@yadsh/dsh-qa-surface 0.4.0 → 0.6.0

package/lib/documents/defaults.js

@@ -0,0 +1,78 @@
+/**
+ * Canonical document defaults (§20).
+ *
+ * This module must stay free of Node builtins: it is the only part of the
+ * document subsystem the browser settings bundle reaches (through the QA
+ * surface's resolved configuration), and the page cannot load `node:*`. The
+ * resolver and everything that touches the filesystem live in `config.ts`,
+ * which imports these values.
+ */
+import { DEFAULT_ASSET_MIME_TYPES } from "./security/mime.js";
+/** The canonical resolved configuration; the schema and resolver both use it. */
+export const DEFAULT_QA_DOCUMENTS_CONFIG = Object.freeze({
+    enabled: true,
+    storage: Object.freeze({
+        root: null,
+        retainSource: true,
+        retainInputs: true,
+        allowedInputRoots: Object.freeze([]),
+    }),
+    templates: Object.freeze({ root: null, default: "default" }),
+    create: Object.freeze({
+        defaultPdfMode: "auto",
+        allowFormats: Object.freeze(["docx", "pdf"]),
+        allowRawMarkup: false,
+        toc: false,
+    }),
+    extraction: Object.freeze({
+        defaultMode: "accurate",
+        ocr: "auto",
+        ocrLanguages: Object.freeze([]),
+        extractImages: true,
+        extractTables: true,
+        preservePageMarkers: false,
+        allowFallback: true,
+        maxInlineChars: 200_000,
+    }),
+    docling: Object.freeze({
+        enabled: true,
+        baseUrl: "http://docling:5001",
+        timeoutMs: 120_000,
+    }),
+    pandoc: Object.freeze({ executable: "pandoc", timeoutMs: 60_000 }),
+    libreoffice: Object.freeze({
+        executable: "libreoffice",
+        timeoutMs: 120_000,
+    }),
+    typst: Object.freeze({
+        enabled: false,
+        executable: "typst",
+        timeoutMs: 120_000,
+    }),
+    markitdown: Object.freeze({
+        enabled: false,
+        executable: "markitdown",
+        timeoutMs: 120_000,
+    }),
+    workers: Object.freeze({
+        renderConcurrency: 2,
+        extractionConcurrency: 2,
+        ocrConcurrency: 1,
+    }),
+    retention: Object.freeze({
+        enabled: true,
+        maxAgeDays: 30,
+        cleanupIntervalHours: 12,
+    }),
+    limits: Object.freeze({
+        maxInputBytes: 52_428_800,
+        maxMarkdownChars: 5_000_000,
+        maxPages: 1_000,
+        maxExtractedImages: 500,
+        maxAssetBytes: 20_971_520,
+        maxResponseBytes: 268_435_456,
+        maxExtractedMarkdownBytes: 16_777_216,
+        allowedAssetMimeTypes: Object.freeze([...DEFAULT_ASSET_MIME_TYPES]),
+    }),
+});
+//# sourceMappingURL=defaults.js.map

package/lib/documents/defaults.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"defaults.js","sourceRoot":"","sources":["../../src/documents/defaults.ts"],"names":[],"mappings":"AAAA;;;;;;;;GAQG;AAEH,OAAO,EAAE,wBAAwB,EAAE,MAAM,oBAAoB,CAAC;AAoK9D,iFAAiF;AACjF,MAAM,CAAC,MAAM,2BAA2B,GACtC,MAAM,CAAC,MAAM,CAAC;IACZ,OAAO,EAAE,IAAI;IACb,OAAO,EAAE,MAAM,CAAC,MAAM,CAAC;QACrB,IAAI,EAAE,IAAI;QACV,YAAY,EAAE,IAAI;QAClB,YAAY,EAAE,IAAI;QAClB,iBAAiB,EAAE,MAAM,CAAC,MAAM,CAAC,EAAE,CAAC;KACrC,CAAC;IACF,SAAS,EAAE,MAAM,CAAC,MAAM,CAAC,EAAE,IAAI,EAAE,IAAI,EAAE,OAAO,EAAE,SAAS,EAAE,CAAC;IAC5D,MAAM,EAAE,MAAM,CAAC,MAAM,CAAC;QACpB,cAAc,EAAE,MAAM;QACtB,YAAY,EAAE,MAAM,CAAC,MAAM,CAAC,CAAC,MAAM,EAAE,KAAK,CAAU,CAAC;QACrD,cAAc,EAAE,KAAK;QACrB,GAAG,EAAE,KAAK;KACX,CAAC;IACF,UAAU,EAAE,MAAM,CAAC,MAAM,CAAC;QACxB,WAAW,EAAE,UAAU;QACvB,GAAG,EAAE,MAAM;QACX,YAAY,EAAE,MAAM,CAAC,MAAM,CAAC,EAAE,CAAC;QAC/B,aAAa,EAAE,IAAI;QACnB,aAAa,EAAE,IAAI;QACnB,mBAAmB,EAAE,KAAK;QAC1B,aAAa,EAAE,IAAI;QACnB,cAAc,EAAE,OAAO;KACxB,CAAC;IACF,OAAO,EAAE,MAAM,CAAC,MAAM,CAAC;QACrB,OAAO,EAAE,IAAI;QACb,OAAO,EAAE,qBAAqB;QAC9B,SAAS,EAAE,OAAO;KACnB,CAAC;IACF,MAAM,EAAE,MAAM,CAAC,MAAM,CAAC,EAAE,UAAU,EAAE,QAAQ,EAAE,SAAS,EAAE,MAAM,EAAE,CAAC;IAClE,WAAW,EAAE,MAAM,CAAC,MAAM,CAAC;QACzB,UAAU,EAAE,aAAa;QACzB,SAAS,EAAE,OAAO;KACnB,CAAC;IACF,KAAK,EAAE,MAAM,CAAC,MAAM,CAAC;QACnB,OAAO,EAAE,KAAK;QACd,UAAU,EAAE,OAAO;QACnB,SAAS,EAAE,OAAO;KACnB,CAAC;IACF,UAAU,EAAE,MAAM,CAAC,MAAM,CAAC;QACxB,OAAO,EAAE,KAAK;QACd,UAAU,EAAE,YAAY;QACxB,SAAS,EAAE,OAAO;KACnB,CAAC;IACF,OAAO,EAAE,MAAM,CAAC,MAAM,CAAC;QACrB,iBAAiB,EAAE,CAAC;QACpB,qBAAqB,EAAE,CAAC;QACxB,cAAc,EAAE,CAAC;KAClB,CAAC;IACF,SAAS,EAAE,MAAM,CAAC,MAAM,CAAC;QACvB,OAAO,EAAE,IAAI;QACb,UAAU,EAAE,EAAE;QACd,oBAAoB,EAAE,EAAE;KACzB,CAAC;IACF,MAAM,EAAE,MAAM,CAAC,MAAM,CAAC;QACpB,aAAa,EAAE,UAAU;QACzB,gBAAgB,EAAE,SAAS;QAC3B,QAAQ,EAAE,KAAK;QACf,kBAAkB,EAAE,GAAG;QACvB,aAAa,EAAE,UAAU;QACzB,gBAAgB,EAAE,WAAW;QAC7B,yBAAyB,EAAE,UAAU;QACrC,qBAAqB,EAAE,MAAM,CAAC,MAAM,CAAC,CAAC,GAAG,wBAAwB,CAAC,CAAC;KACpE,CAAC;CACH,CAAC,CAAC"}

