Fix #794 , add PDF 2.0 disclaimer

pkg/api/api.go

@@ -47,6 +47,25 @@ import (
 	"github.com/pkg/errors"
 )
 
+func logDisclaimerPDF20() {
+	disclaimer := `
+***************************** Disclaimer ****************************
+* PDF 2.0 features are supported on a need basis.                   *
+* (See ISO 32000:2 6.3.2 Conformance of PDF processors)             *
+* At the moment pdfcpu comes with basic PDF 2.0 support.            *
+* Please let us know which feature you would like to see supported, *
+* provide a sample PDF file and create an issue:                    *
+* https://github.com/pdfcpu/pdfcpu/issues/new/choose                *
+*********************************************************************`
+
+	if log.ValidateEnabled() {
+		log.Validate.Println(disclaimer)
+	}
+	if log.CLIEnabled() {
+		log.CLI.Println(disclaimer)
+	}
+}
+
 // ReadContext uses an io.ReadSeeker to build an internal structure holding its cross reference table aka the Context.
 func ReadContext(rs io.ReadSeeker, conf *model.Configuration) (*model.Context, error) {
 	if rs == nil {
@@ -68,6 +87,10 @@ func ReadContextFile(inFile string) (*model.Context, error) {
 		return nil, err
 	}
 
+	if ctx.Version() == model.V20 {
+		logDisclaimerPDF20()
+	}
+
 	if err = validate.XRefTable(ctx.XRefTable); err != nil {
 		return nil, err
 	}
@@ -77,6 +100,11 @@ func ReadContextFile(inFile string) (*model.Context, error) {
 
 // ValidateContext validates ctx.
 func ValidateContext(ctx *model.Context) error {
+
+	if ctx.Version() == model.V20 {
+		logDisclaimerPDF20()
+	}
+
 	return validate.XRefTable(ctx.XRefTable)
 }
 
@@ -130,6 +158,10 @@ func readAndValidate(rs io.ReadSeeker, conf *model.Configuration, from1 time.Tim
 
 	from2 := time.Now()
 
+	if ctx.Version() == model.V20 {
+		logDisclaimerPDF20()
+	}
+
 	if err = validate.XRefTable(ctx.XRefTable); err != nil {
 		return nil, 0, 0, err
 	}
@@ -146,10 +178,6 @@ func ReadValidateAndOptimize(rs io.ReadSeeker, conf *model.Configuration, from1
 		return nil, 0, 0, 0, err
 	}
 
-	if ctx.Version() == model.V20 {
-		return nil, 0, 0, 0, pdfcpu.ErrUnsupportedVersion
-	}
-
 	from3 := time.Now()
 	if err = OptimizeContext(ctx); err != nil {
 		return nil, 0, 0, 0, err

pkg/api/booklet.go

@@ -88,10 +88,6 @@ func Booklet(rs io.ReadSeeker, w io.Writer, imgFiles, selectedPages []string, nu
 			return err
 		}
 
-		if ctx.Version() == model.V20 {
-			return pdfcpu.ErrUnsupportedVersion
-		}
-
 		if err := ctx.EnsurePageCount(); err != nil {
 			return err
 		}

pkg/api/extract.go

@@ -264,7 +264,7 @@ func ExtractPages(rs io.ReadSeeker, outDir, fileName string, selectedPages []str
 
 	if len(pages) == 0 {
 		if log.CLIEnabled() {
-			log.CLI.Println("aborted: nothing to extract!")
+			log.CLI.Println("aborted: missing page numbers!")
 		}
 		return nil
 	}

pkg/api/merge.go

@@ -36,7 +36,7 @@ func appendTo(rs io.ReadSeeker, fName string, ctxDest *model.Context, dividerPag
 		return err
 	}
 
-	if ctxSource.Version() == model.V20 {
+	if ctxDest.Version() < model.V20 && ctxSource.Version() == model.V20 {
 		return pdfcpu.ErrUnsupportedVersion
 	}
 
@@ -93,7 +93,9 @@ func prepDestContext(destFile string, rs io.ReadSeeker, conf *model.Configuratio
 		}
 	}
 
