mirror of
https://github.com/ArchiveBox/ArchiveBox.git
synced 2025-05-21 10:25:11 -04:00
Add tests for generic_rss and pinboard_rss parsers
This commit is contained in:
parent
22f9a289d3
commit
1f828d9441
3 changed files with 124 additions and 0 deletions
tests
|
@ -91,3 +91,71 @@ def test_extract_input_uses_only_passed_extractors(tmp_path, process):
|
|||
|
||||
assert (archived_item_path / "warc").exists()
|
||||
assert not (archived_item_path / "singlefile.html").exists()
|
||||
|
||||
def test_generic_rss(tmp_path, process, disable_extractors_dict):
|
||||
with open('../../mock_server/templates/example.rss', 'r', encoding='utf-8') as f:
|
||||
arg_process = subprocess.run(
|
||||
["archivebox", "add", "--index-only", "--parser=rss"],
|
||||
stdin=f,
|
||||
capture_output=True,
|
||||
env=disable_extractors_dict,
|
||||
)
|
||||
|
||||
conn = sqlite3.connect("index.sqlite3")
|
||||
c = conn.cursor()
|
||||
urls = c.execute("SELECT url from core_snapshot").fetchall()
|
||||
tags = c.execute("SELECT name from core_tag").fetchall()
|
||||
conn.commit()
|
||||
conn.close()
|
||||
|
||||
urls = list(map(lambda x: x[0], urls))
|
||||
assert "http://127.0.0.1:8080/static/example.com.html" in urls
|
||||
# if the following URL appears, we must have fallen back to another parser
|
||||
assert not "http://purl.org/dc/elements/1.1/" in urls
|
||||
|
||||
tags = list(map(lambda x: x[0], tags))
|
||||
assert "Tag1 Tag2" in tags
|
||||
|
||||
def test_pinboard_rss(tmp_path, process, disable_extractors_dict):
|
||||
with open('../../mock_server/templates/example.rss', 'r', encoding='utf-8') as f:
|
||||
arg_process = subprocess.run(
|
||||
["archivebox", "add", "--index-only", "--parser=pinboard_rss"],
|
||||
stdin=f,
|
||||
capture_output=True,
|
||||
env=disable_extractors_dict,
|
||||
)
|
||||
|
||||
conn = sqlite3.connect("index.sqlite3")
|
||||
c = conn.cursor()
|
||||
tags = c.execute("SELECT name from core_tag").fetchall()
|
||||
conn.commit()
|
||||
conn.close()
|
||||
|
||||
tags = list(map(lambda x: x[0], tags))
|
||||
assert "Tag1" in tags
|
||||
assert "Tag2" in tags
|
||||
|
||||
def test_atom(tmp_path, process, disable_extractors_dict):
|
||||
with open('../../mock_server/templates/example.atom', 'r', encoding='utf-8') as f:
|
||||
arg_process = subprocess.run(
|
||||
["archivebox", "add", "--index-only", "--parser=rss"],
|
||||
stdin=f,
|
||||
capture_output=True,
|
||||
env=disable_extractors_dict,
|
||||
)
|
||||
|
||||
conn = sqlite3.connect("index.sqlite3")
|
||||
c = conn.cursor()
|
||||
urls = c.execute("SELECT url from core_snapshot").fetchall()
|
||||
tags = c.execute("SELECT name from core_tag").fetchall()
|
||||
conn.commit()
|
||||
conn.close()
|
||||
|
||||
urls = list(map(lambda x: x[0], urls))
|
||||
assert "http://127.0.0.1:8080/static/example.com.html" in urls
|
||||
# if the following URL appears, we must have fallen back to another parser
|
||||
assert not "http://www.w3.org/2005/Atom" in urls
|
||||
|
||||
tags = list(map(lambda x: x[0], tags))
|
||||
assert "Tag1" in tags
|
||||
assert "Tag2" in tags
|
||||
|
|
Loading…
Add table
Add a link
Reference in a new issue