Implement encoding_name() to get the encoding name

Author: abetomo

grnpy/context.pxd

@@ -23,3 +23,4 @@ cdef class Context:
     cdef grn_ctx *_ctx
 
     cdef grn_ctx *unwrap(self)
+    cdef encoding_name(self)

grnpy/context.pyx

@@ -18,10 +18,13 @@
 # distutils: sources = grnpy/grnpy_ctx.c
 
 from grnpy.grn_ctx cimport grn_ctx
+from grnpy.grn_encoding cimport grn_encoding
 from grnpy.grn_error cimport grn_rc
 from grnpy.grn_id cimport grn_id
 from grnpy.grn_obj cimport grn_obj
 
+cimport grnpy.grn_encoding
+
 from grnpy.object cimport build_object
 
 from .error import Error
@@ -39,6 +42,7 @@ cdef extern from "groonga.h":
 cdef extern from "grnpy_ctx.h":
     grn_rc grnpy_ctx_get_rc(grn_ctx *ctx)
     const char *grnpy_ctx_get_error_message(grn_ctx *ctx)
+    grn_encoding grnpy_ctx_get_encoding(grn_ctx *ctx)
 
 cdef class Context:
     def __cinit__(self, flags=0):
@@ -56,6 +60,23 @@ cdef class Context:
     cdef grn_ctx *unwrap(self):
         return self._ctx
 
+    cdef encoding_name(self):
+        # https://docs.python.org/3/library/codecs.html
+        cdef grn_encoding encoding = grnpy_ctx_get_encoding(self._ctx)
+        if encoding == grnpy.grn_encoding.UTF8:
+            return "utf_8"
+        elif encoding == grnpy.grn_encoding.EUC_JP:
+            return "euc_jp"
+        elif encoding == grnpy.grn_encoding.SJIS:
+            # CP932 has better coverage of special characters.
+            return "cp932"
+        elif encoding == grnpy.grn_encoding.LATIN1:
+            return "latin_1"
+        elif encoding == grnpy.grn_encoding.KOI8R:
+            return "koi8_r"
+        else:
+            raise NotImplementedError(f"unsupported encoding: <{encoding}>")
+
     def rc(self):
         return grnpy_ctx_get_rc(self._ctx)
 

grnpy/grnpy_ctx.c

@@ -29,3 +29,9 @@ grnpy_ctx_get_error_message(grn_ctx *ctx)
 {
   return ctx->errbuf;
 }
+
+grn_encoding
+grnpy_ctx_get_encoding(grn_ctx *ctx)
+{
+  return GRN_CTX_GET_ENCODING(ctx);
+}

grnpy/grnpy_ctx.h

@@ -22,3 +22,4 @@
 
 grn_rc grnpy_ctx_get_rc(grn_ctx *ctx);
 const char *grnpy_ctx_get_error_message(grn_ctx *ctx);
+grn_encoding grnpy_ctx_get_encoding(grn_ctx *ctx);

grnpy/object.pyx

@@ -89,7 +89,7 @@ cdef class Object:
         elif isinstance(value, float):
             rc = grnpy_obj_set_float(ctx, self.unwrap(), id, value)
         elif isinstance(value, str):
-            value_bytes = value.encode()
+            value_bytes = value.encode(context.encoding_name())
             rc = grnpy_obj_set_text(ctx, self.unwrap(), id, value_bytes, len(value_bytes))
         else:
             raise TypeError(f"unsupported value type: <{type(value)}>")