package/lib/documents/errors.js

@@ -0,0 +1,101 @@
+/**
+ * Typed failures of the document pipeline.
+ *
+ * Every failure the agent can observe carries a stable machine-readable code
+ * (§24) and the backend that produced it. Backend output is never forwarded
+ * verbatim: {@link sanitizeBackendOutput} strips ANSI escapes, collapses
+ * whitespace, replaces absolute paths with the placeholder the operator can
+ * recognize as "inside the sandbox" and caps the length, so a converter's
+ * stderr cannot leak the deployment's layout or the document's content into a
+ * model request.
+ */
+export const DOCUMENT_ERROR_CODES = [
+    "INVALID_INPUT",
+    "UNSUPPORTED_FORMAT",
+    "UNSUPPORTED_CONVERSION",
+    "INPUT_TOO_LARGE",
+    "DOCUMENT_TOO_LARGE",
+    "FILE_NOT_FOUND",
+    "TEMPLATE_NOT_FOUND",
+    "INVALID_TEMPLATE",
+    "INVALID_ASSET",
+    "PATH_NOT_ALLOWED",
+    "ENCRYPTED_DOCUMENT",
+    "MACRO_ENABLED_DOCUMENT",
+    "OCR_FAILED",
+    "EXTRACTION_FAILED",
+    "RENDER_FAILED",
+    "CONVERSION_FAILED",
+    "BACKEND_UNAVAILABLE",
+    "BACKEND_TIMEOUT",
+    "BACKEND_FAILED",
+    "ARTIFACT_WRITE_FAILED",
+];
+/** Retryability per code: only transient backend conditions ask for a retry. */
+const RETRYABLE = new Set([
+    "BACKEND_UNAVAILABLE",
+    "BACKEND_TIMEOUT",
+    "BACKEND_FAILED",
+]);
+export class DocumentError extends Error {
+    code;
+    retryable;
+    backend;
+    details;
+    constructor(code, message, fields = {}) {
+        super(`document pipeline (${code}): ${message}`, { cause: fields.cause });
+        this.name = "DocumentError";
+        this.code = code;
+        this.retryable = fields.retryable ?? RETRYABLE.has(code);
+        this.backend = fields.backend;
+        this.details = fields.details;
+    }
+    /** The structured form the manifest and the logs carry (§24). */
+    toShape() {
+        return {
+            code: this.code,
+            message: this.message,
+            retryable: this.retryable,
+            ...(this.backend === undefined ? {} : { backend: this.backend }),
+            ...(this.details === undefined ? {} : { details: this.details }),
+        };
+    }
+}
+/* eslint-disable no-control-regex -- escape and control sequences are the point */
+const ANSI_PATTERN = /\u001B\[[0-9;]*[A-Za-z]/gu;
+const CONTROL_PATTERN = /[\u0000-\u0008\u000B\u000C\u000E-\u001F\u007F]/gu;
+/* eslint-enable no-control-regex */
+// Anchored at a word start and requiring a second separator, so prose such as
+// "and/or" survives while `/tmp/qa-documents/job-1/output.docx` does not.
+const ABSOLUTE_PATH_PATTERN = /(?<=^|\s)(?:[A-Za-z]:[\\/]|\/)(?:[^\s"'`()[\]{},;]+[\\/])+[^\s"'`()[\]{},;]*/gu;
+const DEFAULT_BACKEND_MESSAGE_MAX = 2_000;
+/**
+ * Make a backend's diagnostic text safe to show: no escapes, no absolute
+ * paths, one line per statement, at most `max` characters.
+ */
+export function sanitizeBackendOutput(text, max = DEFAULT_BACKEND_MESSAGE_MAX) {
+    const cleaned = text
+        .replace(ANSI_PATTERN, "")
+        .replace(CONTROL_PATTERN, "")
+        .replace(/\r\n?/gu, "\n")
+        .replace(ABSOLUTE_PATH_PATTERN, "<path>")
+        .split("\n")
+        .map((line) => line.trim())
+        .filter((line) => line !== "")
+        .join("; ")
+        .replace(/\s{2,}/gu, " ");
+    return cleaned.length > max ? `${cleaned.slice(0, max)}…` : cleaned;
+}
+/** Wrap any thrown value into a {@link DocumentError} without losing its code. */
+export function asDocumentError(error, fallbackCode, backend) {
+    if (error instanceof DocumentError)
+        return error;
+    const message = error instanceof Error
+        ? sanitizeBackendOutput(error.message)
+        : sanitizeBackendOutput(String(error));
+    return new DocumentError(fallbackCode, message === "" ? "the operation failed" : message, {
+        ...(backend === undefined ? {} : { backend }),
+        cause: error,
+    });
+}
+//# sourceMappingURL=errors.js.map

package/lib/documents/errors.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"errors.js","sourceRoot":"","sources":["../../src/documents/errors.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;GAUG;AAEH,MAAM,CAAC,MAAM,oBAAoB,GAAG;IAClC,eAAe;IACf,oBAAoB;IACpB,wBAAwB;IACxB,iBAAiB;IACjB,oBAAoB;IACpB,gBAAgB;IAChB,oBAAoB;IACpB,kBAAkB;IAClB,eAAe;IACf,kBAAkB;IAClB,oBAAoB;IACpB,wBAAwB;IACxB,YAAY;IACZ,mBAAmB;IACnB,eAAe;IACf,mBAAmB;IACnB,qBAAqB;IACrB,iBAAiB;IACjB,gBAAgB;IAChB,uBAAuB;CACf,CAAC;AAIX,gFAAgF;AAChF,MAAM,SAAS,GAAmC,IAAI,GAAG,CAAoB;IAC3E,qBAAqB;IACrB,iBAAiB;IACjB,gBAAgB;CACjB,CAAC,CAAC;AASH,MAAM,OAAO,aAAc,SAAQ,KAAK;IAC7B,IAAI,CAAoB;IACxB,SAAS,CAAU;IACnB,OAAO,CAAqB;IAC5B,OAAO,CAAgD;IAEhE,YACE,IAAuB,EACvB,OAAe,EACf,MAAM,GAAwB,EAAE;QAEhC,KAAK,CAAC,sBAAsB,IAAI,MAAM,OAAO,EAAE,EAAE,EAAE,KAAK,EAAE,MAAM,CAAC,KAAK,EAAE,CAAC,CAAC;QAC1E,IAAI,CAAC,IAAI,GAAG,eAAe,CAAC;QAC5B,IAAI,CAAC,IAAI,GAAG,IAAI,CAAC;QACjB,IAAI,CAAC,SAAS,GAAG,MAAM,CAAC,SAAS,IAAI,SAAS,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;QACzD,IAAI,CAAC,OAAO,GAAG,MAAM,CAAC,OAAO,CAAC;QAC9B,IAAI,CAAC,OAAO,GAAG,MAAM,CAAC,OAAO,CAAC;IAChC,CAAC;IAED,iEAAiE;IACjE,OAAO;QAOL,OAAO;YACL,IAAI,EAAE,IAAI,CAAC,IAAI;YACf,OAAO,EAAE,IAAI,CAAC,OAAO;YACrB,SAAS,EAAE,IAAI,CAAC,SAAS;YACzB,GAAG,CAAC,IAAI,CAAC,OAAO,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,OAAO,EAAE,IAAI,CAAC,OAAO,EAAE,CAAC;YAChE,GAAG,CAAC,IAAI,CAAC,OAAO,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,OAAO,EAAE,IAAI,CAAC,OAAO,EAAE,CAAC;SACjE,CAAC;IACJ,CAAC;CACF;AAED,mFAAmF;AACnF,MAAM,YAAY,GAAG,2BAA2B,CAAC;AACjD,MAAM,eAAe,GAAG,kDAAkD,CAAC;AAC3E,oCAAoC;AACpC,8EAA8E;AAC9E,0EAA0E;AAC1E,MAAM,qBAAqB,GACzB,gFAAgF,CAAC;AACnF,MAAM,2BAA2B,GAAG,KAAK,CAAC;AAE1C;;;GAGG;AACH,MAAM,UAAU,qBAAqB,CACnC,IAAY,EACZ,GAAG,GAAW,2BAA2B;IAEzC,MAAM,OAAO,GAAG,IAAI;SACjB,OAAO,CAAC,YAAY,EAAE,EAAE,CAAC;SACzB,OAAO,CAAC,eAAe,EAAE,EAAE,CAAC;SAC5B,OAAO,CAAC,SAAS,EAAE,IAAI,CAAC;SACxB,OAAO,CAAC,qBAAqB,EAAE,QAAQ,CAAC;SACxC,KAAK,CAAC,IAAI,CAAC;SACX,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC;SAC1B,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,KAAK,EAAE,CAAC;SAC7B,IAAI,CAAC,IAAI,CAAC;SACV,OAAO,CAAC,UAAU,EAAE,GAAG,CAAC,CAAC;IAC5B,OAAO,OAAO,CAAC,MAAM,GAAG,GAAG,CAAC,CAAC,CAAC,GAAG,OAAO,CAAC,KAAK,CAAC,CAAC,EAAE,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,OAAO,CAAC;AACtE,CAAC;AAED,kFAAkF;AAClF,MAAM,UAAU,eAAe,CAC7B,KAAc,EACd,YAA+B,EAC/B,OAAgB;IAEhB,IAAI,KAAK,YAAY,aAAa;QAAE,OAAO,KAAK,CAAC;IACjD,MAAM,OAAO,GACX,KAAK,YAAY,KAAK;QACpB,CAAC,CAAC,qBAAqB,CAAC,KAAK,CAAC,OAAO,CAAC;QACtC,CAAC,CAAC,qBAAqB,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,CAAC;IAC3C,OAAO,IAAI,aAAa,CACtB,YAAY,EACZ,OAAO,KAAK,EAAE,CAAC,CAAC,CAAC,sBAAsB,CAAC,CAAC,CAAC,OAAO,EACjD;QACE,GAAG,CAAC,OAAO,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,OAAO,EAAE,CAAC;QAC7C,KAAK,EAAE,KAAK;KACb,CACF,CAAC;AACJ,CAAC"}

