1 | "use strict";
|
2 | const whatwgEncoding = require("whatwg-encoding");
|
3 |
|
4 |
|
5 | module.exports = (uint8Array, { transportLayerEncodingLabel, defaultEncoding = "windows-1252" } = {}) => {
|
6 | let encoding = whatwgEncoding.getBOMEncoding(uint8Array);
|
7 |
|
8 | if (encoding === null && transportLayerEncodingLabel !== undefined) {
|
9 | encoding = whatwgEncoding.labelToName(transportLayerEncodingLabel);
|
10 | }
|
11 |
|
12 | if (encoding === null) {
|
13 | encoding = prescanMetaCharset(uint8Array);
|
14 | }
|
15 |
|
16 | if (encoding === null) {
|
17 | encoding = defaultEncoding;
|
18 | }
|
19 |
|
20 | return encoding;
|
21 | };
|
22 |
|
23 |
|
24 | function prescanMetaCharset(uint8Array) {
|
25 | const l = Math.min(uint8Array.byteLength, 1024);
|
26 | for (let i = 0; i < l; i++) {
|
27 | let c = uint8Array[i];
|
28 | if (c === 0x3C) {
|
29 |
|
30 | const c1 = uint8Array[i + 1];
|
31 | const c2 = uint8Array[i + 2];
|
32 | const c3 = uint8Array[i + 3];
|
33 | const c4 = uint8Array[i + 4];
|
34 | const c5 = uint8Array[i + 5];
|
35 |
|
36 | if (c1 === 0x21 && c2 === 0x2D && c3 === 0x2D) {
|
37 | i += 4;
|
38 | for (; i < l; i++) {
|
39 | c = uint8Array[i];
|
40 | const cMinus1 = uint8Array[i - 1];
|
41 | const cMinus2 = uint8Array[i - 2];
|
42 |
|
43 | if (c === 0x3E && cMinus1 === 0x2D && cMinus2 === 0x2D) {
|
44 | break;
|
45 | }
|
46 | }
|
47 | } else if ((c1 === 0x4D || c1 === 0x6D) &&
|
48 | (c2 === 0x45 || c2 === 0x65) &&
|
49 | (c3 === 0x54 || c3 === 0x74) &&
|
50 | (c4 === 0x41 || c4 === 0x61) &&
|
51 | (isSpaceCharacter(c5) || c5 === 0x2F)) {
|
52 |
|
53 | i += 6;
|
54 | const attributeList = new Set();
|
55 | let gotPragma = false;
|
56 | let needPragma = null;
|
57 | let charset = null;
|
58 |
|
59 | let attrRes;
|
60 | do {
|
61 | attrRes = getAttribute(uint8Array, i, l);
|
62 | if (attrRes.attr && !attributeList.has(attrRes.attr.name)) {
|
63 | attributeList.add(attrRes.attr.name);
|
64 | if (attrRes.attr.name === "http-equiv") {
|
65 | gotPragma = attrRes.attr.value === "content-type";
|
66 | } else if (attrRes.attr.name === "content" && !charset) {
|
67 | charset = extractCharacterEncodingFromMeta(attrRes.attr.value);
|
68 | if (charset !== null) {
|
69 | needPragma = true;
|
70 | }
|
71 | } else if (attrRes.attr.name === "charset") {
|
72 | charset = whatwgEncoding.labelToName(attrRes.attr.value);
|
73 | needPragma = false;
|
74 | }
|
75 | }
|
76 | i = attrRes.i;
|
77 | } while (attrRes.attr);
|
78 |
|
79 | if (needPragma === null) {
|
80 | continue;
|
81 | }
|
82 | if (needPragma === true && gotPragma === false) {
|
83 | continue;
|
84 | }
|
85 | if (charset === null) {
|
86 | continue;
|
87 | }
|
88 |
|
89 | if (charset === "UTF-16LE" || charset === "UTF-16BE") {
|
90 | charset = "UTF-8";
|
91 | }
|
92 | if (charset === "x-user-defined") {
|
93 | charset = "windows-1252";
|
94 | }
|
95 |
|
96 | return charset;
|
97 | } else if ((c1 >= 0x41 && c1 <= 0x5A) || (c1 >= 0x61 && c1 <= 0x7A)) {
|
98 |
|
99 | for (i += 2; i < l; i++) {
|
100 | c = uint8Array[i];
|
101 |
|
102 | if (isSpaceCharacter(c) || c === 0x3E) {
|
103 | break;
|
104 | }
|
105 | }
|
106 | let attrRes;
|
107 | do {
|
108 | attrRes = getAttribute(uint8Array, i, l);
|
109 | i = attrRes.i;
|
110 | } while (attrRes.attr);
|
111 | } else if (c1 === 0x21 || c1 === 0x2F || c1 === 0x3F) {
|
112 |
|
113 | for (i += 2; i < l; i++) {
|
114 | c = uint8Array[i];
|
115 |
|
116 | if (c === 0x3E) {
|
117 | break;
|
118 | }
|
119 | }
|
120 | }
|
121 | }
|
122 | }
|
123 | return null;
|
124 | }
|
125 |
|
126 |
|
127 | function getAttribute(uint8Array, i, l) {
|
128 | for (; i < l; i++) {
|
129 | let c = uint8Array[i];
|
130 |
|
131 | if (isSpaceCharacter(c) || c === 0x2F) {
|
132 | continue;
|
133 | }
|
134 |
|
135 | if (c === 0x3E) {
|
136 | break;
|
137 | }
|
138 | let name = "";
|
139 | let value = "";
