semql-prompt

LLM-facing prompt rendering for a semql Catalog.

semql's compiler is pure — it turns a SemanticQuery into SQL and never renders a prompt. This package is the rendering layer on top:

• Four-role prompt fragments — build_planner_prompt_fragment, build_router_prompt_fragment, build_presenter_prompt_fragment, build_drilldown_prompt_fragment, build_query_generator_prompt_fragment.
• Cacheable segments — CatalogPrompt (viewer-invariant static + per-viewer overlay) for prompt-cache breakpoints, plus prompt_hash.
• Tool-description projection — to_openai_tools / to_langchain_tools / to_openai_function for function-calling clients, and to_bedrock_converse_tools for the Bedrock Converse API (object-rooted inputSchema).
• Prompt-token budgeting — PromptBudget, apply_budget, estimate_tokens.

Install

pip install semql-prompt

Quick start

The catalog-level conveniences take the catalog as their first argument:

from semql import Catalog
from semql_prompt import planner_prompt, planner_prompt_segments, prompt_hash, to_openai_tools

text = planner_prompt(catalog, viewer=viewer)
segs = planner_prompt_segments(catalog)
key  = prompt_hash(catalog)
tools = to_openai_tools(catalog, viewer=viewer)

For the lower-level per-role fragment builders, see API reference.

License

BSD-3-Clause.