Struct HttpExtractionProvider 

pub struct HttpExtractionProvider { /* private fields */ }

HTTP-based extraction provider that calls OpenAI, Anthropic, or Ollama APIs.

Implementations

impl HttpExtractionProvider

pub fn new(config: ExtractionConfig) -> Result<Self, ExtractionError>

pub async fn expand_query( &self, query: &str, ) -> Result<Vec<String>, ExtractionError>

Expand a search query into multiple sub-queries via LLM.

Given a natural language question, identifies the expected answer type and extracts 2-3 targeted search queries. The first line of the response is the answer type (PLACE, DATE, NUMBER, NAME, PERSON, BRAND, etc.), followed by the search queries.

For counting/aggregation/comparison queries, also generates comprehensive category synonyms for exhaustive BM25 sweep.

pub async fn call_with_retry( &self, conversation: &str, system_prompt: &str, ) -> Result<String, ExtractionError>

Execute a request with retry logic for rate limits (HTTP 429). Uses exponential backoff: 1s, 2s, 4s.

pub async fn call_text_with_retry( &self, conversation: &str, system_prompt: &str, ) -> Result<String, ExtractionError>

Like call_with_retry but without forcing JSON response format. Use for prompts that expect plain text output (synthesis, re-ranking, etc).

Trait Implementations

impl ExtractionProvider for HttpExtractionProvider

async fn extract( &self, conversation: &str, system_prompt: &str, ) -> Result<String, ExtractionError>

Send a conversation to the LLM with the given system prompt and return the raw response text (expected to be JSON).