[clang-tidy] Implement alphabetical order test

clang-tools-extra/clang-tidy/tool/clang-tidy-alphabetical-order-check.py

@@ -0,0 +1,285 @@
+#!/usr/bin/env python3
+#
+# ===-----------------------------------------------------------------------===#
+#
+# Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+# See https://llvm.org/LICENSE.txt for license information.
+# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+#
+# ===-----------------------------------------------------------------------===#
+
+"""
+
+ClangTidy Alphabetical Order Checker
+====================================
+
+Normalize clang-tidy docs with deterministic sorting for linting/tests.
+
+Behavior:
+- Sort entries in docs/clang-tidy/checks/list.rst csv-table.
+- Sort key sections in docs/ReleaseNotes.rst. Does not remove duplicate
+  entries; developers should merge duplicates manually when needed.
+
+Flags:
+  -o/--output  Write normalized content to this path instead of updating docs.
+"""
+
+import argparse
+import io
+import os
+import re
+import sys
+from typing import List, Optional, Sequence, Tuple
+
+# Matches a :doc:`label <path>` or :doc:`label` reference anywhere in text and
+# captures the label. Used to sort bullet items alphabetically in ReleaseNotes
+# items by their label.
+DOC_LABEL_RN_RE = re.compile(r":doc:`(?P<label>[^`<]+)\s*(?:<[^>]+>)?`")
+
+# Matches a single csv-table row line in list.rst that begins with a :doc:
+# reference, capturing the label. Used to extract the sort key per row.
+DOC_LINE_RE = re.compile(r"^\s*:doc:`(?P<label>[^`<]+?)\s*<[^>]+>`.*$")
+
+
+def script_dir() -> str:
+    return os.path.dirname(os.path.abspath(__file__))
+
+
+def read_text(path: str) -> List[str]:
+    with io.open(path, "r", encoding="utf-8") as f:
+        return f.read().splitlines(True)
+
+
+def write_text(path: str, content: str) -> None:
+    with io.open(path, "w", encoding="utf-8", newline="") as f:
+        f.write(content)
+
+
+def normalize_list_rst(lines: List[str]) -> str:
+    """Return normalized content of checks list.rst from given lines.
+
+    Input: full file content split into lines.
+    Output: single string with all csv-table rows sorted by :doc: label while
+            preserving non-table content and table options/headers.
+    """
+    out: List[str] = []
+    i = 0
+    n = len(lines)
+
+    def key_for(line: str):
+        m = DOC_LINE_RE.match(line)
+        if not m:
+            return (1, "")
+        return (0, m.group("label"))
+
+    while i < n:
+        line = lines[i]
+        if line.lstrip().startswith(".. csv-table::"):
+            out.append(line)
+            i += 1
+
+            while i < n and (lines[i].startswith(" ") or lines[i].strip() == ""):
+                if DOC_LINE_RE.match(lines[i]):
+                    break
+                out.append(lines[i])
+                i += 1
+
+            entries: List[str] = []
+            while i < n and lines[i].startswith(" "):
+                entries.append(lines[i])
+                i += 1
+
+            entries_sorted = sorted(entries, key=key_for)
+            out.extend(entries_sorted)
+            continue
+
+        out.append(line)
+        i += 1
+
+    return "".join(out)
+
+
+def find_heading(lines: Sequence[str], title: str) -> Optional[int]:
+    """Find heading start index for a section underlined with ^ characters.
+
+    The function looks for a line equal to `title` followed by a line that
+    consists solely of ^, which matches the ReleaseNotes style for subsection
+    headings used here.
+
+    Returns index of the title line, or None if not found.
+    """
+    for i in range(len(lines) - 1):
+        if lines[i].rstrip("\n") == title:
+            underline = lines[i + 1].rstrip("\n")
+            if underline and set(underline) == {"^"} and len(underline) >= len(title):
+                return i
+    return None
+
+
+def extract_label(text: str) -> str:
+    m = DOC_LABEL_RN_RE.search(text)
+    return m.group("label") if m else text
+
+
+def is_bullet_start(line: str) -> bool:
+    return line.startswith("- ")
+
+
+def collect_bullet_blocks(
+    lines: Sequence[str], start: int, end: int
+) -> Tuple[List[str], List[Tuple[str, List[str]]], List[str]]:
+    i = start
+    n = end
+    first_bullet = i
+    while first_bullet < n and not is_bullet_start(lines[first_bullet]):
+        first_bullet += 1
+    prefix = list(lines[i:first_bullet])
+
+    blocks: List[Tuple[str, List[str]]] = []
+    i = first_bullet
+    while i < n:
+        if not is_bullet_start(lines[i]):
+            break
+        bstart = i
+        i += 1
+        while i < n and not is_bullet_start(lines[i]):
+            if (
+                i + 1 < n
+                and set(lines[i + 1].rstrip("\n")) == {"^"}
+                and lines[i].strip()
+            ):
+                break
+            i += 1
+        block = list(lines[bstart:i])
+        key = extract_label(block[0])
+        blocks.append((key, block))
+
+    suffix = list(lines[i:n])
+    return prefix, blocks, suffix
+
+
+def sort_blocks(blocks: List[Tuple[str, List[str]]]) -> List[List[str]]:
+    """Return blocks sorted deterministically by their extracted label.
+
+    Duplicates are preserved; merging is left to authors to handle manually.
+    """
+    return [b for _, b in sorted(blocks, key=lambda kb: kb[0])]
+
+
+def _find_section_bounds(
+    lines: Sequence[str], title: str, next_title: Optional[str]
+) -> Optional[Tuple[int, int, int]]:
+    """Return (h_start, sec_start, sec_end) for section `title`.
+
+    - h_start: index of the section title line
+    - sec_start: index of the first content line after underline
+    - sec_end: index of the first line of the next section title (or end)
+    """
+    h_start = find_heading(lines, title)
+    if h_start is None:
+        return None
+
+    sec_start = h_start + 2
+
+    # Determine end of section either from next_title or by scanning.
+    if next_title is not None:
+        h_end = find_heading(lines, next_title)
+        if h_end is None:
+            # Scan forward to the next heading-like underline.
+            h_end = sec_start
+            while h_end + 1 < len(lines):
+                if lines[h_end].strip() and set(lines[h_end + 1].rstrip("\n")) == {"^"}:
+                    break
+                h_end += 1
+        sec_end = h_end
+    else:
+        # Scan to end or until a heading underline is found.
+        h_end = sec_start
+        while h_end + 1 < len(lines):
+            if lines[h_end].strip() and set(lines[h_end + 1].rstrip("\n")) == {"^"}:
+                break
+            h_end += 1
+        sec_end = h_end
+
+    return h_start, sec_start, sec_end
+
+
+def _normalize_release_notes_section(
+    lines: List[str], title: str, next_title: Optional[str]
+) -> List[str]:
+    """Normalize a single release-notes section and return updated lines."""
+    bounds = _find_section_bounds(lines, title, next_title)
+    if bounds is None:
+        return lines
+    _, sec_start, sec_end = bounds
+
+    prefix, blocks, suffix = collect_bullet_blocks(lines, sec_start, sec_end)
+    sorted_blocks = sort_blocks(blocks)
+
+    new_section: List[str] = []
+    new_section.extend(prefix)
+    for i_b, b in enumerate(sorted_blocks):
+        if i_b > 0 and (
+            not new_section or (new_section and new_section[-1].strip() != "")
+        ):
+            new_section.append("\n")
+        new_section.extend(b)
+    new_section.extend(suffix)
+
+    return lines[:sec_start] + new_section + lines[sec_end:]
+
+
+def normalize_release_notes(lines: List[str]) -> str:
+    sections = ["New checks", "New check aliases", "Changes in existing checks"]
+
+    out = list(lines)
+
+    for idx in range(len(sections) - 1, -1, -1):
+        title = sections[idx]
+        next_title = sections[idx + 1] if idx + 1 < len(sections) else None
+        out = _normalize_release_notes_section(out, title, next_title)
+
+    return "".join(out)
+
+
+def _default_paths() -> Tuple[str, str]:
+    base = os.path.normpath(os.path.join(script_dir(), "..", ".."))
+    list_doc = os.path.join(base, "docs", "clang-tidy", "checks", "list.rst")
+    rn_doc = os.path.join(base, "docs", "ReleaseNotes.rst")
+    return list_doc, rn_doc
+
+
+def main(argv: List[str]) -> int:
+    ap = argparse.ArgumentParser()
+    ap.add_argument("-o", "--output", dest="out", default=None)
+    args = ap.parse_args(argv)
+
+    list_doc, rn_doc = _default_paths()
+
+    if args.out:
+        out_path = args.out
+        out_lower = os.path.basename(out_path).lower()
+        if "release" in out_lower:
+            lines = read_text(rn_doc)
+            normalized = normalize_release_notes(lines)
+            write_text(out_path, normalized)
+            return 0
+        else:
+            lines = read_text(list_doc)
+            normalized = normalize_list_rst(lines)
+            write_text(out_path, normalized)
+            return 0
+
+    list_lines = read_text(list_doc)
+    rn_lines = read_text(rn_doc)
+    list_norm = normalize_list_rst(list_lines)
+    rn_norm = normalize_release_notes(rn_lines)
+    if "".join(list_lines) != list_norm:
+        write_text(list_doc, list_norm)
+    if "".join(rn_lines) != rn_norm:
+        write_text(rn_doc, rn_norm)
+    return 0
+
+
+if __name__ == "__main__":
+    main(sys.argv[1:])

clang-tools-extra/test/clang-tidy/infrastructure/alphabetical-order.cpp

@@ -0,0 +1,5 @@
+// RUN: %python %S/../../../clang-tidy/tool/clang-tidy-alphabetical-order-check.py -o %t.clang-tidy-checks-list.rst
+// RUN: diff --strip-trailing-cr %t.clang-tidy-checks-list.rst %S/../../../docs/clang-tidy/checks/list.rst
+
+// RUN: %python %S/../../../clang-tidy/tool/clang-tidy-alphabetical-order-check.py -o %t.ReleaseNotes.rst
+// RUN: diff --strip-trailing-cr %t.ReleaseNotes.rst %S/../../../docs/ReleaseNotes.rst

clang-tools-extra/test/lit.cfg.py

@@ -57,6 +57,7 @@
 if config.clang_tidy_custom_check:
     config.available_features.add("custom-check")
 python_exec = shlex.quote(config.python_executable)
+config.substitutions.append(("%python", python_exec))
 check_clang_tidy = os.path.join(
     config.test_source_root, "clang-tidy", "check_clang_tidy.py"
 )