src/submission/submit.py

import json
import os
import sys
from datetime import datetime, timezone

import requests

from src.display.formatting import styled_error, styled_message, styled_warning
from src.envs import API, settings
from src.submission.check_validity import (
    already_submitted_models,
    check_model_card,
    get_model_size,
    is_model_on_hub,
)

if sys.version_info < (3, 11):
    UTC = timezone.utc
else:
    from datetime import UTC

REQUESTED_MODELS: set[str] | None = None


def add_new_submit(
    model: str,
    base_model: str,
    revision: str | None,
    precision: str,
    weight_type: str,
    model_type: str,
    json_str: str,
    commit_message: str,
    user_id: str,
):
    """
    Submit a new evaluation request.

    Args:
        model: Model name (e.g., "org/model_name")
        base_model: Base model name (for delta or adapter weights)
        revision: Model revision/commit (defaults to "main" if empty)
        precision: Model precision (e.g., "float16", "bfloat16")
        weight_type: Weight type (e.g., "Original", "Delta", "Adapter")
        model_type: Model type (e.g., "pretrained", "fine-tuned")
        json_str: JSON string containing config and results
        commit_message: Optional commit message
        user_id: Submitter's HuggingFace user ID/username (from OAuth)
    """
    global REQUESTED_MODELS
    if not REQUESTED_MODELS:
        REQUESTED_MODELS, _ = already_submitted_models(settings.EVAL_REQUESTS_PATH.as_posix())

    # Use provided user_id, or extract from model name as fallback

    if " " in precision:
        precision = precision.split(" ")[0]
    # Does the model actually exist?
    revision = revision or None

    # Is the model on the hub?
    if weight_type in ["Delta", "Adapter"]:
        base_model_on_hub, error, _ = is_model_on_hub(
            model_name=base_model,
            revision=revision or "main",
            token=settings.HF_TOKEN.get_secret_value(),
            test_tokenizer=True,
        )
        if not base_model_on_hub:
            return styled_error(f'Base model "{base_model}" {error}')

    if not weight_type == "Adapter":
        model_on_hub, error, _ = is_model_on_hub(
            model_name=model,
            revision=revision or "main",
            token=settings.HF_TOKEN.get_secret_value(),
            test_tokenizer=True,
        )
        if not model_on_hub:
            return styled_error(f'Model "{model}" {error}')

    # Is the model info correctly filled?
    try:
        model_info = API.model_info(repo_id=model, revision=revision)
    except Exception:
        return styled_error("Could not get your model information. Please fill it up properly.")

    # Were the model card and license filled?
    try:
        license = model_info.cardData["license"]
    except Exception:
        return styled_error("Please select a license for your model")

    # Validate required fields
    if not model or not model.strip():
        return styled_error("Model name is required.")
    if not user_id or not user_id.strip():
        return styled_error("User ID/username is required. Please make sure you are logged in.")

    # Get current UTC time for submit_time
    current_time = datetime.now(UTC).strftime("%Y-%m-%dT%H:%M:%SZ")

    # Parse the evaluation results JSON (json_str contains config and results)
    try:
        eval_results = json.loads(json_str)
    except json.JSONDecodeError:
        return styled_error("Invalid evaluation results JSON format.")

    # Organize all fields into a comprehensive JSON structure for the content field
    # This will be the complete JSON that gets uploaded as a file
    model_type = model_type.rpartition(":")[2].strip()  # "⭕ : instruction-tuned" -> "instruction-tuned"
    complete_submission_content = {
        "user_id": user_id,
        "model_id": model,
        "base_model": base_model or "",
        "model_sha": revision,
        "model_dtype": precision,
        "weight_type": weight_type,
        "model_type": model_type or "",
        "submit_time": current_time,
        "commit_message": commit_message,
        # Include the evaluation results (config and results)
        "config": eval_results.get("config", {}),
        "results": eval_results.get("results", {}),
    }

    # Convert the complete submission content to JSON string for the content field
    complete_content_json_str = json.dumps(complete_submission_content, indent=2, ensure_ascii=False)

    # Request JSON for the API call - includes all fields separately
    request_json = {
        "username": user_id,
        "model_id": model,
        "base_model": base_model or "",
        "model_sha": revision,
        "model_dtype": precision,
        "weight_type": weight_type,
        "model_type": model_type or "",
        "content": complete_content_json_str,  # Complete JSON with all fields
        "submit_time": current_time,
        "commit_message": commit_message,
    }

