feat(formats): expose some new customizable changelog formats on the commitizen.changelog_format endpoint (Textile, AsciiDoc and RestructuredText)

Author: noirbizarre

commitizen/changelog.py

@@ -26,37 +26,44 @@
 """
 from __future__ import annotations
 
-import os
 import re
 from collections import OrderedDict, defaultdict
+from dataclasses import dataclass
 from datetime import date
-from typing import TYPE_CHECKING, Callable, Iterable, cast
+from typing import TYPE_CHECKING, Callable, Iterable
 
 from jinja2 import (
     BaseLoader,
     ChoiceLoader,
     Environment,
     FileSystemLoader,
-    PackageLoader,
     Template,
 )
 
 from commitizen import out
 from commitizen.bump import normalize_tag
-from commitizen.defaults import encoding
 from commitizen.exceptions import InvalidConfigurationError, NoCommitsFoundError
 from commitizen.git import GitCommit, GitTag
 from commitizen.version_schemes import (
     DEFAULT_SCHEME,
     BaseVersion,
     InvalidVersion,
-    Pep440,
 )
 
 if TYPE_CHECKING:
     from commitizen.version_schemes import VersionScheme
 
-DEFAULT_TEMPLATE = "CHANGELOG.md.j2"
+
+@dataclass
+class Metadata:
+    """
+    Metadata extracted from the changelog produced by a plugin
+    """
+
+    unreleased_start: int | None = None
+    unreleased_end: int | None = None
+    latest_version: str | None = None
+    latest_version_position: int | None = None
 
 
 def get_commit_tag(commit: GitCommit, tags: list[GitTag]) -> GitTag | None:
@@ -196,100 +203,31 @@ def order_changelog_tree(tree: Iterable, change_type_order: list[str]) -> Iterab
     return sorted_tree
 
 
-def get_changelog_template(
-    loader: BaseLoader | None = None, template: str | None = None
-) -> Template:
+def get_changelog_template(loader: BaseLoader, template: str) -> Template:
     loader = ChoiceLoader(
         [
             FileSystemLoader("."),
-            loader or PackageLoader("commitizen", "templates"),
+            loader,
         ]
     )
     env = Environment(loader=loader, trim_blocks=True)
-    return env.get_template(template or DEFAULT_TEMPLATE)
+    return env.get_template(template)
 
 
 def render_changelog(
     tree: Iterable,
-    loader: BaseLoader | None = None,
-    template: str | None = None,
+    loader: BaseLoader,
+    template: str,
     **kwargs,
 ) -> str:
-    jinja_template = get_changelog_template(loader, template or DEFAULT_TEMPLATE)
+    jinja_template = get_changelog_template(loader, template)
     changelog: str = jinja_template.render(tree=tree, **kwargs)
     return changelog
 
 
-def parse_version_from_markdown(
-    value: str, scheme: VersionScheme = Pep440
-) -> str | None:
-    if not value.startswith("#"):
-        return None
-    m = scheme.parser.search(value)
-    if not m:
-        return None
-    return cast(str, m.group("version"))
-
-
-def parse_title_type_of_line(value: str) -> str | None:
-    md_title_parser = r"^(?P<title>#+)"
-    m = re.search(md_title_parser, value)
-    if not m:
-        return None
-    return m.groupdict().get("title")
-
-
-def get_metadata(
-    filepath: str, scheme: VersionScheme = Pep440, encoding: str = encoding
-) -> dict:
-    unreleased_start: int | None = None
-    unreleased_end: int | None = None
-    unreleased_title: str | None = None
-    latest_version: str | None = None
-    latest_version_position: int | None = None
-    if not os.path.isfile(filepath):
-        return {
-            "unreleased_start": None,
-            "unreleased_end": None,
-            "latest_version": None,
-            "latest_version_position": None,
-        }
-
-    with open(filepath, encoding=encoding) as changelog_file:
-        for index, line in enumerate(changelog_file):
-            line = line.strip().lower()
-
-            unreleased: str | None = None
-            if "unreleased" in line:
-                unreleased = parse_title_type_of_line(line)
-            # Try to find beginning and end lines of the unreleased block
-            if unreleased:
-                unreleased_start = index
-                unreleased_title = unreleased
-                continue
-            elif (
-                isinstance(unreleased_title, str)
-                and parse_title_type_of_line(line) == unreleased_title
-            ):
-                unreleased_end = index
-
-            # Try to find the latest release done
-            version = parse_version_from_markdown(line, scheme)
-            if version:
-                latest_version = version
-                latest_version_position = index
-                break  # there's no need for more info
-        if unreleased_start is not None and unreleased_end is None:
-            unreleased_end = index
-    return {
-        "unreleased_start": unreleased_start,
-        "unreleased_end": unreleased_end,
-        "latest_version": latest_version,
-        "latest_version_position": latest_version_position,
-    }
-
-
-def incremental_build(new_content: str, lines: list[str], metadata: dict) -> list[str]:
+def incremental_build(
+    new_content: str, lines: list[str], metadata: Metadata
+) -> list[str]:
     """Takes the original lines and updates with new_content.
 
