mol_charset_ucf_lib 0.0.3 → 0.0.5
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/README.md +9 -3
- package/node.d.ts +1 -0
- package/node.d.ts.map +1 -1
- package/node.deps.json +1 -1
- package/node.js +53 -58
- package/node.js.map +1 -1
- package/node.mjs +53 -58
- package/node.test.js +63 -66
- package/node.test.js.map +1 -1
- package/package.json +1 -1
- package/web.d.ts +1 -0
- package/web.d.ts.map +1 -1
- package/web.deps.json +1 -1
- package/web.js +53 -58
- package/web.js.map +1 -1
- package/web.mjs +53 -58
- package/web.test.js +10 -8
- package/web.test.js.map +1 -1
package/README.md
CHANGED
|
@@ -2,7 +2,7 @@
|
|
|
2
2
|
|
|
3
3
|
> [UCF](https://page.hyoo.ru/#!=1woher_nlk5d3) - Simple compact unicode text binarization format.
|
|
4
4
|
|
|
5
|
-
|
|
6
6
|
|
|
7
7
|
# Usage
|
|
8
8
|
|
|
@@ -24,8 +24,14 @@ import {
|
|
|
24
24
|
|
|
25
25
|
# Benchmarks
|
|
26
26
|
|
|
27
|
+
System: Chrome 143, Win 10, i7-6600U 2.60GHz
|
|
28
|
+
|
|
27
29
|
## Encoding perf
|
|
28
|
-
|
|
30
|
+
[Online](https://perf.js.hyoo.ru/#!bench=yl2hcs_rfamzi)
|
|
31
|
+
|
|
32
|
+
|
|
29
33
|
|
|
30
34
|
## Decoding perf
|
|
31
|
-
|
|
35
|
+
[Online](https://perf.js.hyoo.ru/#!bench=hlj49x_hdq7va)
|
|
36
|
+
|
|
37
|
+
|
package/node.d.ts
CHANGED
|
@@ -20,6 +20,7 @@ declare namespace $ {
|
|
|
20
20
|
|
|
21
21
|
declare namespace $ {
|
|
22
22
|
function $mol_charset_ucf_encode(str: string): Uint8Array<ArrayBuffer>;
|
|
23
|
+
function $mol_charset_ucf_encode_to(str: string, buf: Uint8Array<ArrayBuffer>, from?: number): number;
|
|
23
24
|
}
|
|
24
25
|
|
|
25
26
|
declare namespace $ {
|
package/node.d.ts.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"sources":["../../../../../mam.d.ts","../../../buffer/buffer.d.ts","../../encode/encode.d.ts","../../decode/decode.d.ts"],"names":[],"mappings":"AAAA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;ACfA;AACA;AACA;AACA;ACHA;AACA;AACA;AACA;
|
|
1
|
+
{"version":3,"sources":["../../../../../mam.d.ts","../../../buffer/buffer.d.ts","../../encode/encode.d.ts","../../decode/decode.d.ts"],"names":[],"mappings":"AAAA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;ACfA;AACA;AACA;AACA;ACHA;AACA;AACA;AACA;AACA;ACJA;AACA;AACA;AACA","sourcesContent":[null,null,null,null]}
|
package/node.deps.json
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"files":["mam.ts","LICENSE","README.md","mam.jam.js","tsfmt.json","package.json","tsconfig.json","lang.lang.tree","meta.lang.tree","sandbox.config.json","mol/CNAME","mol/LICENSE","mol/readme.md","mol/mol.meta.tree","mol/CONTRIBUTING.md","mol/CODE_OF_CONDUCT.md","mol/charset/ucf/readme.md","mol/charset/buffer/buffer.ts","mol/charset/ucf/encode/encode.ts","mol/charset/ucf/decode/decode.ts","mol/charset/ucf/lib/lib.meta.tree"],"mods":{},"deps_in":{"mol/charset/ucf":{"mol/charset/ucf/lib":-999,"mol/charset/ucf/encode":-999,"mol/charset/ucf/decode":-999},"mol/charset":{"mol/charset/ucf":-999,"mol/charset/buffer":-999},"mol":{"mol/charset":-999},"":{"mol":-999},"mol/charset/ucf/encode":{"mol/charset/ucf/lib":-999},"mol/charset/buffer":{"mol/charset/ucf/encode":-2},"mol/charset/ucf/decode":{"mol/charset/ucf/lib":-999}},"deps_out":{"mol/charset/ucf/lib":{"mol/charset/ucf":-999,"mol/charset/ucf/encode":-999,"mol/charset/ucf/decode":-999},"mol/charset/ucf":{"mol/charset":-999},"mol/charset":{"mol":-999},"mol":{"":-999},"mol/charset/ucf/encode":{"mol/charset/ucf":-999,"mol/charset/buffer":-2},"mol/charset/buffer":{"mol/charset":-999},"mol/charset/ucf/decode":{"mol/charset/ucf":-999}},"sloc":{"ts":
|
|
1
|
+
{"files":["mam.ts","LICENSE","README.md","mam.jam.js","tsfmt.json","package.json","tsconfig.json","lang.lang.tree","meta.lang.tree","sandbox.config.json","mol/CNAME","mol/LICENSE","mol/readme.md","mol/mol.meta.tree","mol/CONTRIBUTING.md","mol/CODE_OF_CONDUCT.md","mol/charset/ucf/readme.md","mol/charset/buffer/buffer.ts","mol/charset/ucf/encode/encode.ts","mol/charset/ucf/decode/decode.ts","mol/charset/ucf/lib/lib.meta.tree"],"mods":{},"deps_in":{"mol/charset/ucf":{"mol/charset/ucf/lib":-999,"mol/charset/ucf/encode":-999,"mol/charset/ucf/decode":-999},"mol/charset":{"mol/charset/ucf":-999,"mol/charset/buffer":-999},"mol":{"mol/charset":-999},"":{"mol":-999},"mol/charset/ucf/encode":{"mol/charset/ucf/lib":-999},"mol/charset/buffer":{"mol/charset/ucf/encode":-2},"mol/charset/ucf/decode":{"mol/charset/ucf/lib":-999}},"deps_out":{"mol/charset/ucf/lib":{"mol/charset/ucf":-999,"mol/charset/ucf/encode":-999,"mol/charset/ucf/decode":-999},"mol/charset/ucf":{"mol/charset":-999},"mol/charset":{"mol":-999},"mol":{"":-999},"mol/charset/ucf/encode":{"mol/charset/ucf":-999,"mol/charset/buffer":-2},"mol/charset/buffer":{"mol/charset":-999},"mol/charset/ucf/decode":{"mol/charset/ucf":-999}},"sloc":{"ts":92,"LICENSE":113,"md":489,"js":9,"json":82,"tree":43,"CNAME":1},"deps":{"mol/charset/ucf/lib":{"..":-999,"/mol/charset/ucf/encode":-999,"/mol/charset/ucf/decode":-999},"mol/charset/ucf":{"..":-999},"mol/charset":{"..":-999},"mol":{"..":-999},"":{},"mol/charset/ucf/encode":{"..":-999,"/mol/charset/ucf/encode":-1,"/mol/charset/buffer":-2,"/mol/charset/ucf/encode/to":-1},"mol/charset/buffer":{"..":-999,"/mol/charset/buffer":-1},"mol/charset/ucf/decode":{"..":-999,"/mol/charset/ucf/decode":-1}}}
|
package/node.js
CHANGED
|
@@ -45,96 +45,91 @@ var $;
|
|
|
45
45
|
"use strict";
|
|
46
46
|
var $;
|
|
47
47
|
(function ($) {
|
|
48
|
-
let buf = new Uint8Array(2 ** 12);
|
|
49
48
|
function $mol_charset_ucf_encode(str) {
|
|
50
49
|
const buf = $mol_charset_buffer(str.length * 3);
|
|
51
50
|
return buf.slice(0, $mol_charset_ucf_encode_to(str, buf));
|
|
52
51
|
}
|
|
53
52
|
$.$mol_charset_ucf_encode = $mol_charset_ucf_encode;
|
|
53
|
+
const fast_char = `0123456789.,:;()?!-'" \n`;
|
|
54
|
+
const fast_map = new Array(0x80).fill(0);
|
|
55
|
+
for (let i = 0; i < fast_char.length; ++i)
|