package/lib/documents/index.js

@@ -0,0 +1,80 @@
+/**
+ * The document subsystem of `dsh-qa-surface` (§38).
+ *
+ * Markdown is the canonical source for everything the agent creates, the
+ * backends that turn it into DOCX/PDF (or read it back out of them) live
+ * behind the provider interfaces in `providers/`, and the four semantic tools
+ * in `tools/` are the only surface the model ever sees. Everything else —
+ * artifact bundles, manifests, template resolution, path containment, limits —
+ * exists to make those four tools predictable.
+ *
+ * The public entry point is {@link installDocumentSubsystem}.
+ */
+export { DOCUMENT_ERROR_CODES, DocumentError, asDocumentError, sanitizeBackendOutput, } from "./errors.js";
+export { DEFAULT_QA_DOCUMENTS_CONFIG, applyDocumentsEnvOverrides, resolveDocumentsConfig, retentionIntervalMs, } from "./config.js";
+export { createDocumentRuntime, DocumentRuntime, } from "./runtime.js";
+export { createDocumentTools, registerDocumentTools, DOCUMENT_TOOL_NAMES, DOCUMENT_CONVERT_TOOL, DOCUMENT_CREATE_TOOL, DOCUMENT_INSPECT_TOOL, DOCUMENT_TO_MARKDOWN_TOOL, } from "./tools/index.js";
+export { createProviders, } from "./providers/registry.js";
+export { documentCapabilities, documentHealth } from "./capabilities.js";
+export { ArtifactStore, sha256Hex, sha256OfFile } from "./artifacts/store.js";
+export { createArtifactId, createUlid, isArtifactId } from "./artifacts/ids.js";
+export { MANIFEST_FILE, MANIFEST_SCHEMA_VERSION, readManifest, writeManifest, } from "./artifacts/manifest.js";
+export { BUILTIN_TEMPLATE_NAME, TEMPLATE_MANIFEST_FILE, loadTemplateRegistry, templateNames, } from "./templates/registry.js";
+export { resolveTemplate, } from "./templates/resolver.js";
+export { CONVERSION_ROUTES } from "./orchestrator/convert-document.js";
+export { defaultArtifactRoot, resolveDocumentScope, } from "./orchestrator/scope.js";
+export { silentDocumentLogger, } from "./orchestrator/runtime-deps.js";
+export { assertInsideRoot, canonicalizeForContainment, isInsideRoot, resolveInsideRoot, sanitizeFilename, } from "./security/paths.js";
+export { DEFAULT_ASSET_MIME_TYPES, detectFormatByExtension, describeUnsupported, sniffDocument, supportedFormatOf, } from "./security/file-types.js";
+export { applyDirectives } from "./markdown/directives.js";
+export { normalizeExtractedMarkdown } from "./markdown/normalize.js";
+export { parseFrontMatter } from "./markdown/frontmatter.js";
+export { readPdfFacts, readDocxFacts, readDocxMetadata, } from "./inspect/facts.js";
+export { readZipEntries, readZipEntry, readZipEntryByName, } from "./inspect/zip.js";
+import { resolveDocumentsConfig, retentionIntervalMs, } from "./config.js";
+import { DocumentRuntime } from "./runtime.js";
+import { registerDocumentTools, DOCUMENT_TOOL_NAMES, } from "./tools/index.js";
+/**
+ * Wire the subsystem into a host context: build the runtime, register the four
+ * tools, and schedule the retention sweep when a storage root is pinned.
+ *
+ * `enabled: false` is answered with `undefined` — the caller then leaves both
+ * the tools and the timer absent instead of registering inert definitions.
+ */
+export function installDocumentSubsystem(ctx, options) {
+    const config = resolveDocumentsConfig(options.config);
+    if (!config.enabled)
+        return undefined;
+    const runtime = new DocumentRuntime({
+        config,
+        logger: options.logger,
+        ...(options.seams === undefined ? {} : { seams: options.seams }),
+        ...(options.now === undefined ? {} : { now: options.now }),
+    });
+    const disposeTools = registerDocumentTools({ register: options.register }, { runtime });
+    if (config.retention.enabled && config.storage.root !== null) {
+        const timer = setInterval(() => {
+            void runtime.cleanup().catch((error) => {
+                options.logger.warn("documents.retention.failed", {
+                    message: error instanceof Error ? error.message : String(error),
+                });
+            });
+        }, retentionIntervalMs(config));
+        timer.unref?.();
+        ctx.effect(() => () => clearInterval(timer), "dsh-qa-surface.documents-retention");
+    }
+    options.logger.info("documents.installed", {
+        tools: DOCUMENT_TOOL_NAMES,
+        storageRoot: config.storage.root,
+        templatesDefault: config.templates.default,
+        docling: config.docling.enabled ? config.docling.baseUrl : "disabled",
+    });
+    return {
+        runtime,
+        toolNames: DOCUMENT_TOOL_NAMES,
+        dispose: () => {
+            disposeTools();
+        },
+    };
+}
+//# sourceMappingURL=index.js.map

