Merge pull request #896 from RonnyPfannschmidt/fix-883-robust-mime-parsing

fix mime parsing - ignore potential mime headers after end of headers

Author: RonnyPfannschmidt

CHANGELOG.rst

@@ -39,6 +39,11 @@ features
 * support using rich.logging as console log handler if installed
 * fix #527: type annotation in default version template
 
+bugfixes
+--------
+
+* fix #883: use HeadersParser to ensure only mime metadata in headers is used
+
 v7.1.0
 ======
 

src/setuptools_scm/integration.py

@@ -9,11 +9,22 @@
 log = logging.getLogger(__name__)
 
 
-def data_from_mime(path: _t.PathT) -> dict[str, str]:
-    content = Path(path).read_text(encoding="utf-8")
+def data_from_mime(path: _t.PathT, content: None | str = None) -> dict[str, str]:
+    """return a mapping from mime/pseudo-mime content
+    :param path: path to the mime file
+    :param content: content of the mime file, if None, read from path
+    :rtype: dict[str, str]
+
+    """
+
+    if content is None:
+        content = Path(path).read_text(encoding="utf-8")
     log.debug("mime %s content:\n%s", path, textwrap.indent(content, "    "))
-    # the complex conditions come from reading pseudo-mime-messages
-    data = dict(x.split(": ", 1) for x in content.splitlines() if ": " in x)
 
+    from email.parser import HeaderParser
+
+    parser = HeaderParser()
+    message = parser.parsestr(content)
+    data = dict(message.items())
     log.debug("mime %s data:\n%s", path, data)
     return data

testing/test_regressions.py

@@ -12,6 +12,14 @@
 from setuptools_scm import Configuration
 from setuptools_scm._run_cmd import run
 from setuptools_scm.git import parse
+from setuptools_scm.integration import data_from_mime
+
+
+def test_data_from_mime_ignores_body() -> None:
+    assert data_from_mime(
+        "test",
+        "version: 1.0\r\n\r\nversion: bad",
+    ) == {"version": "1.0"}
 
 
 def test_pkginfo_noscmroot(tmp_path: Path, monkeypatch: pytest.MonkeyPatch) -> None: