gff-nostream 1.3.1
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.
- package/LICENSE +21 -0
- package/README.md +163 -0
- package/dist/api.d.ts +23 -0
- package/dist/api.js +40 -0
- package/dist/api.js.map +1 -0
- package/dist/index.d.ts +3 -0
- package/dist/index.js +6 -0
- package/dist/index.js.map +1 -0
- package/dist/parse.d.ts +55 -0
- package/dist/parse.js +341 -0
- package/dist/parse.js.map +1 -0
- package/dist/util.d.ts +166 -0
- package/dist/util.js +274 -0
- package/dist/util.js.map +1 -0
- package/esm/api.d.ts +23 -0
- package/esm/api.js +42 -0
- package/esm/api.js.map +1 -0
- package/esm/index.d.ts +3 -0
- package/esm/index.js +3 -0
- package/esm/index.js.map +1 -0
- package/esm/parse.d.ts +55 -0
- package/esm/parse.js +317 -0
- package/esm/parse.js.map +1 -0
- package/esm/util.d.ts +166 -0
- package/esm/util.js +268 -0
- package/esm/util.js.map +1 -0
- package/package.json +52 -0
- package/src/api.ts +86 -0
- package/src/index.ts +12 -0
- package/src/parse.ts +400 -0
- package/src/util.ts +425 -0
package/esm/util.js
ADDED
|
@@ -0,0 +1,268 @@
|
|
|
1
|
+
// Fast, low-level functions for parsing and formatting GFF3.
|
|
2
|
+
// JavaScript port of Robert Buels's Bio::GFF3::LowLevel Perl module.
|
|
3
|
+
/**
|
|
4
|
+
* Unescape a string value used in a GFF3 attribute.
|
|
5
|
+
*
|
|
6
|
+
* @param stringVal - Escaped GFF3 string value
|
|
7
|
+
* @returns An unescaped string value
|
|
8
|
+
*/
|
|
9
|
+
export function unescape(stringVal) {
|
|
10
|
+
return stringVal.replaceAll(/%([0-9A-Fa-f]{2})/g, (_match, seq) => String.fromCharCode(parseInt(seq, 16)));
|
|
11
|
+
}
|
|
12
|
+
function _escape(regex, s) {
|
|
13
|
+
return String(s).replace(regex, ch => {
|
|
14
|
+
const hex = ch.charCodeAt(0).toString(16).toUpperCase().padStart(2, '0');
|
|
15
|
+
return `%${hex}`;
|
|
16
|
+
});
|
|
17
|
+
}
|
|
18
|
+
/**
|
|
19
|
+
* Escape a value for use in a GFF3 attribute value.
|
|
20
|
+
*
|
|
21
|
+
* @param rawVal - Raw GFF3 attribute value
|
|
22
|
+
* @returns An escaped string value
|
|
23
|
+
*/
|
|
24
|
+
export function escape(rawVal) {
|
|
25
|
+
return _escape(/[\n;\r\t=%&,\u0000-\u001f\u007f-\u00ff]/g, rawVal);
|
|
26
|
+
}
|
|
27
|
+
/**
|
|
28
|
+
* Escape a value for use in a GFF3 column value.
|
|
29
|
+
*
|
|
30
|
+
* @param rawVal - Raw GFF3 column value
|
|
31
|
+
* @returns An escaped column value
|
|
32
|
+
*/
|
|
33
|
+
export function escapeColumn(rawVal) {
|
|
34
|
+
return _escape(/[\n\r\t%\u0000-\u001f\u007f-\u00ff]/g, rawVal);
|
|
35
|
+
}
|
|
36
|
+
/**
|
|
37
|
+
* Parse the 9th column (attributes) of a GFF3 feature line.