package/lib/documents/index.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.js","sourceRoot":"","sources":["../../src/documents/index.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;GAWG;AAEH,OAAO,EACL,oBAAoB,EACpB,aAAa,EACb,eAAe,EACf,qBAAqB,GAGtB,MAAM,aAAa,CAAC;AACrB,OAAO,EACL,2BAA2B,EAC3B,0BAA0B,EAC1B,sBAAsB,EACtB,mBAAmB,GAGpB,MAAM,aAAa,CAAC;AAErB,OAAO,EACL,qBAAqB,EACrB,eAAe,GAEhB,MAAM,cAAc,CAAC;AACtB,OAAO,EACL,mBAAmB,EACnB,qBAAqB,EACrB,mBAAmB,EACnB,qBAAqB,EACrB,oBAAoB,EACpB,qBAAqB,EACrB,yBAAyB,GAE1B,MAAM,kBAAkB,CAAC;AAC1B,OAAO,EACL,eAAe,GAGhB,MAAM,yBAAyB,CAAC;AACjC,OAAO,EAAE,oBAAoB,EAAE,cAAc,EAAE,MAAM,mBAAmB,CAAC;AACzE,OAAO,EAAE,aAAa,EAAE,SAAS,EAAE,YAAY,EAAE,MAAM,sBAAsB,CAAC;AAC9E,OAAO,EAAE,gBAAgB,EAAE,UAAU,EAAE,YAAY,EAAE,MAAM,oBAAoB,CAAC;AAChF,OAAO,EACL,aAAa,EACb,uBAAuB,EACvB,YAAY,EACZ,aAAa,GACd,MAAM,yBAAyB,CAAC;AACjC,OAAO,EACL,qBAAqB,EACrB,sBAAsB,EACtB,oBAAoB,EACpB,aAAa,GAGd,MAAM,yBAAyB,CAAC;AACjC,OAAO,EACL,eAAe,GAEhB,MAAM,yBAAyB,CAAC;AACjC,OAAO,EAAE,iBAAiB,EAAE,MAAM,oCAAoC,CAAC;AACvE,OAAO,EACL,mBAAmB,EACnB,oBAAoB,GAGrB,MAAM,yBAAyB,CAAC;AACjC,OAAO,EACL,oBAAoB,GAErB,MAAM,gCAAgC,CAAC;AACxC,OAAO,EACL,gBAAgB,EAChB,0BAA0B,EAC1B,YAAY,EACZ,iBAAiB,EACjB,gBAAgB,GACjB,MAAM,qBAAqB,CAAC;AAC7B,OAAO,EACL,wBAAwB,EACxB,uBAAuB,EACvB,mBAAmB,EACnB,aAAa,EACb,iBAAiB,GAElB,MAAM,0BAA0B,CAAC;AAClC,OAAO,EAAE,eAAe,EAAE,MAAM,0BAA0B,CAAC;AAC3D,OAAO,EAAE,0BAA0B,EAAE,MAAM,yBAAyB,CAAC;AACrE,OAAO,EAAE,gBAAgB,EAAE,MAAM,2BAA2B,CAAC;AAC7D,OAAO,EACL,YAAY,EACZ,aAAa,EACb,gBAAgB,GACjB,MAAM,oBAAoB,CAAC;AAC5B,OAAO,EACL,cAAc,EACd,YAAY,EACZ,kBAAkB,GACnB,MAAM,kBAAkB,CAAC;AAI1B,OAAO,EACL,sBAAsB,EACtB,mBAAmB,GAEpB,MAAM,aAAa,CAAC;AAGrB,OAAO,EAAE,eAAe,EAAE,MAAM,cAAc,CAAC;AAC/C,OAAO,EACL,qBAAqB,EACrB,mBAAmB,GAEpB,MAAM,kBAAkB,CAAC;AAkB1B;;;;;;GAMG;AACH,MAAM,UAAU,wBAAwB,CACtC,GAAY,EACZ,OAAwC;IAExC,MAAM,MAAM,GAAG,sBAAsB,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC;IACtD,IAAI,CAAC,MAAM,CAAC,OAAO;QAAE,OAAO,SAAS,CAAC;IAEtC,MAAM,OAAO,GAAG,IAAI,eAAe,CAAC;QAClC,MAAM;QACN,MAAM,EAAE,OAAO,CAAC,MAAM;QACtB,GAAG,CAAC,OAAO,CAAC,KAAK,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,KAAK,EAAE,OAAO,CAAC,KAAK,EAAE,CAAC;QAChE,GAAG,CAAC,OAAO,CAAC,GAAG,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,GAAG,EAAE,OAAO,CAAC,GAAG,EAAE,CAAC;KAC3D,CAAC,CAAC;IACH,MAAM,YAAY,GAAG,qBAAqB,CACxC,EAAE,QAAQ,EAAE,OAAO,CAAC,QAAQ,EAAE,EAC9B,EAAE,OAAO,EAAE,CACZ,CAAC;IAEF,IAAI,MAAM,CAAC,SAAS,CAAC,OAAO,IAAI,MAAM,CAAC,OAAO,CAAC,IAAI,KAAK,IAAI,EAAE,CAAC;QAC7D,MAAM,KAAK,GAAG,WAAW,CAAC,GAAG,EAAE;YAC7B,KAAK,OAAO,CAAC,OAAO,EAAE,CAAC,KAAK,CAAC,CAAC,KAAc,EAAE,EAAE;gBAC9C,OAAO,CAAC,MAAM,CAAC,IAAI,CAAC,4BAA4B,EAAE;oBAChD,OAAO,EAAE,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC;iBAChE,CAAC,CAAC;YACL,CAAC,CAAC,CAAC;QACL,CAAC,EAAE,mBAAmB,CAAC,MAAM,CAAC,CAAC,CAAC;QAChC,KAAK,CAAC,KAAK,EAAE,EAAE,CAAC;QAChB,GAAG,CAAC,MAAM,CACR,GAAG,EAAE,CAAC,GAAG,EAAE,CAAC,aAAa,CAAC,KAAK,CAAC,EAChC,oCAAoC,CACrC,CAAC;IACJ,CAAC;IAED,OAAO,CAAC,MAAM,CAAC,IAAI,CAAC,qBAAqB,EAAE;QACzC,KAAK,EAAE,mBAAmB;QAC1B,WAAW,EAAE,MAAM,CAAC,OAAO,CAAC,IAAI;QAChC,gBAAgB,EAAE,MAAM,CAAC,SAAS,CAAC,OAAO;QAC1C,OAAO,EAAE,MAAM,CAAC,OAAO,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC,OAAO,CAAC,CAAC,CAAC,UAAU;KACtE,CAAC,CAAC;IAEH,OAAO;QACL,OAAO;QACP,SAAS,EAAE,mBAAmB;QAC9B,OAAO,EAAE,GAAG,EAAE;YACZ,YAAY,EAAE,CAAC;QACjB,CAAC;KACF,CAAC;AACJ,CAAC"}

