Upload scripts/vision_ocr.py with huggingface_hub

scripts/vision_ocr.py

@@ -1,3 +1,49 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:fdcdc5154f8a6f015f271fd1ccc7d1f5329503d787b07559f4bff7b9892bcd3d
-size 1889
+#!/usr/bin/env python3
+"""单图验证码识别，走 pie-xian command-a-vision。输出格式兼容 harness 的 recognize()。
+用法: vision_ocr.py file.png ... -> {"file.png": {"vision": "code"}}
+"""
+import base64
+import json
+import re
+import ssl
+import sys
+import urllib.request
+
+API = "https://api.pie-xian.com/v1/chat/completions"
+KEY = "sk-Rel8iERHrmu9NDbvQhnhwTowEUnGqVvCFXMatASH8q5O4gGP"
+MODEL = "command-a-vision-07-2025"
+UA = "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 Chrome/126.0.0.0 Safari/537.36"
+PROMPT = "识别这张验证码图片中的字符。只输出字符本身,不要任何解释或标点。这是登录验证码,可能包含1-4个数字。"
+
+_CTX = ssl.create_default_context()
+_CTX.check_hostname = False
+_CTX.verify_mode = ssl.CERT_NONE
+
+
+def read(path):
+    b64 = base64.b64encode(open(path, "rb").read()).decode()
+    body = json.dumps({"model": MODEL, "messages": [{"role": "user", "content": [
+        {"type": "text", "text": PROMPT},
+        {"type": "image_url", "image_url": {"url": "data:image/png;base64," + b64}}]}],
+        "max_tokens": 20, "temperature": 0}).encode()
+    req = urllib.request.Request(API, data=body, headers={
+        "Authorization": "Bearer " + KEY, "Content-Type": "application/json", "User-Agent": UA})
+    d = json.loads(urllib.request.urlopen(req, timeout=90, context=_CTX).read())
+    return re.sub(r"[^0-9A-Za-z]", "", d["choices"][0]["message"]["content"])
+
+
+def main():
+    out = {}
+    for f in sys.argv[1:]:
+        try:
+            # 2 次读取，一致才返回；不一致返回空（harness 跳过登录）
+            r1, r2 = read(f), read(f)
+            code = r1 if r1 == r2 else ""
+            out[f] = {"vision": code}
+        except Exception as e:
+            out[f] = {"vision": "", "err": str(e)}
+    print(json.dumps(out, ensure_ascii=False))
+
+
+if __name__ == "__main__":
+    main()