adalal/gotenberg
1
0
Fork 0
mirror of https://github.com/gotenberg/gotenberg.git synced 2026-08-17 20:52:14 +01:00
Code Issues Packages Projects Releases Wiki Activity

feat(pdfengines): add split feature

Browse Source
This commit is contained in:
Julien Neuhart
2024-12-20 15:51:57 +01:00
parent 42ee593708
commit c30da805b3
41 changed files with 2153 additions and 346 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
pkg/modules/libreoffice/api/libreoffice.go
View File

@@ -44,7 +44,7 @@ type libreOfficeProcess struct {
func newLibreOfficeProcess(arguments libreOfficeArguments) libreOffice {
p := &libreOfficeProcess{
arguments: arguments,
fs: gotenberg.NewFileSystem(),
fs: gotenberg.NewFileSystem(new(gotenberg.OsMkdirAll)),
}
p.isStarted.Store(false)

26
pkg/modules/libreoffice/api/libreoffice_test.go
View File

@@ -230,7 +230,7 @@ func TestLibreOfficeProcess_pdf(t *testing.T) {
p.isStarted.Store(false)
return p
}(),
fs: gotenberg.NewFileSystem(),
fs: gotenberg.NewFileSystem(new(gotenberg.OsMkdirAll)),
cancelledCtx: false,
start: false,
expectError: true,
@@ -243,7 +243,7 @@ func TestLibreOfficeProcess_pdf(t *testing.T) {
p.isStarted.Store(true)
return p
}(),
fs: gotenberg.NewFileSystem(),
fs: gotenberg.NewFileSystem(new(gotenberg.OsMkdirAll)),
options: Options{PdfFormats: gotenberg.PdfFormats{PdfA: "foo"}},
cancelledCtx: false,
start: false,
@@ -261,7 +261,7 @@ func TestLibreOfficeProcess_pdf(t *testing.T) {
),
options: Options{PageRanges: "foo"},
fs: func() *gotenberg.FileSystem {
fs := gotenberg.NewFileSystem()
fs := gotenberg.NewFileSystem(new(gotenberg.OsMkdirAll))
err := os.MkdirAll(fs.WorkingDirPath(), 0o755)
if err != nil {
@@ -291,7 +291,7 @@ func TestLibreOfficeProcess_pdf(t *testing.T) {
),
options: Options{Password: "foo"},
fs: func() *gotenberg.FileSystem {
fs := gotenberg.NewFileSystem()
fs := gotenberg.NewFileSystem(new(gotenberg.OsMkdirAll))
err := os.MkdirAll(fs.WorkingDirPath(), 0o755)
if err != nil {
@@ -344,7 +344,7 @@ func TestLibreOfficeProcess_pdf(t *testing.T) {
},
),
fs: func() *gotenberg.FileSystem {
fs := gotenberg.NewFileSystem()
fs := gotenberg.NewFileSystem(new(gotenberg.OsMkdirAll))
err := os.MkdirAll(fs.WorkingDirPath(), 0o755)
if err != nil {
@@ -372,7 +372,7 @@ func TestLibreOfficeProcess_pdf(t *testing.T) {
},
),
fs: func() *gotenberg.FileSystem {
fs := gotenberg.NewFileSystem()
fs := gotenberg.NewFileSystem(new(gotenberg.OsMkdirAll))
err := os.MkdirAll(fs.WorkingDirPath(), 0o755)
if err != nil {
@@ -400,7 +400,7 @@ func TestLibreOfficeProcess_pdf(t *testing.T) {
},
),
fs: func() *gotenberg.FileSystem {
fs := gotenberg.NewFileSystem()
fs := gotenberg.NewFileSystem(new(gotenberg.OsMkdirAll))
err := os.MkdirAll(fs.WorkingDirPath(), 0o755)
if err != nil {
@@ -452,7 +452,7 @@ func TestLibreOfficeProcess_pdf(t *testing.T) {
},
),
fs: func() *gotenberg.FileSystem {
fs := gotenberg.NewFileSystem()
fs := gotenberg.NewFileSystem(new(gotenberg.OsMkdirAll))
err := os.MkdirAll(fs.WorkingDirPath(), 0o755)
if err != nil {
@@ -481,7 +481,7 @@ func TestLibreOfficeProcess_pdf(t *testing.T) {
},
),
fs: func() *gotenberg.FileSystem {
fs := gotenberg.NewFileSystem()
fs := gotenberg.NewFileSystem(new(gotenberg.OsMkdirAll))
err := os.MkdirAll(fs.WorkingDirPath(), 0o755)
if err != nil {
@@ -510,7 +510,7 @@ func TestLibreOfficeProcess_pdf(t *testing.T) {
},
),
fs: func() *gotenberg.FileSystem {
fs := gotenberg.NewFileSystem()
fs := gotenberg.NewFileSystem(new(gotenberg.OsMkdirAll))
err := os.MkdirAll(fs.WorkingDirPath(), 0o755)
if err != nil {
@@ -539,7 +539,7 @@ func TestLibreOfficeProcess_pdf(t *testing.T) {
},
),
fs: func() *gotenberg.FileSystem {
fs := gotenberg.NewFileSystem()
fs := gotenberg.NewFileSystem(new(gotenberg.OsMkdirAll))
err := os.MkdirAll(fs.WorkingDirPath(), 0o755)
if err != nil {
@@ -625,7 +625,7 @@ func TestNonBasicLatinCharactersGuard(t *testing.T) {
{
scenario: "basic latin characters",
fs: func() *gotenberg.FileSystem {
fs := gotenberg.NewFileSystem()
fs := gotenberg.NewFileSystem(new(gotenberg.OsMkdirAll))
err := os.MkdirAll(fs.WorkingDirPath(), 0o755)
if err != nil {
@@ -646,7 +646,7 @@ func TestNonBasicLatinCharactersGuard(t *testing.T) {
{
scenario: "non-basic latin characters",
fs: func() *gotenberg.FileSystem {
fs := gotenberg.NewFileSystem()
fs := gotenberg.NewFileSystem(new(gotenberg.OsMkdirAll))
err := os.MkdirAll(fs.WorkingDirPath(), 0o755)
if err != nil {