package/lib/documents/inspect/facts.js

@@ -0,0 +1,161 @@
+/**
+ * Document facts read without converting anything (§11).
+ *
+ * `document_inspect` must be cheap and safe: it reads the bytes it was given
+ * and answers what they claim to be. Two tiny readers cover the formats the
+ * pipeline supports — a PDF dictionary scan and an OOXML package scan — which
+ * is what makes inspect useful on a 400-page file that nobody wants to render.
+ */
+import { findZipEntry, readZipEntries, readZipEntry, } from "./zip.js";
+const PDF_DATE = /^D:(\d{4})(\d{2})(\d{2})(\d{2})?(\d{2})?(\d{2})?/u;
+function normalizePdfDate(raw) {
+    const match = PDF_DATE.exec(raw.trim());
+    if (match === null)
+        return undefined;
+    const [, year, month, day, hour = "00", minute = "00", second = "00"] = match;
+    return `${year}-${month}-${day}T${hour}:${minute}:${second}Z`;
+}
+function decodePdfString(raw) {
+    const literal = /^\((.*)\)$/su.exec(raw.trim());
+    const value = literal?.[1] ?? raw;
+    return value
+        .replace(/\\([nrtbf()\\])/gu, (_all, escape) => {
+        const map = {
+            n: "\n",
+            r: "\r",
+            t: "\t",
+            b: "\b",
+            f: "\f",
+        };
+        return map[escape] ?? escape;
+    })
+        .replace(/\\(\d{1,3})/gu, (_all, octal) => String.fromCharCode(Number.parseInt(octal, 8)))
+        .trim();
+}
+/**
+ * Read a PDF's page count, encryption flag and document-information entries.
+ * This is a scan, not a parser: it is immune to malformed cross-reference
+ * tables (a broken PDF still answers "how many pages does it claim"), and it
+ * never follows an object stream.
+ */
+export function readPdfFacts(buffer) {
+    const text = buffer.toString("latin1");
+    const tail = text.slice(Math.max(0, text.length - 131_072));
+    const encrypted = /\/Encrypt\b/u.test(tail) || /\/Encrypt\b/u.test(text.slice(0, 4_096));
+    const pageMatches = text.match(/\/Type\s*\/Page[^s]/gu);
+    const pages = pageMatches === null ? undefined : pageMatches.length;
+    // The trailer names the document-information object; its dictionary is read
+    // from the object itself, because the trailer's own dictionary holds no
+    // title or author.
+    const infoId = /\/Info\s+(\d+)\s+\d+\s+R/u.exec(tail)?.[1];
+    const info = infoId === undefined
+        ? undefined
+        : new RegExp(`\\b${infoId}\\s+\\d+\\s+obj\\s*<<([\\s\\S]*?)>>`, "u").exec(text)?.[1];
+    const readEntry = (key) => {
+        if (info === undefined)
+            return undefined;
+        const pattern = new RegExp(`/${key}\\s*(\\((?:[^()\\\\]|\\\\.)*\\)|<[0-9A-Fa-f\\s]+>)`, "u");
+        const match = pattern.exec(info);
+        if (match?.[1] === undefined)
+            return undefined;
+        const raw = match[1];
+        if (raw.startsWith("<")) {
+            const hex = raw.slice(1, -1).replace(/\s+/gu, "");
+            const bytes = Buffer.from(hex.length % 2 === 0 ? hex : `${hex}0`, "hex");
+            /* eslint-disable-next-line no-control-regex -- NUL padding is exactly what this strips */
+            const decoded = bytes.toString("utf16le").replace(/\u0000+$/u, "");
+            return decoded.trim() === "" ? undefined : decoded.trim();
+        }
+        const decoded = decodePdfString(raw);
+        return decoded === "" ? undefined : decoded;
+    };
+    const created = readEntry("CreationDate");
+    const modified = readEntry("ModDate");
+    const title = readEntry("Title");
+    const author = readEntry("Author");
+    const streamImages = (/\/Subtype\s*\/Image\b/gu.exec(text) ?? []).length;
+    return {
+        ...(pages === undefined || pages === 0 ? {} : { pages }),
+        encrypted,
+        metadata: {
+            ...(title === undefined ? {} : { title }),
+            ...(author === undefined ? {} : { author }),
+            ...(created === undefined
+                ? {}
+                : { createdAt: normalizePdfDate(created) ?? created }),
+            ...(modified === undefined
+                ? {}
+                : { modifiedAt: normalizePdfDate(modified) ?? modified }),
+        },
+        ...(streamImages > 0 ? { images: streamImages } : {}),
+    };
+}
+const DOCX_FACTS_MAX_PART_BYTES = 32 * 1024 * 1024;
+function readPart(buffer, entries, name) {
+    const entry = findZipEntry(entries, name);
+    if (entry === undefined ||
+        entry.uncompressedSize > DOCX_FACTS_MAX_PART_BYTES) {
+        return undefined;
+    }
+    const payload = readZipEntry(buffer, entry);
+    return payload?.toString("utf8");
+}
+/** Count document structure from the OOXML package, without rendering it. */
+export function readDocxFacts(buffer) {
+    const entries = readZipEntries(buffer);
+    if (entries === undefined)
+        return {};
+    const document = readPart(buffer, entries, "word/document.xml");
+    if (document === undefined)
+        return {};
+    const headingStyles = document.match(/<w:pStyle[^>]*w:val="(?:Heading|heading|Заголовок)[^"]*"/gu) ?? [];
+    const outlineLevels = document.match(/<w:outlineLvl\b/gu) ?? [];
+    const tables = (document.match(/<w:tbl>/gu) ?? []).length;
+    const images = (document.match(/<w:drawing>/gu) ?? []).length +
+        (document.match(/<w:pict>/gu) ?? []).length;
+    const mediaParts = entries.filter((entry) => /^word\/media\//iu.test(entry.name)).length;
+    return {
+        headings: headingStyles.length + outlineLevels.length,
+        tables,
+        images: Math.max(images, mediaParts),
+    };
+}
+/** Metadata recorded in `docProps/core.xml`. */
+export function readDocxMetadata(buffer) {
+    const entries = readZipEntries(buffer);
+    if (entries === undefined)
+        return {};
+    const core = readPart(buffer, entries, "docProps/core.xml");
+    if (core === undefined)
+        return {};
+    const read = (tag) => {
+        const match = new RegExp(`<(?:[A-Za-z]+:)?${tag}[^>]*>([\\s\\S]*?)</(?:[A-Za-z]+:)?${tag}>`, "u").exec(core);
+        const value = match?.[1]?.trim();
+        return value === undefined || value === "" ? undefined : value;
+    };
+    const title = read("title");
+    const author = read("creator");
+    const created = read("created");
+    const modified = read("modified");
+    return {
+        ...(title === undefined ? {} : { title }),
+        ...(author === undefined ? {} : { author }),
+        ...(created === undefined ? {} : { createdAt: created }),
+        ...(modified === undefined ? {} : { modifiedAt: modified }),
+    };
+}
+export function mediaTypeOfDetected(format) {
+    switch (format) {
+        case "pdf":
+            return "application/pdf";
+        case "docx":
+            return "application/vnd.openxmlformats-officedocument.wordprocessingml.document";
+        case "docm":
+            return "application/vnd.ms-word.document.macroEnabled.12";
+        case "markdown":
+            return "text/markdown";
+        default:
+            return "application/octet-stream";
+    }
+}
+//# sourceMappingURL=facts.js.map

