do_postgres 0.2.4 → 0.9.2

data/README

@@ -1,4 +1,4 @@
-do_sqlite3
-=================
+do_postgres
+===========
 
-A plugin for the Merb framework that provides ....
+A PostgreSQL driver for DataObjects

data/Rakefile

@@ -1,36 +1,57 @@
 require 'rubygems'
+require 'rake/clean'
 require 'rake/gempackagetask'
+require 'spec/rake/spectask'
+require 'pathname'
+require Pathname(__FILE__).dirname.expand_path.parent + 'tasks/ext_helper'
 
-PLUGIN = "do_postgres"
-NAME = "do_postgres"
-VERSION = "0.2.4"
-AUTHOR = "Yehuda Katz"
-EMAIL = "wycats@gmail.com"
-HOMEPAGE = "http://dataobjects.devjavu.com"
-SUMMARY = "A DataObject.rb driver for postgres"
+# House-keeping
+CLEAN.include '**/*.o', '**/*.so', '**/*.bundle', '**/*.a',
+  '**/*.log', '{ext,lib}/*.{bundle,so,obj,pdb,lib,def,exp}',
+  'ext/Makefile'
+
+JRUBY   = (RUBY_PLATFORM =~ /java/) rescue nil
+WINDOWS = (RUBY_PLATFORM =~ /mswin|mingw|cygwin/) rescue nil
+          # don't use SUDO with JRuby, for the moment, although this behaviour
+          # is not entirely correct.
+SUDO    = (WINDOWS || JRUBY) ? '' : ('sudo' unless ENV['SUDOLESS'])
 
 spec = Gem::Specification.new do |s|
-  s.name = NAME
-  s.version = VERSION
-  s.platform = Gem::Platform::RUBY
-  s.has_rdoc = true
-  s.extra_rdoc_files = ["README", "LICENSE", 'TODO']
-  s.summary = SUMMARY
-  s.description = s.summary
-  s.author = AUTHOR
-  s.email = EMAIL
-  s.homepage = HOMEPAGE
-  s.add_dependency('data_objects', ["<=0.2.0"])
-  s.require_path = 'lib'
-  s.autorequire = PLUGIN
-  s.extensions = ["ext/extconf.rb"]
-  s.files = %w(LICENSE README Rakefile TODO) + Dir.glob("{lib,specs,ext}/**/*").reject {|x| x =~ /\.(o|bundle)$/ }
+  s.name              = 'do_postgres'
+  s.version           = '0.9.2'
+  s.platform          = Gem::Platform::RUBY
+  s.has_rdoc          = true
+  s.extra_rdoc_files  = %w[ README LICENSE TODO ]
+  s.summary           = 'A DataObject.rb driver for PostgreSQL'
+  s.description       = s.summary
+  s.author            = 'Yehuda Katz'
+  s.email             = 'wycats@gmail.com'
+  s.homepage          = 'http://rubyforge.org/projects/dorb'
+  s.rubyforge_project = 'dorb'
+  s.require_path      = 'lib'
+  s.extensions        = %w[ ext/extconf.rb ]
+  s.files             = FileList[ '{ext,lib,spec}/**/*.{c,h,rb}', 'Rakefile', *s.extra_rdoc_files ]
+  s.add_dependency('data_objects', "=#{s.version}")
 end
 
 Rake::GemPackageTask.new(spec) do |pkg|
   pkg.gem_spec = spec
 end
 
-task :install => [:package] do
-  sh %{sudo gem install pkg/#{NAME}-#{VERSION}}, :verbose => false
-end
+# Use of ext_helper to properly setup compile tasks and native gem generation
+setup_extension "#{spec.name}_ext", spec
+
+task :install => [ :package ] do
+  sh %{#{SUDO} gem install --local pkg/#{spec.name}-#{spec.version} --no-update-sources}, :verbose => false
+end
+
+desc "Uninstall #{spec.name} #{spec.version} (default ruby)"
+task :uninstall => [ :clobber ] do
+  sh "#{SUDO} gem uninstall #{spec.name} -v#{spec.version} -I -x", :verbose => false
+end
+
+desc 'Run specifications'
+Spec::Rake::SpecTask.new(:spec => [ :compile ]) do |t|
+  t.spec_opts << '--options' << 'spec/spec.opts' if File.exists?('spec/spec.opts')
+  t.spec_files = Pathname.glob(Pathname.new(__FILE__).dirname + 'spec/**/*_spec.rb')
+end

