First working version

- Move from separate dirs for upload to a unified one (identify
  media/metadata files by file extension)
- Prevent uploading when an import is already scheduled
- Allow setting custom, parser-specific settings
- Add CLI
- Implement WebDAV upload
- Implement checking of upload folders for uploadable contents

Close #6, close #7, close #9, close #3, close #1, close #4
This commit is contained in:
2025-02-27 17:29:20 +01:00
parent 9a5c432186
commit 7cfd3bb1de
12 changed files with 553 additions and 72 deletions
+115
View File
@@ -0,0 +1,115 @@
package webdavupload
import (
"os"
"sync"
"time"
"github.com/studio-b12/gowebdav"
"gitea.armuli.eu/museum-digital/museum-digital-webdav-uploader/src/configloader"
)
// Takes a path on the webdav remote and reads folder contents from it.
func listWebDavFolderContents(config configloader.MDWebDavUploaderConfig, path string) ([]os.FileInfo, error) {
c := gowebdav.NewClient(getWebDavHost(config), config.Username, config.WebDavAuthToken)
c.Connect()
files, err := c.ReadDir(path)
return files, err
}
// Returns a list of the files on the top level of the remote.
func ListTopLevelContents(config configloader.MDWebDavUploaderConfig) ([]os.FileInfo, error) {
return listWebDavFolderContents(config, ".")
}
// Returns a list of the files in the current metadata dir of the remote.
func ListMetadataDir(config configloader.MDWebDavUploaderConfig) ([]os.FileInfo, error) {
return listWebDavFolderContents(config, "./IMPORT_XML")
}
// Returns a list of the files in the current media dir of the remote.
func ListMediaDir(config configloader.MDWebDavUploaderConfig) ([]os.FileInfo, error) {
return listWebDavFolderContents(config, "./IMPORT_IMG")
}
func checkImportConfigExists(c *gowebdav.Client) bool {
_, err := c.Stat("./import_config.txt")
if gowebdav.IsErrNotFound(err) == true {
return false
}
return true
}
// Checks if a remote directory is free for updating:
// This means it can be read, and it does not contain files more recent than
// 10 minutes.
func checkRemoteDirIsFree(c *gowebdav.Client, path string) bool {
now := time.Now()
// Check metadata dir
files, err := c.ReadDir(path)
if err != nil {
print("Failed to load remote contents")
return false
}
for _, f := range(files) {
if diff := now.Sub(f.ModTime()); diff < 10 * time.Minute {
print("File " + f.Name() + " has been recently renamed")
return false
}
}
return true
}
// Checks if either the metadata or the image folder contains files that have
// been changed in the last 10 minutes.
// In this case, it can be assumed that there is a concurrent upload taking place.
func checkRemoteImportFilesAreTooRecent(c *gowebdav.Client) bool {
var wg sync.WaitGroup
var metadataDirFree bool
var mediaDirFree bool
wg.Add(2)
go func() {
defer wg.Done()
metadataDirFree = checkRemoteDirIsFree(c, "./IMPORT_XML")
}()
go func() {
defer wg.Done()
mediaDirFree = checkRemoteDirIsFree(c, "./IMPORT_IMG")
}()
wg.Wait()
if metadataDirFree == true && mediaDirFree == true {
return false
}
return true
}
// Checks the remote for the existence of an import_config.txt.
func CheckRemoteIsFree(c *gowebdav.Client) bool {
if checkImportConfigExists(c) == true {
return false
}
// Check if the remote media and metadata directories currently
// contain very recent files indicating a concurrent upload.
if checkRemoteImportFilesAreTooRecent(c) == true {
return false
}
return true
}
+35
View File
@@ -0,0 +1,35 @@
package webdavupload
import (
"net/url"
"strconv"
"github.com/studio-b12/gowebdav"
"gitea.armuli.eu/museum-digital/museum-digital-webdav-uploader/src/configloader"
)
// Generates the path for a webdav endpoint for the configured instance and institution.
func getWebDavHost(config configloader.MDWebDavUploaderConfig) string {
parsed, err := url.Parse(config.InstanceLink)
if err != nil {
panic("Your settings must be broken. Failed to load instance URL.")
}
output := url.URL{
Scheme: "https",
Host: parsed.Host,
Path: "/musdb/webdav/" + strconv.Itoa(config.InstitutionId),
}
return output.String()
}
// Returns a WebDAV client for the current config.
func GetWebdavClient(config configloader.MDWebDavUploaderConfig) *gowebdav.Client {
c := gowebdav.NewClient(getWebDavHost(config), config.Username, config.WebDavAuthToken)
c.Connect()
return c
}
+84
View File
@@ -0,0 +1,84 @@
package webdavupload
import (
"fmt"
"path/filepath"
"os"
"runtime"
"sync"
"sync/atomic"
"github.com/studio-b12/gowebdav"
"gitea.armuli.eu/museum-digital/museum-digital-webdav-uploader/src/configloader"
"gitea.armuli.eu/museum-digital/museum-digital-webdav-uploader/src/importconfiggen"
)
// Writes an import config to the remote
func SetImportConfigToTheRemote(c *gowebdav.Client, config configloader.MDWebDavUploaderConfig) error {
importConf := importconfiggen.GenerateImportConfig(config)
return c.Write("import_config.txt", []byte(importConf), 0660)
}
// Uploads a list of files to the target folder.
func uploadFiles(c *gowebdav.Client, files []string, remoteTarget string, outputContext string) {
total := len(files)
var counter atomic.Uint64
// Determine the number of upload tasks to be processed concurrently.
// 10 will be a hard maximum to not spam the server.
maxConcTasks := min(10, runtime.NumCPU())
// Set a semaphore to restrict the number of concurrent upload tasks.
semaphore := make(chan struct{}, maxConcTasks)
wg := &sync.WaitGroup{}
fmt.Printf("Will upload %v files. Processing %v tasks at a time.\n", total, maxConcTasks)
for _, f := range(files) {
semaphore <- struct{}{} // acquire
wg.Add(1)
go func() {
defer wg.Done()
basename := filepath.Base(f)
file, fOpenErr := os.Open(f)
if fOpenErr != nil {
panic("Failed to read file: " + f)
}
defer file.Close()
c.WriteStream("./" + remoteTarget + "/" + basename, file, 0644)
counter.Add(1)
fmt.Printf(outputContext + ". File %v of %v. Done. (File: %v)\n", counter, total, basename)
<-semaphore // release
}()
}
wg.Wait()
fmt.Println("Done")
}
// Uploads the selected metadata files.
func UploadMetadataFiles(c *gowebdav.Client, files []string) {
uploadFiles(c, files, "IMPORT_XML", "Uploading metadata files")
}
// Uploads the selected media files.
func UploadMediaFiles(c *gowebdav.Client, files []string) {
uploadFiles(c, files, "IMPORT_IMG", "Uploading media files")
}