@cairn-tool/cairn 3.1.0 → 3.3.0

package/dist/pdf/types.d.ts

@@ -0,0 +1,205 @@
+import type { DiagnosticSeverity, MappingQuality } from "../mapping-quality.js";
+export type { MappingQuality } from "../mapping-quality.js";
+/**
+ * A PDF finding.
+ *
+ * The third finding family, after `AB###` for agent bundles and `AD###` for ADF
+ * conversion. It is structurally a `ConversionDiagnostic` with `node` and
+ * `location` replaced by `page` and `construct`, because a PDF finding is
+ * positioned by page rather than by a node ancestry trail — stuffing a page
+ * number into a field documented as "slash-joined ancestor node types" would be
+ * worse than a third shape. `quality` and the quality-to-severity rule are
+ * shared with the other two through `src/mapping-quality.ts`.
+ */
+export interface PdfDiagnostic {
+    code: string;
+    severity: DiagnosticSeverity;
+    message: string;
+    quality: MappingQuality;
+    /** 1-based page the finding concerns, when it concerns one page. */
+    page?: number;
+    /** The PDF construct: a structure role, a filter name, a font name. */
+    construct?: string;
+    remediation?: string;
+}
+export type PdfCommand = "inspect" | "text" | "outline" | "validate" | "to-markdown" | "attachments" | "forms";
+/** How much text a page carries, relative to its area. */
+export type TextLayer = "present" | "sparse" | "absent";
+/**
+ * Document-level facts, present on every command that opened the document.
+ *
+ * A deliberate divergence from `AdfResult`, where `inventory` is `inspect`-only:
+ * a conversion from a tagged document and one from a scan are not the same kind
+ * of artifact and must not look identical in the payload. `jq .document.tagged`
+ * has to work on a `to-markdown` result.
+ */
+export interface PdfDocumentInfo {
+    pageCount: number;
+    /**
+     * The document declares `/MarkInfo <</Marked true>>`.
+     *
+     * Not the same question as whether a usable structure tree exists — plenty of
+     * producers set this and ship an empty tree — which is what `structured`
+     * answers by probing.
+     */
+    tagged: boolean;
+    /**
+     * Whether any page yielded a structure tree with content.
+     *
+     * Optional because only the commands that walk pages can answer it; `outline`
+     * never opens one. Omitted rather than defaulted, on the rule that a field
+     * nobody measured must not read as a measurement.
+     */
+    structured?: "struct" | "partial" | "none";
+    encrypted: boolean;
+    /** Omitted for the same reason as `structured`. */
+    textLayer?: TextLayer;
+    pdfVersion?: string;
+    title?: string;
+    author?: string;
+    subject?: string;
+    keywords?: string;
+    creator?: string;
+    producer?: string;
+    /** ISO-8601, when the document's own `D:` string parsed. Omitted, never guessed. */
+    created?: string;
+    modified?: string;
+}
+export interface PdfPageInfo {
+    page: number;
+    /** Points, 72 to the inch, with `/Rotate` already applied. */
+    width: number;
+    height: number;
+    rotation: number;
+    /** Non-whitespace code points on the page: the evidence behind `textLayer`. */
+    characters: number;
+    /** Characters per square inch, rounded to 2 dp. The other half of the evidence. */
+    density: number;
+    textLayer: TextLayer;
+}
+export interface PdfTextPage {
+    page: number;
+    text: string;
+    characters: number;
+}
+export interface PdfOutlineEntry {
+    title: string;
+    /** 1-based nesting depth. */
+    level: number;
+    /** 1-based destination page, or null when it did not resolve. */
+    page: number | null;
+    /** Present only when pdf.js validated the URL's scheme. */
+    url?: string;
+    children: PdfOutlineEntry[];
+}
+/** One embedded file. Emitted by `pdf attachments`. */
+export interface PdfAttachment {
+    /** The name-tree key: the identifier the content lookup takes. */
+    id: string;
+    /**
+     * The basename pdf.js derived. Never used as a path without re-checking —
+     * the sanitization this command applies is its own, not pdf.js's.
+     */
+    filename: string;
+    /** The stored name verbatim, including any traversal it carries. */
+    rawFilename: string;
+    description?: string;
+    /** Byte length of the decoded file. Absent when the content could not be read. */
+    bytes?: number;
+    sha256?: string;
+    /** Executable format, when the magic bytes say so. */
+    binary?: "elf" | "pe" | "macho";
+    /** Absolute path written. Present only under `--extract`. */
+    written?: string;
+}
+/** One AcroForm field. Emitted by `pdf forms`. */
+export interface PdfFormField {
+    /** Fully-qualified field name. */
+    name: string;
+    type: string;
+    /** 1-based, converted from the 0-based index pdf.js reports. */
+    page: number | null;
+    value?: string;
+    defaultValue?: string;
+    readOnly: boolean;
+    hidden: boolean;
+    /**
+     * The field's password flag. The value is still reported: the same bytes are
+     * reachable through `pdf text`, so withholding them would be theatre.
+     */
+    password: boolean;
+    charLimit?: number;
+    exportValues?: string;
+    /** Widgets this field renders as; one field can appear on several pages. */
+    widgets: number;
+}
+export interface PdfForm {
+    /** `xfa` means the values live in an XML packet this does not read. */
+    type: "acroform" | "xfa" | "none";
+    fieldCount: number;
+    fields: PdfFormField[];
+}
+export interface PdfResult {
+    command: PdfCommand;
+    ok: boolean;
+    /** Input path, or `-` for stdin. */
+    source: string;
+    document?: PdfDocumentInfo;
+    /** inspect */
+    pages?: PdfPageInfo[];
+    /** text */
+    text?: PdfTextPage[];
+    /** outline */
+    outline?: PdfOutlineEntry[];
+    /** to-markdown */
+    markdown?: string;
+    /** attachments */
+    attachments?: PdfAttachment[];
+    /** forms */
+    form?: PdfForm;
+    /** text, to-markdown — present only when `--pages` narrowed the document. */
+    selectedPages?: number[];
+    /** Where `--output` wrote. */
+    output?: string;
+    diagnostics: PdfDiagnostic[];
+}
+export type BlockKind = "heading" | "paragraph" | "list" | "listItem" | "table" | "tableRow" | "tableCell" | "code" | "blockquote" | "figure" | "caption";
+export interface InlineSpan {
+    text: string;
+    bold: boolean;
+    italic: boolean;
+    code: boolean;
+}
+export interface BlockProvenance {
+    path: "struct" | "geometric";
+    /** Per block, not per document: one conversion mixes exact and approximate. */
+    quality: MappingQuality;
+    /** Pages the block's runs came from. A stitched paragraph has two. */
+    pages: number[];
+    /** The structure-tree role, when `path` is `"struct"`. */
+    role?: string;
+}
+/**
+ * One block of converted content.
+ *
+ * `spans` and `children` are both always present and mutually exclusive by
+ * kind, rather than a discriminated union, for the reason `AdfNode` is
+ * structural: a new kind must be able to reach a diagnostic instead of failing
+ * to compile at forty call sites.
+ *
+ * A flattened table is a run of sibling `paragraph` blocks — never a `table`
+ * with one cell. The emitter must be structurally unable to produce a table
+ * that was not one.
+ */
+export interface Block {
+    kind: BlockKind;
+    /** heading only, 1..6. */
+    level?: number;
+    /** list only. */
+    ordered?: boolean;
+    /** tableCell only. */
+    header?: boolean;
+    spans: InlineSpan[];
+    children: Block[];
+    provenance: BlockProvenance;
+}