|
140 | nameLoop:for (; i < l; i++) {
|
141 | c = uint8Array[i];
|
142 |
|
143 | if (c === 0x3D && name !== "") {
|
144 | i++;
|
145 | break;
|
146 | }
|
147 |
|
148 | if (isSpaceCharacter(c)) {
|
149 | for (i++; i < l; i++) {
|
150 | c = uint8Array[i];
|
151 |
|
152 | if (isSpaceCharacter(c)) {
|
153 | continue;
|
154 | }
|
155 |
|
156 | if (c !== 0x3D) {
|
157 | return { attr: { name, value }, i };
|
158 | }
|
159 |
|
160 | i++;
|
161 | break nameLoop;
|
162 | }
|
163 | break;
|
164 | }
|
165 |
|
166 | if (c === 0x2F || c === 0x3E) {
|
167 | return { attr: { name, value }, i };
|
168 | }
|
169 |
|
170 | if (c >= 0x41 && c <= 0x5A) {
|
171 | name += String.fromCharCode(c + 0x20);
|
172 | } else {
|
173 | name += String.fromCharCode(c);
|
174 | }
|
175 | }
|
176 | c = uint8Array[i];
|
177 |
|
178 | if (isSpaceCharacter(c)) {
|
179 | for (i++; i < l; i++) {
|
180 | c = uint8Array[i];
|
181 |
|
182 | if (isSpaceCharacter(c)) {
|
183 | continue;
|
184 | } else {
|
185 | break;
|
186 | }
|
187 | }
|
188 | }
|
189 |
|
190 | if (c === 0x22 || c === 0x27) {
|
191 | const quote = c;
|
192 | for (i++; i < l; i++) {
|
193 | c = uint8Array[i];
|
194 |
|
195 | if (c === quote) {
|
196 | i++;
|
197 | return { attr: { name, value }, i };
|
198 | }
|
199 |
|
200 |
|
201 | if (c >= 0x41 && c <= 0x5A) {
|
202 | value += String.fromCharCode(c + 0x20);
|
203 | } else {
|
204 | value += String.fromCharCode(c);
|
205 | }
|
206 | }
|
207 | }
|
208 |
|
209 |
|
210 | if (c === 0x3E) {
|
211 | return { attr: { name, value }, i };
|
212 | }
|
213 |
|
214 |
|
215 | if (c >= 0x41 && c <= 0x5A) {
|
216 | value += String.fromCharCode(c + 0x20);
|
217 | } else {
|
218 | value += String.fromCharCode(c);
|
219 | }
|
220 |
|
221 | for (i++; i < l; i++) {
|
222 | c = uint8Array[i];
|
223 |
|
224 |
|
225 | if (isSpaceCharacter(c) || c === 0x3E) {
|
226 | return { attr: { name, value }, i };
|
227 | }
|
228 |
|
229 |
|
230 | if (c >= 0x41 && c <= 0x5A) {
|
231 | value += String.fromCharCode(c + 0x20);
|
232 | } else {
|
233 | value += String.fromCharCode(c);
|
234 | }
|
235 | }
|
236 | }
|
237 | return { i };
|
238 | }
|
239 |
|
240 | function extractCharacterEncodingFromMeta(string) {
|
241 | let position = 0;
|
242 |
|
243 | while (true) {
|
244 | const indexOfCharset = string.substring(position).search(/charset/ui);
|
245 |
|
246 | if (indexOfCharset === -1) {
|
247 | return null;
|
248 | }
|
249 | let subPosition = position + indexOfCharset + "charset".length;
|
250 |
|
251 | while (isSpaceCharacter(string[subPosition].charCodeAt(0))) {
|
252 | ++subPosition;
|
253 | }
|
254 |
|
255 | if (string[subPosition] !== "=") {
|
256 | position = subPosition - 1;
|
257 | continue;
|
258 | }
|
259 |
|
260 | ++subPosition;
|
261 |
|
262 | while (isSpaceCharacter(string[subPosition].charCodeAt(0))) {
|
263 | ++subPosition;
|
264 | }
|
265 |
|
266 | position = subPosition;
|
267 | break;
|
268 | }
|
269 |
|
270 | if (string[position] === "\"" || string[position] === "'") {
|
271 | const nextIndex = string.indexOf(string[position], position + 1);
|
272 |
|
273 | if (nextIndex !== -1) {
|
274 | return whatwgEncoding.labelToName(string.substring(position + 1, nextIndex));
|
275 | }
|
276 |
|
277 |
|
278 | return null;
|
279 | }
|
280 |
|
281 | if (string.length === position + 1) {
|
282 | return null;
|
283 | }
|
284 |
|
285 | const indexOfASCIIWhitespaceOrSemicolon = string.substring(position + 1).search(/\x09|\x0A|\x0C|\x0D|\x20|;/u);
|
286 | const end = indexOfASCIIWhitespaceOrSemicolon === -1 ?
|
287 | string.length :
|
288 | position + indexOfASCIIWhitespaceOrSemicolon + 1;
|
289 |
|
290 | return whatwgEncoding.labelToName(string.substring(position, end));
|
291 | }
|
292 |
|
293 | function isSpaceCharacter(c) {
|
294 | return c === 0x09 || c === 0x0A || c === 0x0C || c === 0x0D || c === 0x20;
|
295 | }
|