     The metadata governs how to remove the old unreleased section and where to place the
@@ -303,9 +241,9 @@ def incremental_build(new_content: str, lines: list[str], metadata: dict) -> lis
     Returns:
         Updated lines
     """
-    unreleased_start = metadata.get("unreleased_start")
-    unreleased_end = metadata.get("unreleased_end")
-    latest_version_position = metadata.get("latest_version_position")
+    unreleased_start = metadata.unreleased_start
+    unreleased_end = metadata.unreleased_end
+    latest_version_position = metadata.latest_version_position
     skip = False
     output_lines: list[str] = []
     for index, line in enumerate(lines):

commitizen/changelog_formats/__init__.py

@@ -0,0 +1,90 @@
+from __future__ import annotations
+
+from typing import ClassVar, Protocol
+
+import importlib_metadata as metadata
+
+from commitizen.changelog import Metadata
+from commitizen.exceptions import ChangelogFormatUnknown
+from commitizen.config.base_config import BaseConfig
+
+
+CHANGELOG_FORMAT_ENTRYPOINT = "commitizen.changelog_format"
+TEMPLATE_EXTENSION = "j2"
+
+
+class ChangelogFormat(Protocol):
+    extension: ClassVar[str]
+    """Standard known extension associated with this format"""
+
+    alternative_extensions: ClassVar[set[str]]
+    """Known alternatives extensions for this format"""
+
+    config: BaseConfig
+
+    def __init__(self, config: BaseConfig):
+        self.config = config
+
+    @property
+    def ext(self) -> str:
+        """Dotted version of extensions, as in `pathlib` and `os` modules"""
+        return f".{self.extension}"
+
+    @property
+    def template(self) -> str:
+        """Expected template name for this format"""
+        return f"CHANGELOG.{self.extension}.{TEMPLATE_EXTENSION}"
+
+    @property
+    def default_changelog_file(self) -> str:
+        return f"CHANGELOG.{self.extension}"
+
+    def get_metadata(self, filepath: str) -> Metadata:
+        """
+        Extract the changelog metadata.
+        """
+        raise NotImplementedError
+
+
+KNOWN_CHANGELOG_FORMATS: dict[str, type[ChangelogFormat]] = {
+    ep.name: ep.load()
+    for ep in metadata.entry_points(group=CHANGELOG_FORMAT_ENTRYPOINT)
+}
+
+
+def get_changelog_format(
+    config: BaseConfig, filename: str | None = None
+) -> ChangelogFormat:
+    """
+    Get a format from its name
+
+    :raises FormatUnknown: if a non-empty name is provided but cannot be found in the known formats
+    """
+    name: str | None = config.settings.get("changelog_format")
+    format: type[ChangelogFormat] | None = guess_changelog_format(filename)
+
+    if name and name in KNOWN_CHANGELOG_FORMATS:
+        format = KNOWN_CHANGELOG_FORMATS[name]
+
+    if not format:
+        raise ChangelogFormatUnknown(f"Unknown changelog format '{name}'")
+
+    return format(config)
+
+
+def guess_changelog_format(filename: str | None) -> type[ChangelogFormat] | None:
+    """
+    Try guessing the file format from the filename.
+
+    Algorithm is basic, extension-based, and won't work
+    for extension-less file names like `CHANGELOG` or `NEWS`.
+    """
+    if not filename or not isinstance(filename, str):
+        return None
+    for format in KNOWN_CHANGELOG_FORMATS.values():
+        if filename.endswith(f".{format.extension}"):
+            return format
+        for alt_extension in format.alternative_extensions:
+            if filename.endswith(f".{alt_extension}"):
+                return format
+    return None

