Analysis Components

AnalysisComponent are the building blocks of extraction in Pāṇini. Each component is responsible for a specific analysis axis (Morphology, Pedagogical Explanation, Leipzig Glossing, etc.).

---

1. Role and Principles

A component is an isolated unit of work that: 1. Defines its own data structure JSON (via a schema). 2. Provides its own instructions to the AI (via a prompt fragment). 3. Validates and cleans its own data after extraction.

This modular approach allows you to ask the AI only for what you need, reducing costs and improving accuracy.

graph TD
    Registry["Registry (ISO Code)"]
    Registry -->|fra| French[French Definition]
    Registry -->|ara| Arabic[Arabic Definition]
    Registry -->|tur| Turkish[Turkish Definition]

    ComponentSelection["Component Keys (CLI/API)"]
    ComponentSelection --> C1[Morphology]
    ComponentSelection --> C2[Leipzig]
    ComponentSelection --> C3[Pedagogical]

    French & C1 & C2 & C3 --> Pipeline[Extraction Pipeline]

---

2. Usage

Selecting Components (CLI)

By default, all components compatible with the target language are executed. You can restrict the selection with the --components flag:

# Execute only morphology and Leipzig glossing
panini extract --components morphology,leipzig_alignment \
  --text "Dał kotowi mleko." \
  --target "kotowi"

Usage via API (Rust)

let result = registry::extract_erased_with_components(
    "pol", &model, &request,
    Some(&["morphology", "pedagogical_explanation"]),
    0.2, 4096, None, &prompts,
).await?;

---

3. Built-in Components

Key	Struct	Description
morphology	MorphologyAnalysis	Full morphological analysis (POS, lemma, case, gender...).
pedagogical_explanation	PedagogicalExplanation	Structured HTML explanation for the learner.
morpheme_segmentation	MorphemeSegmentation	Morpheme segmentation (Agglutinative languages).
multiword_expressions	MultiwordExpressions	Detection of idioms and multi-word expressions.
leipzig_alignment	LeipzigAlignment	Interlinear glossing (Leipzig Rules).

---

4. Consuming Results

The recommended way to consume results is through the #[derive(PaniniResult)] macro. It automatically maps JSON keys to your struct fields.

#[derive(PaniniResult)]
pub struct FullExtraction<L: LinguisticDefinition> {
    #[component(MorphologyAnalysis)]
    pub morphology: MorphologyResult<L::Morphology>,

    #[component(PedagogicalExplanation)]
    pub explanation: String,
}

Optional Results

If a component is optional (e.g., MorphemeSegmentation, which only runs on compatible languages), use Option<T> in your struct.