pg 0.18.3 → 1.4.3

data/ext/pg_copy_coder.c

@@ -21,17 +21,47 @@ typedef struct {
 
 
 static void
-pg_copycoder_mark( t_pg_copycoder *this )
+pg_copycoder_mark( void *_this )
 {
-	rb_gc_mark(this->typemap);
-	rb_gc_mark(this->null_string);
+	t_pg_copycoder *this = (t_pg_copycoder *)_this;
+	rb_gc_mark_movable(this->typemap);
+	rb_gc_mark_movable(this->null_string);
 }
 
+static size_t
+pg_copycoder_memsize( const void *_this )
+{
+	const t_pg_copycoder *this = (const t_pg_copycoder *)_this;
+	return sizeof(*this);
+}
+
+static void
+pg_copycoder_compact( void *_this )
+{
+	t_pg_copycoder *this = (t_pg_copycoder *)_this;
+	pg_coder_compact(&this->comp);
+	pg_gc_location(this->typemap);
+	pg_gc_location(this->null_string);
+}
+
+static const rb_data_type_t pg_copycoder_type = {
+	"PG::CopyCoder",
+	{
+		pg_copycoder_mark,
+		RUBY_TYPED_DEFAULT_FREE,
+		pg_copycoder_memsize,
+		pg_compact_callback(pg_copycoder_compact),
+	},
+	&pg_coder_type,
+	0,
+	RUBY_TYPED_FREE_IMMEDIATELY,
+};
+
 static VALUE
 pg_copycoder_encoder_allocate( VALUE klass )
 {
 	t_pg_copycoder *this;
-	VALUE self = Data_Make_Struct( klass, t_pg_copycoder, pg_copycoder_mark, -1, this );
+	VALUE self = TypedData_Make_Struct( klass, t_pg_copycoder, &pg_copycoder_type, this );
 	pg_coder_init_encoder( self );
 	this->typemap = pg_typemap_all_strings;
 	this->delimiter = '\t';
@@ -43,7 +73,7 @@ static VALUE
 pg_copycoder_decoder_allocate( VALUE klass )
 {
 	t_pg_copycoder *this;
-	VALUE self = Data_Make_Struct( klass, t_pg_copycoder, pg_copycoder_mark, -1, this );
+	VALUE self = TypedData_Make_Struct( klass, t_pg_copycoder, &pg_copycoder_type, this );
 	pg_coder_init_decoder( self );
 	this->typemap = pg_typemap_all_strings;
 	this->delimiter = '\t';
@@ -62,7 +92,7 @@ pg_copycoder_decoder_allocate( VALUE klass )
 static VALUE
 pg_copycoder_delimiter_set(VALUE self, VALUE delimiter)
 {
-	t_pg_copycoder *this = DATA_PTR(self);
+	t_pg_copycoder *this = RTYPEDDATA_DATA(self);
 	StringValue(delimiter);
 	if(RSTRING_LEN(delimiter) != 1)
 		rb_raise( rb_eArgError, "delimiter size must be one byte");
@@ -79,7 +109,7 @@ pg_copycoder_delimiter_set(VALUE self, VALUE delimiter)
 static VALUE
 pg_copycoder_delimiter_get(VALUE self)
 {
-	t_pg_copycoder *this = DATA_PTR(self);
+	t_pg_copycoder *this = RTYPEDDATA_DATA(self);
 	return rb_str_new(&this->delimiter, 1);
 }
 
@@ -92,7 +122,7 @@ pg_copycoder_delimiter_get(VALUE self)
 static VALUE
 pg_copycoder_null_string_set(VALUE self, VALUE null_string)
 {
-	t_pg_copycoder *this = DATA_PTR(self);
+	t_pg_copycoder *this = RTYPEDDATA_DATA(self);
 	StringValue(null_string);
 	this->null_string = null_string;
 	return null_string;
@@ -104,7 +134,7 @@ pg_copycoder_null_string_set(VALUE self, VALUE null_string)
 static VALUE
 pg_copycoder_null_string_get(VALUE self)
 {
-	t_pg_copycoder *this = DATA_PTR(self);
+	t_pg_copycoder *this = RTYPEDDATA_DATA(self);
 	return this->null_string;
 }
 