|
56
|
+
fast_map[fast_char[i].charCodeAt(0)] = i | 0x80;
|
|
54
57
|
function $mol_charset_ucf_encode_to(str, buf, from = 0) {
|
|
55
58
|
let pos = from;
|
|
56
|
-
let mode =
|
|
59
|
+
let mode = 0x9C;
|
|
57
60
|
for (let i = 0; i < str.length; i++) {
|
|
58
61
|
let code = str.charCodeAt(i);
|
|
59
62
|
if (code >= 0xd800 && code < 0xe000)
|
|
60
63
|
code = ((code - 0xd800) << 10) + str.charCodeAt(++i) + 0x2400;
|
|
61
|
-
if (code <
|
|
62
|
-
if (
|
|
63
|
-
|
|
64
|
-
|
|
65
|
-
|
|
66
|
-
|
|
67
|
-
buf[pos++] = mode =
|
|
68
|
-
buf[pos++] = code | 0x80;
|
|
64
|
+
if (code < 0x80) {
|
|
65
|
+
if (mode !== 0x9C) {
|
|
66
|
+
const fast = fast_map[code];
|
|
67
|
+
if (fast)
|
|
68
|
+
code = fast;
|
|
69
|
+
else
|
|
70
|
+
buf[pos++] = mode = 0x9C;
|
|
69
71
|
}
|
|
72
|
+
buf[pos++] = code;
|
|
73
|
+
}
|
|
74
|
+
else if (code < 0x32_00) {
|
|
75
|
+
const page = (code >> 7) + 0x9C;
|
|
76
|
+
if (mode !== page)
|
|
77
|
+
buf[pos++] = mode = page;
|
|
78
|
+
buf[pos++] = code & 0x7F;
|
|
79
|
+
}
|
|
80
|
+
else if (code < 0x04_20_00) {
|
|
81
|
+
code -= 0x2000;
|
|
82
|
+
const page = (code >> 15) + 0x98;
|
|
83
|
+
if (mode !== page)
|
|
84
|
+
buf[pos++] = mode = page;
|
|
85
|
+
buf[pos++] = code & 0x7F;
|
|
86
|
+
buf[pos++] = code >> 7;
|
|
70
87
|
}
|
|
71
88
|
else {
|
|
72
|
-
if (
|
|
73
|
-
buf[pos++] =
|
|
74
|
-
|
|
75
|
-
|
|
76
|
-
|
|
77
|
-
if (mode !== page)
|
|
78
|
-
buf[pos++] = mode = page;
|
|
79
|
-
buf[pos++] = (code & 0x7F) | 0x80;
|
|
80
|
-
}
|
|
81
|
-
else if (code < 0x04_00_00) {
|
|
82
|
-
const page = code >> 15;
|
|
83
|
-
if (mode !== page)
|
|
84
|
-
buf[pos++] = mode = page;
|
|
85
|
-
buf[pos++] = (code & 0x7F) | 0x80;
|
|
86
|
-
buf[pos++] = code >> 7;
|
|
87
|
-
}
|
|
88
|
-
else {
|
|
89
|
-
if (mode !== 0x08)
|
|
90
|
-
buf[pos++] = mode = 0x08;
|
|
91
|
-
buf[pos++] = (code & 0x7F) | 0x80;
|
|
92
|
-
buf[pos++] = code >> 7;
|
|
93
|
-
buf[pos++] = code >> 15;
|
|
94
|
-
}
|
|
89
|
+
if (mode !== 0x97)
|
|
90
|
+
buf[pos++] = mode = 0x97;
|
|
91
|
+
buf[pos++] = code & 0x7F;
|
|
92
|
+
buf[pos++] = code >> 7;
|
|
93
|
+
buf[pos++] = code >> 15;
|
|
95
94
|
}
|
|
96
95
|
}
|
|
97
96
|
return pos - from;
|
|
98
97
|
}
|
|
98
|
+
$.$mol_charset_ucf_encode_to = $mol_charset_ucf_encode_to;
|
|
99
99
|
})($ || ($ = {}));
|
|
100
100
|
|
|
101
101
|
;
|
|
102
102
|
"use strict";
|
|
103
103
|
var $;
|
|
104
104
|
(function ($) {
|
|
105
|
-
|
|
105
|
+
const fast_char = `0123456789.,:;()?!-'" \n`;
|
|
106
|
+
function $mol_charset_ucf_decode(buffer, mode = 0x9C) {
|
|
106
107
|
let text = '';
|
|
107
108
|
let pos = 0;
|
|
108
109
|
let page_offset = 0;
|
|
109
110
|
while (pos < buffer.length) {
|
|
110
111
|
let code = buffer[pos++];
|
|
111
|
-
if (code <
|
|
112
|
-
if (
|
|
113
|
-
|
|
114
|
-
|
|
115
|
-
|
|
116
|
-
|
|
117
|
-
text += String.fromCodePoint(code);
|
|
118
|
-
}
|
|
119
|
-
else if (code === 0x08) {
|
|