|
|
38
|
+
*
|
|
39
|
+
* @param attrString - String of GFF3 9th column
|
|
40
|
+
* @returns Parsed attributes
|
|
41
|
+
*/
|
|
42
|
+
export function parseAttributes(attrString) {
|
|
43
|
+
if (!attrString.length || attrString === '.') {
|
|
44
|
+
return {};
|
|
45
|
+
}
|
|
46
|
+
const attrs = {};
|
|
47
|
+
attrString
|
|
48
|
+
.replace(/\r?\n$/, '')
|
|
49
|
+
.split(';')
|
|
50
|
+
.forEach(a => {
|
|
51
|
+
var _a;
|
|
52
|
+
const nv = a.split('=', 2);
|
|
53
|
+
if (!((_a = nv[1]) === null || _a === void 0 ? void 0 : _a.length)) {
|
|
54
|
+
return;
|
|
55
|
+
}
|
|
56
|
+
nv[0] = nv[0].trim();
|
|
57
|
+
let arec = attrs[nv[0].trim()];
|
|
58
|
+
if (!arec) {
|
|
59
|
+
arec = [];
|
|
60
|
+
attrs[nv[0]] = arec;
|
|
61
|
+
}
|
|
62
|
+
arec.push(...nv[1]
|
|
63
|
+
.split(',')
|
|
64
|
+
.map(s => s.trim())
|
|
65
|
+
.map(unescape));
|
|
66
|
+
});
|
|
67
|
+
return attrs;
|
|
68
|
+
}
|
|
69
|
+
/**
|
|
70
|
+
* Parse a GFF3 feature line
|
|
71
|
+
*
|
|
72
|
+
* @param line - GFF3 feature line
|
|
73
|
+
* @returns The parsed feature
|
|
74
|
+
*/
|
|
75
|
+
export function parseFeature(line) {
|
|
76
|
+
// split the line into columns and replace '.' with null in each column
|
|
77
|
+
const f = line.split('\t').map(a => (a === '.' || a === '' ? null : a));
|
|
78
|
+
// unescape only the ref, source, and type columns
|
|
79
|
+
const parsed = {
|
|
80
|
+
seq_id: f[0] && unescape(f[0]),
|
|
81
|
+
source: f[1] && unescape(f[1]),
|
|
82
|
+
type: f[2] && unescape(f[2]),
|
|
83
|
+
start: f[3] === null ? null : parseInt(f[3], 10),
|
|
84
|
+
end: f[4] === null ? null : parseInt(f[4], 10),
|
|
85
|
+
score: f[5] === null ? null : parseFloat(f[5]),
|
|
86
|
+
strand: f[6],
|
|
87
|
+
phase: f[7],
|
|
88
|
+
attributes: f[8] === null ? null : parseAttributes(f[8]),
|
|
89
|
+
};
|
|
90
|
+
return parsed;
|
|
91
|
+
}
|
|
92
|
+
/**
|
|
93
|
+
* Parse a GFF3 directive line.
|
|
94
|
+
*
|
|
95
|
+
* @param line - GFF3 directive line
|
|
96
|
+
* @returns The parsed directive
|
|
97
|
+
*/
|
|
98
|
+
export function parseDirective(line) {
|
|
99
|
+
var _a, _b;
|
|
100
|
+
const match = /^\s*##\s*(\S+)\s*(.*)/.exec(line);
|
|
101
|
+
if (!match) {
|
|
102
|
+
return null;
|
|
103
|
+
}
|
|
104
|
+
const [, name] = match;
|
|
105
|
+
let [, , contents] = match;
|
|
106
|
+
const parsed = { directive: name };
|
|
107
|
+
if (contents.length) {
|
|
108
|
+
contents = contents.replace(/\r?\n$/, '');
|
|
109
|
+
parsed.value = contents;
|
|
110
|
+
}
|
|
111
|
+
// do a little additional parsing for sequence-region and genome-build directives
|
|
112
|
+
if (name === 'sequence-region') {
|
|
113
|
+
const c = contents.split(/\s+/, 3);
|
|
114
|
+
return {
|
|
115
|
+
...parsed,
|
|
116
|
+
seq_id: c[0],
|
|
117
|
+
start: (_a = c[1]) === null || _a === void 0 ? void 0 : _a.replaceAll(/\D/g, ''),
|
|
118
|
+
end: (_b = c[2]) === null || _b === void 0 ? void 0 : _b.replaceAll(/\D/g, ''),
|
|
119
|
+
};
|
|
120
|
+
}
|
|
121
|
+
else if (name === 'genome-build') {
|
|
122
|
+
const [source, buildName] = contents.split(/\s+/, 2);
|
|
123
|
+
return {
|
|
124
|
+
...parsed,
|
|
125
|
+
source,
|
|
126
|
+
buildName,
|
|
127
|
+
};
|
|
128
|
+
}
|
|
129
|
+
return parsed;
|
|
130
|
+
}
|
|
131
|
+
/**
|
|
132
|
+
* Format an attributes object into a string suitable for the 9th column of GFF3.
|
|
133
|
+
*
|
|
134
|
+
* @param attrs - Attributes
|
|
135
|
+
* @returns GFF3 9th column string
|
|
136
|
+
*/
|
|
137
|
+
export function formatAttributes(attrs) {
|
|
138
|
+
const attrOrder = [];
|
|
139
|
+
Object.entries(attrs).forEach(([tag, val]) => {
|
|
140
|
+
if (!val) {
|
|
141
|
+
return;
|
|
142
|
+
}
|
|
143
|
+
let valstring;
|
|
144
|
+
if (val.hasOwnProperty('toString')) {
|
|
145
|
+
valstring = escape(val.toString());
|
|
146
|
+
// } else if (Array.isArray(val.values)) {
|
|
147
|
+
// valstring = val.values.map(escape).join(',')
|
|
148
|
+
}
|
|
149
|
+
else if (Array.isArray(val)) {
|
|
150
|
+
valstring = val.map(escape).join(',');
|
|
151
|
+
}
|
|
152
|
+
else {
|
|
153
|
+
valstring = escape(val);
|
|
154
|
+
}
|
|
155
|
+
attrOrder.push(`${escape(tag)}=${valstring}`);
|
|
156
|
+
});
|
|
157
|
+
return attrOrder.length ? attrOrder.join(';') : '.';
|
|
158
|
+
}
|
|
159
|
+
function _formatSingleFeature(f, seenFeature) {
|
|
160
|
+
const attrString = f.attributes === null || f.attributes === undefined
|
|
161
|
+
? '.'