5
pkg/modules/libreoffice/pdfengine/pdfengine.go
View File

@@ -51,6 +51,11 @@ func (engine *LibreOfficePdfEngine) Merge(ctx context.Context, logger *zap.Logge
return fmt.Errorf("merge PDFs with LibreOffice: %w", gotenberg.ErrPdfEngineMethodNotSupported)
}
// Split is not available in this implementation.
func (engine *LibreOfficePdfEngine) Split(ctx context.Context, logger *zap.Logger, mode gotenberg.SplitMode, inputPath, outputDirPath string) ([]string, error) {
return nil, fmt.Errorf("split PDF with LibreOffice: %w", gotenberg.ErrPdfEngineMethodNotSupported)
}
// Convert converts the given PDF to a specific PDF format. Currently, only the
// PDF/A-1b, PDF/A-2b, PDF/A-3b and PDF/UA formats are available. If another
// PDF format is requested, it returns a [gotenberg.ErrPdfFormatNotSupported]

25
pkg/modules/libreoffice/pdfengine/pdfengine_test.go
View File

@@ -118,11 +118,21 @@ func TestLibreOfficePdfEngine_Merge(t *testing.T) {
}
}
func TestLibreOfficePdfEngine_Split(t *testing.T) {
engine := new(LibreOfficePdfEngine)
_, err := engine.Split(context.Background(), zap.NewNop(), gotenberg.SplitMode{}, "", "")
if !errors.Is(err, gotenberg.ErrPdfEngineMethodNotSupported) {
t.Errorf("expected error %v, but got: %v", gotenberg.ErrPdfEngineMethodNotSupported, err)
}
}
func TestLibreOfficePdfEngine_Convert(t *testing.T) {
for _, tc := range []struct {
scenario string
api api.Uno
expectError bool
scenario string
api api.Uno
expectError bool
expectedError error
}{
{
scenario: "convert success",
@@ -134,13 +144,14 @@ func TestLibreOfficePdfEngine_Convert(t *testing.T) {
expectError: false,
},
{
scenario: "invalid PDF format",
scenario: "ErrInvalidPdfFormats",
api: &api.ApiMock{
PdfMock: func(ctx context.Context, logger *zap.Logger, inputPath, outputPath string, options api.Options) error {
return api.ErrInvalidPdfFormats
},
},
expectError: true,
expectError: true,
expectedError: gotenberg.ErrPdfFormatNotSupported,
},
{
scenario: "convert fail",
@@ -163,6 +174,10 @@ func TestLibreOfficePdfEngine_Convert(t *testing.T) {
if tc.expectError && err == nil {
t.Fatal("expected error but got none")
}
if tc.expectedError != nil && !errors.Is(err, tc.expectedError) {
t.Fatalf("expected error %v but got: %v", tc.expectedError, err)
}
})
}
}