package/dist/pdf/types.js

@@ -0,0 +1,2 @@
+export {};
+//# sourceMappingURL=types.js.map

package/dist/pdf/types.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"types.js","sourceRoot":"","sources":["../../src/pdf/types.ts"],"names":[],"mappings":""}

package/dist/pdf/validate.d.ts

@@ -0,0 +1,51 @@
+import type { OpenDocument } from "./document.js";
+import type { PdfDiagnostic } from "./types.js";
+/**
+ * Structural validation, bounded by what pdf.js can actually see.
+ *
+ * Deliberately **not** a PDF/A, PDF/UA, or PDF/X conformance checker. Full
+ * conformance validation is veraPDF's job and is a Java program; claiming it
+ * here would be a lie. This is the same line `jira adf validate` draws when it
+ * reports `AD100` for a node type it does not model rather than pretending to be
+ * Atlassian's schema.
+ *
+ * Specifically **not** checked, and the docs say so in these words: PDF/A or
+ * PDF/UA conformance at any level; digital signature validity, certificate
+ * chains, or timestamps; byte-level spec conformance such as `/Length`
+ * mismatches or trailing garbage, which pdf.js repairs below the warning
+ * threshold; whether the document *renders* correctly, since nothing here
+ * rasterizes; whether a glyph in the text layer actually exists in the embedded
+ * subset, which is only discoverable at render time; colour spaces, transparency
+ * groups, and output intents; and accessibility beyond "a structure tree
+ * exists".
+ */
+export interface WarningPattern {
+    pattern: RegExp;
+    code: string;
+    message: string;
+    quality: "approximate" | "unsupported";
+}
+/**
+ * pdf.js warnings this command reports as findings.
+ *
+ * **The most fragile coupling in the toolset**: these exist only because pdf.js
+ * prints a particular English string. That is contained rather than accepted —
+ * `tests/unit/pdf-validate.test.ts` drives one case per row from a deliberately
+ * damaged fixture, so a pdfjs bump that rewords a message fails the suite
+ * instead of quietly turning the check off.
+ *
+ * An unmatched warning is **not** dropped. It is reported once as `AP120` with
+ * the raw text, so a reworded message degrades to "something was recovered here"
+ * rather than to silence.
+ */
+export declare const WARNING_PATTERNS: readonly WarningPattern[];
+export interface ValidateResult {
+    diagnostics: PdfDiagnostic[];
+}
+/**
+ * Walks every page and reports what the parser could not do.
+ *
+ * No `--pages`: a partial validation that reported "valid" would be a lie, so
+ * the command always covers the whole document.
+ */
+export declare function validateDocument(handle: OpenDocument): Promise<ValidateResult>;

