p-spore 0.1.3 → 0.1.42

package/dist/documents/ole.js

@@ -0,0 +1,189 @@
+/**
+ * The container every pre-2007 Office file is written in.
+ *
+ * A `.doc`, `.xls` or `.ppt` is a Compound File Binary document: a FAT-style
+ * filesystem inside one file, whose named streams hold the actual format. The
+ * reader is here so the three legacy parsers share one implementation of the
+ * container and only differ in how they read their own stream.
+ */
+const SIGNATURE = Buffer.from([0xd0, 0xcf, 0x11, 0xe0, 0xa1, 0xb1, 0x1a, 0xe1]);
+/** Sector chain terminators; anything below them is a real sector number. */
+const MAX_REGULAR_SECTOR = 0xfffffffa;
+const FREE_SECTOR = 0xffffffff;
+const END_OF_CHAIN = 0xfffffffe;
+const DIRECTORY_ENTRY_BYTES = 128;
+const STREAM_TYPE = 2;
+/** DIFAT entries stored in the header before the chain of DIFAT sectors starts. */
+const HEADER_DIFAT_ENTRIES = 109;
+/**
+ * Ceiling on how many sectors one chain may visit. A crafted file can point a
+ * chain at itself and read forever, and these documents are untrusted.
+ */
+const MAX_CHAIN_SECTORS = 1 << 22;
+export class OleDocument {
+    #buffer;
+    #sectorSize;
+    #miniSectorSize;
+    #miniStreamCutoff;
+    #fat;
+    #miniFat;
+    #streams;
+    #miniStream;
+    constructor(buffer) {
+        this.#buffer = buffer;
+        this.#sectorSize = 1 << buffer.readUInt16LE(30);
+        this.#miniSectorSize = 1 << buffer.readUInt16LE(32);
+        this.#miniStreamCutoff = buffer.readUInt32LE(56);
+        if (this.#sectorSize < 128 || this.#sectorSize > 1 << 20) {
+            throw new Error("Compound document declares an unusable sector size");
+        }
+        this.#fat = this.#readFat();
+        this.#miniFat = this.#readChainValues(buffer.readUInt32LE(60), buffer.readUInt32LE(64));
+        const directory = this.#readDirectory();
+        this.#miniStream = directory.miniStream;
+        this.#streams = directory.streams;
+    }
+    static open(buffer) {
+        if (!looksLikeOle(buffer))
+            throw new Error("Not a compound document");
+        return new OleDocument(buffer);
+    }
+    /** Stream names present in the document, for diagnostics and format sniffing. */
+    names() {
+        return [...this.#streams.keys()];
+    }
+    has(name) {
+        return this.#streams.has(name);
+    }
+    /** Reads a whole named stream, or undefined when the document has no such stream. */
+    read(name) {
+        const entry = this.#streams.get(name);
+        if (!entry)
+            return undefined;
+        // Short streams live packed inside the mini stream rather than in sectors of
+        // their own, which is why the cutoff has to be consulted before reading.
+        return entry.size < this.#miniStreamCutoff
+            ? this.#readMini(entry.start, entry.size)
+            : this.#readSectors(entry.start, entry.size);
+    }
+    #readFat() {
+        const count = this.#buffer.readUInt32LE(44);
+        const sectors = [];
+        for (let index = 0; index < Math.min(count, HEADER_DIFAT_ENTRIES); index += 1) {
+            const sector = this.#buffer.readUInt32LE(76 + index * 4);
+            if (sector <= MAX_REGULAR_SECTOR)
+                sectors.push(sector);
+        }
+        // Documents with more than 109 FAT sectors continue the list in a chain of
+        // DIFAT sectors, each ending with a pointer to the next.
+        let difat = this.#buffer.readUInt32LE(68);
+        const difatCount = this.#buffer.readUInt32LE(72);
+        const perSector = this.#sectorSize / 4 - 1;
+        for (let visited = 0; difat <= MAX_REGULAR_SECTOR && visited < difatCount + 1; visited += 1) {
+            const offset = this.#sectorOffset(difat);
+            for (let index = 0; index < perSector; index += 1) {
+                const sector = this.#buffer.readUInt32LE(offset + index * 4);
+                if (sector <= MAX_REGULAR_SECTOR)
+                    sectors.push(sector);
+            }
+            difat = this.#buffer.readUInt32LE(offset + perSector * 4);
+        }
+        const fat = [];
+        for (const sector of sectors) {
+            const offset = this.#sectorOffset(sector);
+            for (let index = 0; index < this.#sectorSize / 4; index += 1) {
+                fat.push(this.#buffer.readUInt32LE(offset + index * 4));
+            }
+        }
+        return fat;
+    }
+    /** Follows a chain and returns every 32-bit value the visited sectors hold. */
+    #readChainValues(start, expectedSectors) {
+        const values = [];
+        let sector = start;
+        for (let visited = 0; sector <= MAX_REGULAR_SECTOR && visited <= expectedSectors; visited += 1) {
+            const offset = this.#sectorOffset(sector);
+            for (let index = 0; index < this.#sectorSize / 4; index += 1) {
+                values.push(this.#buffer.readUInt32LE(offset + index * 4));
+            }
+            sector = this.#fat[sector] ?? END_OF_CHAIN;
+        }
+        return values;
+    }
+    #readDirectory() {
+        const directory = this.#readSectors(this.#buffer.readUInt32LE(48), Number.POSITIVE_INFINITY);
+        const streams = new Map();
+        let miniStreamStart = END_OF_CHAIN;
+        let miniStreamSize = 0;
+        for (let offset = 0; offset + DIRECTORY_ENTRY_BYTES <= directory.length; offset += DIRECTORY_ENTRY_BYTES) {
+            const nameBytes = directory.readUInt16LE(offset + 64);
+            if (nameBytes < 2 || nameBytes > 64)
+                continue;
+            // The stored length counts the terminating NUL, and names are UTF-16LE.
+            const name = directory.subarray(offset + 0, offset + nameBytes - 2).toString("utf16le");
+            const type = directory.readUInt8(offset + 66);
+            const start = directory.readUInt32LE(offset + 116);
+            const size = Number(directory.readBigUInt64LE(offset + 120));
+            // Entry zero is the root, and its "stream" is where mini sectors live.
+            if (offset === 0) {
+                miniStreamStart = start;
+                miniStreamSize = size;
+                continue;
+            }
+            if (type !== STREAM_TYPE || size <= 0)
+                continue;
+            if (!streams.has(name))
+                streams.set(name, { start, size });
+        }
+        const miniStream = miniStreamStart <= MAX_REGULAR_SECTOR
+            ? this.#readSectors(miniStreamStart, miniStreamSize)
+            : Buffer.alloc(0);
+        return { streams, miniStream };
+    }
+    #readSectors(start, size) {
+        const chunks = [];
+        let sector = start;
+        let remaining = size;
+        for (let visited = 0; sector <= MAX_REGULAR_SECTOR && remaining > 0 && visited < MAX_CHAIN_SECTORS; visited += 1) {
+            const offset = this.#sectorOffset(sector);
+            const take = Math.min(this.#sectorSize, remaining);
+            chunks.push(this.#buffer.subarray(offset, offset + take));
+            remaining -= take;
+            const next = this.#fat[sector];
+            if (next === undefined || next === FREE_SECTOR)
+                break;
+            sector = next;
+        }
+        return Buffer.concat(chunks);
+    }
+    #readMini(start, size) {
+        const chunks = [];
+        let sector = start;
+        let remaining = size;
+        for (let visited = 0; sector <= MAX_REGULAR_SECTOR && remaining > 0 && visited < MAX_CHAIN_SECTORS; visited += 1) {
+            const offset = sector * this.#miniSectorSize;
+            if (offset >= this.#miniStream.length)
+                break;
+            const take = Math.min(this.#miniSectorSize, remaining);
+            chunks.push(this.#miniStream.subarray(offset, offset + take));
+            remaining -= take;
+            const next = this.#miniFat[sector];
+            if (next === undefined || next === FREE_SECTOR)
+                break;
+            sector = next;
+        }
+        return Buffer.concat(chunks);
+    }
+    /** Sector numbering starts after the 512-byte header, whatever the sector size. */
+    #sectorOffset(sector) {
+        const offset = 512 + sector * this.#sectorSize;
+        if (offset < 0 || offset + this.#sectorSize > this.#buffer.length) {
+            throw new Error("Compound document points past the end of the file");
+        }
+        return offset;
+    }
+}
+export function looksLikeOle(buffer) {
+    return buffer.length >= 512 && buffer.subarray(0, 8).equals(SIGNATURE);
+}
+//# sourceMappingURL=ole.js.map

