@leemour/cli-messaging 0.178.0 → 0.180.0

package/dist/attachments/container.d.ts

@@ -0,0 +1,7 @@
+import { type Document, type Element, type Node } from "@xmldom/xmldom";
+export declare class ReaderLimit extends Error {
+}
+export declare const readContainer: (bytes: Uint8Array, signal?: AbortSignal) => Promise<Map<string, Uint8Array>>;
+export declare const xmlPart: (archive: Map<string, Uint8Array>, path: string) => Document;
+export declare const elements: (node: Node, name: string, namespace: string) => Element[];
+//# sourceMappingURL=container.d.ts.map

package/dist/attachments/container.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"container.d.ts","sourceRoot":"","sources":["../../src/attachments/container.ts"],"names":[],"mappings":"AAGA,OAAO,EAAa,KAAK,QAAQ,EAAE,KAAK,OAAO,EAAE,KAAK,IAAI,EAAE,MAAM,gBAAgB,CAAA;AAUlF,qBAAa,WAAY,SAAQ,KAAK;CAAG;AAqEzC,eAAO,MAAM,aAAa,UAAiB,UAAU,WAAW,WAAW,KAAG,OAAO,CAAC,GAAG,CAAC,MAAM,EAAE,UAAU,CAAC,CAiD5G,CAAA;AAyCD,eAAO,MAAM,OAAO,YAAa,GAAG,CAAC,MAAM,EAAE,UAAU,CAAC,QAAQ,MAAM,KAAG,QAyBxE,CAAA;AAED,eAAO,MAAM,QAAQ,SAAU,IAAI,QAAQ,MAAM,aAAa,MAAM,KAAG,OAAO,EAS7E,CAAA"}

package/dist/attachments/container.js

@@ -0,0 +1,236 @@
+import { setImmediate as tick } from "node:timers/promises";
+import { crc32 } from "node:zlib";
+import { CliError } from "@leemour/cli-core";
+import { DOMParser } from "@xmldom/xmldom";
+import { Unzip, UnzipInflate } from "fflate";
+import { decodeXml } from "./encoding.js";
+import { MAX_FILE_BYTES as MAX_EXPANDED } from "./limits.js";
+const MAX_PART = 10 * 1024 * 1024;
+const MAX_ENTRIES = 1000;
+const decoder = new TextDecoder("utf-8", { fatal: true });
+export class ReaderLimit extends Error {
+}
+const cancelled = (signal) => {
+    if (signal?.aborted)
+        throw new CliError("cancelled", "document extraction cancelled");
+};
+const safeName = (name) => name.length > 0 &&
+    name.length <= 512 &&
+    !/^[\\/]|\\/.test(name) &&
+    ![...name].some((character) => character.charCodeAt(0) < 32) &&
+    !name.split("/").includes("..");
+const directory = (bytes) => {
+    const view = new DataView(bytes.buffer, bytes.byteOffset, bytes.byteLength);
+    let end = -1;
+    for (let index = bytes.length - 22; index >= Math.max(0, bytes.length - 65_557); index--) {
+        if (view.getUint32(index, true) === 0x06054b50 && index + 22 + view.getUint16(index + 20, true) === bytes.length) {
+            end = index;
+            break;
+        }
+    }
+    if (end < 0 || view.getUint16(end + 4, true) || view.getUint16(end + 6, true))
+        throw new Error("invalid_zip");
+    const count = view.getUint16(end + 10, true);
+    if (count > MAX_ENTRIES)
+        throw new ReaderLimit();
+    if (count !== view.getUint16(end + 8, true))
+        throw new Error("invalid_zip");
+    const stop = view.getUint32(end + 16, true) + view.getUint32(end + 12, true);
+    if (stop > end)
+        throw new Error("invalid_zip");
+    let offset = view.getUint32(end + 16, true);
+    let expanded = 0;
+    const entries = new Map();
+    for (let index = 0; index < count; index++) {
+        if (offset + 46 > stop || view.getUint32(offset, true) !== 0x02014b50)
+            throw new Error("invalid_zip");
+        const flags = view.getUint16(offset + 8, true);
+        const method = view.getUint16(offset + 10, true);
+        const size = view.getUint32(offset + 24, true);
+        const compressed = view.getUint32(offset + 20, true);
+        const length = view.getUint16(offset + 28, true);
+        const next = offset + 46 + length + view.getUint16(offset + 30, true) + view.getUint16(offset + 32, true);
+        if (next > stop)
+            throw new Error("invalid_zip");
+        const name = decoder.decode(bytes.subarray(offset + 46, offset + 46 + length));
+        if (!safeName(name) || entries.has(name) || flags & 1 || (method !== 0 && method !== 8))
+            throw new Error("invalid_zip");
+        if (((view.getUint32(offset + 38, true) >>> 16) & 0xf000) === 0xa000)
+            throw new Error("invalid_zip");
+        expanded += size;
+        if (expanded > MAX_EXPANDED)
+            throw new ReaderLimit();
+        const local = view.getUint32(offset + 42, true);
+        if (local + 30 > offset || view.getUint32(local, true) !== 0x04034b50)
+            throw new Error("invalid_zip");
+        const localLength = view.getUint16(local + 26, true);
+        const start = local + 30 + localLength + view.getUint16(local + 28, true);
+        if (start + compressed > view.getUint32(end + 16, true) ||
+            decoder.decode(bytes.subarray(local + 30, local + 30 + localLength)) !== name ||
+            view.getUint16(local + 6, true) !== flags ||
+            view.getUint16(local + 8, true) !== method)
+            throw new Error("invalid_zip");
+        if (!(flags & 8) &&
+            (view.getUint32(local + 14, true) !== view.getUint32(offset + 16, true) ||
+                view.getUint32(local + 18, true) !== compressed ||
+                view.getUint32(local + 22, true) !== size))
+            throw new Error("invalid_zip");
+        entries.set(name, { size, crc: view.getUint32(offset + 16, true) });
+        offset = next;
+    }
+    if (offset !== stop)
+        throw new Error("invalid_zip");
+    return entries;
+};
+export const readContainer = async (bytes, signal) => {
+    cancelled(signal);
+    const entries = directory(bytes);
+    const result = new Map();
+    const seen = new Set();
+    const done = new Set();
+    let expanded = 0;
+    const archive = new Unzip((file) => {
+        const expected = entries.get(file.name);
+        if (!expected || seen.has(file.name))
+            throw new Error("invalid_zip");
+        seen.add(file.name);
+        const keep = /\.(xml|rels|opf|xhtml|html)$/i.test(file.name) || file.name === "mimetype";
+        let size = 0;
+        let checksum = 0;
+        const parts = [];
+        file.ondata = (error, data, final) => {
+            if (error)
+                throw error;
+            size += data.length;
+            expanded += data.length;
+            if (expanded > MAX_EXPANDED || (keep && size > MAX_PART))
+                throw new ReaderLimit();
+            checksum = crc32(data, checksum);
+            if (keep)
+                parts.push(data.slice());
+            if (final) {
+                if (size !== expected.size || checksum !== expected.crc)
+                    throw new Error("invalid_zip");
+                done.add(file.name);
+                if (keep) {
+                    const content = new Uint8Array(size);
+                    let offset = 0;
+                    for (const part of parts) {
+                        content.set(part, offset);
+                        offset += part.length;
+                    }
+                    result.set(file.name, content);
+                    parts.length = 0;
+                }
+            }
+        };
+        file.start();
+    });
+    archive.register(UnzipInflate);
+    // Small compressed chunks bound allocation before the expanded-byte check runs.