48
pkg/modules/libreoffice/routes.go
View File

@@ -28,8 +28,11 @@ func convertRoute(libreOffice libreofficeapi.Uno, engine gotenberg.PdfEngine) ap
defaultOptions := libreofficeapi.DefaultOptions()
form := ctx.FormData()
splitMode := pdfengines.FormDataPdfSplitMode(form, false)
pdfFormats := pdfengines.FormDataPdfFormats(form)
metadata := pdfengines.FormDataPdfMetadata(form)
metadata := pdfengines.FormDataPdfMetadata(form, false)
zeroValuedSplitMode := gotenberg.SplitMode{}
var (
inputPaths []string
@@ -165,7 +168,9 @@ func convertRoute(libreOffice libreofficeapi.Uno, engine gotenberg.PdfEngine) ap
MaxImageResolution: maxImageResolution,
}
if nativePdfFormats {
if nativePdfFormats && splitMode == zeroValuedSplitMode {
// Only apply natively given PDF formats if we're not
// splitting the PDF later.
options.PdfFormats = pdfFormats
}
@@ -209,11 +214,44 @@ func convertRoute(libreOffice libreofficeapi.Uno, engine gotenberg.PdfEngine) ap
outputPaths = []string{outputPath}
}
if !nativePdfFormats {
outputPaths, err = pdfengines.ConvertStub(ctx, engine, pdfFormats, outputPaths)
if splitMode != zeroValuedSplitMode {
if !merge {
// document.docx -> document.docx.pdf, so that split naming
// document.docx_0.pdf, etc.
for i, inputPath := range inputPaths {
outputPath := fmt.Sprintf("%s.pdf", inputPath)
err = ctx.Rename(outputPaths[i], outputPath)
if err != nil {
return fmt.Errorf("rename output path: %w", err)
}
outputPaths[i] = outputPath
}
}
outputPaths, err = pdfengines.SplitPdfStub(ctx, engine, splitMode, outputPaths)
if err != nil {
return fmt.Errorf("split PDFs: %w", err)
}
}
if !nativePdfFormats || (nativePdfFormats && splitMode != zeroValuedSplitMode) {
convertOutputPaths, err := pdfengines.ConvertStub(ctx, engine, pdfFormats, outputPaths)
if err != nil {
return fmt.Errorf("convert PDFs: %w", err)
}
if splitMode != zeroValuedSplitMode {
// The PDF has been split and split parts have been converted to
// specific formats. We want to keep the split naming.
for i, convertOutputPath := range convertOutputPaths {
err = ctx.Rename(convertOutputPath, outputPaths[i])
if err != nil {
return fmt.Errorf("rename output path: %w", err)
}
}
}
}
err = pdfengines.WriteMetadataStub(ctx, engine, metadata, outputPaths)
@@ -221,7 +259,7 @@ func convertRoute(libreOffice libreofficeapi.Uno, engine gotenberg.PdfEngine) ap
return fmt.Errorf("write metadata: %w", err)
}
if len(outputPaths) > 1 {
if len(outputPaths) > 1 && splitMode == zeroValuedSplitMode {
// If .zip archive, document.docx -> document.docx.pdf.
for i, inputPath := range inputPaths {
outputPath := fmt.Sprintf("%s.pdf", inputPath)

