|
34 | 34 | import static java.nio.charset.StandardCharsets.UTF_8; |
35 | 35 | import static java.util.Arrays.asList; |
36 | 36 | import static java.util.Collections.emptyList; |
| 37 | +import static org.apache.commons.lang.StringUtils.isEmpty; |
37 | 38 |
|
38 | 39 | /** |
39 | 40 | * Service to work with datasets and manifests. |
@@ -99,36 +100,115 @@ public FutureResult<Void> loadDataset(final Project project, final DatasetManife |
99 | 100 | final ByteArrayInputStream inputStream = new ByteArrayInputStream(manifestJson.getBytes(UTF_8)); |
100 | 101 | dataStoreService.upload(dirPath.resolve(MANIFEST_FILE_NAME).toString(), inputStream); |
101 | 102 |
|
102 | | - final PullTask pullTask = restTemplate |
103 | | - .postForObject(Pull.URI, new Pull(dirPath.toString()), PullTask.class, project.getId()); |
104 | | - return new FutureResult<>(this, new SimplePollHandler<Void>(pullTask.getUri(), Void.class) { |
105 | | - @Override |
106 | | - public boolean isFinished(ClientHttpResponse response) throws IOException { |
107 | | - final PullTaskStatus status = extractData(response, PullTaskStatus.class); |
108 | | - final boolean finished = status.isFinished(); |
109 | | - if (finished && !status.isSuccess()) { |
110 | | - final String message = getErrorMessage(status, dirPath); |
111 | | - throw new DatasetException(message, manifest.getDataSet()); |
112 | | - } |
113 | | - return finished; |
114 | | - } |
115 | | - |
116 | | - @Override |
117 | | - protected void onFinish() { |
118 | | - try { |
119 | | - dataStoreService.delete(dirPath.toString() + "/"); |
120 | | - } catch (DataStoreException ignored) { |
121 | | - // todo log? |
122 | | - } |
123 | | - } |
124 | | - }); |
| 103 | + return pullLoad(project, dirPath, manifest.getDataSet()); |
125 | 104 | } catch (IOException e) { |
126 | 105 | throw new DatasetException("Unable to serialize manifest", manifest.getDataSet(), e); |
127 | 106 | } catch (DataStoreException | GoodDataRestException | RestClientException e) { |
128 | 107 | throw new DatasetException("Unable to load", manifest.getDataSet(), e); |
129 | 108 | } |
130 | 109 | } |
131 | 110 |
|
| 111 | + /** |
| 112 | + * Gets DatasetManifest (using {@link #getDatasetManifest(com.gooddata.project.Project, String)} |
| 113 | + * first and then calls {@link #loadDataset(com.gooddata.project.Project, DatasetManifest, java.io.InputStream)} |
| 114 | + * |
| 115 | + * @param project project to which dataset belongs |
| 116 | + * @param datasetId datasetId to obtain a manifest |
| 117 | + * @param dataset dataset to upload |
| 118 | + * @return {@link com.gooddata.FutureResult} of the task |
| 119 | + */ |
| 120 | + public FutureResult<Void> loadDataset(Project project, String datasetId, InputStream dataset) { |
| 121 | + notNull(project, "project"); |
| 122 | + notEmpty(datasetId, "datasetId"); |
| 123 | + notNull(dataset, "dataset"); |
| 124 | + return loadDataset(project, getDatasetManifest(project, datasetId), dataset); |
| 125 | + } |
| 126 | + |
| 127 | + public FutureResult<Void> loadDatasets(final Project project, DatasetManifest... datasets) { |
| 128 | + return loadDatasets(project, asList(datasets)); |
| 129 | + } |
| 130 | + |
| 131 | + /** |
| 132 | + * Loads datasets into platform. Uploads given datasets and their manifests to staging area and triggers ETL pull. |
| 133 | + * The call is asynchronous returning {@link com.gooddata.FutureResult} to let caller wait for results. |
| 134 | + * Uploaded files are deleted from staging area when finished. |
| 135 | + * |
| 136 | + * @param project project to which dataset belongs |
| 137 | + * @param datasets map dataset manifests |
| 138 | + * @return {@link com.gooddata.FutureResult} of the task, which can throw {@link com.gooddata.dataset.DatasetException} |
| 139 | + * in case the ETL pull task fails |
| 140 | + * @throws com.gooddata.dataset.DatasetException if there is a problem to serialize manifest or upload dataset |
| 141 | + * @see <a href="https://developer.gooddata.com/article/multiload-of-csv-data">batch upload reference</a> |
| 142 | + */ |
| 143 | + public FutureResult<Void> loadDatasets(final Project project, final Collection<DatasetManifest> datasets) { |
| 144 | + notNull(project, "project"); |
| 145 | + validateUploadManifests(datasets); |
| 146 | + final List<String> datasetsNames = new ArrayList<>(datasets.size()); |
| 147 | + try { |
| 148 | + final Path dirPath = Paths.get("/", project.getId() + "_" + RandomStringUtils.randomAlphabetic(3), "/"); |
