Upload: All metadata incl. PEP 639

We were previously not uploading all metadata in the formdata of an upload request in the legacy api. Notably, we were missing the PEP 639 license-files field.

I had to switch to pdm due to pypa/hatch#1828

crates/uv-publish/src/lib.rs

@@ -613,20 +613,49 @@ async fn form_metadata(
 ) -> Result<Vec<(&'static str, String)>, PublishPrepareError> {
     let hash_hex = hash_file(file, Hasher::from(HashAlgorithm::Sha256)).await?;
 
-    let metadata = metadata(file, filename).await?;
+    let Metadata23 {
+        metadata_version,
+        name,
+        version,
+        platforms,
+        // Not used by PyPI legacy upload
+        supported_platforms: _,
+        summary,
+        description,
+        description_content_type,
+        keywords,
+        home_page,
+        download_url,
+        author,
+        author_email,
+        maintainer,
+        maintainer_email,
+        license,
+        license_expression,
+        license_files,
+        classifiers,
+        requires_dist,
+        provides_dist,
+        obsoletes_dist,
+        requires_python,
+        requires_external,
+        project_urls,
+        provides_extras,
+        dynamic,
+    } = metadata(file, filename).await?;
 
     let mut form_metadata = vec![
         (":action", "file_upload".to_string()),
         ("sha256_digest", hash_hex.digest.to_string()),
         ("protocol_version", "1".to_string()),
-        ("metadata_version", metadata.metadata_version.clone()),
+        ("metadata_version", metadata_version.clone()),
         // Twine transforms the name with `re.sub("[^A-Za-z0-9.]+", "-", name)`
         // * <https://github.com/pypa/twine/issues/743>
         // * <https://github.com/pypa/twine/blob/5bf3f38ff3d8b2de47b7baa7b652c697d7a64776/twine/package.py#L57-L65>
         // warehouse seems to call `packaging.utils.canonicalize_name` nowadays and has a separate
         // `normalized_name`, so we'll start with this and we'll readjust if there are user reports.
-        ("name", metadata.name.clone()),
-        ("version", metadata.version.clone()),
+        ("name", name.clone()),
+        ("version", version.clone()),
         ("filetype", filename.filetype().to_string()),
     ];
 
@@ -642,41 +671,39 @@ async fn form_metadata(
         }
     };
 
-    add_option("summary", metadata.summary);
-    add_option("description", metadata.description);
-    add_option(
-        "description_content_type",
-        metadata.description_content_type,
-    );
-    add_option("author", metadata.author);
-    add_option("author_email", metadata.author_email);
-    add_option("maintainer", metadata.maintainer);
-    add_option("maintainer_email", metadata.maintainer_email);
-    add_option("license", metadata.license);
-    add_option("keywords", metadata.keywords);
-    add_option("home_page", metadata.home_page);
-    add_option("download_url", metadata.download_url);
+    add_option("author", author);
+    add_option("author_email", author_email);
+    add_option("description", description);
+    add_option("description_content_type", description_content_type);
+    add_option("download_url", download_url);
+    add_option("home_page", home_page);
+    add_option("keywords", keywords);
+    add_option("license", license);
+    add_option("license_expression", license_expression);
+    add_option("maintainer", maintainer);
+    add_option("maintainer_email", maintainer_email);
+    add_option("summary", summary);
 
     // The GitLab PyPI repository API implementation requires this metadata field and twine always
     // includes it in the request, even when it's empty.
-    form_metadata.push((
-        "requires_python",
-        metadata.requires_python.unwrap_or(String::new()),
-    ));
+    form_metadata.push(("requires_python", requires_python.unwrap_or(String::new())));
 
     let mut add_vec = |name, values: Vec<String>| {
         for i in values {
             form_metadata.push((name, i.clone()));
         }
     };
 
