app.py

from strive.reranker import Reranker, EmbeddingType, deduplicate_results
import gradio as gr
import time

def rerank_and_deduplicate(query, corpus_text):
    corpus = [line.strip() for line in corpus_text.split("\n") if line.strip()]

    start = time.perf_counter()
    textual_reranker = Reranker(embedding_type=EmbeddingType.textual)
    semantic_reranker = Reranker(embedding_type=EmbeddingType.semantic)

    textual_results = textual_reranker.rerank_documents(query, corpus, top_k=len(corpus))
    semantic_results = semantic_reranker.rerank_documents(query, corpus, top_k=len(corpus))

    merged_results = textual_results + semantic_results
    deduplicated_results = deduplicate_results(merged_results, top_k=50)
    elapsed_ms = (time.perf_counter() - start) * 1000

    header = f"Reranked {len(corpus)} documents in {elapsed_ms:.1f} ms on CPU\n\n"
    return header + "\n".join([f"{text} (Score: {score:.4f})" for text, score in deduplicated_results])

laptop_corpus = "\n".join([
    "The best budget laptops for programming in 2026 balance a fast CPU, at least 16GB of RAM, and a comfortable keyboard.",
    "A good programming laptop needs enough memory to run an IDE, a browser with dozens of tabs, and a local dev server at the same time.",
    "Affordable laptops for coding usually trade a discrete GPU for a better CPU and more RAM within the same budget.",
    "The best budget laptops for programming in 2026 balance a fast CPU, at least 16GB of RAM, and a comfortable keyboard.",
    "The latest flagship smartphone features a titanium frame and a much larger camera sensor than last year's model.",
    "For gaming, a discrete GPU with at least 8GB of VRAM matters far more than raw CPU clock speed.",
    "A mechanical keyboard with hot-swappable switches is popular among developers who type for long hours.",
    "Budget desktop towers can be upgraded piece by piece, unlike laptops which are mostly fixed at purchase time.",
    "Look for a laptop with a matte 1080p or higher display, since glossy screens cause glare during long coding sessions.",
    "The new wireless earbuds promise 30 hours of battery life with the charging case included.",
    "Cheap laptops with only 8GB of RAM will struggle once you open a Docker container alongside your editor.",
    "A 14-inch or 15-inch laptop is usually the sweet spot between portability and screen real estate for coding.",
])

health_corpus = "\n".join([
    "Common symptoms of vitamin D deficiency include fatigue, bone pain, muscle weakness, and mood changes.",
    "People with low vitamin D levels often report persistent tiredness even after a full night's sleep.",
    "Vitamin D deficiency can cause bone and muscle pain, and in severe cases contributes to osteoporosis.",
    "Common symptoms of vitamin D deficiency include fatigue, bone pain, muscle weakness, and mood changes.",
    "Iron deficiency anemia typically presents with pale skin, shortness of breath, and brittle nails.",
    "Sunlight exposure on bare skin triggers natural vitamin D synthesis, though sunscreen reduces this effect.",
    "A balanced diet rich in leafy greens supports overall micronutrient intake, including iron and folate.",
    "Low vitamin D has also been associated with a weakened immune response and slower wound healing.",
    "Regular cardiovascular exercise improves resting heart rate and long-term cardiovascular health.",
    "Vitamin D supplements are commonly recommended during winter months in regions with limited sunlight.",
    "Chronic fatigue can also stem from thyroid disorders, sleep apnea, or unmanaged stress.",
    "Blood tests measuring 25-hydroxyvitamin D are the standard way to confirm a deficiency diagnosis.",
])

devops_corpus = "\n".join([
    "Reducing cloud infrastructure costs often starts with rightsizing over-provisioned virtual machines.",
    "Switching non-critical workloads to spot or preemptible instances can cut compute costs by over 60%.",
    "Unused storage volumes and orphaned snapshots quietly accumulate cost if nobody audits them regularly.",
    "Reducing cloud infrastructure costs often starts with rightsizing over-provisioned virtual machines.",
    "A well-designed CI/CD pipeline reduces the time between a commit and a production deployment.",
    "Reserved instances offer significant discounts in exchange for a one- or three-year usage commitment.",
    "Container orchestration platforms like Kubernetes can bin-pack workloads to improve node utilization.",
    "Feature flags let teams ship code to production without immediately exposing it to all users.",
    "Setting up autoscaling policies prevents both over-provisioning during quiet periods and outages during spikes.",
    "A service mesh adds observability and traffic control between microservices at the cost of some latency.",
    "Cost allocation tags make it possible to attribute cloud spend back to individual teams or products.",
    "Moving infrequently accessed data to cold storage tiers can reduce storage costs substantially.",
])

recipe_corpus_pt = "\n".join([
    "Uma boa receita de bolo de chocolate leva farinha, cacau em pó, ovos, açúcar e fermento em pó.",
    "Para um bolo de chocolate úmido, é importante não bater a massa demais depois de adicionar a farinha.",
    "O segredo de um bolo de chocolate fofo está no ponto certo do fermento e no tempo de forno.",
    "Uma boa receita de bolo de chocolate leva farinha, cacau em pó, ovos, açúcar e fermento em pó.",
    "A receita de brigadeiro gourmet costuma usar chocolate meio amargo em vez do achocolatado tradicional.",
    "Um bom risoto de camarão exige caldo quente sendo adicionado aos poucos, sem parar de mexer.",
    "Cobertura de ganache é feita derretendo chocolate com creme de leite quente em partes iguais.",
    "A farofa crocante combina bem com carnes assadas e é feita com farinha de mandioca torrada na manteiga.",
    "Para testar se o bolo está pronto, espete um palito no centro; se sair limpo, está assado.",
    "Pão de queijo mineiro tradicional é feito com polvilho azedo e queijo meia cura ralado.",
    "Bater as claras em neve antes de incorporar à massa deixa o bolo de chocolate ainda mais leve.",
    "Salada caprese leva apenas tomate, mussarela de búfala, manjericão fresco e azeite de oliva.",
])

gradio_examples = [
    ["best budget laptop for programming", laptop_corpus],
    ["symptoms of vitamin D deficiency", health_corpus],
    ["how to reduce cloud infrastructure costs", devops_corpus],
    ["receita de bolo de chocolate", recipe_corpus_pt],
]

app = gr.Interface(
    fn=rerank_and_deduplicate,
    inputs=[
        gr.Textbox(label="Query", placeholder="Enter your query here"),
        gr.Textbox(label="Corpus", placeholder="Enter one sentence per line", lines=10)
    ],
    outputs=gr.Textbox(label="Top Ranked Results"),
    title="STRIVE: Semantic Tokenized Ranking via Vectorization & Embeddings",
    description="Enter a query and multiple sentences to test the reranking algorithm.",
    examples=gradio_examples,
)

app.launch()