Add —ignore-invalid-pages option (#45)

* add —ignore-invalid-pages option * Update pdf.go change CLI option description --------- Co-authored-by: jerbob92 <jerbob92@users.noreply.github.com>
klippa-app · Jul 16, 2024 · 1f500d3 · 1f500d3
1 parent c7b5c22
commit 1f500d3
Show file tree

Hide file tree

Showing 9 changed files with 48 additions and 16 deletions.
diff --git a/cmd/explode.go b/cmd/explode.go
@@ -66,7 +66,7 @@ var explodeCmd = &cobra.Command{
 			pageRange = cmd.Flag("pages").Value.String()
 		}
 
-		parsedPageRange, _, err := pdf.NormalizePageRange(pageCount.PageCount, pageRange)
+		parsedPageRange, _, err := pdf.NormalizePageRange(pageCount.PageCount, pageRange, ignoreInvalidPages)
 		if err != nil {
 			handleError(cmd, fmt.Errorf("invalid page range '%s': %w\n", pageRange, err), ExitCodeInvalidPageRange)
 			return

diff --git a/cmd/images.go b/cmd/images.go
@@ -86,7 +86,7 @@ var imagesCmd = &cobra.Command{
 			pageRange = pages
 		}
 
-		parsedPageRange, _, err := pdf.NormalizePageRange(pageCount.PageCount, pageRange)
+		parsedPageRange, _, err := pdf.NormalizePageRange(pageCount.PageCount, pageRange, ignoreInvalidPages)
 		if err != nil {
 			handleError(cmd, fmt.Errorf("invalid page range '%s': %w\n", pageRange, err), ExitCodeInvalidPageRange)
 			return

diff --git a/cmd/merge.go b/cmd/merge.go
@@ -90,7 +90,7 @@ var mergeCmd = &cobra.Command{
 				return
 			}
 
-			pageRange, calculatedPageCount, err := pdf.NormalizePageRange(pageCount.PageCount, "first-last")
+			pageRange, calculatedPageCount, err := pdf.NormalizePageRange(pageCount.PageCount, "first-last", false)
 			if err != nil {
 				closeFunc()
 				handleError(cmd, fmt.Errorf("invalid page range 'first-last': %w\n", err), ExitCodeInvalidPageRange)

diff --git a/cmd/pdf.go b/cmd/pdf.go
@@ -17,9 +17,10 @@ import (
 
 var (
 	// Used for flags.
-	password         string
-	stdFileDelimiter string
-	pages            string
+	password           string
+	stdFileDelimiter   string
+	pages              string
+	ignoreInvalidPages bool
 )
 
 func addGenericPDFOptions(command *cobra.Command) {
@@ -29,6 +30,7 @@ func addGenericPDFOptions(command *cobra.Command) {
 
 func addPagesOption(intro string, command *cobra.Command) {
 	command.Flags().StringVarP(&pages, "pages", "", "first-last", intro+". Ranges are like '1-3,5', which will result in a PDF file with pages 1, 2, 3 and 5. You can use the keywords first and last. You can prepend a page number with r to start counting from the end. Examples: use '2-last' for the second page until the last page, use '3-r1' for page 3 until the second-last page.")
+	command.Flags().BoolVarP(&ignoreInvalidPages, "ignore-invalid-pages", "", false, "Ignore non-existing pages in the pages option.")
 }
 
 func isExperimentalError(err error) bool {

diff --git a/cmd/render.go b/cmd/render.go
@@ -88,7 +88,7 @@ var renderCmd = &cobra.Command{
 			pageRange = pages
 		}
 
-		parsedPageRange, _, err := pdf.NormalizePageRange(pageCount.PageCount, pageRange)
+		parsedPageRange, _, err := pdf.NormalizePageRange(pageCount.PageCount, pageRange, ignoreInvalidPages)
 		if err != nil {
 			handleError(cmd, fmt.Errorf("invalid page range '%s': %w\n", pageRange, err), ExitCodeInvalidPageRange)
 			return

diff --git a/cmd/text.go b/cmd/text.go
@@ -96,7 +96,7 @@ var textCmd = &cobra.Command{
 			pageRange = pages
 		}
 
-		parsedPageRange, _, err := pdf.NormalizePageRange(pageCount.PageCount, pageRange)
+		parsedPageRange, _, err := pdf.NormalizePageRange(pageCount.PageCount, pageRange, ignoreInvalidPages)
 		if err != nil {
 			handleError(cmd, fmt.Errorf("invalid page range '%s': %w\n", pageRange, err), ExitCodeInvalidPageRange)
 			return

diff --git a/cmd/thumbnails.go b/cmd/thumbnails.go
@@ -82,7 +82,7 @@ var thumbnailsCmd = &cobra.Command{
 			pageRange = pages
 		}
 
