|
| 1 | +// Implementation for Dataverse |
| 2 | + |
| 3 | +package doi |
| 4 | + |
| 5 | +import ( |
| 6 | + "context" |
| 7 | + "fmt" |
| 8 | + "net/http" |
| 9 | + "net/url" |
| 10 | + "path" |
| 11 | + "strings" |
| 12 | + "time" |
| 13 | + |
| 14 | + "github.com/rclone/rclone/backend/doi/api" |
| 15 | + "github.com/rclone/rclone/fs" |
| 16 | + "github.com/rclone/rclone/lib/rest" |
| 17 | +) |
| 18 | + |
| 19 | +// Returns true if resolvedURL is likely a DOI hosted on a Dataverse intallation |
| 20 | +func activateDataverse(resolvedURL *url.URL) (isActive bool) { |
| 21 | + queryValues := resolvedURL.Query() |
| 22 | + persistentID := queryValues.Get("persistentId") |
| 23 | + return persistentID != "" |
| 24 | +} |
| 25 | + |
| 26 | +// Resolve the main API endpoint for a DOI hosted on a Dataverse installation |
| 27 | +func resolveDataverseEndpoint(resolvedURL *url.URL) (provider Provider, endpoint *url.URL, err error) { |
| 28 | + queryValues := resolvedURL.Query() |
| 29 | + persistentID := queryValues.Get("persistentId") |
| 30 | + |
| 31 | + query := url.Values{} |
| 32 | + query.Add("persistentId", persistentID) |
| 33 | + endpointURL := resolvedURL.ResolveReference(&url.URL{Path: "/api/datasets/:persistentId/", RawQuery: query.Encode()}) |
| 34 | + |
| 35 | + return Dataverse, endpointURL, nil |
| 36 | +} |
| 37 | + |
| 38 | +// dataverseProvider implements the doiProvider interface for Dataverse installations |
| 39 | +type dataverseProvider struct { |
| 40 | + f *Fs |
| 41 | +} |
| 42 | + |
| 43 | +// ListEntries returns the full list of entries found at the remote, regardless of root |
| 44 | +func (dp *dataverseProvider) ListEntries(ctx context.Context) (entries []*Object, err error) { |
| 45 | + // Use the cache if populated |
| 46 | + cachedEntries, found := dp.f.cache.GetMaybe("files") |
| 47 | + if found { |
| 48 | + parsedEntries, ok := cachedEntries.([]Object) |
| 49 | + if ok { |
| 50 | + for _, entry := range parsedEntries { |
| 51 | + newEntry := entry |
| 52 | + entries = append(entries, &newEntry) |
| 53 | + } |
| 54 | + return entries, nil |
| 55 | + } |
| 56 | + } |
| 57 | + |
| 58 | + filesURL := dp.f.endpoint |
| 59 | + var res *http.Response |
| 60 | + var result api.DataverseDatasetResponse |
| 61 | + opts := rest.Opts{ |
| 62 | + Method: "GET", |
| 63 | + Path: strings.TrimLeft(filesURL.EscapedPath(), "/"), |
| 64 | + Parameters: filesURL.Query(), |
| 65 | + } |
| 66 | + err = dp.f.pacer.Call(func() (bool, error) { |
| 67 | + res, err = dp.f.srv.CallJSON(ctx, &opts, nil, &result) |
| 68 | + return shouldRetry(ctx, res, err) |
| 69 | + }) |
| 70 | + if err != nil { |
| 71 | + return nil, fmt.Errorf("readDir failed: %w", err) |
| 72 | + } |
| 73 | + modTime, modTimeErr := time.Parse(time.RFC3339, result.Data.LatestVersion.LastUpdateTime) |
| 74 | + if modTimeErr != nil { |
| 75 | + fs.Logf(dp.f, "error: could not parse last update time %v", modTimeErr) |
| 76 | + modTime = timeUnset |
| 77 | + } |
| 78 | + for _, file := range result.Data.LatestVersion.Files { |
| 79 | + contentURLPath := fmt.Sprintf("/api/access/datafile/%d", file.DataFile.ID) |
| 80 | + query := url.Values{} |
| 81 | + query.Add("format", "original") |
| 82 | + contentURL := dp.f.endpoint.ResolveReference(&url.URL{Path: contentURLPath, RawQuery: query.Encode()}) |
| 83 | + entry := &Object{ |
| 84 | + fs: dp.f, |
| 85 | + remote: path.Join(file.DirectoryLabel, file.DataFile.Filename), |
| 86 | + contentURL: contentURL.String(), |
| 87 | + size: file.DataFile.FileSize, |
| 88 | + modTime: modTime, |
| 89 | + md5: file.DataFile.MD5, |
| 90 | + contentType: file.DataFile.ContentType, |
| 91 | + } |
| 92 | + if file.DataFile.OriginalFileName != "" { |
| 93 | + entry.remote = path.Join(file.DirectoryLabel, file.DataFile.OriginalFileName) |
| 94 | + entry.size = file.DataFile.OriginalFileSize |
| 95 | + entry.contentType = file.DataFile.OriginalFileFormat |
| 96 | + } |
| 97 | + entries = append(entries, entry) |
| 98 | + } |
| 99 | + // Populate the cache |
| 100 | + cacheEntries := []Object{} |
| 101 | + for _, entry := range entries { |
| 102 | + cacheEntries = append(cacheEntries, *entry) |
| 103 | + } |
| 104 | + dp.f.cache.Put("files", cacheEntries) |
| 105 | + return entries, nil |
| 106 | +} |
| 107 | + |
| 108 | +func newDataverseProvider(f *Fs) doiProvider { |
| 109 | + return &dataverseProvider{ |
| 110 | + f: f, |
| 111 | + } |
| 112 | +} |
0 commit comments