header update

scripts/header_lib.py

@@ -16,6 +16,7 @@ BANNER_START = "/***************************************************************
 BANNER_END = "**********************************************************************/"
 MARKER_CREATED = "Created at:"
 MARKER_UPDATED = "Updated at:"
+MARKER_COMMIT = "Commit:"
 MARKER_DIGEST = "Digest:"
 MARKER_CONTRIBUTORS = "Contributors:"
 MARKER_CLASS = re.compile(r"^\s*\*?\s*(class|interface|enum)\s+(\w+)", re.I)
@@ -62,6 +63,7 @@ class FileMeta:
     last_contributor: str
     contributors_block: str
     language: str
+    commit_sha: str = ""
 
 
 def iter_source_files(root: Path) -> Iterable[Path]:
@@ -151,6 +153,45 @@ def format_identity(name: str, email: str = "") -> str:
     return name
 
 
+def hook_skip_requested() -> bool:
+    import os
+
+    return os.environ.get("HEADER_HOOK_SKIP", "").strip() in {"1", "true", "yes"}
+
+
+def git_head_sha() -> str:
+    return run_git(["rev-parse", "HEAD"], REPO_ROOT)
+
+
+def git_last_commit_sha_for_file(path: Path) -> str:
+    rel = repo_relative(path)
+    return run_git(["log", "-1", "--format=%H", "--", rel], REPO_ROOT)
+
+
+def parse_commit_from_header(text: str) -> str:
+    m = re.search(r"\* Commit:\s*(\S+)", text)
+    return m.group(1).strip() if m else ""
+
+
+def files_in_git_commit(commit: str = "HEAD") -> list[Path]:
+    out = run_git(
+        ["diff-tree", "--no-commit-id", "--name-only", "-r", commit],
+        REPO_ROOT,
+    )
+    paths: list[Path] = []
+    for line in out.splitlines():
+        rel = line.strip()
+        if not rel:
+            continue
+        p = REPO_ROOT / rel
+        if not p.is_file() or p.suffix.lower() not in SOURCE_SUFFIXES:
+            continue
+        if is_excluded_path(p):
+            continue
+        paths.append(p)
+    return paths
+
+
 def format_git_date(raw: str) -> str:
     if not raw:
         return datetime.now().astimezone().strftime("%a %d %b %Y %H:%M:%S %z")
@@ -332,18 +373,45 @@ def package_line_for_file(source_text: str, rel_path: str, suffix: str) -> str:
     return f"/* package {rel_path} */"
 
 
+def strip_existing_header_php(text: str) -> str:
+    """Remove <?php, block-comment header(s), and closing ?> before real body."""
+    work = text.lstrip("\ufeff")
+    while True:
+        changed = False
+        w = work.lstrip()
+        if w.startswith("<?php"):
+            work = w[5:].lstrip("\n")
+            changed = True
+            w = work.lstrip()
+        if w.startswith("?>"):
+            work = w[2:].lstrip("\n")
+            changed = True
+            w = work.lstrip()
+        if w.startswith("/*") or w.startswith("/*****"):
+            close = w.find("*/")
+            if close >= 0:
+                work = w[close + 2 :].lstrip("\n")
+                changed = True
+            else:
+                break
+        if not changed:
+            break
+    return work.lstrip("\n")
+
+
+def php_body_is_markup(body: str) -> bool:
+    """True when body is HTML/templates after the commented header (needs ?>)."""
+    s = body.lstrip()
+    return bool(s) and s[0] == "<"
+
+
 def strip_existing_header(text: str, suffix: str) -> tuple[str, str]:
     """Returns (body without header/package, preserved leading e.g. <?php)."""
+    if suffix == ".php":
+        return strip_existing_header_php(text), ""
+
     leading = ""
     work = text
-    if suffix == ".php" and work.startswith("<?php"):
-        first_nl = work.find("\n")
-        if first_nl >= 0:
-            leading = work[: first_nl + 1]
-            work = work[first_nl + 1 :]
-        else:
-            leading = work + "\n"
-            work = ""
 
     lines = work.splitlines(keepends=True)
     i = 0
@@ -392,6 +460,7 @@ def build_meta(path: Path, body: str, source_text: str = "") -> FileMeta:
     lang = "java" if suffix in {".java", ".aidl"} else ("php" if suffix == ".php" else "c")
 
     class_line = detect_class_line(body, suffix)
+    commit_sha = parse_commit_from_header(source_text) or git_last_commit_sha_for_file(path)
     return FileMeta(
         rel_path=rel,
         filename=path.name,
@@ -402,34 +471,83 @@ def build_meta(path: Path, body: str, source_text: str = "") -> FileMeta:
         last_contributor=author,
         contributors_block=contributors_block(rows),
         language=lang,
+        commit_sha=commit_sha,
     )
 