+    for (let offset = 0; offset < bytes.length; offset += 1024) {
+        cancelled(signal);
+        archive.push(bytes.subarray(offset, offset + 1024), offset + 1024 >= bytes.length);
+        if (offset % 65_536 === 0)
+            await tick();
+    }
+    cancelled(signal);
+    if (seen.size !== entries.size || done.size !== entries.size)
+        throw new Error("incomplete_zip");
+    return result;
+};
+const checkXml = (text) => {
+    if (/<!DOCTYPE|<!ENTITY/i.test(text))
+        throw new Error("unsupported_xml");
+    let depth = 0;
+    let nodes = 0;
+    for (let index = 0; index < text.length; index++) {
+        if (text[index] !== "<")
+            continue;
+        const skip = text.startsWith("<!--", index)
+            ? "-->"
+            : text.startsWith("<![CDATA[", index)
+                ? "]]>"
+                : text.startsWith("<?", index)
+                    ? "?>"
+                    : undefined;
+        if (skip) {
+            if (++nodes > 200_000)
+                throw new ReaderLimit();
+            const end = text.indexOf(skip, index + 2);
+            if (end < 0)
+                throw new Error("invalid_xml");
+            index = end + skip.length - 1;
+            continue;
+        }
+        const start = index;
+        let quote = "";
+        for (index++; index < text.length; index++) {
+            const char = text[index];
+            if (quote) {
+                if (char === quote)
+                    quote = "";
+            }
+            else if (char === '"' || char === "'")
+                quote = char;
+            else if (char === ">")
+                break;
+        }
+        if (index === text.length)
+            throw new Error("invalid_xml");
+        const tag = text.slice(start, index + 1);
+        if (tag.startsWith("</"))
+            depth--;
+        else if (!/\/\s*>$/.test(tag))
+            depth++;
+        if (++nodes > 200_000 || depth > 128)
+            throw new ReaderLimit();
+        if (depth < 0)
+            throw new Error("invalid_xml");
+    }
+    if (depth !== 0)
+        throw new Error("invalid_xml");
+};
+export const xmlPart = (archive, path) => {
+    const bytes = archive.get(path);
+    if (!bytes)
+        throw new Error("missing_part");
+    let text = decodeXml(bytes);
+    if (/\.xhtml$/i.test(path)) {
+        let start = 0;
+        while (start < text.length) {
+            start += /^\s*/.exec(text.slice(start))?.[0].length ?? 0;
+            const end = text.startsWith("<?", start) ? "?>" : text.startsWith("<!--", start) ? "-->" : undefined;
+            if (!end)
+                break;
+            const index = text.indexOf(end, start + 2);
+            if (index < 0)
+                throw new Error("invalid_xml");
+            start = index + end.length;
+        }
+        const doctype = /^<!DOCTYPE\s+html\s*>/i.exec(text.slice(start));
+        if (doctype)
+            text = text.slice(0, start) + text.slice(start + doctype[0].length);
+    }
+    checkXml(text);
+    const document = new DOMParser({
+        onError: () => {
+            throw new Error("invalid_xml");
+        },
+    }).parseFromString(text, "application/xml");
+    if (!document.documentElement)
+        throw new Error("invalid_xml");
+    return document;
+};
+export const elements = (node, name, namespace) => {
+    const found = [];
+    const visit = (current) => {
+        if (current.nodeType === 1 && current.localName === name && current.namespaceURI === namespace)
+            found.push(current);
+        for (let child = current.firstChild; child; child = child.nextSibling)
+            visit(child);
+    };
+    visit(node);
+    return found;
+};
+//# sourceMappingURL=container.js.map

