@zxyycom/vibe-check 0.0.1 → 0.0.3

package/src/package-checks/markdown-link-validation/options-resolution.ts

@@ -9,6 +9,7 @@ import {
   snapshotProjectFileSelection
 } from "../project-files/configuration.ts";
 import { DEFAULT_FINDING_POLICY, resolveFindingPolicy } from "../code-quality-findings/policy.ts";
+import { resolveOptionalLocalCacheOptions } from "../local-cache-options.ts";
 import type { ResolvedMarkdownLinkValidationOptions } from "./options.ts";
 import { validMarkdownLinkValidationOptions } from "./options-validation.ts";
 
@@ -32,8 +33,15 @@ export function resolveMarkdownLinkValidationOptions(
   const limits = resolvedLimits(input.limits);
   const files = resolvedFiles(input.files);
   const findingPolicy = resolveFindingPolicy(input.findingPolicy, DEFAULT_FINDING_POLICY);
-  if (limits === undefined || files === undefined || findingPolicy === undefined) return undefined;
-  const candidate = optionsCandidate(input, files, findingPolicy, limits);
+  const cache = resolveOptionalLocalCacheOptions(input.cache);
+  if (
+    limits === undefined ||
+    files === undefined ||
+    findingPolicy === undefined ||
+    cache === undefined
+  )
+    return undefined;
+  const candidate = optionsCandidate(input, files, findingPolicy, limits, cache);
   return candidate !== undefined && validMarkdownLinkValidationOptions(candidate)
     ? candidate
     : undefined;
@@ -43,7 +51,8 @@ function optionsCandidate(
   input: Readonly<Record<string, unknown>>,
   files: Exclude<ReturnType<typeof resolvedFiles>, undefined>,
   findingPolicy: Exclude<ReturnType<typeof resolveFindingPolicy>, undefined>,
-  limits: Readonly<Record<string, unknown>>
+  limits: Readonly<Record<string, unknown>>,
+  cache: ResolvedMarkdownLinkValidationOptions["cache"]
 ) {
   return canonicalizeJsonObject({
     files,
@@ -58,6 +67,7 @@ function optionsCandidate(
       input.rootExternalTargetMode === undefined ? "report" : input.rootExternalTargetMode,
     requireNonEmptyDirectories:
       input.requireNonEmptyDirectories === undefined ? false : input.requireNonEmptyDirectories,
+    cache,
     limits
   });
 }
@@ -76,6 +86,7 @@ function isMarkdownLinkValidationInput(
         "validateCrossDocumentAnchors",
         "rootExternalTargetMode",
         "requireNonEmptyDirectories",
+        "cache",
         "limits"
       ],
       required: []

package/src/package-checks/markdown-link-validation/options-validation.ts

@@ -1,31 +1,18 @@
-import {
-  hasExactPlainRecordKeys,
-  snapshotClosedRecord
-} from "../../data-boundary/closed-values.ts";
+import { snapshotExactClosedRecord } from "../../data-boundary/closed-values.ts";
+import { isPositiveSafeInteger } from "../../data-boundary/value-shapes.ts";
 import { validProjectFileSelection } from "../project-files/configuration.ts";
 import { validFindingPolicy } from "../code-quality-findings/policy.ts";
+import { validLocalCacheOptions } from "../local-cache-options.ts";
 import type { ResolvedMarkdownLinkValidationOptions } from "./options.ts";
 
-function exactRecord(
-  value: unknown,
-  keys: readonly string[]
-): Readonly<Record<string, unknown>> | undefined {
-  const record = snapshotClosedRecord(value);
-  return record !== undefined && hasExactPlainRecordKeys(record, keys) ? record : undefined;
-}
-
-function positiveSafeInteger(value: unknown): value is number {
-  return typeof value === "number" && Number.isSafeInteger(value) && value > 0;
-}
-
 function boundedPositiveSafeInteger(value: unknown, maximum: number): value is number {
-  return positiveSafeInteger(value) && value <= maximum;
+  return isPositiveSafeInteger(value) && value <= maximum;
 }
 
