@file-type/xml 0.2.1 → 0.3.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (4) hide show
  1. package/README.md +1 -1
  2. package/lib/index.d.ts +26 -0
  3. package/lib/index.js +133 -151
  4. package/package.json +55 -46
package/README.md CHANGED
@@ -24,7 +24,7 @@ console.log(fileType);
24
24
 
25
25
  You can also use the XML detector outside file-type:
26
26
  ```js
27
- import {XmlTextDetector} from 'index.js';
27
+ import {XmlTextDetector} from '@file-type/xml';
28
28
 
29
29
  xmlTextDetector.write('<svg xmlns="http://www.w3.org/2000/svg"><path fill="#00CD9F"/></svg>');
30
30
  const fileType = xmlTextDetector.fileType;
package/lib/index.d.ts ADDED
@@ -0,0 +1,26 @@
1
+ import { ITokenizer } from 'strtok3';
2
+ /**
3
+ * Temporary, should be taken from `file-type`
4
+ */
5
+ interface FileTypeResult {
6
+ ext: string;
7
+ mime: string;
8
+ }
9
+ export type Detector = (tokenizer: ITokenizer, fileType?: FileTypeResult) => Promise<FileTypeResult | undefined>;
10
+ interface IXmlTextDetectorOptions {
11
+ fullScan?: boolean;
12
+ }
13
+ export declare class XmlTextDetector {
14
+ private options;
15
+ private firstTag;
16
+ private parser;
17
+ private nesting;
18
+ onEnd: boolean;
19
+ fileType?: FileTypeResult;
20
+ constructor(options?: IXmlTextDetectorOptions);
21
+ write(text: string): void;
22
+ close(): void;
23
+ isValid(): boolean;
24
+ }
25
+ export declare const detectXml: Detector;
26
+ export {};
package/lib/index.js CHANGED
@@ -1,172 +1,154 @@
1
-
2
1
  import sax from 'sax';
3
-
4
2
  function startsWith(array, prefix) {
5
- if (prefix.length > array.length) {
6
- return false;
7
- }
8
- for (let i = 0; i < prefix.length; i++) {
9
- if (array[i] !== prefix[i]) {
10
- return false;
11
- }
12
- }
13
- return true;
3
+ if (prefix.length > array.length) {
4
+ return false;
5
+ }
6
+ for (let i = 0; i < prefix.length; i++) {
7
+ if (array[i] !== prefix[i]) {
8
+ return false;
9
+ }
10
+ }
11
+ return true;
14
12
  }
15
-
16
13
  function isXml(array) {
17
- if (startsWith(array,[60, 63, 120, 109, 108, 32])) {
18
- return {xml: true, encoding: 'utf-8', offset: 0}
19
- } else if (startsWith(array,[0xEF, 0xBB, 0xBF, 60, 63, 120, 109, 108, 32])) { // UTF-8 BOM
20
- return {xml: true, encoding: 'utf-8', offset: 3}
21
- } else if (startsWith(array,[0xFE, 0xFF, 0, 60, 0, 63, 0, 120, 0, 109, 0, 108, 0, 32 ])) {
22
- return {xml: true, encoding: 'utf-16be', offset: 2}
23
- } else if (startsWith(array,[0xFF, 0xFE, 60, 0, 63, 0, 120, 0, 109, 0, 108, 0, 32, 0 ])) {
24
- return {xml: true, encoding: 'utf-16le', offset: 2}
25
- } else if (startsWith(array,[0, 60, 0, 63, 0, 120, 0, 109, 0, 108, 0, 32 ])) {
26
- return {xml: true, encoding: 'utf-16be', offset: 0}
27
