Skip to content

fix: stabilize test_text_query_word_weights with structural assertions #541

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
thakoreh wants to merge 3 commits into
base: main
Choose a base branch
from
+38 −5
Open

fix: stabilize test_text_query_word_weights with structural assertions #541

Changes from 2 commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
8ee962b
fix: stabilize test_text_query_word_weights with structural assertions
thakoreh Mar 25, 2026
517c61f
fix: strengthen test assertions per Copilot review
thakoreh Mar 25, 2026
9680523
fix: define missing suffix variable in test
thakoreh Mar 25, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
42 changes: 37 additions & 5 deletions tests/unit/test_query_types.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -335,7 +335,6 @@ def test_text_query_with_string_filter():
assert "AND" not in query_string_wildcard


@pytest.mark.skip("Test is flaking")
def test_text_query_word_weights():
# verify word weights get added into the raw Redis query syntax
query = TextQuery(
Expand All @@ -344,10 +343,43 @@ def test_text_query_word_weights():
text_weights={"alpha": 2, "delta": 0.555, "gamma": 0.95},
)

assert (
str(query)
== "@description:(query | string | alpha=>{$weight:2} | bravo | delta=>{$weight:0.555} | tango | alpha=>{$weight:2}) SCORER BM25STD WITHSCORES DIALECT 2 LIMIT 0 10"
)
# Check query components with structural guarantees,
# not exact token ordering (which is non-deterministic).
Comment on lines +346 to +347
Copy link

Copilot AI Mar 25, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The comment about token ordering being non-deterministic is likely misleading here: TextQuery._tokenize_and_escape_query() preserves the input token order and only does in-place substitutions for weighted tokens. If the goal is to avoid brittleness, consider rewording this to the actual nondeterminism/flakiness source (or just say the test avoids exact full-string matching).

Suggested change
# Check query components with structural guarantees,
# not exact token ordering (which is non-deterministic).
# Check query components with structural guarantees rather than relying
# on exact full-string or token-order matching, to avoid brittle tests.

Copilot uses AI. Check for mistakes.
query_str = str(query)

# Description clause is properly delimited
assert "@description:(" in query_str
desc_start = query_str.index("@description:(")
desc_close = query_str.index(")", desc_start)
desc_clause = query_str[desc_start + len("@description:(") : desc_close]

# Weighted terms appear inside the description clause
assert "delta=>{$weight:0.555}" in desc_clause

# alpha appears twice and both occurrences are weighted
alpha_weighted = "alpha=>{$weight:2}"
assert desc_clause.count(alpha_weighted) == 2
# Ensure no unweighted 'alpha' tokens slipped through
idx = 0
Comment on lines +359 to +363
Copy link

Copilot AI Mar 25, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

desc_clause.count(alpha_weighted) and the subsequent find("alpha") loop rely on substring matching, which can behave unexpectedly if token escaping changes or a different token contains alpha as a substring. Since tokens are joined with " | ", consider splitting desc_clause into tokens (strip whitespace) and asserting exact token equality + frequency via a list/Counter.

Copilot uses AI. Check for mistakes.
while True:
idx = desc_clause.find("alpha", idx)
if idx == -1:
break
assert desc_clause.startswith(alpha_weighted, idx)
idx += len("alpha")

# Unweighted terms are present
for term in ["query", "string", "bravo", "tango"]:
assert term in desc_clause
Comment on lines +371 to +373
Copy link

Copilot AI Mar 25, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

assert term in desc_clause checks substrings, not whole tokens, so it may pass even if the tokenization/output changes in a way that shouldn’t be acceptable (or fail for escaped tokens). Consider tokenizing desc_clause (split on | and strip) and asserting the expected token set is present.

Suggested change
# Unweighted terms are present
for term in ["query", "string", "bravo", "tango"]:
assert term in desc_clause
# Unweighted terms are present as standalone tokens within the clause
desc_tokens = {token.strip() for token in desc_clause.split("|") if token.strip()}
for term in ["query", "string", "bravo", "tango"]:
assert term in desc_tokens

Copilot uses AI. Check for mistakes.

# Post-query modifiers follow after the closing paren in expected order
suffix_stripped = suffix.lstrip()
assert suffix_stripped.startswith("SCORER BM25STD")
scorer_idx = suffix_stripped.index("SCORER BM25STD")
withscores_idx = suffix_stripped.index("WITHSCORES")
dialect_idx = suffix_stripped.index("DIALECT 2")
limit_idx = suffix_stripped.index("LIMIT 0 10")
assert scorer_idx <= withscores_idx <= dialect_idx <= limit_idx

# raise an error if weights are not positive floats
with pytest.raises(ValueError):
Expand Down