package/dist/pdf/validate.js

@@ -0,0 +1,136 @@
+import { CODES, DiagnosticSink } from "./diagnostics.js";
+import { hasStructContent } from "./inspect.js";
+/**
+ * pdf.js warnings this command reports as findings.
+ *
+ * **The most fragile coupling in the toolset**: these exist only because pdf.js
+ * prints a particular English string. That is contained rather than accepted —
+ * `tests/unit/pdf-validate.test.ts` drives one case per row from a deliberately
+ * damaged fixture, so a pdfjs bump that rewords a message fails the suite
+ * instead of quietly turning the check off.
+ *
+ * An unmatched warning is **not** dropped. It is reported once as `AP120` with
+ * the raw text, so a reworded message degrades to "something was recovered here"
+ * rather than to silence.
+ */
+export const WARNING_PATTERNS = [
+    {
+        pattern: /Indexing all PDF objects/i,
+        code: CODES.xrefReconstructed,
+        message: "The cross-reference table was unusable and the parser rebuilt it by scanning every object",
+        quality: "approximate",
+    },
+    {
+        pattern: /(fallback|standard) font|Failed to load font|Cannot load system font|font.*not found/i,
+        code: CODES.fontSubstituted,
+        message: "A font is not embedded or failed to load, and a substitute was used",
+        quality: "approximate",
+    },
+    {
+        pattern: /Unsupported (filter|image|stream)|filter.*not supported|JPX|JBIG2/i,
+        code: CODES.filterUnsupported,
+        message: "A stream uses a filter the parser could not decode",
+        quality: "unsupported",
+    },
+    {
+        pattern: /Unable to read mark info|Invalid Metadata|Skipping invalid Metadata/i,
+        code: CODES.metadataUnreadable,
+        message: "The document's metadata could not be read",
+        quality: "approximate",
+    },
+];
+/**
+ * Walks every page and reports what the parser could not do.
+ *
+ * No `--pages`: a partial validation that reported "valid" would be a lie, so
+ * the command always covers the whole document.
+ */
+export async function validateDocument(handle) {
+    const sink = new DiagnosticSink();
+    const { doc } = handle;
+    let taggedClaim;
+    try {
+        const markInfo = await handle.within(() => doc.getMarkInfo());
+        taggedClaim = markInfo instanceof Map && markInfo.get("Marked") === true;
+    }
+    catch {
+        taggedClaim = false;
+    }
+    let structured = 0;
+    let probed = 0;
+    for (let page = 1; page <= doc.numPages; page += 1) {
+        try {
+            const proxy = await handle.within(() => doc.getPage(page));
+            try {
+                await handle.within(() => proxy.getTextContent());
+            }
+            catch (error) {
+                sink.add({
+                    code: CODES.contentUndecodable,
+                    severity: "error",
+                    quality: "unsupported",
+                    message: `The page's content stream could not be decoded: ${error.message}`,
+                    page,
+                });
+            }
+            try {
+                const tree = await handle.within(() => proxy.getStructTree());
+                probed += 1;
+                if (hasStructContent(tree))
+                    structured += 1;
+            }
+            catch {
+                probed += 1;
+            }
+            proxy.cleanup();
+        }
+        catch (error) {
+            // The raw message, not `classify()`'s: that prefixes "Page N:" for callers
+            // with nowhere else to put it, and the finding already carries `page`.
+            sink.add({
+                code: CODES.pageUnreadable,
+                severity: "error",
+                quality: "unsupported",
+                message: error.message,
+                page,
+            });
+        }
+    }
+    if (taggedClaim && structured === 0)
+        sink.add({
+            code: CODES.taggedButEmpty,
+            quality: "approximate",
+            message: "The document declares /MarkInfo <</Marked true>> but no page carries a usable structure tree",
+        });
+    if (structured > 0 && structured < probed)
+        sink.add({
+            code: CODES.structPartial,
+            quality: "exact",
+            message: `A structure tree is present on ${structured} of ${probed} pages`,
+        });
+    if (handle.encrypted)
+        sink.add({
+            code: CODES.encryptedOpenPassword,
+            quality: "exact",
+            message: "The document is encrypted but opened without a password, so its restrictions are advisory only",
+        });
+    // The captured-warning channel. Everything above is a structural probe; these
+    // are conditions only the parser can see, and it only reports them in prose.
+    const unmatched = [];
+    for (const notice of handle.notices) {
+        const match = WARNING_PATTERNS.find((candidate) => candidate.pattern.test(notice));
+        if (match)
+            sink.add({ code: match.code, quality: match.quality, message: match.message });
+        else if (/^Warning:/i.test(notice))
+            unmatched.push(notice);
+    }
+    for (const notice of unmatched.slice(0, 10))
+        sink.add({
+            code: CODES.parserWarning,
+            quality: "approximate",
+            message: `The parser recovered from a condition this tool does not classify: ${notice.slice(0, 300)}`,
+            construct: notice.slice(0, 60),
+        });
+    return { diagnostics: sink.all() };
+}
+//# sourceMappingURL=validate.js.map

