Fixing issue with file: URLs

README.md

@@ -112,7 +112,7 @@ additional aliases, and the related Go driver:
 | MySQL                | `mysql`         | `my`, `maria`, `aurora`, `mariadb`, `percona`   | [github.com/go-sql-driver/mysql][d-mysql]                                   |
 | Microsoft SQL Server | `sqlserver`     | `ms`, `mssql`, `azuresql`                       | [github.com/microsoft/go-mssqldb][d-sqlserver]                              |
 | Oracle Database      | `oracle`        | `or`, `ora`, `oci`, `oci8`, `odpi`, `odpi-c`    | [github.com/sijms/go-ora/v2][d-oracle]                                      |
-| SQLite3              | `sqlite3`       | `sq`, `file`, `sqlite`                          | [github.com/mattn/go-sqlite3][d-sqlite3] <sup>[†][f-cgo]</sup>              |
+| SQLite3              | `sqlite3`       | `sq`, `sqlite`, `file`                          | [github.com/mattn/go-sqlite3][d-sqlite3] <sup>[†][f-cgo]</sup>              |
 | CSVQ                 | `csvq`          | `cs`, `csv`, `tsv`, `json`                      | [github.com/mithrandie/csvq-driver][d-csvq]                                 |
 |                      |                 |                                                 |                                                                             |
 | Alibaba MaxCompute   | `maxcompute`    | `mc`                                            | [sqlflow.org/gomaxcompute][d-maxcompute]                                    |
@@ -126,7 +126,7 @@ additional aliases, and the related Go driver:
 | Couchbase            | `couchbase`     | `n1`, `n1ql`                                    | [github.com/couchbase/go_n1ql][d-couchbase]                                 |
 | Cznic QL             | `ql`            | `cznic`, `cznicql`                              | [modernc.org/ql][d-ql]                                                      |
 | Databend             | `databend`      | `dd`, `bend`                                    | [github.com/databendcloud/databend-go][d-databend]                          |
-| DuckDB               | `duckdb`        | `dk`, `ddb`, `duck`                             | [github.com/marcboeker/go-duckdb][d-duckdb] <sup>[†][f-cgo]</sup>           |
+| DuckDB               | `duckdb`        | `dk`, `ddb`, `duck`, `file`                     | [github.com/marcboeker/go-duckdb][d-duckdb] <sup>[†][f-cgo]</sup>           |
 | Exasol               | `exasol`        | `ex`, `exa`                                     | [github.com/exasol/exasol-driver-go][d-exasol]                              |
 | Firebird             | `firebird`      | `fb`, `firebirdsql`                             | [github.com/nakagami/firebirdsql][d-firebird]                               |
 | FlightSQL            | `flightsql`     | `fl`, `flight`                                  | [github.com/apache/arrow/go/v12/arrow/flight/flightsql/driver][d-flightsql] |

dburl.go

@@ -12,6 +12,7 @@ package dburl
 import (
 	"database/sql"
 	"net/url"
+	"os"
 	"strings"
 )
 
@@ -73,10 +74,10 @@ type URL struct {
 func Parse(urlstr string) (*URL, error) {
 	// parse url
 	v, err := url.Parse(urlstr)
-	if err != nil {
+	switch {
+	case err != nil:
 		return nil, err
-	}
-	if v.Scheme == "" {
+	case v.Scheme == "":
 		return nil, ErrInvalidDatabaseScheme
 	}
 	// create url
@@ -97,6 +98,15 @@ func Parse(urlstr string) (*URL, error) {
 	if !ok {
 		return nil, ErrUnknownDatabaseScheme
 	}
+	// load real scheme for file:
+	if scheme.Driver == "file" {
+		typ, err := SchemeType(u.Opaque)
+		if err == nil {
+			if s, ok := schemeMap[typ]; ok {
+				scheme = s
+			}
+		}
+	}
 	// if scheme does not understand opaque URLs, retry parsing after building
 	// fully qualified URL
 	if !scheme.Opaque && u.Opaque != "" {
@@ -230,6 +240,25 @@ func (u *URL) Normalize(sep, empty string, cut int) string {
 	return strings.Join(s, sep)
 }
 
+// SchemeType returns the scheme type for a file on disk.
+func SchemeType(name string) (string, error) {
+	f, err := os.OpenFile(name, os.O_RDONLY, 0)
+	if err != nil {
+		return "", err
+	}
+	defer f.Close()
+	buf := make([]byte, 128)
+	if _, err := f.Read(buf); err != nil {
+		return "", err
+	}
+	for _, header := range headerTypes {
+		if header.f(buf) {
+			return header.driver, nil
+		}
+	}
+	return "", ErrUnknownFileHeader
+}
+
 // Error is an error.
 type Error string
 
@@ -244,6 +273,8 @@ const (
 	ErrInvalidDatabaseScheme Error = "invalid database scheme"
 	// ErrUnknownDatabaseScheme is the unknown database type error.
 	ErrUnknownDatabaseScheme Error = "unknown database scheme"
+	// ErrUnknownFileHeader is the unknown file header error.
+	ErrUnknownFileHeader Error = "unknown file header"
 	// ErrInvalidTransportProtocol is the invalid transport protocol error.
 	ErrInvalidTransportProtocol Error = "invalid transport protocol"
 	// ErrRelativePathNotSupported is the relative paths not supported error.