120
|
-
mode = 0x08;
|
|
121
|
-
page_offset = 0;
|
|
122
|
-
}
|
|
123
|
-
else {
|
|
124
|
-
mode = code;
|
|
125
|
-
page_offset = mode << 15;
|
|
126
|
-
}
|
|
112
|
+
if (code < 0x80) {
|
|
113
|
+
if (mode < 0x9C)
|
|
114
|
+
code |= buffer[pos++] << 7;
|
|
115
|
+
if (mode === 0x97)
|
|
116
|
+
code |= buffer[pos++] << 15;
|
|
117
|
+
text += String.fromCodePoint(page_offset + code);
|
|
127
118
|
}
|
|
128
|
-
else if (code <
|
|
129
|
-
text +=
|
|
119
|
+
else if (code < 0x97) {
|
|
120
|
+
text += fast_char[code - 0x80];
|
|
121
|
+
}
|
|
122
|
+
else if (code >= 0x9C) {
|
|
123
|
+
mode = code;
|
|
124
|
+
page_offset = (mode - 0x9C) << 7;
|
|
125
|
+
}
|
|
126
|
+
else if (code === 0x97) {
|
|
127
|
+
mode = code;
|
|
128
|
+
page_offset = 0;
|
|
130
129
|
}
|
|
131
130
|
else {
|
|
132
|
-
|
|
133
|
-
|
|
134
|
-
code |= buffer[pos++] << 7;
|
|
135
|
-
if (mode === 0x08)
|
|
136
|
-
code |= buffer[pos++] << 15;
|
|
137
|
-
text += String.fromCodePoint(page_offset | code);
|
|
131
|
+
mode = code;
|
|
132
|
+
page_offset = ((mode - 0x98) << 15) + 0x20_00;
|
|
138
133
|
}
|
|
139
134
|
}
|
|
140
135
|
return text;
|
package/node.js.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"sources":["-","../../../../../mam.ts","../../../../../mol/charset/buffer/buffer.ts","../../../../../mol/charset/ucf/encode/encode.ts","../../../../../mol/charset/ucf/decode/decode.ts"],"names":[],"mappings":";;;AAAA;AACA;AACA;AACA;;ACHA,KAAK,CAAC,eAAe,GAAG,EAAE,CAAC;AAK3B,IAAU,CAAC,CAQV;AARD,WAAU,CAAC;AAQX,CAAC,EARS,CAAC,KAAD,CAAC,QAQV;AAED,MAAM,CAAC,OAAO,GAAG,CAAC,CAAA;;;ADflB;AACA;AACA;;;;;;;;;;;;;;;AEFA,IAAU,CAAC,CAUV;AAVD,WAAU,CAAC;IAEV,IAAI,GAAG,GAAG,IAAI,UAAU,CAAE,CAAC,IAAE,EAAE,CAAE,CAAA;IAGjC,SAAgB,mBAAmB,CAAE,IAAY;QAChD,IAAI,GAAG,CAAC,UAAU,GAAG,IAAI;YAAG,GAAG,GAAG,IAAI,UAAU,CAAE,IAAI,CAAE,CAAA;QACxD,OAAO,GAAG,CAAA;IACX,CAAC;IAHe,qBAAmB,sBAGlC,CAAA;AAEF,CAAC,EAVS,CAAC,KAAD,CAAC,QAUV;;;;ACVD,IAAU,CAAC,
|
|
1
|
+
{"version":3,"sources":["-","../../../../../mam.ts","../../../../../mol/charset/buffer/buffer.ts","../../../../../mol/charset/ucf/encode/encode.ts","../../../../../mol/charset/ucf/decode/decode.ts"],"names":[],"mappings":";;;AAAA;AACA;AACA;AACA;;ACHA,KAAK,CAAC,eAAe,GAAG,EAAE,CAAC;AAK3B,IAAU,CAAC,CAQV;AARD,WAAU,CAAC;AAQX,CAAC,EARS,CAAC,KAAD,CAAC,QAQV;AAED,MAAM,CAAC,OAAO,GAAG,CAAC,CAAA;;;ADflB;AACA;AACA;;;;;;;;;;;;;;;AEFA,IAAU,CAAC,CAUV;AAVD,WAAU,CAAC;IAEV,IAAI,GAAG,GAAG,IAAI,UAAU,CAAE,CAAC,IAAE,EAAE,CAAE,CAAA;IAGjC,SAAgB,mBAAmB,CAAE,IAAY;QAChD,IAAI,GAAG,CAAC,UAAU,GAAG,IAAI;YAAG,GAAG,GAAG,IAAI,UAAU,CAAE,IAAI,CAAE,CAAA;QACxD,OAAO,GAAG,CAAA;IACX,CAAC;IAHe,qBAAmB,sBAGlC,CAAA;AAEF,CAAC,EAVS,CAAC,KAAD,CAAC,QAUV;;;;ACVD,IAAU,CAAC,CA2DV;AA3DD,WAAU,CAAC;IAGV,SAAgB,uBAAuB,CAAE,GAAW;QACnD,MAAM,GAAG,GAAG,mBAAmB,CAAE,GAAG,CAAC,MAAM,GAAG,CAAC,CAAE,CAAA;QACjD,OAAO,GAAG,CAAC,KAAK,CAAE,CAAC,EAAE,0BAA0B,CAAE,GAAG,EAAE,GAAG,CAAE,CAAE,CAAA;IAC9D,CAAC;IAHe,yBAAuB,0BAGtC,CAAA;IAED,MAAM,SAAS,GAAG,0BAA0B,CAAA;IAC5C,MAAM,QAAQ,GAAG,IAAI,KAAK,CAAE,IAAI,CAAE,CAAC,IAAI,CAAE,CAAC,CAAE,CAAA;IAC5C,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,SAAS,CAAC,MAAM,EAAE,EAAE,CAAC;QAAG,QAAQ,CAAE,SAAS,CAAE,CAAC,CAAE,CAAC,UAAU,CAAC,CAAC,CAAC,CAAE,GAAG,CAAC,GAAG,IAAI,CAAA;IAE/F,SAAgB,0BAA0B,CAAE,GAAW,EAAE,GAA8B,EAAE,IAAI,GAAG,CAAC;QAEhG,IAAI,GAAG,GAAG,IAAI,CAAA;QACd,IAAI,IAAI,GAAG,IAAI,CAAA;QAEf,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,GAAG,CAAC,MAAM,EAAE,CAAC,EAAE,EAAG,CAAC;YAEtC,IAAI,IAAI,GAAG,GAAG,CAAC,UAAU,CAAE,CAAC,CAAE,CAAA;YAC9B,IAAI,IAAI,IAAI,MAAM,IAAI,IAAI,GAAG,MAAM;gBAAG,IAAI,GAAG,CAAE,CAAE,IAAI,GAAG,MAAM,CAAE,IAAI,EAAE,CAAE,GAAG,GAAG,CAAC,UAAU,CAAE,EAAG,CAAC,CAAE,GAAG,MAAM,CAAA;YAE1G,IAAI,IAAI,GAAG,IAAI,EAAG,CAAC;gBAElB,IAAI,IAAI,KAAK,IAAI,EAAG,CAAC;oBACpB,MAAM,IAAI,GAAG,QAAQ,CAAE,IAAI,CAAE,CAAA;oBAC7B,IAAI,IAAI;wBAAG,IAAI,GAAG,IAAI,CAAA;;wBACjB,GAAG,CAAE,GAAG,EAAG,CAAE,GAAG,IAAI,GAAG,IAAI,CAAA;gBACjC,CAAC;gBACD,GAAG,CAAE,GAAG,EAAG,CAAE,GAAG,IAAI,CAAA;YAErB,CAAC;iBAAM,IAAI,IAAI,GAAG,OAAO,EAAG,CAAC;gBAE5B,MAAM,IAAI,GAAG,CAAE,IAAI,IAAI,CAAC,CAAE,GAAG,IAAI,CAAA;gBACjC,IAAI,IAAI,KAAK,IAAI;oBAAG,GAAG,CAAE,GAAG,EAAG,CAAE,GAAG,IAAI,GAAG,IAAI,CAAA;gBAC/C,GAAG,CAAE,GAAG,EAAG,CAAE,GAAG,IAAI,GAAG,IAAI,CAAA;YAE5B,CAAC;iBAAM,IAAI,IAAI,GAAG,UAAU,EAAG,CAAC;gBAE/B,IAAI,IAAI,MAAM,CAAA;gBACd,MAAM,IAAI,GAAG,CAAE,IAAI,IAAI,EAAE,CAAE,GAAG,IAAI,CAAA;gBAClC,IAAI,IAAI,KAAK,IAAI;oBAAG,GAAG,CAAE,GAAG,EAAG,CAAE,GAAG,IAAI,GAAG,IAAI,CAAA;gBAC/C,GAAG,CAAE,GAAG,EAAG,CAAE,GAAG,IAAI,GAAG,IAAI,CAAA;gBAC3B,GAAG,CAAE,GAAG,EAAG,CAAE,GAAG,IAAI,IAAI,CAAC,CAAA;YAE1B,CAAC;iBAAM,CAAC;gBAEP,IAAI,IAAI,KAAK,IAAI;oBAAG,GAAG,CAAE,GAAG,EAAG,CAAE,GAAG,IAAI,GAAG,IAAI,CAAA;gBAC/C,GAAG,CAAE,GAAG,EAAG,CAAE,GAAG,IAAI,GAAG,IAAI,CAAA;gBAC3B,GAAG,CAAE,GAAG,EAAG,CAAE,GAAG,IAAI,IAAI,CAAC,CAAA;gBACzB,GAAG,CAAE,GAAG,EAAG,CAAE,GAAG,IAAI,IAAI,EAAE,CAAA;YAE3B,CAAC;QAEF,CAAC;QAED,OAAO,GAAG,GAAG,IAAI,CAAA;IAClB,CAAC;IA7Ce,4BAA0B,6BA6CzC,CAAA;AAEF,CAAC,EA3DS,CAAC,KAAD,CAAC,QA2DV;;;;AC3DD,IAAU,CAAC,CA+CV;AA/CD,WAAU,CAAC;IAEV,MAAM,SAAS,GAAG,0BAA0B,CAAA;IAG5C,SAAgB,uBAAuB,CAAE,MAAiC,EAAE,IAAI,GAAG,IAAI;QAEtF,IAAI,IAAI,GAAG,EAAE,CAAA;QACb,IAAI,GAAG,GAAG,CAAC,CAAA;QACX,IAAI,WAAW,GAAG,CAAC,CAAA;QAEnB,OAAO,GAAG,GAAG,MAAM,CAAC,MAAM,EAAG,CAAC;YAE7B,IAAI,IAAI,GAAG,MAAM,CAAE,GAAG,EAAG,CAAE,CAAA;YAE3B,IAAI,IAAI,GAAG,IAAI,EAAG,CAAC;gBAElB,IAAI,IAAI,GAAG,IAAI;oBAAG,IAAI,IAAI,MAAM,CAAE,GAAG,EAAG,CAAE,IAAI,CAAC,CAAA;gBAC/C,IAAI,IAAI,KAAK,IAAI;oBAAG,IAAI,IAAI,MAAM,CAAE,GAAG,EAAG,CAAE,IAAI,EAAE,CAAA;gBAClD,IAAI,IAAI,MAAM,CAAC,aAAa,CAAE,WAAW,GAAG,IAAI,CAAE,CAAA;YAEnD,CAAC;iBAAM,IAAI,IAAI,GAAG,IAAI,EAAG,CAAC;gBAEzB,IAAI,IAAI,SAAS,CAAE,IAAI,GAAG,IAAI,CAAE,CAAA;YAEjC,CAAC;iBAAM,IAAI,IAAI,IAAI,IAAI,EAAG,CAAC;gBAE1B,IAAI,GAAG,IAAI,CAAA;gBACX,WAAW,GAAG,CAAE,IAAI,GAAG,IAAI,CAAE,IAAI,CAAC,CAAA;YAEnC,CAAC;iBAAM,IAAI,IAAI,KAAK,IAAI,EAAG,CAAC;gBAE3B,IAAI,GAAG,IAAI,CAAA;gBACX,WAAW,GAAG,CAAC,CAAA;YAEhB,CAAC;iBAAM,CAAC;gBAEP,IAAI,GAAG,IAAI,CAAA;gBACX,WAAW,GAAG,CAAE,CAAE,IAAI,GAAG,IAAI,CAAE,IAAI,EAAE,CAAE,GAAG,OAAO,CAAA;YAElD,CAAC;QAEF,CAAC;QAED,OAAO,IAAI,CAAA;IACZ,CAAC;IAxCe,yBAAuB,0BAwCtC,CAAA;AAEF,CAAC,EA/CS,CAAC,KAAD,CAAC,QA+CV;;","sourcesContent":[null,"Error.stackTraceLimit = 50;\n\ndeclare let _$_: { new(): {} } & typeof globalThis\ndeclare class $ extends _$_ {}\n\nnamespace $ {\n\texport type $ = typeof $$\n\texport declare class $$ extends $ {\n\t\tstatic $: $\n\t}\n\tnamespace $$ {\n\t\texport type $$ = $\n\t}\n}\n\nmodule.exports = $\n","namespace $ {\n\t\n\tlet buf = new Uint8Array( 2**12 ) // 4KB Mem Page\n\t\n\t/** Temporary buffer. Recursive usage isn't supported. */\n\texport function $mol_charset_buffer( size: number ) {\n\t\tif( buf.byteLength < size ) buf = new Uint8Array( size )\n\t\treturn buf\n\t}\n\t\n}\n","namespace $ {\n\t\n\t/** Encode text to Unicode Compact Format. */\n\texport function $mol_charset_ucf_encode( str: string ) {\n\t\tconst buf = $mol_charset_buffer( str.length * 3 )\n\t\treturn buf.slice( 0, $mol_charset_ucf_encode_to( str, buf ) )\n\t}\n\t\n\tconst fast_char = `0123456789.,:;()?!-'\" \\n`\n\tconst fast_map = new Array( 0x80 ).fill( 0 )\n\tfor( let i = 0; i < fast_char.length; ++i ) fast_map[ fast_char[ i ].charCodeAt(0) ] = i | 0x80\n\n\texport function $mol_charset_ucf_encode_to( str: string, buf: Uint8Array< ArrayBuffer >, from = 0 ) {\n\n\t\tlet pos = from\n\t\tlet mode = 0x9C\n\n\t\tfor( let i = 0; i < str.length; i++ ) {\n\t\t\t\n\t\t\tlet code = str.charCodeAt( i )\n\t\t\tif( code >= 0xd800 && code < 0xe000 ) code = ( ( code - 0xd800 ) << 10 ) + str.charCodeAt( ++ i ) + 0x2400\n\t\t\t\n\t\t\tif( code < 0x80 ) { // ASCII\n\t\t\t\t\n\t\t\t\tif( mode !== 0x9C ) {\n\t\t\t\t\tconst fast = fast_map[ code ]\n\t\t\t\t\tif( fast ) code = fast\n\t\t\t\t\telse buf[ pos ++ ] = mode = 0x9C\n\t\t\t\t}\n\t\t\t\tbuf[ pos ++ ] = code\n\t\t\t\t\n\t\t\t} else if( code < 0x32_00 ) { // Tiny\n\t\t\t\t\n\t\t\t\tconst page = ( code >> 7 ) + 0x9C\n\t\t\t\tif( mode !== page ) buf[ pos ++ ] = mode = page\n\t\t\t\tbuf[ pos ++ ] = code & 0x7F\n\t\t\t\t\n\t\t\t} else if( code < 0x04_20_00 ) { // Wide\n\t\t\t\t\n\t\t\t\tcode -= 0x2000\n\t\t\t\tconst page = ( code >> 15 ) + 0x98\n\t\t\t\tif( mode !== page ) buf[ pos ++ ] = mode = page\n\t\t\t\tbuf[ pos ++ ] = code & 0x7F\n\t\t\t\tbuf[ pos ++ ] = code >> 7\n\t\t\t\t\n\t\t\t} else { // Full\n\t\t\t\t\n\t\t\t\tif( mode !== 0x97 ) buf[ pos ++ ] = mode = 0x97\n\t\t\t\tbuf[ pos ++ ] = code & 0x7F\n\t\t\t\tbuf[ pos ++ ] = code >> 7\n\t\t\t\tbuf[ pos ++ ] = code >> 15\n\t\t\t\t\n\t\t\t}\n\n\t\t}\n\n\t\treturn pos - from\n\t}\n\t\n}\n","namespace $ {\n\n\tconst fast_char = `0123456789.,:;()?!