 export function validMarkdownLinkValidationOptions(
   value: unknown
 ): value is ResolvedMarkdownLinkValidationOptions {
-  const options = exactRecord(value, [
+  const options = snapshotExactClosedRecord(value, [
     "files",
     "findingPolicy",
     "requireExistingTargets",
@@ -33,6 +20,7 @@ export function validMarkdownLinkValidationOptions(
     "validateCrossDocumentAnchors",
     "rootExternalTargetMode",
     "requireNonEmptyDirectories",
+    "cache",
     "limits"
   ]);
   return options !== undefined && validOptionFields(options);
@@ -43,6 +31,7 @@ function validOptionFields(options: Readonly<Record<string, unknown>>): boolean
     validProjectFileSelection(options.files) &&
     validFindingPolicy(options.findingPolicy) &&
     validAnchorValidationOptions(options) &&
+    validLocalCacheOptions(options.cache) &&
     validMarkdownLinkLimits(options.limits)
   );
 }
@@ -62,7 +51,11 @@ function validRootExternalTargetMode(value: unknown): boolean {
 }
 
 function validMarkdownLinkLimits(value: unknown): boolean {
-  const limits = exactRecord(value, ["maxMarkdownBytes", "maxOccurrences", "maxTargetReads"]);
+  const limits = snapshotExactClosedRecord(value, [
+    "maxMarkdownBytes",
+    "maxOccurrences",
+    "maxTargetReads"
+  ]);
   return (
     limits !== undefined &&
     boundedPositiveSafeInteger(limits.maxMarkdownBytes, 16_777_216) &&

package/src/package-checks/markdown-link-validation/options.ts

@@ -1,11 +1,16 @@
+import type { PackageCheckAuthoringOptions } from "../check-authoring.ts";
 import type {
   ProjectFileSelection,
   ProjectFileSelectionOptions
 } from "../project-files/configuration.ts";
 import type { FindingPolicy } from "../code-quality-findings/policy.ts";
+import type { LocalCacheOptions } from "../local-cache-options.ts";
 
 export type MarkdownRootExternalTargetMode = "ignore" | "report" | "validate";
 
+/** 调用方显式拥有的 Link-private parse-facts local cache policy。 */
+type MarkdownLinkValidationCacheOptions = LocalCacheOptions;
+
 /** 可省略的 Markdown validation work limits。 */
 export interface MarkdownLinkValidationLimitOptions {
   readonly maxMarkdownBytes?: number;
@@ -14,7 +19,9 @@ export interface MarkdownLinkValidationLimitOptions {
 }
 
 /** `markdownLinkValidation(options?)` 接受的可省略 authoring policy。 */
-export interface MarkdownLinkValidationOptions {
+export interface MarkdownLinkValidationOptions<
+  Id extends string = string
+> extends PackageCheckAuthoringOptions<Id> {
   /** 参与本 Check 的 Markdown source selection；省略字段使用 package defaults。 */
   readonly files?: ProjectFileSelectionOptions;
   /** local-reference finding 是否使本 Check failed；省略时为 non-blocking。 */
@@ -31,6 +38,12 @@ export interface MarkdownLinkValidationOptions {
   readonly requireNonEmptyDirectories?: boolean;
   /** 每次运行的 Markdown 内容、occurrence 和 direct target work 上限。 */
   readonly limits?: MarkdownLinkValidationLimitOptions;
+  /**
+   * 显式启用的、调用方拥有的 Link parse-facts persistent cache。省略时关闭；enabled directory 必须 absolute、可信且可删除。
+   * 本 Check 只在其中使用 `markdown-link-parse-facts-v1.jsonl` 保存 source-derived facts；调用方负责容量和删除，且不提供
+   * confidentiality、automatic cleanup、concurrency 或 durability guarantees。cache 不可用时 fresh-parse，Check 结果不变。
+   */
+  readonly cache?: MarkdownLinkValidationCacheOptions;
 }
 