data/TODO

@@ -0,0 +1,4 @@
+TODO
+====
+
+  * Add JDBC-based version of this driver.

data/ext/do_postgres_ext.c

@@ -0,0 +1,614 @@
+#include <ruby.h>
+#include <version.h>
+#include <string.h>
+#include <math.h>
+#include <time.h>
+#include <locale.h>
+#include <libpq-fe.h>
+#include "type-oids.h"
+
+#define ID_CONST_GET rb_intern("const_get")
+#define ID_PATH rb_intern("path")
+#define ID_NEW rb_intern("new")
+#define ID_ESCAPE rb_intern("escape_sql")
+
+#define RUBY_STRING(char_ptr) rb_str_new2(char_ptr)
+#define TAINTED_STRING(name) rb_tainted_str_new2(name)
+#define CONST_GET(scope, constant) (rb_funcall(scope, ID_CONST_GET, 1, rb_str_new2(constant)))
+#define POSTGRES_CLASS(klass, parent) (rb_define_class_under(mPostgres, klass, parent))
+#define DEBUG(value) data_objects_debug(value)
+
+#ifdef _WIN32
+#define do_int64 signed __int64
+#else
+#define do_int64 signed long long int
+#endif
+
+// To store rb_intern values
+static ID ID_NEW_DATE;
+static ID ID_LOGGER;
+static ID ID_DEBUG;
+static ID ID_LEVEL;
+
+static VALUE mDO;
+static VALUE cDO_Quoting;
+static VALUE cDO_Connection;
+static VALUE cDO_Command;
+static VALUE cDO_Result;
+static VALUE cDO_Reader;
+
+static VALUE rb_cDate;
+static VALUE rb_cDateTime;
+static VALUE rb_cRational;
+static VALUE rb_cBigDecimal;
+
+static VALUE mPostgres;
+static VALUE cConnection;
+static VALUE cCommand;
+static VALUE cResult;
+static VALUE cReader;
+
+static VALUE ePostgresError;
+
+static void data_objects_debug(VALUE string) {
+	VALUE logger = rb_funcall(mPostgres, ID_LOGGER, 0);
+	int log_level = NUM2INT(rb_funcall(logger, ID_LEVEL, 0));
+	
+	if (0 == log_level) {
+		rb_funcall(logger, ID_DEBUG, 1, string);
+	}
+}
+
+/* ====== Time/Date Parsing Helper Functions ====== */
+static void reduce( do_int64 *numerator, do_int64 *denominator ) {
+	do_int64 a, b, c;
+	a = *numerator;
+	b = *denominator;
+	while ( a != 0 ) {
+		c = a; a = b % a; b = c;
+	}
+	*numerator = *numerator / b;
+	*denominator = *denominator / b;
+}
+
+// Generate the date integer which Date.civil_to_jd returns
+static int jd_from_date(int year, int month, int day) {
+	int a, b;
+	if ( month <= 2 ) {
+		year -= 1;
+		month += 12;
+	}
+	a = year / 100;
+	b = 2 - a + (a / 4);
+	return floor(365.25 * (year + 4716)) + floor(30.6001 * (month + 1)) + day + b - 1524;
+}
+
+static VALUE parse_date(const char *date) {
+	int year, month, day;
+	int jd, ajd;
+	VALUE rational;
+	
+	sscanf(date, "%4d-%2d-%2d", &year, &month, &day);
+	
+	jd = jd_from_date(year, month, day);
+	
+	// Math from Date.jd_to_ajd
+	ajd = jd * 2 - 1;
+	rational = rb_funcall(rb_cRational, rb_intern("new!"), 2, INT2NUM(ajd), INT2NUM(2));
+	
+	return rb_funcall(rb_cDate, ID_NEW_DATE, 3, rational, INT2NUM(0), INT2NUM(2299161));
+}
+
+// Creates a Rational for use as a Timezone offset to be passed to DateTime.new!
+static VALUE seconds_to_offset(do_int64 num) {