package/dist/pdf/validate.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"validate.js","sourceRoot":"","sources":["../../src/pdf/validate.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,KAAK,EAAE,cAAc,EAAE,MAAM,kBAAkB,CAAC;AAEzD,OAAO,EAAE,gBAAgB,EAAE,MAAM,cAAc,CAAC;AA8BhD;;;;;;;;;;;;GAYG;AACH,MAAM,CAAC,MAAM,gBAAgB,GAA8B;IACzD;QACE,OAAO,EAAE,2BAA2B;QACpC,IAAI,EAAE,KAAK,CAAC,iBAAiB;QAC7B,OAAO,EACL,2FAA2F;QAC7F,OAAO,EAAE,aAAa;KACvB;IACD;QACE,OAAO,EACL,uFAAuF;QACzF,IAAI,EAAE,KAAK,CAAC,eAAe;QAC3B,OAAO,EAAE,qEAAqE;QAC9E,OAAO,EAAE,aAAa;KACvB;IACD;QACE,OAAO,EAAE,oEAAoE;QAC7E,IAAI,EAAE,KAAK,CAAC,iBAAiB;QAC7B,OAAO,EAAE,oDAAoD;QAC7D,OAAO,EAAE,aAAa;KACvB;IACD;QACE,OAAO,EAAE,sEAAsE;QAC/E,IAAI,EAAE,KAAK,CAAC,kBAAkB;QAC9B,OAAO,EAAE,2CAA2C;QACpD,OAAO,EAAE,aAAa;KACvB;CACF,CAAC;AAMF;;;;;GAKG;AACH,MAAM,CAAC,KAAK,UAAU,gBAAgB,CAAC,MAAoB;IACzD,MAAM,IAAI,GAAG,IAAI,cAAc,EAAE,CAAC;IAClC,MAAM,EAAE,GAAG,EAAE,GAAG,MAAM,CAAC;IAEvB,IAAI,WAAoB,CAAC;IACzB,IAAI,CAAC;QACH,MAAM,QAAQ,GAAG,MAAM,MAAM,CAAC,MAAM,CAAC,GAAG,EAAE,CAAC,GAAG,CAAC,WAAW,EAAE,CAAC,CAAC;QAC9D,WAAW,GAAG,QAAQ,YAAY,GAAG,IAAI,QAAQ,CAAC,GAAG,CAAC,QAAQ,CAAC,KAAK,IAAI,CAAC;IAC3E,CAAC;IAAC,MAAM,CAAC;QACP,WAAW,GAAG,KAAK,CAAC;IACtB,CAAC;IAED,IAAI,UAAU,GAAG,CAAC,CAAC;IACnB,IAAI,MAAM,GAAG,CAAC,CAAC;IAEf,KAAK,IAAI,IAAI,GAAG,CAAC,EAAE,IAAI,IAAI,GAAG,CAAC,QAAQ,EAAE,IAAI,IAAI,CAAC,EAAE,CAAC;QACnD,IAAI,CAAC;YACH,MAAM,KAAK,GAAG,MAAM,MAAM,CAAC,MAAM,CAAC,GAAG,EAAE,CAAC,GAAG,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC;YAC3D,IAAI,CAAC;gBACH,MAAM,MAAM,CAAC,MAAM,CAAC,GAAG,EAAE,CAAC,KAAK,CAAC,cAAc,EAAE,CAAC,CAAC;YACpD,CAAC;YAAC,OAAO,KAAK,EAAE,CAAC;gBACf,IAAI,CAAC,GAAG,CAAC;oBACP,IAAI,EAAE,KAAK,CAAC,kBAAkB;oBAC9B,QAAQ,EAAE,OAAO;oBACjB,OAAO,EAAE,aAAa;oBACtB,OAAO,EAAE,mDAAoD,KAAe,CAAC,OAAO,EAAE;oBACtF,IAAI;iBACL,CAAC,CAAC;YACL,CAAC;YACD,IAAI,CAAC;gBACH,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,MAAM,CAAC,GAAG,EAAE,CAAC,KAAK,CAAC,aAAa,EAAE,CAAC,CAAC;gBAC9D,MAAM,IAAI,CAAC,CAAC;gBACZ,IAAI,gBAAgB,CAAC,IAAI,CAAC;oBAAE,UAAU,IAAI,CAAC,CAAC;YAC9C,CAAC;YAAC,MAAM,CAAC;gBACP,MAAM,IAAI,CAAC,CAAC;YACd,CAAC;YACD,KAAK,CAAC,OAAO,EAAE,CAAC;QAClB,CAAC;QAAC,OAAO,KAAK,EAAE,CAAC;YACf,2EAA2E;YAC3E,uEAAuE;YACvE,IAAI,CAAC,GAAG,CAAC;gBACP,IAAI,EAAE,KAAK,CAAC,cAAc;gBAC1B,QAAQ,EAAE,OAAO;gBACjB,OAAO,EAAE,aAAa;gBACtB,OAAO,EAAG,KAAe,CAAC,OAAO;gBACjC,IAAI;aACL,CAAC,CAAC;QACL,CAAC;IACH,CAAC;IAED,IAAI,WAAW,IAAI,UAAU,KAAK,CAAC;QACjC,IAAI,CAAC,GAAG,CAAC;YACP,IAAI,EAAE,KAAK,CAAC,cAAc;YAC1B,OAAO,EAAE,aAAa;YACtB,OAAO,EACL,8FAA8F;SACjG,CAAC,CAAC;IACL,IAAI,UAAU,GAAG,CAAC,IAAI,UAAU,GAAG,MAAM;QACvC,IAAI,CAAC,GAAG,CAAC;YACP,IAAI,EAAE,KAAK,CAAC,aAAa;YACzB,OAAO,EAAE,OAAO;YAChB,OAAO,EAAE,kCAAkC,UAAU,OAAO,MAAM,QAAQ;SAC3E,CAAC,CAAC;IAEL,IAAI,MAAM,CAAC,SAAS;QAClB,IAAI,CAAC,GAAG,CAAC;YACP,IAAI,EAAE,KAAK,CAAC,qBAAqB;YACjC,OAAO,EAAE,OAAO;YAChB,OAAO,EACL,gGAAgG;SACnG,CAAC,CAAC;IAEL,8EAA8E;IAC9E,6EAA6E;IAC7E,MAAM,SAAS,GAAa,EAAE,CAAC;IAC/B,KAAK,MAAM,MAAM,IAAI,MAAM,CAAC,OAAO,EAAE,CAAC;QACpC,MAAM,KAAK,GAAG,gBAAgB,CAAC,IAAI,CAAC,CAAC,SAAS,EAAE,EAAE,CAAC,SAAS,CAAC,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC;QACnF,IAAI,KAAK;YAAE,IAAI,CAAC,GAAG,CAAC,EAAE,IAAI,EAAE,KAAK,CAAC,IAAI,EAAE,OAAO,EAAE,KAAK,CAAC,OAAO,EAAE,OAAO,EAAE,KAAK,CAAC,OAAO,EAAE,CAAC,CAAC;aACrF,IAAI,YAAY,CAAC,IAAI,CAAC,MAAM,CAAC;YAAE,SAAS,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;IAC7D,CAAC;IACD,KAAK,MAAM,MAAM,IAAI,SAAS,CAAC,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC;QACzC,IAAI,CAAC,GAAG,CAAC;YACP,IAAI,EAAE,KAAK,CAAC,aAAa;YACzB,OAAO,EAAE,aAAa;YACtB,OAAO,EAAE,sEAAsE,MAAM,CAAC,KAAK,CAAC,CAAC,EAAE,GAAG,CAAC,EAAE;YACrG,SAAS,EAAE,MAAM,CAAC,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC;SAC/B,CAAC,CAAC;IAEL,OAAO,EAAE,WAAW,EAAE,IAAI,CAAC,GAAG,EAAE,EAAE,CAAC;AACrC,CAAC"}

package/dist/serve/pdf-tools.d.ts

@@ -0,0 +1,2 @@
+import type { ServeTool } from "./types.js";
+export declare const PDF_SERVE_TOOLS: readonly ServeTool[];