-'\" \\n`\n\t\n\t/** Decode text from Unicode Compact Format. */\n\texport function $mol_charset_ucf_decode( buffer: Uint8Array< ArrayBuffer >, mode = 0x9C ) {\n\t\t\n\t\tlet text = ''\n\t\tlet pos = 0\n\t\tlet page_offset = 0\n\t\t\n\t\twhile( pos < buffer.length ) {\n\t\t\t\n\t\t\tlet code = buffer[ pos ++ ]\n\t\t\t\n\t\t\tif( code < 0x80 ) { // Char Code\n\t\t\t\t\n\t\t\t\tif( mode < 0x9C ) code |= buffer[ pos ++ ] << 7\n\t\t\t\tif( mode === 0x97 ) code |= buffer[ pos ++ ] << 15\n\t\t\t\ttext += String.fromCodePoint( page_offset + code )\n\t\t\t\t\n\t\t\t} else if( code < 0x97 ) { // ASCII\n\t\t\t\t\n\t\t\t\ttext += fast_char[ code - 0x80 ]\n\t\t\t\t\n\t\t\t} else if( code >= 0x9C ) { // Tiny Set\n\t\t\t\t\n\t\t\t\tmode = code\n\t\t\t\tpage_offset = ( mode - 0x9C ) << 7\n\t\t\t\t\n\t\t\t} else if( code === 0x97 ) { // Full Set\n\t\t\t\t\n\t\t\t\tmode = code\n\t\t\t\tpage_offset = 0\n\t\t\t\t\n\t\t\t} else { // Wide Set\n\t\t\t\t\n\t\t\t\tmode = code\n\t\t\t\tpage_offset = ( ( mode - 0x98 ) << 15 ) + 0x20_00\n\t\t\t\t\n\t\t\t}\n\t\t\t\n\t\t}\n\t\t\n\t\treturn text\n\t}\n\n}\n"]}
|
package/node.mjs
CHANGED
|
@@ -45,96 +45,91 @@ var $;
|
|
|
45
45
|
"use strict";
|
|
46
46
|
var $;
|
|
47
47
|
(function ($) {
|
|
48
|
-
let buf = new Uint8Array(2 ** 12);
|
|
49
48
|
function $mol_charset_ucf_encode(str) {
|
|
50
49
|
const buf = $mol_charset_buffer(str.length * 3);
|
|
51
50
|
return buf.slice(0, $mol_charset_ucf_encode_to(str, buf));
|
|
52
51
|
}
|
|
53
52
|
$.$mol_charset_ucf_encode = $mol_charset_ucf_encode;
|
|
53
|
+
const fast_char = `0123456789.,:;()?!-'" \n`;
|
|
54
|
+
const fast_map = new Array(0x80).fill(0);
|
|
55
|
+
for (let i = 0; i < fast_char.length; ++i)
|
|
56
|
+
fast_map[fast_char[i].charCodeAt(0)] = i | 0x80;
|
|
54
57
|
function $mol_charset_ucf_encode_to(str, buf, from = 0) {
|
|
55
58
|
let pos = from;
|
|
56
|
-
let mode =
|
|
59
|
+
let mode = 0x9C;
|
|
57
60
|
for (let i = 0; i < str.length; i++) {
|
|
58
61
|
let code = str.charCodeAt(i);
|
|
59
62
|
if (code >= 0xd800 && code < 0xe000)
|
|
60
63
|
code = ((code - 0xd800) << 10) + str.charCodeAt(++i) + 0x2400;
|
|
61
|
-
if (code <
|
|
62
|
-
if (
|
|
63
|
-
|
|
64
|
-
|
|
65
|
-
|
|
66
|
-
|
|
67
|
-
buf[pos++] = mode =
|
|
68
|
-
buf[pos++] = code | 0x80;
|
|
64
|
+
if (code < 0x80) {
|
|
65
|
+
if (mode !== 0x9C) {
|
|
66
|
+
const fast = fast_map[code];
|
|
67
|
+
if (fast)
|
|
68
|
+
code = fast;
|
|
69
|
+
else
|
|
70
|
+
buf[pos++] = mode = 0x9C;
|
|
69
71
|
}
|
|
72
|
+
buf[pos++] = code;
|
|
73
|
+
}
|
|
74
|
+
else if (code < 0x32_00) {
|
|
75
|
+
const page = (code >> 7) + 0x9C;
|
|
76
|
+
if (mode !== page)
|
|
77
|
+
buf[pos++] = mode = page;
|
|
78
|
+
buf[pos++] = code & 0x7F;
|
|
79
|
+
}
|
|
80
|
+
else if (code < 0x04_20_00) {
|
|
81
|
+
code -= 0x2000;
|
|
82
|
+
const page = (code >> 15) + 0x98;
|
|
83
|
+
if (mode !== page)
|
|
84
|
+
buf[pos++] = mode = page;
|
|
85
|
+
buf[pos++] = code & 0x7F;
|
|
86
|
+
buf[pos++] = code >> 7;
|
|
70
87
|
}
|
|
71
88
|
else {
|
|
72
|
-
if (
|
|
73
|
-
buf[pos++] =
|
|
74
|
-
|
|
75
|
-
|
|
76
|
-
|
|
77
|
-
if (mode !== page)
|
|
78
|
-
buf[pos++] = mode = page;
|
|
79
|
-
buf[pos++] = (code & 0x7F) | 0x80;
|