|
|
162
|
+
: formatAttributes(f.attributes);
|
|
163
|
+
const fields = [
|
|
164
|
+
f.seq_id === null ? '.' : escapeColumn(f.seq_id),
|
|
165
|
+
f.source === null ? '.' : escapeColumn(f.source),
|
|
166
|
+
f.type === null ? '.' : escapeColumn(f.type),
|
|
167
|
+
f.start === null ? '.' : escapeColumn(f.start),
|
|
168
|
+
f.end === null ? '.' : escapeColumn(f.end),
|
|
169
|
+
f.score === null ? '.' : escapeColumn(f.score),
|
|
170
|
+
f.strand === null ? '.' : escapeColumn(f.strand),
|
|
171
|
+
f.phase === null ? '.' : escapeColumn(f.phase),
|
|
172
|
+
attrString,
|
|
173
|
+
];
|
|
174
|
+
const formattedString = `${fields.join('\t')}\n`;
|
|
175
|
+
// if we have already output this exact feature, skip it
|
|
176
|
+
if (seenFeature[formattedString]) {
|
|
177
|
+
return '';
|
|
178
|
+
}
|
|
179
|
+
seenFeature[formattedString] = true;
|
|
180
|
+
return formattedString;
|
|
181
|
+
}
|
|
182
|
+
function _formatFeature(feature, seenFeature) {
|
|
183
|
+
if (Array.isArray(feature)) {
|
|
184
|
+
return feature.map(f => _formatFeature(f, seenFeature)).join('');
|
|
185
|
+
}
|
|
186
|
+
const strings = [_formatSingleFeature(feature, seenFeature)];
|
|
187
|
+
if (_isFeatureLineWithRefs(feature)) {
|
|
188
|
+
strings.push(...feature.child_features.map(f => _formatFeature(f, seenFeature)), ...feature.derived_features.map(f => _formatFeature(f, seenFeature)));
|
|
189
|
+
}
|
|
190
|
+
return strings.join('');
|
|
191
|
+
}
|
|
192
|
+
/**
|
|
193
|
+
* Format a feature object or array of feature objects into one or more lines of
|
|
194
|
+
* GFF3.
|
|
195
|
+
*
|
|
196
|
+
* @param featureOrFeatures - A feature object or array of feature objects
|
|
197
|
+
* @returns A string of one or more GFF3 lines
|
|
198
|
+
*/
|
|
199
|
+
export function formatFeature(featureOrFeatures) {
|
|
200
|
+
const seen = {};
|
|
201
|
+
return _formatFeature(featureOrFeatures, seen);
|
|
202
|
+
}
|
|
203
|
+
/**
|
|
204
|
+
* Format a directive into a line of GFF3.
|
|
205
|
+
*
|
|
206
|
+
* @param directive - A directive object
|
|
207
|
+
* @returns A directive line string
|
|
208
|
+
*/
|
|
209
|
+
export function formatDirective(directive) {
|
|
210
|
+
let str = `##${directive.directive}`;
|
|
211
|
+
if (directive.value) {
|
|
212
|
+
str += ` ${directive.value}`;
|
|
213
|
+
}
|
|
214
|
+
str += '\n';
|
|
215
|
+
return str;
|
|
216
|
+
}
|
|
217
|
+
/**
|
|
218
|
+
* Format a comment into a GFF3 comment.
|
|
219
|
+
* Yes I know this is just adding a # and a newline.
|
|
220
|
+
*
|
|
221
|
+
* @param comment - A comment object
|
|
222
|
+
* @returns A comment line string
|
|
223
|
+
*/
|
|
224
|
+
export function formatComment(comment) {
|
|
225
|
+
return `# ${comment.comment}\n`;
|
|
226
|
+
}
|
|
227
|
+
/**
|
|
228
|
+
* Format a sequence object as FASTA
|
|
229
|
+
*
|
|
230
|
+
* @param seq - A sequence object
|
|
231
|
+
* @returns Formatted single FASTA sequence string
|
|
232
|
+
*/
|
|
233
|
+
export function formatSequence(seq) {
|
|
234
|
+
return `>${seq.id}${seq.description ? ` ${seq.description}` : ''}\n${seq.sequence}\n`;
|
|
235
|
+
}
|
|
236
|
+
/**
|
|
237
|
+
* Format a directive, comment, sequence, or feature, or array of such items,
|
|
238
|
+
* into one or more lines of GFF3.
|
|
239
|
+
*
|
|
240
|
+
* @param itemOrItems - A comment, sequence, or feature, or array of such items
|
|
241
|
+
* @returns A formatted string or array of strings
|
|
242
|
+
*/
|
|
243
|
+
export function formatItem(itemOrItems) {
|
|
244
|
+
function formatSingleItem(item) {
|
|
245
|
+
if ('attributes' in item) {
|
|
246
|
+
return formatFeature(item);
|
|
247
|
+
}
|
|
248
|
+
if ('directive' in item) {
|
|
249
|
+
return formatDirective(item);
|
|
250
|
+
}
|
|
251
|
+
if ('sequence' in item) {
|
|
252
|
+
return formatSequence(item);
|
|
253
|
+
}
|
|
254
|
+
if ('comment' in item) {
|
|
255
|
+
return formatComment(item);
|
|
256
|
+
}
|
|
257
|
+
return '# (invalid item found during format)\n';
|
|
258
|
+
}
|
|
259
|
+
if (Array.isArray(itemOrItems)) {
|
|
260
|
+
return itemOrItems.map(formatSingleItem);
|
|
261
|
+
}
|
|
262
|
+
return formatSingleItem(itemOrItems);
|
|
263
|
+
}
|
|
264
|
+
function _isFeatureLineWithRefs(featureLine) {
|
|
265
|
+
return (featureLine.child_features !== undefined &&
|
|
266
|
+
featureLine.derived_features !== undefined);
|
|
267
|
+
}
|
|
268
|
+
//# sourceMappingURL=util.js.map
|
package/esm/util.js.map
ADDED
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"util.js","sourceRoot":"","sources":["../src/util.ts"],"names":[],"mappings":"AAAA,6DAA6D;AAC7D,qEAAqE;AAErE;;;;;GAKG;AACH,MAAM,UAAU,QAAQ,CAAC,SAAiB;IACxC,OAAO,SAAS,CAAC,UAAU,CAAC,oBAAoB,EAAE,CAAC,MAAM,EAAE,GAAG,EAAE,EAAE,CAChE,MAAM,CAAC,YAAY,CAAC,QAAQ,CAAC,GAAG,EAAE,EAAE,CAAC,CAAC,CACvC,CAAA;AACH,CAAC;AAED,SAAS,OAAO,CAAC,KAAa,EAAE,CAAkB;IAChD,OAAO,MAAM,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,KAAK,EAAE,EAAE,CAAC,EAAE;QACnC,MAAM,GAAG,GAAG,EAAE,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,EAAE,CAAC,CAAC,WAAW,EAAE,CAAC,QAAQ,CAAC,CAAC,EAAE,GAAG,CAAC,CAAA;QACxE,OAAO,IAAI,GAAG,EAAE,CAAA;IAClB,CAAC,CAAC,CAAA;AACJ,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,MAAM,CAAC,MAAuB;IAC5C,OAAO,OAAO,CAAC,0CAA0C,EAAE,MAAM,CAAC,CAAA;AACpE,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,YAAY,CAAC,MAAuB;IAClD,OAAO,OAAO,CAAC,sCAAsC,EAAE,MAAM,CAAC,CAAA;AAChE,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,eAAe,CAAC,UAAkB;IAChD,IAAI,CAAC,UAAU,CAAC,MAAM,IAAI,UAAU,KAAK,GAAG,EAAE,CAAC;QAC7C,OAAO,EAAE,CAAA;IACX,CAAC;IAED,MAAM,KAAK,GAAmB,EAAE,CAAA;IAEhC,UAAU;SACP,OAAO,CAAC,QAAQ,EAAE,EAAE,CAAC;SACrB,KAAK,CAAC,GAAG,CAAC;SACV,OAAO,CAAC,CAAC,CAAC,EAAE;;QACX,MAAM,EAAE,GAAG,CAAC,CAAC,KAAK,CAAC,GAAG,EAAE,CAAC,CAAC,CAAA;QAC1B,IAAI,CAAC,CAAA,MAAA,EAAE,CAAC,CAAC,CAAC,0CAAE,MAAM,CAAA,EAAE,CAAC;YACnB,OAAM;QACR,CAAC;QAED,EAAE,CAAC,CAAC,CAAC,GAAG,EAAE,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAA;QACpB,IAAI,IAAI,GAAG,KAAK,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAA;QAC9B,IAAI,CAAC,IAAI,EAAE,CAAC;YACV,IAAI,GAAG,EAAE,CAAA;YACT,KAAK,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,GAAG,IAAI,CAAA;QACrB,CAAC;QAED,IAAI,CAAC