| 1 | """Quoted proper-noun phrases must survive into the provider query. |
| 2 | |
| 3 | Two compounding defects meant a phrase search never actually happened: |
| 4 | |
| 5 | 1. planner._keyword_query emitted the quoted compound *and* a core that |
| 6 | already contained the same words: '"Peter Steinberger" peter steinberger |
| 7 | steipete'. |
| 8 | 2. bird_x stripped the quotes before building the X query, degrading an |
| 9 | intended phrase match into a bare token conjunction |
| 10 | (peter AND steinberger AND steipete). |
| 11 | |
| 12 | On the measured 'Peter Steinberger steipete' baseline the primary X lane |
| 13 | returned nothing as a result, and the whole run depended on the supplement |
| 14 | lanes. |
| 15 | """ |
| 16 | |
| 17 | from lib import bird_x, planner |
| 18 | |
| 19 | |
| 20 | def test_quoted_phrase_words_are_not_repeated_unquoted(): |
| 21 | q = planner._keyword_query("Peter Steinberger steipete", "peter steinberger steipete") |
| 22 | assert q.lower().count("steinberger") == 1, ( |
| 23 | f"'steinberger' appears more than once in {q!r}; the quoted compound " |
| 24 | "is being emitted alongside a core that already contains it" |
| 25 | ) |
| 26 | assert '"Peter Steinberger"' in q, "the proper-noun phrase must stay quoted" |
| 27 | |
| 28 | |
| 29 | def test_distinct_topic_tokens_survive_alongside_the_phrase(): |
| 30 | q = planner._keyword_query("Peter Steinberger steipete", "peter steinberger steipete") |
| 31 | assert "steipete" in q.lower(), ( |
| 32 | f"deduplicating the phrase must not drop unrelated topic tokens: {q!r}" |
| 33 | ) |
| 34 | |
| 35 | |
| 36 | def test_single_word_topic_is_unchanged(): |
| 37 | q = planner._keyword_query("bentgo", "bentgo") |
| 38 | assert q.strip() == "bentgo" |
| 39 | |
| 40 | |
| 41 | def test_topic_with_no_title_cased_compound_is_unquoted(): |
| 42 | q = planner._keyword_query("open source llm tooling", "open source llm tooling") |
| 43 | assert '"' not in q |
| 44 | |
| 45 | |
| 46 | def test_provider_query_preserves_quoted_phrase(): |
| 47 | """bird_x must not strip quotes out of the phrase before querying X. |
| 48 | |
| 49 | X advanced search supports quoted phrases natively, so passing them |
| 50 | through is strictly better retrieval than a token conjunction. |
| 51 | """ |
| 52 | built = bird_x.build_topic_query('"Peter Steinberger" steipete', "2026-07-14") |
| 53 | assert '"Peter Steinberger"' in built, ( |
| 54 | f"quoted phrase was stripped before reaching X: {built!r}" |
| 55 | ) |
| 56 | assert "since:2026-07-14" in built |
| 57 | |
| 58 | |
| 59 | def test_provider_query_keeps_bare_tokens(): |
| 60 | built = bird_x.build_topic_query("bentgo lunch", "2026-07-14") |
| 61 | assert "bentgo" in built and "lunch" in built |
| 62 | assert '"' not in built |
| 63 | |
| 64 | |
| 65 | def test_provider_query_drops_grouping_syntax_but_not_phrases(): |
| 66 | """Bird grouping characters are still noise; phrase quotes are not.""" |
| 67 | built = bird_x.build_topic_query('("Claude Code") review', "2026-07-14") |
| 68 | assert '"Claude Code"' in built |
| 69 | assert "(" not in built and ")" not in built |
| 70 | |
| 71 | |
| 72 | def test_unbalanced_quote_falls_back_to_bare_tokens(): |
| 73 | """An orphan quote must never reach X. |
| 74 | |
| 75 | Upstream trimming (core-subject extraction, the retry ladder shortening a |
| 76 | query) can cut a topic mid-phrase, leaving 'berlin "mixed-use'. X reads the |
| 77 | orphan as an unterminated phrase and matches nothing, so a query that |
| 78 | cannot be balanced is safer unquoted. |
| 79 | """ |
| 80 | built = bird_x.build_topic_query('immobilienmakler berlin "mixed-use', "2026-07-12") |
| 81 | assert '"' not in built, f"unbalanced quote survived into the query: {built!r}" |
| 82 | assert "immobilienmakler" in built and "mixed-use" in built |
| 83 | |
| 84 | |
| 85 | def test_balanced_quotes_are_still_preserved(): |
| 86 | built = bird_x.build_topic_query('"Claude Code" "Peter Steinberger"', "2026-07-14") |
| 87 | assert built.count('"') == 4 |
| 88 |