feat(parser): 배포용 문서 지원

Co-authored-by: Karam KIM <zntuszntus@gmail.com>

packages/parser/package.json

@@ -37,6 +37,7 @@
   },
   "devDependencies": {
     "@hwp.js/eslint-config": "workspace:*",
+    "@types/aes-js": "^3.1.4",
     "@types/node": "^20",
     "@types/pako": "^1.0.1",
     "eslint": "^8",
@@ -45,6 +46,7 @@
     "typescript": "^5"
   },
   "dependencies": {
+    "aes-js": "^3.1.2",
     "cfb": "^1.2.2",
     "pako": "^2.1.0"
   }

packages/parser/src/models/document.ts

@@ -42,12 +42,7 @@ export class HWPDocument {
     const sections: Section[] = []
 
     for (let i = 0; i < docInfo.properties.sections; i += 1) {
-      const entry = find(container, `Root Entry/BodyText/Section${i}`)
-
-      if (!entry) {
-        throw new Error('Section not exist')
-      }
-      sections.push(Section.fromEntry(entry, header, options))
+      sections.push(Section.fromCFB(container, i, header, options))
     }
 
     const binDataList: BinData[] = []

packages/parser/src/models/section.ts

@@ -14,8 +14,9 @@
  * limitations under the License.
  */
 
-import type { CFB$Entry } from 'cfb'
+import { find, type CFB$Container, type CFB$Entry } from 'cfb'
 import { inflate } from 'pako'
+import * as aesjs from 'aes-js'
 
 import type { HWPHeader } from './header.js'
 import { Paragraph } from './paragraph.js'
@@ -26,9 +27,53 @@ import type { ParseOptions } from '../types/parser.js'
 export class Section {
   constructor(public paragraphs: Paragraph[]) {}
 
-  static fromEntry(entry: CFB$Entry, header: HWPHeader, options: ParseOptions): Section {
-    const content = Uint8Array.from(entry.content)
+  static fromCFB(
+    container: CFB$Container,
+    index: number,
+    header: HWPHeader,
+    options: ParseOptions,
+  ) {
+    if (header.flags.distributed) {
+      return Section.fromDistributed(container, index, header, options)
+    }
+    return Section.fromNormal(container, index, header, options)
+  }
+
+  static fromNormal(
+    container: CFB$Container,
+    index: number,
+    header: HWPHeader,
+    options: ParseOptions,
+  ) {
+    const entry = find(container, `Root Entry/BodyText/Section${index}`)
+
+    if (!entry) {
+      throw new Error('Section not exist')
+    }
+
+    return Section.fromBytes(Uint8Array.from(entry.content), header, options)
+  }
+
+  static fromDistributed(
+    container: CFB$Container,
+    index: number,
+    header: HWPHeader,
+    options: ParseOptions,
+  ) {
+    const entry = find(container, `Root Entry/ViewText/Section${index}`)
+    if (!entry) {
+      throw new Error('Section not exist')
+    }
+
+    const content = decodeContent(entry)
+    return Section.fromBytes(content, header, options)
+  }
 
+  static fromBytes(
+    content: Uint8Array,
+    header: HWPHeader,
+    options: ParseOptions,
+  ): Section {
     if (header.flags.compressed) {
       const decoded = inflate(content, { windowBits: -15 })
       return Section.fromBuffer(decoded.buffer, header, options)
@@ -37,7 +82,11 @@ export class Section {
     return Section.fromBuffer(content.buffer, header, options)
   }
 
-  static fromBuffer(buffer: ArrayBuffer, header: HWPHeader, options: ParseOptions): Section {
+  static fromBuffer(
+    buffer: ArrayBuffer,
+    header: HWPHeader,
+    options: ParseOptions,
+  ): Section {
     const reader = new ByteReader(buffer)
     const records = new PeekableIterator(reader.records())
     const paragraphs: Paragraph[] = []
@@ -47,3 +96,45 @@ export class Section {
     return new Section(paragraphs)
   }
 }
+
+function createRand(seed = 1) {
+  let randomSeed = seed
+  return () => {
+    randomSeed = (randomSeed * 214013 + 2531011) & 0xffffffff
+    return (randomSeed >> 16) & 0x7fff
+  }
+}
+
+function decrypt(cipherText: ArrayBuffer, decKey: ArrayBuffer) {
+  // eslint-disable-next-line new-cap
+  const aesEcb = new aesjs.ModeOfOperation.ecb(new Uint8Array(decKey))
+  const decryptedBytes = aesEcb.decrypt(new Uint8Array(cipherText))
+  return decryptedBytes
+}
+
+function getDecryptionKey(data: ArrayBuffer): ArrayBuffer {
+  const sha1Encoded = new Uint8Array(data)
+  const sha1Decoded = new Uint8Array(sha1Encoded.length)
+  const seed = new DataView(data.slice(0, 4)).getInt32(0, true)
+  const offset = 4 + (seed & 0xf)
+  const rand = createRand(seed)
+  for (let j = 0, n = 0, k = 0; j < 256; j += 1, n -= 1) {
+    if (n === 0) {
+      k = rand() & 0xff
+      n = (rand() & 0xf) + 1
+    }
+    sha1Decoded[j] = sha1Encoded[j] ^ k
+  }
+  const sha1ucsstr = sha1Decoded.slice(offset, 80)
+  return sha1ucsstr.slice(0, 16)
+}
+
+function decodeContent(entry: CFB$Entry) {
+  const content = new Uint8Array(entry.content)
+  const reader = new ByteReader(content.buffer)
+  const { data } = reader.readRecord()
+  const encryptedData = reader.read(reader.remainByte())
+  const decKey = getDecryptionKey(data)
+  const decrypted = decrypt(encryptedData, decKey)
+  return decrypted
+}