,IAAI,CACP,GAAG,EAAE,CAAC,CAAC,CAAC;aACL,KAAK,CAAC,GAAG,CAAC;aACV,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;aAClB,GAAG,CAAC,QAAQ,CAAC,CACjB,CAAA;IACH,CAAC,CAAC,CAAA;IACJ,OAAO,KAAK,CAAA;AACd,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,YAAY,CAAC,IAAY;IACvC,uEAAuE;IACvE,MAAM,CAAC,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,KAAK,GAAG,IAAI,CAAC,KAAK,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAA;IAEvE,kDAAkD;IAClD,MAAM,MAAM,GAAoB;QAC9B,MAAM,EAAE,CAAC,CAAC,CAAC,CAAC,IAAI,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;QAC9B,MAAM,EAAE,CAAC,CAAC,CAAC,CAAC,IAAI,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;QAC9B,IAAI,EAAE,CAAC,CAAC,CAAC,CAAC,IAAI,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;QAC5B,KAAK,EAAE,CAAC,CAAC,CAAC,CAAC,KAAK,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC;QAChD,GAAG,EAAE,CAAC,CAAC,CAAC,CAAC,KAAK,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC;QAC9C,KAAK,EAAE,CAAC,CAAC,CAAC,CAAC,KAAK,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;QAC9C,MAAM,EAAE,CAAC,CAAC,CAAC,CAAC;QACZ,KAAK,EAAE,CAAC,CAAC,CAAC,CAAC;QACX,UAAU,EAAE,CAAC,CAAC,CAAC,CAAC,KAAK,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,eAAe,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;KACzD,CAAA;IACD,OAAO,MAAM,CAAA;AACf,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,cAAc,CAC5B,IAAY;;IAMZ,MAAM,KAAK,GAAG,uBAAuB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;IAChD,IAAI,CAAC,KAAK,EAAE,CAAC;QACX,OAAO,IAAI,CAAA;IACb,CAAC;IAED,MAAM,CAAC,EAAE,IAAI,CAAC,GAAG,KAAK,CAAA;IACtB,IAAI,CAAC,EAAE,AAAD,EAAG,QAAQ,CAAC,GAAG,KAAK,CAAA;IAE1B,MAAM,MAAM,GAAkB,EAAE,SAAS,EAAE,IAAI,EAAE,CAAA;IACjD,IAAI,QAAQ,CAAC,MAAM,EAAE,CAAC;QACpB,QAAQ,GAAG,QAAQ,CAAC,OAAO,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAA;QACzC,MAAM,CAAC,KAAK,GAAG,QAAQ,CAAA;IACzB,CAAC;IAED,iFAAiF;IACjF,IAAI,IAAI,KAAK,iBAAiB,EAAE,CAAC;QAC/B,MAAM,CAAC,GAAG,QAAQ,CAAC,KAAK,CAAC,KAAK,EAAE,CAAC,CAAC,CAAA;QAClC,OAAO;YACL,GAAG,MAAM;YACT,MAAM,EAAE,CAAC,CAAC,CAAC,CAAC;YACZ,KAAK,EAAE,MAAA,CAAC,CAAC,CAAC,CAAC,0CAAE,UAAU,CAAC,KAAK,EAAE,EAAE,CAAC;YAClC,GAAG,EAAE,MAAA,CAAC,CAAC,CAAC,CAAC,0CAAE,UAAU,CAAC,KAAK,EAAE,EAAE,CAAC;SACF,CAAA;IAClC,CAAC;SAAM,IAAI,IAAI,KAAK,cAAc,EAAE,CAAC;QACnC,MAAM,CAAC,MAAM,EAAE,SAAS,CAAC,GAAG,QAAQ,CAAC,KAAK,CAAC,KAAK,EAAE,CAAC,CAAC,CAAA;QACpD,OAAO;YACL,GAAG,MAAM;YACT,MAAM;YACN,SAAS;SACkB,CAAA;IAC/B,CAAC;IAED,OAAO,MAAM,CAAA;AACf,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,gBAAgB,CAAC,KAAqB;IACpD,MAAM,SAAS,GAAa,EAAE,CAAA;IAC9B,MAAM,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,GAAG,EAAE,GAAG,CAAC,EAAE,EAAE;QAC3C,IAAI,CAAC,GAAG,EAAE,CAAC;YACT,OAAM;QACR,CAAC;QACD,IAAI,SAAS,CAAA;QACb,IAAI,GAAG,CAAC,cAAc,CAAC,UAAU,CAAC,EAAE,CAAC;YACnC,SAAS,GAAG,MAAM,CAAC,GAAG,CAAC,QAAQ,EAAE,CAAC,CAAA;YAClC,0CAA0C;YAC1C,iDAAiD;QACnD,CAAC;aAAM,IAAI,KAAK,CAAC,OAAO,CAAC,GAAG,CAAC,EAAE,CAAC;YAC9B,SAAS,GAAG,GAAG,