Files
cs249r_book/interviews/vault/scripts/render_visuals.py
Vijay Janapa Reddi 542aaf95d2 cleanup(vault): release-ready Phase A — schema hardening + lint calibration + chain repair
Closes the cleanup arc (A.1–A.10 in RESUME_PLAN_RELEASE.md). Every
gate is now green: vault check --strict, vault lint, vault doctor,
vault codegen --check, staffml validate-vault, Playwright (9/9), tsc.

A.1 mobile-1962.svg: renamed `Edge` → `RegEdge` in graphviz source
    (`Edge` is a reserved keyword); SVG renders cleanly. Also fixed
    tinyml-1570.py (missing `import numpy as np`) which the new failure
    log surfaced.

A.2 render_visuals.py: structured per-ID failure log written to
    `_validation_results/render_failures.json` on every run; non-zero
    exit on any per-item crash; new `--fail-fast` and `--failure-log`
    CLI options. Replaces the prior silent-failure mode.

A.3 LinkML visual schema: typed as a structured sub-schema. New
    `VisualKind` enum (svg only — `mermaid` was reserved but never
    shipped, dropped to keep the enum honest). Path regex tightened
    to `^[a-z0-9-]+\.svg$`. Alt minimum length 10, caption required
    minimum length 5. TypeScript Visual interface + Question.visual
    field added to staffml-vault-types/index.ts.

A.4 Pydantic Visual + Question validators:
    - Visual.kind hard-rejects anything but `svg`
    - Visual.path enforces the new regex
    - Visual.alt min 10 chars, caption required min 5 chars
    - Question.model_validator: visual.path MUST resolve to a real
      file under interviews/vault/visuals/<track>/. Skipped in
      production deploys where the working tree is absent.

A.5 Registry repair + doctor split:
    - tools: repair_registry.py appended 5,269 missing IDs
      (the rename refactor at 8a5c3ff3c left the append-only registry
      unsynced; this brings disk-coverage to 100%). Header block in
      id-registry.yaml documents the rebuild rationale.
    - doctor.py: split symmetric `registry-integrity` check into
      `disk-coverage` (HARD FAIL if any disk YAML id is unregistered)
      and `registry-history` (INFO ONLY for retired ids — the registry
      is by design an audit log, retired ids are normal). Pre-existing
      `_check_schema_version` bug (`versions == {1}` vs string `"1.0"`)
      fixed.

A.6 Lint calibration via 4-expert consensus + bloom-canonical
    reclassification:
    - Spawned 4 experts (Vijay Reddi, Chip Huyen, Jeff Dean,
      education-reviewer) on 42 disputed (zone, level) pairs;
      consensus-builder aggregated to 15 valid / 19 invalid / 8
      borderline.
    - User arbitrated 8 borderlines: 7 widen / 1 reclassify.
    - Built ZONE_BLOOM_AFFINITY matrix (Education-Reviewer's idea):
      every zone admits its dominant Bloom verb + adjacent verbs,
      rejects clear hierarchy violations.
    - reclassify_zone_bloom_mismatch.py applied 576 deterministic
      zone fixes via BLOOM_CANONICAL_ZONE mapping (e.g. fluency+analyze
      → analyze, recall+analyze → analyze, evaluation+apply → implement).
    - Question.model_validator(_zone_bloom_compatible): hard-rejects
      future zone-bloom mismatches at write time. Generated drafts
      can no longer ship a self-contradicting classification.
    - ZONE_LEVEL_AFFINITY widened per consensus + arbitration +
      post-reclassification adjustments. Lint warnings: 1,308 → 0.

A.7 Chain integrity:
    - repair_chains.py: drops chain refs when a chain has <2 published
      members (chain ceases to exist), renumbers all members of any
      chain whose positions are non-sequential / duplicated /
      non-monotonic-by-level. Sort key: level ascending, then old
      position, then qid (deterministic).
    - validate-vault.py: relaxed sequential check to unique-positions
      check. Position gaps from mid-chain deletions are normal; what
      matters is uniqueness + bloom-monotonicity (vault check --strict
      enforces both from YAML source-of-truth).

A.8 Practice page visual + zoom modal:
    - QuestionVisual.tsx: wraps the `<img>` in `<Zoom>` from
      react-medium-image-zoom (4 KB). Click image → fullscreen
      `<dialog data-rmiz-modal>`; ESC closes. Added test-id
      `question-visual-img` for stable selector.
    - New Playwright test: 9th in the suite, deep-links cloud-4492,
      asserts the dialog opens on click and closes on ESC.
    - TypeScript: removed `mermaid` from local Visual types in
      corpus.ts and corpus-vault.ts; tsc clean.

A.9 All gates green:
    - vault check --strict: 0 errors / 0 invariant failures
    - vault lint: 0 errors / 0 warnings (was 1,308 warnings)
    - vault codegen --check: artifacts in sync (hash baseline updated)
    - vault doctor: 0 fails (registry-history info, git-state warn
      on uncommitted state-pre-this-commit)
    - staffml validate-vault: 0 errors / 0 warnings, deployment-ready
    - Playwright: 9/9 pass (was 8; +zoom modal test)
    - render_visuals: 0 errors (was 2 silent failures pre-A.2)
    - tsc: clean

Distribution after reclassification: 9,544 published unchanged;
576 items moved zone via bloom-canonical mapping (full per-item
report at /tmp/reclassify_changes.csv). Chain count 879 → 850
after orphan-singleton drops. release_hash updated.