| 149 | + for (DatasetManifest datasetManifest : datasets) { |
| 150 | + datasetsNames.add(datasetManifest.getDataSet()); |
| 151 | + dataStoreService.upload(dirPath.resolve(datasetManifest.getFile()).toString(), datasetManifest.getSource()); |
| 152 | + } |
| 153 | + |
| 154 | + final String manifestJson = mapper.writeValueAsString(new DatasetManifests(datasets)); |
| 155 | + final ByteArrayInputStream inputStream = new ByteArrayInputStream(manifestJson.getBytes(UTF_8)); |
| 156 | + dataStoreService.upload(dirPath.resolve(MANIFEST_FILE_NAME).toString(), inputStream); |
| 157 | + |
| 158 | + return pullLoad(project, dirPath, datasetsNames); |
| 159 | + } catch (IOException e) { |
| 160 | + throw new DatasetException("Unable to serialize manifest", datasetsNames, e); |
| 161 | + } catch (DataStoreException | GoodDataRestException | RestClientException e) { |
| 162 | + throw new DatasetException("Unable to load", datasetsNames, e); |
| 163 | + } |
| 164 | + } |
| 165 | + |
| 166 | + private void validateUploadManifests(final Collection<DatasetManifest> datasets) { |
| 167 | + notEmpty(datasets, "datasets"); |
| 168 | + for (DatasetManifest datasetManifest : datasets) { |
| 169 | + if (datasetManifest.getSource() == null) { |
| 170 | + throw new IllegalArgumentException(format("Source for dataset '%s' is null", datasetManifest.getDataSet())); |
| 171 | + } |
| 172 | + if (datasetManifest.getFile() == null) { |
| 173 | + throw new IllegalArgumentException(format("File for dataset '%s' is null", datasetManifest.getDataSet())); |
| 174 | + } |
| 175 | + if (isEmpty(datasetManifest.getDataSet())) { |
| 176 | + throw new IllegalArgumentException("Dataset name is empty."); |
| 177 | + } |
| 178 | + } |
| 179 | + } |
| 180 | + |
| 181 | + private FutureResult<Void> pullLoad(Project project, final Path dirPath, final String dataset) { |
| 182 | + return pullLoad(project, dirPath, asList(dataset)); |
| 183 | + } |
| 184 | + |
| 185 | + private FutureResult<Void> pullLoad(Project project, final Path dirPath, final Collection<String> datasets) { |
| 186 | + final PullTask pullTask = restTemplate |
| 187 | + .postForObject(Pull.URI, new Pull(dirPath.toString()), PullTask.class, project.getId()); |
| 188 | + return new FutureResult<>(this, new SimplePollHandler<Void>(pullTask.getUri(), Void.class) { |
| 189 | + @Override |
| 190 | + public boolean isFinished(ClientHttpResponse response) throws IOException { |
| 191 | + final PullTaskStatus status = extractData(response, PullTaskStatus.class); |
| 192 | + final boolean finished = status.isFinished(); |
| 193 | + if (finished && !status.isSuccess()) { |
| 194 | + final String message = getErrorMessage(status, dirPath); |
| 195 | + throw new DatasetException(message, datasets); |
| 196 | + } |
| 197 | + return finished; |
| 198 | + } |
| 199 | + |
| 200 | + @Override |
| 201 | + protected void onFinish() { |
| 202 | + try { |
| 203 | + dataStoreService.delete(dirPath.toString() + "/"); |
| 204 | + } catch (DataStoreException ignored) { |
| 205 | + // todo log? |
| 206 | + } |
| 207 | + } |
| 208 | + }); |
| 209 | + |
| 210 | + } |
| 211 | + |
132 | 212 | private String getErrorMessage(final PullTaskStatus status, final Path dirPath) { |
133 | 213 | String message = "status: " + status.getStatus(); |
134 | 214 | try { |
@@ -166,22 +246,6 @@ private <T> T download(final Path path, final Class<T> type) throws IOException |
166 | 246 | } |
167 | 247 | } |
168 | 248 |
|
169 | | - /** |
170 | | - * Gets DatasetManifest (using {@link #getDatasetManifest(com.gooddata.project.Project, String)} |
171 | | - * first and then calls {@link #loadDataset(com.gooddata.project.Project, DatasetManifest, java.io.InputStream)} |
172 | | - * |
173 | | - * @param project project to which dataset belongs |
174 | | - * @param datasetId datasetId to obtain a manifest |
175 | | - * @param dataset dataset to upload |
176 | | - * @return {@link com.gooddata.FutureResult} of the task |
177 | | - */ |
178 | | - public FutureResult<Void> loadDataset(Project project, String datasetId, InputStream dataset) { |
179 | | - notNull(project, "project"); |
180 | | - notEmpty(datasetId, "datasetId"); |
181 | | - notNull(dataset, "dataset"); |
182 | | - return loadDataset(project, getDatasetManifest(project, datasetId), dataset); |
183 | | - } |
184 | | - |
185 | 249 | /** |
186 | 250 | * Lists datasets in project. Returns empty list in case there are no datasets. |
187 | 251 | * |
|
0 commit comments