diff options
Diffstat (limited to 'wflib/check.py')
| -rw-r--r-- | wflib/check.py | 378 |
1 files changed, 378 insertions, 0 deletions
diff --git a/wflib/check.py b/wflib/check.py new file mode 100644 index 0000000..f86b2fe --- /dev/null +++ b/wflib/check.py @@ -0,0 +1,378 @@ +"""Validation: everything `wf check` reports.""" +from __future__ import annotations + +import re +import subprocess +import time +from dataclasses import dataclass +from pathlib import Path + +from . import areas, refs, tasks +from .config import NAME, Config + +NUMBERED_RE = re.compile(r"^\d+\. ") +NUMBER_REF_RE = re.compile(r"(?<![\w/&#])#\d+\b") +MD_LINK_RE = re.compile(r"\[[^\]]*\]\(([^)\s]+)\)") +TASK_LINK_RE = re.compile(r"^[ta]-") +EFFORTS = ", ".join(tasks.EFFORTS) +KINDS = "a- items belong in Awaiting, t- elsewhere" +STALE_DAYS = 30 + + +@dataclass(frozen=True) +class Problem: + path: str + line: int | None + id: str + message: str + + @property + def key(self) -> tuple[str, str, str]: + return (self.path, self.id, self.message) + + def __str__(self) -> str: + where = self.path + (f":{self.line}" if self.line else "") + return f"{where}: " + (f"{self.id}: " if self.id else "") + self.message + + +def _read(path: Path) -> str: + return path.read_text(encoding="utf-8", errors="replace").replace("\r\n", "\n") + + +def _prose_lines(text: str): + """(1-based line number, line) outside code fences.""" + fenced = False + for n, line in enumerate(text.split("\n"), 1): + if refs.FENCE_RE.match(line): + fenced = not fenced + elif not fenced: + yield n, line + + +def _link_problems(path: str, text: str, known: set[str]) -> list[Problem]: + return [Problem(path, n, f"[[{id}]]", "no such id in TASKS or archive") + for n, line in _prose_lines(text) + for id in dict.fromkeys(refs.links(line)) + if TASK_LINK_RE.match(id) and id not in known] + + +def _cycle(start: str, edges: dict[str, list[str]]) -> list[str] | None: + def walk(node: str, path: list[str]) -> list[str] | None: + for nxt in edges.get(node, []): + if nxt == start: + return path + [nxt] + if nxt not in path and (found := walk(nxt, path + [nxt])): + return found + return None + return walk(start, [start]) + + +def _index_sections(cfg: Config) -> list[refs.DocSection] | None: + """Headings under the anchors index section; None when no [anchors].""" + if not cfg.anchors_index or not cfg.anchors_index.is_file(): + return None + found = refs.sections(_read(cfg.anchors_index)) + if not cfg.anchors_section: + return [s for s in found if s.level] + top = next((s for s in found if s.level and s.heading == cfg.anchors_section), None) + if top is None: + return [] + return [s for s in found if s.level > top.level and top.start < s.start < top.end] + + +def check_tasks(text: str, archive_text: str, cfg: Config) -> tuple[list[Problem], list[Problem]]: + path = cfg.rel(cfg.tasks) + errors: list[Problem] = [] + warnings: list[Problem] = [] + doc = tasks.parse(text) + archived = tasks.archive_ids(archive_text) + + def err(line, id, message): + errors.append(Problem(path, line, id, message)) + + for key, heading in tasks.SECTIONS.items(): + if not any(s.key == key for s in doc.sections): + err(None, "", f"no '## {heading}' section") + + errors += _link_problems(path, text, doc.ids() | archived) + + index = _index_sections(cfg) + index_anchors = {a for s in index for a in s.anchors} if index is not None else None + open_awaiting = {i.id for s in doc.sections if s.key == "awaiting" for i in s.items} + deferred = {i.id for s in doc.sections if s.key == "deferred" for i in s.items} + edges = {i.id: [a for a in i.after if a in doc.ids()] for i in doc.all_items()} + seen: dict[str, int] = {} + in_cycle: set[str] = set() + + for section in doc.sections: + if section.key is None: + continue + for n, line in enumerate(section.prefix, section.line + 1): + if NUMBERED_RE.match(line): + err(n, "", "old numbered item (wf migrate)") + for n, line in enumerate(section.suffix, section.suffix_line or 0): + if