package/dist/documents/ole.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"ole.js","sourceRoot":"","sources":["../../src/documents/ole.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AAEH,MAAM,SAAS,GAAG,MAAM,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,CAAC,CAAC,CAAC;AAEhF,6EAA6E;AAC7E,MAAM,kBAAkB,GAAG,UAAU,CAAC;AACtC,MAAM,WAAW,GAAG,UAAU,CAAC;AAC/B,MAAM,YAAY,GAAG,UAAU,CAAC;AAEhC,MAAM,qBAAqB,GAAG,GAAG,CAAC;AAClC,MAAM,WAAW,GAAG,CAAC,CAAC;AAEtB,mFAAmF;AACnF,MAAM,oBAAoB,GAAG,GAAG,CAAC;AAEjC;;;GAGG;AACH,MAAM,iBAAiB,GAAG,CAAC,IAAI,EAAE,CAAC;AAElC,MAAM,OAAO,WAAW;IACb,OAAO,CAAS;IAChB,WAAW,CAAS;IACpB,eAAe,CAAS;IACxB,iBAAiB,CAAS;IAC1B,IAAI,CAAW;IACf,QAAQ,CAAW;IACnB,QAAQ,CAAuD;IAC/D,WAAW,CAAS;IAE7B,YAAoB,MAAc;QAChC,IAAI,CAAC,OAAO,GAAG,MAAM,CAAC;QACtB,IAAI,CAAC,WAAW,GAAG,CAAC,IAAI,MAAM,CAAC,YAAY,CAAC,EAAE,CAAC,CAAC;QAChD,IAAI,CAAC,eAAe,GAAG,CAAC,IAAI,MAAM,CAAC,YAAY,CAAC,EAAE,CAAC,CAAC;QACpD,IAAI,CAAC,iBAAiB,GAAG,MAAM,CAAC,YAAY,CAAC,EAAE,CAAC,CAAC;QACjD,IAAI,IAAI,CAAC,WAAW,GAAG,GAAG,IAAI,IAAI,CAAC,WAAW,GAAG,CAAC,IAAI,EAAE,EAAE,CAAC;YACzD,MAAM,IAAI,KAAK,CAAC,oDAAoD,CAAC,CAAC;QACxE,CAAC;QACD,IAAI,CAAC,IAAI,GAAG,IAAI,CAAC,QAAQ,EAAE,CAAC;QAC5B,IAAI,CAAC,QAAQ,GAAG,IAAI,CAAC,gBAAgB,CAAC,MAAM,CAAC,YAAY,CAAC,EAAE,CAAC,EAAE,MAAM,CAAC,YAAY,CAAC,EAAE,CAAC,CAAC,CAAC;QACxF,MAAM,SAAS,GAAG,IAAI,CAAC,cAAc,EAAE,CAAC;QACxC,IAAI,CAAC,WAAW,GAAG,SAAS,CAAC,UAAU,CAAC;QACxC,IAAI,CAAC,QAAQ,GAAG,SAAS,CAAC,OAAO,CAAC;IACpC,CAAC;IAED,MAAM,CAAC,IAAI,CAAC,MAAc;QACxB,IAAI,CAAC,YAAY,CAAC,MAAM,CAAC;YAAE,MAAM,IAAI,KAAK,CAAC,yBAAyB,CAAC,CAAC;QACtE,OAAO,IAAI,WAAW,CAAC,MAAM,CAAC,CAAC;IACjC,CAAC;IAED,iFAAiF;IACjF,KAAK;QACH,OAAO,CAAC,GAAG,IAAI,CAAC,QAAQ,CAAC,IAAI,EAAE,CAAC,CAAC;IACnC,CAAC;IAED,GAAG,CAAC,IAAY;QACd,OAAO,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;IACjC,CAAC;IAED,qFAAqF;IACrF,IAAI,CAAC,IAAY;QACf,MAAM,KAAK,GAAG,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;QACtC,IAAI,CAAC,KAAK;YAAE,OAAO,SAAS,CAAC;QAC7B,6EAA6E;QAC7E,yEAAyE;QACzE,OAAO,KAAK,CAAC,IAAI,GAAG,IAAI,CAAC,iBAAiB;YACxC,CAAC,CAAC,IAAI,CAAC,SAAS,CAAC,KAAK,CAAC,KAAK,EAAE,KAAK,CAAC,IAAI,CAAC;YACzC,CAAC,CAAC,IAAI,CAAC,YAAY,CAAC,KAAK,CAAC,KAAK,EAAE,KAAK,CAAC,IAAI,CAAC,CAAC;IACjD,CAAC;IAED,QAAQ;QACN,MAAM,KAAK,GAAG,IAAI,CAAC,OAAO,CAAC,YAAY,CAAC,EAAE,CAAC,CAAC;QAC5C,MAAM,OAAO,GAAa,EAAE,CAAC;QAC7B,KAAK,IAAI,KAAK,GAAG,CAAC,EAAE,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,KAAK,EAAE,oBAAoB,CAAC,EAAE,KAAK,IAAI,CAAC,EAAE,CAAC;YAC9E,MAAM,MAAM,GAAG,IAAI,CAAC,OAAO,CAAC,YAAY,CAAC,EAAE,GAAG,KAAK,GAAG,CAAC,CAAC,CAAC;YACzD,IAAI,MAAM,IAAI,kBAAkB;gBAAE,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;QACzD,CAAC;QACD,2EAA2E;QAC3E,yDAAyD;QACzD,IAAI,KAAK,GAAG,IAAI,CAAC,OAAO,CAAC,YAAY,CAAC,EAAE,CAAC,CAAC;QAC1C,MAAM,UAAU,GAAG,IAAI,CAAC,OAAO,CAAC,YAAY,CAAC,EAAE,CAAC,CAAC;QACjD,MAAM,SAAS,GAAG,IAAI,CAAC,WAAW,GAAG,CAAC,GAAG,CAAC,CAAC;QAC3C,KAAK,IAAI,OAAO,GAAG,CAAC,EAAE,KAAK,IAAI,kBAAkB,IAAI,OAAO,GAAG,UAAU,GAAG,CAAC,EAAE,OAAO,IAAI,CAAC,EAAE,CAAC;YAC5F,MAAM,MAAM,GAAG,IAAI,CAAC,aAAa,CAAC,KAAK,CAAC,CAAC;YACzC,KAAK,IAAI,KAAK,GAAG,CAAC,EAAE,KAAK,GAAG,SAAS,EAAE,KAAK,IAAI,CAAC,EAAE,CAAC;gBAClD,MAAM,MAAM,GAAG,IAAI,CAAC,OAAO,CAAC,YAAY,CAAC,MAAM,GAAG,KAAK,GAAG,CAAC,CAAC,CAAC;gBAC7D,IAAI,MAAM,IAAI,kBAAkB;oBAAE,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;YACzD,CAAC;YACD,KAAK,GAAG,IAAI,CAAC,OAAO,CAAC,YAAY,CAAC,MAAM,GAAG,SAAS,GAAG,CAAC,CAAC,CAAC;QAC5D,CAAC;QAED,MAAM,GAAG,GAAa,EAAE,CAAC;QACzB,KAAK,MAAM,MAAM,IAAI,OAAO,EAAE,CAAC;YAC7B,MAAM,MAAM,GAAG,IAAI,CAAC,aAAa,CAAC,MAAM,CAAC,CAAC;YAC1C,KAAK,IAAI,KAAK,GAAG,CAAC,EAAE,KAAK,GAAG,IAAI,CAAC,WAAW,GAAG,CAAC,EAAE,KAAK,IAAI,CAAC,EAAE,CAAC;gBAC7D,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,OAAO,CAAC,YAAY,CAAC,MAAM,GAAG,KAAK,GAAG,CAAC,CAAC,CAAC,CAAC;YAC1D,CAAC;QACH,CAAC;QACD,OAAO,GAAG,CAAC;IACb,CAAC;IAED,+EAA+E;IAC/E,gBAAgB,CAAC,KAAa,EAAE,eAAuB;QACrD,MAAM,MAAM,GAAa,EAAE,CAAC;QAC5B,IAAI,MAAM,GAAG,KAAK,CAAC;QACnB,KAAK,IAAI,OAAO,GAAG,CAAC,EAAE,MAAM,IAAI,kBAAkB,IAAI,OAAO,IAAI,eAAe,EAAE,OAAO,IAAI,CAAC,EAAE,CAAC;YAC/F,MAAM,MAAM,GAAG,IAAI,CAAC,aAAa,CAAC,MAAM,CAAC,CAAC;YAC1C,KAAK,IAAI,KAAK,GAAG,CAAC,EAAE,KAAK,GAAG,IAAI,CAAC,WAAW,GAAG,CAAC,EAAE,KAAK,IAAI,CAAC,EAAE,CAAC;gBAC7D,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,OAAO,CAAC,YAAY,CAAC,MAAM,GAAG,KAAK,GAAG,CAAC,CAAC,CAAC,CAAC;YAC7D,CAAC;YACD,MAAM,GAAG,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,YAAY,CAAC;QAC7C,CAAC;QACD,OAAO,MAAM,CAAC;IAChB,CAAC;IAED,cAAc;QACZ,MAAM,SAAS,GAAG,IAAI,CAAC,YAAY,CAAC,IAAI,CAAC,OAAO,CAAC,YAAY,CAAC,EAAE,CAAC,EAAE,MAAM,CAAC,iBAAiB,CAAC,CAAC;QAC7F,MAAM,OAAO,GAAG,IAAI,GAAG,EAA2C,CAAC;QACnE,IAAI,eAAe,GAAG,YAAY,CAAC;QACnC,IAAI,cAAc,GAAG,CAAC,CAAC;QACvB,KAAK,IAAI,MAAM,GAAG,CAAC,EAAE,MAAM,GAAG,qBAAqB,IAAI,SAAS,CAAC,MAAM,EAAE,MAAM,IAAI,qBAAqB,EAAE,CAAC;YACzG,MAAM,SAAS,GAAG,SAAS,CAAC,YAAY,CAAC,MAAM,GAAG,EAAE,CAAC,CAAC;YACtD,IAAI,SAAS,GAAG,CAAC,IAAI,SAAS,GAAG,EAAE;gBAAE,SAAS;YAC9C,wEAAwE;YACxE,MAAM,IAAI,GAAG,SAAS,CAAC,QAAQ,CAAC,MAAM,GAAG,CAAC,EAAE,MAAM,GAAG,SAAS,GAAG,CAAC,CAAC,CAAC,QAAQ,CAAC,SAAS,CAAC,CAAC;YACxF,MAAM,IAAI,GAAG,SAAS,CAAC,SAAS,CAAC,MAAM,GAAG,EAAE,CAAC,CAAC;YAC9C,MAAM,KAAK,GAAG,SAAS,CAAC,YAAY,CAAC,MAAM,GAAG,GAAG,CAAC,CAAC;YACnD,MAAM,IAAI,GAAG,MAAM,CAAC,SAAS,CAAC,eAAe,CAAC,MAAM,GAAG,GAAG,CAAC,CAAC,CAAC;YAC7D,uEAAuE;YACvE,IAAI,MAAM,KAAK,CAAC,EAAE,CAAC;gBACjB,eAAe,GAAG,KAAK,CAAC;gBACxB,cAAc,GAAG,IAAI,CAAC;gBACtB,SAAS;YACX,CAAC;YACD,IAAI,IAAI,KAAK,WAAW,IAAI,IAAI,IAAI,CAAC;gBAAE,SAAS;YAChD,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,IAAI,CAAC;gBAAE,OAAO,CAAC,GAAG,CAAC,IAAI,EAAE,EAAE,KAAK,EAAE,IAAI,EAAE,CAAC,CAAC;QAC7D,CAAC;QACD,MAAM,UAAU,GAAG,eAAe,IAAI,kBAAkB;YACtD,CAAC,CAAC,IAAI,CAAC,YAAY,CAAC,eAAe,EAAE,cAAc,CAAC;YACpD,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;QACpB,OAAO,EAAE,OAAO,EAAE,UAAU,EAAE,CAAC;IACjC,CAAC;IAED,YAAY,CAAC,KAAa,EAAE,IAAY;QACtC,MAAM,MAAM,GAAa,EAAE,CAAC;QAC5B,IAAI,MAAM,GAAG,KAAK,CAAC;QACnB,IAAI,SAAS,GAAG,IAAI,CAAC;QACrB,KAAK,IAAI,OAAO,GAAG,CAAC,EAAE,MAAM,IAAI,kBAAkB,IAAI,SAAS,GAAG,CAAC,IAAI,OAAO,GAAG,iBAAiB,EAAE,OAAO,IAAI,CAAC,EAAE,CAAC;YACjH,MAAM,MAAM,GAAG,IAAI,CAAC,aAAa,CAAC,MAAM,CAAC,CAAC;YAC1C,MAAM,IAAI,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,WAAW,EAAE,SAAS,CAAC,CAAC;YACnD,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,OAAO,CAAC,QAAQ,CAAC,MAAM,EAAE,MAAM,GAAG,IAAI,CAAC,CAAC,CAAC;YAC1D,SAAS,IAAI,IAAI,CAAC;YAClB,MAAM,IAAI,GAAG,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;YAC/B,IAAI,IAAI,KAAK,SAAS,IAAI,IAAI,KAAK,WAAW;gBAAE,MAAM;YACtD,MAAM,GAAG,IAAI,CAAC;QAChB,CAAC;QACD,OAAO,MAAM,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC;IAC/B,CAAC;IAED,SAAS,CAAC,KAAa,EAAE,IAAY;QACnC,MAAM,MAAM,GAAa,EAAE,CAAC;QAC5B,IAAI,MAAM,GAAG,KAAK,CAAC;QACnB,IAAI,SAAS,GAAG,IAAI,CAAC;QACrB,KAAK,IAAI,OAAO,GAAG,CAAC,EAAE,MAAM,IAAI,kBAAkB,IAAI,SAAS,GAAG,CAAC,IAAI,OAAO,GAAG,iBAAiB,EAAE,OAAO,IAAI,CAAC,EAAE,CAAC;YACjH,MAAM,MAAM,GAAG,MAAM,GAAG,IAAI,CAAC,eAAe,CAAC;YAC7C,IAAI,MAAM,IAAI,IAAI,CAAC,WAAW,CAAC,MAAM;gBAAE,MAAM;YAC7C,MAAM,IAAI,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,eAAe,EAAE,SAAS,CAAC,CAAC;YACvD,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,WAAW,CAAC,QAAQ,CAAC,MAAM,EAAE,MAAM,GAAG,IAAI,CAAC,CAAC,CAAC;YAC9D,SAAS,IAAI,IAAI,CAAC;YAClB,MAAM,IAAI,GAAG,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC;YACnC,IAAI,IAAI,KAAK,SAAS,IAAI,IAAI,KAAK,WAAW;gBAAE,MAAM;YACtD,MAAM,GAAG,IAAI,CAAC;QAChB,CAAC;QACD,OAAO,MAAM,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC;IAC/B,CAAC;IAED,mFAAmF;IACnF,aAAa,CAAC,MAAc;QAC1B,MAAM,MAAM,GAAG,GAAG,GAAG,MAAM,GAAG,IAAI,CAAC,WAAW,CAAC;QAC/C,IAAI,MAAM,GAAG,CAAC,IAAI,MAAM,GAAG,IAAI,CAAC,WAAW,GAAG,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE,CAAC;YAClE,MAAM,IAAI,KAAK,CAAC,mDAAmD,CAAC,CAAC;QACvE,CAAC;QACD,OAAO,MAAM,CAAC;IAChB,CAAC;CACF;AAED,MAAM,UAAU,YAAY,CAAC,MAAc;IACzC,OAAO,MAAM,CAAC,MAAM,IAAI,GAAG,IAAI,MAAM,CAAC,QAAQ,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,MAAM,CAAC,SAAS,CAAC,CAAC;AACzE,CAAC"}