dburl_test.go

@@ -43,8 +43,9 @@ func TestBadParse(t *testing.T) {
 		{`file+tcp://`, ErrInvalidTransportProtocol},
 		{`file://`, ErrMissingPath},
 		{`ql://`, ErrMissingPath},
+		{`duckdb://`, ErrMissingPath},
 		{`mssql+tcp://user:pass@host/dbname`, ErrInvalidTransportProtocol},
-		{`mssql+aoeu://`, ErrInvalidTransportProtocol},
+		{`mssql+foobar://`, ErrInvalidTransportProtocol},
 		{`mssql+unix:/var/run/mssql.sock`, ErrInvalidTransportProtocol},
 		{`mssql+udp:localhost:155`, ErrInvalidTransportProtocol},
 		{`adodb+foo+bar://provider/database`, ErrInvalidTransportProtocol},
@@ -140,7 +141,6 @@ func TestParse(t *testing.T) {
 			`oo+Postgres+Unicode://user:pass@host:5432/dbname`, `adodb`,
 			`Provider=MSDASQL.1;Extended Properties="Database=dbname;Driver={Postgres Unicode};PWD=pass;Port=5432;Server=host;UID=user"`, ``,
 		},
-		{`file:/path/to/file.sqlite3`, `sqlite3`, `/path/to/file.sqlite3`, ``}, // 33
 		{`sqlite:///path/to/file.sqlite3`, `sqlite3`, `/path/to/file.sqlite3`, ``},
 		{`sq://path/to/file.sqlite3`, `sqlite3`, `path/to/file.sqlite3`, ``},
 		{`sq:path/to/file.sqlite3`, `sqlite3`, `path/to/file.sqlite3`, ``},
@@ -210,6 +210,8 @@ func TestParse(t *testing.T) {
 		{`flightsql://user:pass@localhost?timeout=3s&token=foobar&tls=enabled`, `flightsql`, `flightsql://user:pass@localhost?timeout=3s&token=foobar&tls=enabled`, ``},
 		{`duckdb:/path/to/foo.db?access_mode=read_only&threads=4`, `duckdb`, `/path/to/foo.db?access_mode=read_only&threads=4`, ``},
 		{`dk:///path/to/foo.db?access_mode=read_only&threads=4`, `duckdb`, `/path/to/foo.db?access_mode=read_only&threads=4`, ``},
+		{`file:./testdata/test.sqlite3?a=b`, `sqlite3`, `./testdata/test.sqlite3?a=b`, ``},
+		{`file:./testdata/test.duckdb?a=b`, `duckdb`, `./testdata/test.duckdb?a=b`, ``},
 	}
 	for i, test := range tests {
 		u, err := Parse(test.s)

scheme.go

@@ -1,7 +1,9 @@
 package dburl
 
 import (
+	"bytes"
 	"fmt"
+	"regexp"
 	"sort"
 )
 
@@ -48,6 +50,12 @@ type Scheme struct {
 // BaseSchemes returns the supported base schemes.
 func BaseSchemes() []Scheme {
 	return []Scheme{
+		{
+			"file",
+			GenOpaque, 0, true,
+			[]string{"file"},
+			"",
+		},
 		// core databases
 		{
 			"mysql",
@@ -71,7 +79,7 @@ func BaseSchemes() []Scheme {
 		{
 			"sqlite3",
 			GenOpaque, 0, true,
-			[]string{"sqlite", "file"},
+			[]string{"sqlite"},
 			"",
 		},
 		{
@@ -88,8 +96,7 @@ func BaseSchemes() []Scheme {
 			"postgres",
 		},
 		{
-			"memsql",
-			GenMysql, 0, false, nil, "mysql",
+			"memsql", GenMysql, 0, false, nil, "mysql",
 		},
 		{
 			"redshift",
@@ -331,6 +338,8 @@ func init() {
 	for _, scheme := range schemes {
 		Register(scheme)
 	}
+	RegisterHeaderType("duckdb", isDuckdbHeader)
+	RegisterHeaderType("sqlite3", isSqlite3Header)
 }
 
 // schemeMap is the map of registered schemes.
@@ -431,6 +440,32 @@ func RegisterAlias(name, alias string) {
 	registerAlias(name, alias, true)
 }
 
+// headerTypes are registered header recognition funcs.
+var headerTypes []headerType
+
+// RegisterHeaderType registers a file header recognition func.
+func RegisterHeaderType(driver string, f func([]byte) bool) {
+	headerTypes = append(headerTypes, headerType{
+		driver: driver,
+		f:      f,
+	})
+}
+
+// headerType wraps a header recognition func.
+type headerType struct {
+	driver string
+	f      func([]byte) bool
+}
+
+// HeaderTypes returns the registered header types.
+func HeaderTypes() []string {
+	var v []string
+	for _, header := range headerTypes {
+		v = append(v, header.driver)
+	}
+	return v
+}
+
 // Protocols returns list of all valid protocol aliases for a registered
 // [Scheme] name.
 func Protocols(name string) []string {
@@ -474,6 +509,27 @@ func ShortAlias(name string) string {
 	return schemeMap[name].Aliases[0]
 }
 
+// isSqlite3Header returns true when the passed header is empty or starts with
+// the SQLite3 header.
+//
+// See: https://www.sqlite.org/fileformat.html
+func isSqlite3Header(buf []byte) bool {
+	return len(buf) == 0 || bytes.HasPrefix(buf, sqlite3Header)
+}
+
+// sqlite3Header is the sqlite3 header.
+var sqlite3Header = []byte("SQLite format 3\000")
+
+// isDuckdbHeader returns true when the passed header is a DuckDB header.
+//
+// See: https://duckdb.org/internals/storage
+func isDuckdbHeader(buf []byte) bool {
+	return duckdbRE.Match(buf)
+}
+
+// duckdbRE is the duckdb storage header regexp.
+var duckdbRE = regexp.MustCompile(`^.{8}DUCK.{8}`)
+
 // contains determines if v contains s.
 func contains(v []string, s string) bool {
 	for _, z := range v {