package/dist/attachments/container.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"container.js","sourceRoot":"","sources":["../../src/attachments/container.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,YAAY,IAAI,IAAI,EAAE,MAAM,sBAAsB,CAAA;AAC3D,OAAO,EAAE,KAAK,EAAE,MAAM,WAAW,CAAA;AACjC,OAAO,EAAE,QAAQ,EAAE,MAAM,mBAAmB,CAAA;AAC5C,OAAO,EAAE,SAAS,EAA0C,MAAM,gBAAgB,CAAA;AAClF,OAAO,EAAE,KAAK,EAAE,YAAY,EAAE,MAAM,QAAQ,CAAA;AAC5C,OAAO,EAAE,SAAS,EAAE,MAAM,eAAe,CAAA;AAEzC,OAAO,EAAE,cAAc,IAAI,YAAY,EAAE,MAAM,aAAa,CAAA;AAE5D,MAAM,QAAQ,GAAG,EAAE,GAAG,IAAI,GAAG,IAAI,CAAA;AACjC,MAAM,WAAW,GAAG,IAAI,CAAA;AACxB,MAAM,OAAO,GAAG,IAAI,WAAW,CAAC,OAAO,EAAE,EAAE,KAAK,EAAE,IAAI,EAAE,CAAC,CAAA;AAEzD,MAAM,OAAO,WAAY,SAAQ,KAAK;CAAG;AACzC,MAAM,SAAS,GAAG,CAAC,MAAoB,EAAE,EAAE;IACzC,IAAI,MAAM,EAAE,OAAO;QAAE,MAAM,IAAI,QAAQ,CAAC,WAAW,EAAE,+BAA+B,CAAC,CAAA;AACvF,CAAC,CAAA;AACD,MAAM,QAAQ,GAAG,CAAC,IAAY,EAAE,EAAE,CAChC,IAAI,CAAC,MAAM,GAAG,CAAC;IACf,IAAI,CAAC,MAAM,IAAI,GAAG;IAClB,CAAC,WAAW,CAAC,IAAI,CAAC,IAAI,CAAC;IACvB,CAAC,CAAC,GAAG,IAAI,CAAC,CAAC,IAAI,CAAC,CAAC,SAAS,EAAE,EAAE,CAAC,SAAS,CAAC,UAAU,CAAC,CAAC,CAAC,GAAG,EAAE,CAAC;IAC5D,CAAC,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAA;AAEjC,MAAM,SAAS,GAAG,CAAC,KAAiB,EAAE,EAAE;IACtC,MAAM,IAAI,GAAG,IAAI,QAAQ,CAAC,KAAK,CAAC,MAAM,EAAE,KAAK,CAAC,UAAU,EAAE,KAAK,CAAC,UAAU,CAAC,CAAA;IAC3E,IAAI,GAAG,GAAG,CAAC,CAAC,CAAA;IACZ,KAAK,IAAI,KAAK,GAAG,KAAK,CAAC,MAAM,GAAG,EAAE,EAAE,KAAK,IAAI,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,KAAK,CAAC,MAAM,GAAG,MAAM,CAAC,EAAE,KAAK,EAAE,EAAE,CAAC;QACzF,IAAI,IAAI,CAAC,SAAS,CAAC,KAAK,EAAE,IAAI,CAAC,KAAK,UAAU,IAAI,KAAK,GAAG,EAAE,GAAG,IAAI,CAAC,SAAS,CAAC,KAAK,GAAG,EAAE,EAAE,IAAI,CAAC,KAAK,KAAK,CAAC,MAAM,EAAE,CAAC;YACjH,GAAG,GAAG,KAAK,CAAA;YACX,MAAK;QACP,CAAC;IACH,CAAC;IACD,IAAI,GAAG,GAAG,CAAC,IAAI,IAAI,CAAC,SAAS,CAAC,GAAG,GAAG,CAAC,EAAE,IAAI,CAAC,IAAI,IAAI,CAAC,SAAS,CAAC,GAAG,GAAG,CAAC,EAAE,IAAI,CAAC;QAAE,MAAM,IAAI,KAAK,CAAC,aAAa,CAAC,CAAA;IAC7G,MAAM,KAAK,GAAG,IAAI,CAAC,SAAS,CAAC,GAAG,GAAG,EAAE,EAAE,IAAI,CAAC,CAAA;IAC5C,IAAI,KAAK,GAAG,WAAW;QAAE,MAAM,IAAI,WAAW,EAAE,CAAA;IAChD,IAAI,KAAK,KAAK,IAAI,CAAC,SAAS,CAAC,GAAG,GAAG,CAAC,EAAE,IAAI,CAAC;QAAE,MAAM,IAAI,KAAK,CAAC,aAAa,CAAC,CAAA;IAC3E,MAAM,IAAI,GAAG,IAAI,CAAC,SAAS,CAAC,GAAG,GAAG,EAAE,EAAE,IAAI,CAAC,GAAG,IAAI,CAAC,SAAS,CAAC,GAAG,GAAG,EAAE,EAAE,IAAI,CAAC,CAAA;IAC5E,IAAI,IAAI,GAAG,GAAG;QAAE,MAAM,IAAI,KAAK,CAAC,aAAa,CAAC,CAAA;IAC9C,IAAI,MAAM,GAAG,IAAI,CAAC,SAAS,CAAC,GAAG,GAAG,EAAE,EAAE,IAAI,CAAC,CAAA;IAC3C,IAAI,QAAQ,GAAG,CAAC,CAAA;IAChB,MAAM,OAAO,GAAG,IAAI,GAAG,EAAyC,CAAA;IAChE,KAAK,IAAI,KAAK,GAAG,CAAC,EAAE,KAAK,GAAG,KAAK,EAAE,KAAK,EAAE,EAAE,CAAC;QAC3C,IAAI,MAAM,GAAG,EAAE,GAAG,IAAI,IAAI,IAAI,CAAC,SAAS,CAAC,MAAM,EAAE,IAAI,CAAC,KAAK,UAAU;YAAE,MAAM,IAAI,KAAK,CAAC,aAAa,CAAC,CAAA;QACrG,MAAM,KAAK,GAAG,IAAI,CAAC,SAAS,CAAC,MAAM,GAAG,CAAC,EAAE,IAAI,CAAC,CAAA;QAC9C,MAAM,MAAM,GAAG,IAAI,CAAC,SAAS,CAAC,MAAM,GAAG,EAAE,EAAE,IAAI,CAAC,CAAA;QAChD,MAAM,IAAI,GAAG,IAAI,CAAC,SAAS,CAAC,MAAM,GAAG,EAAE,EAAE,IAAI,CAAC,CAAA;QAC9C,MAAM,UAAU,GAAG,IAAI,CAAC,SAAS,CAAC,MAAM,GAAG,EAAE,EAAE,IAAI,CAAC,CAAA;QACpD,MAAM,MAAM,GAAG,IAAI,CAAC,SAAS,CAAC,MAAM,GAAG,EAAE,EAAE,IAAI,CAAC,CAAA;QAChD,MAAM,IAAI,GAAG,MAAM,GAAG,EAAE,GAAG,MAAM,GAAG,IAAI,CAAC,SAAS,CAAC,MAAM,GAAG,EAAE,EAAE,IAAI,CAAC,GAAG,IAAI,CAAC,SAAS,CAAC,MAAM,GAAG,EAAE,EAAE,IAAI,CAAC,CAAA;QACzG,IAAI,IAAI,GAAG,IAAI;YAAE,MAAM,IAAI,KAAK,CAAC,aAAa,CAAC,CAAA;QAC/C,MAAM,IAAI,GAAG,OAAO,CAAC,MAAM,CAAC,KAAK,CAAC,QAAQ,CAAC,MAAM,GAAG,EAAE,EAAE,MAAM,GAAG,EAAE,GAAG,MAAM,CAAC,CAAC,CAAA;QAC9E,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,OAAO,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,KAAK,GAAG,CAAC,IAAI,CAAC,MAAM,KAAK,CAAC,IAAI,MAAM,KAAK,CAAC,CAAC;YACrF,MAAM,IAAI,KAAK,CAAC,aAAa,CAAC,CAAA;QAChC,IAAI,CAAC,CAAC,IAAI,CAAC,SAAS,CAAC,MAAM,GAAG,EAAE,EAAE,IAAI,CAAC,KAAK,EAAE,CAAC,GAAG,MAAM,CAAC,KAAK,MAAM;YAAE,MAAM,IAAI,KAAK,CAAC,aAAa,CAAC,CAAA;QACpG,QAAQ,IAAI,IAAI,CAAA;QAChB,IAAI,QAAQ,GAAG,YAAY;YAAE,MAAM,IAAI,WAAW,EAAE,CAAA;QACpD,MAAM,KAAK,GAAG,IAAI,CAAC,SAAS,CAAC,MAAM,GAAG,EAAE,EAAE,IAAI,CAAC,CAAA;QAC/C,IAAI,KAAK,GAAG,EAAE,GAAG,MAAM,IAAI,IAAI,CAAC,SAAS,CAAC,KAAK,EAAE,IAAI,CAAC,KAAK,UAAU;YAAE,MAAM,IAAI,KAAK,CAAC,aAAa,CAAC,CAAA;QACrG,MAAM,WAAW,GAAG,IAAI,CAAC,SAAS,CAAC,KAAK,GAAG,EAAE,EAAE,IAAI,CAAC,CAAA;QACpD,MAAM,KAAK,GAAG,KAAK,GAAG,EAAE,GAAG,WAAW,GAAG,IAAI,CAAC,SAAS,CAAC,KAAK,GAAG,EAAE,EAAE,IAAI,CAAC,CAAA;QACzE,IACE,KAAK,GAAG,UAAU,GAAG,IAAI,CAAC,SAAS,CAAC,GAAG,GAAG,EAAE,EAAE,IAAI,CAAC;YACnD,OAAO,CAAC,MAAM,CAAC,KAAK,CAAC,QAAQ,CAAC,KAAK,GAAG,EAAE,EAAE,KAAK,GAAG,EAAE,GAAG,WAAW,CAAC,CAAC,KAAK,IAAI;YAC7E,IAAI,CAAC,SAAS,CAAC,KAAK,GAAG,CAAC,EAAE,IAAI,CAAC,KAAK,KAAK;YACzC,IAAI,CAAC,SAAS,CAAC,KAAK,GAAG,CAAC,EAAE,IAAI,CAAC,KAAK,MAAM;YAE1C,MAAM,IAAI,KAAK,CAAC,aAAa,CAAC,CAAA;QAChC,IACE,CAAC,CAAC,KAAK,GAAG,CAAC,CAAC;YACZ,CAAC,IAAI,CAAC,SAAS,CAAC,KAAK,GAAG,EAAE,EAAE,IAAI,CAAC,KAAK,IAAI,CAAC,SAAS,CAAC,MAAM,GAAG,EAAE,EAAE,IAAI,CAAC;gBACrE,IAAI,CAAC,SAAS,CAAC,KAAK,GAAG,EAAE,EAAE,IAAI,CAAC,KAAK,UAAU;gBAC/C,IAAI,CAAC,SAAS,CAAC,KAAK,GAAG,EAAE,EAAE,IAAI,CAAC,KAAK,IAAI,CAAC;YAE5C,MAAM,IAAI,KAAK,CAAC,aAAa,CAAC,CAAA;QAChC,OAAO,CAAC,GAAG,CAAC,IAAI,EAAE,EAAE,IAAI,EAAE,GAAG,EAAE,IAAI,CAAC,SAAS,CAAC,MAAM,GAAG,EAAE,EAAE,IAAI,CAAC,EAAE,CAAC,CAAA;QACnE,MAAM,GAAG,IAAI,CAAA;IACf,CAAC;IACD,IAAI,MAAM,KAAK,IAAI;QAAE,MAAM,IAAI,KAAK,CAAC,aAAa,CAAC,CAAA;IACnD,OAAO,OAAO,CAAA;AAChB,CAAC,CAAA;AAED,MAAM,CAAC,MAAM,aAAa,GAAG,KAAK,EAAE,KAAiB,EAAE,MAAoB,EAAoC,EAAE;IAC/G,SAAS,CAAC,MAAM,CAAC,CAAA;IACjB,MAAM,OAAO,GAAG,SAAS,CAAC,KAAK,CAAC,CAAA;IAChC,MAAM,MAAM,GAAG,IAAI,GAAG,EAAsB,CAAA;IAC5C,MAAM,IAAI,GAAG,IAAI,GAAG,EAAU,CAAA;IAC9B,MAAM,IAAI,GAAG,IAAI,GAAG,EAAU,CAAA;IAC9B,IAAI,QAAQ,GAAG,CAAC,CAAA;IAChB,MAAM,OAAO,GAAG,IAAI,KAAK,CAAC,CAAC,IAAI,EAAE,EAAE;QACjC,MAAM,QAAQ,GAAG,OAAO,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;QACvC,IAAI,CAAC,QAAQ,IAAI,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC;YAAE,MAAM,IAAI,KAAK,CAAC,aAAa,CAAC,CAAA;QACpE,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;QACnB,MAAM,IAAI,GAAG,+BAA+B,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,IAAI,CAAC,IAAI,KAAK,UAAU,CAAA;QACxF,IAAI,IAAI,GAAG,CAAC,CAAA;QACZ,IAAI,QAAQ,GAAG,CAAC,CAAA;QAChB,MAAM,KAAK,GAAiB,EAAE,CAAA;QAC9B,IAAI,CAAC,MAAM,GAAG,CAAC,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE,EAAE;YACnC,IAAI,KAAK;gBAAE,MAAM,KAAK,CAAA;YACtB,IAAI,IAAI,IAAI,CAAC,MAAM,CAAA;YACnB,QAAQ,IAAI,IAAI,CAAC,MAAM,CAAA;YACvB,IAAI,QAAQ,GAAG,YAAY,IAAI,CAAC,IAAI,IAAI,IAAI,GAAG,QAAQ,CAAC;gBAAE,MAAM,IAAI,WAAW,EAAE,CAAA;YACjF,QAAQ,GAAG,KAAK,CAAC,IAAI,EAAE,QAAQ,CAAC,CAAA;YAChC,IAAI,IAAI;gBAAE,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,KAAK,EAAE,CAAC,CAAA;YAClC,IAAI,KAAK,EAAE,CAAC;gBACV,IAAI,IAAI,KAAK,QAAQ,CAAC,IAAI,IAAI,QAAQ,KAAK,QAAQ,CAAC,GAAG;oBAAE,MAAM,IAAI,KAAK,CAAC,aAAa,CAAC,CAAA;gBACvF,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;gBACnB,IAAI,IAAI,EAAE,CAAC;oBACT,MAAM,OAAO,GAAG,IAAI,UAAU,CAAC,IAAI,CAAC,CAAA;oBACpC,IAAI,MAAM,GAAG,CAAC,CAAA;oBACd,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;wBACzB,OAAO,CAAC,GAAG,CAAC,IAAI,EAAE,MAAM,CAAC,CAAA;wBACzB,MAAM,IAAI,IAAI,CAAC,MAAM,CAAA;oBACvB,CAAC;oBACD,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,EAAE,OAAO,CAAC,CAAA;oBAC9B,KAAK,CAAC,MAAM,GAAG,CAAC,CAAA;gBAClB,CAAC;YACH,CAAC;QACH,CAAC,CAAA;QACD,IAAI,CAAC,KAAK,EAAE,CAAA;IACd,CAAC,CAAC,CAAA;IACF,OAAO,CAAC,QAAQ,CAAC,YAAY,CAAC,CAAA;IAC9B,gFAAgF;IAChF,KAAK,IAAI,MAAM,GAAG,CAAC,EAAE,MAAM,GAAG,KAAK,CAAC,MAAM,EAAE,MAAM,IAAI,IAAI,EAAE,CAAC;QAC3D,SAAS,CAAC,MAAM,CAAC,CAAA;QACjB,OAAO,CAAC,IAAI,CAAC,KAAK,CAAC,QAAQ,CAAC,MAAM,EAAE,MAAM,GAAG,IAAI,CAAC,EAAE,MAAM,GAAG,IAAI,IAAI,KAAK,CAAC,MAAM,CAAC,CAAA;QAClF,IAAI,MAAM,GAAG,MAAM,KAAK,CAAC;YAAE,MAAM,IAAI,EAAE,CAAA;IACzC,CAAC;IACD,SAAS,CAAC,MAAM,CAAC,CAAA;IACjB,IAAI,IAAI,CAAC,IAAI,KAAK,OAAO,CAAC,IAAI,IAAI,IAAI,CAAC,IAAI,KAAK,OAAO,CAAC,IAAI;