collective/mediawiki_dump_tools
18
0
Fork 0
Code Issues Pull Requests Activity

finish support for paragraph move.

Browse Source
This commit is contained in:
Nathan TeBlunthuis
2025-07-01 11:16:08 -07:00
parent 20de5b93f9
commit c4acc711d2
5 changed files with 666 additions and 63 deletions
Download Patch File Download Diff File Expand all files Collapse all files

34
test/test_wiki_diff_matcher.py
View File

@@ -27,7 +27,11 @@ async def start_stop_server():
def assert_equal_enough(tokens:List[Token], rev):
# the tokens exclude newlines
# we allow extra whitespace at the beginning or end
assert ''.join([str(t) for t in tokens]).strip() == rev.replace('\n','').strip()
token_doc = ''.join(str(t) for t in tokens).strip()
rev_doc = rev.replace('\n','').strip()
print(token_doc, file = open('token','w'))
print(rev_doc, file = open('rev','w'))
assert token_doc == rev_doc
def assert_correct_equal_section(ops, expected_equal_lines, expected_equal_tokens):
@@ -128,8 +132,6 @@ def test_delete():
expected_equal_tokens=9323)
# first lets test that we properly build the operations.
# then we can test if the state seems to work as intended.
def test_addition():
@@ -170,3 +172,29 @@ def test_addition():
assert n_inserted_tokens == last_b2 - initial_equal_tokens == 292
assert n_inserts == 2
def test_paragraph_move():
rev1 = open("test/test_diff_revisions/1295229484").read()
rev2 = open("test/test_diff_revisions/1295229484_parmove").read()
matcher = WikiDiffMatcher([rev1,rev2])
diff_processor = matcher.processor()
# note that a and b are constructed from the diffs.
# so they reflect the state of the text according to the diff processor
ops, a, b = diff_processor.process(rev1)
ops, a, b = diff_processor.process(rev2)
assert_equal_enough(b, rev2)
assert_equal_enough(a, rev1)
def test_paragraph_move_and_change():
rev1 = open("test/test_diff_revisions/1295229484").read()
rev2 = open("test/test_diff_revisions/1295229484_parmove_and_change").read()
matcher = WikiDiffMatcher([rev1,rev2])
diff_processor = matcher.processor()
# note that a and b are constructed from the diffs.
# so they reflect the state of the text according to the diff processor
ops, a, b = diff_processor.process(rev1)
ops, a, b = diff_processor.process(rev2)
assert_equal_enough(b, rev2)
assert_equal_enough(a, rev1)