Add support for piping

README.md

@@ -19,6 +19,8 @@
 * Extracting attachments from PDFs
 * Extracting thumbnails from PDFs
 * Extracting JavaScripts from PDFs
+* Piping input through stdin when the input is one file
+* Piping output through stdout when the output is one file
 
 ## PDFium & Wazero
 

cmd/attachments.go

@@ -19,13 +19,13 @@ func init() {
 var attachmentsCmd = &cobra.Command{
 	Use:   "attachments [input] [output-folder]",
 	Short: "Extract the attachments of a PDF",
-	Long:  "Extract the attachments of a PDF and store them as file.",
+	Long:  "Extract the attachments of a PDF and store them as file.\n[input] can either be a file path or - for stdin.",
 	Args: func(cmd *cobra.Command, args []string) error {
 		if err := cobra.ExactArgs(2)(cmd, args); err != nil {
 			return err
 		}
 
-		if _, err := os.Stat(args[0]); err != nil {
+		if err := validFile(args[0]); err != nil {
 			return fmt.Errorf("could not open input file %s: %w\n", args[0], err)
 		}
 
@@ -48,19 +48,12 @@ var attachmentsCmd = &cobra.Command{
 		}
 		defer pdf.ClosePdfium()
 
-		file, err := os.Open(args[0])
+		document, closeFile, err := openFile(args[0])
 		if err != nil {
 			cmd.PrintErrf("could not open input file %s: %w\n", args[0], err)
 			return
 		}
-		defer file.Close()
-
-		document, err := openFile(file)
-		if err != nil {
-			cmd.PrintErrf("could not open input file %s: %w\n", args[0], err)
-			return
-		}
-		defer pdf.PdfiumInstance.FPDF_CloseDocument(&requests.FPDF_CloseDocument{Document: document.Document})
+		defer closeFile()
 
 		attachments, err := pdf.PdfiumInstance.GetAttachments(&requests.GetAttachments{
 			Document: document.Document,

cmd/explode.go

@@ -2,6 +2,7 @@ package cmd
 
 import (
 	"fmt"
+	"io"
 	"os"
 	"strings"
 
@@ -21,17 +22,17 @@ func init() {
 var explodeCmd = &cobra.Command{
 	Use:   "explode [input] [output]",
 	Short: "Explode a PDF into multiple PDFs",
-	Long:  "Explode a PDF into multiple PDFs, the output filename should contain a \"%d\" placeholder for the page number, e.g. split invoice.pdf invoice-%d.pdf, the result for a 2-page PDF will be invoice-1.pdf and invoice-2.pdf.",
+	Long:  "Explode a PDF into multiple PDFs.\n[input] can either be a file path or - for stdin.\n[output] can either be a file path or - for stdout, stdout is only supported when the output will be one file. The output filename should contain a \"%d\" placeholder for the page number, e.g. split invoice.pdf invoice-%d.pdf, the result for a 2-page PDF will be invoice-1.pdf and invoice-2.pdf.",
 	Args: func(cmd *cobra.Command, args []string) error {
 		if err := cobra.ExactArgs(2)(cmd, args); err != nil {
 			return err
 		}
 
-		if _, err := os.Stat(args[0]); err != nil {
+		if err := validFile(args[0]); err != nil {
 			return fmt.Errorf("could not open input file %s: %w\n", args[0], err)
 		}
 
-		if !strings.Contains(args[1], "%d") {
+		if args[1] != stdFilename && !strings.Contains(args[1], "%d") {
 			return fmt.Errorf("output string %s should contain page pattern %%d\n", args[1])
 		}
 
@@ -45,19 +46,12 @@ var explodeCmd = &cobra.Command{
 		}
 		defer pdf.ClosePdfium()
 
-		file, err := os.Open(args[0])
+		document, closeFile, err := openFile(args[0])
 		if err != nil {
 			cmd.PrintErrf("could not open input file %s: %w\n", args[0], err)
 			return
 		}
-		defer file.Close()
-
-		document, err := openFile(file)
-		if err != nil {
-			cmd.PrintErrf("could not open input file %s: %w\n", args[0], err)
-			return
-		}
-		defer pdf.PdfiumInstance.FPDF_CloseDocument(&requests.FPDF_CloseDocument{Document: document.Document})
+		defer closeFile()
 
 		pageCount, err := pdf.PdfiumInstance.FPDF_GetPageCount(&requests.FPDF_GetPageCount{
 			Document: document.Document,
@@ -79,6 +73,12 @@ var explodeCmd = &cobra.Command{
 		}
 