QAAE,MAAM,IAAI,KAAK,CAAC,gBAAgB,CAAC,CAAA;IAC/F,OAAO,MAAM,CAAA;AACf,CAAC,CAAA;AAED,MAAM,QAAQ,GAAG,CAAC,IAAY,EAAE,EAAE;IAChC,IAAI,qBAAqB,CAAC,IAAI,CAAC,IAAI,CAAC;QAAE,MAAM,IAAI,KAAK,CAAC,iBAAiB,CAAC,CAAA;IACxE,IAAI,KAAK,GAAG,CAAC,CAAA;IACb,IAAI,KAAK,GAAG,CAAC,CAAA;IACb,KAAK,IAAI,KAAK,GAAG,CAAC,EAAE,KAAK,GAAG,IAAI,CAAC,MAAM,EAAE,KAAK,EAAE,EAAE,CAAC;QACjD,IAAI,IAAI,CAAC,KAAK,CAAC,KAAK,GAAG;YAAE,SAAQ;QACjC,MAAM,IAAI,GAAG,IAAI,CAAC,UAAU,CAAC,MAAM,EAAE,KAAK,CAAC;YACzC,CAAC,CAAC,KAAK;YACP,CAAC,CAAC,IAAI,CAAC,UAAU,CAAC,WAAW,EAAE,KAAK,CAAC;gBACnC,CAAC,CAAC,KAAK;gBACP,CAAC,CAAC,IAAI,CAAC,UAAU,CAAC,IAAI,EAAE,KAAK,CAAC;oBAC5B,CAAC,CAAC,IAAI;oBACN,CAAC,CAAC,SAAS,CAAA;QACjB,IAAI,IAAI,EAAE,CAAC;YACT,IAAI,EAAE,KAAK,GAAG,OAAO;gBAAE,MAAM,IAAI,WAAW,EAAE,CAAA;YAC9C,MAAM,GAAG,GAAG,IAAI,CAAC,OAAO,CAAC,IAAI,EAAE,KAAK,GAAG,CAAC,CAAC,CAAA;YACzC,IAAI,GAAG,GAAG,CAAC;gBAAE,MAAM,IAAI,KAAK,CAAC,aAAa,CAAC,CAAA;YAC3C,KAAK,GAAG,GAAG,GAAG,IAAI,CAAC,MAAM,GAAG,CAAC,CAAA;YAC7B,SAAQ;QACV,CAAC;QACD,MAAM,KAAK,GAAG,KAAK,CAAA;QACnB,IAAI,KAAK,GAAG,EAAE,CAAA;QACd,KAAK,KAAK,EAAE,EAAE,KAAK,GAAG,IAAI,CAAC,MAAM,EAAE,KAAK,EAAE,EAAE,CAAC;YAC3C,MAAM,IAAI,GAAG,IAAI,CAAC,KAAK,CAAC,CAAA;YACxB,IAAI,KAAK,EAAE,CAAC;gBACV,IAAI,IAAI,KAAK,KAAK;oBAAE,KAAK,GAAG,EAAE,CAAA;YAChC,CAAC;iBAAM,IAAI,IAAI,KAAK,GAAG,IAAI,IAAI,KAAK,GAAG;gBAAE,KAAK,GAAG,IAAI,CAAA;iBAChD,IAAI,IAAI,KAAK,GAAG;gBAAE,MAAK;QAC9B,CAAC;QACD,IAAI,KAAK,KAAK,IAAI,CAAC,MAAM;YAAE,MAAM,IAAI,KAAK,CAAC,aAAa,CAAC,CAAA;QACzD,MAAM,GAAG,GAAG,IAAI,CAAC,KAAK,CAAC,KAAK,EAAE,KAAK,GAAG,CAAC,CAAC,CAAA;QACxC,IAAI,GAAG,CAAC,UAAU,CAAC,IAAI,CAAC;YAAE,KAAK,EAAE,CAAA;aAC5B,IAAI,CAAC,SAAS,CAAC,IAAI,CAAC,GAAG,CAAC;YAAE,KAAK,EAAE,CAAA;QACtC,IAAI,EAAE,KAAK,GAAG,OAAO,IAAI,KAAK,GAAG,GAAG;YAAE,MAAM,IAAI,WAAW,EAAE,CAAA;QAC7D,IAAI,KAAK,GAAG,CAAC;YAAE,MAAM,IAAI,KAAK,CAAC,aAAa,CAAC,CAAA;IAC/C,CAAC;IACD,IAAI,KAAK,KAAK,CAAC;QAAE,MAAM,IAAI,KAAK,CAAC,aAAa,CAAC,CAAA;AACjD,CAAC,CAAA;AAED,MAAM,CAAC,MAAM,OAAO,GAAG,CAAC,OAAgC,EAAE,IAAY,EAAY,EAAE;IAClF,MAAM,KAAK,GAAG,OAAO,CAAC,GAAG,CAAC,IAAI,CAAC,CAAA;IAC/B,IAAI,CAAC,KAAK;QAAE,MAAM,IAAI,KAAK,CAAC,cAAc,CAAC,CAAA;IAC3C,IAAI,IAAI,GAAG,SAAS,CAAC,KAAK,CAAC,CAAA;IAC3B,IAAI,WAAW,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;QAC3B,IAAI,KAAK,GAAG,CAAC,CAAA;QACb,OAAO,KAAK,GAAG,IAAI,CAAC,MAAM,EAAE,CAAC;YAC3B,KAAK,IAAI,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,MAAM,IAAI,CAAC,CAAA;YACxD,MAAM,GAAG,GAAG,IAAI,CAAC,UAAU,CAAC,IAAI,EAAE,KAAK,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,UAAU,CAAC,MAAM,EAAE,KAAK,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,SAAS,CAAA;YACpG,IAAI,CAAC,GAAG;gBAAE,MAAK;YACf,MAAM,KAAK,GAAG,IAAI,CAAC,OAAO,CAAC,GAAG,EAAE,KAAK,GAAG,CAAC,CAAC,CAAA;YAC1C,IAAI,KAAK,GAAG,CAAC;gBAAE,MAAM,IAAI,KAAK,CAAC,aAAa,CAAC,CAAA;YAC7C,KAAK,GAAG,KAAK,GAAG,GAAG,CAAC,MAAM,CAAA;QAC5B,CAAC;QACD,MAAM,OAAO,GAAG,wBAAwB,CAAC,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,CAAA;QAChE,IAAI,OAAO;YAAE,IAAI,GAAG,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,KAAK,CAAC,GAAG,IAAI,CAAC,KAAK,CAAC,KAAK,GAAG,OAAO,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,CAAA;IAClF,CAAC;IACD,QAAQ,CAAC,IAAI,CAAC,CAAA;IACd,MAAM,QAAQ,GAAG,IAAI,SAAS,CAAC;QAC7B,OAAO,EAAE,GAAG,EAAE;YACZ,MAAM,IAAI,KAAK,CAAC,aAAa,CAAC,CAAA;QAChC,CAAC;KACF,CAAC,CAAC,eAAe,CAAC,IAAI,EAAE,iBAAiB,CAAC,CAAA;IAC3C,IAAI,CAAC,QAAQ,CAAC,eAAe;QAAE,MAAM,IAAI,KAAK,CAAC,aAAa,CAAC,CAAA;IAC7D,OAAO,QAAQ,CAAA;AACjB,CAAC,CAAA;AAED,MAAM,CAAC,MAAM,QAAQ,GAAG,CAAC,IAAU,EAAE,IAAY,EAAE,SAAiB,EAAa,EAAE;IACjF,MAAM,KAAK,GAAc,EAAE,CAAA;IAC3B,MAAM,KAAK,GAAG,CAAC,OAAa,EAAE,EAAE;QAC9B,IAAI,OAAO,CAAC,QAAQ,KAAK,CAAC,IAAI,OAAO,CAAC,SAAS,KAAK,IAAI,IAAI,OAAO,CAAC,YAAY,KAAK,SAAS;YAC5F,KAAK,CAAC,IAAI,CAAC,OAAkB,CAAC,CAAA;QAChC,KAAK,IAAI,KAAK,GAAG,OAAO,CAAC,UAAU,EAAE,KAAK,EAAE,KAAK,GAAG,KAAK,CAAC,WAAW;YAAE,KAAK,CAAC,KAAK,CAAC,CAAA;IACrF,CAAC,CAAA;IACD,KAAK,CAAC,IAAI,CAAC,CAAA;IACX,OAAO,KAAK,CAAA;AACd,CAAC,CAAA"}

