|
| 1 | +""" |
| 2 | +Helpers for Tutor commands and "do" commands. |
| 3 | +""" |
| 4 | +from zipfile import ZipFile |
| 5 | +import glob |
| 6 | +import os |
| 7 | +import re |
| 8 | +import yaml |
| 9 | + |
| 10 | +import click |
| 11 | + |
| 12 | +FILE_NAME_ATTRIBUTE = "_file_name" |
| 13 | + |
| 14 | +PLUGIN_PATH = os.path.dirname(os.path.abspath(__file__)) |
| 15 | +ASSETS_PATH = os.path.join(PLUGIN_PATH, "templates", "openedx-assets", "assets") |
| 16 | + |
| 17 | + |
| 18 | +class SupersetCommandError(Exception): |
| 19 | + """ |
| 20 | + An error we can use for these methods. |
| 21 | + """ |
| 22 | + |
| 23 | + |
| 24 | +class Asset: |
| 25 | + """ |
| 26 | + Base class for asset types used in import. |
| 27 | + """ |
| 28 | + |
| 29 | + path = None |
| 30 | + assets_path = None |
| 31 | + templated_vars = None |
| 32 | + |
| 33 | + def __init__(self): |
| 34 | + if not self.path: |
| 35 | + raise NotImplementedError("Asset is an abstract class.") |
| 36 | + |
| 37 | + self.assets_path = os.path.join(ASSETS_PATH, self.path) |
| 38 | + |
| 39 | + def get_path(self): |
| 40 | + """ |
| 41 | + Returns the full path to the asset file type. |
| 42 | + """ |
| 43 | + if self.assets_path: |
| 44 | + return self.assets_path |
| 45 | + raise NotImplementedError |
| 46 | + |
| 47 | + def get_templated_vars(self): |
| 48 | + """ |
| 49 | + Returns a list of variables which should have templated variables. |
| 50 | +
|
| 51 | + This allows us to alert users when they might be submitting hard coded |
| 52 | + values instead of something like {{ TABLE_NAME }}. |
| 53 | + """ |
| 54 | + return self.templated_vars or [] |
| 55 | + |
| 56 | + |
| 57 | +class ChartAsset(Asset): |
| 58 | + """ |
| 59 | + Chart assets. |
| 60 | + """ |
| 61 | + |
| 62 | + path = "charts" |
| 63 | + |
| 64 | + |
| 65 | +class DashboardAsset(Asset): |
| 66 | + """ |
| 67 | + Dashboard assets. |
| 68 | + """ |
| 69 | + |
| 70 | + path = "dashboards" |
| 71 | + |
| 72 | + |
| 73 | +class DatasetAsset(Asset): |
| 74 | + """ |
| 75 | + Dataset assets. |
| 76 | + """ |
| 77 | + |
| 78 | + path = "datasets" |
| 79 | + templated_vars = ["schema", "table_name", "sql"] |
| 80 | + |
| 81 | + |
| 82 | +class DatabaseAsset(Asset): |
| 83 | + """ |
| 84 | + Database assets. |
| 85 | + """ |
| 86 | + |
| 87 | + path = "databases" |
| 88 | + templated_vars = ["sqlalchemy_uri"] |
| 89 | + |
| 90 | + |
| 91 | +ASSET_TYPE_MAP = { |
| 92 | + "slice_name": ChartAsset(), |
| 93 | + "dashboard_title": DashboardAsset(), |
| 94 | + "table_name": DatasetAsset(), |
| 95 | + "database_name": DatabaseAsset(), |
| 96 | +} |
| 97 | + |
| 98 | + |
| 99 | +def import_asset_file(asset_path, content, echo): |
| 100 | + orig_filename = os.path.basename(asset_path) |
| 101 | + out_filename = re.sub(r'(_\d*)\.yaml', '.yaml', orig_filename) |
| 102 | + content[FILE_NAME_ATTRIBUTE] = out_filename |
| 103 | + |
| 104 | + out_path = None |
| 105 | + needs_review = False |
| 106 | + for key, cls in ASSET_TYPE_MAP.items(): |
| 107 | + if key in content: |
| 108 | + out_path = cls.get_path() |
| 109 | + |
| 110 | + for var in cls.get_templated_vars(): |
