File
Blob: src/workerd/api/pyodide/pyodide-test.c++
| 1 | // Copyright (c) 2017-2022 Cloudflare, Inc. |
| 2 | // Licensed under the Apache 2.0 license found in the LICENSE file or at: |
| 3 | // https://opensource.org/licenses/Apache-2.0 |
| 4 | |
| 5 | #include "pyodide.h" |
| 6 | |
| 7 | #include <kj/test.h> |
| 8 | |
| 9 | namespace workerd::api { |
| 10 | namespace { |
| 11 | |
| 12 | KJ_TEST("getPythonSnapshotRelease") { |
| 13 | capnp::MallocMessageBuilder arena; |
| 14 | // TODO(beta): Factor out FeatureFlags from WorkerBundle. |
| 15 | auto featureFlags = arena.initRoot<CompatibilityFlags>(); |
| 16 | |
| 17 | { |
| 18 | auto res = getPythonSnapshotRelease(featureFlags); |
| 19 | KJ_ASSERT(res == kj::none); |
| 20 | } |
| 21 | |
| 22 | featureFlags.setPythonWorkers(true); |
| 23 | { |
| 24 | auto res = KJ_ASSERT_NONNULL(getPythonSnapshotRelease(featureFlags)); |
| 25 | KJ_ASSERT(res.getPyodide() == "0.26.0a2"); |
| 26 | KJ_ASSERT(res.getFlagName() == "pythonWorkers"); |
| 27 | } |
| 28 | |
| 29 | featureFlags.setPythonWorkersDevPyodide(true); |
| 30 | { |
| 31 | auto res = KJ_ASSERT_NONNULL(getPythonSnapshotRelease(featureFlags)); |
| 32 | KJ_ASSERT(res.getPyodide() == "dev"); |
| 33 | KJ_ASSERT(res.getFlagName() == "pythonWorkersDevPyodide"); |
| 34 | } |
| 35 | |
| 36 | featureFlags.setPythonWorkers(false); |
| 37 | { |
| 38 | auto res = KJ_ASSERT_NONNULL(getPythonSnapshotRelease(featureFlags)); |
| 39 | KJ_ASSERT(res.getPyodide() == "dev"); |
| 40 | KJ_ASSERT(res.getFlagName() == "pythonWorkersDevPyodide"); |
| 41 | } |
| 42 | |
| 43 | featureFlags.setPythonWorkers20250116(true); |
| 44 | { |
| 45 | auto res = KJ_ASSERT_NONNULL(getPythonSnapshotRelease(featureFlags)); |
| 46 | KJ_ASSERT(res.getPyodide() == "0.28.2"); |
| 47 | KJ_ASSERT(res.getFlagName() == "pythonWorkers20250116"); |
| 48 | } |
| 49 | |
| 50 | featureFlags.setPythonWorkersDevPyodide(false); |
| 51 | { |
| 52 | auto res = KJ_ASSERT_NONNULL(getPythonSnapshotRelease(featureFlags)); |
| 53 | KJ_ASSERT(res.getPyodide() == "0.28.2"); |
| 54 | KJ_ASSERT(res.getFlagName() == "pythonWorkers20250116"); |
| 55 | } |
| 56 | } |
| 57 | |
| 58 | KJ_TEST("basic `import` tests") { |
| 59 | auto files = kj::heapArrayBuilder<kj::String>(2); |
| 60 | files.add(kj::str("import a\nimport z")); |
| 61 | files.add(kj::str("import b")); |
| 62 | auto result = pyodide::PythonModuleInfo::parsePythonScriptImports(files.finish()); |
| 63 | KJ_REQUIRE(result.size() == 3); |
| 64 | KJ_REQUIRE(result[0] == "a"); |
| 65 | KJ_REQUIRE(result[1] == "z"); |
| 66 | KJ_REQUIRE(result[2] == "b"); |
| 67 | } |
| 68 | |
| 69 | KJ_TEST("supports whitespace") { |
| 70 | auto files = kj::heapArrayBuilder<kj::String>(1); |
| 71 | files.add(kj::str("import a\nimport \n\tz")); |
| 72 | auto result = pyodide::PythonModuleInfo::parsePythonScriptImports(files.finish()); |
| 73 | KJ_REQUIRE(result.size() == 2); |
| 74 | KJ_REQUIRE(result[0] == "a"); |
| 75 | KJ_REQUIRE(result[1] == "z"); |
| 76 | } |
| 77 | |
| 78 | KJ_TEST("supports windows newlines") { |
| 79 | auto files = kj::heapArrayBuilder<kj::String>(1); |
