pg 0.18.0 → 1.1.4

data/ext/pg_copy_coder.c

@@ -167,9 +167,15 @@ pg_copycoder_type_map_get(VALUE self)
  *     conn.put_copy_data ["string2", 42, true]
  *   end
  * This creates +my_table+ and inserts two rows.
+ *
+ * It is possible to manually assign a type encoder for each column per PG::TypeMapByColumn,
+ * or to make use of PG::BasicTypeMapBasedOnResult to assign them based on the table OIDs.
+ *
+ * See also PG::TextDecoder::CopyRow for the decoding direction with
+ * PG::Connection#get_copy_data .
  */
 static int
-pg_text_enc_copy_row(t_pg_coder *conv, VALUE value, char *out, VALUE *intermediate)
+pg_text_enc_copy_row(t_pg_coder *conv, VALUE value, char *out, VALUE *intermediate, int enc_idx)
 {
 	t_pg_copycoder *this = (t_pg_copycoder *)conv;
 	t_pg_coder_enc_func enc_func;
@@ -184,6 +190,7 @@ pg_text_enc_copy_row(t_pg_coder *conv, VALUE value, char *out, VALUE *intermedia
 
 	/* Allocate a new string with embedded capacity and realloc exponential when needed. */
 	PG_RB_STR_NEW( *intermediate, current_out, end_capa_ptr );
+	PG_ENCODING_SET_NOCHECK(*intermediate, enc_idx);
 
 	for( i=0; i<RARRAY_LEN(value); i++){
 		char *ptr1;
@@ -211,7 +218,7 @@ pg_text_enc_copy_row(t_pg_coder *conv, VALUE value, char *out, VALUE *intermedia
 				enc_func = pg_coder_enc_func(p_elem_coder);
 
 				/* 1st pass for retiving the required memory space */
-				strlen = enc_func(p_elem_coder, entry, NULL, &subint);
+				strlen = enc_func(p_elem_coder, entry, NULL, &subint, enc_idx);
 
 				if( strlen == -1 ){
 					/* we can directly use String value in subint */
@@ -234,7 +241,7 @@ pg_text_enc_copy_row(t_pg_coder *conv, VALUE value, char *out, VALUE *intermedia
 					PG_RB_STR_ENSURE_CAPA( *intermediate, strlen * 2, current_out, end_capa_ptr );
 
 					/* Place the unescaped string at current output position. */
-					strlen = enc_func(p_elem_coder, entry, current_out, &subint);
+					strlen = enc_func(p_elem_coder, entry, current_out, &subint, enc_idx);
 
 					ptr1 = current_out;
 					ptr2 = current_out + strlen;
@@ -301,15 +308,38 @@ GetDecimalFromHex(char hex)
  * strings by PG::TextDecoder::String.
  *
  * Example with default type map ( TypeMapAllStrings ):
+ *   conn.exec("CREATE TABLE my_table AS VALUES('astring', 7, FALSE), ('string2', 42, TRUE) ")
+ *
  *   deco = PG::TextDecoder::CopyRow.new
  *   conn.copy_data "COPY my_table TO STDOUT", deco do
  *     while row=conn.get_copy_data
  *       p row
  *     end
  *   end
- * This prints all rows of +my_table+ to stdout:
+ * This prints all rows of +my_table+ :
  *   ["astring", "7", "f"]
  *   ["string2", "42", "t"]
+ *
+ * Example with column based type map:
+ *   tm = PG::TypeMapByColumn.new( [
+ *     PG::TextDecoder::String.new,
+ *     PG::TextDecoder::Integer.new,
+ *     PG::TextDecoder::Boolean.new] )
+ *   deco = PG::TextDecoder::CopyRow.new( type_map: tm )
+ *   conn.copy_data "COPY my_table TO STDOUT", deco do
+ *     while row=conn.get_copy_data
+ *       p row
+ *     end
+ *   end
+ * This prints the rows with type casted columns:
+ *   ["astring", 7, false]
+ *   ["string2", 42, true]
+ *
+ * Instead of manually assigning a type decoder for each column, PG::BasicTypeMapForResults
+ * can be used to assign them based on the table OIDs.
+ *
+ * See also PG::TextEncoder::CopyRow for the encoding direction with
+ * PG::Connection#put_copy_data .
  */
 /*
  * Parse the current line into separate attributes (fields),
@@ -324,7 +354,7 @@ GetDecimalFromHex(char hex)
  * src/backend/commands/copy.c
  */
 static VALUE
-pg_text_dec_copy_row(t_pg_coder *conv, char *input_line, int len, int _tuple, int _field, int enc_idx)
+pg_text_dec_copy_row(t_pg_coder *conv, const char *input_line, int len, int _tuple, int _field, int enc_idx)
 {
 	t_pg_copycoder *this = (t_pg_copycoder *)conv;
 
@@ -338,8 +368,8 @@ pg_text_dec_copy_row(t_pg_coder *conv, char *input_line, int len, int _tuple, in
 	int fieldno;
 	int expected_fields;
 	char *output_ptr;
-	char *cur_ptr;
-	char *line_end_ptr;
+	const char *cur_ptr;
+	const char *line_end_ptr;
 	char *end_capa_ptr;
 	t_typemap *p_typemap;
 
@@ -362,8 +392,8 @@ pg_text_dec_copy_row(t_pg_coder *conv, char *input_line, int len, int _tuple, in
 	for (;;)
 	{
 		int found_delim = 0;
-		char *start_ptr;
-		char *end_ptr;
+		const char *start_ptr;
+		const char *end_ptr;
 		int input_len;
 
 		/* Remember start of field on input side */
@@ -555,7 +585,12 @@ init_pg_copycoder()
 	/* rb_mPG_TextEncoder = rb_define_module_under( rb_mPG, "TextEncoder" ); */
 	/* dummy = rb_define_class_under( rb_mPG_TextEncoder, "CopyRow", rb_cPG_CopyEncoder ); */
 	pg_define_coder( "CopyRow", pg_text_enc_copy_row, rb_cPG_CopyEncoder, rb_mPG_TextEncoder );
+	rb_include_module( rb_cPG_CopyEncoder, rb_mPG_BinaryFormatting );
+
 	/* rb_mPG_TextDecoder = rb_define_module_under( rb_mPG, "TextDecoder" ); */
 	/* dummy = rb_define_class_under( rb_mPG_TextDecoder, "CopyRow", rb_cPG_CopyDecoder ); */
 	pg_define_coder( "CopyRow", pg_text_dec_copy_row, rb_cPG_CopyDecoder, rb_mPG_TextDecoder );
+	/* Although CopyRow is a text decoder, data can contain zero bytes and are not zero terminated.
+	 * They are handled like binaries. So format is set to 1 (binary). */
+	rb_include_module( rb_cPG_CopyDecoder, rb_mPG_BinaryFormatting );
 }