- } else if (startsWith(array,[60, 0, 63, 0, 120, 0, 109, 0, 108, 0, 32, 0 ])) {
28
- return {xml: true, encoding: 'utf-16le', offset: 0}
29
- }
30
- return {xml: false, encoding: undefined}
14
+ if (startsWith(array, [60, 63, 120, 109, 108, 32])) {
15
+ return { xml: true, encoding: 'utf-8', offset: 0 };
16
+ }
17
+ else if (startsWith(array, [0xEF, 0xBB, 0xBF, 60, 63, 120, 109, 108, 32])) { // UTF-8 BOM
18
+ return { xml: true, encoding: 'utf-8', offset: 3 };
19
+ }
20
+ else if (startsWith(array, [0xFE, 0xFF, 0, 60, 0, 63, 0, 120, 0, 109, 0, 108, 0, 32])) {
21
+ return { xml: true, encoding: 'utf-16be', offset: 2 };
22
+ }
23
+ else if (startsWith(array, [0xFF, 0xFE, 60, 0, 63, 0, 120, 0, 109, 0, 108, 0, 32, 0])) {
24
+ return { xml: true, encoding: 'utf-16le', offset: 2 };
25
+ }
26
+ else if (startsWith(array, [0, 60, 0, 63, 0, 120, 0, 109, 0, 108, 0, 32])) {
27
+ return { xml: true, encoding: 'utf-16be', offset: 0 };
28
+ }
29
+ else if (startsWith(array, [60, 0, 63, 0, 120, 0, 109, 0, 108, 0, 32, 0])) {
30
+ return { xml: true, encoding: 'utf-16le', offset: 0 };
31
+ }
32
+ return { xml: false };
31
33
  }
32
-
33
34
  function extractNsElement(node) {
34
- const parts = node.name.split(':');
35
- if(parts.length === 1) {
36
- return {name: parts[0], ns: node.attributes['xmlns']};
37
- } else if (parts.length === 2) {
38
- return {name: parts[1], ns: node.attributes[`xmlns:${parts[0]}`]};
39
- }
35
+ const parts = node.name.split(':');
36
+ if (parts.length === 1) {
37
+ return { name: parts[0], ns: node.attributes['xmlns'] };
38
+ }
39
+ else if (parts.length === 2) {
40
+ return { name: parts[1], ns: node.attributes[`xmlns:${parts[0]}`] };
41
+ }
40
42
  }
41
-
42
43
  /**
43
44
  * Maps the root element namespace to corresponding file-type
44
45
  */
45
46
  const namespaceMapping = {
46
- 'http://www.w3.org/2000/svg': {
47
- ext: 'svg',
48
- mime: 'image/svg+xml',
49
- },
50
- 'http://www.w3.org/1999/xhtml': {
51
- ext: 'xhtml',
52
- mime: 'application/xhtml+xml',
53
- },
54
- 'http://www.opengis.net/kml/2.2': {
55
- ext: 'kml',
56
- mime: 'application/vnd.google-earth.kml+xml',
57
- },
58
- 'http://www.opengis.net/gml': {
59
- ext: 'gml',
60
- mime: 'application/gml+xml',
61
- }
62
- }
63
-
47
+ 'http://www.w3.org/2000/svg': {
48
+ ext: 'svg',
49
+ mime: 'image/svg+xml'
50
+ },
51
+ 'http://www.w3.org/1999/xhtml': {
52
+ ext: 'xhtml',
53
+ mime: 'application/xhtml+xml'
54
+ },
55
+ 'http://www.opengis.net/kml/2.2': {
56
+ ext: 'kml',
57
+ mime: 'application/vnd.google-earth.kml+xml'
58
+ },
59
+ 'http://www.opengis.net/gml': {
60
+ ext: 'gml',
61
+ mime: 'application/gml+xml'
62
+ }
63
+ };
64
64
  /**
65
65
  * Maps the root element name to corresponding file-type.
