contrib/relnotes

storageutil: extract copy metadata retrieval out of filelog

As part of implementing an alternate storage backend, I found myself
reinventing this wheel.

Let's create a utility function for doing the work.

Differential Revision: https://phab.mercurial-scm.org/D4800

#!/usr/bin/env python3
"""Generate release notes from our commit log.

This uses the relnotes extension directives when they're available,
and falls back to our old pre-relnotes logic that used to live in the
release-tools repo.
"""
import argparse
import re
import subprocess

# Regenerate this list with
#   hg export 'grep("\.\. [a-z]+::")' | grep '^\.\.' | \
#     sed 's/.. //;s/::.*//' | sort -u
rnsections = ["api", "bc", "container", "feature", "fix", "note", "perf"]

rules = {
    # keep
    r"\(issue": 100,
    r"\(BC\)": 100,
    r"\(API\)": 100,
    # core commands, bump up
    r"(commit|files|log|pull|push|patch|status|tag|summary)(|s|es):": 20,
    r"(annotate|alias|branch|bookmark|clone|graft|import|verify).*:": 20,
    # extensions, bump up
    r"(mq|shelve|rebase):": 20,
    # newsy
    r": deprecate": 20,
    r"(option|feature|command|support)": 10,
    # bug-like?
    r"(fix|don't break|improve)": 7,
    # boring stuff, bump down
    r"^contrib": -5,
    r"debug": -5,
    r"help": -5,
    r"(doc|bundle2|obsolete|obsmarker|rpm|setup|debug\S+:)": -15,
    r"(check-code|check-commit|import-checker)": -20,
    # cleanups and refactoring
    r"(cleanup|whitespace|nesting|indent|spelling|comment)": -20,
    r"(typo|hint|note|style:|correct doc)": -20,
    r"_": -10,
    r"(argument|absolute_import|attribute|assignment|mutable)": -15,
    r"(unused|useless|unnecessary|duplicate|deprecated|scope|True|False)": -10,
    r"(redundant|pointless|confusing|uninitialized|meaningless|dead)": -10,
    r": (drop|remove|inherit|rename|simplify|naming|inline)": -10,
    r"(docstring|document .* method)": -20,
    r"(factor|extract|prepare|split|replace| import)": -20,
    r": add.*(function|method|implementation|test|example)": -10,
    r": (move|extract) .* (to|into|from)": -20,
    r": implement ": -5,
    r": use .* implementation": -20,
    r"\S\S\S+\.\S\S\S\S+": -5,
    r": use .* instead of": -20,
    r"__": -5,
    # dumb keywords
    r"\S+/\S+:": -10,
    r"\S+\.\S+:": -10,
    # drop
    r"^i18n-": -50,
    r"^i18n:.*(hint|comment)": -50,
    r"perf:": -50,
    r"check-code:": -50,
    r"Added.*for changeset": -50,
    r"tests?:": -50,
    r"test-": -50,
    r"add.* tests": -50,
    r"^_": -50,
}

cutoff = 10
commits = []

groupings = [
    (r"util|parsers|repo|ctx|context|revlog|filelog|alias|cmdutil", "core"),
    (r"revset|templater|ui|dirstate|hook|i18n|transaction|wire", "core"),
    (r"color|pager", "core"),
    (r"hgweb|paper|coal|gitweb", "hgweb"),
    (r"pull|push|revert|resolve|annotate|bookmark|branch|clone", "commands"),
    (r"commands|commit|config|files|graft|import|log|merge|patch", "commands"),
    (r"phases|status|summary|amend|tag|help|verify", "commands"),
    (r"rebase|mq|convert|eol|histedit|largefiles", "extensions"),
    (r"shelve|unshelve", "extensions"),
]

def main():
    ap = argparse.ArgumentParser()
    ap.add_argument(
        "startrev",
        metavar="REV",
        type=str,
        nargs=1,
        help=(
            "Starting revision for the release notes. This revision "
            "won't be included, but later revisions will."
        ),
    )
    ap.add_argument(
        "--stoprev",
        metavar="REV",
        type=str,
        default="@",
        nargs=1,
        help=(
            "Stop revision for release notes. This revision will be included,"
            " but no later revisions will. This revision needs to be "
            "a descendant of startrev."
        ),
    )
    args = ap.parse_args()
    fromext = subprocess.check_output(
        [
            "hg",
            "--config",
            "extensions.releasenotes=",
            "releasenotes",
            "-r",
            "%s::%s" % (args.startrev[0], args.stoprev[0]),
        ]
    ).decode("utf-8")
    # Find all release notes from un-relnotes-flagged commits.
    for entry in sorted(
        subprocess.check_output(
            [
                "hg",
                "log",
                "-r",
                r'%s::%s - merge() - grep("\n\.\. (%s)::")'
                % (args.startrev[0], args.stoprev[0], "|".join(rnsections)),
                "-T",
                r"{desc|firstline}\n",
            ]
        )
        .decode("utf-8")
        .splitlines()
    ):
        desc = entry.replace("`", "'")

        score = 0
        for rule, val in rules.items():
            if re.search(rule, desc):
                score += val

        desc = desc.replace("(issue", "(Bts:issue")

        if score >= cutoff:
            commits.append(desc)
    # Group unflagged notes.
    groups = {}
    bcs = []
    apis = []

    for d in commits:
        if "(BC)" in d:
            bcs.append(d)
        if "(API)" in d:
            apis.append(d)
        for rule, g in groupings:
            if re.match(rule, d):
                groups.setdefault(g, []).append(d)
                break
        else:
            groups.setdefault("unsorted", []).append(d)
    print(fromext)
    # print legacy release notes sections
    for g in sorted(groups):
        print("\n=== %s ===" % g)
        for d in sorted(groups[g]):
            print(" * %s" % d)

    print("\n=== BC ===\n")

    for d in sorted(bcs):
        print(" * %s" % d)

    print("\n=== API Changes ===\n")

    for d in sorted(apis):
        print(" * %s" % d)

if __name__ == "__main__":
    main()