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAA;QACvC,CAAC;aAAM,CAAC;YACN,SAAS,GAAG,MAAM,CAAC,GAAG,CAAC,CAAA;QACzB,CAAC;QACD,SAAS,CAAC,IAAI,CAAC,GAAG,MAAM,CAAC,GAAG,CAAC,IAAI,SAAS,EAAE,CAAC,CAAA;IAC/C,CAAC,CAAC,CAAA;IACF,OAAO,SAAS,CAAC,MAAM,CAAC,CAAC,CAAC,SAAS,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,GAAG,CAAA;AACrD,CAAC;AAED,SAAS,oBAAoB,CAC3B,CAA4C,EAC5C,WAAgD;IAEhD,MAAM,UAAU,GACd,CAAC,CAAC,UAAU,KAAK,IAAI,IAAI,CAAC,CAAC,UAAU,KAAK,SAAS;QACjD,CAAC,CAAC,GAAG;QACL,CAAC,CAAC,gBAAgB,CAAC,CAAC,CAAC,UAAU,CAAC,CAAA;IAEpC,MAAM,MAAM,GAAG;QACb,CAAC,CAAC,MAAM,KAAK,IAAI,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,MAAM,CAAC;QAChD,CAAC,CAAC,MAAM,KAAK,IAAI,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,MAAM,CAAC;QAChD,CAAC,CAAC,IAAI,KAAK,IAAI,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,IAAI,CAAC;QAC5C,CAAC,CAAC,KAAK,KAAK,IAAI,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,KAAK,CAAC;QAC9C,CAAC,CAAC,GAAG,KAAK,IAAI,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,GAAG,CAAC;QAC1C,CAAC,CAAC,KAAK,KAAK,IAAI,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,KAAK,CAAC;QAC9C,CAAC,CAAC,MAAM,KAAK,IAAI,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,MAAM,CAAC;QAChD,CAAC,CAAC,KAAK,KAAK,IAAI,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,KAAK,CAAC;QAC9C,UAAU;KACX,CAAA;IAED,MAAM,eAAe,GAAG,GAAG,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAA;IAEhD,wDAAwD;IACxD,IAAI,WAAW,CAAC,eAAe,CAAC,EAAE,CAAC;QACjC,OAAO,EAAE,CAAA;IACX,CAAC;IAED,WAAW,CAAC,eAAe,CAAC,GAAG,IAAI,CAAA;IACnC,OAAO,eAAe,CAAA;AACxB,CAAC;AAED,SAAS,cAAc,CACrB,OAGiD,EACjD,WAAgD;IAEhD,IAAI,KAAK,CAAC,OAAO,CAAC,OAAO,CAAC,EAAE,CAAC;QAC3B,OAAO,OAAO,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,cAAc,CAAC,CAAC,EAAE,WAAW,CAAC,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,CAAA;IAClE,CAAC;IAED,MAAM,OAAO,GAAG,CAAC,oBAAoB,CAAC,OAAO,EAAE,WAAW,CAAC,CAAC,CAAA;IAC5D,IAAI,sBAAsB,CAAC,OAAO,CAAC,EAAE,CAAC;QACpC,OAAO,CAAC,IAAI,CACV,GAAG,OAAO,CAAC,cAAc,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,cAAc,CAAC,CAAC,EAAE,WAAW,CAAC,CAAC,EAClE,GAAG,OAAO,CAAC,gBAAgB,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,cAAc,CAAC,CAAC,EAAE,WAAW,CAAC,CAAC,CACrE,CAAA;IACH,CAAC;IACD,OAAO,OAAO,CAAC,IAAI,CAAC,EAAE,CAAC,CAAA;AACzB,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,aAAa,CAC3B,iBAGiD;IAEjD,MAAM,IAAI,GAAG,EAAE,CAAA;IACf,OAAO,cAAc,CAAC,iBAAiB,EAAE,IAAI,CAAC,CAAA;AAChD,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,eAAe,CAAC,SAAwB;IACtD,IAAI,GAAG,GAAG,KAAK,SAAS,CAAC,SAAS,EAAE,CAAA;IACpC,IAAI,SAAS,CAAC,KAAK,EAAE,CAAC;QACpB,GAAG,IAAI,IAAI,SAAS,CAAC,KAAK,EAAE,CAAA;IAC9B,CAAC;IACD,GAAG,IAAI,IAAI,CAAA;IACX,OAAO,GAAG,CAAA;AACZ,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,aAAa,CAAC,OAAoB;IAChD,OAAO,KAAK,OAAO,CAAC,OAAO,IAAI,CAAA;AACjC,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,cAAc,CAAC,GAAiB;IAC9C,OAAO,IAAI,GAAG,CAAC,EAAE,GAAG,GAAG,CAAC,WAAW,CAAC,CAAC,CAAC,IAAI,GAAG,CAAC,WAAW,EAAE,CAAC,CAAC,CAAC,EAAE,KAC9D,GAAG,CAAC,QACN,IAAI,CAAA;AACN,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,UAAU,CACxB,WAK4E;IAE5E,SAAS,gBAAgB,CACvB,IAA0E;QAE1E,IAAI,YAAY,IAAI,IAAI,EAAE,CAAC;YACzB,OAAO,aAAa,CAAC,IAAI,CAAC,CAAA;QAC5B,CAAC;QACD,IAAI,WAAW,IAAI,IAAI,EAAE,CAAC;YACxB,OAAO,eAAe,CAAC,IAAI,CAAC,CAAA;QAC9B,CAAC;QACD,IAAI,UAAU,IAAI,IAAI,EAAE,CAAC;YACvB,OAAO,cAAc,CAAC,IAAI,CAAC,CAAA;QAC7B,CAAC;QACD,IAAI,SAAS,IAAI,IAAI,EAAE,CAAC;YACtB,OAAO,aAAa,CAAC,IAAI,CAAC,CAAA;QAC5B,CAAC;QACD,OAAO,wCAAwC,CAAA;IACjD,CAAC;IAED,IAAI,KAAK,CAAC,OAAO,CAAC,WAAW,CAAC,EAAE,CAAC;QAC/B,OAAO,WAAW,CAAC,GAAG,CAAC,gBAAgB,CAAC,CAAA;IAC1C,CAAC;IACD,OAAO,gBAAgB,CAAC,WAAW,CAAC,CAAA;AACtC,CAAC;AAsCD,SAAS,sBAAsB,CAC7B,WAAsD;IAEtD,OAAO,CACJ,WAAuC,CAAC,cAAc,KAAK,SAAS;QACpE,WAAuC,CAAC,gBAAgB,KAAK,SAAS,CACxE,CAAA;AACH,CAAC"}
|
package/package.json
ADDED
|
@@ -0,0 +1,52 @@
|
|
|
1
|
+
{
|
|
2
|
+
"name": "gff-nostream",
|
|
3
|
+
"version": "1.3.1",
|
|
4
|
+
"description": "utilities to read GFF3 data",
|
|
5
|
+
"license": "MIT",
|
|
6
|
+
"repository": "cmdcolin/gff-nostream",
|
|
7
|
+
"main": "dist/index.js",
|
|
8
|
+
"module": "esm/index.js",
|
|
9
|
+
"author": {
|
|
10
|
+
"name": "Colin Diesh",
|
|
11
|
+
"email": "colin.diesh@gmail.com",
|
|
12
|
+
"url": "https://github.com/cmdcolin"
|
|
13
|
+
},
|
|
14
|
+
"files": [
|
|
15
|
+
"dist",
|
|
16
|
+
"esm",
|
|
17
|
+
"src"
|
|
18
|
+
],
|
|
19
|
+
"scripts": {
|
|
20
|
+
"test": "vitest",
|
|
21
|
+
"clean": "rimraf dist esm",
|
|
22
|
+
"lint": "eslint src test",
|
|
23
|
+
"docs": "documentation readme src/api.ts --section=API --shallow && npm run format",
|
|
24
|
+
"format": "prettier --write .",
|
|
25
|
+
"prebuild": "npm run clean",
|
|
26
|
+
"build:esm": "tsc --target es2018 --outDir esm",
|
|
27
|
+
"build:es5": "tsc --target es2015 --module commonjs --outDir dist",
|
|
28
|
+
"build": "npm run build:es5 && npm run build:esm",
|
|
29
|
+
"prepublishOnly": "npm run test run && npm run build",
|
|
30
|
+
"postversion": "git push --follow-tags"
|
|
31
|
+
},
|
|
32
|
+
"keywords": [
|
|
33
|
+
"bionode",
|
|
34
|
+
"biojs",
|
|
35
|
+
"gff3",
|
|
36
|
+
"gff",
|
|
37
|
+
"genomics"
|
|
38
|
+
],
|
|
39
|
+
"devDependencies": {
|
|
40
|
+
"@types/node": "^20.14.7",
|
|
41
|
+
"@typescript-eslint/eslint-plugin": "^8.4.0",
|
|
42
|
+
"@typescript-eslint/parser": "^8.4.0",
|
|
43
|
+
"@vitest/coverage-v8": "^2.0.5",
|
|
44
|
+
"documentation": "^14.0.1",
|
|
45
|
+
"eslint": "^9.9.1",
|
|
46
|
+
"prettier": "^3.3.3",
|
|
47
|
+
"rimraf": "^6.0.1",
|
|
48
|
+
"typescript": "^5.5.4",
|
|
49
|
+
"typescript-eslint": "^8.4.0",
|
|
50
|
+
"vitest": "^2.0.5"
|
|
51
|
+
}
|
|
52
|
+
}
|
package/src/api.ts
ADDED
|
@@ -0,0 +1,86 @@
|
|
|
1
|
+
import Parser from './parse'
|
|
2
|
+
import { GFF3Item } from './util'
|
|
3
|
+
|
|
4
|
+
/** Parser options */
|
|
5
|
+
export interface ParseOptions {