package/lib/documents/inspect/facts.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"facts.js","sourceRoot":"","sources":["../../../src/documents/inspect/facts.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AAGH,OAAO,EACL,YAAY,EACZ,cAAc,EACd,YAAY,GAEb,MAAM,UAAU,CAAC;AAclB,MAAM,QAAQ,GAAG,mDAAmD,CAAC;AAErE,SAAS,gBAAgB,CAAC,GAAW;IACnC,MAAM,KAAK,GAAG,QAAQ,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC,CAAC;IACxC,IAAI,KAAK,KAAK,IAAI;QAAE,OAAO,SAAS,CAAC;IACrC,MAAM,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE,GAAG,EAAE,IAAI,GAAG,IAAI,EAAE,MAAM,GAAG,IAAI,EAAE,MAAM,GAAG,IAAI,CAAC,GAAG,KAAK,CAAC;IAC9E,OAAO,GAAG,IAAI,IAAI,KAAK,IAAI,GAAG,IAAI,IAAI,IAAI,MAAM,IAAI,MAAM,GAAG,CAAC;AAChE,CAAC;AAED,SAAS,eAAe,CAAC,GAAW;IAClC,MAAM,OAAO,GAAG,cAAc,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC,CAAC;IAChD,MAAM,KAAK,GAAG,OAAO,EAAE,CAAC,CAAC,CAAC,IAAI,GAAG,CAAC;IAClC,OAAO,KAAK;SACT,OAAO,CAAC,mBAAmB,EAAE,CAAC,IAAI,EAAE,MAAc,EAAE,EAAE;QACrD,MAAM,GAAG,GAA2B;YAClC,CAAC,EAAE,IAAI;YACP,CAAC,EAAE,IAAI;YACP,CAAC,EAAE,IAAI;YACP,CAAC,EAAE,IAAI;YACP,CAAC,EAAE,IAAI;SACR,CAAC;QACF,OAAO,GAAG,CAAC,MAAM,CAAC,IAAI,MAAM,CAAC;IAC/B,CAAC,CAAC;SACD,OAAO,CAAC,eAAe,EAAE,CAAC,IAAI,EAAE,KAAa,EAAE,EAAE,CAChD,MAAM,CAAC,YAAY,CAAC,MAAM,CAAC,QAAQ,CAAC,KAAK,EAAE,CAAC,CAAC,CAAC,CAC/C;SACA,IAAI,EAAE,CAAC;AACZ,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,YAAY,CAAC,MAAc;IACzC,MAAM,IAAI,GAAG,MAAM,CAAC,QAAQ,CAAC,QAAQ,CAAC,CAAC;IACvC,MAAM,IAAI,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,MAAM,GAAG,OAAO,CAAC,CAAC,CAAC;IAC5D,MAAM,SAAS,GACb,cAAc,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,cAAc,CAAC,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,KAAK,CAAC,CAAC,CAAC;IAEzE,MAAM,WAAW,GAAG,IAAI,CAAC,KAAK,CAAC,uBAAuB,CAAC,CAAC;IACxD,MAAM,KAAK,GAAG,WAAW,KAAK,IAAI,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,WAAW,CAAC,MAAM,CAAC;IAEpE,4EAA4E;IAC5E,wEAAwE;IACxE,mBAAmB;IACnB,MAAM,MAAM,GAAG,2BAA2B,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC;IAC3D,MAAM,IAAI,GACR,MAAM,KAAK,SAAS;QAClB,CAAC,CAAC,SAAS;QACX,CAAC,CAAC,IAAI,MAAM,CAAC,MAAM,MAAM,qCAAqC,EAAE,GAAG,CAAC,CAAC,IAAI,CACrE,IAAI,CACL,EAAE,CAAC,CAAC,CAAC,CAAC;IACb,MAAM,SAAS,GAAG,CAAC,GAAW,EAAsB,EAAE;QACpD,IAAI,IAAI,KAAK,SAAS;YAAE,OAAO,SAAS,CAAC;QACzC,MAAM,OAAO,GAAG,IAAI,MAAM,CACxB,IAAI,GAAG,oDAAoD,EAC3D,GAAG,CACJ,CAAC;QACF,MAAM,KAAK,GAAG,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QACjC,IAAI,KAAK,EAAE,CAAC,CAAC,CAAC,KAAK,SAAS;YAAE,OAAO,SAAS,CAAC;QAC/C,MAAM,GAAG,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC;QACrB,IAAI,GAAG,CAAC,UAAU,CAAC,GAAG,CAAC,EAAE,CAAC;YACxB,MAAM,GAAG,GAAG,GAAG,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,OAAO,EAAE,EAAE,CAAC,CAAC;YAClD,MAAM,KAAK,GAAG,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,MAAM,GAAG,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,GAAG,GAAG,GAAG,EAAE,KAAK,CAAC,CAAC;YACzE,0FAA0F;YAC1F,MAAM,OAAO,GAAG,KAAK,CAAC,QAAQ,CAAC,SAAS,CAAC,CAAC,OAAO,CAAC,WAAW,EAAE,EAAE,CAAC,CAAC;YACnE,OAAO,OAAO,CAAC,IAAI,EAAE,KAAK,EAAE,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC;QAC5D,CAAC;QACD,MAAM,OAAO,GAAG,eAAe,CAAC,GAAG,CAAC,CAAC;QACrC,OAAO,OAAO,KAAK,EAAE,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,OAAO,CAAC;IAC9C,CAAC,CAAC;IAEF,MAAM,OAAO,GAAG,SAAS,CAAC,cAAc,CAAC,CAAC;IAC1C,MAAM,QAAQ,GAAG,SAAS,CAAC,SAAS,CAAC,CAAC;IACtC,MAAM,KAAK,GAAG,SAAS,CAAC,OAAO,CAAC,CAAC;IACjC,MAAM,MAAM,GAAG,SAAS,CAAC,QAAQ,CAAC,CAAC;IACnC,MAAM,YAAY,GAAG,CAAC,yBAAyB,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC,CAAC,MAAM,CAAC;IACzE,OAAO;QACL,GAAG,CAAC,KAAK,KAAK,SAAS,IAAI,KAAK,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,KAAK,EAAE,CAAC;QACxD,SAAS;QACT,QAAQ,EAAE;YACR,GAAG,CAAC,KAAK,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,KAAK,EAAE,CAAC;YACzC,GAAG,CAAC,MAAM,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,MAAM,EAAE,CAAC;YAC3C,GAAG,CAAC,OAAO,KAAK,SAAS;gBACvB,CAAC,CAAC,EAAE;gBACJ,CAAC,CAAC,EAAE,SAAS,EAAE,gBAAgB,CAAC,OAAO,CAAC,IAAI,OAAO,EAAE,CAAC;YACxD,GAAG,CAAC,QAAQ,KAAK,SAAS;gBACxB,CAAC,CAAC,EAAE;gBACJ,CAAC,CAAC,EAAE,UAAU,EAAE,gBAAgB,CAAC,QAAQ,CAAC,IAAI,QAAQ,EAAE,CAAC;SAC5D;QACD,GAAG,CAAC,YAAY,GAAG,CAAC,CAAC,CAAC,CAAC,EAAE,MAAM,EAAE,YAAY,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;KACtD,CAAC;AACJ,CAAC;AAQD,MAAM,yBAAyB,GAAG,EAAE,GAAG,IAAI,GAAG,IAAI,CAAC;AAEnD,SAAS,QAAQ,CACf,MAAc,EACd,OAA4B,EAC5B,IAAY;IAEZ,MAAM,KAAK,GAAG,YAAY,CAAC,OAAO,EAAE,IAAI,CAAC,CAAC;IAC1C,IACE,KAAK,KAAK,SAAS;QACnB,KAAK,CAAC,gBAAgB,GAAG,yBAAyB,EAClD,CAAC;QACD,OAAO,SAAS,CAAC;IACnB,CAAC;IACD,MAAM,OAAO,GAAG,YAAY,CAAC,MAAM,EAAE,KAAK,CAAC,CAAC;IAC5C,OAAO,OAAO,EAAE,QAAQ,CAAC,MAAM,CAAC,CAAC;AACnC,CAAC;AAED,6EAA6E;AAC7E,MAAM,UAAU,aAAa,CAAC,MAAc;IAC1C,MAAM,OAAO,GAAG,cAAc,CAAC,MAAM,CAAC,CAAC;IACvC,IAAI,OAAO,KAAK,SAAS;QAAE,OAAO,EAAE,CAAC;IACrC,MAAM,QAAQ,GAAG,QAAQ,CAAC,MAAM,EAAE,OAAO,EAAE,mBAAmB,CAAC,CAAC;IAChE,IAAI,QAAQ,KAAK,SAAS;QAAE,OAAO,EAAE,CAAC;IACtC,MAAM,aAAa,GACjB,QAAQ,CAAC,KAAK,CACZ,4DAA4D,CAC7D,IAAI,EAAE,CAAC;IACV,MAAM,aAAa,GAAG,QAAQ,CAAC,KAAK,CAAC,mBAAmB,CAAC,IAAI,EAAE,CAAC;IAChE,MAAM,MAAM,GAAG,CAAC,QAAQ,CAAC,KAAK,CAAC,WAAW,CAAC,IAAI,EAAE,CAAC,CAAC,MAAM,CAAC;IAC1D,MAAM,MAAM,GACV,CAAC,QAAQ,CAAC,KAAK,CAAC,eAAe,CAAC,IAAI,EAAE,CAAC,CAAC,MAAM;QAC9C,CAAC,QAAQ,CAAC,KAAK,CAAC,YAAY,CAAC,IAAI,EAAE,CAAC,CAAC,MAAM,CAAC;IAC9C,MAAM,UAAU,GAAG,OAAO,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,CAC1C,kBAAkB,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CACpC,CAAC,MAAM,CAAC;IACT,OAAO;QACL,QAAQ,EAAE,aAAa,CAAC,MAAM,GAAG,aAAa,CAAC,MAAM;QACrD,MAAM;QACN,MAAM,EAAE,IAAI,CAAC,GAAG,CAAC,MAAM,EAAE,UAAU,CAAC;KACrC,CAAC;AACJ,CAAC;AAED,gDAAgD;AAChD,MAAM,UAAU,gBAAgB,CAAC,MAAc;IAM7C,MAAM,OAAO,GAAG,cAAc,CAAC,MAAM,CAAC,CAAC;IACvC,IAAI,OAAO,KAAK,SAAS;QAAE,OAAO,EAAE,CAAC;IACrC,MAAM,IAAI,GAAG,QAAQ,CAAC,MAAM,EAAE,OAAO,EAAE,mBAAmB,CAAC,CAAC;IAC5D,IAAI,IAAI,KAAK,SAAS;QAAE,OAAO,EAAE,CAAC;IAClC,MAAM,IAAI,GAAG,CAAC,GAAW,EAAsB,EAAE;QAC/C,MAAM,KAAK,GAAG,IAAI,MAAM,CACtB,mBAAmB,GAAG,sCAAsC,GAAG,GAAG,EAClE,GAAG,CACJ,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QACb,MAAM,KAAK,GAAG,KAAK,EAAE,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,CAAC;QACjC,OAAO,KAAK,KAAK,SAAS,IAAI,KAAK,KAAK,EAAE,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,KAAK,CAAC;IACjE,CAAC,CAAC;IACF,MAAM,KAAK,GAAG,IAAI,CAAC,OAAO,CAAC,CAAC;IAC5B,MAAM,MAAM,GAAG,IAAI,CAAC,SAAS,CAAC,CAAC;IAC/B,MAAM,OAAO,GAAG,IAAI,CAAC,SAAS,CAAC,CAAC;IAChC,MAAM,QAAQ,GAAG,IAAI,CAAC,UAAU,CAAC,CAAC;IAClC,OAAO;QACL,GAAG,CAAC,KAAK,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,KAAK,EAAE,CAAC;QACzC,GAAG,CAAC,MAAM,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,MAAM,EAAE,CAAC;QAC3C,GAAG,CAAC,OAAO,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,SAAS,EAAE,OAAO,EAAE,CAAC;QACxD,GAAG,CAAC,QAAQ,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,UAAU,EAAE,QAAQ,EAAE,CAAC;KAC5D,CAAC;AACJ,CAAC;AAED,MAAM,UAAU,mBAAmB,CAAC,MAAsB;IACxD,QAAQ,MAAM,EAAE,CAAC;QACf,KAAK,KAAK;YACR,OAAO,iBAAiB,CAAC;QAC3B,KAAK,MAAM;YACT,OAAO,yEAAyE,CAAC;QACnF,KAAK,MAAM;YACT,OAAO,kDAAkD,CAAC;QAC5D,KAAK,UAAU;YACb,OAAO,eAAe,CAAC;QACzB;YACE,OAAO,0BAA0B,CAAC;IACtC,CAAC;AACH,CAAC"}

