Basic type mapper

4 files changed, 249 insertions, 1 deletions

diff --git a/lib/output/pq/typeMapper.cpp b/lib/output/pq/typeMapper.cpp
new file mode 100644
index 0000000..5583319
--- /dev/null
+++ b/lib/output/pq/typeMapper.cpp
@@ -0,0 +1,72 @@
+#include "typeMapper.h"
+#include <compileTimeFormatter.h>
+
+namespace MyGrate::Output::Pq {
+	TypeMapper::RegexMapper::RegexMapper(std::optional<ObsRegex> t, std::optional<ObsRegex> c, std::string tt) :
+		typeMatch {std::move(t)}, columnMatch {std::move(c)}, targetType {std::move(tt)}
+	{
+	}
+
+	std::optional<std::string_view>
+	TypeMapper::RegexMapper::map(std::string_view typeName, std::string_view columnName) const
+	{
+		if ((!typeMatch || std::regex_match(typeName.begin(), typeName.end(), *typeMatch))
+				&& (!columnMatch || std::regex_match(columnName.begin(), columnName.end(), *columnMatch))) {
+			if (targetType.empty()) {
+				return {typeName};
+			}
+			return {targetType};
+		}
+		return {};
+	}
+
+	TypeMapper::TypeMapper()
+	{
+		const auto add = [this](auto r, auto t) {
+			mappings.emplace_back(std::make_unique<RegexMapper>(r, std::nullopt, t));
+		};
+		// https://dev.mysql.com/doc/refman/8.0/en/numeric-types.html
+		add(R"(bit\(\d+\))"_r, "");
+		add(R"(tinyint\(1\)( (un)?signed)?)"_r, "boolean");
+		add(R"((tiny|small)int(\(\d+\))?( (un)?signed)?)"_r, "smallint");
+		add(R"((medium)?int(\(\d+\))?( (un)?signed)?)"_r, "int");
+		add(R"(bigint(\(\d+\))?( (un)?signed)?)"_r, "bigint");
+		add(R"((numeric|decimal)(\(\d+(,\d+)?\))?)"_r, "");
+		add(R"(float(\(\d+,\d+\))?)"_r, "float");
+		add(R"((real|double( precision)?)(\(\d+,\d+\))?)"_r, "double");
+		// https://dev.mysql.com/doc/refman/8.0/en/date-and-time-types.html
+		add(R"(date)"_r, "date");
+		add(R"(datetime)"_r, "timestamp without time zone");
+		add(R"(datetime(\(\d+\))?)"_r, "timestamp without time zone"); // drops precision
+		add(R"(timestamp)"_r, "timestamp without time zone");
+		add(R"(timestamp(\(\d+\))?)"_r, "timestamp without time zone"); // drops precision
+		add(R"(time)"_r, "time without time zone");
+		add(R"(time(\(\d+\))?)"_r, "time without time zone"); // drops precision
+		add(R"(year)"_r, "smallint");
+		// https://dev.mysql.com/doc/refman/8.0/en/string-types.html
+		add(R"((var)?binary(\(\d+\))?)"_r, "bytea");
+		add(R"((var)?char(\(\d+\))?)"_r, "");
+		add(R"((tiny||medium|long)blob(\(\d+\))?)"_r, "bytea");
+		add(R"((tiny||medium|long)text(\(\d+\))?)"_r, "text");
+		add(R"(enum(.*))"_r, "text"); // not really
+		// set
+		// https://dev.mysql.com/doc/refman/8.0/en/json.html
+		add(R"(json)"_r, "json");
+	}
+
+	std::string_view
+	TypeMapper::map(std::string_view t, std::string_view n) const
+	{
+		for (const auto & m : mappings) {
+			if (const auto rt {m->map(t, n)}) {
+				return *rt;
+			}
+		}
+		throw std::runtime_error(scprintf<"No mapping for %? (%?)">(n, t));
+	}
+
+	TypeMapper::ObsRegex operator""_r(const char * input, std::size_t)
+	{
+		return MyGrate::Output::Pq::TypeMapper::ObsRegex {input, std::regex::icase};
+	}
+}
diff --git a/lib/output/pq/typeMapper.h b/lib/output/pq/typeMapper.h
new file mode 100644
index 0000000..7005e4f
--- /dev/null