+	do_int64 den = 86400;	
+	reduce(&num, &den);
+	return rb_funcall(rb_cRational, rb_intern("new!"), 2, rb_ll2inum(num), rb_ll2inum(den));	
+}
+
+static VALUE timezone_to_offset(int hour_offset, int minute_offset) {
+	do_int64 seconds = 0;
+
+	seconds += hour_offset * 3600;
+	seconds += minute_offset * 60;
+
+	return seconds_to_offset(seconds);
+}
+
+static VALUE parse_date_time(const char *date) {
+	VALUE ajd, offset;
+
+	int year, month, day, hour, min, sec, usec, hour_offset, minute_offset;
+	int jd;
+	do_int64 num, den;
+	
+	long int gmt_offset;
+	int is_dst;
+	
+	time_t rawtime;
+	struct tm * timeinfo;
+
+	int tokens_read, max_tokens;
+	
+	if (0 != strchr(date, '.')) {
+		// This is a datetime with sub-second precision
+		tokens_read = sscanf(date, "%4d-%2d-%2d %2d:%2d:%2d.%d%3d:%2d", &year, &month, &day, &hour, &min, &sec, &usec, &hour_offset, &minute_offset);
+		max_tokens = 9;
+	} else {
+		// This is a datetime second precision
+		tokens_read = sscanf(date, "%4d-%2d-%2d %2d:%2d:%2d%3d:%2d", &year, &month, &day, &hour, &min, &sec, &hour_offset, &minute_offset);
+		max_tokens = 8;
+	}
+	
+	if (max_tokens == tokens_read) {
+		// We read the Date, Time, and Timezone info
+		minute_offset *= hour_offset < 0 ? -1 : 1;
+	} else if ((max_tokens - 1) == tokens_read) {
+		// We read the Date and Time, but no Minute Offset
+		minute_offset = 0;
+	} else if (tokens_read == 3) {
+		return parse_date(date);
+	} else if (tokens_read >= (max_tokens - 3)) {
+		// We read the Date and Time, default to the current locale's offset
+		
+		// Get localtime
+		time(&rawtime);
+		timeinfo = localtime(&rawtime);
+		
+		is_dst = timeinfo->tm_isdst * 3600;
+
+		// Reset to GM Time
+		timeinfo = gmtime(&rawtime);
+
+		gmt_offset = mktime(timeinfo) - rawtime;
+
+		if ( is_dst > 0 )
+			gmt_offset -= is_dst;
+
+		hour_offset = -(gmt_offset / 3600);
+		minute_offset = -(gmt_offset % 3600 / 60);
+
+	} else {
+		// Something went terribly wrong
+		rb_raise(ePostgresError, "Couldn't parse date: %s", date);
+	}
+
+	jd = jd_from_date(year, month, day);
+
+	// Generate ajd with fractional days for the time
+	// Extracted from Date#jd_to_ajd, Date#day_fraction_to_time, and Rational#+ and #-
+	num = (hour * 1440) + (min * 24);
+
+	// Modify the numerator so when we apply the timezone everything works out
+	num -= (hour_offset * 1440) + (minute_offset * 24);
+	
+	den = (24 * 1440);
+	reduce(&num, &den);
+
+	num = (num * 86400) + (sec * den);
+	den = den * 86400;
+	reduce(&num, &den);
+
+	num = (jd * den) + num;
+
+	num = num * 2;
+	num = num - den;
+	den = den * 2;
+
+	reduce(&num, &den);
+
+	ajd = rb_funcall(rb_cRational, rb_intern("new!"), 2, rb_ull2inum(num), rb_ull2inum(den));
+	offset = timezone_to_offset(hour_offset, minute_offset);
+	
+	return rb_funcall(rb_cDateTime, ID_NEW_DATE, 3, ajd, offset, INT2NUM(2299161));