commitizen/changelog_formats/asciidoc.py

@@ -0,0 +1,27 @@
+from __future__ import annotations
+
+import re
+
+from .base import BaseFormat
+
+
+class AsciiDoc(BaseFormat):
+    extension = "adoc"
+
+    RE_TITLE = re.compile(r"^(?P<level>=+) (?P<title>.*)$")
+
+    def parse_version_from_title(self, line: str) -> str | None:
+        m = self.RE_TITLE.match(line)
+        if not m:
+            return None
+        # Capture last match as AsciiDoc use postfixed URL labels
+        matches = list(re.finditer(self.version_parser, m.group("title")))
+        if not matches:
+            return None
+        return matches[-1].group("version")
+
+    def parse_title_level(self, line: str) -> int | None:
+        m = self.RE_TITLE.match(line)
+        if not m:
+            return None
+        return len(m.group("level"))

commitizen/changelog_formats/base.py

@@ -0,0 +1,81 @@
+from __future__ import annotations
+
+import os
+from abc import ABCMeta
+from re import Pattern
+from typing import IO, Any, ClassVar
+
+from commitizen.changelog import Metadata
+from commitizen.config.base_config import BaseConfig
+from commitizen.version_schemes import get_version_scheme
+
+from . import ChangelogFormat
+
+
+class BaseFormat(ChangelogFormat, metaclass=ABCMeta):
+    """
+    Base class to extend to implement a changelog file format.
+    """
+
+    extension: ClassVar[str] = ""
+    alternative_extensions: ClassVar[set[str]] = set()
+
+    def __init__(self, config: BaseConfig):
+        # Constructor needs to be redefined because `Protocol` prevent instantiation by default
+        # See: https://bugs.python.org/issue44807
+        self.config = config
+
+    @property
+    def version_parser(self) -> Pattern:
+        return get_version_scheme(self.config).parser
+
+    def get_metadata(self, filepath: str) -> Metadata:
+        if not os.path.isfile(filepath):
+            return Metadata()
+
+        with open(filepath) as changelog_file:
+            return self.get_metadata_from_file(changelog_file)
+
+    def get_metadata_from_file(self, file: IO[Any]) -> Metadata:
+        meta = Metadata()
+        unreleased_level: int | None = None
+        for index, line in enumerate(file):
+            line = line.strip().lower()
+
+            unreleased: int | None = None
+            if "unreleased" in line:
+                unreleased = self.parse_title_level(line)
+            # Try to find beginning and end lines of the unreleased block
+            if unreleased:
+                meta.unreleased_start = index
+                unreleased_level = unreleased
+                continue
+            elif unreleased_level and self.parse_title_level(line) == unreleased_level:
+                meta.unreleased_end = index
+
+            # Try to find the latest release done
+            version = self.parse_version_from_title(line)
+            if version:
+                meta.latest_version = version
+                meta.latest_version_position = index
+                break  # there's no need for more info
+        if meta.unreleased_start is not None and meta.unreleased_end is None:
+            meta.unreleased_end = index
+
+        return meta
+
+    def parse_version_from_title(self, line: str) -> str | None:
+        """
+        Extract the version from a title line if any
+        """
+        raise NotImplementedError(
+            "Default `get_metadata_from_file` requires `parse_version_from_changelog` to be implemented"
+        )
+
+    def parse_title_level(self, line: str) -> int | None:
+        """
+        Get the title level/type of a line if any
+        """
+        raise NotImplementedError(
+            "Default `get_metadata_from_file` requires `parse_title_type_of_line` to be implemented"
+        )