Merge branch 'hotfix/1.3.3'

Author: shahramn

CHANGELOG.rst

@@ -2,11 +2,18 @@
 Changelog for eccodes-python
 ============================
 
+1.3.3 (2021-mm-dd)
+--------------------
+
+- ECC-1246: UnicodeDecodeError when parsing BUFR file
+
+
 1.3.2 (2021-04-16)
 --------------------
 
 - Restore the experimental high-level interface
 
+
 1.3.1 (2021-04-16)
 --------------------
 

gribapi/bindings.py

@@ -20,7 +20,7 @@
 
 import cffi
 
-__version__ = "1.3.2"
+__version__ = "1.3.3"
 
 LOG = logging.getLogger(__name__)
 

gribapi/gribapi.py

@@ -491,7 +491,7 @@ def grib_get_string(msgid, key):
     length_p = ffi.new("size_t *", length)
     err = lib.grib_get_string(h, key.encode(ENC), values, length_p)
     GRIB_CHECK(err)
-    return ffi.string(values, length_p[0]).decode(ENC)
+    return _decode_bytes(values, length_p[0])
 
 
 @require(msgid=int, key=str, value=str)
@@ -1174,6 +1174,19 @@ def grib_get_double_array(msgid, key):
     return arr
 
 
+# See ECC-1246
+def _decode_bytes(binput, maxlen=None):
+    if maxlen:
+        a_str = ffi.string(binput, maxlen)
+    else:
+        a_str = ffi.string(binput)
+    # Check for a MISSING value i.e., each character has all its bits=1
+    if all(x == 255 for x in a_str):
+        return ""
+    # Replace with a suitable replacement character rather than throw an exception
+    return a_str.decode(ENC, "replace")
+
+
 @require(msgid=int, key=str)
 def grib_get_string_array(msgid, key):
     """
@@ -1192,7 +1205,7 @@ def grib_get_string_array(msgid, key):
     size_p = ffi.new("size_t *", size)
     err = lib.grib_get_string_array(h, key.encode(ENC), values, size_p)
     GRIB_CHECK(err)
-    return [ffi.string(values[i]).decode(ENC) for i in range(size_p[0])]
+    return [_decode_bytes(values[i]) for i in range(size_p[0])]
 
 
 @require(msgid=int, key=str)