-    add_vec("classifiers", metadata.classifiers);
-    add_vec("platform", metadata.platforms);
-    add_vec("requires_dist", metadata.requires_dist);
-    add_vec("provides_dist", metadata.provides_dist);
-    add_vec("obsoletes_dist", metadata.obsoletes_dist);
-    add_vec("requires_external", metadata.requires_external);
-    add_vec("project_urls", metadata.project_urls);
+    add_vec("classifiers", classifiers);
+    add_vec("dynamic", dynamic);
+    add_vec("license_file", license_files);
+    add_vec("obsoletes_dist", obsoletes_dist);
+    add_vec("platform", platforms);
+    add_vec("project_urls", project_urls);
+    add_vec("provides_dist", provides_dist);
+    add_vec("provides_extra", provides_extras);
+    add_vec("requires_dist", requires_dist);
+    add_vec("requires_external", requires_external);
 
     Ok(form_metadata)
 }

scripts/publish/test_publish.py

@@ -52,6 +52,7 @@
 
 import os
 import re
+import shutil
 import time
 from argparse import ArgumentParser
 from dataclasses import dataclass
@@ -70,6 +71,30 @@
 
 TEST_PYPI_PUBLISH_URL = "https://test.pypi.org/legacy/"
 PYTHON_VERSION = os.environ.get("UV_TEST_PUBLISH_PYTHON_VERSION", "3.12")
+# `pyproject.toml` contents using all supported metadata fields, except for the
+# generated header with `[project]`, name and version.
+PYPROJECT_TAIL = """
+authors = [{ name = "konstin", email = "[email protected]" }]
+classifiers = ["Topic :: Software Development :: Testing"]
+# Empty for simplicity with the `uv compile` check, anyio still tests,
+# optional-dependencies still test the `Requires-Dist` field.
+dependencies = []
+description = "Add your description here"
+dynamic = ["gui-scripts", "scripts"]
+keywords = ["test", "publish"]
+license = "MIT OR Apache-2.0"
+license-files = ["LICENSE*"]
+maintainers = [{ name = "konstin", email = "[email protected]" }]
+optional-dependencies = { "async" = ["anyio>=4,<5"] }
+readme = "README.md"
+requires-python = ">=3.12"
+urls = { "github" = "https://github.com/astral-sh/uv" }
+
+# https://github.com/pypa/hatch/issues/1828
+[build-system]
+requires = ["pdm-backend"]
+build-backend = "pdm.backend"
+""".lstrip()
 
 cwd = Path(__file__).parent
 
@@ -200,12 +225,22 @@ def build_project_at_version(
         [uv, "init", "-p", PYTHON_VERSION, "--lib", "--name", project_name, dir_name],
         cwd=cwd,
     )
-    pyproject_toml = project_root.joinpath("pyproject.toml")
-
-    # Set to an unclaimed version
-    toml = pyproject_toml.read_text()
-    toml = re.sub('version = ".*"', f'version = "{version}"', toml)
-    pyproject_toml.write_text(toml)
+    project_root.joinpath("pyproject.toml").write_text(
+        "[project]\n"
+        + f'name = "{project_name}"\n'
+        # Set to an unclaimed version
+        + f'version = "{version}"\n'
+        # Add all supported metadata
+        + PYPROJECT_TAIL
+    )
+    shutil.copy(
+        cwd.parent.parent.joinpath("LICENSE-APACHE"),
+        cwd.joinpath(dir_name).joinpath("LICENSE-APACHE"),
+    )
+    shutil.copy(
+        cwd.parent.parent.joinpath("LICENSE-MIT"),
+        cwd.joinpath(dir_name).joinpath("LICENSE-MIT"),
+    )
 
     # Modify the code so we get a different source dist and wheel
     if modified:
@@ -262,8 +297,11 @@ def wait_for_index(
             break
 
         print(
-            f"uv pip compile not updated, missing 2 files for {version}: `{output.replace("\\\n    ", "")}`, "
-            f"sleeping for 2s: `{index_url}`"
+            f"uv pip compile not updated, missing 2 files for {version}, "
+            + f"sleeping for 2s: `{index_url}`:\n"
+            + f"```\n"
+            + output.replace("\\\n    ", "")
+            + "```"
         )
         sleep(2)