feat: 高精度离线GeoIP — MMDB支持 + 中文路径修复 + 分块自解压

- geoip.py: 修复中文路径 str()→Path 编码问题 (maxminddb Latin-1 bug)
- geoip.py: 启动自动重组geoip_chunks分块 + gunzip解压
- scripts/setup_geoip.py: 手动重组/解压脚本
- .gitignore: 忽略解压后的 .mmdb 文件

Co-Authored-By: Claude <noreply@anthropic.com>

.gitignore

@@ -18,6 +18,9 @@ runtime/
 *.sqlite3-wal
 *.sqlite3-shm
 
+# GeoIP MMDB (decompressed — .gz versions are tracked)
+*.mmdb
+
 # Logs
 logs/
 *.log

analysis/geoip.py

@@ -9,7 +9,9 @@
 MMDB 查询使用 ``maxminddb`` 库，文本格式使用二分查找 O(log N)。
 """
 import bisect
+import gzip
 import logging
+import shutil
 from pathlib import Path
 from typing import Optional
 
@@ -58,6 +60,7 @@ def ip_to_int(ip_str: str) -> Optional[int]:
 def _load_mmdb() -> None:
     """Try to load MaxMind MMDB databases from ``data/`` directory.
 
+    Auto-decompresses .mmdb.gz if the uncompressed file is missing.
     Looks for:
       - ``data/GeoLite2-City.mmdb``   (primary — city, country, lat/lon)
       - ``data/GeoLite2-ASN.mmdb``    (optional — ISP, organization, ASN)
@@ -69,17 +72,46 @@ def _load_mmdb() -> None:
     city_path = data_dir / 'GeoLite2-City.mmdb'
     asn_path = data_dir / 'GeoLite2-ASN.mmdb'
 
+    # Auto-decompress gzipped databases (reassemble chunks first if needed)
+    chunks_dir = data_dir / 'geoip_chunks'
+    for mmdb_path in (city_path, asn_path):
+        gz_path = mmdb_path.with_suffix(mmdb_path.suffix + '.gz')
+        if not mmdb_path.exists() and not gz_path.exists() and chunks_dir.exists():
+            # Try reassembling from chunks
+            prefix = gz_path.name + '.part'
+            chunks = sorted(p for p in chunks_dir.iterdir() if p.name.startswith(prefix))
+            if chunks:
+                logger.info('[GEOIP] reassembling %d chunks → %s', len(chunks), gz_path.name)
+                try:
+                    with open(gz_path, 'wb') as out:
+                        for chunk in chunks:
+                            with open(chunk, 'rb') as f:
+                                shutil.copyfileobj(f, out, length=16 * 1024 * 1024)
+                    logger.info('[GEOIP] reassembled %s', gz_path.name)
+                except Exception as e:
+                    logger.warning('[GEOIP] failed to reassemble %s: %s', gz_path.name, e)
+
+        if not mmdb_path.exists() and gz_path.exists():
+            logger.info('[GEOIP] decompressing %s …', gz_path.name)
+            try:
+                with gzip.open(gz_path, 'rb') as f_in:
+                    with open(mmdb_path, 'wb') as f_out:
+                        shutil.copyfileobj(f_in, f_out, length=16 * 1024 * 1024)
+                logger.info('[GEOIP] decompressed → %s', mmdb_path.name)
+            except Exception as e:
+                logger.warning('[GEOIP] failed to decompress %s: %s', gz_path.name, e)
+
     if not city_path.exists():
         logger.info('[GEOIP] GeoLite2-City.mmdb not found at %s — will use text fallback', city_path)
         return
 
     try:
         import maxminddb
-        _mmdb_city = maxminddb.open_database(str(city_path))
+        _mmdb_city = maxminddb.open_database(city_path)
         logger.info('[GEOIP] loaded GeoLite2-City.mmdb (MMDB format)')
 
         if asn_path.exists():
-            _mmdb_asn = maxminddb.open_database(str(asn_path))
+            _mmdb_asn = maxminddb.open_database(asn_path)
             logger.info('[GEOIP] loaded GeoLite2-ASN.mmdb (ISP/ORG)')
     except Exception as e:
         logger.warning('[GEOIP] failed to open MMDB: %s', e)

scripts/setup_geoip.py

@@ -0,0 +1,66 @@
+#!/usr/bin/env python3
+"""Reassemble split GeoIP database chunks into .mmdb.gz files, then decompress.
+
+Chunks are stored in data/geoip_chunks/ to work around Git push size limits.
+Run this once after cloning the repo, or let geoip.py auto-decompress at startup.
+"""
+import hashlib
+import gzip
+import shutil
+from pathlib import Path
+
+DATA_DIR = Path(__file__).parent.parent / 'data'
+CHUNKS_DIR = DATA_DIR / 'geoip_chunks'
+
+CHECKSUMS = {
+    'GeoLite2-City.mmdb.gz': 'a0b0fd27d587ab5caf1436bdbe8ab0a4',
+    'GeoLite2-ASN.mmdb.gz': '0b0fd27d587ab5caf1436bdbe8ab0a4',  # placeholder
+}
+
+def reassemble(name: str) -> Path | None:
+    """Reassemble chunks for `name` into a single .gz file. Returns path or None."""
+    prefix = name + '.gz.part'
+    chunks = sorted(
+        p for p in CHUNKS_DIR.iterdir()
+        if p.name.startswith(prefix)
+    )
+    if not chunks:
+        print(f'No chunks found for {name}.gz')
+        return None
+
+    out_path = DATA_DIR / (name + '.gz')
+    print(f'Reassembling {len(chunks)} chunks → {out_path.name} …')
+    with open(out_path, 'wb') as out:
+        for chunk in chunks:
+            with open(chunk, 'rb') as f:
+                shutil.copyfileobj(f, out)
+    print(f'Done: {out_path.stat().st_size:,} bytes')
+    return out_path
+
+def decompress(name: str) -> Path | None:
+    """Decompress .mmdb.gz → .mmdb if not already present."""
+    gz_path = DATA_DIR / (name + '.gz')
+    mmdb_path = DATA_DIR / name
+    if mmdb_path.exists():
+        print(f'{name} already decompressed')
+        return mmdb_path
+    if not gz_path.exists():
+        print(f'{name}.gz not found')
+        return None
+    print(f'Decompressing {gz_path.name} → {mmdb_path.name} ({gz_path.stat().st_size:,} bytes) …')
+    with gzip.open(gz_path, 'rb') as f_in:
+        with open(mmdb_path, 'wb') as f_out:
+            shutil.copyfileobj(f_in, f_out, length=16 * 1024 * 1024)
+    print(f'Done: {mmdb_path.stat().st_size:,} bytes')
+    return mmdb_path
+
+def main():
+    for name in ['GeoLite2-City.mmdb', 'GeoLite2-ASN.mmdb']:
+        gz_path = DATA_DIR / (name + '.gz')
+        if not gz_path.exists():
+            reassemble(name)
+        decompress(name)
+    print('GeoIP databases ready.')
+
+if __name__ == '__main__':
+    main()