File
Blob: src/rust/python-parser/import_parsing.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 <workerd/rust/python-parser/lib.rs.h> |
| 6 | |
| 7 | #include <kj-rs/kj-rs.h> |
| 8 | |
| 9 | #include <kj/test.h> |
| 10 | |
| 11 | using namespace kj_rs; |
| 12 | |
| 13 | using ::edgeworker::rust::python_parser::get_imports; |
| 14 | |
| 15 | kj::Array<kj::String> parseImports(kj::ArrayPtr<kj::StringPtr> cpp_modules) { |
| 16 | auto rust_modules = kj::heapArrayBuilder<::rust::Str const>(cpp_modules.size()); |
| 17 | for (auto& entry: cpp_modules) { |
| 18 | rust_modules.add(entry.cStr()); |
| 19 | } |
| 20 | ::rust::Slice<::rust::Str const> rust_slice(rust_modules.begin(), rust_modules.size()); |
| 21 | auto rust_result = get_imports(rust_slice); |
| 22 | return kj::from<RustCopy>(rust_result); |
| 23 | } |
| 24 | |
| 25 | namespace workerd::api { |
| 26 | namespace { |
| 27 | |
| 28 | KJ_TEST("basic `import` tests") { |
| 29 | auto result = parseImports(kj::arr("import a\nimport z"_kj, "import b"_kj)); |
| 30 | KJ_REQUIRE(result.size() == 3); |
| 31 | KJ_REQUIRE(result[0] == "a"); |
| 32 | KJ_REQUIRE(result[1] == "b"); |
| 33 | KJ_REQUIRE(result[2] == "z"); |
| 34 | } |
| 35 | |
| 36 | KJ_TEST("supports whitespace") { |
| 37 | auto result = parseImports(kj::arr("import a\nimport \\\n\tz"_kj)); |
| 38 | KJ_REQUIRE(result.size() == 2); |
| 39 | KJ_REQUIRE(result[0] == "a"); |
| 40 | KJ_REQUIRE(result[1] == "z"); |
| 41 | } |
| 42 | |
| 43 | KJ_TEST("supports windows newlines") { |
| 44 | auto result = parseImports(kj::arr("import a\r\nimport \\\r\n\tz"_kj)); |
| 45 | KJ_REQUIRE(result.size() == 2); |
| 46 | KJ_REQUIRE(result[0] == "a"); |
| 47 | KJ_REQUIRE(result[1] == "z"); |
| 48 | } |
| 49 | |
| 50 | KJ_TEST("basic `from` test") { |
| 51 | auto result = parseImports(kj::arr("from x import a,b\nfrom z import y"_kj)); |
| 52 | KJ_REQUIRE(result.size() == 2); |
| 53 | KJ_REQUIRE(result[0] == "x"); |
| 54 | KJ_REQUIRE(result[1] == "z"); |
| 55 | } |
| 56 | |
| 57 | KJ_TEST("ignores indented blocks") { |
| 58 | auto result = parseImports(kj::arr("import a\nif True:\n import x\nimport y"_kj)); |
| 59 | KJ_REQUIRE(result.size() == 2); |
| 60 | KJ_REQUIRE(result[0] == "a"); |
| 61 | KJ_REQUIRE(result[1] == "y"); |
| 62 | } |
| 63 | |
| 64 | KJ_TEST("supports nested imports") { |
| 65 | auto result = parseImports(kj::arr("import a.b\nimport z.x.y.i"_kj)); |
| 66 | KJ_REQUIRE(result.size() == 2); |
| 67 | KJ_REQUIRE(result[0] == "a.b"); |
| 68 | KJ_REQUIRE(result[1] == "z.x.y.i"); |
| 69 | } |
| 70 | |
| 71 | KJ_TEST("nested `from` test") { |
| 72 | auto result = parseImports(kj::arr("from x.y.z import a,b\nfrom z import y"_kj)); |
| 73 | KJ_REQUIRE(result.size() == 2); |
| 74 | KJ_REQUIRE(result[0] == "x.y.z"); |
| 75 | KJ_REQUIRE(result[1] == "z"); |
| 76 | } |
| 77 | |
| 78 | KJ_TEST("ignores trailing period") { |
| 79 | auto result = parseImports(kj::arr("import a.b.\nimport z.x.y.i."_kj)); |
| 80 | KJ_REQUIRE(result.size() == 0); |
| 81 | } |
| 82 | |
| 83 | KJ_TEST("ignores relative import") { |
