def test_slice_ql_on_matches_and_chunks():
docs = random_docs_with_chunks_and_matches(10)
driver = SliceQL(start=0, end=2, traversal_paths=('cc', 'c', 'r', 'mm', 'm'))
assert len(docs) == 10
assert len(docs[0].chunks) == 10
assert len(docs[-1].chunks) == 10
assert len(docs[0].matches) == 10
assert len(docs[-1].matches) == 10
assert len(docs[0].matches[0].chunks) == 10
assert len(docs[0].matches[-1].chunks) == 10
assert len(docs[-1].matches[0].chunks) == 10
assert len(docs[-1].matches[-1].chunks) == 10
assert len(docs[0].chunks[0].chunks) == 10
assert len(docs[0].chunks[0].matches) == 10
assert len(docs[0].chunks[0].matches[0].chunks) == 10
assert len(docs[0].chunks[0].matches[-1].chunks) == 10
assert len(docs[0].chunks[-1].matches[0].chunks) == 10
assert len(docs[0].chunks[-1].matches[-1].chunks) == 10
assert len(docs[0].chunks[-1].chunks) == 10
assert len(docs[0].chunks[-1].matches) == 10
assert len(docs[-1].chunks[0].chunks) == 10
assert len(docs[-1].chunks[0].matches) == 10
assert len(docs[-1].chunks[-1].chunks) == 10
assert len(docs[-1].chunks[-1].matches) == 10
driver._traverse_apply(docs)
assert len(docs) == 2
assert len(docs[0].chunks) == 2 # slice on level 1
assert len(docs[0].matches) == 2 # slice on level 1
assert len(docs[0].chunks[0].chunks) == 2 # slice on level 2 for chunks
assert len(docs[0].chunks[-1].chunks) == 2 # slice on level 2 for chunks
assert len(docs[0].chunks[0].matches) == 10 # traverses directly on matches
assert len(docs[0].chunks[0].matches[0].chunks) == 10
assert len(docs[0].chunks[0].matches[-1].chunks) == 10
assert len(docs[0].chunks[-1].matches) == 10 # traverses directly on matches
assert len(docs[0].chunks[-1].matches[0].chunks) == 10
assert len(docs[0].chunks[-1].matches[-1].chunks) == 10
assert len(docs[0].matches[0].chunks) == 10
assert len(docs[0].matches[-1].chunks) == 10
assert len(docs[-1].chunks) == 2 # slice on level 1 of chunks
assert len(docs[-1].matches) == 2 # slice on level 1 of chunks
assert len(docs[-1].chunks[0].chunks) == 2 # slice on level 2 for matches of chunks
assert len(docs[-1].chunks[-1].chunks) == 2 # slice on level 2 for matches of chunks
assert len(docs[-1].chunks[0].matches) == 10 # traverses directly on matches
assert len(docs[-1].chunks[0].matches[0].chunks) == 10
assert len(docs[-1].chunks[0].matches[-1].chunks) == 10
assert len(docs[-1].chunks[-1].matches) == 10 # traverses directly on matches
assert len(docs[-1].chunks[-1].matches[0].chunks) == 10
assert len(docs[-1].chunks[-1].matches[-1].chunks) == 10
assert len(docs[-1].matches[0].chunks) == 10
assert len(docs[-1].matches[-1].chunks) == 10
def test_slice_ql_on_matches_and_chunks():
docs = random_docs_with_chunks_and_matches(10)
driver = SliceQL(start=0, end=2, granularity_range=(0, 2), adjacency_range=(0, 2))
assert len(docs) == 10
assert len(docs[0].chunks) == 10
assert len(docs[-1].chunks) == 10
assert len(docs[0].matches) == 10
assert len(docs[-1].matches) == 10
assert len(docs[0].matches[0].chunks) == 10
assert len(docs[0].matches[-1].chunks) == 10
assert len(docs[-1].matches[0].chunks) == 10
assert len(docs[-1].matches[-1].chunks) == 10
assert len(docs[0].chunks[0].chunks) == 10
assert len(docs[0].chunks[0].matches) == 10
assert len(docs[0].chunks[0].matches[0].chunks) == 10
assert len(docs[0].chunks[0].matches[-1].chunks) == 10
assert len(docs[0].chunks[-1].matches[0].chunks) == 10
assert len(docs[0].chunks[-1].matches[-1].chunks) == 10
assert len(docs[0].chunks[-1].chunks) == 10
assert len(docs[0].chunks[-1].matches) == 10
assert len(docs[-1].chunks[0].chunks) == 10
assert len(docs[-1].chunks[0].matches) == 10
assert len(docs[-1].chunks[-1].chunks) == 10
assert len(docs[-1].chunks[-1].matches) == 10
driver._traverse_apply(docs)
assert len(docs) == 2
assert len(docs[0].chunks) == 2 # slice on level 1
assert len(docs[0].matches) == 2 # slice on level 1
assert len(docs[0].chunks[0].chunks) == 2 # slice on level 2 for chunks
