使用Cartopy遇struct.error: unpack需8字节缓冲区错误的解决咨询

运行基于Cartopy的地图程序时，在shapefile.py中触发struct.error: unpack requires a buffer of 8 bytes错误，按照开发者建议回退到Python3.8后问题依旧。

完整错误堆栈 #

(venv_with_python3.8) lnx@lnx-System-Product-Name:~/Desktop/globus$ python3 globus.py
2022-10-08 14:57:29,347 start
/home/lnx/virtual_env/venv_with_python3.8/lib/python3.8/site-packages/shapely/geometry/base.py:250: ShapelyDeprecationWarning: Setting the 'coords' to mutate a Geometry in place is deprecated, and will not be possible any more in Shapely 2.0
  super().__setattr__(name, value)
2022-10-08 14:57:32,114 reading the land polygons
2022-10-08 14:57:32,571 marking the ocean on axes 0
2022-10-08 14:57:47,094 marking the ocean on axes 1
2022-10-08 14:57:56,265 marking the ocean on axes 2
2022-10-08 14:58:11,742 marking the ocean on axes 3
2022-10-08 14:58:18,259 marking the ocean on axes 4
2022-10-08 14:58:24,119 marking the ocean on axes 5
2022-10-08 14:58:34,041 projecting the raster map on axes 0
2022-10-08 14:58:35,313 projecting the raster map on axes 1
2022-10-08 14:58:36,484 projecting the raster map on axes 2
2022-10-08 14:58:36,574 projecting the raster map on axes 3
2022-10-08 14:58:36,638 projecting the raster map on axes 4
2022-10-08 14:58:36,721 projecting the raster map on axes 5
2022-10-08 14:58:36,791 reading features
Traceback (most recent call last):
  File "globus.py", line 495, in <module>
    main()
  File "globus.py", line 468, in main
    axes.add_features()
  File "globus.py", line 236, in add_features
    borders = reader.ReadBorders()
  File "/home/lnx/Desktop/globus/reader.py", line 310, in ReadBorders
    b.geometry for b in ReadRecords('shapes/ne_10m_admin_0_boundary_lines_land')
  File "/home/lnx/Desktop/globus/reader.py", line 209, in ReadRecords
    return list(shapereader.Reader(name).records())
  File "/home/lnx/virtual_env/venv_with_python3.8/lib/python3.8/site-packages/cartopy/io/shapereader.py", line 167, in records
    for shape_record in self._reader.iterShapeRecords():
  File "/home/lnx/virtual_env/venv_with_python3.8/lib/python3.8/site-packages/shapefile.py", line 1771, in iterShapeRecords
    for shape, record in izip(self.iterShapes(), self.iterRecords(fields=fields)):
  File "/home/lnx/virtual_env/venv_with_python3.8/lib/python3.8/site-packages/shapefile.py", line 1483, in iterShapes
    shape = self.__shape(oid=i, bbox=bbox)
  File "/home/lnx/virtual_env/venv_with_python3.8/lib/python3.8/site-packages/shapefile.py", line 1309, in __shape
    (recNum, recLength) = unpack(">2i", f.read(8))
struct.error: unpack requires a buffer of 8 bytes
(venv_with_python3.8) lnx@lnx-System-Product-Name:~/Desktop/globus$

出错的shapefile.py代码片段 #

(recNum, recLength) = unpack(">2i", f.read(8))
        # Determine the start of the next record
        next = f.tell() + (2 * recLength)
        shapeType = unpack("<i", f.read(4))[0]
        record.shapeType = shapeType
        # For Null shapes create an empty points list for consistency
        if shapeType == 0:
            record.points = []
        # All shape types capable of having a bounding box
        elif shapeType in (3,5,8,13,15,18,23,25,28,31):
            record.bbox = _Array('d', unpack("<4d", f.read(32)))
            # if bbox specified and no overlap, skip this shape
            if bbox is not None and not bbox_overlap(bbox, record.bbox):
                # because we stop parsing this shape, skip to beginning of
                # next shape before we return
                f.seek(next)
                return None

修复方案 #

• 检查并重新下载shapefile文件：错误根源大概率是ne_10m_admin_0_boundary_lines_land相关的shp/shx/dbf文件损坏或下载不完整，重新获取完整的文件包替换原文件。
• 降级shapefile库版本：当前使用的pyshp（即shapefile库）版本可能存在兼容性问题，执行以下命令安装稳定版本：

  pip install pyshp==2.3.1
  

• 添加读取容错逻辑：如果不想更换库版本，可以修改shapefile.py第1309行附近代码，增加读取长度校验，避免因文件损坏导致崩溃：

  data = f.read(8)
  if len(data) != 8:
      # 跳到文件末尾终止循环，或根据需求自定义处理逻辑
      f.seek(os.path.getsize(f.name))
      return None
  (recNum, recLength) = unpack(">2i", data)
  

• 验证文件路径完整性：确保shapes/ne_10m_admin_0_boundary_lines_land路径下的所有配套文件（.shp、.shx、.dbf、.prj等）都存在且版本匹配。

内容的提问来源于stack exchange，提问作者RemiDocs