From 8e2bed5343cc5ac47485c94b09aa0b9d43d76d5c Mon Sep 17 00:00:00 2001
From: Christopher Szucko <cszucko@gmail.com>
Date: Sun, 11 Aug 2019 09:30:35 -0500
Subject: [PATCH 61/71] Add unit tests for HTML metadata imports
---
src/calibre/ebooks/metadata/html.py | 182 ++++++++++++++++++++++++++++
1 file changed, 182 insertions(+)
diff --git a/src/calibre/ebooks/metadata/html.py b/src/calibre/ebooks/metadata/html.py
index dad218ae19..c6d4693baf 100644
--- a/src/calibre/ebooks/metadata/html.py
+++ b/src/calibre/ebooks/metadata/html.py
@@ -9,6 +9,7 @@ Try to read metadata from an HTML file.
'''
import re
+import unittest
from collections import defaultdict
from HTMLParser import HTMLParser
@@ -251,3 +252,184 @@ def get_metadata_(src, encoding=None):
mi.set_identifier(k, v[0])
return mi
+
+
+class MetadataHtmlTest(unittest.TestCase):
+
+ def compare_metadata(self, meta_a, meta_b):
+ for attr in ('title', 'authors', 'publisher', 'isbn', 'languages', 'pubdate', 'timestamp', 'series', 'series_index', 'rating', 'comments', 'tags', 'identifiers'):
+ self.assertEqual(getattr(meta_a, attr), getattr(meta_b, attr))
+
+ def get_stream(self, test):
+ from io import BytesIO
+
+ raw = b'''\
+<html>
+ <head>
+'''
+
+ if test in {'title', 'meta_single', 'meta_multi', 'comment_single', 'comment_multi'}:
+ raw += b'''\
+ }
+ <title>A Title Tag &amp; Title Ⓒ</title>
+'''
+
+ if test in {'meta_single', 'meta_multi', 'comment_single', 'comment_multi'}:
+ raw += b'''\
+ <meta name="dc:title" content="A Meta Tag &amp; Title Ⓒ" />
+ <meta name="dcterms.creator.aut" content="George Washington" />
+ <meta name="dc.publisher" content="Publisher A" />
+ <meta name="isbn" content="1234567890" />
+ <meta name="dc.language" content="English" />
+ <meta name="dc.date.published" content="2019-01-01" />
+ <meta name="dcterms.created" content="2018-01-01" />
+ <meta name="series" content="Meta Series" />
+ <meta name="seriesnumber" content="1" />
+ <meta name="rating" content="" />
+ <meta name="dc.description" content="" />
+ <meta name="tags" content="tag a, tag b" />
+ <meta name="dc.identifier.url" content="" />
+ <meta name="dc.identifier" scheme="" content="invalid" />
+ <meta name="dc.identifier." content="still invalid" />
+ <meta name="dc.identifier.conflicting" scheme="schemes" content="are also invalid" />
+ <meta name="dc.identifier.custom.subid" content="invalid too" />
+'''
+
+ if test in {'meta_multi', 'comment_single', 'comment_multi'}:
+ raw += b'''\
+ <meta name="title" content="A Different Meta Tag &amp; Title Ⓒ" />
+ <meta name="author" content="John Adams with Thomas Jefferson" />
+ <meta name="publisher" content="Publisher B" />
+ <meta name="isbn" content="2345678901" />
+ <meta name="dcterms.language" content="Spanish" />
+ <meta name="date of publication" content="2017-01-01" />
+ <meta name="timestamp" content="2016-01-01" />
+ <meta name="series" content="Another Meta Series" />
+ <meta name="series.index" content="2" />
+ <meta name="rating" content="8" />
+ <meta name="comments" content="meta "comments" ♥ HTML &amp;" />
+ <meta name="tags" content="tag c" />
+ <meta name="dc.identifier.url" content="http://google.com/search?q=calibre" />
+'''
+
+ if test in {'comment_single', 'comment_multi'}:
+ raw += b'''\
+ <!-- TITLE="A Comment Tag &amp; Title Ⓒ" -->
+ <!-- AUTHOR="James Madison and James Monroe" -->
+ <!-- PUBLISHER="Publisher C" -->
+ <!-- ISBN="3456789012" -->
+ <!-- LANGUAGE="French" -->
+ <!-- PUBDATE="2015-01-01" -->
+ <!-- TIMESTAMP="2014-01-01" -->
+ <!-- SERIES="Comment Series" -->
+ <!-- SERIESNUMBER="3" -->
+ <!-- RATING="20" -->
+ <!-- COMMENTS="comment "comments" ♥ HTML too &amp;" -->
+ <!-- TAGS="tag d" -->
+'''
+
+ if test in {'comment_multi'}:
+ raw += b'''\