package/lib/documents/inspect/zip.js

@@ -0,0 +1,108 @@
+/**
+ * Minimal ZIP reader (central directory + per-entry inflate).
+ *
+ * The document pipeline inspects OOXML containers — `document_inspect` counts
+ * headings and images, `security/file-types.ts` decides DOCX vs macro-enabled
+ * DOCM — and neither may pull a ZIP dependency into the host plugin. Only the
+ * two compression methods OOXML writers actually produce are supported
+ * (stored and deflate); anything else is reported as unreadable instead of
+ * being guessed at.
+ */
+import { inflateRawSync } from "node:zlib";
+const EOCD_SIGNATURE = 0x06054b50;
+const CENTRAL_SIGNATURE = 0x02014b50;
+const LOCAL_SIGNATURE = 0x04034b50;
+/** The end-of-central-directory record sits in the last 64 KiB (plus comment). */
+const EOCD_SEARCH_BYTES = 65_557;
+/** Guard against absurd declared sizes in a hostile archive. */
+const MAX_ENTRY_BYTES = 64 * 1024 * 1024;
+function findEndOfCentralDirectory(buffer) {
+    const start = Math.max(0, buffer.length - EOCD_SEARCH_BYTES);
+    for (let offset = buffer.length - 22; offset >= start; offset -= 1) {
+        if (buffer.readUInt32LE(offset) === EOCD_SIGNATURE)
+            return offset;
+    }
+    return -1;
+}
+/**
+ * Parse the central directory. Returns `undefined` when the buffer is not a
+ * ZIP archive (or is truncated) — the caller decides whether that is an error.
+ */
+export function readZipEntries(buffer) {
+    if (buffer.length < 22)
+        return undefined;
+    const eocd = findEndOfCentralDirectory(buffer);
+    if (eocd < 0)
+        return undefined;
+    const count = buffer.readUInt16LE(eocd + 10);
+    const centralOffset = buffer.readUInt32LE(eocd + 16);
+    if (centralOffset === 0xffffffff || count === 0xffff)
+        return undefined;
+    const entries = [];
+    let cursor = centralOffset;
+    for (let index = 0; index < count; index += 1) {
+        if (cursor + 46 > buffer.length)
+            return undefined;
+        if (buffer.readUInt32LE(cursor) !== CENTRAL_SIGNATURE)
+            return undefined;
+        const nameLength = buffer.readUInt16LE(cursor + 28);
+        const extraLength = buffer.readUInt16LE(cursor + 30);
+        const commentLength = buffer.readUInt16LE(cursor + 32);
+        const nameStart = cursor + 46;
+        if (nameStart + nameLength > buffer.length)
+            return undefined;
+        entries.push({
+            name: buffer.toString("utf8", nameStart, nameStart + nameLength),
+            compressionMethod: buffer.readUInt16LE(cursor + 10),
+            compressedSize: buffer.readUInt32LE(cursor + 20),
+            uncompressedSize: buffer.readUInt32LE(cursor + 24),
+            localHeaderOffset: buffer.readUInt32LE(cursor + 42),
+        });
+        cursor = nameStart + nameLength + extraLength + commentLength;
+    }
+    return entries;
+}
+export function findZipEntry(entries, name) {
+    const wanted = name.toLowerCase();
+    return entries.find((entry) => entry.name.toLowerCase() === wanted);
+}
+/** Decompress one entry, or `undefined` when it is unreadable/oversized. */
+export function readZipEntry(buffer, entry) {
+    if (entry.compressedSize > MAX_ENTRY_BYTES ||
+        entry.uncompressedSize > MAX_ENTRY_BYTES) {
+        return undefined;
+    }
+    const offset = entry.localHeaderOffset;
+    if (offset + 30 > buffer.length)
+        return undefined;
+    if (buffer.readUInt32LE(offset) !== LOCAL_SIGNATURE)
+        return undefined;
+    const nameLength = buffer.readUInt16LE(offset + 26);
+    const extraLength = buffer.readUInt16LE(offset + 28);
+    const start = offset + 30 + nameLength + extraLength;
+    const end = start + entry.compressedSize;
+    if (end > buffer.length)
+        return undefined;
+    const payload = buffer.subarray(start, end);
+    try {
+        if (entry.compressionMethod === 0)
+            return Buffer.from(payload);
+        if (entry.compressionMethod === 8)
+            return inflateRawSync(payload);
+        return undefined;
+    }
+    catch {
+        return undefined;
+    }
+}
+/** Convenience: read a named entry from an archive in one call. */
+export function readZipEntryByName(buffer, name) {
+    const entries = readZipEntries(buffer);
+    if (entries === undefined)
+        return undefined;
+    const entry = findZipEntry(entries, name);
+    if (entry === undefined)
+        return undefined;
+    return readZipEntry(buffer, entry);
+}
+//# sourceMappingURL=zip.js.map