Carry-forward to next session (Phase B):
- Priority gap closure for parallelism cells + global L4-L6+
  (the run that produced this corpus did not close the targeted
  cells; B.3 needs specialized prompts per cell-class)
- 120 NEEDS_FIX items from coverage_loop/20260425_150712/ still
  carry judge fix_suggestions; spawn fix-agent in Phase C
2026-04-25 15:12:51 -04:00

276 lines
10 KiB
Python

#!/usr/bin/env python3
"""Render question visuals to ship-ready SVG.
The schema the website cares about is minimal::
visual:
kind: svg # always svg — that's what the website ships
path: <id>.svg # the static asset
alt: <text>
caption: <text>
# Build metadata (optional, ignored by website):
source_format: dot | matplotlib | hand # default: hand
The runtime story for the practice page is "load `<id>.svg` as an
`<img>`". The renderer's job is to produce that SVG when it's a build
artifact (DOT or matplotlib source). For hand-authored SVGs, the source
IS the asset, no build needed.
Source files live next to the asset by naming convention:
interviews/vault/visuals/<track>/<id>.svg # the asset (always)
interviews/vault/visuals/<track>/<id>.dot # iff source_format=dot
interviews/vault/visuals/<track>/<id>.py # iff source_format=matplotlib
Usage:
python3 render_visuals.py # render all stale
python3 render_visuals.py --force # force re-render
python3 render_visuals.py --id cloud-2846 # single question
python3 render_visuals.py --dry-run # plan only
Architecture: see interviews/vault/visuals/ARCHITECTURE.md.
"""
from __future__ import annotations
import argparse
import json
import re
import subprocess
import sys
from datetime import datetime, timezone
from pathlib import Path
from typing import Any
import yaml
VAULT_DIR = Path(__file__).resolve().parent.parent
QUESTIONS_DIR = VAULT_DIR / "questions"
VISUALS_DIR = VAULT_DIR / "visuals"
DEFAULT_FAILURE_LOG = VAULT_DIR / "_validation_results" / "render_failures.json"
VALID_KINDS = {"svg"} # what the website renders
# Source format is inferred from the filesystem: a `<basename>.dot` next
# to `<basename>.svg` means DOT-built; `.py` means matplotlib-built; no
# sibling means hand-authored. We do not encode this in the YAML schema.
SOURCE_EXT_TO_FORMAT = {".dot": "dot", ".py": "matplotlib"}
# Book SVG style: enforce these properties on every rendered SVG so DOT,
# matplotlib, and hand-SVG outputs render identically in the practice page.
SVG_FONT_FAMILY = "Helvetica Neue, Helvetica, Arial, sans-serif"
# ---------------------------------------------------------------------------
# Discovery
# ---------------------------------------------------------------------------
def discover_visuals() -> list[dict[str, Any]]:
"""Return one record per question with a `visual:` block.
Reads only the production-schema fields: kind, path, alt, caption.
Build metadata: source_format (optional).
"""
records = []
for yaml_path in QUESTIONS_DIR.glob("**/*.yaml"):
try:
data = yaml.safe_load(yaml_path.read_text(encoding="utf-8"))
except Exception as exc:
print(f" ! {yaml_path}: parse error {exc}", file=sys.stderr)
continue
if not data or "visual" not in data:
continue
v = data["visual"]
if not isinstance(v, dict):
continue
kind = v.get("kind", "svg")
if kind not in VALID_KINDS:
print(f" ! {data.get('id')}: unsupported kind={kind!r} "
f"(only {VALID_KINDS} ship)", file=sys.stderr)
continue
path = v.get("path")
if not path:
print(f" ! {data.get('id')}: missing visual.path", file=sys.stderr)
continue
track = data.get("track", "global")
track_dir = VISUALS_DIR / track
asset_path = track_dir / path
# Infer source format by filesystem: look for a sibling file with
# a known build-source extension. Absent => hand-authored.
basename = asset_path.with_suffix("")
source_path = None
source_format = "hand"
for ext, fmt in SOURCE_EXT_TO_FORMAT.items():
cand = basename.with_suffix(ext)
if cand.exists():
source_path = cand
source_format = fmt
break
records.append({
"id": data["id"],
"track": track,
"asset_path": asset_path,
"source_path": source_path,
"source_format": source_format,
"yaml_path": yaml_path,
})
return records
# ---------------------------------------------------------------------------
# Rendering
# ---------------------------------------------------------------------------
def render_one(rec: dict[str, Any], force: bool = False, dry_run: bool = False) -> str:
"""Render or pass through a single visual. Returns 'rendered'|'skipped'|error."""
qid = rec["id"]
fmt = rec["source_format"]
src = rec["source_path"]
out = rec["asset_path"]
if fmt == "hand":