package/dist/serve/pdf-tools.js

@@ -0,0 +1,196 @@
+import fs from "node:fs";
+import { readAttachments } from "../pdf/attachments.js";
+import { withDocument } from "../pdf/document.js";
+import { readForm } from "../pdf/forms.js";
+import { documentSummary, inspectDocument } from "../pdf/inspect.js";
+import { readOutline } from "../pdf/outline.js";
+import { MAX_INPUT_BYTES, parsePageRange, readInput } from "../pdf/read.js";
+import { extractPage, runsToText } from "../pdf/text.js";
+import { toMarkdown } from "../pdf/to-markdown.js";
+import { confine, PathRejected, relativeTo } from "./paths.js";
+/**
+ * Read-only PDF tools.
+ *
+ * Kept out of `tools.ts` because that file is uniformly the Markdown workspace
+ * engine and these are not; `SERVE_TOOLS` spreads this array so registration
+ * stays one list.
+ *
+ * Three properties hold across every tool here:
+ *
+ * **Confined like everything else.** A PDF must live under `--root`. That is a
+ * real narrowing — a PDF handed to the CLI has nothing to do with a workspace —
+ * but the alternative is a second boundary on a surface whose whole claim is
+ * that there is one.
+ *
+ * **Nothing writes.** `list_pdf_attachments` inventories and there is no path
+ * to `--extract` here, the same line that keeps `scripts run` off this surface.
+ *
+ * **No command action is ever called.** Those terminate the process on a
+ * finding, which on a long-lived stdio server would end the session; these call
+ * the `src/pdf` modules directly, all of which are pure.
+ */
+/**
+ * Bounds, as constants rather than flags: `ServeContext` carries none, and a
+ * host cannot pass `--max-bytes`. They match the CLI defaults so the same
+ * document behaves the same way on both surfaces.
+ */
+const LIMITS = { maxBytes: MAX_INPUT_BYTES, timeoutMs: 30_000, maxPages: 5_000 };
+/** A confined, existing, regular file. Mirrors `fileArgument` in `tools.ts`. */
+function pdfArgument(args, context) {
+    const value = args.file;
+    if (typeof value !== "string" || value === "")
+        throw new PathRejected("file is required");
+    // `confine` refuses "-" before it can reach `readInput`, whose stdin branch
+    // would otherwise read fd 0 — the JSON-RPC channel — and deadlock the server.
+    const target = confine(context.root, value, "file");
+    if (!fs.existsSync(target) || !fs.statSync(target).isFile())
+        throw new PathRejected(`File not found: ${relativeTo(context.root, target)}`);
+    return target;
+}
+/**
+ * Opens the document, applying the same bounded read the CLI does.
+ *
+ * Project configuration is deliberately not consulted, even though
+ * `context.config` is populated on this surface: a `.cairn.yml` has no say over
+ * how a document named in a tool call is parsed, which is the same rule that
+ * keeps `pdf` out of `servesWorkspace`.
+ */
+async function open(args, context, body) {
+    const file = pdfArgument(args, context);
+    const read = await readInput(file, LIMITS);
+    if (!read.ok)
+        throw new Error(read.diagnostic.message);
+    return withDocument(read.bytes, LIMITS, (handle) => body(handle, file));
+}
+function pages(args, handle) {
+    const spec = args.pages;
+    if (typeof spec !== "string" || spec === "")
+        return undefined;
+    const parsed = parsePageRange(spec, handle.doc.numPages);
+    if (!parsed.ok)
+        throw new Error(parsed.diagnostic.message);
+    return parsed.pages;
+}
+const FILE = {
+    type: "string",
+    description: "PDF file relative to the served root.",
+};
+const PAGES = {
+    type: "string",
+    description: "1-based pages to read, e.g. 1,3,5-8. Every page by default.",
+};
+const inspectPdf = {
+    name: "inspect_pdf",
+    description: "Report a PDF's page count, metadata, tagging, and per-page text-layer classification. Read this before converting: `tagged` says whether structure can be trusted, and a page classified `absent` carries no text to extract. Mirrors `pdf inspect`.",
+    inputSchema: { type: "object", properties: { file: FILE }, required: ["file"] },
+    handler: (args, context) => open(args, context, async (handle, file) => {
+        const inspected = await inspectDocument(handle, { probeStructure: true });
+        return {
+            file: relativeTo(context.root, file),
+            document: inspected.document,
+            pages: inspected.pages,
+            diagnostics: inspected.diagnostics,
+        };
+    }),
+};
+const readPdfText = {
+    name: "read_pdf_text",
+    description: "Extract the text layer a PDF already carries, page by page. Never recognizes text in an image: a scanned page comes back empty with a diagnostic saying so rather than silently blank. Mirrors `pdf text`.",
+    inputSchema: { type: "object", properties: { file: FILE, pages: PAGES }, required: ["file"] },
+    handler: (args, context) => open(args, context, async (handle, file) => {
+        const summary = await documentSummary(handle);
+        const selected = pages(args, handle);
+        const wanted = selected ?? Array.from({ length: handle.doc.numPages }, (_, i) => i + 1);
+        const text = [];
+        const diagnostics = [...summary.diagnostics];
+        for (const page of wanted) {
+            const extracted = await extractPage(handle, page);
+            const body = runsToText(extracted.runs);
+            text.push({ page, text: body, characters: extracted.characters });
+        }
+        return {
+            file: relativeTo(context.root, file),
+            document: summary.document,
+            ...(selected ? { selectedPages: selected } : {}),
+            text,
+            diagnostics,
+        };
+    }),
+};
+const convertPdfToMarkdown = {
+    name: "convert_pdf_to_markdown",
+    description: "Convert a PDF's content to Markdown, reporting per construct what was inferred or lost. On an untagged page every block boundary is inferred from geometry, so read `document.tagged` and the diagnostics before trusting the structure. Mirrors `pdf to-markdown`.",
+    inputSchema: { type: "object", properties: { file: FILE, pages: PAGES }, required: ["file"] },
+    handler: (args, context) => open(args, context, async (handle, file) => {
+        const summary = await documentSummary(handle);
+        const selected = pages(args, handle);
+        const outline = await readOutline(handle);
+        const converted = await toMarkdown(handle, {
+            ...(selected ? { pages: selected } : {}),
+            headingLevels: outline.headingLevels,
+        });
+        return {
+            file: relativeTo(context.root, file),
+            document: summary.document,
+            ...(selected ? { selectedPages: selected } : {}),
+            markdown: converted.markdown,
+            diagnostics: [...summary.diagnostics, ...converted.diagnostics],
+        };
+    }),
+};
+const getPdfOutline = {
+    name: "get_pdf_outline",
+    description: "Read a PDF's declared outline (bookmarks) as a heading tree with resolved page numbers. A document with no outline returns an empty tree, which is an answer rather than a failure. Mirrors `pdf outline`.",
+    inputSchema: { type: "object", properties: { file: FILE }, required: ["file"] },
+    handler: (args, context) => open(args, context, async (handle, file) => {
+        const summary = await documentSummary(handle);
+        const outline = await readOutline(handle);
+        return {
+            file: relativeTo(context.root, file),
+            document: summary.document,
+            outline: outline.outline,
+            diagnostics: [...summary.diagnostics, ...outline.diagnostics],
+        };
+    }),
+};
+const listPdfAttachments = {
+    name: "list_pdf_attachments",
+    description: "List the files embedded inside a PDF, with size and SHA-256. Inventory only: this surface never writes, so there is no way to extract them here. Mirrors `pdf attachments` without --extract.",
+    inputSchema: { type: "object", properties: { file: FILE }, required: ["file"] },
+    handler: (args, context) => open(args, context, async (handle, file) => {
+        const summary = await documentSummary(handle);
+        // No `extract` option is passed, and none can be: the schema declares no
+        // such property, so a host cannot reach the writing path from here.
+        const read = await readAttachments(handle);
+        return {
+            file: relativeTo(context.root, file),
+            document: summary.document,
+            attachments: read.attachments,
+            diagnostics: [...summary.diagnostics, ...read.diagnostics],
+        };
+    }),
+};
+const listPdfFormFields = {
+    name: "list_pdf_form_fields",
+    description: "List a PDF's AcroForm field names, types, and current values. Reads and never writes. An XFA form reports type `xfa` with no fields rather than an empty list, because its values are not readable here. Mirrors `pdf forms`.",
+    inputSchema: { type: "object", properties: { file: FILE }, required: ["file"] },
+    handler: (args, context) => open(args, context, async (handle, file) => {
+        const summary = await documentSummary(handle);
+        const read = await readForm(handle);
+        return {
+            file: relativeTo(context.root, file),
+            document: summary.document,
+            form: read.form,
+            diagnostics: [...summary.diagnostics, ...read.diagnostics],
+        };
+    }),
+};
+export const PDF_SERVE_TOOLS = [
+    inspectPdf,
+    readPdfText,
+    convertPdfToMarkdown,
+    getPdfOutline,
+    listPdfAttachments,
+    listPdfFormFields,
+];
+//# sourceMappingURL=pdf-tools.js.map