+}
+
+static VALUE parse_time(char *date) {
+
+	int year, month, day, hour, min, sec, usec;
+	char subsec[7];
+
+	if (0 != strchr(date, '.')) {
+		// right padding usec with 0. e.g. '012' will become 12000 microsecond, since Time#local use microsecond
+	  sscanf(date, "%4d-%2d-%2d %2d:%2d:%2d.%s", &year, &month, &day, &hour, &min, &sec, subsec);
+      usec = atoi(subsec);
+      usec *= pow(10, (6 - strlen(subsec)));
+	} else {
+		sscanf(date, "%4d-%2d-%2d %2d:%2d:%2d", &year, &month, &day, &hour, &min, &sec);
+		usec = 0;
+	}
+
+	return rb_funcall(rb_cTime, rb_intern("local"), 7, INT2NUM(year), INT2NUM(month), INT2NUM(day), INT2NUM(hour), INT2NUM(min), INT2NUM(sec), INT2NUM(usec));
+}
+
+/* ===== Typecasting Functions ===== */
+
+static VALUE infer_ruby_type(Oid type) {
+	char *ruby_type = "String";
+	switch(type) {
+		case INT2OID:
+		case INT4OID:
+		case INT8OID: {
+			ruby_type = "Fixnum";
+			break;
+		}
+		case FLOAT4OID:
+		case FLOAT8OID: {
+			ruby_type = "Float";
+			break;
+		}
+		case BOOLOID: {
+			ruby_type = "TrueClass";
+			break;
+		}
+		case TIMESTAMPOID: {
+			ruby_type = "DateTime";
+			break;
+		}
+		// case TIMESTAMPTZOID
+		// case TIMETZOID
+		case TIMEOID: {
+			ruby_type = "Time";
+			break;
+		}
+		case DATEOID: {
+			ruby_type = "Date";
+			break;
+		}
+	}
+	return rb_str_new2(ruby_type);
+}
+
+static VALUE typecast(char *value, char *type) {
+	
+	if ( strcmp(type, "Class") == 0) {
+	  return rb_funcall(mDO, rb_intern("find_const"), 1, TAINTED_STRING(value));
+	} else if ( strcmp(type, "Integer") == 0 || strcmp(type, "Fixnum") == 0 || strcmp(type, "Bignum") == 0 ) {
+		return rb_cstr2inum(value, 10);
+	} else if ( strcmp(type, "Float") == 0 ) {
+		return rb_float_new(rb_cstr_to_dbl(value, Qfalse));
+	} else if (0 == strcmp("BigDecimal", type) ) {
+		return rb_funcall(rb_cBigDecimal, ID_NEW, 1, TAINTED_STRING(value));
+	} else if ( strcmp(type, "TrueClass") == 0 ) {
+		return *value == 't' ? Qtrue : Qfalse;
+	} else if ( strcmp(type, "Date") == 0 ) {
+		return parse_date(value);
+	} else if ( strcmp(type, "DateTime") == 0 ) {
+		return parse_date_time(value);
+	} else if ( strcmp(type, "Time") == 0 ) {
+		return parse_time(value);
+	} else {
+		return TAINTED_STRING(value);
+	}
+
+}
+
+/* ====== Public API ======= */
+
+static VALUE cConnection_initialize(VALUE self, VALUE uri) {
+	VALUE r_host, r_user, r_password, r_path, r_port;
+	char *host = NULL, *user = NULL, *password = NULL, *path;
+	char *database = "", *port = "5432";
+
+	PGconn *db;
+
+	r_host = rb_funcall(uri, rb_intern("host"), 0);
+	if ( Qnil != r_host && "localhost" != StringValuePtr(r_host) ) {
+		host = StringValuePtr(r_host);
+	}
+
+	r_user = rb_funcall(uri, rb_intern("user"), 0);
+	if (Qnil != r_user) {
+		user = StringValuePtr(r_user);
+	}
+
+	r_password = rb_funcall(uri, rb_intern("password"), 0);
+	if (Qnil != r_password) {