package/dist/documents/ooxml.d.ts

@@ -0,0 +1,23 @@
+import { ZipArchive } from "./zip.js";
+/**
+ * Text out of Office Open XML packages: `.docx`, `.xlsx` and `.pptx`, plus
+ * anything WPS wrote in the same containers under a different extension.
+ *
+ * These are read with a scanner rather than a DOM: only character data and a
+ * handful of break elements matter, the parts are machine-generated, and a
+ * general XML parser would be a dependency handling untrusted documents.
+ */
+export type OoxmlKind = "docx" | "xlsx" | "pptx";
+export interface OoxmlDocument {
+    kind: OoxmlKind;
+    text: string;
+    /** One entry per worksheet or slide; empty for a word-processing document. */
+    sections: readonly {
+        name: string;
+        text: string;
+    }[];
+}
+/** Which OOXML format a package holds, decided by the parts it carries. */
+export declare function detectOoxmlKind(archive: ZipArchive): OoxmlKind | undefined;
+export declare function extractOoxml(archive: ZipArchive, kind: OoxmlKind): OoxmlDocument;
+//# sourceMappingURL=ooxml.d.ts.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"ooxml.d.ts","sourceRoot":"","sources":["../../src/documents/ooxml.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,UAAU,EAAE,MAAM,UAAU,CAAC;AAEtC;;;;;;;GAOG;AAEH,MAAM,MAAM,SAAS,GAAG,MAAM,GAAG,MAAM,GAAG,MAAM,CAAC;AAEjD,MAAM,WAAW,aAAa;IAC5B,IAAI,EAAE,SAAS,CAAC;IAChB,IAAI,EAAE,MAAM,CAAC;IACb,8EAA8E;IAC9E,QAAQ,EAAE,SAAS;QAAE,IAAI,EAAE,MAAM,CAAC;QAAC,IAAI,EAAE,MAAM,CAAA;KAAE,EAAE,CAAC;CACrD;AAED,2EAA2E;AAC3E,wBAAgB,eAAe,CAAC,OAAO,EAAE,UAAU,GAAG,SAAS,GAAG,SAAS,CAK1E;AAED,wBAAgB,YAAY,CAAC,OAAO,EAAE,UAAU,EAAE,IAAI,EAAE,SAAS,GAAG,aAAa,CAIhF"}

package/dist/documents/ooxml.js

@@ -0,0 +1,197 @@
+import { ZipArchive } from "./zip.js";
+/** Which OOXML format a package holds, decided by the parts it carries. */
+export function detectOoxmlKind(archive) {
+    if (archive.has("word/document.xml"))
+        return "docx";
+    if (archive.has("xl/workbook.xml"))
+        return "xlsx";
+    if (archive.has("ppt/presentation.xml"))
+        return "pptx";
+    return undefined;
+}
+export function extractOoxml(archive, kind) {
+    if (kind === "docx")
+        return { kind, text: extractWord(archive), sections: [] };
+    if (kind === "pptx")
+        return extractSlides(archive);
+    return extractSheets(archive);
+}
+/**
+ * Word body text. Paragraphs end a line, `w:tab` and `w:br` keep the breaks the
+ * author typed, and everything else is markup the reader does not need.
+ */
+function extractWord(archive) {
+    const xml = archive.readText("word/document.xml") ?? "";
+    return collapseBlankLines(harvestText(xml, {
+        text: /^w:t$/,
+        lineBreak: /^w:(?:br|cr)$/,
+        tab: /^w:tab$/,
+        paragraph: /^w:p$/,
+    }));
+}
+/**
+ * Slide text in slide order, with each slide named so a reader can cite one.
+ * Speaker notes are deliberately left out: they are not what is on the slide.
+ */
+function extractSlides(archive) {
+    const sections = archive.matching(/^ppt\/slides\/slide\d+\.xml$/).map((part, index) => ({
+        name: `幻灯片 ${index + 1}`,
+        text: collapseBlankLines(harvestText(archive.readText(part) ?? "", {
+            text: /^a:t$/,
+            lineBreak: /^a:br$/,
+            tab: /^a:tab$/,
+            paragraph: /^a:p$/,
+        })),
+    }));
+    return { kind: "pptx", sections, text: joinSections(sections) };
+}
+/**
+ * Worksheet cells as tab-separated rows, which keeps a table readable without
+ * inventing a layout. Values live in the shared string table often enough that
+ * a sheet read on its own would come back as a page of integers.
+ */
+function extractSheets(archive) {
+    const sharedStrings = readSharedStrings(archive);
+    const names = readSheetNames(archive);
+    const parts = archive.matching(/^xl\/worksheets\/sheet\d+\.xml$/);
+    const sections = parts.map((part, index) => ({
+        name: names[index] ?? `工作表 ${index + 1}`,
+        text: readSheet(archive.readText(part) ?? "", sharedStrings),
+    }));
+    return { kind: "xlsx", sections, text: joinSections(sections) };
+}
+function readSharedStrings(archive) {
+    const xml = archive.readText("xl/sharedStrings.xml");
+    if (xml === undefined)
+        return [];
+    // Each <si> is one string, possibly split across runs that only differ in
+    // formatting, so the runs inside one entry are concatenated without a break.
+    return [...xml.matchAll(/<si\b[^>]*>([\s\S]*?)<\/si>/g)]
+        .map((match) => harvestText(match[1] ?? "", { text: /^t$/ }).replace(/\n/g, ""));
+}
+function readSheetNames(archive) {
+    const xml = archive.readText("xl/workbook.xml") ?? "";
+    return [...xml.matchAll(/<sheet\b[^>]*\bname="([^"]*)"/g)].map((match) => decodeEntities(match[1] ?? ""));
+}
+function readSheet(xml, sharedStrings) {
+    const lines = [];
+    for (const rowMatch of xml.matchAll(/<row\b[^>]*>([\s\S]*?)<\/row>/g)) {
+        const cells = [];
+        for (const cellMatch of (rowMatch[1] ?? "").matchAll(/<c\b([^>]*)>([\s\S]*?)<\/c>/g)) {
+            cells.push(readCell(cellMatch[1] ?? "", cellMatch[2] ?? "", sharedStrings));
+        }
+        // A row of empty cells carries no information once the layout is gone.
+        if (cells.some((cell) => cell !== ""))
+            lines.push(cells.join("\t"));
+    }
+    return lines.join("\n");
+}
+function readCell(attributes, body, sharedStrings) {
+    const type = /\bt="([^"]*)"/.exec(attributes)?.[1];
+    if (type === "s") {
+        const index = Number(/<v>([\s\S]*?)<\/v>/.exec(body)?.[1] ?? Number.NaN);
+        return Number.isInteger(index) ? sharedStrings[index] ?? "" : "";
+    }
+    if (type === "inlineStr")
+        return harvestText(body, { text: /^t$/ }).replace(/\n/g, "");
+    // A formula cell keeps its cached result: the formula itself is not the value
+    // the document is showing anyone.
+    return decodeEntities(/<v>([\s\S]*?)<\/v>/.exec(body)?.[1] ?? "").trim();
+}
+/**
+ * Walks the markup once, emitting character data from the elements that carry
+ * text and a break wherever the format says a line ends.
+ *
+ * Written as a scanner because these parts are large, machine-generated and
+ * untrusted: it never builds a tree, so a deeply nested document costs no more
+ * than a flat one.
+ */
+function harvestText(xml, rules) {
+    let out = "";
+    let cursor = 0;
+    let capturing = 0;
+    while (cursor < xml.length) {
+        const open = xml.indexOf("<", cursor);
+        if (open < 0)
+            break;
+        if (capturing > 0 && open > cursor)
+            out += decodeEntities(xml.slice(cursor, open));
+        const close = xml.indexOf(">", open);
+        if (close < 0)
+            break;
+        const tag = xml.slice(open + 1, close);
+        cursor = close + 1;
+        // Comments, CDATA, processing instructions and doctypes carry no text.
+        if (tag.startsWith("!") || tag.startsWith("?"))
+            continue;
+        const selfClosing = tag.endsWith("/");
+        const closing = tag.startsWith("/");
+        const name = elementName(tag);
+        // A break is written either way round: `<w:br/>` and `<w:br></w:br>` are the
+        // same instruction, and only one of them may appear in a given document.
+        if (selfClosing || closing) {
+            if (rules.lineBreak?.test(name))
+                out += "\n";
+            else if (rules.tab?.test(name))
+                out += "\t";
+        }
+        if (selfClosing) {
+            // An empty paragraph is written `<w:p/>`, and it is still a blank line.
+            if (rules.paragraph?.test(name))
+                out += "\n";
+            continue;
+        }
+        if (closing) {
+            if (rules.text.test(name) && capturing > 0)
+                capturing -= 1;
+            if (rules.paragraph?.test(name))
+                out += "\n";
+            continue;
+        }
+        if (rules.text.test(name))
+            capturing += 1;
+    }
+    return out;
+}
+function elementName(tag) {
+    const body = tag.replace(/^\//, "").replace(/\/$/, "").trim();
+    const end = body.search(/[\s/]/);
+    return end < 0 ? body : body.slice(0, end);
+}
+const NAMED_ENTITIES = {
+    amp: "&",
+    lt: "<",
+    gt: ">",
+    quot: '"',
+    apos: "'",
+};
+function decodeEntities(value) {
+    if (!value.includes("&"))
+        return value;
+    return value.replace(/&(#x?[0-9a-fA-F]+|[a-zA-Z]+);/g, (match, entity) => {
+        if (entity.startsWith("#")) {
+            const code = entity.startsWith("#x") || entity.startsWith("#X")
+                ? Number.parseInt(entity.slice(2), 16)
+                : Number.parseInt(entity.slice(1), 10);
+            // A reference outside Unicode, or to a lone surrogate, stays as written
+            // rather than becoming a character no reader asked for.
+            if (!Number.isInteger(code) || code < 0 || code > 0x10ffff)
+                return match;
+            if (code >= 0xd800 && code <= 0xdfff)
+                return match;
+            return String.fromCodePoint(code);
+        }
+        return NAMED_ENTITIES[entity] ?? match;
+    });
+}
+function joinSections(sections) {
+    return sections
+        .filter((section) => section.text.trim() !== "")
+        .map((section) => `## ${section.name}\n${section.text}`)
+        .join("\n\n");
+}
+/** Office writes a paragraph per line break, so runs of empty ones pile up. */
+function collapseBlankLines(value) {
+    return value.replace(/\n{3,}/g, "\n\n").trim();
+}
+//# sourceMappingURL=ooxml.js.map