package/dist/serve/pdf-tools.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"pdf-tools.js","sourceRoot":"","sources":["../../src/serve/pdf-tools.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,MAAM,SAAS,CAAC;AACzB,OAAO,EAAE,eAAe,EAAE,MAAM,uBAAuB,CAAC;AACxD,OAAO,EAAE,YAAY,EAAE,MAAM,oBAAoB,CAAC;AAElD,OAAO,EAAE,QAAQ,EAAE,MAAM,iBAAiB,CAAC;AAC3C,OAAO,EAAE,eAAe,EAAE,eAAe,EAAE,MAAM,mBAAmB,CAAC;AACrE,OAAO,EAAE,WAAW,EAAE,MAAM,mBAAmB,CAAC;AAChD,OAAO,EAAE,eAAe,EAAE,cAAc,EAAE,SAAS,EAAE,MAAM,gBAAgB,CAAC;AAC5E,OAAO,EAAE,WAAW,EAAE,UAAU,EAAE,MAAM,gBAAgB,CAAC;AACzD,OAAO,EAAE,UAAU,EAAE,MAAM,uBAAuB,CAAC;AAEnD,OAAO,EAAE,OAAO,EAAE,YAAY,EAAE,UAAU,EAAE,MAAM,YAAY,CAAC;AAG/D;;;;;;;;;;;;;;;;;;;;GAoBG;AAEH;;;;GAIG;AACH,MAAM,MAAM,GAAG,EAAE,QAAQ,EAAE,eAAe,EAAE,SAAS,EAAE,MAAM,EAAE,QAAQ,EAAE,KAAK,EAAE,CAAC;AAEjF,gFAAgF;AAChF,SAAS,WAAW,CAAC,IAA6B,EAAE,OAAqB;IACvE,MAAM,KAAK,GAAG,IAAI,CAAC,IAAI,CAAC;IACxB,IAAI,OAAO,KAAK,KAAK,QAAQ,IAAI,KAAK,KAAK,EAAE;QAAE,MAAM,IAAI,YAAY,CAAC,kBAAkB,CAAC,CAAC;IAC1F,4EAA4E;IAC5E,8EAA8E;IAC9E,MAAM,MAAM,GAAG,OAAO,CAAC,OAAO,CAAC,IAAI,EAAE,KAAK,EAAE,MAAM,CAAC,CAAC;IACpD,IAAI,CAAC,EAAE,CAAC,UAAU,CAAC,MAAM,CAAC,IAAI,CAAC,EAAE,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC,MAAM,EAAE;QACzD,MAAM,IAAI,YAAY,CAAC,mBAAmB,UAAU,CAAC,OAAO,CAAC,IAAI,EAAE,MAAM,CAAC,EAAE,CAAC,CAAC;IAChF,OAAO,MAAM,CAAC;AAChB,CAAC;AAED;;;;;;;GAOG;AACH,KAAK,UAAU,IAAI,CACjB,IAA6B,EAC7B,OAAqB,EACrB,IAAwD;IAExD,MAAM,IAAI,GAAG,WAAW,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC;IACxC,MAAM,IAAI,GAAG,MAAM,SAAS,CAAC,IAAI,EAAE,MAAM,CAAC,CAAC;IAC3C,IAAI,CAAC,IAAI,CAAC,EAAE;QAAE,MAAM,IAAI,KAAK,CAAC,IAAI,CAAC,UAAU,CAAC,OAAO,CAAC,CAAC;IACvD,OAAO,YAAY,CAAC,IAAI,CAAC,KAAK,EAAE,MAAM,EAAE,CAAC,MAAM,EAAE,EAAE,CAAC,IAAI,CAAC,MAAM,EAAE,IAAI,CAAC,CAAC,CAAC;AAC1E,CAAC;AAED,SAAS,KAAK,CAAC,IAA6B,EAAE,MAAoB;IAChE,MAAM,IAAI,GAAG,IAAI,CAAC,KAAK,CAAC;IACxB,IAAI,OAAO,IAAI,KAAK,QAAQ,IAAI,IAAI,KAAK,EAAE;QAAE,OAAO,SAAS,CAAC;IAC9D,MAAM,MAAM,GAAG,cAAc,CAAC,IAAI,EAAE,MAAM,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAC;IACzD,IAAI,CAAC,MAAM,CAAC,EAAE;QAAE,MAAM,IAAI,KAAK,CAAC,MAAM,CAAC,UAAU,CAAC,OAAO,CAAC,CAAC;IAC3D,OAAO,MAAM,CAAC,KAAK,CAAC;AACtB,CAAC;AAED,MAAM,IAAI,GAAG;IACX,IAAI,EAAE,QAAQ;IACd,WAAW,EAAE,uCAAuC;CAC5C,CAAC;AAEX,MAAM,KAAK,GAAG;IACZ,IAAI,EAAE,QAAQ;IACd,WAAW,EAAE,6DAA6D;CAClE,CAAC;AAEX,MAAM,UAAU,GAAc;IAC5B,IAAI,EAAE,aAAa;IACnB,WAAW,EACT,sPAAsP;IACxP,WAAW,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,UAAU,EAAE,EAAE,IAAI,EAAE,IAAI,EAAE,EAAE,QAAQ,EAAE,CAAC,MAAM,CAAC,EAAE;IAC/E,OAAO,EAAE,CAAC,IAAI,EAAE,OAAO,EAAE,EAAE,CACzB,IAAI,CAAC,IAAI,EAAE,OAAO,EAAE,KAAK,EAAE,MAAM,EAAE,IAAI,EAAE,EAAE;QACzC,MAAM,SAAS,GAAG,MAAM,eAAe,CAAC,MAAM,EAAE,EAAE,cAAc,EAAE,IAAI,EAAE,CAAC,CAAC;QAC1E,OAAO;YACL,IAAI,EAAE,UAAU,CAAC,OAAO,CAAC,IAAI,EAAE,IAAI,CAAC;YACpC,QAAQ,EAAE,SAAS,CAAC,QAAQ;YAC5B,KAAK,EAAE,SAAS,CAAC,KAAK;YACtB,WAAW,EAAE,SAAS,CAAC,WAAW;SACnC,CAAC;IACJ,CAAC,CAAC;CACL,CAAC;AAEF,MAAM,WAAW,GAAc;IAC7B,IAAI,EAAE,eAAe;IACrB,WAAW,EACT,4MAA4M;IAC9M,WAAW,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,UAAU,EAAE,EAAE,IAAI,EAAE,IAAI,EAAE,KAAK,EAAE,KAAK,EAAE,EAAE,QAAQ,EAAE,CAAC,MAAM,CAAC,EAAE;IAC7F,OAAO,EAAE,CAAC,IAAI,EAAE,OAAO,EAAE,EAAE,CACzB,IAAI,CAAC,IAAI,EAAE,OAAO,EAAE,KAAK,EAAE,MAAM,EAAE,IAAI,EAAE,EAAE;QACzC,MAAM,OAAO,GAAG,MAAM,eAAe,CAAC,MAAM,CAAC,CAAC;QAC9C,MAAM,QAAQ,GAAG,KAAK,CAAC,IAAI,EAAE,MAAM,CAAC,CAAC;QACrC,MAAM,MAAM,GAAG,QAAQ,IAAI,KAAK,CAAC,IAAI,CAAC,EAAE,MAAM,EAAE,MAAM,CAAC,GAAG,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;QACxF,MAAM,IAAI,GAAkB,EAAE,CAAC;QAC/B,MAAM,WAAW,GAAG,CAAC,GAAG,OAAO,CAAC,WAAW,CAAC,CAAC;QAC7C,KAAK,MAAM,IAAI,IAAI,MAAM,EAAE,CAAC;YAC1B,MAAM,SAAS,GAAG,MAAM,WAAW,CAAC,MAAM