 		splitPages := strings.Split(*parsedPageRange, ",")
+
+		if args[1] == stdFilename && len(splitPages) > 1 {
+			cmd.PrintErrf("could not explode into multiple pages with output to stdout\n")
+			return
+		}
+
 		for _, page := range splitPages {
 			newDocument, err := pdf.PdfiumInstance.FPDF_CreateNewDocument(&requests.FPDF_CreateNewDocument{})
 			if err != nil {
@@ -102,27 +102,39 @@ var explodeCmd = &cobra.Command{
 			}
 
 			newFilePath := strings.Replace(args[1], "%d", page, -1)
-			createdFile, err := os.Create(newFilePath)
-			if err != nil {
-				cmd.PrintErrf("could not save document for page %s: %w\n", page, err)
-				return
+
+			var fileWriter io.Writer
+			if args[1] == stdFilename {
+				fileWriter = os.Stdout
+			} else {
+				createdFile, err := os.Create(newFilePath)
+				if err != nil {
+					cmd.PrintErrf("could not save document for page %s: %w\n", page, err)
+					return
+				}
+				fileWriter = createdFile
+				originalCloseFunc := closeFunc
+				closeFunc = func() {
+					originalCloseFunc()
+					createdFile.Close()
+				}
 			}
 
 			_, err = pdf.PdfiumInstance.FPDF_SaveAsCopy(&requests.FPDF_SaveAsCopy{
 				Document:   newDocument.Document,
-				FileWriter: createdFile,
+				FileWriter: fileWriter,
 			})
 			if err != nil {
 				closeFunc()
-				createdFile.Close()
 				cmd.PrintErrf("could not save document for page %s: %w\n", page, err)
 				return
 			}
 
 			closeFunc()
-			createdFile.Close()
 
-			cmd.Printf("Exploded page %s into %s\n", page, newFilePath)
+			if args[1] != stdFilename {
+				cmd.Printf("Exploded page %s into %s\n", page, newFilePath)
+			}
 		}
 	},
 }

cmd/images.go

@@ -34,13 +34,13 @@ func init() {
 var imagesCmd = &cobra.Command{
 	Use:   "images [input] [output-folder]",
 	Short: "Extract the images of a PDF",
-	Long:  "Extract the images of a PDF and store them as file.",
+	Long:  "Extract the images of a PDF and store them as file.\n[input] can either be a file path or - for stdin.",
 	Args: func(cmd *cobra.Command, args []string) error {
 		if err := cobra.ExactArgs(2)(cmd, args); err != nil {
 			return err
 		}
 
-		if _, err := os.Stat(args[0]); err != nil {
+		if err := validFile(args[0]); err != nil {
 			return fmt.Errorf("could not open input file %s: %w\n", args[0], err)
 		}
 
@@ -63,19 +63,12 @@ var imagesCmd = &cobra.Command{
 		}
 		defer pdf.ClosePdfium()
 
-		file, err := os.Open(args[0])
+		document, closeFile, err := openFile(args[0])
 		if err != nil {
 			cmd.PrintErrf("could not open input file %s: %w\n", args[0], err)
 			return
 		}
-		defer file.Close()
-
-		document, err := openFile(file)
-		if err != nil {
-			cmd.PrintErrf("could not open input file %s: %w\n", args[0], err)
-			return
-		}
-		defer pdf.PdfiumInstance.FPDF_CloseDocument(&requests.FPDF_CloseDocument{Document: document.Document})
+		defer closeFile()
 
 		pageCount, err := pdf.PdfiumInstance.FPDF_GetPageCount(&requests.FPDF_GetPageCount{
 			Document: document.Document,

cmd/info.go

@@ -2,7 +2,6 @@ package cmd
 
 import (
 	"fmt"
-	"os"
 	"strings"
 
 	"github.com/klippa-app/pdfium-cli/pdf"
@@ -25,7 +24,7 @@ var infoCmd = &cobra.Command{
 			return err
 		}
 
-		if _, err := os.Stat(args[0]); err != nil {
+		if err := validFile(args[0]); err != nil {
 			return fmt.Errorf("could not open input file %s: %w\n", args[0], err)
 		}
 
@@ -39,18 +38,14 @@ var infoCmd = &cobra.Command{
 		}
 		defer pdf.ClosePdfium()
 
-		file, err := os.Open(args[0])
+		document, closeFile, err := openFile(args[0])
 		if err != nil {
 			cmd.PrintErrf("could not open input file %s: %w\n", args[0], err)
 			return
 		}
-		defer file.Close()
 