+++ b/lib/output/pq/typeMapper.h
@@ -0,0 +1,48 @@
+#ifndef MYGRATE_OUTPUT_PQ_TYPEMAPPER_H
+#define MYGRATE_OUTPUT_PQ_TYPEMAPPER_H
+
+#include <optional>
+#include <regex>
+#include <string>
+#include <string_view>
+#include <vector>
+
+namespace MyGrate::Output::Pq {
+	class TypeMapper {
+	public:
+		class ObsRegex : public std::regex {
+		public:
+			explicit ObsRegex(const std::string & src, std::regex_constants::syntax_option_type sot = {}) :
+				std::regex {src, sot}, src {src}
+			{
+			}
+			const std::string src;
+		};
+
+		struct Mapping {
+			virtual std::optional<std::string_view> map(std::string_view t, std::string_view n) const = 0;
+		};
+		using MappingPtr = std::unique_ptr<Mapping>;
+
+		struct RegexMapper : public Mapping {
+			RegexMapper(std::optional<ObsRegex>, std::optional<ObsRegex>, std::string);
+
+			std::optional<std::string_view> map(std::string_view t, std::string_view n) const override;
+
+			std::optional<ObsRegex> typeMatch;
+			std::optional<ObsRegex> columnMatch;
+			std::string targetType;
+		};
+
+		TypeMapper();
+
+		std::string_view map(std::string_view t, std::string_view n) const;
+
+	private:
+		std::vector<MappingPtr> mappings;
+	};
+
+	TypeMapper::ObsRegex operator""_r(const char * input, std::size_t);
+}
+
+#endif
diff --git a/test/Jamfile.jam b/test/Jamfile.jam
index 55dcadf..d0b0e70 100644
--- a/test/Jamfile.jam
+++ b/test/Jamfile.jam
@@ -18,4 +18,4 @@ run test-streams.cpp ;
 run test-misc.cpp ;
 run test-mysql.cpp : : : <library>testdb ;
 run test-postgresql.cpp : -- : ../db/schema.sql : <library>testdb ;
-run test-e2e.cpp : -- : ../db/schema.sql : <library>testdb ;
+run test-mapping.cpp : : : <library>testdb <implicit-dependency>testdb ;
diff --git a/test/test-mapping.cpp b/test/test-mapping.cpp
new file mode 100644
index 0000000..0ecc62f
--- /dev/null
+++ b/test/test-mapping.cpp
@@ -0,0 +1,128 @@
+#define BOOST_TEST_MODULE TypeMappers
+#include <boost/test/data/test_case.hpp>
+#include <boost/test/unit_test.hpp>
+
+#include "helpers.h"
+#include <compileTimeFormatter.h>
+#include <output/pq/typeMapper.h>
+
+using M = MyGrate::Output::Pq::TypeMapper::RegexMapper;
+
+namespace std {
+	ostream &
+	operator<<(ostream & s, const MyGrate::Output::Pq::TypeMapper::ObsRegex & r)
+	{
+		return s << r.src;
+	}
+	ostream &
+	operator<<(ostream & s, const M & m)
+	{
+		return MyGrate::scprintf<"{ type: %?, column: %?, target: %? }">(s, m.typeMatch, m.columnMatch, m.targetType);
+	}
+}
+
+using regex_mapping_test_data = std::tuple<M, std::string_view, std::string_view, std::optional<std::string>>;
+using namespace MyGrate::Output::Pq;
+BOOST_DATA_TEST_CASE(regex_mapper,
+		boost::unit_test::data::make<regex_mapping_test_data>({
+				// Types
+				{{R"(int\(\d+\))"_r, {}, "int"}, "int(10)", "any", "int"},
+				{{R"(int(\(\d+\))?)"_r, {}, "int"}, "int", "any", "int"},
+				{{R"(int\(\d+\))"_r, {}, "int"}, "int(1)", "any", "int"},
+				{{R"(int\(\d+\))"_r, {}, "int"}, "varchar(1)", "any", {}},
+				{{R"(varchar\(\d+\))"_r, {}, ""}, "varchar(1)", "any", "varchar(1)"},
+				{{R"(varchar\(\d+\))"_r, {}, ""}, "varchar(10)", "any", "varchar(10)"},