@@ -112,16 +142,17 @@ pg_copycoder_null_string_get(VALUE self)
  * call-seq:
  *    coder.type_map = map
  *
+ * Defines how single columns are encoded or decoded.
  * +map+ must be a kind of PG::TypeMap .
  *
  * Defaults to a PG::TypeMapAllStrings , so that PG::TextEncoder::String respectively
- * PG::TextDecoder::String is used for encoding/decoding of all columns.
+ * PG::TextDecoder::String is used for encoding/decoding of each column.
  *
  */
 static VALUE
 pg_copycoder_type_map_set(VALUE self, VALUE type_map)
 {
-	t_pg_copycoder *this = DATA_PTR( self );
+	t_pg_copycoder *this = RTYPEDDATA_DATA( self );
 
 	if ( !rb_obj_is_kind_of(type_map, rb_cTypeMap) ){
 		rb_raise( rb_eTypeError, "wrong elements type %s (expected some kind of PG::TypeMap)",
@@ -136,11 +167,12 @@ pg_copycoder_type_map_set(VALUE self, VALUE type_map)
  * call-seq:
  *    coder.type_map -> PG::TypeMap
  *
+ * The PG::TypeMap that will be used for encoding and decoding of columns.
  */
 static VALUE
 pg_copycoder_type_map_get(VALUE self)
 {
-	t_pg_copycoder *this = DATA_PTR( self );
+	t_pg_copycoder *this = RTYPEDDATA_DATA( self );
 
 	return this->typemap;
 }
@@ -167,9 +199,15 @@ 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;
@@ -179,11 +217,12 @@ pg_text_enc_copy_row(t_pg_coder *conv, VALUE value, char *out, VALUE *intermedia
 	char *current_out;
 	char *end_capa_ptr;
 
-	p_typemap = DATA_PTR( this->typemap );
+	p_typemap = RTYPEDDATA_DATA( this->typemap );
 	p_typemap->funcs.fit_to_query( this->typemap, value );
 
 	/* 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;
@@ -211,11 +250,11 @@ 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 */
-					strlen = RSTRING_LEN(subint);
+					strlen = RSTRING_LENINT(subint);
 
 					/* size of string assuming the worst case, that every character must be escaped. */
 					PG_RB_STR_ENSURE_CAPA( *intermediate, strlen * 2, current_out, end_capa_ptr );
@@ -234,7 +273,7 @@ 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 +340,38 @@ 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 +386,7 @@ 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;
 
@@ -338,12 +400,12 @@ 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;
 
-	p_typemap = DATA_PTR( this->typemap );
+	p_typemap = RTYPEDDATA_DATA( this->typemap );
 	expected_fields = p_typemap->funcs.fit_to_copy_get( this->typemap );
 
 	/* The received input string will probably have this->nfields fields. */
@@ -351,7 +413,7 @@ pg_text_dec_copy_row(t_pg_coder *conv, char *input_line, int len, int _tuple, in
 
 	/* Allocate a new string with embedded capacity and realloc later with
 	 * exponential growing size when needed. */
-	PG_RB_TAINTED_STR_NEW( field_str, output_ptr, end_capa_ptr );
+	PG_RB_STR_NEW( field_str, output_ptr, end_capa_ptr );
 
 	/* set pointer variables for loop */
 	cur_ptr = input_line;
@@ -362,9 +424,9 @@ 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;
-		int input_len;
+		const char *start_ptr;
+		const char *end_ptr;
+		long input_len;
 
 		/* Remember start of field on input side */
 		start_ptr = cur_ptr;
@@ -513,7 +575,7 @@ pg_text_dec_copy_row(t_pg_coder *conv, char *input_line, int len, int _tuple, in
 			if( field_value == field_str ){
 				/* Our output string will be send to the user, so we can not reuse
 				 * it for the next field. */
-				PG_RB_TAINTED_STR_NEW( field_str, output_ptr, end_capa_ptr );
+				PG_RB_STR_NEW( field_str, output_ptr, end_capa_ptr );
 			}
 		}
 		/* Reset the pointer to the start of the output/buffer string. */
@@ -555,7 +617,12 @@ 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 );
 }