,EAAE,IAAI,CAAC,CAAC;YAClD,MAAM,IAAI,GAAG,UAAU,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC;YACxC,IAAI,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,UAAU,EAAE,SAAS,CAAC,UAAU,EAAE,CAAC,CAAC;QACpE,CAAC;QACD,OAAO;YACL,IAAI,EAAE,UAAU,CAAC,OAAO,CAAC,IAAI,EAAE,IAAI,CAAC;YACpC,QAAQ,EAAE,OAAO,CAAC,QAAQ;YAC1B,GAAG,CAAC,QAAQ,CAAC,CAAC,CAAC,EAAE,aAAa,EAAE,QAAQ,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;YAChD,IAAI;YACJ,WAAW;SACZ,CAAC;IACJ,CAAC,CAAC;CACL,CAAC;AAEF,MAAM,oBAAoB,GAAc;IACtC,IAAI,EAAE,yBAAyB;IAC/B,WAAW,EACT,qQAAqQ;IACvQ,WAAW,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,UAAU,EAAE,EAAE,IAAI,EAAE,IAAI,EAAE,KAAK,EAAE,KAAK,EAAE,EAAE,QAAQ,EAAE,CAAC,MAAM,CAAC,EAAE;IAC7F,OAAO,EAAE,CAAC,IAAI,EAAE,OAAO,EAAE,EAAE,CACzB,IAAI,CAAC,IAAI,EAAE,OAAO,EAAE,KAAK,EAAE,MAAM,EAAE,IAAI,EAAE,EAAE;QACzC,MAAM,OAAO,GAAG,MAAM,eAAe,CAAC,MAAM,CAAC,CAAC;QAC9C,MAAM,QAAQ,GAAG,KAAK,CAAC,IAAI,EAAE,MAAM,CAAC,CAAC;QACrC,MAAM,OAAO,GAAG,MAAM,WAAW,CAAC,MAAM,CAAC,CAAC;QAC1C,MAAM,SAAS,GAAG,MAAM,UAAU,CAAC,MAAM,EAAE;YACzC,GAAG,CAAC,QAAQ,CAAC,CAAC,CAAC,EAAE,KAAK,EAAE,QAAQ,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;YACxC,aAAa,EAAE,OAAO,CAAC,aAAa;SACrC,CAAC,CAAC;QACH,OAAO;YACL,IAAI,EAAE,UAAU,CAAC,OAAO,CAAC,IAAI,EAAE,IAAI,CAAC;YACpC,QAAQ,EAAE,OAAO,CAAC,QAAQ;YAC1B,GAAG,CAAC,QAAQ,CAAC,CAAC,CAAC,EAAE,aAAa,EAAE,QAAQ,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;YAChD,QAAQ,EAAE,SAAS,CAAC,QAAQ;YAC5B,WAAW,EAAE,CAAC,GAAG,OAAO,CAAC,WAAW,EAAE,GAAG,SAAS,CAAC,WAAW,CAAC;SAChE,CAAC;IACJ,CAAC,CAAC;CACL,CAAC;AAEF,MAAM,aAAa,GAAc;IAC/B,IAAI,EAAE,iBAAiB;IACvB,WAAW,EACT,4MAA4M;IAC9M,WAAW,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,UAAU,EAAE,EAAE,IAAI,EAAE,IAAI,EAAE,EAAE,QAAQ,EAAE,CAAC,MAAM,CAAC,EAAE;IAC/E,OAAO,EAAE,CAAC,IAAI,EAAE,OAAO,EAAE,EAAE,CACzB,IAAI,CAAC,IAAI,EAAE,OAAO,EAAE,KAAK,EAAE,MAAM,EAAE,IAAI,EAAE,EAAE;QACzC,MAAM,OAAO,GAAG,MAAM,eAAe,CAAC,MAAM,CAAC,CAAC;QAC9C,MAAM,OAAO,GAAG,MAAM,WAAW,CAAC,MAAM,CAAC,CAAC;QAC1C,OAAO;YACL,IAAI,EAAE,UAAU,CAAC,OAAO,CAAC,IAAI,EAAE,IAAI,CAAC;YACpC,QAAQ,EAAE,OAAO,CAAC,QAAQ;YAC1B,OAAO,EAAE,OAAO,CAAC,OAAO;YACxB,WAAW,EAAE,CAAC,GAAG,OAAO,CAAC,WAAW,EAAE,GAAG,OAAO,CAAC,WAAW,CAAC;SAC9D,CAAC;IACJ,CAAC,CAAC;CACL,CAAC;AAEF,MAAM,kBAAkB,GAAc;IACpC,IAAI,EAAE,sBAAsB;IAC5B,WAAW,EACT,+LAA+L;IACjM,WAAW,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,UAAU,EAAE,EAAE,IAAI,EAAE,IAAI,EAAE,EAAE,QAAQ,EAAE,CAAC,MAAM,CAAC,EAAE;IAC/E,OAAO,EAAE,CAAC,IAAI,EAAE,OAAO,EAAE,EAAE,CACzB,IAAI,CAAC,IAAI,EAAE,OAAO,EAAE,KAAK,EAAE,MAAM,EAAE,IAAI,EAAE,EAAE;QACzC,MAAM,OAAO,GAAG,MAAM,eAAe,CAAC,MAAM,CAAC,CAAC;QAC9C,yEAAyE;QACzE,oEAAoE;QACpE,MAAM,IAAI,GAAG,MAAM,eAAe,CAAC,MAAM,CAAC,CAAC;QAC3C,OAAO;YACL,IAAI,EAAE,UAAU,CAAC,OAAO,CAAC,IAAI,EAAE,IAAI,CAAC;YACpC,QAAQ,EAAE,OAAO,CAAC,QAAQ;YAC1B,WAAW,EAAE,IAAI,CAAC,WAAW;YAC7B,WAAW,EAAE,CAAC,GAAG,OAAO,CAAC,WAAW,EAAE,GAAG,IAAI,CAAC,WAAW,CAAC;SAC3D,CAAC;IACJ,CAAC,CAAC;CACL,CAAC;AAEF,MAAM,iBAAiB,GAAc;IACnC,IAAI,EAAE,sBAAsB;IAC5B,WAAW,EACT,+NAA+N;IACjO,WAAW,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,UAAU,EAAE,EAAE,IAAI,EAAE,IAAI,EAAE,EAAE,QAAQ,EAAE,CAAC,MAAM,CAAC,EAAE;IAC/E,OAAO,EAAE,CAAC,IAAI,EAAE,OAAO,EAAE,EAAE,CACzB,IAAI,CAAC,IAAI,EAAE,OAAO,EAAE,KAAK,EAAE,MAAM,EAAE,IAAI,EAAE,EAAE;QACzC,MAAM,OAAO,GAAG,MAAM,eAAe,CAAC,MAAM,CAAC,CAAC;QAC9C,MAAM,IAAI,GAAG,MAAM,QAAQ,CAAC,MAAM,CAAC,CAAC;QACpC,OAAO;YACL,IAAI,EAAE,UAAU,CAAC,OAAO,CAAC,IAAI,EAAE,IAAI,CAAC;YACpC,QAAQ,EAAE,OAAO,CAAC,QAAQ;YAC1B,IAAI,EAAE,IAAI,CAAC,IAAI;YACf,WAAW,EAAE,CAAC,GAAG,OAAO,CAAC,WAAW,EAAE,GAAG,IAAI,CAAC,WAAW,CAAC;SAC3D,CAAC;IACJ,CAAC,CAAC;CACL,CAAC;AAEF,MAAM,CAAC,MAAM,eAAe,GAAyB;IACnD,UAAU;IACV,WAAW;IACX,oBAAoB;IACpB,aAAa;IACb,kBAAkB;IAClB,iBAAiB;CAClB,CAAC"}