| 80 | files.add(kj::str("import a\r\nimport \r\n\tz")); |
| 81 | auto result = pyodide::PythonModuleInfo::parsePythonScriptImports(files.finish()); |
| 82 | KJ_REQUIRE(result.size() == 2); |
| 83 | KJ_REQUIRE(result[0] == "a"); |
| 84 | KJ_REQUIRE(result[1] == "z"); |
| 85 | } |
| 86 | |
| 87 | KJ_TEST("basic `from` test") { |
| 88 | auto files = kj::heapArrayBuilder<kj::String>(1); |
| 89 | files.add(kj::str("from x import a,b\nfrom z import y")); |
| 90 | auto result = pyodide::PythonModuleInfo::parsePythonScriptImports(files.finish()); |
| 91 | KJ_REQUIRE(result.size() == 2); |
| 92 | KJ_REQUIRE(result[0] == "x"); |
| 93 | KJ_REQUIRE(result[1] == "z"); |
| 94 | } |
| 95 | |
| 96 | KJ_TEST("ignores indented blocks") { |
| 97 | auto files = kj::heapArrayBuilder<kj::String>(1); |
| 98 | files.add(kj::str("import a\nif True:\n import x\nimport y")); |
| 99 | auto result = pyodide::PythonModuleInfo::parsePythonScriptImports(files.finish()); |
| 100 | KJ_REQUIRE(result.size() == 2); |
| 101 | KJ_REQUIRE(result[0] == "a"); |
| 102 | KJ_REQUIRE(result[1] == "y"); |
| 103 | } |
| 104 | |
| 105 | KJ_TEST("supports nested imports") { |
| 106 | auto files = kj::heapArrayBuilder<kj::String>(1); |
| 107 | files.add(kj::str("import a.b\nimport z.x.y.i")); |
| 108 | auto result = pyodide::PythonModuleInfo::parsePythonScriptImports(files.finish()); |
| 109 | KJ_REQUIRE(result.size() == 2); |
| 110 | KJ_REQUIRE(result[0] == "a.b"); |
| 111 | KJ_REQUIRE(result[1] == "z.x.y.i"); |
| 112 | } |
| 113 | |
| 114 | KJ_TEST("nested `from` test") { |
| 115 | auto files = kj::heapArrayBuilder<kj::String>(1); |
| 116 | files.add(kj::str("from x.y.z import a,b\nfrom z import y")); |
| 117 | auto result = pyodide::PythonModuleInfo::parsePythonScriptImports(files.finish()); |
| 118 | KJ_REQUIRE(result.size() == 2); |
| 119 | KJ_REQUIRE(result[0] == "x.y.z"); |
| 120 | KJ_REQUIRE(result[1] == "z"); |
| 121 | } |
| 122 | |
| 123 | KJ_TEST("ignores trailing period") { |
| 124 | auto files = kj::heapArrayBuilder<kj::String>(1); |
| 125 | files.add(kj::str("import a.b.\nimport z.x.y.i.")); |
| 126 | auto result = pyodide::PythonModuleInfo::parsePythonScriptImports(files.finish()); |
| 127 | KJ_REQUIRE(result.size() == 0); |
| 128 | } |
| 129 | |
| 130 | KJ_TEST("ignores relative import") { |
| 131 | // This is where we diverge from the old AST-based approach. It would have returned `y` in the |
| 132 | // input below. |
| 133 | auto files = kj::heapArrayBuilder<kj::String>(1); |
| 134 | files.add(kj::str("import .a.b\nimport ..z.x\nfrom .y import x")); |
| 135 | auto result = pyodide::PythonModuleInfo::parsePythonScriptImports(files.finish()); |
| 136 | KJ_REQUIRE(result.size() == 0); |
| 137 | } |
| 138 | |
| 139 | KJ_TEST("supports commas") { |
| 140 | auto files = kj::heapArrayBuilder<kj::String>(1); |
| 141 | files.add(kj::str("import a,b")); |
| 142 | auto result = pyodide::PythonModuleInfo::parsePythonScriptImports(files.finish()); |
| 143 | KJ_REQUIRE(result.size() == 2); |
| 144 | KJ_REQUIRE(result[0] == "a"); |
| 145 | KJ_REQUIRE(result[1] == "b"); |
| 146 | } |