-		document, err := openFile(file)
-		if err != nil {
-			cmd.PrintErrf("could not open input file %s: %w\n", args[0], err)
-			return
-		}
+		defer closeFile()
+
 		defer pdf.PdfiumInstance.FPDF_CloseDocument(&requests.FPDF_CloseDocument{Document: document.Document})
 
 		fileVersion, err := pdf.PdfiumInstance.FPDF_GetFileVersion(&requests.FPDF_GetFileVersion{

cmd/javascripts.go

@@ -19,13 +19,13 @@ func init() {
 var javascriptsCmd = &cobra.Command{
 	Use:   "javascripts [input] [output-folder]",
 	Short: "Extract the javascripts of a PDF",
-	Long:  "Extract the javascripts of a PDF and store them as file.",
+	Long:  "Extract the javascripts of a PDF and store them as file.\n[input] can either be a file path or - for stdin.",
 	Args: func(cmd *cobra.Command, args []string) error {
 		if err := cobra.ExactArgs(2)(cmd, args); err != nil {
 			return err
 		}
 
-		if _, err := os.Stat(args[0]); err != nil {
+		if err := validFile(args[0]); err != nil {
 			return fmt.Errorf("could not open input file %s: %w\n", args[0], err)
 		}
 
@@ -48,19 +48,12 @@ var javascriptsCmd = &cobra.Command{
 		}
 		defer pdf.ClosePdfium()
 
-		file, err := os.Open(args[0])
+		document, closeFile, err := openFile(args[0])
 		if err != nil {
 			cmd.PrintErrf("could not open input file %s: %w\n", args[0], err)
 			return
 		}
-		defer file.Close()
-
-		document, err := openFile(file)
-		if err != nil {
-			cmd.PrintErrf("could not open input file %s: %w\n", args[0], err)
-			return
-		}
-		defer pdf.PdfiumInstance.FPDF_CloseDocument(&requests.FPDF_CloseDocument{Document: document.Document})
+		defer closeFile()
 
 		javascripts, err := pdf.PdfiumInstance.GetJavaScriptActions(&requests.GetJavaScriptActions{
 			Document: document.Document,

cmd/merge.go

@@ -2,6 +2,7 @@ package cmd
 
 import (
 	"fmt"
+	"io"
 	"os"
 
 	"github.com/klippa-app/go-pdfium/requests"
@@ -18,7 +19,7 @@ func init() {
 var mergeCmd = &cobra.Command{
 	Use:   "merge [input] [input] ([input]...) [output]",
 	Short: "Merge multiple PDFs into a single PDF",
-	Long:  "Merge multiple PDFs into a single PDF",
+	Long:  "Merge multiple PDFs into a single PDF.\n[output] can either be a file path or - for stdout.",
 	Args: func(cmd *cobra.Command, args []string) error {
 		if err := cobra.MinimumNArgs(3)(cmd, args); err != nil {
 			return err
@@ -48,21 +49,14 @@ var mergeCmd = &cobra.Command{
 
 		mergedPageCount := 0
 		for i := 0; i < len(args)-1; i++ {
-			file, err := os.Open(args[i])
-			if err != nil {
-				cmd.PrintErrf("could not open input file %s: %w", args[i], err)
-				return
-			}
-			defer file.Close()
-
-			document, err := openFile(file)
+			document, closeFile, err := openFile(args[i])
 			if err != nil {
 				cmd.PrintErrf("could not open input file %s: %w", args[i], err)
 				return
 			}
 
 			closeFunc := func() {
-				pdf.PdfiumInstance.FPDF_CloseDocument(&requests.FPDF_CloseDocument{Document: newDocument.Document})
+				closeFile()
 			}
 
 			pageCount, err := pdf.PdfiumInstance.FPDF_GetPageCount(&requests.FPDF_GetPageCount{
@@ -107,17 +101,23 @@ var mergeCmd = &cobra.Command{
 			closeFunc()
 		}
 
-		createdFile, err := os.Create(args[len(args)-1])
-		if err != nil {
-			cmd.PrintErrf("could not save document: %w", err)
-			return
-		}
+		var fileWriter io.Writer
+		if args[len(args)-1] == stdFilename {
+			fileWriter = os.Stdout
+		} else {
+			createdFile, err := os.Create(args[len(args)-1])
+			if err != nil {
+				cmd.PrintErrf("could not save document: %w", err)
+				return
+			}
 
-		defer createdFile.Close()
+			defer createdFile.Close()
+			fileWriter = createdFile
+		}
 
 		_, err = pdf.PdfiumInstance.FPDF_SaveAsCopy(&requests.FPDF_SaveAsCopy{
 			Document:   newDocument.Document,
-			FileWriter: createdFile,
+			FileWriter: fileWriter,
 		})
 		if err != nil {
 			cmd.PrintErrf("could not save new document %s: %w", err)