cognis-core 0.3.0

Foundation traits and types for Cognis: typed Runnable<I, O>, Message, errors, streaming, prompts, output parsers, callbacks. Zero external integrations.
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43

//! Typed parsers that convert LLM string output into structured values.
//!
//! Each parser implements:
//! - [`OutputParser<T>`] — for direct parsing.
//! - `Runnable<String, T>` — for chain composition (`prompt | model | parser`).
//!
//! Parsers also expose a `format_instructions()` snippet that can be embedded
//! in a prompt to tell the LLM how to format its output.

pub mod boolean;
pub mod fixing;
pub mod json;
pub mod list;
pub mod string;
pub mod structured;
pub mod xml;

pub use boolean::BooleanParser;
pub use fixing::{OutputFixingParser, RetryParser};
pub use json::JsonParser;
pub use list::{CommaListParser, NumberedListParser};
pub use string::StringParser;
pub use structured::{
    JsonExtraction, JsonExtractor, StructuredOutputConfig, StructuredOutputParser,
};
pub use xml::XmlParser;

use crate::Result;

/// Parses LLM string output into a typed value `T`.
pub trait OutputParser<T>: Send + Sync
where
    T: Send + 'static,
{
    /// Attempt to parse the LLM output.
    fn parse(&self, text: &str) -> Result<T>;

    /// Optional human-readable instructions to embed in a prompt to tell
    /// the LLM how to format its output.
    fn format_instructions(&self) -> Option<String> {
        None
    }
}