app.py

import gradio as gr, numpy as np
from PIL import Image, ImageOps, ImageDraw, ImageFont
from pathlib import Path
import os, requests
from features import binarize, feat_vec, cosine_sim, stroke_normalize, _ensure_ink_true
from features_preproc import crop_and_center as crop_ref, LO
from huggingface_hub import hf_hub_download
from skimage.color import rgb2gray
import numpy as np
ASSET_REPO = "raycosine/detangutify-data"
FONT_PATH = "data/NotoSerifTangut-Regular.ttf"
URL = "https://notofonts.github.io/tangut/fonts/NotoSerifTangut/full/ttf/NotoSerifTangut-Regular.ttf"

if not os.path.exists(FONT_PATH):
    os.makedirs(os.path.dirname(FONT_PATH), exist_ok=True)
    r = requests.get(URL)
    with open(FONT_PATH, "wb") as f:
        f.write(r.content)
DATA_PATH = hf_hub_download(repo_id=ASSET_REPO, repo_type="dataset",
                                filename="templates_aug.npz")
DATA = np.load(DATA_PATH)
X = DATA["X"] 
Y = DATA["y"] 
MEAN = DATA.get("mean", None)
STD = DATA.get("std", None)
SIZE = 64


THUMB = 128
_font = ImageFont.truetype(FONT_PATH, THUMB)

def render_glyph(cp:int) -> Image.Image:
    ch = chr(cp)
    canvas = Image.new("L", (THUMB, THUMB), 0)
    d = ImageDraw.Draw(canvas)
    bbox = d.textbbox((0,0), ch, font=_font)
    w, h = bbox[2]-bbox[0], bbox[3]-bbox[1]
    x = (THUMB - w)//2 - bbox[0]
    y = (THUMB - h)//2 - bbox[1]
    d.text((x, y), ch, fill=255, font=_font)

    return ImageOps.invert(canvas)
import numpy as np
from scipy.ndimage import center_of_mass
from PIL import Image

def crop_and_center_deprecated(bw, size=64, pad=2):
    coords = np.argwhere(bw > 0)
    if coords.size == 0:
        return np.zeros((size, size), dtype=np.float32)

    y0, x0 = coords.min(axis=0)
    y1, x1 = coords.max(axis=0) + 1
    cropped = bw[y0:y1, x0:x1]

    pil = Image.fromarray((cropped*255).astype(np.uint8))
    pil = pil.resize((size - 2*pad, size - 2*pad), Image.BILINEAR)

    canvas = Image.new("L", (size, size), 0)
    canvas.paste(pil, (pad, pad))
    arr = np.array(canvas, dtype=np.float32) / 255.0

    cy, cx = center_of_mass(arr)
    shift_y, shift_x = int(size/2 - cy), int(size/2 - cx)
    arr = np.roll(arr, shift_y, axis=0)
    arr = np.roll(arr, shift_x, axis=1)
    return arr
def _to_gray_uint8(arr: np.ndarray) -> np.uint8:

    if arr.ndim == 2:
        if arr.dtype != np.uint8:
            a = arr.astype(np.float32)
            if a.max() <= 1.0: a *= 255.0
            arr = np.clip(a, 0, 255).astype(np.uint8)
        return arr

    if arr.ndim == 3:
        a = arr.astype(np.float32)
        if a.max() > 1.0:
            a /= 255.0


        if a.shape[2] == 4:
            rgb = a[..., :3]
            alpha = a[..., 3:4]
            a = rgb * alpha + (1.0 - alpha) * 1.0

        elif a.shape[2] >= 3:
            a = a[..., :3] 

        g = rgb2gray(a)
        return (g * 255.0).astype(np.uint8)

    # 其它奇怪形状：兜底到 uint8
    return np.clip(arr, 0, 255).astype(np.uint8)
def infer(img):
    if img is None:
        return [], None

    if isinstance(img, dict):
        picked = None
        for k in ("image", "composite", "background"):
            v = img.get(k, None)
            if v is not None:
                picked = v
                break
        if picked is None:
            return [], None
        img = picked

    if isinstance(img, Image.Image):
        arr = np.array(img)
    else:
        arr = np.asarray(img)
    arr = _to_gray_uint8(arr)

    if arr.dtype != np.uint8:
        arr = np.clip(arr, 0, 255).astype(np.uint8)

    #pil = Image.fromarray(arr, mode="L").resize((SIZE, SIZE), Image.BILINEAR)
    #bw = binarize(np.array(pil, dtype=np.uint8))
    #bw = crop_and_center(bw, SIZE)
    #bw = stroke_normalize(bw, target_px=3)
    #bw = crop_ref(bw, out_size=LO, margin_frac=0.08)  # 用训练同款
    #bw = stroke_normalize(bw, target_px=2)
    bw0 = binarize(arr)
    bw0 = _ensure_ink_true(bw0)
    bw  = crop_ref(bw0, out_size=LO, margin_frac=0.08)
    bw  = stroke_normalize(bw, target_px=3)
    viz_img = Image.fromarray((bw*255).astype(np.uint8))
    q = feat_vec(bw)
    if MEAN is not None and STD is not None:
        q = (q - MEAN.ravel()) / STD.ravel()
    s = cosine_sim(q, X)
    idxs = np.argsort(-s)[:10]
    top, sec = float(s[idxs[0]]), float(s[idxs[1]]) if len(idxs)>1 else (float(s[idxs[0]]), -1)
    low_conf = (top < 0.58) or (top - sec < 0.05)
    seen = set()
    gallery_items = []
    results_json = []
    for idx in idxs:
        cp = int(Y[idx]); sc = float(s[idx])
        if cp in seen:
            continue
        seen.add(cp)
        glyph_img = render_glyph(cp)
        caption = f"U+{cp:05X} {chr(cp)}\nScore: {sc:.6f}"
        gallery_items.append((glyph_img, caption))
        results_json.append({"cp": cp, "char": chr(cp), "score": sc})
        if len(gallery_items) >= 10:
            break
    return gallery_items, viz_img, results_json


with gr.Blocks() as demo:
    
    gr.Markdown("### Detangutify (Tangut Character classifier)")
    with gr.Row():
        canvas = gr.Sketchpad(
            label="Draw here",
            image_mode="L",
            brush=gr.Brush(colors="black", default_size=2),
            canvas_size=(192,192),
            type="numpy",
        )

        upload = gr.Image(
            label="Upload a character image",
            type="numpy",   
            #image_mode="L",  
            #sources=["upload"],
            height=192
        )
    gallery = gr.Gallery(
        label="Top-10 Results",
        columns=10,
        preview=False,
        height=320
    )
    preview = gr.Image(label="stroke_normalize result", type="pil")
    jsonout = gr.JSON(label="Top-10 (JSON)", visible=False)
    btn_draw = gr.Button("Search (draw)")
    btn_draw.click(fn=infer, inputs=canvas, outputs=[gallery, preview, jsonout], api_name="predict")

    btn_upload = gr.Button("Search (upload)")
    btn_upload.click(fn=infer, inputs=upload, outputs=[gallery, preview, jsonout], api_name="predict_upload")
    upload.change(fn=infer, inputs=upload, outputs=[gallery, preview, jsonout])

    #canvas.change(fn=infer, inputs=canvas, outputs=gallery)
    api_img = gr.Image(type="pil", visible=False)
    api_btn = gr.Button(visible=False)
    api_btn.click(fn=infer, inputs=api_img, outputs=[gallery, preview, jsonout], api_name="predict_img")

if __name__ == "__main__":
    demo.launch()