package/dist/documents/ooxml.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"ooxml.js","sourceRoot":"","sources":["../../src/documents/ooxml.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,UAAU,EAAE,MAAM,UAAU,CAAC;AAoBtC,2EAA2E;AAC3E,MAAM,UAAU,eAAe,CAAC,OAAmB;IACjD,IAAI,OAAO,CAAC,GAAG,CAAC,mBAAmB,CAAC;QAAE,OAAO,MAAM,CAAC;IACpD,IAAI,OAAO,CAAC,GAAG,CAAC,iBAAiB,CAAC;QAAE,OAAO,MAAM,CAAC;IAClD,IAAI,OAAO,CAAC,GAAG,CAAC,sBAAsB,CAAC;QAAE,OAAO,MAAM,CAAC;IACvD,OAAO,SAAS,CAAC;AACnB,CAAC;AAED,MAAM,UAAU,YAAY,CAAC,OAAmB,EAAE,IAAe;IAC/D,IAAI,IAAI,KAAK,MAAM;QAAE,OAAO,EAAE,IAAI,EAAE,IAAI,EAAE,WAAW,CAAC,OAAO,CAAC,EAAE,QAAQ,EAAE,EAAE,EAAE,CAAC;IAC/E,IAAI,IAAI,KAAK,MAAM;QAAE,OAAO,aAAa,CAAC,OAAO,CAAC,CAAC;IACnD,OAAO,aAAa,CAAC,OAAO,CAAC,CAAC;AAChC,CAAC;AAED;;;GAGG;AACH,SAAS,WAAW,CAAC,OAAmB;IACtC,MAAM,GAAG,GAAG,OAAO,CAAC,QAAQ,CAAC,mBAAmB,CAAC,IAAI,EAAE,CAAC;IACxD,OAAO,kBAAkB,CAAC,WAAW,CAAC,GAAG,EAAE;QACzC,IAAI,EAAE,OAAO;QACb,SAAS,EAAE,eAAe;QAC1B,GAAG,EAAE,SAAS;QACd,SAAS,EAAE,OAAO;KACnB,CAAC,CAAC,CAAC;AACN,CAAC;AAED;;;GAGG;AACH,SAAS,aAAa,CAAC,OAAmB;IACxC,MAAM,QAAQ,GAAG,OAAO,CAAC,QAAQ,CAAC,8BAA8B,CAAC,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,KAAK,EAAE,EAAE,CAAC,CAAC;QACtF,IAAI,EAAE,OAAO,KAAK,GAAG,CAAC,EAAE;QACxB,IAAI,EAAE,kBAAkB,CAAC,WAAW,CAAC,OAAO,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,EAAE,EAAE;YACjE,IAAI,EAAE,OAAO;YACb,SAAS,EAAE,QAAQ;YACnB,GAAG,EAAE,SAAS;YACd,SAAS,EAAE,OAAO;SACnB,CAAC,CAAC;KACJ,CAAC,CAAC,CAAC;IACJ,OAAO,EAAE,IAAI,EAAE,MAAM,EAAE,QAAQ,EAAE,IAAI,EAAE,YAAY,CAAC,QAAQ,CAAC,EAAE,CAAC;AAClE,CAAC;AAED;;;;GAIG;AACH,SAAS,aAAa,CAAC,OAAmB;IACxC,MAAM,aAAa,GAAG,iBAAiB,CAAC,OAAO,CAAC,CAAC;IACjD,MAAM,KAAK,GAAG,cAAc,CAAC,OAAO,CAAC,CAAC;IACtC,MAAM,KAAK,GAAG,OAAO,CAAC,QAAQ,CAAC,iCAAiC,CAAC,CAAC;IAClE,MAAM,QAAQ,GAAG,KAAK,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,KAAK,EAAE,EAAE,CAAC,CAAC;QAC3C,IAAI,EAAE,KAAK,CAAC,KAAK,CAAC,IAAI,OAAO,KAAK,GAAG,CAAC,EAAE;QACxC,IAAI,EAAE,SAAS,CAAC,OAAO,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,EAAE,EAAE,aAAa,CAAC;KAC7D,CAAC,CAAC,CAAC;IACJ,OAAO,EAAE,IAAI,EAAE,MAAM,EAAE,QAAQ,EAAE,IAAI,EAAE,YAAY,CAAC,QAAQ,CAAC,EAAE,CAAC;AAClE,CAAC;AAED,SAAS,iBAAiB,CAAC,OAAmB;IAC5C,MAAM,GAAG,GAAG,OAAO,CAAC,QAAQ,CAAC,sBAAsB,CAAC,CAAC;IACrD,IAAI,GAAG,KAAK,SAAS;QAAE,OAAO,EAAE,CAAC;IACjC,0EAA0E;IAC1E,6EAA6E;IAC7E,OAAO,CAAC,GAAG,GAAG,CAAC,QAAQ,CAAC,8BAA8B,CAAC,CAAC;SACrD,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,WAAW,CAAC,KAAK,CAAC,CAAC,CAAC,IAAI,EAAE,EAAE,EAAE,IAAI,EAAE,KAAK,EAAE,CAAC,CAAC,OAAO,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,CAAC;AACrF,CAAC;AAED,SAAS,cAAc,CAAC,OAAmB;IACzC,MAAM,GAAG,GAAG,OAAO,CAAC,QAAQ,CAAC,iBAAiB,CAAC,IAAI,EAAE,CAAC;IACtD,OAAO,CAAC,GAAG,GAAG,CAAC,QAAQ,CAAC,gCAAgC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,cAAc,CAAC,KAAK,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC;AAC5G,CAAC;AAED,SAAS,SAAS,CAAC,GAAW,EAAE,aAAgC;IAC9D,MAAM,KAAK,GAAa,EAAE,CAAC;IAC3B,KAAK,MAAM,QAAQ,IAAI,GAAG,CAAC,QAAQ,CAAC,gCAAgC,CAAC,EAAE,CAAC;QACtE,MAAM,KAAK,GAAa,EAAE,CAAC;QAC3B,KAAK,MAAM,SAAS,IAAI,CAAC,QAAQ,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,QAAQ,CAAC,8BAA8B,CAAC,EAAE,CAAC;YACrF,KAAK,CAAC,IAAI,CAAC,QAAQ,CAAC,SAAS,CAAC,CAAC,CAAC,IAAI,EAAE,EAAE,SAAS,CAAC,CAAC,CAAC,IAAI,EAAE,EAAE,aAAa,CAAC,CAAC,CAAC;QAC9E,CAAC;QACD,uEAAuE;QACvE,IAAI,KAAK,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,KAAK,EAAE,CAAC;YAAE,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC;IACtE,CAAC;IACD,OAAO,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;AAC1B,CAAC;AAED,SAAS,QAAQ,CAAC,UAAkB,EAAE,IAAY,EAAE,aAAgC;IAClF,MAAM,IAAI,GAAG,eAAe,CAAC,IAAI,CAAC,UAAU,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC;IACnD,IAAI,IAAI,KAAK,GAAG,EAAE,CAAC;QACjB,MAAM,KAAK,GAAG,MAAM,CAAC,oBAAoB,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