Add tag_filter as a way to pre-filter tags

Fixes #219

.github/workflows/ci.yml

@@ -36,6 +36,27 @@ jobs:
         run: |
           python -m pip install -U pip pytest setuptools editables
           pip install .
+      - name: Install Mercurial
+        shell: bash
+        run: |
+          case "$RUNNER_OS" in
+            "Linux")
+              sudo apt install mercurial
+              ;;
+            "Windows")
+              choco install hg
+              ;;
+            "macOS")
+              brew install mercurial
+              ;;
+            "*")
+              echo "$RUNNER_OS not supported"
+              exit 1
+              ;;
+          esac
+
+          echo "[ui]" >> ~/.hgrc
+          echo "username = \"John Doe <[email protected]>\"" >> ~/.hgrc
       - name: Run Tests
         run: |
           git config --global user.name "John Doe"

docs/metadata.md

@@ -53,14 +53,25 @@ Alternatively, you can specify a default version in the configuration:
 fallback_version = "0.0.0"
 ```
 
-You can specify another regex pattern to match the SCM tag, in which a `version` group is required:
+To control which scm tags are used to generate the version, you can use two
+fields: `tag_filter` and `tag_regex`.
 
 ```toml
 [tool.pdm.version]
 source = "scm"
-tag_regex = '^(?:\D*)?(?P<version>([1-9][0-9]*!)?(0|[1-9][0-9]*)(\.(0|[1-9][0-9]*))*((a|b|c|rc)(0|[1-9][0-9]*))?(\.post(0|[1-9][0-9]*))?(\.dev(0|[1-9][0-9]*))?$)$'
+tag_filter = "test/*"
+tag_regex = '^test/(?:\D*)?(?P<version>([1-9][0-9]*!)?(0|[1-9][0-9]*)(\.(0|[1-9][0-9]*))*((a|b|c|rc)(0|[1-9][0-9]*))?(\.post(0|[1-9][0-9]*))?(\.dev(0|[1-9][0-9]*))?$)$'
 ```
 
+`tag_filter` filters the set of tags which are considered as candidates to
+capture your project's version. For `git` repositories, this field is a glob
+matched against the tag. For `hg` repositories, it is a regular expression used
+with the `latesttag` function.
+
+`tag_regex` configures how you extract a version from a tag. It is applied after
+`tag_filter` extracts candidate tags to extract the version from that tag. It is
+a python style regular expression.
+
 +++ 2.2.0
 
 To customize the format of the version string, specify the `version_format` option with a format function:
@@ -117,10 +128,10 @@ write_template = "__version__ = '{}'"
 ```
 
 !!! note
-    The path in `write_to` is relative to the root of the wheel file, hence the `package-dir` part should be stripped.
+The path in `write_to` is relative to the root of the wheel file, hence the `package-dir` part should be stripped.
 
 !!! note
-    `pdm-backend` will rewrite the whole file each time, so you can't have additional contents in that file.
+`pdm-backend` will rewrite the whole file each time, so you can't have additional contents in that file.
 