package/dist/attachments/documents.d.ts

@@ -0,0 +1,6 @@
+import type { Extraction, FileHint } from "./extract.js";
+export type DocumentKind = "odt" | "ods" | "xlsx" | "pptx" | "epub";
+export declare const documentKind: (hint: FileHint) => DocumentKind | undefined;
+export declare const documentExtractor: (kind: DocumentKind) => string;
+export declare const readDocument: (bytes: Uint8Array, kind: DocumentKind, signal?: AbortSignal) => Promise<Extraction>;
+//# sourceMappingURL=documents.d.ts.map

package/dist/attachments/documents.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"documents.d.ts","sourceRoot":"","sources":["../../src/attachments/documents.ts"],"names":[],"mappings":"AAIA,OAAO,KAAK,EAAE,UAAU,EAAE,QAAQ,EAAE,MAAM,cAAc,CAAA;AAexD,MAAM,MAAM,YAAY,GAAG,KAAK,GAAG,KAAK,GAAG,MAAM,GAAG,MAAM,GAAG,MAAM,CAAA;AASnE,eAAO,MAAM,YAAY,SAAU,QAAQ,KAAG,YAAY,GAAG,SAQ5D,CAAA;AACD,eAAO,MAAM,iBAAiB,SAAU,YAAY,WAA0B,CAAA;AA2R9E,eAAO,MAAM,YAAY,UAChB,UAAU,QACX,YAAY,WACT,WAAW,KACnB,OAAO,CAAC,UAAU,CAgBpB,CAAA"}

package/dist/attachments/documents.js

@@ -0,0 +1,355 @@
+import { posix } from "node:path";
+import { CliError } from "@leemour/cli-core";
+import { elements, ReaderLimit, readContainer, xmlPart } from "./container.js";
+import { MAX_TEXT_CHARS as MAX_TEXT } from "./limits.js";
+const OFFICE = "urn:oasis:names:tc:opendocument:xmlns:office:1.0";
+const TEXT = "urn:oasis:names:tc:opendocument:xmlns:text:1.0";
+const TABLE = "urn:oasis:names:tc:opendocument:xmlns:table:1.0";
+const PACKAGE_REL = "http://schemas.openxmlformats.org/package/2006/relationships";
+const REL = "http://schemas.openxmlformats.org/officeDocument/2006/relationships";
+const STRICT_REL = "http://purl.oclc.org/ooxml/officeDocument/relationships";
+const SHEET = "http://schemas.openxmlformats.org/spreadsheetml/2006/main";
+const SLIDE = "http://schemas.openxmlformats.org/presentationml/2006/main";
+const DRAW = "http://schemas.openxmlformats.org/drawingml/2006/main";
+const OPF = "http://www.idpf.org/2007/opf";
+const XHTML = "http://www.w3.org/1999/xhtml";
+const TYPES = {
+    odt: "application/vnd.oasis.opendocument.text",
+    ods: "application/vnd.oasis.opendocument.spreadsheet",
+    xlsx: "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
+    pptx: "application/vnd.openxmlformats-officedocument.presentationml.presentation",
+    epub: "application/epub+zip",
+};
+export const documentKind = (hint) => {
+    const extension = posix
+        .extname(hint.name ?? hint.path)
+        .slice(1)
+        .toLowerCase();
+    return Object.keys(TYPES).find((kind) => kind === extension || TYPES[kind] === hint.mime?.split(";")[0]?.trim().toLowerCase());
+};
+export const documentExtractor = (kind) => `document:v1:${kind}`;
+class Output {
+    parts = [];
+    length = 0;
+    real = false;
+    add(value, source = true) {
+        this.length += value.length;
+        if (this.length > MAX_TEXT)
+            throw new ReaderLimit();
+        this.parts.push(value);
+        if (source && value.trim())
+            this.real = true;
+    }
+    section(label) {
+        this.add(`\n\n[${label}]\n`, false);
+    }
+    text() {
+        return this.parts.join("").trim();
+    }
+}
+const attribute = (node, name, namespace) => (namespace ? node.getAttributeNS(namespace, name) : node.getAttribute(name)) ?? "";
+const required = (value) => {
+    if (!value)
+        throw new Error("missing_value");
+    return value;
+};
+const count = (value) => {
+    if (!value)
+        return 1;
+    if (!/^[1-9]\d{0,7}$/.test(value))
+        throw new ReaderLimit();
+    return Number(value);
+};
+const rootIs = (node, name, namespaces) => {
+    const root = node.nodeType === 9 ? node.firstChild : node;
+    // XML declarations/comments can precede the document element.