-	ctxDest.EnsureVersionForWriting()
+	if ctxDest.Version() < model.V20 {
+		ctxDest.EnsureVersionForWriting()
+	}
 
 	return ctxDest, nil
 }
@@ -135,9 +137,6 @@ func Merge(destFile string, inFiles []string, w io.Writer, conf *model.Configura
 	if err != nil {
 		return err
 	}
-	if ctxDest.Version() == model.V20 {
-		return pdfcpu.ErrUnsupportedVersion
-	}
 
 	for _, fName := range inFiles {
 		if err := func() error {
@@ -181,6 +180,7 @@ func MergeCreateFile(inFiles []string, outFile string, dividerPage bool, conf *m
 				return
 			}
 			os.Remove(outFile)
+			return
 		}
 		if err = f.Close(); err != nil {
 			return

pkg/api/nup.go

@@ -119,10 +119,6 @@ func NUp(rs io.ReadSeeker, w io.Writer, imgFiles, selectedPages []string, nup *m
 			return err
 		}
 
-		if ctx.Version() == model.V20 {
-			return pdfcpu.ErrUnsupportedVersion
-		}
-
 		if err := ctx.EnsurePageCount(); err != nil {
 			return err
 		}

pkg/api/page.go

@@ -19,9 +19,11 @@ package api
 import (
 	"io"
 	"os"
+	"sort"
 	"time"
 
 	"github.com/pdfcpu/pdfcpu/pkg/log"
+	"github.com/pdfcpu/pdfcpu/pkg/pdfcpu"
 	"github.com/pdfcpu/pdfcpu/pkg/pdfcpu/model"
 	"github.com/pdfcpu/pdfcpu/pkg/pdfcpu/types"
 	"github.com/pkg/errors"
@@ -134,7 +136,7 @@ func RemovePages(rs io.ReadSeeker, w io.Writer, selectedPages []string, conf *mo
 	conf.Cmd = model.REMOVEPAGES
 
 	fromStart := time.Now()
-	ctx, durRead, durVal, durOpt, err := ReadValidateAndOptimize(rs, conf, fromStart)
+	ctx, _, _, _, err := ReadValidateAndOptimize(rs, conf, fromStart)
 	if err != nil {
 		return err
 	}
@@ -143,31 +145,38 @@ func RemovePages(rs io.ReadSeeker, w io.Writer, selectedPages []string, conf *mo
 		return err
 	}
 
-	fromWrite := time.Now()
-
-	pages, err := PagesForPageSelection(ctx.PageCount, selectedPages, false, true)
+	pages, err := RemainingPagesForPageRemoval(ctx.PageCount, selectedPages, true)
 	if err != nil {
 		return err
 	}
 
-	// ctx.Pagecount gets set during validation.
-	if len(pages) >= ctx.PageCount {
-		return errors.New("pdfcpu: operation invalid")
+	if len(pages) == 0 {
+		if log.CLIEnabled() {
+			log.CLI.Println("aborted: missing page numbers!")
+		}
+		return nil
 	}
 
-	// No special context processing required.
-	// WriteContext decides which pages get written by checking conf.Cmd
+	var pageNrs []int
+	for k, v := range pages {
+		if v {
+			pageNrs = append(pageNrs, k)
+		}
+	}
+	sort.Ints(pageNrs)
 
-	ctx.Write.SelectedPages = pages
-	if err = WriteContext(ctx, w); err != nil {
+	ctxDest, err := pdfcpu.ExtractPages(ctx, pageNrs, false)
+	if err != nil {
 		return err
 	}
 
-	durWrite := time.Since(fromWrite).Seconds()
-	durTotal := time.Since(fromStart).Seconds()
-	logOperationStats(ctx, "remove pages, write", durRead, durVal, durOpt, durWrite, durTotal)
+	if conf.ValidationMode != model.ValidationNone {
+		if err = ValidateContext(ctxDest); err != nil {
+			return err
+		}
+	}
 
-	return nil
+	return WriteContext(ctxDest, w)
 }
 
 // RemovePagesFile removes selected inFile pages and writes the result to outFile..

pkg/api/pageLayout.go

@@ -21,7 +21,6 @@ import (
 	"os"
 	"time"
 
-	"github.com/pdfcpu/pdfcpu/pkg/pdfcpu"
 	"github.com/pdfcpu/pdfcpu/pkg/pdfcpu/model"
 	"github.com/pdfcpu/pdfcpu/pkg/pdfcpu/types"
 	"github.com/pkg/errors"
@@ -113,10 +112,6 @@ func SetPageLayout(rs io.ReadSeeker, w io.Writer, val model.PageLayout, conf *mo
 		return err
 	}
 
-	if ctx.Version() == model.V20 {
-		return pdfcpu.ErrUnsupportedVersion
-	}
-
 	ctx.RootDict["PageLayout"] = types.Name(val.String())
 
 	if err = WriteContext(ctx, w); err != nil {
@@ -182,10 +177,6 @@ func ResetPageLayout(rs io.ReadSeeker, w io.Writer, conf *model.Configuration) e
 		return err
 	}
 
-	if ctx.Version() == model.V20 {
-		return pdfcpu.ErrUnsupportedVersion
-	}
-
 	delete(ctx.RootDict, "PageLayout")
 
 	if err = WriteContext(ctx, w); err != nil {

pkg/api/pageMode.go

@@ -21,7 +21,6 @@ import (
 	"os"
 	"time"
 
-	"github.com/pdfcpu/pdfcpu/pkg/pdfcpu"
 	"github.com/pdfcpu/pdfcpu/pkg/pdfcpu/model"
 	"github.com/pdfcpu/pdfcpu/pkg/pdfcpu/types"
 	"github.com/pkg/errors"
@@ -113,10 +112,6 @@ func SetPageMode(rs io.ReadSeeker, w io.Writer, val model.PageMode, conf *model.
 		return err
 	}
 
-	if ctx.Version() == model.V20 {
-		return pdfcpu.ErrUnsupportedVersion
-	}
-
 	ctx.RootDict["PageMode"] = types.Name(val.String())
 
 	if err = WriteContext(ctx, w); err != nil {
@@ -182,10 +177,6 @@ func ResetPageMode(rs io.ReadSeeker, w io.Writer, conf *model.Configuration) err
 		return err
 	}
 
-	if ctx.Version() == model.V20 {
-		return pdfcpu.ErrUnsupportedVersion
-	}
-
 	delete(ctx.RootDict, "PageMode")
 
 	if err = WriteContext(ctx, w); err != nil {

pkg/api/permission.go

@@ -21,6 +21,7 @@ import (
 	"os"
 	"time"
 
+	"github.com/pdfcpu/pdfcpu/pkg/pdfcpu"
 	"github.com/pdfcpu/pdfcpu/pkg/pdfcpu/model"
 	"github.com/pkg/errors"
 )
@@ -68,6 +69,10 @@ func SetPermissions(rs io.ReadSeeker, w io.Writer, conf *model.Configuration) er
 		return err
 	}
 
+	if ctx.Version() == model.V20 {
+		return pdfcpu.ErrUnsupportedVersion
+	}
+
 	fromWrite := time.Now()
 	if err = WriteContext(ctx, w); err != nil {
 		return err
@@ -141,6 +146,11 @@ func GetPermissions(rs io.ReadSeeker, conf *model.Configuration) (*int16, error)
 	if err != nil {
 		return nil, err
 	}
+
+	if ctx.Version() == model.V20 {
+		return nil, pdfcpu.ErrUnsupportedVersion
+	}
+
 	if ctx.E == nil {
 		// Full access - permissions don't apply.
 		return nil, nil

pkg/api/selectPages.go

@@ -391,6 +391,26 @@ func PagesForPageSelection(pageCount int, pageSelection []string, ensureAllforNo
 	return m, nil
 }
 
+func RemainingPagesForPageRemoval(pageCount int, pageSelection []string, log bool) (types.IntSet, error) {
+	pagesToRemove, err := selectedPages(pageCount, pageSelection, log)
+	if err != nil {
+		return nil, err
+	}
+
+	m := types.IntSet{}
+	for i := 1; i <= pageCount; i++ {
+		m[i] = true
+	}
+
+	for k, v := range pagesToRemove {
+		if v {
+			m[k] = false
+		}
+	}
+
+	return m, nil
+}
+
 func deletePageFromCollection(cp *[]int, p int) {
 	a := []int{}
 	for _, i := range *cp {