+ <!-- TITLE="Another Comment Tag &amp; Title Ⓒ" -->
+ <!-- AUTHOR="John Quincy Adams" -->
+ <!-- PUBLISHER="Publisher D" -->
+ <!-- ISBN="4567890123" -->
+ <!-- LANGUAGE="Japanese" -->
+ <!-- PUBDATE="2013-01-01" -->
+ <!-- TIMESTAMP="2012-01-01" -->
+ <!-- SERIES="Comment Series 2" -->
+ <!-- SERIESNUMBER="4" -->
+ <!-- RATING="1" -->
+ <!-- COMMENTS="comment "comments" ♥ HTML too &amp; for sure" -->
+ <!-- TAGS="tag e, tag f" -->
+'''
+
+ raw += b'''\
+ </head>
+ <body>
+ </body>
+</html>
+'''
+ return BytesIO(raw)
+
+
+ def test_input_title(self):
+ stream_meta = get_metadata(self.get_stream('title'))
+ canon_meta = Metadata('A Title Tag & Title Ⓒ', [_('Unknown')])
+ self.compare_metadata(stream_meta, canon_meta)
+
+
+ def test_input_meta_single(self):
+ stream_meta = get_metadata(self.get_stream('meta_single'))
+ canon_meta = Metadata('A Meta Tag & Title Ⓒ', ['George Washington'])
+ canon_meta.publisher = 'Publisher A'
+ canon_meta.languages = ['English']
+ canon_meta.pubdate = parse_date('2019-01-01')
+ canon_meta.timestamp = parse_date('2018-01-01')
+ canon_meta.series = 'Meta Series'
+ canon_meta.series_index = float(1)
+ # canon_meta.rating = float(0)
+ # canon_meta.comments = ''
+ canon_meta.tags = ['tag a', 'tag b']
+ canon_meta.set_identifiers({'isbn': '1234567890'})
+ self.compare_metadata(stream_meta, canon_meta)
+
+
+ def test_input_meta_multi(self):
+ stream_meta = get_metadata(self.get_stream('meta_multi'))
+ canon_meta = Metadata('A Meta Tag & Title Ⓒ', ['George Washington', 'John Adams', 'Thomas Jefferson'])
+ canon_meta.publisher = 'Publisher A'
+ canon_meta.languages = ['English', 'Spanish']
+ canon_meta.pubdate = parse_date('2019-01-01')
+ canon_meta.timestamp = parse_date('2018-01-01')
+ canon_meta.series = 'Meta Series'
+ canon_meta.series_index = float(1)
+ canon_meta.rating = float(8)
+ canon_meta.comments = 'meta "comments" ♥ HTML &amp;'
+ canon_meta.tags = ['tag a', 'tag b', 'tag c']
+ canon_meta.set_identifiers({'isbn': '1234567890', 'url': 'http://google.com/search?q=calibre'})
+ self.compare_metadata(stream_meta, canon_meta)
+
+
+ def test_input_comment_single(self):
+ stream_meta = get_metadata(self.get_stream('comment_single'))
+ canon_meta = Metadata('A Comment Tag & Title Ⓒ', ['James Madison', 'James Monroe'])
+ canon_meta.publisher = 'Publisher C'
+ canon_meta.languages = ['French']
+ canon_meta.pubdate = parse_date('2015-01-01')
+ canon_meta.timestamp = parse_date('2014-01-01')
+ canon_meta.series = 'Comment Series'
+ canon_meta.series_index = float(3)
+ canon_meta.rating = float(0)
+ canon_meta.comments = 'comment "comments" ♥ HTML too &amp;'
+ canon_meta.tags = ['tag d']
+ canon_meta.set_identifiers({'isbn': '3456789012', 'url': 'http://google.com/search?q=calibre'})
+ self.compare_metadata(stream_meta, canon_meta)
+
+
+ def test_input_comment_multi(self):
+ stream_meta = get_metadata(self.get_stream('comment_multi'))
+ canon_meta = Metadata('A Comment Tag & Title Ⓒ', ['James Madison', 'James Monroe', 'John Quincy Adams'])
+ canon_meta.publisher = 'Publisher C'
+ canon_meta.languages = ['French', 'Japanese']
+ canon_meta.pubdate = parse_date('2015-01-01')
+ canon_meta.timestamp = parse_date('2014-01-01')
+ canon_meta.series = 'Comment Series'
+ canon_meta.series_index = float(3)
+ canon_meta.rating = float(0)
+ canon_meta.comments = 'comment "comments" ♥ HTML too &amp;'
+ canon_meta.tags = ['tag d', 'tag e', 'tag f']
+ canon_meta.set_identifiers({'isbn': '3456789012', 'url': 'http://google.com/search?q=calibre'})
+ self.compare_metadata(stream_meta, canon_meta)
+
+
+def suite():
+ return unittest.TestLoader().loadTestsFromTestCase(MetadataHtmlTest)
+
+
+def test():
+ unittest.TextTestRunner(verbosity=2).run(suite())
+
+
+if __name__ == '__main__':
+ test()