| 111 | + # If this is a variable we expect to be templated, |
| 112 | + # check that it is. |
| 113 | + if ( |
| 114 | + content[var] |
| 115 | + and not content[var].startswith("{{") |
| 116 | + and not content[var].startswith("{%") |
| 117 | + ): |
| 118 | + echo( |
| 119 | + click.style( |
| 120 | + f"WARN: {orig_filename} has " |
| 121 | + f"{var} set to {content[var]} instead of a " |
| 122 | + f"setting.", |
| 123 | + fg="yellow", |
| 124 | + ) |
| 125 | + ) |
| 126 | + needs_review = True |
| 127 | + break |
| 128 | + return out_path, needs_review |
| 129 | + |
| 130 | + |
| 131 | +def import_superset_assets(file, echo): |
| 132 | + """ |
| 133 | + Import assets from a Superset export zip file to the openedx-assets directory. |
| 134 | + """ |
| 135 | + written_assets = [] |
| 136 | + review_files = set() |
| 137 | + err = 0 |
| 138 | + |
| 139 | + with ZipFile(file.name) as zip_file: |
| 140 | + for asset_path in zip_file.namelist(): |
| 141 | + if "metadata.yaml" in asset_path: |
| 142 | + continue |
| 143 | + with zip_file.open(asset_path) as asset_file: |
| 144 | + content = yaml.safe_load(asset_file) |
| 145 | + out_path, needs_review = import_asset_file(asset_path, content, echo) |
| 146 | + |
| 147 | + # This can happen if it's an unknown asset type |
| 148 | + if not out_path: |
| 149 | + continue |
| 150 | + |
| 151 | + if needs_review: |
| 152 | + review_files.add(content[FILE_NAME_ATTRIBUTE]) |
| 153 | + |
| 154 | + out_path = os.path.join(out_path, content[FILE_NAME_ATTRIBUTE]) |
| 155 | + written_assets.append(out_path) |
| 156 | + |
| 157 | + with open(out_path, "w", encoding="utf-8") as out_f: |
| 158 | + yaml.dump(content, out_f, encoding="utf-8") |
| 159 | + |
| 160 | + if review_files: |
| 161 | + echo() |
| 162 | + echo( |
| 163 | + click.style( |
| 164 | + f"{len(review_files)} files had warnings and need review:", fg="red" |
| 165 | + ) |
| 166 | + ) |
| 167 | + for filename in review_files: |
| 168 | + echo(f" - {filename}") |
| 169 | + |
| 170 | + raise SupersetCommandError( |
| 171 | + "Warnings found, please review then run " |
| 172 | + "'tutor aspects check_superset_assets'" |
| 173 | + ) |
| 174 | + |
| 175 | + echo() |
| 176 | + echo(f"Serialized {len(written_assets)} assets") |
| 177 | + |
| 178 | + return err |
| 179 | + |
| 180 | + |
| 181 | +def _get_asset_files(): |
| 182 | + for file_name in glob.iglob(ASSETS_PATH + "/**/*.yaml", recursive=True): |
| 183 | + with open(file_name, "r", encoding="utf-8") as file: |
| 184 | + # We have to remove the jinja for it to parse |
| 185 | + file_str = file.read() |
| 186 | + file_str = file_str.replace("{{", "").replace("}}", "") |
| 187 | + asset = yaml.safe_load(file_str) |
| 188 | + |
| 189 | + # Some asset types are lists of one element for some reason |
| 190 | + if isinstance(asset, list): |
| 191 | + asset = asset[0] |
| 192 | + |
| 193 | + yield file_name, asset |
| 194 | + |
| 195 | + |
| 196 | +def _deduplicate_asset_files(existing, found, echo): |
| 197 | + if existing["modified"] == found["modified"]: |
| 198 | + short_existing = os.path.basename(existing["file_name"]) |