assert len(docs[0].chunks[-1].chunks) == 2 # slice on level 2 for chunks
assert len(docs[0].chunks[0].matches) == 10 # traverses directly on matches
assert len(docs[0].chunks[0].matches[0].chunks) == 10
assert len(docs[0].chunks[0].matches[-1].chunks) == 10
assert len(docs[0].chunks[-1].matches) == 10 # traverses directly on matches
assert len(docs[0].chunks[-1].matches[0].chunks) == 10
assert len(docs[0].chunks[-1].matches[-1].chunks) == 10
assert len(docs[0].matches[0].chunks) == 10
assert len(docs[0].matches[-1].chunks) == 10
assert len(docs[-1].chunks) == 2 # slice on level 1 of chunks
assert len(docs[-1].matches) == 2 # slice on level 1 of chunks
assert len(docs[-1].chunks[0].chunks) == 2 # slice on level 2 for matches of chunks
assert len(docs[-1].chunks[-1].chunks) == 2 # slice on level 2 for matches of chunks
assert len(docs[-1].chunks[0].matches) == 10 # traverses directly on matches
assert len(docs[-1].chunks[0].matches[0].chunks) == 10
assert len(docs[-1].chunks[0].matches[-1].chunks) == 10
assert len(docs[-1].chunks[-1].matches) == 10 # traverses directly on matches
assert len(docs[-1].chunks[-1].matches[0].chunks) == 10
assert len(docs[-1].chunks[-1].matches[-1].chunks) == 10
assert len(docs[-1].matches[0].chunks) == 10
assert len(docs[-1].matches[-1].chunks) == 10
def test_selection():
docs = build_docs()
driver = SliceQL(start=0, end=1, traversal_paths=['cmm', 'mcm'])
driver._traverse_apply(docs)
assert docs[0].chunks[0].matches[0].matches[0].granularity == 1
assert docs[0].chunks[0].matches[0].matches[0].adjacency == 2
assert len(docs[0].chunks[0].matches[0].matches) == 1
assert docs[0].matches[0].chunks[0].matches[0].granularity == 1
assert docs[0].matches[0].chunks[0].matches[0].adjacency == 2
assert len(docs[0].matches[0].chunks[0].matches) == 1
def test_match_chunk():
docs = build_docs()
driver = SliceQL(start=0, end=1, traversal_paths=['mc'])
driver._traverse_apply(docs)
assert len(docs) == DOCUMENTS_PER_LEVEL
assert len(docs[0].chunks) == DOCUMENTS_PER_LEVEL
assert len(docs[0].chunks[0].matches) == DOCUMENTS_PER_LEVEL
assert len(docs[0].matches) == DOCUMENTS_PER_LEVEL
assert len(docs[0].matches[0].chunks) == 1
assert len(docs[0].matches[0].matches) == DOCUMENTS_PER_LEVEL
assert len(docs[0].matches[0].matches[0].chunks) == DOCUMENTS_PER_LEVEL
def test_slice_ql_on_chunks():
docs = random_docs_with_chunks(10)
driver = SliceQL(start=0, end=2, recur_on='chunks', recur_range=(0, 2))
driver._traverse_apply(docs)
assert len(docs) == 2
assert len(docs[0].chunks) == 2 # slice on level 1
assert len(docs[0].chunks[0].chunks) == 2 # slice on level 2 for chunks
assert len(docs[0].chunks[-1].chunks) == 2 # slice on level 2 for chunks
assert len(docs[-1].chunks) == 2 # slice on level 1
assert len(docs[-1].chunks[0].chunks) == 2 # slice on level 2 for chunks
assert len(docs[-1].chunks[-1].chunks) == 2 # slice on level 2 for chunks
def test_slice_ql_on_chunks():
docs = random_docs_with_chunks(10)
driver = SliceQL(start=0, end=2, traversal_paths=('cc', 'c', 'r'))
driver._traverse_apply(docs)
assert len(docs) == 2
assert len(docs[0].chunks) == 2 # slice on level 1
assert len(docs[0].chunks[0].chunks) == 2 # slice on level 2 for chunks
assert len(docs[0].chunks[-1].chunks) == 2 # slice on level 2 for chunks
assert len(docs[-1].chunks) == 2 # slice on level 1
assert len(docs[-1].chunks[0].chunks) == 2 # slice on level 2 for chunks
assert len(docs[-1].chunks[-1].chunks) == 2 # slice on level 2 for chunks
def test_traverse_apply():
docs = build_docs()
doc = docs[0]
doc.ClearField('chunks')
docs = [
doc,
]
driver = SliceQL(start=0, end=1, traversal_paths=['mcm'])
assert docs[0].matches[0].chunks[0].matches[0].granularity == 1
assert docs[0].matches[0].chunks[0].matches[0].adjacency == 2
driver._traverse_apply(docs)
assert len(docs[0].matches[0].chunks[0].matches) == 1
def test_both_from_0():