+		password = StringValuePtr(r_password);
+	}
+
+	r_path = rb_funcall(uri, rb_intern("path"), 0);
+	path = StringValuePtr(r_path);
+	if (Qnil != r_path) {
+		database = strtok(path, "/");
+	}
+
+	if (NULL == database || 0 == strlen(database)) {
+		rb_raise(ePostgresError, "Database must be specified");
+	}
+
+	r_port = rb_funcall(uri, rb_intern("port"), 0);
+	if (Qnil != r_port) {
+		r_port = rb_funcall(r_port, rb_intern("to_s"), 0);
+		port = StringValuePtr(r_port);
+	}
+
+	db = PQsetdbLogin(
+		host, 
+		port, 
+		NULL, 
+		NULL, 
+		database, 
+		user, 
+		password
+	);
+
+	if ( PQstatus(db) == CONNECTION_BAD ) {
+		rb_raise(ePostgresError, PQerrorMessage(db));
+	}
+
+	rb_iv_set(self, "@uri", uri);
+	rb_iv_set(self, "@connection", Data_Wrap_Struct(rb_cObject, 0, 0, db));
+
+	return Qtrue;
+}
+
+static VALUE cConnection_dispose(VALUE self) {
+	PGconn *db = DATA_PTR(rb_iv_get(self, "@connection"));
+	PQfinish(db);
+	return Qtrue;
+}
+
+static VALUE cCommand_set_types(VALUE self, VALUE array) {
+	rb_iv_set(self, "@field_types", array);
+	return array;
+}
+
+static VALUE build_query_from_args(VALUE klass, int count, VALUE *args[]) {
+	VALUE query = rb_iv_get(klass, "@text");
+	if ( count > 0 ) {
+		int i;
+		VALUE array = rb_ary_new();
+		for ( i = 0; i < count; i++) {
+			rb_ary_push(array, (VALUE)args[i]);
+		}
+		query = rb_funcall(klass, ID_ESCAPE, 1, array);
+	}
+	return query;
+}
+
+static VALUE cCommand_quote_string(VALUE self, VALUE string) {
+	PGconn *db = DATA_PTR(rb_iv_get(rb_iv_get(self, "@connection"), "@connection"));
+
+	size_t length;
+	const char *source = StringValuePtr(string);
+	char *escaped;
+	int quoted_length = 0;
+	VALUE result;
+	
+	length = strlen(source);
+
+	// Allocate space for the escaped version of 'string'
+	// http://www.postgresql.org/docs/8.3/static/libpq-exec.html#LIBPQ-EXEC-ESCAPE-STRING
+	escaped = (char *)calloc(strlen(source) * 2 + 3, sizeof(char));
+
+	// Escape 'source' using the current charset in use on the conection 'db'
+	quoted_length = PQescapeStringConn(db, escaped + 1, source, length, NULL);
+
+	// Wrap the escaped string in single-quotes, this is DO's convention
+	escaped[quoted_length + 1] = escaped[0] = '\'';
+
+	result = rb_str_new(escaped, quoted_length + 2);
+	free(escaped);
+	return result;
+}
+
+static VALUE cCommand_execute_non_query(int argc, VALUE *argv[], VALUE self) {
+	PGconn *db = DATA_PTR(rb_iv_get(rb_iv_get(self, "@connection"), "@connection"));
+	PGresult *response;
+	int status;
+	
+	int affected_rows;
+	int insert_id;
+	
+	VALUE query = build_query_from_args(self, argc, argv);
+	data_objects_debug(query);
+	
+	response = PQexec(db, StringValuePtr(query));
+	
+	status = PQresultStatus(response);
+	
+	if ( status == PGRES_TUPLES_OK ) {
+		insert_id = atoi(PQgetvalue(response, 0, 0));
+		affected_rows = 1;
+	}
+	else if ( status == PGRES_COMMAND_OK ) {