227
pkg/modules/libreoffice/routes_test.go
View File

@@ -3,7 +3,10 @@ package libreoffice
import (
"context"
"errors"
"fmt"
"net/http"
"os"
"path/filepath"
"slices"
"testing"
@@ -300,6 +303,40 @@ func TestConvertRoute(t *testing.T) {
expectHttpError: false,
expectOutputPathsCount: 0,
},
{
scenario: "PDF engine split error",
ctx: func() *api.ContextMock {
ctx := &api.ContextMock{Context: new(api.Context)}
ctx.SetFiles(map[string]string{
"document.docx": "/document.docx",
})
ctx.SetValues(map[string][]string{
"splitMode": {
gotenberg.SplitModeIntervals,
},
"splitSpan": {
"1",
},
})
return ctx
}(),
libreOffice: &libreofficeapi.ApiMock{
PdfMock: func(ctx context.Context, logger *zap.Logger, inputPath, outputPath string, options libreofficeapi.Options) error {
return nil
},
ExtensionsMock: func() []string {
return []string{".docx"}
},
},
engine: &gotenberg.PdfEngineMock{
SplitMock: func(ctx context.Context, logger *zap.Logger, mode gotenberg.SplitMode, inputPath, outputDirPath string) ([]string, error) {
return nil, errors.New("foo")
},
},
expectError: true,
expectHttpError: false,
expectOutputPathsCount: 0,
},
{
scenario: "PDF engine convert error",
ctx: func() *api.ContextMock {
@@ -365,32 +402,6 @@ func TestConvertRoute(t *testing.T) {
expectHttpError: false,
expectOutputPathsCount: 0,
},
{
scenario: "cannot rename many files",
ctx: func() *api.ContextMock {
ctx := &api.ContextMock{Context: new(api.Context)}
ctx.SetFiles(map[string]string{
"document.docx": "/document.docx",
"document2.docx": "/document2.docx",
"document2.doc": "/document2.doc",
})
ctx.SetPathRename(&gotenberg.PathRenameMock{RenameMock: func(oldpath, newpath string) error {
return errors.New("cannot rename")
}})
return ctx
}(),
libreOffice: &libreofficeapi.ApiMock{
PdfMock: func(ctx context.Context, logger *zap.Logger, inputPath, outputPath string, options libreofficeapi.Options) error {
return nil
},
ExtensionsMock: func() []string {
return []string{".docx", ".doc"}
},
},
expectError: true,
expectHttpError: false,
expectOutputPathsCount: 0,
},
{
scenario: "cannot add output paths",
ctx: func() *api.ContextMock {
@@ -550,9 +561,173 @@ func TestConvertRoute(t *testing.T) {
expectHttpError: false,
expectOutputPathsCount: 1,
},
{
scenario: "success with split (many files)",
ctx: func() *api.ContextMock {
ctx := &api.ContextMock{Context: new(api.Context)}
ctx.SetFiles(map[string]string{
"document.docx": "/document.docx",
"document2.docx": "/document2.docx",
})
ctx.SetValues(map[string][]string{
"splitMode": {
gotenberg.SplitModeIntervals,
},
"splitSpan": {
"1",
},
})
return ctx
}(),
libreOffice: &libreofficeapi.ApiMock{
PdfMock: func(ctx context.Context, logger *zap.Logger, inputPath, outputPath string, options libreofficeapi.Options) error {
return nil
},
ExtensionsMock: func() []string {
return []string{".docx"}
},
},
engine: &gotenberg.PdfEngineMock{
MergeMock: func(ctx context.Context, logger *zap.Logger, inputPaths []string, outputPath string) error {
return nil
},
SplitMock: func(ctx context.Context, logger *zap.Logger, mode gotenberg.SplitMode, inputPath, outputDirPath string) ([]string, error) {
inputPathNoExt := inputPath[:len(inputPath)-len(filepath.Ext(inputPath))]
filenameNoExt := filepath.Base(inputPathNoExt)
return []string{
fmt.Sprintf(
"%s/%s_%d.pdf",
outputDirPath, filenameNoExt, 0,
),
fmt.Sprintf(
"%s/%s_%d.pdf",
outputDirPath, filenameNoExt, 1,
),
}, nil
},
},
expectError: false,