package/lib/documents/inspect/zip.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"zip.js","sourceRoot":"","sources":["../../../src/documents/inspect/zip.ts"],"names":[],"mappings":"AAAA;;;;;;;;;GASG;AAEH,OAAO,EAAE,cAAc,EAAE,MAAM,WAAW,CAAC;AAE3C,MAAM,cAAc,GAAG,UAAU,CAAC;AAClC,MAAM,iBAAiB,GAAG,UAAU,CAAC;AACrC,MAAM,eAAe,GAAG,UAAU,CAAC;AACnC,kFAAkF;AAClF,MAAM,iBAAiB,GAAG,MAAM,CAAC;AACjC,gEAAgE;AAChE,MAAM,eAAe,GAAG,EAAE,GAAG,IAAI,GAAG,IAAI,CAAC;AAUzC,SAAS,yBAAyB,CAAC,MAAc;IAC/C,MAAM,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,MAAM,CAAC,MAAM,GAAG,iBAAiB,CAAC,CAAC;IAC7D,KAAK,IAAI,MAAM,GAAG,MAAM,CAAC,MAAM,GAAG,EAAE,EAAE,MAAM,IAAI,KAAK,EAAE,MAAM,IAAI,CAAC,EAAE,CAAC;QACnE,IAAI,MAAM,CAAC,YAAY,CAAC,MAAM,CAAC,KAAK,cAAc;YAAE,OAAO,MAAM,CAAC;IACpE,CAAC;IACD,OAAO,CAAC,CAAC,CAAC;AACZ,CAAC;AAED;;;GAGG;AACH,MAAM,UAAU,cAAc,CAAC,MAAc;IAC3C,IAAI,MAAM,CAAC,MAAM,GAAG,EAAE;QAAE,OAAO,SAAS,CAAC;IACzC,MAAM,IAAI,GAAG,yBAAyB,CAAC,MAAM,CAAC,CAAC;IAC/C,IAAI,IAAI,GAAG,CAAC;QAAE,OAAO,SAAS,CAAC;IAC/B,MAAM,KAAK,GAAG,MAAM,CAAC,YAAY,CAAC,IAAI,GAAG,EAAE,CAAC,CAAC;IAC7C,MAAM,aAAa,GAAG,MAAM,CAAC,YAAY,CAAC,IAAI,GAAG,EAAE,CAAC,CAAC;IACrD,IAAI,aAAa,KAAK,UAAU,IAAI,KAAK,KAAK,MAAM;QAAE,OAAO,SAAS,CAAC;IACvE,MAAM,OAAO,GAAe,EAAE,CAAC;IAC/B,IAAI,MAAM,GAAG,aAAa,CAAC;IAC3B,KAAK,IAAI,KAAK,GAAG,CAAC,EAAE,KAAK,GAAG,KAAK,EAAE,KAAK,IAAI,CAAC,EAAE,CAAC;QAC9C,IAAI,MAAM,GAAG,EAAE,GAAG,MAAM,CAAC,MAAM;YAAE,OAAO,SAAS,CAAC;QAClD,IAAI,MAAM,CAAC,YAAY,CAAC,MAAM,CAAC,KAAK,iBAAiB;YAAE,OAAO,SAAS,CAAC;QACxE,MAAM,UAAU,GAAG,MAAM,CAAC,YAAY,CAAC,MAAM,GAAG,EAAE,CAAC,CAAC;QACpD,MAAM,WAAW,GAAG,MAAM,CAAC,YAAY,CAAC,MAAM,GAAG,EAAE,CAAC,CAAC;QACrD,MAAM,aAAa,GAAG,MAAM,CAAC,YAAY,CAAC,MAAM,GAAG,EAAE,CAAC,CAAC;QACvD,MAAM,SAAS,GAAG,MAAM,GAAG,EAAE,CAAC;QAC9B,IAAI,SAAS,GAAG,UAAU,GAAG,MAAM,CAAC,MAAM;YAAE,OAAO,SAAS,CAAC;QAC7D,OAAO,CAAC,IAAI,CAAC;YACX,IAAI,EAAE,MAAM,CAAC,QAAQ,CAAC,MAAM,EAAE,SAAS,EAAE,SAAS,GAAG,UAAU,CAAC;YAChE,iBAAiB,EAAE,MAAM,CAAC,YAAY,CAAC,MAAM,GAAG,EAAE,CAAC;YACnD,cAAc,EAAE,MAAM,CAAC,YAAY,CAAC,MAAM,GAAG,EAAE,CAAC;YAChD,gBAAgB,EAAE,MAAM,CAAC,YAAY,CAAC,MAAM,GAAG,EAAE,CAAC;YAClD,iBAAiB,EAAE,MAAM,CAAC,YAAY,CAAC,MAAM,GAAG,EAAE,CAAC;SACpD,CAAC,CAAC;QACH,MAAM,GAAG,SAAS,GAAG,UAAU,GAAG,WAAW,GAAG,aAAa,CAAC;IAChE,CAAC;IACD,OAAO,OAAO,CAAC;AACjB,CAAC;AAED,MAAM,UAAU,YAAY,CAC1B,OAA4B,EAC5B,IAAY;IAEZ,MAAM,MAAM,GAAG,IAAI,CAAC,WAAW,EAAE,CAAC;IAClC,OAAO,OAAO,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,IAAI,CAAC,WAAW,EAAE,KAAK,MAAM,CAAC,CAAC;AACtE,CAAC;AAED,4EAA4E;AAC5E,MAAM,UAAU,YAAY,CAC1B,MAAc,EACd,KAAe;IAEf,IACE,KAAK,CAAC,cAAc,GAAG,eAAe;QACtC,KAAK,CAAC,gBAAgB,GAAG,eAAe,EACxC,CAAC;QACD,OAAO,SAAS,CAAC;IACnB,CAAC;IACD,MAAM,MAAM,GAAG,KAAK,CAAC,iBAAiB,CAAC;IACvC,IAAI,MAAM,GAAG,EAAE,GAAG,MAAM,CAAC,MAAM;QAAE,OAAO,SAAS,CAAC;IAClD,IAAI,MAAM,CAAC,YAAY,CAAC,MAAM,CAAC,KAAK,eAAe;QAAE,OAAO,SAAS,CAAC;IACtE,MAAM,UAAU,GAAG,MAAM,CAAC,YAAY,CAAC,MAAM,GAAG,EAAE,CAAC,CAAC;IACpD,MAAM,WAAW,GAAG,MAAM,CAAC,YAAY,CAAC,MAAM,GAAG,EAAE,CAAC,CAAC;IACrD,MAAM,KAAK,GAAG,MAAM,GAAG,EAAE,GAAG,UAAU,GAAG,WAAW,CAAC;IACrD,MAAM,GAAG,GAAG,KAAK,GAAG,KAAK,CAAC,cAAc,CAAC;IACzC,IAAI,GAAG,GAAG,MAAM,CAAC,MAAM;QAAE,OAAO,SAAS,CAAC;IAC1C,MAAM,OAAO,GAAG,MAAM,CAAC,QAAQ,CAAC,KAAK,EAAE,GAAG,CAAC,CAAC;IAC5C,IAAI,CAAC;QACH,IAAI,KAAK,CAAC,iBAAiB,KAAK,CAAC;YAAE,OAAO,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;QAC/D,IAAI,KAAK,CAAC,iBAAiB,KAAK,CAAC;YAAE,OAAO,cAAc,CAAC,OAAO,CAAC,CAAC;QAClE,OAAO,SAAS,CAAC;IACnB,CAAC;IAAC,MAAM,CAAC;QACP,OAAO,SAAS,CAAC;IACnB,CAAC;AACH,CAAC;AAED,mEAAmE;AACnE,MAAM,UAAU,kBAAkB,CAChC,MAAc,EACd,IAAY;IAEZ,MAAM,OAAO,GAAG,cAAc,CAAC,MAAM,CAAC,CAAC;IACvC,IAAI,OAAO,KAAK,SAAS;QAAE,OAAO,SAAS,CAAC;IAC5C,MAAM,KAAK,GAAG,YAAY,CAAC,OAAO,EAAE,IAAI,CAAC,CAAC;IAC1C,IAAI,KAAK,KAAK,SAAS;QAAE,OAAO,SAAS,CAAC;IAC1C,OAAO,YAAY,CAAC,MAAM,EAAE,KAAK,CAAC,CAAC;AACrC,CAAC"}