docs = build_docs()
driver = SliceQL(start=0,
end=1,
traversal_paths=['r', 'c', 'm', 'cc', 'mm'])
driver._traverse_apply(docs)
assert len(docs) == 1
assert len(docs[0].chunks) == 1
assert len(docs[0].chunks[0].chunks) == 1
assert len(docs[0].chunks[0].chunks[0].matches) == DOCUMENTS_PER_LEVEL
assert len(docs[0].chunks[0].matches) == DOCUMENTS_PER_LEVEL
assert len(docs[0].matches) == 1
assert len(docs[0].matches[0].chunks) == DOCUMENTS_PER_LEVEL
assert len(docs[0].matches[0].matches) == 1
assert len(docs[0].matches[0].matches[0].chunks) == DOCUMENTS_PER_LEVEL
def test_only_adjacency():
docs = build_docs()
driver = SliceQL(start=0,
end=1,
adjacency_range=(0, 2),
granularity_range=(0, 0),
recur_on=["matches"])
driver._traverse_apply(docs)
assert len(docs) == 1
assert len(docs[0].chunks) == DOCUMENTS_PER_LEVEL
assert len(docs[0].chunks[0].matches) == DOCUMENTS_PER_LEVEL
assert len(docs[0].matches) == 1
assert len(docs[0].matches[0].chunks) == DOCUMENTS_PER_LEVEL
assert len(docs[0].matches[0].matches) == 1
assert len(docs[0].matches[0].matches[0].chunks) == DOCUMENTS_PER_LEVEL
def test_adjacency0_granularity1():
docs = build_docs()
driver = SliceQL(start=0, end=1, traversal_paths=['c', 'cc', 'cm', 'cmm'])
driver._traverse_apply(docs)
assert len(docs) == DOCUMENTS_PER_LEVEL
assert len(docs[0].chunks) == 1
assert len(docs[0].chunks[0].chunks) == 1
assert len(docs[0].chunks[0].chunks[0].matches) == DOCUMENTS_PER_LEVEL
assert len(docs[0].chunks[0].matches) == 1
assert len(docs[0].chunks[0].matches[0].chunks) == DOCUMENTS_PER_LEVEL
assert len(docs[0].chunks[0].matches[0].matches) == 1
assert len(
docs[0].chunks[0].matches[0].matches[0].chunks) == DOCUMENTS_PER_LEVEL
assert len(docs[0].matches) == DOCUMENTS_PER_LEVEL
assert len(docs[0].matches[0].chunks) == DOCUMENTS_PER_LEVEL
assert len(docs[0].matches[0].matches) == DOCUMENTS_PER_LEVEL
assert len(docs[0].matches[0].matches[0].chunks) == DOCUMENTS_PER_LEVEL
def test_slice_ql_on_matches(self):
docs = random_docs_with_matches(10)
driver = SliceQL(start=0,
end=2,
traverse_on='matches',
depth_range=(0, 2))
driver._traverse_apply(docs)
assert len(docs) == 10 # traverses directly on matches
assert len(docs[0].matches) == 2 # slice on level 1
assert len(
docs[0].matches[0].matches) == 2 # slice on level 2 for matches
assert len(
docs[0].matches[-1].matches) == 2 # slice on level 2 for matches
assert len(docs[-1].matches) == 2 # slice on level 1
assert len(
docs[-1].matches[0].matches) == 2 # slice on level 2 for matches
assert len(
docs[-1].matches[-1].matches) == 2 # slice on level 2 for matches
def test_traverse_apply():
docs = build_docs()
doc = docs[0]
doc.ClearField('chunks')
docs = [doc, ]
driver = SliceQL(
start=0,
end=1,
adjacency_range=(2, 2),
granularity_range=(1, 1),
recur_on=["matches", ]
)
# check we have a match with (g=1, a=2)
assert docs[0].matches[0].chunks[0].matches[0].granularity == 1
assert docs[0].matches[0].chunks[0].matches[0].adjacency == 2
# the following part will cause IndexError
driver._traverse_apply(docs)
assert len(docs[0].matches[0].chunks[0].matches) == 1
def test_selection():
docs = build_docs()
driver = SliceQL(
start=0,
end=1,
granularity_range=(1, 1),
adjacency_range=(2, 2),
recur_on=['matches', ]
)
driver._traverse_apply(docs)
# check the granularity and adjacency
assert docs[0].chunks[0].matches[0].matches[0].granularity == 1
assert docs[0].chunks[0].matches[0].matches[0].adjacency == 2
assert len(docs[0].chunks[0].matches[0].matches) == 1
# check the granularity and adjacency
assert docs[0].matches[0].chunks[0].matches[0].granularity == 1
assert docs[0].matches[0].chunks[0].matches[0].adjacency == 2
# For the chunks of the matches, the matches of these chunks are not effected by the apply()
assert len(docs[0].matches[0].chunks[0].matches) == 1
def test_adjacency_chunks():
"""
This combination of `range` and `recur_on` will only `_apply_all` to the root node.