| 84 | // This is where we diverge from the old AST-based approach. It would have returned `y` in the |
| 85 | // input below. |
| 86 | auto result = parseImports(kj::arr("import .a.b\nimport ..z.x\nfrom .y import x"_kj)); |
| 87 | KJ_REQUIRE(result.size() == 0); |
| 88 | } |
| 89 | |
| 90 | KJ_TEST("supports commas") { |
| 91 | auto result = parseImports(kj::arr("import a,b"_kj)); |
| 92 | KJ_REQUIRE(result.size() == 2); |
| 93 | KJ_REQUIRE(result[0] == "a"); |
| 94 | KJ_REQUIRE(result[1] == "b"); |
| 95 | } |
| 96 | |
| 97 | KJ_TEST("supports backslash") { |
| 98 | // clang-format off |
| 99 | auto result = parseImports(kj::arr( |
| 100 | "import a\\\n,b"_kj, |
| 101 | "import\\\n q,w"_kj, |
| 102 | "from \\\nx import y"_kj, |
| 103 | "from \\\n c import y"_kj |
| 104 | )); |
| 105 | // clang-format on |
| 106 | KJ_REQUIRE(result.size() == 6); |
| 107 | KJ_REQUIRE(result[0] == "a"); |
| 108 | KJ_REQUIRE(result[1] == "b"); |
| 109 | KJ_REQUIRE(result[2] == "c"); |
| 110 | KJ_REQUIRE(result[3] == "q"); |
| 111 | KJ_REQUIRE(result[4] == "w"); |
| 112 | KJ_REQUIRE(result[5] == "x"); |
| 113 | } |
| 114 | |
| 115 | KJ_TEST("multiline-strings ignored") { |
| 116 | // clang-format off |
| 117 | auto files = kj::arr(R"SCRIPT( |
| 118 | FOO=""" |
| 119 | import x |
| 120 | from y import z |
| 121 | """ |
| 122 | )SCRIPT"_kj, |
| 123 | R"SCRIPT( |
| 124 | FOO=''' |
| 125 | import f |
| 126 | from g import z |
| 127 | ''' |
| 128 | )SCRIPT"_kj, |
| 129 | R"SCRIPT(FOO = "\ |
| 130 | import b \ |
| 131 | ")SCRIPT"_kj, |
| 132 | "FOO=\"\"\" \n"_kj, |
| 133 | R"SCRIPT(import x |
| 134 | from y import z |
| 135 | """)SCRIPT"_kj); |
| 136 | // clang-format on |
| 137 | auto result = parseImports(files); |
| 138 | KJ_REQUIRE(result.size() == 0); |
| 139 | } |
| 140 | |
| 141 | KJ_TEST("multiline-strings with imports in-between") { |
| 142 | auto files = kj::arr( |
| 143 | R"SCRIPT(FOO=""" |
| 144 | import x |
| 145 | from y import z |
| 146 | """ |
| 147 | import q |
| 148 | import w |
| 149 | BAR=""" |
| 150 | import e |
| 151 | """ |
| 152 | from t import u)SCRIPT"_kj); |
| 153 | auto result = parseImports(files); |
| 154 | KJ_REQUIRE(result.size() == 3); |
| 155 | KJ_REQUIRE(result[0] == "q"); |
| 156 | KJ_REQUIRE(result[1] == "t"); |
| 157 | KJ_REQUIRE(result[2] == "w"); |
| 158 | } |
| 159 | |
| 160 | KJ_TEST("import after string literal") { |
| 161 | auto files = kj::arr(R"SCRIPT(import a |
| 162 | "import b")SCRIPT"_kj); |
| 163 | auto result = parseImports(files); |
| 164 | KJ_REQUIRE(result.size() == 1); |
| 165 | KJ_REQUIRE(result[0] == "a"); |
| 166 | } |
| 167 | |
| 168 | KJ_TEST("langchain import") { |
| 169 | auto files = kj::arr(R"SCRIPT(from js import Response, console, URL |
| 170 | from langchain.chat_models import ChatOpenAI |
| 171 | import openai)SCRIPT"_kj); |
| 172 | auto result = parseImports(files); |
| 173 | KJ_REQUIRE(result.size() == 3); |
| 174 | KJ_REQUIRE(result[0] == "js"); |
| 175 | KJ_REQUIRE(result[1] == "langchain.chat_models"); |
| 176 | KJ_REQUIRE(result[2] == "openai"); |
| 177 | } |
| 178 | |
| 179 | } // namespace |
| 180 | } // namespace workerd::api |