,MAAM,CAAC,GAAG,CAAC,CAAC;QACzE,OAAO,MAAM,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,aAAa,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;IACnE,CAAC;IACD,IAAI,IAAI,KAAK,WAAW;QAAE,OAAO,WAAW,CAAC,IAAI,EAAE,EAAE,IAAI,EAAE,KAAK,EAAE,CAAC,CAAC,OAAO,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC;IACvF,8EAA8E;IAC9E,kCAAkC;IAClC,OAAO,cAAc,CAAC,oBAAoB,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC;AAC3E,CAAC;AAWD;;;;;;;GAOG;AACH,SAAS,WAAW,CAAC,GAAW,EAAE,KAAmB;IACnD,IAAI,GAAG,GAAG,EAAE,CAAC;IACb,IAAI,MAAM,GAAG,CAAC,CAAC;IACf,IAAI,SAAS,GAAG,CAAC,CAAC;IAClB,OAAO,MAAM,GAAG,GAAG,CAAC,MAAM,EAAE,CAAC;QAC3B,MAAM,IAAI,GAAG,GAAG,CAAC,OAAO,CAAC,GAAG,EAAE,MAAM,CAAC,CAAC;QACtC,IAAI,IAAI,GAAG,CAAC;YAAE,MAAM;QACpB,IAAI,SAAS,GAAG,CAAC,IAAI,IAAI,GAAG,MAAM;YAAE,GAAG,IAAI,cAAc,CAAC,GAAG,CAAC,KAAK,CAAC,MAAM,EAAE,IAAI,CAAC,CAAC,CAAC;QACnF,MAAM,KAAK,GAAG,GAAG,CAAC,OAAO,CAAC,GAAG,EAAE,IAAI,CAAC,CAAC;QACrC,IAAI,KAAK,GAAG,CAAC;YAAE,MAAM;QACrB,MAAM,GAAG,GAAG,GAAG,CAAC,KAAK,CAAC,IAAI,GAAG,CAAC,EAAE,KAAK,CAAC,CAAC;QACvC,MAAM,GAAG,KAAK,GAAG,CAAC,CAAC;QACnB,uEAAuE;QACvE,IAAI,GAAG,CAAC,UAAU,CAAC,GAAG,CAAC,IAAI,GAAG,CAAC,UAAU,CAAC,GAAG,CAAC;YAAE,SAAS;QAEzD,MAAM,WAAW,GAAG,GAAG,CAAC,QAAQ,CAAC,GAAG,CAAC,CAAC;QACtC,MAAM,OAAO,GAAG,GAAG,CAAC,UAAU,CAAC,GAAG,CAAC,CAAC;QACpC,MAAM,IAAI,GAAG,WAAW,CAAC,GAAG,CAAC,CAAC;QAC9B,6EAA6E;QAC7E,yEAAyE;QACzE,IAAI,WAAW,IAAI,OAAO,EAAE,CAAC;YAC3B,IAAI,KAAK,CAAC,SAAS,EAAE,IAAI,CAAC,IAAI,CAAC;gBAAE,GAAG,IAAI,IAAI,CAAC;iBACxC,IAAI,KAAK,CAAC,GAAG,EAAE,IAAI,CAAC,IAAI,CAAC;gBAAE,GAAG,IAAI,IAAI,CAAC;QAC9C,CAAC;QACD,IAAI,WAAW,EAAE,CAAC;YAChB,wEAAwE;YACxE,IAAI,KAAK,CAAC,SAAS,EAAE,IAAI,CAAC,IAAI,CAAC;gBAAE,GAAG,IAAI,IAAI,CAAC;YAC7C,SAAS;QACX,CAAC;QACD,IAAI,OAAO,EAAE,CAAC;YACZ,IAAI,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,SAAS,GAAG,CAAC;gBAAE,SAAS,IAAI,CAAC,CAAC;YAC3D,IAAI,KAAK,CAAC,SAAS,EAAE,IAAI,CAAC,IAAI,CAAC;gBAAE,GAAG,IAAI,IAAI,CAAC;YAC7C,SAAS;QACX,CAAC;QACD,IAAI,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC;YAAE,SAAS,IAAI,CAAC,CAAC;IAC5C,CAAC;IACD,OAAO,GAAG,CAAC;AACb,CAAC;AAED,SAAS,WAAW,CAAC,GAAW;IAC9B,MAAM,IAAI,GAAG,GAAG,CAAC,OAAO,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,OAAO,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC;IAC9D,MAAM,GAAG,GAAG,IAAI,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC;IACjC,OAAO,GAAG,GAAG,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC;AAC7C,CAAC;AAED,MAAM,cAAc,GAAqC;IACvD,GAAG,EAAE,GAAG;IACR,EAAE,EAAE,GAAG;IACP,EAAE,EAAE,GAAG;IACP,IAAI,EAAE,GAAG;IACT,IAAI,EAAE,GAAG;CACV,CAAC;AAEF,SAAS,cAAc,CAAC,KAAa;IACnC,IAAI,CAAC,KAAK,CAAC,QAAQ,CAAC,GAAG,CAAC;QAAE,OAAO,KAAK,CAAC;IACvC,OAAO,KAAK,CAAC,OAAO,CAAC,gCAAgC,EAAE,CAAC,KAAK,EAAE,MAAc,EAAE,EAAE;QAC/E,IAAI,MAAM,CAAC,UAAU,CAAC,GAAG,CAAC,EAAE,CAAC;YAC3B,MAAM,IAAI,GAAG,MAAM,CAAC,UAAU,CAAC,IAAI,CAAC,IAAI,MAAM,CAAC,UAAU,CAAC,IAAI,CAAC;gBAC7D,CAAC,CAAC,MAAM,CAAC,QAAQ,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC;gBACtC,CAAC,CAAC,MAAM,CAAC,QAAQ,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC;YACzC,wEAAwE;YACxE,wDAAwD;YACxD,IAAI,CAAC,MAAM,CAAC,SAAS,CAAC,IAAI,CAAC,IAAI,IAAI,GAAG,CAAC,IAAI,IAAI,GAAG,QAAQ;gBAAE,OAAO,KAAK,CAAC;YACzE,IAAI,IAAI,IAAI,MAAM,IAAI,IAAI,IAAI,MAAM;gBAAE,OAAO,KAAK,CAAC;YACnD,OAAO,MAAM,CAAC,aAAa,CAAC,IAAI,CAAC,CAAC;QACpC,CAAC;QACD,OAAO,cAAc,CAAC,MAAM,CAAC,IAAI,KAAK,CAAC;IACzC,CAAC,CAAC,CAAC;AACL,CAAC;AAED,SAAS,YAAY,CAAC,QAAmD;IACvE,OAAO,QAAQ;SACZ,MAAM,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC,OAAO,CAAC,IAAI,CAAC,IAAI,EAAE,KAAK,EAAE,CAAC;SAC/C,GAAG,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC,MAAM,OAAO,CAAC,IAAI,KAAK,OAAO,CAAC,IAAI,EAAE,CAAC;SACvD,IAAI,CAAC,MAAM,CAAC,CAAC;AAClB,CAAC;AAED,+EAA+E;AAC/E,SAAS,kBAAkB,CAAC,KAAa;IACvC,OAAO,KAAK,CAAC,OAAO,CAAC,SAAS,EAAE,MAAM,CAAC,CAAC,IAAI,EAAE,CAAC;AACjD,CAAC"}