 
-def render_header(meta: FileMeta, digest_hex: str | None) -> str:
+def _inner_header_lines(meta: FileMeta, *, for_digest: bool) -> list[str]:
+    """Banner lines shared by PHP and C-style headers."""
     lines = [
-        meta.package_line,
-        "",
-        BANNER_START,
         f" * {meta.filename}",
         f" * {MARKER_CREATED} {meta.created_at}",
         f" * {MARKER_UPDATED} {meta.updated_at} by {meta.last_contributor}",
     ]
+    if not for_digest and meta.commit_sha:
+        lines.append(f" * {MARKER_COMMIT} {meta.commit_sha}")
     if meta.class_line:
         lines.append(meta.class_line)
     lines.append(f" * {MARKER_CONTRIBUTORS}")
-    lines.append(meta.contributors_block)
-    if digest_hex:
+    lines.extend(meta.contributors_block.splitlines())
+    return lines
+
+
+def render_php_header(meta: FileMeta, digest_hex: str | None, *, for_digest: bool = False) -> str:
+    """
+    PHP: header is one block comment inside <?php ... ?> so the engine ignores it.
+    No nested '/***** ... */' banners (inner */ would break the comment).
+    """
+    inner = [f" * package {meta.rel_path}"]
+    inner.extend(_inner_header_lines(meta, for_digest=for_digest))
+    if digest_hex and not for_digest:
+        inner.append(f" * {MARKER_DIGEST} SHA256\t{digest_hex}")
+    return "<?php\n/*\n" + "\n".join(inner) + "\n */\n"
+
+
+def render_header(
+        meta: FileMeta, digest_hex: str | None, suffix: str = "", *, for_digest: bool = False) -> str:
+    if suffix == ".php" or meta.language == "php":
+        return render_php_header(meta, digest_hex, for_digest=for_digest)
+    lines = [
+        meta.package_line,
+        "",
+        BANNER_START,
+        *_inner_header_lines(meta, for_digest=for_digest),
+    ]
+    if digest_hex and not for_digest:
         lines.append(f" * {MARKER_DIGEST} SHA256\t{digest_hex}")
     lines.append(BANNER_END)
     lines.append("")
     return "\n".join(lines)
 
 
-def compute_digest(prefix: str, body: str) -> str:
-    payload = prefix + body
+def wrap_php_header_and_body(header: str, body: str) -> str:
+    """
+    Pure PHP: <?php /* header */ code... (declare stays first statement after <?php).
+    Templates: <?php /* header */ ?> then HTML.
+    """
+    body = body.lstrip("\n")
+    if body.startswith("<?php"):
+        body = body[5:].lstrip("\n")
+    if php_body_is_markup(body):
+        return header + "?>\n" + body
+    return header + body
+
+
+def compute_digest(header_prefix: str, body: str) -> str:
+    """Hash file body + header without Digest/Commit lines."""
+    payload = header_prefix + body
     return hashlib.sha256(payload.encode("utf-8")).hexdigest()
 
 
+def finalize_header(meta: FileMeta, body: str, suffix: str, leading: str = "") -> str:
+    prefix_for_digest = render_header(meta, None, suffix, for_digest=True)
+    digest = compute_digest(prefix_for_digest, body)
+    prefix = render_header(meta, digest, suffix, for_digest=False)
+    if suffix == ".php":
+        return wrap_php_header_and_body(prefix, body)
+    return leading + prefix + body
+
+
 def apply_header_to_content(path: Path, text: str, update_only: bool = False) -> str:
     suffix = path.suffix.lower()
     body, leading = strip_existing_header(text, suffix)
@@ -440,14 +558,15 @@ def apply_header_to_content(path: Path, text: str, update_only: bool = False) ->
         if m:
             meta.created_at = m.group(1).strip()
 
-    prefix_without_digest = render_header(meta, None)
-    digest = compute_digest(prefix_without_digest, body)
-    prefix = render_header(meta, digest)
-    return leading + prefix + body
+    return finalize_header(meta, body, suffix, leading)
 
 
 def has_valid_header(text: str) -> bool:
-    return MARKER_CREATED in text and BANNER_START in text and BANNER_END in text
+    if MARKER_CREATED not in text:
+        return False
+    if text.lstrip().startswith("<?php") and "/*" in text[:800]:
+        return "*/" in text[:1200]
+    return BANNER_START in text and BANNER_END in text
 
 
 def header_needs_identity_refresh(text: str, path: Path) -> bool:
@@ -475,7 +594,8 @@ def header_needs_identity_refresh(text: str, path: Path) -> bool:
     return False
 
 
-def update_header_timestamps(text: str, path: Path, contributor: str | None = None) -> str:
+def update_header_timestamps(
+        text: str, path: Path, contributor: str | None = None, *, touch_commit: bool = False) -> str:
     suffix = path.suffix.lower()
     body, leading = strip_existing_header(text, suffix)
     if contributor:
@@ -492,8 +612,27 @@ def update_header_timestamps(text: str, path: Path, contributor: str | None = No
     meta.created_at = created
     meta.updated_at = updated_line
     meta.last_contributor = who
+    if touch_commit:
+        meta.commit_sha = git_head_sha()
+    else:
+        meta.commit_sha = parse_commit_from_header(text)
 
-    prefix_without_digest = render_header(meta, None)
-    digest = compute_digest(prefix_without_digest, body)
-    prefix = render_header(meta, digest)
-    return leading + prefix + body
+    return finalize_header(meta, body, suffix, leading)
+
+
+def update_header_commit(text: str, path: Path, commit_sha: str) -> str:
+    """Set Commit: line after post-commit; recomputes Digest (Commit excluded from hash)."""
+    if parse_commit_from_header(text) == commit_sha:
+        return text
+    suffix = path.suffix.lower()
+    body, leading = strip_existing_header(text, suffix)
+    meta = build_meta(path, body, text)
+    m_created = re.search(r"\* Created at: (.+)", text)
+    m_updated = re.search(r"\* Updated at: (.+?) by (.+)$", text, re.MULTILINE)
+    if m_created:
+        meta.created_at = m_created.group(1).strip()
+    if m_updated:
+        meta.updated_at = m_updated.group(1).strip()
+        meta.last_contributor = m_updated.group(2).strip()
+    meta.commit_sha = commit_sha
+    return finalize_header(meta, body, suffix, leading)