+				{{R"(varchar\(\d+\))"_r, {}, ""}, "int(10)", "any", {}},
+				{{R"(varchar\(\d{4,}\))"_r, {}, "text"}, "varchar(1024)", "any", "text"},
+				// Names
+				{{{}, R"(.*\.id)"_r, "bigint"}, "int(10)", "all.id", "bigint"},
+				{{{}, R"(.*\.name)"_r, "text"}, "int(10)", "all.name", "text"},
+		}),
+		m, inType, inCol, exp)
+{
+	BOOST_CHECK_EQUAL(m.map(inType, inCol), exp);
+}
+
+using mapping_test_data = std::tuple<std::string_view, std::string_view>;
+BOOST_DATA_TEST_CASE(default_type_mapping,
+		boost::unit_test::data::make<mapping_test_data>({
+				// https://dev.mysql.com/doc/refman/8.0/en/numeric-types.html
+				{"bit(10)", "bit(10)"},
+				{"tinyint", "smallint"},
+				{"tinyint(1)", "boolean"},
+				{"tinyint(2)", "smallint"},
+				{"smallint", "smallint"},
+				{"smallint(1)", "smallint"},
+				{"smallint(2)", "smallint"},
+				{"mediumint", "int"},
+				{"mediumint(5)", "int"},
+				{"int", "int"},
+				{"int(10)", "int"},
+				{"bigint", "bigint"},
+				{"bigint(10)", "bigint"},
+				{"tinyint unsigned", "smallint"},
+				{"tinyint(1) unsigned", "boolean"},
+				{"tinyint(2) unsigned", "smallint"},
+				{"smallint unsigned", "smallint"},
+				{"smallint(1) unsigned", "smallint"},
+				{"smallint(2) unsigned", "smallint"},
+				{"mediumint unsigned", "int"},
+				{"mediumint(5) unsigned", "int"},
+				{"int unsigned", "int"},
+				{"int(10) unsigned", "int"},
+				{"bigint unsigned", "bigint"},
+				{"bigint(10) unsigned", "bigint"},
+				{"decimal", "decimal"},
+				{"decimal(1,2)", "decimal(1,2)"},
+				{"decimal(1)", "decimal(1)"},
+				{"numeric", "numeric"},
+				{"numeric(1,2)", "numeric(1,2)"},
+				{"numeric(1)", "numeric(1)"},
+				{"float", "float"},
+				{"float(1,2)", "float"},
+				{"real", "double"},
+				{"real(1,2)", "double"},
+				{"double", "double"},
+				{"double(1,2)", "double"},
+				{"double precision", "double"},
+				{"double precision(1,2)", "double"},
+				// https://dev.mysql.com/doc/refman/8.0/en/date-and-time-types.html
+				{"date", "date"},
+				{"datetime", "timestamp without time zone"},
+				{"datetime(1)", "timestamp without time zone"},
+				{"timestamp", "timestamp without time zone"},
+				{"timestamp(1)", "timestamp without time zone"},
+				{"time", "time without time zone"},
+				{"time(1)", "time without time zone"},
+				{"year", "smallint"},
+				{"time(1)", "time without time zone"},
+				{"year", "smallint"},
+				// https://dev.mysql.com/doc/refman/8.0/en/string-types.html
+				{"binary", "bytea"},
+				{"binary(10)", "bytea"},
+				{"varbinary", "bytea"},
+				{"varbinary(10)", "bytea"},
+				{"char", "char"},
+				{"char(10)", "char(10)"},
+				{"varchar", "varchar"},
+				{"varchar(10)", "varchar(10)"},
+				{"blob", "bytea"},
+				{"blob(100)", "bytea"},
+				{"tinyblob", "bytea"},
+				{"mediumblob", "bytea"},
+				{"longblob", "bytea"},
+				{"text", "text"},
+				{"text(100)", "text"},
+				{"tinytext", "text"},
+				{"mediumtext", "text"},
+				{"longtext", "text"},
+				// enum
+				// set
+				// https://dev.mysql.com/doc/refman/8.0/en/json.html
+				{"json", "json"},
+		}),
+		inType, exp)
+{
+	TypeMapper tm;
+	BOOST_CHECK_EQUAL(tm.map(inType, {}), exp);
+}