 ## Variables expansion
 
@@ -159,7 +170,7 @@ dependencies = [
 ```
 
 !!! note
-    The triple slashes `///` is required for the compatibility of Windows and POSIX systems.
+The triple slashes `///` is required for the compatibility of Windows and POSIX systems.
 
 !!! note
-    The relative paths will be expanded into the absolute paths on the local machine. So it makes no sense to include them in a distribution, since others who install the package will not have the same paths.
+The relative paths will be expanded into the absolute paths on the local machine. So it makes no sense to include them in a distribution, since others who install the package will not have the same paths.

src/pdm/backend/hooks/version/__init__.py

@@ -75,6 +75,7 @@ def resolve_version_from_scm(
         write_to: str | None = None,
         write_template: str = "{}\n",
         tag_regex: str | None = None,
+        tag_filter: str | None = None,
         version_format: str | None = None,
         fallback_version: str | None = None,
     ) -> str:
@@ -88,7 +89,10 @@ def resolve_version_from_scm(
             else:
                 version_formatter = None
             version = get_version_from_scm(
-                context.root, tag_regex=tag_regex, version_formatter=version_formatter
+                context.root,
+                tag_regex=tag_regex,
+                version_formatter=version_formatter,
+                tag_filter=tag_filter,
             )
             if version is None:
                 if fallback_version is not None:

src/pdm/backend/hooks/version/scm.py

@@ -14,7 +14,7 @@
 from dataclasses import dataclass
 from datetime import datetime, timezone
 from pathlib import Path
-from typing import TYPE_CHECKING, Callable, Iterable, NamedTuple
+from typing import TYPE_CHECKING, Callable, NamedTuple
 
 from pdm.backend._vendor.packaging.version import Version
 
@@ -29,6 +29,7 @@
 @dataclass(frozen=True)
 class Config:
     tag_regex: re.Pattern
+    tag_filter: str | None
 
 
 def _subprocess_call(
@@ -164,28 +165,27 @@ def tag_to_version(config: Config, tag: str) -> Version:
     return Version(version)
 
 
-def tags_to_versions(config: Config, tags: Iterable[str]) -> list[Version]:
-    """
-    take tags that might be prefixed with a keyword and return only the version part
-    :param tags: an iterable of tags
-    :param config: optional configuration object
-    """
-    return [tag_to_version(config, tag) for tag in tags if tag]
-
-
 def git_parse_version(root: StrPath, config: Config) -> SCMVersion | None:
-    GIT = shutil.which("git")
-    if not GIT:
+    git = shutil.which("git")
+    if not git:
         return None
 
-    ret, repo, _ = _subprocess_call([GIT, "rev-parse", "--show-toplevel"], root)
+    ret, repo, _ = _subprocess_call([git, "rev-parse", "--show-toplevel"], root)
     if ret or not repo:
         return None
 
     if os.path.isfile(os.path.join(repo, ".git/shallow")):
         warnings.warn(f"{repo!r} is shallow and may cause errors")
-    describe_cmd = [GIT, "describe", "--dirty", "--tags", "--long", "--match", "*.*"]
-    ret, output, err = _subprocess_call(describe_cmd, repo)
+    describe_cmd = [
+        git,
+        "describe",
+        "--dirty",
+        "--tags",
+        "--long",
+        "--match",
+        config.tag_filter or "*.*",
+    ]
+    ret, output, _ = _subprocess_call(describe_cmd, repo)
     branch = _git_get_branch(repo)
 
     if ret:
@@ -201,54 +201,44 @@ def git_parse_version(root: StrPath, config: Config) -> SCMVersion | None:
         return meta(config, tag, number or None, dirty, node, branch)
 
 
-def get_latest_normalizable_tag(root: StrPath) -> str:
-    # Gets all tags containing a '.' from oldest to newest
-    cmd = [
-        "hg",
-        "log",
-        "-r",
-        "ancestors(.) and tag('re:\\.')",
-        "--template",
-        "{tags}\n",
-    ]
-    _, output, _ = _subprocess_call(cmd, root)
-    outlines = output.split()
-    if not outlines:
-        return "null"
-    tag = outlines[-1].split()[-1]
-    return tag
+def get_distance_revset(tag: str | None) -> str:
+    return (
+        "(branch(.)"  # look for revisions in this branch only
+        " and {rev}::."  # after the last tag
+        # ignore commits that only modify .hgtags and nothing else:
+        " and (merge() or file('re:^(?!\\.hgtags).*$'))"
+        " and not {rev})"  # ignore the tagged commit itself
+    ).format(rev=f"tag({tag!r})" if tag is not None else "null")
 
 
-def hg_get_graph_distance(root: StrPath, rev1: str, rev2: str = ".") -> int:
-    cmd = ["hg", "log", "-q", "-r", f"{rev1}::{rev2}"]
+def hg_get_graph_distance(root: StrPath, tag: str | None) -> int:
+    cmd = ["hg", "log", "-q", "-r", get_distance_revset(tag)]
     _, out, _ = _subprocess_call(cmd, root)
-    return len(out.strip().splitlines()) - 1
+    return len(out.strip().splitlines())
 
 
 def _hg_tagdist_normalize_tagcommit(
-    config: Config, root: StrPath, tag: str, dist: int, node: str, branch: str
+    config: Config,
+    root: StrPath,
+    tag: str,
+    dist: int,
+    node: str,
+    branch: str,
+    dirty: bool,
 ) -> SCMVersion:
-    dirty = node.endswith("+")
-    node = "h" + node.strip("+")
-
     # Detect changes since the specified tag
-    revset = (
-        "(branch(.)"  # look for revisions in this branch only
-        " and tag({tag!r})::."  # after the last tag
-        # ignore commits that only modify .hgtags and nothing else:
-        " and (merge() or file('re:^(?!\\.hgtags).*$'))"
-        " and not tag({tag!r}))"  # ignore the tagged commit itself
-    ).format(tag=tag)
     if tag != "0.0":
         _, commits, _ = _subprocess_call(
-            ["hg", "log", "-r", revset, "--template", "{node|short}"],
+            ["hg", "log", "-r", get_distance_revset(tag), "--template", "{node|short}"],
             root,
         )
     else:
         commits = "True"
 
     if commits or dirty:
-        return meta(config, tag, distance=dist, node=node, dirty=dirty, branch=branch)
+        return meta(
+            config, tag, distance=dist or None, node=node, dirty=dirty, branch=branch
+        )
     else:
         return meta(config, tag)
 
@@ -280,32 +270,40 @@ def _bump_regex(version: str) -> str:
 
 
 def hg_parse_version(root: StrPath, config: Config) -> SCMVersion | None:
-    if not shutil.which("hg"):
+    hg = shutil.which("hg")
+    if not hg:
         return None
-    _, output, _ = _subprocess_call("hg id -i -b -t", root)
-    identity_data = output.split()
-    if not identity_data:
-        return None
-    node = identity_data.pop(0)
-    branch = identity_data.pop(0)
-    if "tip" in identity_data:
-        # tip is not a real tag
-        identity_data.remove("tip")
-    tags = tags_to_versions(config, identity_data)
-    dirty = node[-1] == "+"
-    if tags:
-        return meta(config, tags[0], dirty=dirty, branch=branch)
-
-    if node.strip("+") == "0" * 12:
-        return meta(config, "0.0", dirty=dirty, branch=branch)
 
+    tag_filter = config.tag_filter or "\\."
+    _, output, _ = _subprocess_call(
+        [
+            hg,
+            "log",
+            "-r",
+            ".",
+            "--template",
+            f"{{latesttag(r're:{tag_filter}')}}-{{node|short}}-{{branch}}",
+        ],
+        root,
+    )
+    tag: str | None
+    tag, node, branch = output.rsplit("-", 2)
+    # If no tag exists passes the tag filter.
+    if tag == "null":
+        tag = None
+
+    _, id_output, _ = _subprocess_call(
+        [hg, "id", "-i"],
+        root,
+    )
+    dirty = id_output.endswith("+")
     try:
-        tag = get_latest_normalizable_tag(root)
         dist = hg_get_graph_distance(root, tag)
-        if tag == "null":
+        if tag is None:
             tag = "0.0"
-            dist = int(dist) + 1
-        return _hg_tagdist_normalize_tagcommit(config, root, tag, dist, node, branch)
+        return _hg_tagdist_normalize_tagcommit(
+            config, root, tag, dist, node, branch, dirty=dirty
+        )
     except ValueError:
         return None  # unpacking failed, old hg
 
@@ -332,11 +330,15 @@ def get_version_from_scm(
     root: str | Path,
     *,
     tag_regex: str | None = None,
+    tag_filter: str | None = None,
     version_formatter: Callable[[SCMVersion], str] | None = None,
 ) -> str | None:
-    config = Config(tag_regex=re.compile(tag_regex) if tag_regex else DEFAULT_TAG_REGEX)
+    config = Config(
+        tag_regex=re.compile(tag_regex) if tag_regex else DEFAULT_TAG_REGEX,
+        tag_filter=tag_filter,
+    )
     for func in (git_parse_version, hg_parse_version):
-        version = func(root, config)  # type: ignore
+        version = func(root, config)
         if version:
             if version_formatter is None:
                 version_formatter = format_version

src/pdm/backend/utils.py

@@ -107,7 +107,7 @@ def _build_filter(patterns: Iterable[str]) -> Callable[[str], bool]:
 
 
 @contextmanager
-def cd(path: str) -> Generator[None, None, None]:
+def cd(path: str | Path) -> Generator[None, None, None]:
     _old_cwd = os.getcwd()
     os.chdir(path)
     try: