Improved MS Word document name extraction

Use document filename and not H1 title inside document.

Supports both Linux & MSFT.

domain/conversion/conversion.go

@@ -17,6 +17,7 @@ import (
 	"fmt"
 	"io"
 	"net/http"
+	"path/filepath"
 	"strings"
 
 	api "github.com/documize/community/core/convapi"
@@ -284,11 +285,11 @@ func convertFileResult(filename string, fileResult *api.DocumentConversionRespon
 	document.Job = ""
 	document.Location = filename
 
+	// Make document name from filename minus extension.
+	document.Name = GetDocumentNameFromFilename(filename)
+	document.Slug = stringutil.MakeSlug(document.Name)
+
 	if fileResult != nil {
-		if len(fileResult.Pages) > 0 {
-			document.Name = fileResult.Pages[0].Title
-			document.Slug = stringutil.MakeSlug(fileResult.Pages[0].Title)
-		}
 		document.Excerpt = fileResult.Excerpt
 	}
 
@@ -296,3 +297,24 @@ func convertFileResult(filename string, fileResult *api.DocumentConversionRespon
 
 	return document
 }
+
+// GetDocumentNameFromFilename strips path and extension.
+func GetDocumentNameFromFilename(filename string) (dn string) {
+	dn = filename
+
+	// First try Linux separator.
+	t := strings.SplitAfter(filename, "/")
+	if len(t) > 1 {
+		dn = t[len(t)-1]
+	} else {
+		// Now try Linux separator.
+		t = strings.SplitAfter(filename, "\\")
+		if len(t) > 1 {
+			dn = t[len(t)-1]
+		}
+	}
+
+	// Remove file extension.
+	dn = strings.TrimSuffix(dn, filepath.Ext(dn))
+	return
+}

domain/conversion/conversion_test.go

@@ -0,0 +1,53 @@
+package conversion
+
+import (
+	"testing"
+)
+
+// TestFilename validates filename extraction from path
+func TestFilename(t *testing.T) {
+	fn := "/var/folders/vx/lyhy36cn5kl994qj0pt6hgb80000gp/T/documize/_uploads/970f9d07-9bea-48a2-4333-b3c50bca42cd/Demo.docx"
+	t.Run("Test filename "+fn, func(t *testing.T) {
+		f := GetDocumentNameFromFilename(fn)
+		if f != "Demo" {
+			t.Error("Expected Demo, got " + f)
+		}
+		t.Log(f)
+	})
+
+	fn = "/var/Demo Docs.docx"
+	t.Run("Test filename "+fn, func(t *testing.T) {
+		f := GetDocumentNameFromFilename(fn)
+		if f != "Demo Docs" {
+			t.Error("Expected Demo Docs, got " + f)
+		}
+		t.Log(f)
+	})
+
+	fn = "Demo Docs.docx"
+	t.Run("Test filename "+fn, func(t *testing.T) {
+		f := GetDocumentNameFromFilename(fn)
+		if f != "Demo Docs" {
+			t.Error("Expected Demo Docs, got " + f)
+		}
+		t.Log(f)
+	})
+
+	fn = "/DemoDocs.docx"
+	t.Run("Test filename "+fn, func(t *testing.T) {
+		f := GetDocumentNameFromFilename(fn)
+		if f != "DemoDocs" {
+			t.Error("Expected DemoDocs, got " + f)
+		}
+		t.Log(f)
+	})
+
+	fn = "a\\b\\c\\DemoDocs.docx.ppt"
+	t.Run("Test filename "+fn, func(t *testing.T) {
+		f := GetDocumentNameFromFilename(fn)
+		if f != "DemoDocs.docx" {
+			t.Error("Expected DemoDocs.docx, got " + f)
+		}
+		t.Log(f)
+	})
+}