Add multiple new search types to AzureCosmosDBNoSqlVectorSearch

[allenkim0129]

Description

This PR significantly extends llama-index-vector-stores-azurecosmosnosql with multiple new search types, a fully parameterised query engine, comprehensive test coverage, and an updated README.

What changed

base.py — new search engine

• Replaced the old _query method (single hard-coded vector search) with a new _search_query method that supports 6 search types via a search_type kwarg
• All query components (TOP/OFFSET-LIMIT, SELECT, WHERE, ORDER BY) are built from composable helpers using a ParamMapping helper to produce fully parameterised CosmosDB SQL queries (no string interpolation of user values)
• Backward-compatible: the legacy pre_filter={"where_clause": ..., "limit_offset_clause": ...} dict is still accepted and mapped to the new where / offset_limit kwargs
• Added _generate_projection_fields, _generate_limit_clause, _generate_order_by_clause, _generate_order_by_component_with_full_text_rank_filter, _construct_search_query, _execute_search_query, _validate_search_args, _is_vector_search_type, _is_full_text_search_type, _is_vector_search_with_threshold helpers

utils.py — new helper module (extracted from base.py)

• Constants — all field/parameter name literals in one place
• ParamMapping — builds @param-style parameterised queries; handles projections, vector distance, inline vector literals for ORDER BY RANK contexts
• AzureCosmosDBNoSqlVectorSearchType — enum for all supported search types

Search types added

search_type	Description
vector	Standard nearest-neighbour via VectorDistance, ORDER BY
vector_score_threshold	Vector search filtered by a minimum similarity score
full_text_search	FullTextContains WHERE predicate, no ranking
full_text_ranking	ORDER BY RANK FullTextScore(...), single or multi-field RRF
hybrid	ORDER BY RANK RRF(FullTextScore, VectorDistance)
hybrid_score_threshold	Hybrid RRF + client-side score threshold filter
weighted_hybrid_search	Hybrid RRF + client-side per-component weight re-ranking (weight= is not supported in CosmosDB SQL syntax, so weights are applied after results are returned)

Query options added

• where — raw CosmosDB SQL WHERE predicate
• offset_limit — OFFSET x LIMIT y pagination (replaces TOP for hybrid types)
• projection_mapping — project specific document fields as node metadata
• return_with_vectors — include raw embeddings in results
• full_text_rank_filter — list of {search_field, search_text} dicts for full-text ranking components
• threshold — minimum similarity score for *_score_threshold types
• weights — per-component float list for weighted_hybrid_search (client-side RRF re-ranking)

Tests

• test_unit.py — 93 unit tests covering all helpers without any DB access:
  • TestParamMapping, TestGenerateProjectionFields, TestGenerateLimitClause, TestGenerateOrderByComponentWithFullTextRankFilter, TestGenerateOrderByClause, TestConstructSearchQuery, TestIsHelpers, TestSearchQueryPreFilterCompat
• test_azurecosmosnosql.py — 15 integration tests against a live CosmosDB account:
  • test_add_and_delete, test_query, test_query_vector, test_query_vector_with_k, test_query_vector_with_projection_mapping, test_query_vector_with_offset_limit, test_query_vector_with_where_filter, test_query_full_text_search, test_query_full_text_ranking, test_query_full_text_ranking_multiple_fields, test_query_hybrid_with_score_threshold, test_query_hybrid_with_weights, test_query_weighted_hybrid_search, test_cosmos_client_with_host_and_key, test_cosmos_client_with_connection_string

README

• Rewrote from scratch to cover: installation, quick start, all 7 search types with code examples, all query options, full text search container setup, weighted hybrid search with weight notation

---

New Package?

• Yes
• No

Version Bump?

Did I bump the version in the pyproject.toml?

• Yes
• No

Note for reviewer: the changes are additive and backward-compatible (old pre_filter callers are unaffected). Please advise if a minor or patch bump is required.

Type of Change

• New feature (non-breaking change which adds functionality)
• This change requires a documentation update

How Has This Been Tested?

• I added new unit tests to cover this change

Unit tests (no DB): pytest tests/test_unit.py → 93 passed

Integration tests (live CosmosDB):

export AZURE_COSMOSDB_URI="https://<account>.documents.azure.com:443/"
export AZURE_COSMOSDB_KEY="<key>"
pytest tests/test_azurecosmosnosql.py

→ 15 passed

Suggested Checklist

• I have performed a self-review of my own code
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• I have added Google Colab support for the newly added notebooks
• My changes generate no new warnings
• I have added tests that prove my fix is effective or that my feature works
• New and existing unit tests pass locally with my changes
• I ran uv run make format; uv run make lint to appease the lint gods