feat: add ViewText (distribution document) parser

package.json

@@ -47,6 +47,7 @@
     "@rollup/plugin-commonjs": "^15.0.0",
     "@rollup/plugin-node-resolve": "^9.0.0",
     "@rollup/plugin-strip": "^2.0.0",
+    "@types/aes-js": "^3",
     "@types/jest": "^26.0.10",
     "@types/pako": "^1.0.1",
     "@typescript-eslint/eslint-plugin": "^3.6.1",
@@ -63,6 +64,7 @@
     "typescript": "^4.0.2"
   },
   "dependencies": {
+    "aes-js": "^3.1.2",
     "cfb": "^1.2.0",
     "pako": "^1.0.11"
   }

src/parser/parse.ts

@@ -30,6 +30,7 @@ import HWPVersion from '../models/version'
 import Section from '../models/section'
 import DocInfoParser from './DocInfoParser'
 import SectionParser from './SectionParser'
+import parseViewText from './parseViewText'
 import ByteReader from '../utils/byteReader'
 import { getBitValue } from '../utils/bitUtils'
 
@@ -119,17 +120,25 @@ function parseSection(container: CFB$Container, sectionNumber: number): Section
   return new SectionParser(decodedContent).parse()
 }
 
+function parseBodyText(container: CFB$Container, docInfo: DocInfo): Section[] {
+  const sections: Section[] = []
+
+  for (let i = 0; i < docInfo.sectionSize; i += 1) {
+    sections.push(parseSection(container, i))
+  }
+
+  return sections
+}
+
 function parse(input: CFB$Blob, options?: CFB$ParsingOptions): HWPDocument {
   const container: CFB$Container = read(input, options)
 
   const header = parseFileHeader(container)
   const docInfo = parseDocInfo(container, header)
 
-  const sections: Section[] = []
-
-  for (let i = 0; i < docInfo.sectionSize; i += 1) {
-    sections.push(parseSection(container, i))
-  }
+  const sections: Section[] = header.properties.distribution
+    ? parseViewText(container)
+    : parseBodyText(container, docInfo)
 
   return new HWPDocument(header, docInfo, sections)
 }

src/parser/parseViewText.ts

@@ -0,0 +1,78 @@
+// Reference 1 : https://cdn.hancom.com/link/docs/%ED%95%9C%EA%B8%80%EB%AC%B8%EC%84%9C%ED%8C%8C%EC%9D%BC%ED%98%95%EC%8B%9D_%EB%B0%B0%ED%8F%AC%EC%9A%A9%EB%AC%B8%EC%84%9C_revision1.2.hwp
+// Reference 2 : https://groups.google.com/g/hwp-foss/c/d2KL2ypR89Q
+
+import {
+  find,
+  CFB$Container,
+  CFB$Entry,
+} from 'cfb'
+import { inflate } from 'pako'
+import * as aesjs from 'aes-js'
+
+import Section from '../models/section'
+import SectionParser from './SectionParser'
+import ByteReader from '../utils/byteReader'
+
+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 parseViewTextSection(entry: CFB$Entry): Section {
+  const content = new Uint8Array(entry.content)
+  const reader = new ByteReader(content.buffer)
+  const [, , size] = reader.readRecord()
+  const distDocData = reader.read(size)
+  const encryptedData = reader.read(reader.remainByte())
+  const decKey = getDecryptionKey(distDocData)
+  const decrypted = decrypt(encryptedData, decKey)
+  const decodedContent: Uint8Array = inflate(decrypted, { windowBits: -15 })
+  const section: Section = new SectionParser(decodedContent).parse()
+  return section
+}
+
+function parseViewText(container: CFB$Container): Section[] {
+  const view = find(container, 'Root Entry/ViewText/')
+  const viewPaths = container.FullPaths.filter((e: string) => e.startsWith('Root Entry/ViewText/Section'))
+  const sections: Section[] = []
+
+  if (view && viewPaths.length > 0) {
+    for (let i = 0; i < viewPaths.length; i += 1) {
+      const entry = find(container, `Root Entry/ViewText/Section${i}`)
+      if (entry != null) {
+        sections.push(parseViewTextSection(entry))
+      }
+    }
+  }
+
+  return sections
+}
+
+export default parseViewText

yarn.lock

@@ -3393,6 +3393,13 @@ __metadata:
   languageName: node
   linkType: hard
 
+"@types/aes-js@npm:^3":
+  version: 3.1.1
+  resolution: "@types/aes-js@npm:3.1.1"
+  checksum: 388d21adbaa70548f9e94947430f53b54d13f99408fc6461227c2529b9dbaa5397a135e8bb4584206c65a3ef837b868b34602816dc7c20f1537ea9ef23a74bd9
+  languageName: node
+  linkType: hard
+
 "@types/babel__core@npm:^7.0.0, @types/babel__core@npm:^7.1.7":
   version: 7.1.9
   resolution: "@types/babel__core@npm:7.1.9"
@@ -4324,6 +4331,13 @@ __metadata:
   languageName: node
   linkType: hard
 
+"aes-js@npm:^3.1.2":
+  version: 3.1.2
+  resolution: "aes-js@npm:3.1.2"
+  checksum: 062154d50b1e433cc8c3b8ca7879f3a6375d5e79c2a507b2b6c4ec920b4cd851bf2afa7f65c98761a9da89c0ab618cbe6529e8e9a1c71f93290b53128fb8f712
+  languageName: node
+  linkType: hard
+
 "agent-base@npm:6, agent-base@npm:^6.0.2":
   version: 6.0.2
   resolution: "agent-base@npm:6.0.2"
@@ -10935,10 +10949,12 @@ fsevents@~2.3.2:
     "@rollup/plugin-commonjs": ^15.0.0
     "@rollup/plugin-node-resolve": ^9.0.0
     "@rollup/plugin-strip": ^2.0.0
+    "@types/aes-js": ^3
     "@types/jest": ^26.0.10
     "@types/pako": ^1.0.1
     "@typescript-eslint/eslint-plugin": ^3.6.1
     "@typescript-eslint/parser": ^3.9.1
+    aes-js: ^3.1.2
     babel-jest: ^26.3.0
     cfb: ^1.2.0
     eslint: ^7.7.0