NUMBERED_RE.match(line): + err(n, "", "old numbered item (wf migrate)") + elif line.startswith(tasks.ITEM_START): + err(n, tasks.parse_header(line).id, + f"item after a flush-left prose line (line {section.suffix_line}): indent or move the prose") + for at, item in enumerate(section.items): + if item.error: + err(item.line, item.id, item.error) + continue + if not tasks.ID_RE.match(item.id): + err(item.line, item.id, "bad id (want t-… or a-…, lowercase a-z 0-9 -)") + continue + if item.id in seen: + err(item.line, item.id, f"duplicate id (also line {seen[item.id]})") + seen.setdefault(item.id, item.line) + if item.id in archived: + err(item.line, item.id, "id already used in the archive (ids are never reused)") + if item.id.startswith("a-") != (section.key == "awaiting"): + err(item.line, item.id, f"{item.id[:2]} item in {section.heading} ({KINDS})") + continue + if section.key == "awaiting": + continue + if item.prio is None: + err(item.line, item.id, "no priority [P0]-[P3]") + elif item.prio > 3: + err(item.line, item.id, f"priority P{item.prio} (want P0-P3)") + if item.effort is None: + err(item.line, item.id, f"no effort (want {EFFORTS})") + elif item.effort not in tasks.EFFORTS: + err(item.line, item.id, f"effort '{item.effort}' (want {EFFORTS})") + if item.status and item.status.startswith("blocked:") and item.blocked_on not in open_awaiting: + err(item.line, item.id, f"blocked on '{item.blocked_on}', which is not an open Awaiting item") + if item.status and re.fullmatch(r"in progress:\s*", item.status): + warnings.append(Problem(path, item.line, item.id, "in progress without a branch or note")) + if item.id not in in_cycle and (cycle := _cycle(item.id, edges)): + in_cycle.update(cycle) + err(item.line, item.id, "After: cycle " + " → ".join(cycle)) + later = {o.id for o in section.items[at + 1:]} + for dep in item.after: + if dep in later: + err(item.line, item.id, f"placed before '{dep}', which it is After:") + for label, pattern in (("After", tasks.AFTER_RE), ("Slices", tasks.SLICES_RE)): + if (i := item._line(pattern)) is None: + continue + rest = tasks.LINK_RE.sub(" ", pattern.match(item.body[i]).group(1)) + for word in re.split(r"[\s,;]+", rest): + if tasks.ID_RE.match(word): + err(item.line + 1 + i, item.id, f"{label}: '{word}' is not a link (want [[{word}]])") + if (i := item._line(tasks.AFTER_RE)) is not None: + rest = tasks.LINK_RE.sub(" ", tasks.AFTER_RE.match(item.body[i]).group(1)) + if re.search(r"[A-Za-z0-9]", rest): + warnings.append(Problem(path, item.line + 1 + i, item.id, + "After: line has prose; every [[id]] in it is a dependency")) + for dep in item.after: + if dep in deferred and section.key != "deferred": + warnings.append(Problem(path, item.line + 1 + i, item.id, + f"After: '{dep}' is deferred (never runs; blocks this task)")) + models = [(n, m.group(1).split()) for n, l in enumerate(item.body) if (m := tasks.MODEL_RE.match(l))] + for n, words in models[:1]: + if not words or words[0] not in tasks.MODELS: + err(item.line + 1 + n, item.id, + f"Model '{' '.join(words)}' (want {', '.join(tasks.MODELS)})") + elif len(words) > 1: + warnings.append(Problem(path, item.line + 1 + n, item.id, f"Model line '{' '.join(words)}': " + f"write 'Model: {words[0]}' (wf set --model)")) + if len(models) > 1: + err(item.line + 1 + models[1][0], item.id, "two Model lines") + sess = [(n, m.group(1).split()) for n, l in enumerate(item.body) if (m := tasks.SESSIONS_RE.match(l))] + for n, words in sess[:1]: + if not words or words[0] not in tasks.SESSIONS: + err(item.line + 1 + n, item.id, + f"Sessions '{' '.join(words)}' (want {', '.join(tasks.SESSIONS)})") + if len(sess) > 1: + err(item.line + 1 + sess[1][0], item.id, "two Sessions lines") + clouds = [(n, m.group(1).split()) for n, l in enumerate(item.body) if (m := tasks.CLOUD_RE.match(l))] + for n, words in clouds[:1]: + if not words or words[0] not in tasks.CLOUDS: + err(item.line + 1 + n, item.id, f"Cloud '{' '.join(words)}' (want {', '.join(tasks.CLOUDS)})") + if