fix(pdfcpu): switch to CLI

This commit is contained in:
Julien Neuhart
2024-10-06 18:32:09 +02:00
parent aa57b17254
commit bf205c579d
11 changed files with 127 additions and 53 deletions

View File

@@ -1,5 +1,5 @@
// Package pdfcpu provides an implementation of the gotenberg.PdfEngine
// interface using the pdfcpu library. This package allows for:
// interface using the pdfcpu command-line tool. This package allows for:
//
// 1. The merging of PDF files.
//

View File

@@ -2,11 +2,10 @@ package pdfcpu
import (
"context"
"errors"
"fmt"
"os"
pdfcpuAPI "github.com/pdfcpu/pdfcpu/pkg/api"
pdfcpuLog "github.com/pdfcpu/pdfcpu/pkg/log"
pdfcpuConfig "github.com/pdfcpu/pdfcpu/pkg/pdfcpu/model"
"go.uber.org/zap"
"github.com/gotenberg/gotenberg/v8/pkg/gotenberg"
@@ -16,10 +15,10 @@ func init() {
gotenberg.MustRegisterModule(new(PdfCpu))
}
// PdfCpu abstracts the pdfcpu library and implements the [gotenberg.PdfEngine]
// interface.
// PdfCpu abstracts the CLI tool pdfcpu and implements the
// [gotenberg.PdfEngine] interface.
type PdfCpu struct {
conf *pdfcpuConfig.Configuration
binPath string
}
// Descriptor returns a [PdfCpu]'s module descriptor.
@@ -32,16 +31,38 @@ func (engine *PdfCpu) Descriptor() gotenberg.ModuleDescriptor {
// Provision sets the engine properties.
func (engine *PdfCpu) Provision(ctx *gotenberg.Context) error {
pdfcpuConfig.ConfigPath = "disable"
pdfcpuLog.DisableLoggers()
engine.conf = pdfcpuConfig.NewDefaultConfiguration()
binPath, ok := os.LookupEnv("PDFCPU_BIN_PATH")
if !ok {
return errors.New("PDFCPU_BIN_PATH environment variable is not set")
}
engine.binPath = binPath
return nil
}
// Validate validates the module properties.
func (engine *PdfCpu) Validate() error {
_, err := os.Stat(engine.binPath)
if os.IsNotExist(err) {
return fmt.Errorf("pdfcpu binary path does not exist: %w", err)
}
return nil
}
// Merge combines multiple PDFs into a single PDF.
func (engine *PdfCpu) Merge(ctx context.Context, logger *zap.Logger, inputPaths []string, outputPath string) error {
err := pdfcpuAPI.MergeCreateFile(inputPaths, outputPath, false, engine.conf)
var args []string
args = append(args, "merge", outputPath)
args = append(args, inputPaths...)
cmd, err := gotenberg.CommandContext(ctx, logger, engine.binPath, args...)
if err != nil {
return fmt.Errorf("create command: %w", err)
}
_, err = cmd.Exec()
if err == nil {
return nil
}
@@ -68,5 +89,6 @@ func (engine *PdfCpu) WriteMetadata(ctx context.Context, logger *zap.Logger, met
var (
_ gotenberg.Module = (*PdfCpu)(nil)
_ gotenberg.Provisioner = (*PdfCpu)(nil)
_ gotenberg.Validator = (*PdfCpu)(nil)
_ gotenberg.PdfEngine = (*PdfCpu)(nil)
)

View File

@@ -33,14 +33,59 @@ func TestPdfCpu_Provision(t *testing.T) {
}
}
func TestPdfCpu_Validate(t *testing.T) {
for _, tc := range []struct {
scenario string
binPath string
expectError bool
}{
{
scenario: "empty bin path",
binPath: "",
expectError: true,
},
{
scenario: "bin path does not exist",
binPath: "/foo",
expectError: true,
},
{
scenario: "validate success",
binPath: os.Getenv("PDFTK_BIN_PATH"),
expectError: false,
},
} {
t.Run(tc.scenario, func(t *testing.T) {
engine := new(PdfCpu)
engine.binPath = tc.binPath
err := engine.Validate()
if !tc.expectError && err != nil {
t.Fatalf("expected no error but got: %v", err)
}
if tc.expectError && err == nil {
t.Fatal("expected error but got none")
}
})
}
}
func TestPdfCpu_Merge(t *testing.T) {
for _, tc := range []struct {
scenario string
ctx context.Context
inputPaths []string
expectError bool
}{
{
scenario: "invalid context",
ctx: nil,
expectError: true,
},
{
scenario: "invalid input path",
ctx: context.TODO(),
inputPaths: []string{
"foo",
},
@@ -48,6 +93,7 @@ func TestPdfCpu_Merge(t *testing.T) {
},
{
scenario: "single file success",
ctx: context.TODO(),
inputPaths: []string{
"/tests/test/testdata/pdfengines/sample1.pdf",
},
@@ -55,10 +101,12 @@ func TestPdfCpu_Merge(t *testing.T) {
},
{
scenario: "many files success",
ctx: context.TODO(),
inputPaths: []string{
"/tests/test/testdata/pdfengines/sample1.pdf",
"/tests/test/testdata/pdfengines/sample2.pdf",
},
expectError: false,
},
} {
t.Run(tc.scenario, func(t *testing.T) {
@@ -81,7 +129,7 @@ func TestPdfCpu_Merge(t *testing.T) {
}
}()
err = engine.Merge(nil, nil, tc.inputPaths, outputDir+"/foo.pdf")
err = engine.Merge(tc.ctx, zap.NewNop(), tc.inputPaths, outputDir+"/foo.pdf")
if !tc.expectError && err != nil {
t.Fatalf("expected no error but got: %v", err)

View File

@@ -45,7 +45,7 @@ func (engine *PdfTk) Provision(ctx *gotenberg.Context) error {
func (engine *PdfTk) Validate() error {
_, err := os.Stat(engine.binPath)
if os.IsNotExist(err) {
return fmt.Errorf("PdfTk binary path does not exist: %w", err)
return fmt.Errorf("PDFtk binary path does not exist: %w", err)
}
return nil

View File

@@ -97,6 +97,7 @@ func TestPdfTk_Merge(t *testing.T) {
inputPaths: []string{
"/tests/test/testdata/pdfengines/sample1.pdf",
},
expectError: false,
},
{
scenario: "many files success",
@@ -105,6 +106,7 @@ func TestPdfTk_Merge(t *testing.T) {
"/tests/test/testdata/pdfengines/sample1.pdf",
"/tests/test/testdata/pdfengines/sample2.pdf",
},
expectError: false,
},
} {
t.Run(tc.scenario, func(t *testing.T) {

View File

@@ -97,6 +97,7 @@ func TestQPdf_Merge(t *testing.T) {
inputPaths: []string{
"/tests/test/testdata/pdfengines/sample1.pdf",
},
expectError: false,
},
{
scenario: "many files success",
@@ -105,6 +106,7 @@ func TestQPdf_Merge(t *testing.T) {
"/tests/test/testdata/pdfengines/sample1.pdf",
"/tests/test/testdata/pdfengines/sample2.pdf",
},
expectError: false,
},
} {
t.Run(tc.scenario, func(t *testing.T) {