expectHttpError: false,
expectOutputPathsCount: 4,
expectOutputPaths: []string{"/document_docx/document.docx_0.pdf", "/document_docx/document.docx_1.pdf", "/document2_docx/document2.docx_0.pdf", "/document2_docx/document2.docx_1.pdf"},
},
{
scenario: "success with merge and split",
ctx: func() *api.ContextMock {
ctx := &api.ContextMock{Context: new(api.Context)}
ctx.SetFiles(map[string]string{
"document.docx": "/document.docx",
"document2.docx": "/document2.docx",
})
ctx.SetValues(map[string][]string{
"merge": {
"true",
},
"splitMode": {
gotenberg.SplitModeIntervals,
},
"splitSpan": {
"1",
},
})
return ctx
}(),
libreOffice: &libreofficeapi.ApiMock{
PdfMock: func(ctx context.Context, logger *zap.Logger, inputPath, outputPath string, options libreofficeapi.Options) error {
return nil
},
ExtensionsMock: func() []string {
return []string{".docx"}
},
},
engine: &gotenberg.PdfEngineMock{
MergeMock: func(ctx context.Context, logger *zap.Logger, inputPaths []string, outputPath string) error {
return nil
},
SplitMock: func(ctx context.Context, logger *zap.Logger, mode gotenberg.SplitMode, inputPath, outputDirPath string) ([]string, error) {
inputPathNoExt := inputPath[:len(inputPath)-len(filepath.Ext(inputPath))]
filenameNoExt := filepath.Base(inputPathNoExt)
return []string{
fmt.Sprintf(
"%s/%s_%d.pdf",
outputDirPath, filenameNoExt, 0,
),
fmt.Sprintf(
"%s/%s_%d.pdf",
outputDirPath, filenameNoExt, 1,
),
}, nil
},
},
expectError: false,
expectHttpError: false,
expectOutputPathsCount: 2,
},
{
scenario: "success with split and native PDF/A & PDF/UA (many files)",
ctx: func() *api.ContextMock {
ctx := &api.ContextMock{Context: new(api.Context)}
ctx.SetFiles(map[string]string{
"document.docx": "/document.docx",
"document2.docx": "/document2.docx",
})
ctx.SetValues(map[string][]string{
"splitMode": {
gotenberg.SplitModeIntervals,
},
"splitSpan": {
"1",
},
"pdfa": {
gotenberg.PdfA1b,
},
"pdfua": {
"true",
},
})
return ctx
}(),
libreOffice: &libreofficeapi.ApiMock{
PdfMock: func(ctx context.Context, logger *zap.Logger, inputPath, outputPath string, options libreofficeapi.Options) error {
return nil
},
ExtensionsMock: func() []string {
return []string{".docx"}
},
},
engine: &gotenberg.PdfEngineMock{
SplitMock: func(ctx context.Context, logger *zap.Logger, mode gotenberg.SplitMode, inputPath, outputDirPath string) ([]string, error) {
inputPathNoExt := inputPath[:len(inputPath)-len(filepath.Ext(inputPath))]
filenameNoExt := filepath.Base(inputPathNoExt)
return []string{
fmt.Sprintf(
"%s/%s_%d.pdf",
outputDirPath, filenameNoExt, 0,
),
fmt.Sprintf(
"%s/%s_%d.pdf",
outputDirPath, filenameNoExt, 1,
),
}, nil
},
ConvertMock: func(ctx context.Context, logger *zap.Logger, formats gotenberg.PdfFormats, inputPath, outputPath string) error {
return nil
},
},
expectError: false,
expectHttpError: false,
expectOutputPathsCount: 4,
expectOutputPaths: []string{"/document_docx/document.docx_0.pdf", "/document_docx/document.docx_1.pdf", "/document2_docx/document2.docx_0.pdf", "/document2_docx/document2.docx_1.pdf"},
},
} {
t.Run(tc.scenario, func(t *testing.T) {
tc.ctx.SetLogger(zap.NewNop())
tc.ctx.SetMkdirAll(&gotenberg.MkdirAllMock{MkdirAllMock: func(path string, perm os.FileMode) error {
return nil
}})
tc.ctx.SetPathRename(&gotenberg.PathRenameMock{RenameMock: func(oldpath, newpath string) error {
return nil
}})
c := echo.New().NewContext(nil, nil)
c.Set("context", tc.ctx.Context)