66
66
  * Used for Non-namespaced XML
67
67
  * @type {{rss: {ext: string, mime: string}}}
68
68
  */
69
69
  const rootNameMapping = {
70
- rss: {
71
- ext: 'rss',
72
- mime: 'application/rss+xml',
73
- },
74
- 'score-partwise': {
75
- ext: 'musicxml',
76
- mime: 'application/vnd.recordare.musicxml+xml',
77
- },
78
- svg: {
79
- ext: 'svg',
80
- mime: 'image/svg+xml',
81
- },
82
- }
83
-
70
+ rss: {
71
+ ext: 'rss',
72
+ mime: 'application/rss+xml'
73
+ },
74
+ 'score-partwise': {
75
+ ext: 'musicxml',
76
+ mime: 'application/vnd.recordare.musicxml+xml'
77
+ },
78
+ svg: {
79
+ ext: 'svg',
80
+ mime: 'image/svg+xml'
81
+ }
82
+ };
84
83
  export class XmlTextDetector {
85
-
86
- constructor(options) {
87
- this.options = options ?? {};
88
- this.firstTag = true;
89
- this.onEnd = false;
90
- this.parser = sax.parser(true);
91
- this.nesting = 0;
92
-
93
- this.parser.onerror = e => {
94
- if (e.message.startsWith('Invalid character entity')) { // Allow entity reference
95
- return;
96
- }
97
- this.fileType = undefined;
98
- this.onEnd = true;
99
- };
100
- this.parser.onopentag = node => {
101
- ++this.nesting;
102
- if (!this.firstTag || this.onEnd) {
103
- return;
104
- }
105
- this.firstTag = false;
106
- const nsNode = extractNsElement(node);
107
- if (nsNode.ns) {
108
- // Resolve file-type boot root element namespace
109
- this.fileType = namespaceMapping[nsNode.ns.toLowerCase()];
110
- } else {
111
- // Fall back on element name if there is no namespace
112
- this.fileType = rootNameMapping[nsNode.name?.toLowerCase()];
113
- }
114
-
115
- if (this.fileType && !this.options.fullScan) {
84
+ constructor(options) {
85
+ this.options = options ?? {};
86
+ this.firstTag = true;
87
+ this.onEnd = false;
88
+ this.parser = sax.parser(true);
89
+ this.nesting = 0;
90
+ this.parser.onerror = e => {
91
+ if (e.message.startsWith('Invalid character entity')) { // Allow entity reference
92
+ return;
93
+ }
94
+ this.fileType = undefined;
95
+ this.onEnd = true;
96
+ };
97
+ this.parser.onopentag = node => {
98
+ ++this.nesting;
99
+ if (!this.firstTag || this.onEnd) {
100
+ return;
101
+ }
102
+ this.firstTag = false;
103
+ const nsNode = extractNsElement(node);
104
+ if (nsNode?.ns) {
105
+ // Resolve file-type boot root element namespace
106
+ this.fileType = namespaceMapping[nsNode.ns.toLowerCase()];
107
+ }
108
+ else if (nsNode && nsNode.name) {
109
+ // Fall back on element name if there is no namespace
110
+ this.fileType = rootNameMapping[nsNode.name.toLowerCase()];
111
+ }
112
+ if (this.fileType && !this.options.fullScan) {
113
+ this.onEnd = true;
114
+ }
115
+ };
116
+ this.parser.onclosetag = () => {
117
+ --this.nesting;
118
+ };
119
+ }
120
+ write(text) {
121
+ this.parser.write(text);
122
+ }
123
+ close() {
124
+ this.parser.close();
116
125
  this.onEnd = true;
117
- }
118
- };
119
-
120
- this.parser.onclosetag = () => {
121
- --this.nesting;
122
- }
123
- }
124
-
125
- write(text) {
126
- this.parser.write(text);
127
- }
128
-
129
- close() {
130
- this.parser.close();
131
- this.onEnd = true;
132
- }
133
-
134
- isValid() {
135
- return this.nesting === 0;
136
- }
126
+ }
127
+ isValid() {
128
+ return this.nesting === 0;
129
+ }
137
130
  }
138
-
139
- export const detectXml = async tokenizer => {
140
-
141
- const buffer = new Uint8Array(512);
142
-
143
- // Increase sample size from 12 to 256.
144
- await tokenizer.peekBuffer(buffer, {length: 128, mayBeLess: true});
145
-
146
- const {xml, encoding, offset} = isXml(buffer);
147
-
148
- if (xml) {
149
- await tokenizer.ignore(offset);
150
-
151
- const xmlTextDetector = new XmlTextDetector();
152
-
153
- const textDecoder = new TextDecoder(encoding);
154
-
155
- do {
156
- const len = await tokenizer.readBuffer(buffer, {mayBeLess: true});
157
- const portion = buffer.subarray(0, len);
158
- const text = textDecoder.decode(portion);
159
- xmlTextDetector.write(text);
160
- if (len < buffer.length) {
161
- xmlTextDetector.close();
162
- }
163
- } while(!xmlTextDetector.onEnd)
164
-
165
- return xmlTextDetector.fileType ?? {
166
- ext: 'xml',
167
- mime: 'application/xml',
168
- }
169
- }
170
-
131
+ export const detectXml = async (tokenizer) => {
132
+ const buffer = new Uint8Array(512);
133
+ // Increase sample size from 12 to 256.