    # Check for duplicate submission
    if f"{model}_{revision}_{precision}" in REQUESTED_MODELS:
        return styled_warning("This model has been already submitted.")

    try:
        response = requests.post(
            url=f"http://localhost:{settings.BACKEND_PORT}/api/v1/hf/community/submit/",
            json=request_json,  # 使用 json 参数发送 JSON body
            headers={"Content-Type": "application/json"},
        )
        print("response: ", response)  # print response content for debugging
        if response.status_code == 200:
            data = response.json()
            print("returned data: ", data)
            if data.get("code") == 0:
                return styled_message(
                    "Your request has been submitted to the evaluation queue!\nPlease wait for the model to show in the PENDING list."
                )
        return styled_error("Submission unsuccessful.")
    except Exception:
        return styled_error("Submission unsuccessful.")


def add_new_eval(
    model: str,
    base_model: str,
    revision: str,
    precision: str,
    weight_type: str,
    model_type: str,
):
    global REQUESTED_MODELS
    if not REQUESTED_MODELS:
        REQUESTED_MODELS, _ = already_submitted_models(settings.EVAL_REQUESTS_PATH.as_posix())

    user_name = ""
    model_path = model
    if "/" in model:
        user_name = model.split("/")[0]
        model_path = model.split("/")[1]

    precision = precision.split(" ")[0]
    current_time = datetime.now(UTC).strftime("%Y-%m-%dT%H:%M:%SZ")

    if model_type is None or model_type == "":
        return styled_error("Please select a model type.")

    # Does the model actually exist?
    if revision == "":
        revision = "main"

    # Is the model on the hub?
    if weight_type in ["Delta", "Adapter"]:
        base_model_on_hub, error, _ = is_model_on_hub(
            model_name=base_model, revision=revision, token=settings.HF_TOKEN.get_secret_value(), test_tokenizer=True
        )
        if not base_model_on_hub:
            return styled_error(f'Base model "{base_model}" {error}')

    if not weight_type == "Adapter":
        model_on_hub, error, _ = is_model_on_hub(
            model_name=model, revision=revision, token=settings.HF_TOKEN.get_secret_value(), test_tokenizer=True
        )
        if not model_on_hub:
            return styled_error(f'Model "{model}" {error}')

    # Is the model info correctly filled?
    try:
        model_info = API.model_info(repo_id=model, revision=revision)
    except Exception:
        return styled_error("Could not get your model information. Please fill it up properly.")

    model_size = get_model_size(model_info=model_info, precision=precision)

    # Were the model card and license filled?
    try:
        license = model_info.cardData["license"]
    except Exception:
        return styled_error("Please select a license for your model")

    modelcard_OK, error_msg = check_model_card(model)
    if not modelcard_OK:
        return styled_error(error_msg)

    # Seems good, creating the eval
    print("Adding new eval")

    eval_entry = {
        "model": model,
        "base_model": base_model,
        "revision": revision,
        "precision": precision,
        "weight_type": weight_type,
        "status": "PENDING",
        "submitted_time": current_time,
        "model_type": model_type,
        "likes": model_info.likes,
        "params": model_size,
        "license": license,
        "private": False,
    }

    # Check for duplicate submission
    if f"{model}_{revision}_{precision}" in REQUESTED_MODELS:
        return styled_warning("This model has been already submitted.")

    print("Creating eval file")
    OUT_DIR = f"{settings.EVAL_REQUESTS_PATH}/{user_name}"
    os.makedirs(OUT_DIR, exist_ok=True)
    out_path = f"{OUT_DIR}/{model_path}_eval_request_False_{precision}_{weight_type}.json"

    with open(out_path, "w") as f:
        f.write(json.dumps(eval_entry))

    print("Uploading eval file")
    API.upload_file(
        path_or_fileobj=out_path,
        path_in_repo=out_path.split("eval-queue/")[1],
        repo_id=settings.QUEUE_REPO_ID,
        repo_type="dataset",
        commit_message=f"Add {model} to eval queue",
    )

    # Remove the local file
    os.remove(out_path)

    return styled_message(
        "Your request has been submitted to the evaluation queue!\nPlease wait for up to an hour for the model to show in the PENDING list."
    )