+    let element = root;
+    while (element && element.nodeType !== 1)
+        element = element.nextSibling;
+    if (!element || element.localName !== name || !namespaces.includes(element.namespaceURI ?? ""))
+        throw new Error("wrong_format");
+    return element.namespaceURI;
+};
+const target = (owner, reference) => {
+    const decoded = decodeURIComponent(reference.split(/[?#]/)[0] ?? "");
+    if (/^[a-z][a-z0-9+.-]*:|^[\\/]|\\/i.test(decoded) || [...decoded].some((character) => character.charCodeAt(0) < 32))
+        throw new Error("external_part");
+    const path = posix.normalize(posix.join(posix.dirname(owner), decoded));
+    if (path.startsWith("../") || !path || path === ".")
+        throw new Error("external_part");
+    return path;
+};
+const relationships = (archive, owner) => {
+    const path = owner ? posix.join(posix.dirname(owner), "_rels", `${posix.basename(owner)}.rels`) : "_rels/.rels";
+    const document = xmlPart(archive, path);
+    rootIs(document, "Relationships", [PACKAGE_REL]);
+    const result = new Map();
+    for (const entry of elements(document, "Relationship", PACKAGE_REL)) {
+        if (attribute(entry, "TargetMode") === "External")
+            continue;
+        const id = required(attribute(entry, "Id"));
+        if (result.has(id))
+            throw new Error("duplicate_relationship");
+        result.set(id, {
+            path: target(owner, required(attribute(entry, "Target"))),
+            type: required(attribute(entry, "Type")),
+        });
+    }
+    return result;
+};
+const relationId = (element) => required(attribute(element, "id", REL) || attribute(element, "id", STRICT_REL));
+const mainPart = (archive, contentType) => {
+    const types = xmlPart(archive, "[Content_Types].xml");
+    const namespace = "http://schemas.openxmlformats.org/package/2006/content-types";
+    rootIs(types, "Types", [namespace]);
+    const matches = [...relationships(archive, "").values()].filter((entry) => entry.type.endsWith("/officeDocument"));
+    if (matches.length !== 1)
+        throw new Error("missing_main_part");
+    const path = matches[0]?.path;
+    if (!elements(types, "Override", namespace).some((entry) => attribute(entry, "PartName") === `/${path}` && attribute(entry, "ContentType") === contentType))
+        throw new Error("wrong_format");
+    return path;
+};
+const odfText = (node) => {
+    if (node.nodeType === 3 || node.nodeType === 4)
+        return node.nodeValue ?? "";
+    if (node.nodeType === 1 && node.namespaceURI === TEXT) {
+        if (node.localName === "s") {
+            const repeated = count(attribute(node, "c", TEXT));
+            if (repeated > MAX_TEXT)
+                throw new ReaderLimit();
+            return " ".repeat(repeated);
+        }
+        if (node.localName === "tab")
+            return "\t";
+        if (node.localName === "line-break")
+            return "\n";
+    }
+    let value = "";
+    for (let child = node.firstChild; child; child = child.nextSibling) {
+        value += odfText(child);
+        if (value.length > MAX_TEXT)
+            throw new ReaderLimit();
+    }
+    return value;
+};
+const odfParagraphs = (node) => {
+    const result = [];
+    const visit = (current) => {
+        if (current.nodeType === 1 &&
+            current.namespaceURI === TEXT &&
+            (current.localName === "p" || current.localName === "h")) {
+            result.push(current);
+            return;
+        }
+        for (let child = current.firstChild; child; child = child.nextSibling)
+            visit(child);
+    };
+    visit(node);
+    return result;
+};
+const readOdf = (archive, kind, output) => {
+    if (new TextDecoder().decode(archive.get("mimetype")) !== TYPES[kind])
+        throw new Error("wrong_format");
+    const document = xmlPart(archive, "content.xml");
+    rootIs(document, "document-content", [OFFICE]);
+    if (archive.has("META-INF/manifest.xml")) {
+        const manifest = xmlPart(archive, "META-INF/manifest.xml");
+        if (elements(manifest, "encryption-data", "urn:oasis:names:tc:opendocument:xmlns:manifest:1.0").length)
+            throw new Error("encrypted_document");
+    }
+    const body = elements(document, kind === "odt" ? "text" : "spreadsheet", OFFICE);
+    if (body.length !== 1)
+        throw new Error("wrong_format");
+    if (kind === "odt") {
+        for (const paragraph of odfParagraphs(body[0]))
+            output.add(`${odfText(paragraph)}\n`);
+        return;
+    }
+    for (const sheet of elements(body[0], "table", TABLE)) {
+        output.section(`Sheet: ${required(attribute(sheet, "name", TABLE))}`);
+        let rowNumber = 1;
+        for (const row of elements(sheet, "table-row", TABLE)) {
+            const repeatedRows = count(attribute(row, "number-rows-repeated", TABLE));
+            const values = [];
+            let column = 1;
+            for (let child = row.firstChild; child; child = child.nextSibling) {
+                if (child.nodeType !== 1 ||
+                    child.namespaceURI !== TABLE ||
+                    !["table-cell", "covered-table-cell"].includes(child.localName ?? ""))
+                    continue;
+                const cell = child;
+                const repeats = count(attribute(cell, "number-columns-repeated", TABLE));
+                const paragraphs = odfParagraphs(cell).map(odfText).join("\n");
+                const value = paragraphs ||
+                    attribute(cell, "string-value", OFFICE) ||
+                    attribute(cell, "value", OFFICE) ||
+                    attribute(cell, "date-value", OFFICE) ||
+                    attribute(cell, "time-value", OFFICE) ||
+                    attribute(cell, "boolean-value", OFFICE);
+                const formula = attribute(cell, "formula", TABLE);
+                if (value || formula)
+                    values.push({ column, repeats, value: `${value}${formula ? ` [formula: ${formula}]` : ""}` });
+                column += repeats;
+            }
+            if (values.length) {
+                for (let repeat = 0; repeat < repeatedRows; repeat++) {
+                    for (const cell of values) {
+                        if (cell.repeats * repeatedRows * (cell.value.length + 5) > MAX_TEXT)
+                            throw new ReaderLimit();
+                        for (let offset = 0; offset < cell.repeats; offset++)
+                            output.add(`R${rowNumber + repeat}C${cell.column + offset}: ${cell.value}\n`);
+                    }
+                }
+            }
+            rowNumber += repeatedRows;
+        }
+    }
+};
+const readSheet = (archive, output) => {
+    const path = mainPart(archive, `${TYPES.xlsx}.main+xml`);
+    const document = xmlPart(archive, path);
+    const namespace = rootIs(document, "workbook", [SHEET, "http://purl.oclc.org/ooxml/spreadsheetml/main"]);