|
|
6
|
+
/** Whether to resolve references to derives from features */
|
|
7
|
+
disableDerivesFromReferences?: boolean
|
|
8
|
+
/** Text encoding of the input GFF3. default 'utf8' */
|
|
9
|
+
encoding?: BufferEncoding
|
|
10
|
+
/** Whether to parse features, default true */
|
|
11
|
+
parseFeatures?: boolean
|
|
12
|
+
/** Whether to parse directives, default false */
|
|
13
|
+
parseDirectives?: boolean
|
|
14
|
+
/** Whether to parse comments, default false */
|
|
15
|
+
parseComments?: boolean
|
|
16
|
+
/** Whether to parse sequences, default true */
|
|
17
|
+
parseSequences?: boolean
|
|
18
|
+
/**
|
|
19
|
+
* Parse all features, directives, comments, and sequences. Overrides other
|
|
20
|
+
* parsing options. Default false.
|
|
21
|
+
*/
|
|
22
|
+
parseAll?: boolean
|
|
23
|
+
}
|
|
24
|
+
|
|
25
|
+
type ParseOptionsProcessed = Required<Omit<ParseOptions, 'parseAll'>>
|
|
26
|
+
|
|
27
|
+
// shared arg processing for the parse routines
|
|
28
|
+
function _processParseOptions(options: ParseOptions): ParseOptionsProcessed {
|
|
29
|
+
const out = {
|
|
30
|
+
encoding: 'utf8' as const,
|
|
31
|
+
parseFeatures: true,
|
|
32
|
+
parseDirectives: false,
|
|
33
|
+
parseSequences: true,
|
|
34
|
+
parseComments: false,
|
|
35
|
+
disableDerivesFromReferences: false,
|
|
36
|
+
...options,
|
|
37
|
+
}
|
|
38
|
+
|
|
39
|
+
if (options.parseAll) {
|
|
40
|
+
out.parseFeatures = true
|
|
41
|
+
out.parseDirectives = true
|
|
42
|
+
out.parseComments = true
|
|
43
|
+
out.parseSequences = true
|
|
44
|
+
}
|
|
45
|
+
|
|
46
|
+
return out
|
|
47
|
+
}
|
|
48
|
+
|
|
49
|
+
export function parseStringSync(
|
|
50
|
+
str: string,
|
|
51
|
+
inputOptions: ParseOptions = {},
|
|
52
|
+
): GFF3Item[] {
|
|
53
|
+
if (!str) {
|
|
54
|
+
return []
|
|
55
|
+
}
|
|
56
|
+
|
|
57
|
+
const options = _processParseOptions(inputOptions)
|
|
58
|
+
const items: GFF3Item[] = []
|
|
59
|
+
const push = items.push.bind(items)
|
|
60
|
+
|
|
61
|
+
const parser = new Parser({
|
|
62
|
+
featureCallback: options.parseFeatures ? push : undefined,
|
|
63
|
+
directiveCallback: options.parseDirectives ? push : undefined,
|
|
64
|
+
commentCallback: options.parseComments ? push : undefined,
|
|
65
|
+
sequenceCallback: options.parseSequences ? push : undefined,
|
|
66
|
+
disableDerivesFromReferences: options.disableDerivesFromReferences || false,
|
|
67
|
+
errorCallback: err => {
|
|
68
|
+
throw err
|
|
69
|
+
},
|
|
70
|
+
})
|
|
71
|
+
|
|
72
|
+
str.split(/\r?\n/).forEach(parser.addLine.bind(parser))
|
|
73
|
+
parser.finish()
|
|
74
|
+
|
|
75
|
+
return items
|
|
76
|
+
}
|
|
77
|
+
|
|
78
|
+
export {
|
|
79
|
+
type GFF3FeatureLine,
|
|
80
|
+
type GFF3Comment,
|
|
81
|
+
type GFF3FeatureLineWithRefs,
|
|
82
|
+
type GFF3Directive,
|
|
83
|
+
type GFF3Sequence,
|
|
84
|
+
type GFF3Feature,
|
|
85
|
+
type GFF3Item,
|
|
86
|
+
} from './util'
|