| 147 | |
| 148 | KJ_TEST("supports backslash") { |
| 149 | auto files = kj::heapArrayBuilder<kj::String>(4); |
| 150 | files.add(kj::str("import a\\\n,b")); |
| 151 | files.add(kj::str("import\\\n q,w")); |
| 152 | files.add(kj::str("from \\\nx import y")); |
| 153 | files.add(kj::str("from \\\n c import y")); |
| 154 | auto result = pyodide::PythonModuleInfo::parsePythonScriptImports(files.finish()); |
| 155 | KJ_REQUIRE(result.size() == 6); |
| 156 | KJ_REQUIRE(result[0] == "a"); |
| 157 | KJ_REQUIRE(result[1] == "b"); |
| 158 | KJ_REQUIRE(result[2] == "q"); |
| 159 | KJ_REQUIRE(result[3] == "w"); |
| 160 | KJ_REQUIRE(result[4] == "x"); |
| 161 | KJ_REQUIRE(result[5] == "c"); |
| 162 | } |
| 163 | |
| 164 | KJ_TEST("multiline-strings ignored") { |
| 165 | auto files = kj::heapArrayBuilder<kj::String>(4); |
| 166 | files.add(kj::str(R"SCRIPT( |
| 167 | FOO=""" |
| 168 | import x |
| 169 | from y import z |
| 170 | """ |
| 171 | )SCRIPT")); |
| 172 | files.add(kj::str(R"SCRIPT( |
| 173 | FOO=''' |
| 174 | import f |
| 175 | from g import z |
| 176 | ''' |
| 177 | )SCRIPT")); |
| 178 | files.add(kj::str(R"SCRIPT(FOO = "\ |
| 179 | import b \ |
| 180 | ")SCRIPT")); |
| 181 | files.add(kj::str("FOO=\"\"\" \n", R"SCRIPT(import x |
| 182 | from y import z |
| 183 | """)SCRIPT")); |
| 184 | auto result = pyodide::PythonModuleInfo::parsePythonScriptImports(files.finish()); |
| 185 | KJ_REQUIRE(result.size() == 0); |
| 186 | } |
| 187 | |
| 188 | KJ_TEST("multiline-strings with imports in-between") { |
| 189 | auto files = kj::heapArrayBuilder<kj::String>(1); |
| 190 | files.add(kj::str( |
| 191 | R"SCRIPT(FOO=""" |
| 192 | import x |
| 193 | from y import z |
| 194 | """import q |
| 195 | import w |
| 196 | BAR=""" |
| 197 | import e |
| 198 | """ |
| 199 | from t import u)SCRIPT")); |
| 200 | auto result = pyodide::PythonModuleInfo::parsePythonScriptImports(files.finish()); |
| 201 | KJ_REQUIRE(result.size() == 2); |
| 202 | KJ_REQUIRE(result[0] == "w"); |
| 203 | KJ_REQUIRE(result[1] == "t"); |
| 204 | } |
| 205 | |
| 206 | KJ_TEST("import after string literal") { |
| 207 | auto files = kj::heapArrayBuilder<kj::String>(1); |
| 208 | files.add(kj::str(R"SCRIPT(import a |
| 209 | "import b)SCRIPT")); |
| 210 | auto result = pyodide::PythonModuleInfo::parsePythonScriptImports(files.finish()); |
| 211 | KJ_REQUIRE(result.size() == 1); |
| 212 | KJ_REQUIRE(result[0] == "a"); |
| 213 | } |
| 214 | |
| 215 | KJ_TEST("import after `i`") { |
| 216 | auto files = kj::heapArrayBuilder<kj::String>(1); |
| 217 | files.add(kj::str(R"SCRIPT(import a |
| 218 | iimport b)SCRIPT")); |
| 219 | auto result = pyodide::PythonModuleInfo::parsePythonScriptImports(files.finish()); |
| 220 | KJ_REQUIRE(result.size() == 1); |
| 221 | KJ_REQUIRE(result[0] == "a"); |
| 222 | } |
| 223 | |
| 224 | KJ_TEST("langchain import") { |
| 225 | auto files = kj::heapArrayBuilder<kj::String>(1); |
| 226 | files.add(kj::str(R"SCRIPT(from js import Response, console, URL |
| 227 | from langchain.chat_models import ChatOpenAI |
| 228 | import openai)SCRIPT")); |