 /** `markdown-link-validation` execution 消费的完整、冻结 options。 */
@@ -42,6 +55,7 @@ export interface ResolvedMarkdownLinkValidationOptions {
   readonly validateCrossDocumentAnchors: boolean;
   readonly rootExternalTargetMode: MarkdownRootExternalTargetMode;
   readonly requireNonEmptyDirectories: boolean;
+  readonly cache: MarkdownLinkValidationCacheOptions;
   readonly limits: Readonly<{
     readonly maxMarkdownBytes: number;
     readonly maxOccurrences: number;

package/src/package-checks/markdown-link-validation/parse-facts-cache-payload.ts

@@ -0,0 +1,245 @@
+import { createHash } from "node:crypto";
+
+import {
+  snapshotClosedArray,
+  snapshotExactClosedRecord
+} from "../../data-boundary/closed-values.ts";
+import { canonicalJsonText } from "../../data-boundary/canonical-data.ts";
+import { isPositiveSafeInteger } from "../../data-boundary/value-shapes.ts";
+import type {
+  MarkdownHeading,
+  MarkdownLinkOccurrence,
+  MarkdownSourcePosition,
+  MarkdownSourceRange,
+  ParsedMarkdownLinkFacts
+} from "./markdown-parser.ts";
+
+const CACHE_FORMAT_VERSION = "markdown-link-parse-facts-jsonl-v1";
+const CACHE_PAYLOAD_VERSION = "1";
+/** Version the resolver memo and persistent identity share when parser facts change. */
+export const MARKDOWN_LINK_PARSE_FACTS_PARSER_CONTRACT_VERSION =
+  "mdast-gfm-frontmatter-github-slug-v1";
+const MAX_FACTS_PER_DOCUMENT = 100_000;
+const MAX_FACT_STRING_LENGTH = 16_777_216;
+
+type MarkdownLinkParseFactsPayload = Readonly<{
+  readonly occurrences: readonly MarkdownLinkOccurrence[];
+  readonly headings: readonly MarkdownHeading[];
+}>;
+
+/**
+ * Serializes one complete Link-private JSONL envelope from immutable parser facts.
+ * The session owns when this line is appended; this codec owns its shape and identity.
+ */
+export function cacheEnvelopeLine(identityDigest: string, facts: ParsedMarkdownLinkFacts): string {
+  return `${canonicalJsonText({
+    cacheFormatVersion: CACHE_FORMAT_VERSION,
+    identityDigest,
+    parserContractVersion: MARKDOWN_LINK_PARSE_FACTS_PARSER_CONTRACT_VERSION,
+    payloadVersion: CACHE_PAYLOAD_VERSION,
+    payload: projectMarkdownLinkParseFactsPayload(facts)
+  })}\n`;
+}
+
+/** Restores one complete, current-version JSONL envelope or rejects it as a cache miss. */
+export function parseCacheEnvelopeLine(
+  line: string
+):
+  | Readonly<{ readonly identityDigest: string; readonly facts: ParsedMarkdownLinkFacts }>
+  | undefined {
+  try {
+    const envelope = snapshotExactClosedRecord(JSON.parse(line), [
+      "cacheFormatVersion",
+      "identityDigest",
+      "parserContractVersion",
+      "payloadVersion",
+      "payload"
+    ]);
+    if (!isCurrentCacheEnvelope(envelope)) return undefined;
+    return Object.freeze({
+      identityDigest: envelope.identityDigest,
+      facts: parseMarkdownLinkParseFactsPayload(envelope.payload)
+    });
+  } catch {
+    return undefined;
+  }
+}
+
+/** SHA-256 identity over the current, exact authorized Markdown bytes. */
+export function markdownBytesDigest(bytes: Uint8Array): string {
+  return createHash("sha256").update(bytes).digest("hex");
+}
+
+/** Converts parser-owned immutable facts to the closed persistent JSON payload. */
