Add support for STL and other 3D file formats

internal/magic/geo.go

@@ -1,8 +1,18 @@
 package magic
 
 import (
+	"bufio"
 	"bytes"
 	"encoding/binary"
+	"strconv"
+	"strings"
+)
+
+var (
+	// Mtl matches a Material Library format file by Wavefront Technologies.
+	// https://www.loc.gov/preservation/digital/formats/fdd/fdd000508.shtml
+	// https://www.iana.org/assignments/media-types/model/mtl
+	Mtl = prefix([]byte("newmtl"))
 )
 
 // Shp matches a shape format file.
@@ -53,3 +63,63 @@ func Shp(raw []byte, limit uint32) bool {
 func Shx(raw []byte, limit uint32) bool {
 	return bytes.HasPrefix(raw, []byte{0x00, 0x00, 0x27, 0x0A})
 }
+
+// Stl matches a StereoLithography file.
+// STL is available in ASCII as well as Binary representations for compact file format.
+// https://docs.fileformat.com/cad/stl/
+// https://www.iana.org/assignments/media-types/model/stl
+func Stl(raw []byte, limit uint32) bool {
+	// ASCII check.
+	if bytes.HasPrefix(raw, []byte("solid")) {
+		// If the full file content was provided, check file last line.
+		if len(raw) < int(limit) {
+			return bytes.Contains(lastNonWSLine(raw), []byte("endsolid"))
+		}
+		return true
+	}
+
+	// Binary check.
+	return bytes.HasPrefix(raw, bytes.Repeat([]byte{0x20}, 80))
+}
+
+// Obj matches a 3D object model format by Wavefront Technologies.
+// https://www.loc.gov/preservation/digital/formats/fdd/fdd000507.shtml
+// https://www.iana.org/assignments/media-types/model/obj
+func Obj(raw []byte, limit uint32) bool {
+	s := bufio.NewScanner(bytes.NewReader(raw))
+	for s.Scan() {
+		fs := strings.Fields(s.Text())
+		// Check if match a geometric vertice format "v x y z [w]".
+		if (len(fs) == 4 || len(fs) == 5) && fs[0] == "v" {
+			for _, f := range fs[1:] {
+				if _, err := strconv.ParseFloat(f, 64); err != nil {
+					return false
+				}
+				return true
+			}
+		}
+	}
+	return false
+}
+
+// Ply matches a Polygon File Format or the Stanford Triangle Format file.
+// https://www.loc.gov/preservation/digital/formats/fdd/fdd000501.shtml
+func Ply(raw []byte, limit uint32) bool {
+	s := bufio.NewScanner(bytes.NewReader(raw))
+
+	// First line must be "ply".
+	if !s.Scan() {
+		return false
+	}
+	if s.Text() != "ply" {
+		return false
+	}
+
+	// Second line declares the subtype.
+	if !s.Scan() {
+		return false
+	}
+	return s.Text() == "format ascii 1.0" ||
+		s.Text() == "format binary_little_endian 1.0" ||
+		s.Text() == "format binary_big_endian 1.0"
+}

internal/magic/magic.go

@@ -220,6 +220,16 @@ func firstLine(in []byte) []byte {
 	return in[:lineEnd]
 }
 
+func lastNonWSLine(in []byte) []byte {
+	s := bytes.Split(in, []byte("\n"))
+	for i := len(s) - 1; i >= 0; i-- {
+		if len(trimLWS(trimRWS(s[i]))) > 0 {
+			return s[i]
+		}
+	}
+	return []byte{}
+}
+
 func isWS(b byte) bool {
 	return b == '\t' || b == '\n' || b == '\x0c' || b == '\r' || b == ' '
 }

