 include versioneer.py
 include web_monitoring/_version.py
 include web_monitoring/tests/cassettes/*
+include web_monitoring/example_data/*
diff --git a/requirements.txt b/requirements.txt
@@ -2,6 +2,7 @@ beautifulsoup4
 python-dateutil
 diff_match_patch_python
 docopt
+git+https://github.com/danielballan/htmltreediff@py3-compat
 lxml
 pandas
 requests

diff --git a/setup.py b/setup.py
@@ -22,6 +22,7 @@ def read(fname):
       version=versioneer.get_version(),
       cmdclass=versioneer.get_cmdclass(),
       packages=['web_monitoring'],
+      package_data={'web_monitoring': ['example_data/*']},
       scripts=glob.glob('scripts/*'),
       install_requires=read('requirements.txt').splitlines(),
       package_data={'web_monitoring': ['web_monitoring/tests/cassettes/*']},

diff --git a/web_monitoring/example_data/add-paragraph.after b/web_monitoring/example_data/add-paragraph.after
@@ -0,0 +1,8 @@
+<html>
+  <head>
+  </head>
+  <body>
+    <p>One</p>
+    <p>Two</p>
+  </body>
+<html>
diff --git a/web_monitoring/example_data/add-paragraph.before b/web_monitoring/example_data/add-paragraph.before
@@ -0,0 +1,7 @@
+<html>
+  <head>
+  </head>
+  <body>
+    <p>One</p>
+  </body>
+<html>
diff --git a/web_monitoring/example_data/change-href.after b/web_monitoring/example_data/change-href.after
@@ -0,0 +1 @@
+<a href="/elsewhere">Link Text</a>
diff --git a/web_monitoring/example_data/change-href.before b/web_monitoring/example_data/change-href.before
@@ -0,0 +1 @@
+<a href="/somewhere">Link Text</a>
diff --git a/web_monitoring/example_data/change-link-text.after b/web_monitoring/example_data/change-link-text.after
@@ -0,0 +1 @@
+<a href="/somewhere">After</a>
diff --git a/web_monitoring/example_data/change-link-text.before b/web_monitoring/example_data/change-link-text.before
@@ -0,0 +1 @@
+<a href="/somewhere">Before</a>
diff --git a/web_monitoring/example_data/change-tag.after b/web_monitoring/example_data/change-tag.after
@@ -0,0 +1 @@
+<h2>whatever</h2>
diff --git a/web_monitoring/example_data/change-tag.before b/web_monitoring/example_data/change-tag.before
@@ -0,0 +1 @@
+<h1>whatever</h1>
diff --git a/web_monitoring/example_data/change-word-in-paragraph.after b/web_monitoring/example_data/change-word-in-paragraph.after
@@ -0,0 +1 @@
+<p>Beware the Jabberwock my son, the jaws that byte, the claws that catch.</p>
diff --git a/web_monitoring/example_data/change-word-in-paragraph.before b/web_monitoring/example_data/change-word-in-paragraph.before
@@ -0,0 +1 @@
+<p>Beware the Jabberwock my son, the jaws that bite, the claws that catch.</p>
diff --git a/web_monitoring/example_data/ins-in-source.after b/web_monitoring/example_data/ins-in-source.after
@@ -0,0 +1 @@
+<ins>after</ins>
diff --git a/web_monitoring/example_data/ins-in-source.before b/web_monitoring/example_data/ins-in-source.before
@@ -0,0 +1 @@
+<ins>before</ins>
diff --git a/web_monitoring/example_data/two-paragraphs.after b/web_monitoring/example_data/two-paragraphs.after
@@ -0,0 +1,8 @@
+<html>
+  <head>
+  </head>
+  <body>
+    <p>After 1</p>
+    <p>After 2</p>
+  </body>
+<html>
diff --git a/web_monitoring/example_data/two-paragraphs.before b/web_monitoring/example_data/two-paragraphs.before
@@ -0,0 +1,8 @@
+<html>
+  <head>
+  </head>
+  <body>
+    <p>Before 1</p>
+    <p>Before 2</p>
+  </body>
+<html>
diff --git a/web_monitoring/tests/test_html_diff.py b/web_monitoring/tests/test_html_diff.py
@@ -0,0 +1,39 @@
+from pkg_resources import resource_filename
+import pytest
+import htmltreediff
+
+
+def lookup_pair(fn):
+    """Read example data named {fn}.before and {fn}.after"""
+    fn1 = 'example_data/{}.before'.format(fn)
+    fn2 = 'example_data/{}.after'.format(fn)
+    with open(resource_filename('web_monitoring', fn1)) as f:
+        before = f.read()
+    with open(resource_filename('web_monitoring', fn2)) as f:
+        after = f.read()
+    return before, after
+
+
+@pytest.mark.parametrize('fn',
+                         ['change-tag',
+                          'change-href',
+                          'change-link-text',
+                          'ins-in-source',
+                          'two-paragraphs',
+                          'add-paragraph',
+                          'change-word-in-paragraph',
+                         ])
+def test_change(fn):
+    # For now it's unclear what the 'expected' result should be, so this
+    # test will never FAIL (but it can still ERROR). Run pytest with the -s
+    # flag to see these outputs.
+    before, after = lookup_pair(fn)
+    d = htmltreediff.diff(before, after)
+    print("""
+BEFORE:
+{}
+AFTER:
+{}
+DIFF:
+{}
+""".format(before, after, d))