+export function projectMarkdownLinkParseFactsPayload(
+  facts: ParsedMarkdownLinkFacts
+): MarkdownLinkParseFactsPayload {
+  return Object.freeze({
+    occurrences: Object.freeze(
+      facts.occurrences.map((occurrence) =>
+        Object.freeze({
+          kind: occurrence.kind,
+          rawDestination: occurrence.rawDestination,
+          range: projectSourceRange(occurrence.range)
+        })
+      )
+    ),
+    headings: Object.freeze(
+      facts.headings.map((heading) =>
+        Object.freeze({ slug: heading.slug, range: projectSourceRange(heading.range) })
+      )
+    )
+  });
+}
+
+/** Strictly restores the Link-private persistent JSON payload to immutable parser facts. */
+export function parseMarkdownLinkParseFactsPayload(value: unknown): ParsedMarkdownLinkFacts {
+  const payload = snapshotExactClosedRecord(value, ["occurrences", "headings"]);
+  if (payload === undefined) throw new TypeError("Markdown Link cache payload must be closed");
+  const occurrences = parseOccurrences(payload.occurrences);
+  const headings = parseHeadings(payload.headings);
+  if (occurrences === undefined || headings === undefined) {
+    throw new TypeError("Markdown Link cache payload contains invalid parse facts");
+  }
+  return Object.freeze({ occurrences, headings });
+}
+
+function isCurrentCacheEnvelope(
+  envelope: Readonly<Record<string, unknown>> | undefined
+): envelope is Readonly<{
+  readonly identityDigest: string;
+  readonly payload: unknown;
+}> {
+  return (
+    envelope?.cacheFormatVersion === CACHE_FORMAT_VERSION &&
+    validIdentityDigest(envelope?.identityDigest) &&
+    envelope?.parserContractVersion === MARKDOWN_LINK_PARSE_FACTS_PARSER_CONTRACT_VERSION &&
+    envelope?.payloadVersion === CACHE_PAYLOAD_VERSION
+  );
+}
+
+function validIdentityDigest(value: unknown): value is string {
+  return typeof value === "string" && /^[a-f0-9]{64}$/u.test(value);
+}
+
+function parseOccurrences(value: unknown): readonly MarkdownLinkOccurrence[] | undefined {
+  const values = boundedPayloadArray(value);
+  if (values === undefined) return undefined;
+  const occurrences: MarkdownLinkOccurrence[] = [];
+  for (const candidate of values) {
+    const occurrence = snapshotExactClosedRecord(candidate, ["kind", "rawDestination", "range"]);
+    if (
+      occurrence === undefined ||
+      (occurrence.kind !== "image" && occurrence.kind !== "link") ||
+      !validFactString(occurrence.rawDestination)
+    ) {
+      return undefined;
+    }
+    const range = parseSourceRange(occurrence.range);
+    if (range === undefined) return undefined;
+    occurrences.push(
+      Object.freeze({ kind: occurrence.kind, rawDestination: occurrence.rawDestination, range })
+    );
+  }
+  return Object.freeze(occurrences);
+}
+
+function parseHeadings(value: unknown): readonly MarkdownHeading[] | undefined {
+  const values = boundedPayloadArray(value);
+  if (values === undefined) return undefined;
+  const headings: MarkdownHeading[] = [];
+  for (const candidate of values) {
+    const heading = snapshotExactClosedRecord(candidate, ["slug", "range"]);
+    if (heading === undefined || !validFactString(heading.slug)) return undefined;
+    const range = parseSourceRange(heading.range);
+    if (range === undefined) return undefined;
+    headings.push(Object.freeze({ slug: heading.slug, range }));
+  }
+  return Object.freeze(headings);
+}
+
+function boundedPayloadArray(value: unknown): readonly unknown[] | undefined {
+  const values = snapshotClosedArray(value);
+  return values !== undefined && values.length <= MAX_FACTS_PER_DOCUMENT ? values : undefined;
+}
+
+function validFactString(value: unknown): value is string {
+  return (
+    typeof value === "string" &&
+    value.length <= MAX_FACT_STRING_LENGTH &&
+    !hasUnpairedSurrogate(value)
+  );
+}
+
+function parseSourceRange(value: unknown): MarkdownSourceRange | undefined {
+  const range = snapshotExactClosedRecord(value, ["startOffset", "endOffset", "start", "end"]);
+  if (
+    range === undefined ||
+    !boundedNonNegativeSafeInteger(range.startOffset) ||
+    !boundedNonNegativeSafeInteger(range.endOffset) ||
+    range.endOffset < range.startOffset
+  ) {
+    return undefined;
+  }
+  const start = parseSourcePosition(range.start);
+  const end = parseSourcePosition(range.end);
+  if (start === undefined || end === undefined || !sourcePositionFollows(end, start)) {
+    return undefined;
+  }
+  return Object.freeze({ startOffset: range.startOffset, endOffset: range.endOffset, start, end });
+}
+
+function sourcePositionFollows(
+  end: MarkdownSourcePosition,
+  start: MarkdownSourcePosition
+): boolean {
+  return end.line > start.line || (end.line === start.line && end.column >= start.column);
+}
+
+function parseSourcePosition(value: unknown): MarkdownSourcePosition | undefined {
+  const position = snapshotExactClosedRecord(value, ["line", "column"]);
+  if (
+    position === undefined ||
+    !boundedPositiveSafeInteger(position.line) ||
+    !boundedPositiveSafeInteger(position.column)
+  ) {
+    return undefined;
+  }
+  return Object.freeze({ line: position.line, column: position.column });
+}
+
+function projectSourceRange(range: MarkdownSourceRange): MarkdownSourceRange {
+  return Object.freeze({
+    startOffset: range.startOffset,
+    endOffset: range.endOffset,
+    start: Object.freeze({ line: range.start.line, column: range.start.column }),
+    end: Object.freeze({ line: range.end.line, column: range.end.column })
+  });
+}
+
+function boundedNonNegativeSafeInteger(value: unknown): value is number {
+  return (
+    typeof value === "number" &&
+    Number.isSafeInteger(value) &&
+    value >= 0 &&
+    value <= MAX_FACT_STRING_LENGTH
+  );
+}
+
+function boundedPositiveSafeInteger(value: unknown): value is number {
+  return isPositiveSafeInteger(value) && value <= MAX_FACT_STRING_LENGTH + 1;
+}
+
+function hasUnpairedSurrogate(value: string): boolean {
+  for (let index = 0; index < value.length; index += 1) {
+    const codeUnit = value.charCodeAt(index);
+    if (codeUnit >= 0xd800 && codeUnit <= 0xdbff) {
+      const nextCodeUnit = value.charCodeAt(index + 1);
+      if (!(nextCodeUnit >= 0xdc00 && nextCodeUnit <= 0xdfff)) return true;
+      index += 1;
+      continue;
+    }
+    if (codeUnit >= 0xdc00 && codeUnit <= 0xdfff) return true;
+  }
+  return false;
+}