| 199 | + short_found = os.path.basename(found["file_name"]) |
| 200 | + raise SupersetCommandError( |
| 201 | + "Modified dates are identical. You will need to " |
| 202 | + f"remove either {short_existing} or " |
| 203 | + f" {short_found} and run again." |
| 204 | + ) |
| 205 | + |
| 206 | + newer_file = existing if existing["modified"] > found["modified"] else found |
| 207 | + old_file = existing if existing["modified"] < found["modified"] else found |
| 208 | + |
| 209 | + short_new = os.path.basename(newer_file["file_name"]) |
| 210 | + short_old = os.path.basename(old_file["file_name"]) |
| 211 | + |
| 212 | + echo(f"{short_new} is newer, {short_old} will be deleted") |
| 213 | + |
| 214 | + os.remove(old_file["file_name"]) |
| 215 | + |
| 216 | + return newer_file |
| 217 | + |
| 218 | + |
| 219 | +def deduplicate_superset_assets(echo): |
| 220 | + """ |
| 221 | + Check for duplicated UUIDs in openedx-assets, delete the older file. |
| 222 | +
|
| 223 | + Superset exports use the name of the asset in the filename, so if you |
| 224 | + rename a chart or dashboard a new file will be created with the same |
| 225 | + UUID, causing import issues. This tries to fix that. |
| 226 | + """ |
| 227 | + echo("De-duplicating assets...") |
| 228 | + err = 0 |
| 229 | + uuid_file_map = {} |
| 230 | + |
| 231 | + for file_name, asset in _get_asset_files(): |
| 232 | + curr_uuid = asset["uuid"] |
| 233 | + if curr_uuid in uuid_file_map: |
| 234 | + echo() |
| 235 | + echo( |
| 236 | + click.style(f"WARN: Duplicate UUID found {asset['uuid']}", fg="yellow") |
| 237 | + ) |
| 238 | + |
| 239 | + new_file = {"file_name": file_name, "modified": os.stat(file_name)[8]} |
| 240 | + old_file = uuid_file_map[curr_uuid] |
| 241 | + |
| 242 | + try: |
| 243 | + uuid_file_map[curr_uuid] = _deduplicate_asset_files( |
| 244 | + old_file, new_file, echo |
| 245 | + ) |
| 246 | + except SupersetCommandError as ex: |
| 247 | + echo(click.style(f"ERROR: {ex}", fg="red")) |
| 248 | + err += 1 |
| 249 | + continue |
| 250 | + else: |
| 251 | + uuid_file_map[asset["uuid"]] = { |
| 252 | + "file_name": file_name, |
| 253 | + "modified": os.stat(file_name)[8], |
| 254 | + } |
| 255 | + |
| 256 | + if err: |
| 257 | + echo() |
| 258 | + echo() |
| 259 | + echo(click.style(f"{err} errors found!", fg="red")) |
| 260 | + |
| 261 | + echo("Deduplication complete.") |
| 262 | + |
| 263 | + |
| 264 | +def check_asset_names(echo): |
| 265 | + """ |
| 266 | + Warn about any duplicate asset names. |
| 267 | + """ |
| 268 | + echo("Looking for duplicate names...") |
| 269 | + warn = 0 |
| 270 | + |
| 271 | + names = set() |
| 272 | + for file_name, asset in _get_asset_files(): |
| 273 | + for k in ("slice_name", "dashboard_title", "database_name"): |
| 274 | + if k in asset: |
| 275 | + if asset[k] in names: |
| 276 | + warn += 1 |
| 277 | + echo( |
| 278 | + f"WARNING: Duplicate name {asset[k]} in {file_name}, this " |
| 279 | + f"could confuse users, consider changing it." |
| 280 | + ) |
| 281 | + names.add(asset[k]) |
| 282 | + break |
| 283 | + |
| 284 | + echo(f"{warn} duplicate names detected.") |
0 commit comments