len(clouds) > 1: + err(item.line + 1 + clouds[1][0], item.id, "two Cloud lines") + if item.interactive: + warnings.append(Problem(path, item.line, item.id, "'interactive' flag: write 'Sessions: owner' " + f"(wf set {item.id} --sessions owner)")) + if section.key == "pending" and item.human_done_match and item.sessions != "owner": + warnings.append(Problem(path, item.line, item.id, f"not runner-ready: Done reads as human action " + f"('{item.human_done_match}'); reword or set Sessions: owner")) + ref_at = item._line(tasks.REF_RE) + ref_line = item.line + 1 + ref_at if ref_at is not None else item.line + for target, anchor in item.refs: + shown = target + (f"#{anchor}" if anchor else "") + file = cfg.root / target + if not file.exists(): + err(ref_line, item.id, f"Ref '{shown}' does not exist") + elif anchor and file.is_file(): + if anchor not in refs.anchors(_read(file)): + err(ref_line, item.id, f"Ref '{shown}': no such anchor") + elif index_anchors is not None and file.resolve() == cfg.anchors_index.resolve() \ + and anchor not in index_anchors: + err(ref_line, item.id, f"Ref '{shown}' is not a heading under '{cfg.anchors_section}'") + + for n, line in _prose_lines(text): + for m in NUMBER_REF_RE.finditer(refs.strip_code(line)): + warnings.append(Problem(path, n, "", f"'{m.group(0)}': number ref (tasks have ids: [[t-…]])")) + return errors, warnings + + +def _anchor_problems(cfg: Config) -> list[Problem]: + index = _index_sections(cfg) + if index is None: + return [] + out: list[Problem] = [] + path = cfg.rel(cfg.anchors_index) + lines = _read(cfg.anchors_index).split("\n") + where = f"'{cfg.anchors_section}'" if cfg.anchors_section else "the index" + seen: dict[str, int] = {} + for s in index: + slug = refs.slugify(s.heading) + if slug in seen: + out.append(Problem(path, s.start + 1, "", f"two {where} headings give anchor '{slug}' (also line {seen[slug]})")) + seen.setdefault(slug, s.start + 1) + for n in range(s.start, s.end): + for target in MD_LINK_RE.findall(refs.strip_code(lines[n])): + if re.match(r"^[a-z][a-z0-9+.-]*:", target): + continue + file_part, _, frag = target.partition("#") + file = (cfg.anchors_index.parent / file_part) if file_part else cfg.anchors_index + if not file.exists(): + out.append(Problem(path, n + 1, "", f"link '{target}': file does not exist")) + elif frag and file.suffix == ".md" and frag not in refs.anchors(_read(file)): + out.append(Problem(path, n + 1, "", f"link '{target}': no such anchor")) + if cfg.anchors_specs and cfg.anchors_specs.is_dir(): + known = {a for s in index for a in s.anchors} + for spec in sorted(cfg.anchors_specs.rglob("*.md")): + for n, line in _prose_lines(_read(spec)): + for anchor in refs.ANCHOR_RE.findall(line): + if anchor not in known: + out.append(Problem(cfg.rel(spec), n, "", + f"explicit anchor '{anchor}' has no heading under {where} in {path}")) + return out + + +def doc_files(cfg: Config) -> list[Path]: + out: list[Path] = [] + for d in cfg.docs: + if d.is_file(): + out.append(d) + elif d.is_dir(): + out += sorted(p for p in d.rglob("*.md") if p.is_file()) + skip = {cfg.tasks.resolve(), cfg.archive.resolve()} + return [p for p in dict.fromkeys(out) if p.resolve() not in skip] + + +def _age_days(root: Path, path: Path, line: int) -> float | None: + try: + out = subprocess.run(["git", "-C", str(root), "blame", "-L", f"{line},{line}", "--porcelain", "--", str(path)], + capture_output=True, text=True, timeout=10) + except (OSError, subprocess.SubprocessError): + return None + m = re.search(r"^author-time (\d+)$", out.stdout, re.M) if out.returncode == 0 else None + if not m or re.match(r"^0{40}", out.stdout): + return None + return (time.time() - int(m.group(1))) / 86400 + + +def _stale_awaiting(cfg: Config, text: str) -> list[Problem]: + doc = tasks.parse(text) + waited = {i.blocked_on for i in doc.all_items()} | {l for i in doc.all_items() if i.id.startswith("t-") + for l in refs.links("\n".join(i.lines()))} + out = [] + for s in doc.sections: + if s.key != "awaiting": + continue + for item in s.items: + if item.id in waited or item.error: + continue + age = _age_days(cfg.root, cfg.tasks, item.line) + if age is not None and age > STALE_DAYS: + out.append(Problem(cfg.rel(cfg.tasks), item.line, item.id, + f"waiting {int(age)} days, no task references it")) + return out + + +def _git(root: Path, *args: str) -> str | None: + try: + out = subprocess.run(["git", "-C", str(root), *args], capture_output=True, text=True, timeout=10) + except (OSError, subprocess.SubprocessError): + return None + return out.stdout.strip() if out.returncode == 0 else None + + +def merged_tool_worktrees(tool_root: Path) -> list[Problem]: + """Worktrees of the wf tool repo whose branch is behind master (merged, leftover after release). + A branch equal to master, or a worktree created < 2h ago, is not flagged.""" + listing = _git(tool_root, "worktree", "list", "--porcelain") + master = _git(tool_root, "rev-parse", "master") + if not listing or not master: + return [] + out = [] + for block in listing.split("\n\n"): + path = branch = head = None + for l in block.splitlines(): + if l.startswith("worktree "): + path = l[9:] + elif l.startswith("HEAD "): + head = l[5:] + elif l.startswith("branch refs/heads/"): + branch = l[18:] + if not path or not branch or branch == "master" or head == master: + continue + try: # created < 2h ago: a running worker's fresh worktree, not a leftover + if time.time() - (Path(path) / ".git").stat().st_mtime < 7200: + continue + except OSError: + pass + if _git(tool_root, "merge-base", "--is-ancestor", branch, "master") is not None: + out.append(Problem("wf-tool", None, "", f"tool worktree '{path}' (branch {branch}) is merged into " + f"master: git worktree remove it, git branch -d {branch}")) + return out + + +def _area_problems(cfg: Config) -> list[Problem]: + file = cfg.areas_file + if not file.is_file(): + return [] + out = [] + for a in areas.parse(_read(file)): + for tok in areas.missing(cfg.code, a): + out.append(Problem(cfg.rel(file), a.line, "", f"area {a.name}: anchor {tok} not found")) + if a.checked and areas.commits_since(cfg.code, a) is None: + out.append(Problem(cfg.rel(file), a.line, "", f"area {a.name}: Checked {a.checked} unknown")) + return out + + +def check(cfg: Config, tasks_text: str | None = None, archive_text: str | None = None, + slow: bool = True) -> tuple[list[Problem], list[Problem]]: + """All errors and warnings of the project. `tasks_text` / `archive_text` + replace what is on disk (to judge a change before it is written); + `slow=False` leaves out the git-based warnings.""" + errors: list[Problem] = [] + for name, path in (("tasks", cfg.tasks), ("archive", cfg.archive)): + given = tasks_text if name == "tasks" else archive_text + if given is None and not path.is_file(): + errors.append(Problem(NAME, None, "", f"{name} '{cfg.rel(path)}' does not exist")) + for name, path in [("docs", d) for d in cfg.docs] + [("anchors.index", cfg.anchors_index), + ("anchors.specs", cfg.anchors_specs)]: + if path is not None and not path.exists(): + errors.append(Problem(NAME, None, "", f"{name} '{cfg.rel(path)}' does not exist")) + if tasks_text is None and not cfg.tasks.is_file(): + return errors, [] + text = (_read(cfg.tasks) if tasks_text is None else tasks_text).replace("\r\n", "\n") + archive = archive_text if archive_text is not None else (_read(cfg.archive) if cfg.archive.is_file() else "") + found, warnings = check_tasks(text, archive, cfg) + errors += found + known = tasks.parse(text).ids() | tasks.archive_ids(archive) + for file in doc_files(cfg): + errors += _link_problems(cfg.rel(file), _read(file), known) + errors += _anchor_problems(cfg) + if slow: + warnings += _stale_awaiting(cfg, text) + warnings += _area_problems(cfg) + order: dict[str, int] = {} + for p in errors + warnings: + order.setdefault(p.path, len(order)) + by_place = lambda p: (order[p.path], p.line or 0) + return sorted(errors, key=by_place), sorted(warnings, key=by_place) |