internal/magic/magic_test.go

@@ -113,6 +113,23 @@ func TestDropLastLine(t *testing.T) {
 	}
 }
 
+func TestLastNonWSLine(t *testing.T) {
+	tests := []struct {
+		raw []byte
+		res []byte
+	}{
+		{[]byte(""), []byte{}},
+		{[]byte("\n\n\n"), []byte{}},
+		{[]byte("a\nb\ncd \n  \t\n"), []byte("cd ")},
+	}
+	for i, tt := range tests {
+		got := lastNonWSLine(tt.raw)
+		if string(got) != string(tt.res) {
+			t.Errorf("lastNonWSLine %d error: expected %q; got %q", i, tt.res, got)
+		}
+	}
+}
+
 func BenchmarkSrt(b *testing.B) {
 	const subtitle = `1
 00:02:16,612 --> 00:02:19,376

mimetype_test.go

@@ -128,10 +128,12 @@ var files = map[string]string{
 	"mrc.mrc":            "application/marc",
 	"msi.msi":            "application/x-ms-installer",
 	"msg.msg":            "application/vnd.ms-outlook",
+	"mtl.mtl":            "model/mtl",
 	"ndjson.xl.ndjson":   "application/x-ndjson",
 	"ndjson.ndjson":      "application/x-ndjson",
 	"nes.nes":            "application/vnd.nintendo.snes.rom",
 	"elfobject":          "application/x-object",
+	"obj.obj":            "model/obj",
 	"odf.odf":            "application/vnd.oasis.opendocument.formula",
 	"sxc.sxc":            "application/vnd.sun.xml.calc",
 	"odg.odg":            "application/vnd.oasis.opendocument.graphics",
@@ -152,6 +154,7 @@ var files = map[string]string{
 	"pdf.pdf":            "application/pdf",
 	"php.php":            "text/x-php",
 	"pl.pl":              "text/x-perl",
+	"ply.ply":            "text/plain; charset=utf-8",
 	"png.png":            "image/png",
 	"ppt.ppt":            "application/vnd.ms-powerpoint",
 	"pptx.pptx":          "application/vnd.openxmlformats-officedocument.presentationml.presentation",
@@ -173,6 +176,9 @@ var files = map[string]string{
 	"shx.shx":            "application/octet-stream",
 	"so.so":              "application/x-sharedlib",
 	"sqlite.sqlite":      "application/vnd.sqlite3",
+	"stl.ascii.stl":      "model/stl",
+	"stl.ascii2.stl":     "model/stl",
+	"stl.binary.stl":     "model/stl",
 	"srt.srt":            "application/x-subrip",
 	// not.srt.txt uses periods instead of commas for the decimal separators of
 	// the timestamps.

supported_mimes.md

@@ -1,4 +1,4 @@
-## 170 Supported MIME types
+## 174 Supported MIME types
 This file is automatically generated when running tests. Do not edit manually.
 
 Extension | MIME type | Aliases
@@ -138,6 +138,9 @@ Extension | MIME type | Aliases
 **.gbr** | image/x-gimp-gbr | -
 **.glb** | model/gltf-binary | -
 **.avif** | image/avif | -
+**.mtl** | model/mtl | -
+**.obj** | model/obj | -
+**.stl** | model/stl | -
 **.txt** | text/plain | -
 **.html** | text/html | -
 **.svg** | image/svg+xml | -
@@ -173,3 +176,4 @@ Extension | MIME type | Aliases
 **.ics** | text/calendar | -
 **.warc** | application/warc | -
 **.vtt** | text/vtt | -
+**.ply** | text/plain | application/octet-stream

testdata/mtl.mtl

@@ -0,0 +1,17 @@
+newmtl flatwhite
+Ka  0.5000  0.5000  0.5000
+Kd  1.0000  1.0000  1.0000
+illum 1
+
+newmtl shinyred
+Ka  0.1985  0.0000  0.0000
+Kd  0.5921  0.0167  0.0000
+Ks  0.5973  0.2083  0.2083
+illum 2
+Ns 100.2235
+
+newmtl clearblue
+Ka  0.0394  0.0394  0.3300
+Kd  0.1420  0.1420  0.9500
+illum 1
+Tr 0.4300