package/dist/serve/tools.d.ts

@@ -1,19 +1,4 @@
-import type { Tool } from "@modelcontextprotocol/sdk/types.js";
-import type { ResolvedConfig } from "../config.js";
-import type { Workspace } from "../workspace.js";
-export interface ServeContext {
-    workspace: Workspace;
-    config: ResolvedConfig;
-    /** Confinement boundary, already resolved through symlinks. */
-    root: string;
-    /** Upper bound on parallel lints, so a large audit cannot starve the transport. */
-    concurrency: number;
-}
-export interface ServeTool {
-    name: string;
-    description: string;
-    inputSchema: Tool["inputSchema"];
-    handler: (args: Record<string, unknown>, context: ServeContext) => Promise<unknown> | unknown;
-}
+import type { ServeTool } from "./types.js";
+export type { ServeContext, ServeTool } from "./types.js";
 export declare const SERVE_TOOLS: readonly ServeTool[];
 export declare const TOOL_BY_NAME: ReadonlyMap<string, ServeTool>;

package/dist/serve/tools.js

@@ -12,6 +12,7 @@ import { buildPlan } from "../query/plan.js";
 import { executePlan } from "../query/execute.js";
 import { ENTITY_KINDS } from "../query/entities.js";
 import { confine, PathRejected, relativeTo } from "./paths.js";
+import { PDF_SERVE_TOOLS } from "./pdf-tools.js";
 // --- argument readers -------------------------------------------------------
 // Arguments have already been validated against the tool's schema, so these
 // only narrow types; they never re-report a shape error.
@@ -592,6 +593,9 @@ export const SERVE_TOOLS = [
     listTasks,
     listCodeBlocks,
     findReferences,
+    // The PDF tools live in their own module: this file is uniformly the Markdown
+    // workspace engine, and registration is still one list.
+    ...PDF_SERVE_TOOLS,
 ];
 export const TOOL_BY_NAME = new Map(SERVE_TOOLS.map((tool) => [tool.name, tool]));
 //# sourceMappingURL=tools.js.map