+		insert_id = 0;
+		affected_rows = atoi(PQcmdTuples(response));
+	}
+	else {
+		char *message = PQresultErrorMessage(response);
+		PQclear(response);
+		rb_raise(ePostgresError, message);
+	}
+	
+	PQclear(response);
+	
+	return rb_funcall(cResult, ID_NEW, 3, self, INT2NUM(affected_rows), INT2NUM(insert_id));
+}
+
+static VALUE cCommand_execute_reader(int argc, VALUE *argv[], VALUE self) {
+	VALUE reader, query;
+	VALUE field_names, field_types;
+
+	int i;
+	int field_count;
+	int infer_types = 0;
+
+	PGconn *db = DATA_PTR(rb_iv_get(rb_iv_get(self, "@connection"), "@connection"));
+	PGresult *response;
+
+	query = build_query_from_args(self, argc, argv);
+	data_objects_debug(query);
+
+	response = PQexec(db, StringValuePtr(query));
+
+	if ( PQresultStatus(response) != PGRES_TUPLES_OK ) {
+		char *message = PQresultErrorMessage(response);
+		PQclear(response);
+		rb_raise(ePostgresError, message);
+	}
+
+	field_count = PQnfields(response);
+
+	reader = rb_funcall(cReader, ID_NEW, 0);
+	rb_iv_set(reader, "@reader", Data_Wrap_Struct(rb_cObject, 0, 0, response));
+	rb_iv_set(reader, "@field_count", INT2NUM(field_count));
+	rb_iv_set(reader, "@row_count", INT2NUM(PQntuples(response)));
+
+	field_names = rb_ary_new();
+	field_types = rb_iv_get(self, "@field_types");
+
+	if ( field_types == Qnil || RARRAY(field_types)->len == 0 ) {
+		field_types = rb_ary_new();
+		infer_types = 1;
+	}
+
+	for ( i = 0; i < field_count; i++ ) {
+		rb_ary_push(field_names, rb_str_new2(PQfname(response, i)));
+		if ( infer_types == 1 ) {
+			rb_ary_push(field_types, infer_ruby_type(PQftype(response, i)));
+		}
+	}
+
+	rb_iv_set(reader, "@position", INT2NUM(0));
+	rb_iv_set(reader, "@fields", field_names);
+	rb_iv_set(reader, "@field_types", field_types);
+
+	return reader;
+}
+
+static VALUE cReader_close(VALUE self) {
+	VALUE reader_container = rb_iv_get(self, "@reader");
+
+	PGresult *reader;
+
+	if (Qnil == reader_container)
+		return Qfalse;
+
+	reader = DATA_PTR(reader_container);
+
+	if (NULL == reader)
+		return Qfalse;
+
+	PQclear(reader);
+	rb_iv_set(self, "@reader", Qnil);
+	return Qtrue;
+}
+
+static VALUE cReader_next(VALUE self) {
+	PGresult *reader = DATA_PTR(rb_iv_get(self, "@reader"));
+
+	int field_count;
+	int row_count;
+	int i;
+	int position;
+
+	char *type = "";
+
+	VALUE array = rb_ary_new();
+	VALUE field_types, ruby_type;
+	VALUE value;
+
+	row_count = NUM2INT(rb_iv_get(self, "@row_count"));
+	field_count = NUM2INT(rb_iv_get(self, "@field_count"));
+	field_types = rb_iv_get(self, "@field_types");
+	position = NUM2INT(rb_iv_get(self, "@position"));
+
+	if ( position > (row_count-1) ) {
+		return Qnil;
+	}
+
+	for ( i = 0; i < field_count; i++ ) {
+		ruby_type = RARRAY(field_types)->ptr[i];
+		
+		if ( TYPE(ruby_type) == T_STRING ) {
+			type = RSTRING(ruby_type)->ptr;
+		}
+		else {