package/dist/documents/pdf.d.ts

@@ -0,0 +1,20 @@
+/**
+ * PDF text through pdf.js.
+ *
+ * A PDF does not contain text so much as instructions for placing glyphs, and a
+ * Chinese one places them through an embedded CID font whose codes mean nothing
+ * without the font's own ToUnicode table. Reimplementing that is how extraction
+ * ends up returning confident mojibake, so this is the one format that leans on
+ * a library — used for text only, with no rendering and no network fetches.
+ */
+export interface PdfDocument {
+    text: string;
+    pageCount: number;
+    sections: readonly {
+        name: string;
+        text: string;
+    }[];
+}
+export declare function extractPdf(data: Buffer): Promise<PdfDocument>;
+export declare function looksLikePdf(buffer: Buffer): boolean;
+//# sourceMappingURL=pdf.d.ts.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"pdf.d.ts","sourceRoot":"","sources":["../../src/documents/pdf.ts"],"names":[],"mappings":"AAOA;;;;;;;;GAQG;AAEH,MAAM,WAAW,WAAW;IAC1B,IAAI,EAAE,MAAM,CAAC;IACb,SAAS,EAAE,MAAM,CAAC;IAClB,QAAQ,EAAE,SAAS;QAAE,IAAI,EAAE,MAAM,CAAC;QAAC,IAAI,EAAE,MAAM,CAAA;KAAE,EAAE,CAAC;CACrD;AA+DD,wBAAsB,UAAU,CAAC,IAAI,EAAE,MAAM,GAAG,OAAO,CAAC,WAAW,CAAC,CAwCnE;AAED,wBAAgB,YAAY,CAAC,MAAM,EAAE,MAAM,GAAG,OAAO,CAIpD"}

package/dist/documents/pdf.js

@@ -0,0 +1,95 @@
+import { existsSync } from "node:fs";
+import { createRequire } from "node:module";
+import { dirname, join } from "node:path";
+import { pathToFileURL } from "node:url";
+import { UnsupportedDocumentError } from "./errors.js";
+/** Loaded on first use: pdf.js is large, and most documents are not PDFs. */
+let engine;
+async function loadEngine() {
+    try {
+        // Resolved rather than imported by specifier so the same code works from the
+        // build output and from an Electron package, where the package sits in
+        // different places relative to this module.
+        const entry = createRequire(import.meta.url).resolve("pdfjs-dist/legacy/build/pdf.mjs");
+        return await import(pathToFileURL(entry).href);
+    }
+    catch (error) {
+        // The standalone CLI is one bundled file with no package tree beside it, so
+        // it cannot reach pdf.js. Every other format still works there, and saying
+        // which build this is beats reporting a module resolution failure.
+        throw new UnsupportedDocumentError(`本次运行的 pSpore 构建中没有 PDF 解析引擎（${error instanceof Error ? error.message : String(error)}）。`
+            + "桌面版与服务进程支持 PDF；独立 CLI 二进制不支持，请改用桌面端或先把 PDF 转成文本。");
+    }
+}
+/**
+ * Data directories pdf.js loads from disk rather than from its own bundle.
+ *
+ * `cmaps` is what makes a Chinese PDF readable: one that names a predefined
+ * CMap such as `UniGB-UCS2-H` maps its glyph codes through that table, and
+ * without it the page comes back empty or as the wrong characters. Under Node
+ * these are read with `fs`, so they are plain paths — a `file://` URL fails.
+ */
+function dataDirectories() {
+    try {
+        const require = createRequire(import.meta.url);
+        const root = dirname(dirname(dirname(require.resolve("pdfjs-dist/legacy/build/pdf.mjs"))));
+        const cmaps = join(root, "cmaps");
+        const fonts = join(root, "standard_fonts");
+        return {
+            ...(existsSync(cmaps) ? { cMapUrl: `${cmaps}/` } : {}),
+            ...(existsSync(fonts) ? { standardFontDataUrl: `${fonts}/` } : {}),
+        };
+    }
+    catch {
+        // A build that ships pdf.js without its data files still reads any document
+        // whose fonts are embedded, which is most of them.
+        return {};
+    }
+}
+export async function extractPdf(data) {
+    engine ??= loadEngine();
+    const pdfjs = await engine;
+    const task = pdfjs.getDocument({
+        // pdf.js takes ownership of the array it is given, so it gets its own copy.
+        data: new Uint8Array(data),
+        // Nothing here should reach the network or the font system, and no document
+        // should be able to talk this process into evaluating its own code.
+        isEvalSupported: false,
+        disableFontFace: true,
+        useSystemFonts: false,
+        cMapPacked: true,
+        ...dataDirectories(),
+    });
+    const document = await task.promise;
+    try {
+        const sections = [];
+        for (let page = 1; page <= document.numPages; page += 1) {
+            const content = await (await document.getPage(page)).getTextContent();
+            // Items are glyph runs, not lines; pdf.js marks the ones that end a line.
+            const text = content.items
+                .map((item) => `${item.str ?? ""}${item.hasEOL ? "\n" : ""}`)
+                .join("")
+                .replace(/[ \t]+\n/g, "\n")
+                .replace(/\n{3,}/g, "\n\n")
+                .trim();
+            sections.push({ name: `第 ${page} 页`, text });
+        }
+        return {
+            pageCount: document.numPages,
+            sections,
+            text: sections
+                .filter((section) => section.text !== "")
+                .map((section) => `## ${section.name}\n${section.text}`)
+                .join("\n\n"),
+        };
+    }
+    finally {
+        await task.destroy().catch(() => undefined);
+    }
+}
+export function looksLikePdf(buffer) {
+    // The header may sit behind a few junk bytes; readers tolerate that, and a
+    // file that only reaches here because of its extension should too.
+    return buffer.subarray(0, 1024).includes("%PDF-");
+}
+//# sourceMappingURL=pdf.js.map