| 229 | auto result = pyodide::PythonModuleInfo::parsePythonScriptImports(files.finish()); |
| 230 | KJ_REQUIRE(result.size() == 3); |
| 231 | KJ_REQUIRE(result[0] == "js"); |
| 232 | KJ_REQUIRE(result[1] == "langchain.chat_models"); |
| 233 | KJ_REQUIRE(result[2] == "openai"); |
| 234 | } |
| 235 | |
| 236 | KJ_TEST("quote in multiline string") { |
| 237 | auto files = kj::heapArrayBuilder<kj::String>(1); |
| 238 | files.add(kj::str(R"SCRIPT(temp = """ |
| 239 | w["h |
| 240 | """)SCRIPT")); |
| 241 | auto result = pyodide::PythonModuleInfo::parsePythonScriptImports(files.finish()); |
| 242 | KJ_REQUIRE(result.size() == 0); |
| 243 | } |
| 244 | |
| 245 | using pyodide::PythonModuleInfo; |
| 246 | |
| 247 | template <typename... Params> |
| 248 | kj::Array<kj::String> strArray(Params&&... params) { |
| 249 | return kj::arr(kj::str(params)...); |
| 250 | } |
| 251 | |
| 252 | template <typename... Params> |
| 253 | kj::Array<kj::Array<kj::byte>> bytesArray(Params&&... params) { |
| 254 | return kj::arr(kj::heapArray<kj::byte>(kj::str(params).asBytes())...); |
| 255 | } |
| 256 | |
| 257 | template <typename... Params> |
| 258 | kj::HashSet<kj::String> strSet(Params&&... params) { |
| 259 | auto array = strArray(params...); |
| 260 | kj::HashSet<kj::String> set; |
| 261 | for (auto& str: array) { |
| 262 | set.insert(kj::mv(str)); |
| 263 | } |
| 264 | return set; |
| 265 | } |
| 266 | |
| 267 | KJ_TEST("basic test of getPackageSnapshotImports") { |
| 268 | auto a = pyodide::PythonModuleInfo(strArray("a.py"), |
| 269 | bytesArray("from js import Response\n" |
| 270 | "import asyncio\n" |
| 271 | "import numbers\n" |
| 272 | "def on_fetch(request):\n" |
| 273 | " return Response.new('Hello')\n")); |
| 274 | auto result = a.getPackageSnapshotImports("0.26.0a2"); |
| 275 | KJ_REQUIRE(result.size() == 1); |
| 276 | KJ_REQUIRE(result[0] == "numbers"); |
| 277 | }; |
| 278 | |
| 279 | KJ_TEST("basic test of getPackageSnapshotImports user module") { |
| 280 | auto a = pyodide::PythonModuleInfo(strArray("a.py", "numbers.py"), |
| 281 | bytesArray("from js import Response\n" |
| 282 | "import asyncio\n" |
| 283 | "import numbers\n" |
| 284 | "def on_fetch(request):\n" |
| 285 | " return Response.new('Hello')\n", |
| 286 | "")); |
| 287 | auto result = a.getPackageSnapshotImports("0.26.0a2"); |
| 288 | KJ_REQUIRE(result.size() == 0); |
| 289 | }; |
| 290 | |
| 291 | kj::Array<kj::String> filterPythonScriptImports( |
| 292 | kj::Array<kj::String> names, kj::ArrayPtr<kj::String> imports, kj::StringPtr version) { |
| 293 | auto contentsBuilder = kj::heapArrayBuilder<kj::Array<kj::byte>>(names.size()); |
| 294 | for (auto _: kj::zeroTo(names.size())) { |
| 295 | (void)_; |
| 296 | contentsBuilder.add(kj::Array<kj::byte>(nullptr)); |
| 297 | } |
| 298 | auto modInfo = pyodide::PythonModuleInfo(kj::mv(names), contentsBuilder.finish()); |
| 299 | auto modSet = modInfo.getWorkerModuleSet(); |
| 300 | return PythonModuleInfo::filterPythonScriptImports(kj::mv(modSet), kj::mv(imports), version); |
| 301 | } |
| 302 | |
| 303 | KJ_TEST("Simple pass through") { |