package/src/package-checks/markdown-link-validation/parse-facts-cache.ts

@@ -0,0 +1,154 @@
+import { isUtf8 } from "node:buffer";
+import { appendFile, mkdir, readFile } from "node:fs/promises";
+import path from "node:path";
+
+import {
+  cacheEnvelopeLine,
+  markdownBytesDigest,
+  parseCacheEnvelopeLine
+} from "./parse-facts-cache-payload.ts";
+import {
+  parseMarkdownLinkFacts,
+  type MarkdownLinkParseResult,
+  type ParsedMarkdownLinkFacts
+} from "./markdown-parser.ts";
+import type { ResolvedMarkdownLinkValidationOptions } from "./options.ts";
+
+const CACHE_FILE_NAME = "markdown-link-parse-facts-v1.jsonl";
+
+/**
+ * Link-private state for one resolver invocation's parse-facts cache lifecycle.
+ * It restores at most once, collects successful fresh facts by identity, and
+ * publishes at the execution terminal boundary.
+ */
+export class MarkdownLinkParseFactsSession {
+  readonly #cache: ResolvedMarkdownLinkValidationOptions["cache"];
+  readonly #restored = new Map<string, ParsedMarkdownLinkFacts>();
+  readonly #dirty = new Map<string, ParsedMarkdownLinkFacts>();
+  #initialRead: Promise<void> | undefined;
+  #canPublish = true;
+  #needsPartialTailIsolation = false;
+  #finalization: Promise<void> | undefined;
+
+  constructor(cache: ResolvedMarkdownLinkValidationOptions["cache"]) {
+    this.#cache = cache;
+  }
+
+  async parse(
+    bytes: Uint8Array,
+    signal: AbortSignal
+  ): Promise<MarkdownLinkParseResult | undefined> {
+    if (signal.aborted || !isUtf8(bytes)) return undefined;
+    if (!this.#cache.enabled) return parseCurrentMarkdownBytes(bytes, signal);
+
+    await this.#restoreOnce();
+    if (signal.aborted) return undefined;
+    const identityDigest = markdownBytesDigest(bytes);
+    const cachedFacts = this.#cachedFacts(identityDigest);
+    if (cachedFacts !== undefined) return parsedFacts(cachedFacts);
+    return this.#parseFreshFacts(bytes, identityDigest, signal);
+  }
+
+  finalize(signal: AbortSignal): Promise<void> {
+    this.#finalization ??= this.#publish(signal);
+    return this.#finalization;
+  }
+
+  #cachedFacts(identityDigest: string): ParsedMarkdownLinkFacts | undefined {
+    return this.#restored.get(identityDigest) ?? this.#dirty.get(identityDigest);
+  }
+
+  async #parseFreshFacts(
+    bytes: Uint8Array,
+    identityDigest: string,
+    signal: AbortSignal
+  ): Promise<MarkdownLinkParseResult | undefined> {
+    const parsed = parseCurrentMarkdownBytes(bytes, signal);
+    if (parsed === undefined || !parsed.ok) return parsed;
+    if (signal.aborted) return undefined;
+    this.#dirty.set(identityDigest, parsed.facts);
+    return parsed;
+  }
+
+  async #restoreOnce(): Promise<void> {
+    this.#initialRead ??= this.#restore();
+    await this.#initialRead;
+  }
+
+  async #restore(): Promise<void> {
+    if (!this.#cache.enabled) return;
+    let contents: Uint8Array;
+    try {
+      contents = await readFile(cacheFilePath(this.#cache.directory));
+    } catch (error) {
+      if (isMissingFile(error)) return;
+      this.#canPublish = false;
+      return;
+    }
+
+    this.#needsPartialTailIsolation =
+      contents.length > 0 && contents[contents.length - 1] !== "\n".charCodeAt(0);
+    let text: string;
+    try {
+      text = new TextDecoder("utf-8", { fatal: true }).decode(contents);
+    } catch {
+      this.#canPublish = false;
+      return;
+    }
+    const lastNewline = text.lastIndexOf("\n");
+    if (lastNewline < 0) return;
+    for (const line of text.slice(0, lastNewline).split("\n")) {
+      const restored = parseCacheEnvelopeLine(line);
+      if (restored !== undefined) this.#restored.set(restored.identityDigest, restored.facts);
+    }
+  }
+
+  async #publish(signal: AbortSignal): Promise<void> {
+    if (signal.aborted || !this.#cache.enabled || !this.#canPublish || this.#dirty.size === 0)
+      return;
+    try {
+      await mkdir(this.#cache.directory, { recursive: true });
+      if (signal.aborted) return;
+      const block = `${this.#needsPartialTailIsolation ? "\n" : ""}${[...this.#dirty.entries()]
+        .map(([identityDigest, facts]) => cacheEnvelopeLine(identityDigest, facts))
+        .join("")}`;
+      // appendFile both creates a missing file and appends the invocation's one complete block.
+      await appendFile(cacheFilePath(this.#cache.directory), block);
+    } catch {
+      // Cache publication is best effort and never changes Check settlement.
+    }
+  }
+}
+
+function parseCurrentMarkdownBytes(
+  bytes: Uint8Array,
+  signal: AbortSignal
+): MarkdownLinkParseResult | undefined {
+  const markdown = decodeMarkdownBytes(bytes);
+  return markdown === undefined || signal.aborted ? undefined : parseMarkdownLinkFacts(markdown);
+}
+
+function parsedFacts(facts: ParsedMarkdownLinkFacts): MarkdownLinkParseResult {
+  return Object.freeze({ ok: true as const, facts });
+}
+
+function decodeMarkdownBytes(bytes: Uint8Array): string | undefined {
+  try {
+    return new TextDecoder("utf-8", { fatal: true }).decode(bytes);
+  } catch {
+    return undefined;
+  }
+}
+
+function cacheFilePath(directory: string): string {
+  return path.join(directory, CACHE_FILE_NAME);
+}
+
+function isMissingFile(error: unknown): boolean {
+  return (
+    typeof error === "object" &&
+    error !== null &&
+    "code" in error &&
+    (error as Readonly<{ readonly code?: unknown }>).code === "ENOENT"
+  );
+}