Thus, the remaining document structure should still be intact.
In practice this combination should not be chosen, anyhow it nicely demonstrates the behaviour.
"""
docs = build_docs()
driver = SliceQL(start=0,
end=1,
adjacency_range=(0, 1),
granularity_range=(0, 0),
recur_on=["chunks"])
driver._traverse_apply(docs)
assert len(docs) == 1
assert len(docs[0].chunks) == DOCUMENTS_PER_LEVEL
assert len(docs[0].chunks[0].matches) == DOCUMENTS_PER_LEVEL
assert len(docs[0].matches) == DOCUMENTS_PER_LEVEL
assert len(docs[0].matches[0].chunks) == DOCUMENTS_PER_LEVEL
assert len(docs[0].matches[0].matches) == DOCUMENTS_PER_LEVEL
assert len(docs[0].matches[0].matches[0].chunks) == DOCUMENTS_PER_LEVEL
def test_slice_ql_on_matches_and_chunks(self):
docs = random_docs_with_chunks_and_matches(10)
driver = SliceQL(start=0,
end=2,
traverse_on=['chunks', 'matches'],
depth_range=(0, 2))
assert len(docs) == 10
assert len(docs[0].chunks) == 10
assert len(docs[-1].chunks) == 10
assert len(docs[0].matches) == 10
assert len(docs[-1].matches) == 10
assert len(docs[0].matches[0].chunks) == 10
assert len(docs[0].matches[-1].chunks) == 10
assert len(docs[-1].matches[0].chunks) == 10
assert len(docs[-1].matches[-1].chunks) == 10
assert len(docs[0].chunks[0].chunks) == 10
assert len(docs[0].chunks[0].matches) == 10
assert len(docs[0].chunks[0].matches[0].chunks) == 10
assert len(docs[0].chunks[0].matches[-1].chunks) == 10
assert len(docs[0].chunks[-1].matches[0].chunks) == 10
assert len(docs[0].chunks[-1].matches[-1].chunks) == 10
assert len(docs[0].chunks[-1].chunks) == 10
assert len(docs[0].chunks[-1].matches) == 10
assert len(docs[-1].chunks[0].chunks) == 10
assert len(docs[-1].chunks[0].matches) == 10
assert len(docs[-1].chunks[-1].chunks) == 10
assert len(docs[-1].chunks[-1].matches) == 10
driver._traverse_apply(docs)
assert len(docs) == 2
assert len(docs[0].chunks) == 2 # slice on level 1
assert len(docs[0].matches) == 2 # slice on level 1
assert len(
docs[0].chunks[0].chunks) == 2 # slice on level 2 for chunks
assert len(
docs[0].chunks[-1].chunks) == 2 # slice on level 2 for chunks
assert len(
docs[0].chunks[0].matches) == 10 # traverses directly on matches
assert len(docs[0].chunks[0].matches[0].chunks) == 10
assert len(docs[0].chunks[0].matches[-1].chunks) == 10
assert len(
docs[0].chunks[-1].matches) == 10 # traverses directly on matches
assert len(docs[0].chunks[-1].matches[0].chunks) == 10
assert len(docs[0].chunks[-1].matches[-1].chunks) == 10
assert len(docs[0].matches[0].chunks) == 10
assert len(docs[0].matches[-1].chunks) == 10
assert len(docs[-1].chunks) == 2 # slice on level 1 of chunks
assert len(docs[-1].matches) == 2 # slice on level 1 of chunks
assert len(docs[-1].chunks[0].chunks
) == 2 # slice on level 2 for matches of chunks
assert len(docs[-1].chunks[-1].chunks
) == 2 # slice on level 2 for matches of chunks
assert len(
docs[-1].chunks[0].matches) == 10 # traverses directly on matches
assert len(docs[-1].chunks[0].matches[0].chunks) == 10
assert len(docs[-1].chunks[0].matches[-1].chunks) == 10
assert len(
docs[-1].chunks[-1].matches) == 10 # traverses directly on matches
assert len(docs[-1].chunks[-1].matches[0].chunks) == 10
assert len(docs[-1].chunks[-1].matches[-1].chunks) == 10
assert len(docs[-1].matches[0].chunks) == 10
assert len(docs[-1].matches[-1].chunks) == 10