| 304 | auto imports = strArray("b", "c"); |
| 305 | auto result = filterPythonScriptImports({}, kj::mv(imports), ""); |
| 306 | KJ_REQUIRE(result.size() == 2); |
| 307 | KJ_REQUIRE(result[0] == "b"); |
| 308 | KJ_REQUIRE(result[1] == "c"); |
| 309 | } |
| 310 | |
| 311 | KJ_TEST("pyodide and submodules") { |
| 312 | auto imports = strArray("pyodide", "pyodide.ffi"); |
| 313 | auto result = filterPythonScriptImports({}, kj::mv(imports), "0.26.0a2"); |
| 314 | KJ_REQUIRE(result.size() == 0); |
| 315 | } |
| 316 | |
| 317 | KJ_TEST("js and submodules") { |
| 318 | auto imports = strArray("js", "js.crypto"); |
| 319 | auto result = filterPythonScriptImports({}, kj::mv(imports), "0.26.0a2"); |
| 320 | KJ_REQUIRE(result.size() == 0); |
| 321 | } |
| 322 | |
| 323 | KJ_TEST("importlib and submodules") { |
| 324 | // importlib and importlib.metadata are imported into the baseline snapshot, but importlib.resources is not. |
| 325 | auto imports = strArray("importlib", "importlib.metadata", "importlib.resources"); |
| 326 | auto result = filterPythonScriptImports({}, kj::mv(imports), ""); |
| 327 | KJ_REQUIRE(result.size() == 1); |
| 328 | KJ_REQUIRE(result[0] == "importlib.resources"); |
| 329 | } |
| 330 | |
| 331 | KJ_TEST("Filter worker .py files") { |
| 332 | auto workerModules = strArray("b.py", "c.py"); |
| 333 | auto imports = strArray("b", "c", "d"); |
| 334 | auto result = filterPythonScriptImports(kj::mv(workerModules), kj::mv(imports), ""); |
| 335 | KJ_REQUIRE(result.size() == 1); |
| 336 | KJ_REQUIRE(result[0] == "d"); |
| 337 | } |
| 338 | |
| 339 | KJ_TEST("Filter worker module/__init__.py") { |
| 340 | auto workerModules = strArray("a/__init__.py", "b/__init__.py", "c/a.py"); |
| 341 | auto imports = strArray("a", "b", "c"); |
| 342 | auto result = filterPythonScriptImports(kj::mv(workerModules), kj::mv(imports), ""); |
| 343 | KJ_REQUIRE(result.size() == 0); |
| 344 | } |
| 345 | |
| 346 | KJ_TEST("Filters out subdir/submodule") { |
| 347 | auto workerModules = strArray("subdir/submodule.py"); |
| 348 | auto imports = strArray("subdir.submodule"); |
| 349 | auto result = filterPythonScriptImports(kj::mv(workerModules), kj::mv(imports), ""); |
| 350 | KJ_REQUIRE(result.size() == 0); |
| 351 | } |
| 352 | |
| 353 | KJ_TEST("Filters out so") { |
| 354 | auto workerModules = strArray("a.so", "b.txt"); |
| 355 | auto imports = strArray("a", "b"); |
| 356 | auto result = filterPythonScriptImports(kj::mv(workerModules), kj::mv(imports), ""); |
| 357 | KJ_REQUIRE(result.size() == 1); |
| 358 | KJ_REQUIRE(result[0] == "b"); |
| 359 | } |
| 360 | |
| 361 | KJ_TEST("Filters out vendor stuff") { |
| 362 | auto workerModules = strArray("python_modules/a.py", "python_modules/package/b.py", |
| 363 | "python_modules/c.so", "python_modules/x.txt"); |
| 364 | auto imports = strArray("a", "package", "x"); |
| 365 | auto result = filterPythonScriptImports(kj::mv(workerModules), kj::mv(imports), ""); |
| 366 | KJ_REQUIRE(result.size() == 1); |
| 367 | KJ_REQUIRE(result[0] == "x"); |
| 368 | } |
| 369 | |
| 370 | } // namespace |
| 371 | } // namespace workerd::api |