if not out.exists():
return f"error:{qid}:hand-authored asset missing at {out}"
return "skipped"
# Build artifact path: needs source
if not src or not src.exists():
return f"error:{qid}:{fmt} source missing at {src}"
if not force and out.exists() and out.stat().st_mtime >= src.stat().st_mtime:
return "skipped"
if dry_run:
print(f" + would render {qid}: {src.name} -> {out.name}")
return "rendered"
out.parent.mkdir(parents=True, exist_ok=True)
if fmt == "dot":
_render_dot(src, out)
else:
_render_matplotlib(src, out)
_normalize_svg(out)
return "rendered"
def _render_dot(src: Path, out: Path) -> None:
result = subprocess.run(
["dot", "-Tsvg", str(src), "-o", str(out)],
capture_output=True, text=True, timeout=30,
)
if result.returncode != 0:
raise RuntimeError(f"dot failed: {result.stderr.strip()}")
def _render_matplotlib(src: Path, out: Path) -> None:
"""Execute the source script with VISUAL_OUT_PATH env var."""
import os
env = dict(os.environ)
env["VISUAL_OUT_PATH"] = str(out)
result = subprocess.run(
["python3", str(src)],
capture_output=True, text=True, timeout=60, env=env,
)
if result.returncode != 0:
raise RuntimeError(f"matplotlib script failed: {result.stderr.strip()}")
if not out.exists():
raise RuntimeError(
f"matplotlib script ran but did not write to {out}; "
"did the script use os.environ['VISUAL_OUT_PATH']?"
)
def _normalize_svg(path: Path) -> None:
"""Apply book-style normalization to a rendered SVG."""
text = path.read_text(encoding="utf-8")
if "data:image/" in text or "<image " in text:
raise RuntimeError(f"{path} contains embedded raster — not allowed")
text = re.sub(r"<!--\s*Generated by [^>]*?-->\s*", "", text)
if "font-family=" not in text.split(">", 1)[0]:
text = re.sub(
r"<svg(\s[^>]*?)>",
lambda m: f'<svg{m.group(1)} font-family="{SVG_FONT_FAMILY}">',
text, count=1,
)
path.write_text(text, encoding="utf-8")
# ---------------------------------------------------------------------------
# CLI
# ---------------------------------------------------------------------------
def main() -> int:
parser = argparse.ArgumentParser(description=__doc__)
parser.add_argument("--force", action="store_true",
help="Re-render even if output is fresh.")
parser.add_argument("--dry-run", action="store_true")
parser.add_argument("--id", help="Render only this question id.")
parser.add_argument("--fail-fast", action="store_true",
help="Abort on first per-item failure instead of "
"continuing through the batch.")
parser.add_argument("--failure-log", type=Path, default=DEFAULT_FAILURE_LOG,
help=f"Path for structured per-ID failure log "
f"(default: {DEFAULT_FAILURE_LOG}). Always "
"written on completion; empty array if no "
"failures.")
args = parser.parse_args()
recs = discover_visuals()
if args.id:
recs = [r for r in recs if r["id"] == args.id]
if not recs:
print(f"No visual found for id={args.id}", file=sys.stderr)
return 1
print(f"Discovered {len(recs)} visual(s).")
counts = {"rendered": 0, "skipped": 0, "error": 0}
failures: list[dict[str, Any]] = []
for rec in recs:
try:
status = render_one(rec, force=args.force, dry_run=args.dry_run)
except Exception as exc:
status = f"error:{rec['id']}:{exc}"
if status.startswith("error"):
print(f"{status}")
counts["error"] += 1
# Parse "error:<qid>:<message>" into structured record
parts = status.split(":", 2)
qid = parts[1] if len(parts) > 1 else rec["id"]
err_msg = parts[2] if len(parts) > 2 else status
failures.append({
"id": qid,
"track": rec["track"],
"source_format": rec["source_format"],
"source_path": str(rec["source_path"]) if rec["source_path"] else None,
"asset_path": str(rec["asset_path"]),
"error": err_msg.strip(),
})
if args.fail_fast:
print(" ! --fail-fast set; aborting batch", file=sys.stderr)
break
else:
print(f" {'' if status == 'rendered' else '·'} {rec['id']:30s} "
f"[{rec['source_format']:11s}] {status}")
counts[status] += 1
# Always write the failure log so downstream consumers (CI, judge step,
# release_gate.sh) can read a stable path. Empty array if no failures.
args.failure_log.parent.mkdir(parents=True, exist_ok=True)
args.failure_log.write_text(json.dumps({
"generated_at": datetime.now(timezone.utc).isoformat(),
"total_visuals": len(recs),
"rendered": counts["rendered"],
"skipped": counts["skipped"],
"errors": counts["error"],
"failures": failures,
}, indent=2), encoding="utf-8")
print(f"\nFailure log: {args.failure_log}")
print(f"Summary: rendered={counts['rendered']} "
f"skipped={counts['skipped']} errors={counts['error']}")
return 1 if counts["error"] else 0
if __name__ == "__main__":
sys.exit(main())