-		parsedPageRange, _, err := pdf.NormalizePageRange(pageCount.PageCount, pageRange)
+		parsedPageRange, _, err := pdf.NormalizePageRange(pageCount.PageCount, pageRange, ignoreInvalidPages)
 		if err != nil {
 			handleError(cmd, fmt.Errorf("invalid page range '%s': %w\n", pageRange, err), ExitCodeInvalidPageRange)
 			return

diff --git a/pdf/pdf.go b/pdf/pdf.go
@@ -27,7 +27,7 @@ func ClosePdfium() {
 // supports simple instructions like 1-5 or just a page number. This method
 // can automatically calculate ends and reverse pages for example.
 // This way we can also properly validate page ranges.
-func NormalizePageRange(pageCount int, pageRange string) (*string, *int, error) {
+func NormalizePageRange(pageCount int, pageRange string, ignoreInvalidPages bool) (*string, *int, error) {
 	calculatedPageCount := 0
 	var calculatedPageNumbers []string
 	seenPageNumbers := map[int]bool{}
@@ -48,21 +48,35 @@ func NormalizePageRange(pageCount int, pageRange string) (*string, *int, error)
 			} else if strings.HasPrefix(pageRangeParts[pageRangePartI], "r") {
 				parsedPageNumber, err := strconv.Atoi(strings.TrimPrefix(pageRangeParts[pageRangePartI], "r"))
 				if err != nil {
+					if ignoreInvalidPages {
+						continue
+					}
 					return nil, nil, fmt.Errorf("%s is not a valid page number", strings.TrimPrefix(pageRangeParts[pageRangePartI], "r"))
 				}
 
 				if pageCount-parsedPageNumber < 1 || pageCount-parsedPageNumber > pageCount {
+					if ignoreInvalidPages {
+						pageNumbers = append(pageNumbers, 1)
+						continue
+					}
 					return nil, nil, fmt.Errorf("%d is not a valid page number, the document has %d page(s)", pageCount-parsedPageNumber, pageCount)
 				}
 
 				pageNumbers = append(pageNumbers, pageCount-parsedPageNumber)
 			} else {
 				parsedPageNumber, err := strconv.Atoi(pageRangeParts[pageRangePartI])
 				if err != nil {
+					if ignoreInvalidPages {
+						continue
+					}
 					return nil, nil, fmt.Errorf("%s is not a valid page number", pageRangeParts[pageRangePartI])
 				}
 
 				if parsedPageNumber < 1 || parsedPageNumber > pageCount {
+					if ignoreInvalidPages {
+						pageNumbers = append(pageNumbers, pageCount)
+						continue
+					}
 					return nil, nil, fmt.Errorf("%s is not a valid page number, the document has %d page(s)", pageRangeParts[pageRangePartI], pageCount)
 				}
 

diff --git a/pdf/pdf_test.go b/pdf/pdf_test.go
@@ -6,66 +6,82 @@ import (
 
 func TestNormalizePageRange(t *testing.T) {
 	tests := []struct {
-		name      string
-		pageCount int
-		pageRange string
-		want      string
-		wantErr   string
+		name               string
+		pageCount          int
+		pageRange          string
+		ignoreInvalidPages bool
+		want               string
+		wantErr            string
 	}{
 		{
 			"test first-last",
 			5,
 			"first-last",
+			false,
 			"1,2,3,4,5",
 			"",
 		},
 		{
 			"test page-range",
 			5,
 			"1-5",
+			false,
 			"1,2,3,4,5",
 			"",
 		},
 		{
 			"test out of range page-range",
 			5,
 			"1-10",
+			false,
 			"1,2,3,4,5",
 			"10 is not a valid page number, the document has 5 page(s)",
 		},
 		{
 			"test out of range page-range",
 			20,
 			"1,2,4-22",
+			false,
 			"1,2,3,4,5",
 			"22 is not a valid page number, the document has 20 page(s)",
 		},
 		{
 			"test reverse page-range",
 			5,
 			"1-r2",
+			false,
 			"1,2,3",
 			"",
 		},
 		{
 			"test negative reverse page-range",
 			5,
 			"1-r6",
+			false,
 			"1,2,3",
 			"-1 is not a valid page number, the document has 5 page(s)",
 		},
 		{
 			"test removal of duplicate pages",
 			5,
 			"1-3,first-last,2,3",
+			false,
 			"1,2,3,4,5",
 			"",
 		},
+		{
+			"test ignore invalid pages",
+			5,
+			"3-10,6,2",
+			true,
+			"3,4,5,2",
+			"",
+		},
 	}
 
 	for i := range tests {
 		t.Run(tests[i].name, func(t *testing.T) {
-			normalizedPageRange, _, err := NormalizePageRange(tests[i].pageCount, tests[i].pageRange)
+			normalizedPageRange, _, err := NormalizePageRange(tests[i].pageCount, tests[i].pageRange, tests[i].ignoreInvalidPages)
 			if tests[i].wantErr == "" && err != nil {
 				t.Errorf("expected no error but got error %s", err.Error())
 			} else if tests[i].wantErr != "" && err == nil {