134
+ await tokenizer.peekBuffer(buffer, { length: 128, mayBeLess: true });
135
+ const xmlDetection = isXml(buffer);
136
+ if (xmlDetection.xml) {
137
+ await tokenizer.ignore(xmlDetection.offset);
138
+ const xmlTextDetector = new XmlTextDetector();
139
+ const textDecoder = new TextDecoder(xmlDetection.encoding);
140
+ do {
141
+ const len = await tokenizer.readBuffer(buffer, { mayBeLess: true });
142
+ const portion = buffer.subarray(0, len);
143
+ const text = textDecoder.decode(portion);
144
+ xmlTextDetector.write(text);
145
+ if (len < buffer.length) {
146
+ xmlTextDetector.close();
147
+ }
148
+ } while (!xmlTextDetector.onEnd);
149
+ return xmlTextDetector.fileType ?? {
150
+ ext: 'xml',
151
+ mime: 'application/xml'
152
+ };
153
+ }
171
154
  };
172
-
package/package.json CHANGED
@@ -1,48 +1,57 @@
1
1
  {
2
- "name": "@file-type/xml",
3
- "version": "0.2.1",
4
- "description": "XML detection plugin",
5
- "type": "module",
6
- "exports": "./lib/index.js",
7
- "scripts": {
8
- "test": "mocha"
9
- },
10
- "keywords": [
11
- "file-type",
12
- "detect",
13
- "detection",
14
- "detector",
15
- "XML",
16
- "signature",
17
- "namespace",
18
- "SVG",
19
- "XHTML",
20
- "RSS",
21
- "KML",
22
- "GML",
23
- "MusicXML"
24
- ],
25
- "dependencies": {
26
- "sax": "^1.4.1",
27
- "strtok3": "^10.0.1"
28
- },
29
- "devDependencies": {
30
- "chai": "^5.1.2",
31
- "mocha": "^11.0.1"
32
- },
33
- "files": [
34
- "lib/**/*.js",
35
- "lib/**/*.d.ts",
36
- "lib/*.cjs"
37
- ],
38
- "author": {
39
- "name": "Borewit",
40
- "url": "https://github.com/Borewit"
41
- },
42
- "repository": {
43
- "type": "git",
44
- "url": "git+https://github.com/Borewit/file-type-xml.git"
45
- },
46
- "license": "MIT",
47
- "packageManager": "yarn@4.5.3"
2
+ "name": "@file-type/xml",
3
+ "version": "0.3.0",
4
+ "description": "XML detection plugin",
5
+ "type": "module",
6
+ "types": "./lib/index.d.ts",
7
+ "exports": "./lib/index.js",
8
+ "scripts": {
9
+ "clean": "del-cli 'lib/**/*.js' 'lib/**/*.js.map' 'lib/**/*.d.ts' 'src/**/*.d.ts'",
10
+ "compile-src": "tsc -p lib",
11
+ "compile": "yarn run compile-src",
12
+ "build": "yarn run clean && yarn compile",
13
+ "test": "mocha"
14
+ },
15
+ "keywords": [
16
+ "file-type",
17
+ "detect",
18
+ "detection",
19
+ "detector",
20
+ "XML",
21
+ "signature",
22
+ "namespace",
23
+ "SVG",
24
+ "XHTML",
25
+ "RSS",
26
+ "KML",
27
+ "GML",
28
+ "MusicXML"
29
+ ],
30
+ "dependencies": {
31
+ "sax": "^1.4.1",
32
+ "strtok3": "^10.0.1"
33
+ },
34
+ "devDependencies": {
35
+ "@types/sax": "^1.2.7",
36
+ "chai": "^5.1.2",
37
+ "del-cli": "^6.0.0",
38
+ "file-type": "^19.6.0",
39
+ "mocha": "^11.0.1",
40
+ "typescript": "^5.7.2"
41
+ },
42
+ "files": [
43
+ "lib/**/*.js",
44
+ "lib/**/*.d.ts",
45
+ "lib/*.cjs"
46
+ ],
47
+ "author": {
48
+ "name": "Borewit",
49
+ "url": "https://github.com/Borewit"
50
+ },
51
+ "repository": {
52
+ "type": "git",
53
+ "url": "git+https://github.com/Borewit/file-type-xml.git"
54
+ },
55
+ "license": "MIT",
56
+ "packageManager": "yarn@4.5.3"
48
57
  }