Upload scripts/ocr.py with huggingface_hub

scripts/ocr.py

@@ -0,0 +1,51 @@
+#!/usr/bin/env python3
+"""批量验证码识别器（准确性测试用）。
+加载 old / beta 两个 ddddocr 模型，各跑一次完整字符集和一次字母数字限定字符集，
+对每个输入图片输出 4 个变体的识别结果 JSON。
+用法: ocr.py file.png ... > result.json
+"""
+import json
+import sys
+
+import ddddocr
+
+ALNUM = "0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ"
+
+_VARIANTS = {}
+
+
+def _make(**kwargs):
+    ocr = ddddocr.DdddOcr(show_ad=False, **kwargs)
+    ocr.set_ranges(ALNUM)
+    return ocr
+
+
+def _load():
+    if _VARIANTS:
+        return _VARIANTS
+    _VARIANTS["old"] = _make(old=True)
+    _VARIANTS["old_r"] = _make(old=True)
+    _VARIANTS["beta"] = _make(beta=True)
+    _VARIANTS["beta_r"] = _make(beta=True)
+    return _VARIANTS
+
+
+def main():
+    files = sys.argv[1:]
+    if not files:
+        print("usage: ocr.py file.png ... > result.json", file=sys.stderr)
+        sys.exit(2)
+    variants = _load()
+    out = {}
+    for f in files:
+        try:
+            data = open(f, "rb").read()
+            row = {k: variants[k].classification(data) for k in variants}
+        except Exception as e:  # noqa: BLE001
+            row = {"err": str(e)}
+        out[f] = row
+    print(json.dumps(out, ensure_ascii=False))
+
+
+if __name__ == "__main__":
+    main()