+    const rels = relationships(archive, path);
+    const shared = [...rels.values()].find((entry) => entry.type.endsWith("/sharedStrings"));
+    const strings = shared
+        ? elements(xmlPart(archive, shared.path), "si", namespace).map((entry) => elements(entry, "t", namespace)
+            .map((part) => part.textContent ?? "")
+            .join(""))
+        : [];
+    for (const sheet of elements(document, "sheet", namespace)) {
+        output.section(`Sheet: ${required(attribute(sheet, "name"))}`);
+        const relationship = rels.get(relationId(sheet));
+        if (!relationship?.type.endsWith("/worksheet"))
+            throw new Error("missing_sheet");
+        const worksheet = xmlPart(archive, relationship.path);
+        rootIs(worksheet, "worksheet", [namespace]);
+        for (const cell of elements(worksheet, "c", namespace)) {
+            const reference = required(attribute(cell, "r"));
+            if (!/^[A-Z]+[1-9]\d*$/.test(reference))
+                throw new Error("invalid_cell");
+            const type = attribute(cell, "t");
+            const saved = elements(cell, "v", namespace)[0]?.textContent ?? "";
+            let value = type === "inlineStr"
+                ? elements(cell, "t", namespace)
+                    .map((part) => part.textContent ?? "")
+                    .join("")
+                : saved;
+            if (type === "s") {
+                if (!/^\d+$/.test(saved) || strings[Number(saved)] === undefined)
+                    throw new Error("missing_shared_string");
+                value = strings[Number(saved)];
+            }
+            const formula = elements(cell, "f", namespace)[0]?.textContent;
+            if (value || formula)
+                output.add(`${reference}: ${value}${formula ? ` [formula: ${formula}]` : ""}\n`);
+        }
+    }
+};
+const drawingText = (node, namespace) => {
+    if (node.nodeType === 1 && node.namespaceURI === namespace) {
+        if (node.localName === "t")
+            return node.textContent ?? "";
+        if (node.localName === "br")
+            return "\n";
+        if (node.localName === "tab")
+            return "\t";
+    }
+    let value = "";
+    for (let child = node.firstChild; child; child = child.nextSibling) {
+        value += drawingText(child, namespace);
+        if (value.length > MAX_TEXT)
+            throw new ReaderLimit();
+    }
+    return value;
+};
+const readSlides = (archive, output) => {
+    const path = mainPart(archive, `${TYPES.pptx}.main+xml`);
+    const document = xmlPart(archive, path);
+    const namespace = rootIs(document, "presentation", [SLIDE, "http://purl.oclc.org/ooxml/presentationml/main"]);
+    const rels = relationships(archive, path);
+    let number = 0;
+    for (const slide of elements(document, "sldId", namespace)) {
+        const relationship = rels.get(relationId(slide));
+        if (!relationship?.type.endsWith("/slide"))
+            throw new Error("missing_slide");
+        const content = xmlPart(archive, relationship.path);
+        rootIs(content, "sld", [namespace]);
+        output.section(`Slide ${++number}`);
+        const drawing = namespace === SLIDE ? DRAW : "http://purl.oclc.org/ooxml/drawingml/main";
+        for (const paragraph of elements(content, "p", drawing))
+            output.add(`${drawingText(paragraph, drawing)}\n`);
+    }
+};
+const readBook = (archive, output) => {
+    if (new TextDecoder().decode(archive.get("mimetype")) !== TYPES.epub || archive.has("META-INF/encryption.xml"))
+        throw new Error("wrong_format");
+    const container = xmlPart(archive, "META-INF/container.xml");
+    const roots = elements(container, "rootfile", "urn:oasis:names:tc:opendocument:xmlns:container");
+    const root = roots.find((entry) => attribute(entry, "media-type") === "application/oebps-package+xml");
+    const path = target("", required(root ? attribute(root, "full-path") : ""));
+    const book = xmlPart(archive, path);
+    rootIs(book, "package", [OPF]);
+    const manifest = new Map();
+    for (const item of elements(book, "item", OPF)) {
+        const id = required(attribute(item, "id"));
+        if (manifest.has(id))
+            throw new Error("duplicate_chapter");
+        manifest.set(id, { path: target(path, required(attribute(item, "href"))), type: attribute(item, "media-type") });
+    }
+    let number = 0;
+    const chapters = elements(book, "itemref", OPF);
+    if (!chapters.length)
+        throw new Error("missing_chapters");
+    for (const chapter of chapters) {
+        const item = manifest.get(required(attribute(chapter, "idref")));
+        if (item?.type !== "application/xhtml+xml")
+            throw new Error("unsupported_chapter");
+        const document = xmlPart(archive, item.path);
+        const bodies = elements(document, "body", XHTML);
+        if (bodies.length !== 1)
+            throw new Error("missing_chapter");
+        output.section(`Chapter ${++number}`);
+        const visit = (node) => {
+            if (node.nodeType === 3 || node.nodeType === 4) {
+                output.add(node.nodeValue ?? "");
+                return;
+            }
+            if (node.nodeType === 1 && ["script", "style", "head"].includes(node.localName ?? ""))
+                return;
+            for (let child = node.firstChild; child; child = child.nextSibling)
+                visit(child);
+            if (node.nodeType === 1 && ["td", "th"].includes(node.localName ?? ""))
+                output.add("\t", false);
+            if (node.nodeType === 1 && ["p", "div", "li", "h1", "h2", "h3", "br", "tr"].includes(node.localName ?? ""))
+                output.add("\n", false);
+        };
+        visit(bodies[0]);
+    }
+};
+export const readDocument = async (bytes, kind, signal) => {
+    const extractor = documentExtractor(kind);
+    try {
+        const archive = await readContainer(bytes, signal);
+        const output = new Output();
+        if (kind === "odt" || kind === "ods")
+            readOdf(archive, kind, output);
+        else if (kind === "xlsx")
+            readSheet(archive, output);
+        else if (kind === "pptx")
+            readSlides(archive, output);
+        else
+            readBook(archive, output);
+        if (signal?.aborted)
+            throw new CliError("cancelled", "document extraction cancelled");
+        return output.real ? { status: "extracted", extractor, text: output.text() } : { status: "needs-agent", extractor };
+    }
+    catch (error) {
+        if (error instanceof CliError)
+            throw error;
+        if (error instanceof ReaderLimit)
+            return { status: "too-large" };
+        return { status: "unreadable", extractor, error: "invalid_document" };
+    }
+};
+//# sourceMappingURL=documents.js.map