+			type = rb_class2name(ruby_type);
+		}
+		
+		// Always return nil if the value returned from Postgres is null
+		if (!PQgetisnull(reader, position, i)) {
+			value = typecast(PQgetvalue(reader, position, i), type);
+		} else {
+			value = Qnil;
+		}
+
+		rb_ary_push(array, value);
+	}
+
+	rb_iv_set(self, "@values", array);
+	rb_iv_set(self, "@position", INT2NUM(position+1));
+
+	return Qtrue;
+}
+
+static VALUE cReader_values(VALUE self) {
+	
+	int position = rb_iv_get(self, "@position");
+	int row_count = NUM2INT(rb_iv_get(self, "@row_count"));
+	
+	if ( position == Qnil || NUM2INT(position) > row_count ) {
+		rb_raise(ePostgresError, "Reader not initialized");
+	}
+	else {
+		return rb_iv_get(self, "@values");
+	}
+}
+
+static VALUE cReader_fields(VALUE self) {
+	return rb_iv_get(self, "@fields");
+}
+
+void Init_do_postgres_ext() {
+	rb_require("rubygems");
+	rb_require("date");
+	
+	// Get references classes needed for Date/Time parsing 
+	rb_cDate = CONST_GET(rb_mKernel, "Date");
+	rb_cDateTime = CONST_GET(rb_mKernel, "DateTime");
+	rb_cTime = CONST_GET(rb_mKernel, "Time");
+	rb_cRational = CONST_GET(rb_mKernel, "Rational");
+	rb_cBigDecimal = CONST_GET(rb_mKernel, "BigDecimal");
+	
+	rb_funcall(rb_mKernel, rb_intern("require"), 1, rb_str_new2("data_objects"));
+	
+	ID_NEW_DATE = RUBY_VERSION_CODE < 186 ? rb_intern("new0") : rb_intern("new!");
+	ID_LOGGER = rb_intern("logger");
+	ID_DEBUG = rb_intern("debug");
+	ID_LEVEL = rb_intern("level");
+
+	// Get references to the DataObjects module and its classes
+	mDO = CONST_GET(rb_mKernel, "DataObjects");
+	cDO_Quoting = CONST_GET(mDO, "Quoting");
+	cDO_Connection = CONST_GET(mDO, "Connection");
+	cDO_Command = CONST_GET(mDO, "Command");
+	cDO_Result = CONST_GET(mDO, "Result");
+	cDO_Reader = CONST_GET(mDO, "Reader");
+	
+	mPostgres = rb_define_module_under(mDO, "Postgres");
+	ePostgresError = rb_define_class("PostgresError", rb_eStandardError);
+	
+	cConnection = POSTGRES_CLASS("Connection", cDO_Connection);
+	rb_define_method(cConnection, "initialize", cConnection_initialize, 1);
+	rb_define_method(cConnection, "dispose", cConnection_dispose, 0);
+	
+	cCommand = POSTGRES_CLASS("Command", cDO_Command);
+	rb_include_module(cCommand, cDO_Quoting);
+	rb_define_method(cCommand, "set_types", cCommand_set_types, 1);
+	rb_define_method(cCommand, "execute_non_query", cCommand_execute_non_query, -1);
+	rb_define_method(cCommand, "execute_reader", cCommand_execute_reader, -1);
+	rb_define_method(cCommand, "quote_string", cCommand_quote_string, 1);
+	
+	cResult = POSTGRES_CLASS("Result", cDO_Result);
+	
+	cReader = POSTGRES_CLASS("Reader", cDO_Reader);
+	rb_define_method(cReader, "close", cReader_close, 0);
+	rb_define_method(cReader, "next!", cReader_next, 0);
+	rb_define_method(cReader, "values", cReader_values, 0);
+	rb_define_method(cReader, "fields", cReader_fields, 0);
+	
+}
+