File
Blob: src/workerd/api/pyodide/requirements.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 | #include "requirements.h" |
| 5 | |
| 6 | #include <capnp/compat/json.h> |
| 7 | #include <capnp/message.h> |
| 8 | #include <kj/debug.h> |
| 9 | |
| 10 | #include <cctype> |
| 11 | |
| 12 | namespace workerd::api::pyodide { |
| 13 | |
| 14 | // getField gets a field of a JSON object by key |
| 15 | capnp::json::Value::Reader getField( |
| 16 | capnp::List<::capnp::json::Value::Field, capnp::Kind::STRUCT>::Reader &object, |
| 17 | kj::StringPtr name) { |
| 18 | for (const auto &ent: object) { |
| 19 | if (ent.getName() == name) { |
| 20 | return ent.getValue(); |
| 21 | } |
| 22 | } |
| 23 | |
| 24 | KJ_FAIL_ASSERT("Expected key in JSON object", name); |
| 25 | } |
| 26 | |
| 27 | kj::String canonicalizePythonPackageName(kj::StringPtr name) { |
| 28 | kj::Vector<char> res(name.size()); |
| 29 | |
| 30 | auto isSeparator = [](char c) { return c == '-' || c == '_' || c == '.'; }; |
| 31 | |
| 32 | for (int i = 0; i < name.size(); i++) { |
| 33 | if (isSeparator(name[i])) { |
| 34 | res.add('-'); |
| 35 | // make i point to the last separator in the sequence |
| 36 | while (isSeparator(name[i])) i++; |
| 37 | i--; |
| 38 | continue; |
| 39 | } |
| 40 | |
| 41 | res.add(std::tolower(name[i])); |
| 42 | } |
| 43 | |
| 44 | res.add(0); // NUL terminator |
| 45 | |
| 46 | return kj::String(res.releaseAsArray()); |
| 47 | } |
| 48 | |
| 49 | // getDepMapFromPackagesLock computes a dependency map (a mapping from requirement to list of dependencies) from the Pyodide lock file JSON |
| 50 | DepMap getDepMapFromPackagesLock( |
| 51 | capnp::List<capnp::json::Value::Field, capnp::Kind::STRUCT>::Reader &packages) { |
| 52 | DepMap res; |
| 53 | |
| 54 | for (const auto &ent: packages) { |
| 55 | auto packageObj = ent.getValue().getObject(); |
| 56 | auto depends = getField(packageObj, "depends").getArray(); |
| 57 | |
| 58 | auto &[_, deps] = res.insert(kj::str(ent.getName()), kj::Vector<kj::String>(depends.size())); |
| 59 | |
| 60 | for (const auto &dep: depends) { |
| 61 | deps.add(kj::str(dep.getString())); |
| 62 | } |
| 63 | } |
| 64 | |
| 65 | return res; |
| 66 | } |
| 67 | |
| 68 | // addWithRecursiveDependencies adds a requirement along with all its dependencies (according to the dependency map) to the requirements set |
| 69 | void addWithRecursiveDependencies( |
| 70 | kj::StringPtr requirement, const DepMap &depMap, kj::HashSet<kj::String> &requirementsSet) { |
| 71 | auto normalizedName = canonicalizePythonPackageName(requirement); |
| 72 | if (requirementsSet.contains(normalizedName)) { |
| 73 | return; |
| 74 | } |
| 75 | |
| 76 | requirementsSet.insert(kj::str(normalizedName)); |
| 77 | |
| 78 | KJ_IF_SOME(deps, depMap.find(normalizedName)) { |
| 79 | for (const auto &dep: deps) { |
| 80 | addWithRecursiveDependencies(dep, depMap, requirementsSet); |
| 81 | } |
| 82 | } |
| 83 | } |
| 84 | |
| 85 | kj::Own<capnp::List<capnp::json::Value::Field>::Reader> parseLockFile( |
| 86 | kj::StringPtr lockFileContents) { |
| 87 | capnp::JsonCodec json; |
| 88 | capnp::MallocMessageBuilder message; |
| 89 | |
| 90 | auto lock = message.initRoot<capnp::JsonValue>(); |
| 91 | json.decodeRaw(lockFileContents, lock); |
| 92 | |
| 93 | auto object = lock.getObject().asReader(); |
| 94 | auto packages = getField(object, "packages").getObject(); |
| 95 | return capnp::clone(packages); |
| 96 | } |
| 97 | |
| 98 | kj::HashSet<kj::String> getPythonPackageNames( |
| 99 | capnp::List<capnp::json::Value::Field>::Reader packages, |
| 100 | const DepMap &depMap, |
| 101 | kj::ArrayPtr<kj::String> requirements, |
| 102 | kj::StringPtr packagesVersion) { |
| 103 | |
| 104 | kj::HashSet<kj::String> allRequirements; // Requirements including their recursive dependencies. |
| 105 | |
| 106 | // Potentially add the stdlib packages and their recursive dependencies. |
| 107 | // TODO: Loading stdlib and its dependencies breaks package snapshots on "20240829.4". |
| 108 | // Remove this version check once a new package/python release is made. |
| 109 | if (packagesVersion != "20240829.4") { |
| 110 | // We need to scan the packages list for any packages that need to be included because they |
| 111 | // are part of Python's stdlib (hashlib etc). These need to be implicitly treated as part of |
| 112 | // our `requirements`. |
| 113 | for (const auto &ent: packages) { |
| 114 | auto name = ent.getName(); |
| 115 | auto obj = ent.getValue().getObject(); |
| 116 | auto packageType = getField(obj, "package_type").getString(); |
| 117 | |
| 118 | if (packageType == "cpython_module"_kj) { |
| 119 | addWithRecursiveDependencies(name, depMap, allRequirements); |
| 120 | } |
| 121 | } |
| 122 | } |
| 123 | |
| 124 | // Add all recursive dependencies of each requirement. |
| 125 | for (const auto &req: requirements) { |
| 126 | addWithRecursiveDependencies(req, depMap, allRequirements); |
| 127 | } |
| 128 | |
| 129 | return allRequirements; |
| 130 | } |
| 131 | |
| 132 | } // namespace workerd::api::pyodide |