package/dist/documents/pdf.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"pdf.js","sourceRoot":"","sources":["../../src/documents/pdf.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,UAAU,EAAE,MAAM,SAAS,CAAC;AACrC,OAAO,EAAE,aAAa,EAAE,MAAM,aAAa,CAAC;AAC5C,OAAO,EAAE,OAAO,EAAE,IAAI,EAAE,MAAM,WAAW,CAAC;AAC1C,OAAO,EAAE,aAAa,EAAE,MAAM,UAAU,CAAC;AAEzC,OAAO,EAAE,wBAAwB,EAAE,MAAM,aAAa,CAAC;AAkBvD,6EAA6E;AAC7E,IAAI,MAAsC,CAAC;AAiB3C,KAAK,UAAU,UAAU;IACvB,IAAI,CAAC;QACH,6EAA6E;QAC7E,uEAAuE;QACvE,4CAA4C;QAC5C,MAAM,KAAK,GAAG,aAAa,CAAC,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,OAAO,CAAC,iCAAiC,CAAC,CAAC;QACxF,OAAO,MAAM,MAAM,CAAC,aAAa,CAAC,KAAK,CAAC,CAAC,IAAI,CAAc,CAAC;IAC9D,CAAC;IAAC,OAAO,KAAK,EAAE,CAAC;QACf,4EAA4E;QAC5E,2EAA2E;QAC3E,mEAAmE;QACnE,MAAM,IAAI,wBAAwB,CAChC,+BAA+B,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,IAAI;cACvF,kDAAkD,CACrD,CAAC;IACJ,CAAC;AACH,CAAC;AAED;;;;;;;GAOG;AACH,SAAS,eAAe;IACtB,IAAI,CAAC;QACH,MAAM,OAAO,GAAG,aAAa,CAAC,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;QAC/C,MAAM,IAAI,GAAG,OAAO,CAAC,OAAO,CAAC,OAAO,CAAC,OAAO,CAAC,OAAO,CAAC,iCAAiC,CAAC,CAAC,CAAC,CAAC,CAAC;QAC3F,MAAM,KAAK,GAAG,IAAI,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC;QAClC,MAAM,KAAK,GAAG,IAAI,CAAC,IAAI,EAAE,gBAAgB,CAAC,CAAC;QAC3C,OAAO;YACL,GAAG,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,OAAO,EAAE,GAAG,KAAK,GAAG,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;YACtD,GAAG,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,mBAAmB,EAAE,GAAG,KAAK,GAAG,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;SACnE,CAAC;IACJ,CAAC;IAAC,MAAM,CAAC;QACP,4EAA4E;QAC5E,mDAAmD;QACnD,OAAO,EAAE,CAAC;IACZ,CAAC;AACH,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,UAAU,CAAC,IAAY;IAC3C,MAAM,KAAK,UAAU,EAAE,CAAC;IACxB,MAAM,KAAK,GAAG,MAAM,MAAM,CAAC;IAC3B,MAAM,IAAI,GAAG,KAAK,CAAC,WAAW,CAAC;QAC7B,4EAA4E;QAC5E,IAAI,EAAE,IAAI,UAAU,CAAC,IAAI,CAAC;QAC1B,4EAA4E;QAC5E,oEAAoE;QACpE,eAAe,EAAE,KAAK;QACtB,eAAe,EAAE,IAAI;QACrB,cAAc,EAAE,KAAK;QACrB,UAAU,EAAE,IAAI;QAChB,GAAG,eAAe,EAAE;KACrB,CAAC,CAAC;IAEH,MAAM,QAAQ,GAAG,MAAM,IAAI,CAAC,OAAO,CAAC;IACpC,IAAI,CAAC;QACH,MAAM,QAAQ,GAAqC,EAAE,CAAC;QACtD,KAAK,IAAI,IAAI,GAAG,CAAC,EAAE,IAAI,IAAI,QAAQ,CAAC,QAAQ,EAAE,IAAI,IAAI,CAAC,EAAE,CAAC;YACxD,MAAM,OAAO,GAAG,MAAM,CAAC,MAAM,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,cAAc,EAAE,CAAC;YACtE,0EAA0E;YAC1E,MAAM,IAAI,GAAG,OAAO,CAAC,KAAK;iBACvB,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,GAAG,IAAI,CAAC,GAAG,IAAI,EAAE,GAAG,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC;iBAC5D,IAAI,CAAC,EAAE,CAAC;iBACR,OAAO,CAAC,WAAW,EAAE,IAAI,CAAC;iBAC1B,OAAO,CAAC,SAAS,EAAE,MAAM,CAAC;iBAC1B,IAAI,EAAE,CAAC;YACV,QAAQ,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,KAAK,IAAI,IAAI,EAAE,IAAI,EAAE,CAAC,CAAC;QAC/C,CAAC;QACD,OAAO;YACL,SAAS,EAAE,QAAQ,CAAC,QAAQ;YAC5B,QAAQ;YACR,IAAI,EAAE,QAAQ;iBACX,MAAM,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC,OAAO,CAAC,IAAI,KAAK,EAAE,CAAC;iBACxC,GAAG,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC,MAAM,OAAO,CAAC,IAAI,KAAK,OAAO,CAAC,IAAI,EAAE,CAAC;iBACvD,IAAI,CAAC,MAAM,CAAC;SAChB,CAAC;IACJ,CAAC;YAAS,CAAC;QACT,MAAM,IAAI,CAAC,OAAO,EAAE,CAAC,KAAK,CAAC,GAAG,EAAE,CAAC,SAAS,CAAC,CAAC;IAC9C,CAAC;AACH,CAAC;AAED,MAAM,UAAU,YAAY,CAAC,MAAc;IACzC,2EAA2E;IAC3E,mEAAmE;IACnE,OAAO,MAAM,CAAC,QAAQ,CAAC,CAAC,EAAE,IAAI,CAAC,CAAC,QAAQ,CAAC,OAAO,CAAC,CAAC;AACpD,CAAC"}

package/dist/documents/plain-text.d.ts

@@ -0,0 +1,22 @@
+/**
+ * Decoding for files that are already text.
+ *
+ * Encoding cannot be assumed to be UTF-8 here. Documents that reach this agent
+ * are the user's own, written on Windows or by WPS, where a Chinese plain-text
+ * file is routinely GB18030; decoding one as UTF-8 yields a page of replacement
+ * characters that reads as corruption rather than as a wrong guess.
+ */
+export interface DecodedText {
+    text: string;
+    /** Encoding the bytes were read as, so a reader can see what was assumed. */
+    encoding: string;
+}
+export declare function decodeText(buffer: Buffer): DecodedText;
+/**
+ * True when the bytes hold no NUL, which is the standard test for text.
+ *
+ * It is also the test that matters downstream: a NUL is the one character
+ * clients refuse, and `read` rejects files carrying one for the same reason.
+ */
+export declare function looksLikeText(buffer: Buffer): boolean;
+//# sourceMappingURL=plain-text.d.ts.map

package/dist/documents/plain-text.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"plain-text.d.ts","sourceRoot":"","sources":["../../src/documents/plain-text.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AAEH,MAAM,WAAW,WAAW;IAC1B,IAAI,EAAE,MAAM,CAAC;IACb,6EAA6E;IAC7E,QAAQ,EAAE,MAAM,CAAC;CAClB;AAKD,wBAAgB,UAAU,CAAC,MAAM,EAAE,MAAM,GAAG,WAAW,CAYtD;AAwBD;;;;;GAKG;AACH,wBAAgB,aAAa,CAAC,MAAM,EAAE,MAAM,GAAG,OAAO,CAErD"}

package/dist/documents/plain-text.js

@@ -0,0 +1,57 @@
+/**
+ * Decoding for files that are already text.
+ *
+ * Encoding cannot be assumed to be UTF-8 here. Documents that reach this agent
+ * are the user's own, written on Windows or by WPS, where a Chinese plain-text
+ * file is routinely GB18030; decoding one as UTF-8 yields a page of replacement
+ * characters that reads as corruption rather than as a wrong guess.
+ */
+/** Legacy encoding tried when the bytes are not valid UTF-8. */
+const CHINESE_FALLBACK = "gb18030";
+export function decodeText(buffer) {
+    const withBom = decodeByteOrderMark(buffer);
+    if (withBom)
+        return withBom;
+    // UTF-8 is self-verifying: a byte sequence that decodes without producing a
+    // replacement character is UTF-8 in practice, so it is tried before guessing.
+    const utf8 = decodeStrict(buffer, "utf-8");
+    if (utf8 !== undefined)
+        return { text: utf8, encoding: "utf-8" };
+    const legacy = decodeStrict(buffer, CHINESE_FALLBACK);
+    if (legacy !== undefined)
+        return { text: legacy, encoding: CHINESE_FALLBACK };
+    // Nothing decoded cleanly. UTF-8 with replacement characters is the honest
+    // result: it keeps the ASCII structure a reader can still use.
+    return { text: buffer.toString("utf8"), encoding: "utf-8 (有无法解码的字节)" };
+}
+function decodeByteOrderMark(buffer) {
+    if (buffer.length >= 3 && buffer[0] === 0xef && buffer[1] === 0xbb && buffer[2] === 0xbf) {
+        return { text: buffer.subarray(3).toString("utf8"), encoding: "utf-8" };
+    }
+    if (buffer.length >= 2 && buffer[0] === 0xff && buffer[1] === 0xfe) {
+        return { text: new TextDecoder("utf-16le").decode(buffer.subarray(2)), encoding: "utf-16le" };
+    }
+    if (buffer.length >= 2 && buffer[0] === 0xfe && buffer[1] === 0xff) {
+        return { text: new TextDecoder("utf-16be").decode(buffer.subarray(2)), encoding: "utf-16be" };
+    }
+    return undefined;
+}
+/** Decodes only when every byte was representable; otherwise reports failure. */
+function decodeStrict(buffer, encoding) {
+    try {
+        return new TextDecoder(encoding, { fatal: true }).decode(buffer);
+    }
+    catch {
+        return undefined;
+    }
+}
+/**
+ * True when the bytes hold no NUL, which is the standard test for text.
+ *
+ * It is also the test that matters downstream: a NUL is the one character
+ * clients refuse, and `read` rejects files carrying one for the same reason.
+ */
+export function looksLikeText(buffer) {
+    return !buffer.includes(0);
+}
+//# sourceMappingURL=plain-text.js.map