|
80
|
-
}
|
|
81
|
-
else if (code < 0x04_00_00) {
|
|
82
|
-
const page = code >> 15;
|
|
83
|
-
if (mode !== page)
|
|
84
|
-
buf[pos++] = mode = page;
|
|
85
|
-
buf[pos++] = (code & 0x7F) | 0x80;
|
|
86
|
-
buf[pos++] = code >> 7;
|
|
87
|
-
}
|
|
88
|
-
else {
|
|
89
|
-
if (mode !== 0x08)
|
|
90
|
-
buf[pos++] = mode = 0x08;
|
|
91
|
-
buf[pos++] = (code & 0x7F) | 0x80;
|
|
92
|
-
buf[pos++] = code >> 7;
|
|
93
|
-
buf[pos++] = code >> 15;
|
|
94
|
-
}
|
|
89
|
+
if (mode !== 0x97)
|
|
90
|
+
buf[pos++] = mode = 0x97;
|
|
91
|
+
buf[pos++] = code & 0x7F;
|
|
92
|
+
buf[pos++] = code >> 7;
|
|
93
|
+
buf[pos++] = code >> 15;
|
|
95
94
|
}
|
|
96
95
|
}
|
|
97
96
|
return pos - from;
|
|
98
97
|
}
|
|
98
|
+
$.$mol_charset_ucf_encode_to = $mol_charset_ucf_encode_to;
|
|
99
99
|
})($ || ($ = {}));
|
|
100
100
|
|
|
101
101
|
;
|
|
102
102
|
"use strict";
|
|
103
103
|
var $;
|
|
104
104
|
(function ($) {
|
|
105
|
-
|
|
105
|
+
const fast_char = `0123456789.,:;()?!-'" \n`;
|
|
106
|
+
function $mol_charset_ucf_decode(buffer, mode = 0x9C) {
|
|
106
107
|
let text = '';
|
|
107
108
|
let pos = 0;
|
|
108
109
|
let page_offset = 0;
|
|
109
110
|
while (pos < buffer.length) {
|
|
110
111
|
let code = buffer[pos++];
|
|
111
|
-
if (code <
|
|
112
|
-
if (
|
|
113
|
-
|
|
114
|
-
|
|
115
|
-
|
|
116
|
-
|
|
117
|
-
text += String.fromCodePoint(code);
|
|
118
|
-
}
|
|
119
|
-
else if (code === 0x08) {
|
|
120
|
-
mode = 0x08;
|
|
121
|
-
page_offset = 0;
|
|
122
|
-
}
|
|
123
|
-
else {
|
|
124
|
-
mode = code;
|
|
125
|
-
page_offset = mode << 15;
|
|
126
|
-
}
|
|
112
|
+
if (code < 0x80) {
|
|
113
|
+
if (mode < 0x9C)
|
|
114
|
+
code |= buffer[pos++] << 7;
|
|
115
|
+
if (mode === 0x97)
|
|
116
|
+
code |= buffer[pos++] << 15;
|
|
117
|
+
text += String.fromCodePoint(page_offset + code);
|
|
127
118
|
}
|
|
128
|
-
else if (code <
|
|
129
|
-
text +=
|
|
119
|
+
else if (code < 0x97) {
|
|
120
|
+
text += fast_char[code - 0x80];
|
|
121
|
+
}
|
|
122
|
+
else if (code >= 0x9C) {
|
|
123
|
+
mode = code;
|
|
124
|
+
page_offset = (mode - 0x9C) << 7;
|
|
125
|
+
}
|
|
126
|
+
else if (code === 0x97) {
|
|
127
|
+
mode = code;
|
|
128
|
+
page_offset = 0;
|
|
130
129
|
}
|
|
131
130
|
else {
|
|
132
|
-
|
|
133
|
-
|
|
134
|
-
code |= buffer[pos++] << 7;
|
|
135
|
-
if (mode === 0x08)
|
|
136
|
-
code |= buffer[pos++] << 15;
|
|
137
|
-
text += String.fromCodePoint(page_offset | code);
|
|
131
|
+
mode = code;
|
|
132
|
+
page_offset = ((mode - 0x98) << 15) + 0x20_00;
|
|
138
133
|
}
|
|
139
134
|
}
|
|
140
135
|
return text;
|
package/node.test.js
CHANGED
|
@@ -36,96 +36,91 @@ var $;
|
|
|
36
36
|
"use strict";
|
|
37
37
|
var $;
|
|
38
38
|
(function ($) {
|
|
39
|
-
let buf = new Uint8Array(2 ** 12);
|
|
40
39
|
function $mol_charset_ucf_encode(str) {
|
|
41
40
|
const buf = $mol_charset_buffer(str.length * 3);
|
|
42
41
|
return buf.slice(0, $mol_charset_ucf_encode_to(str, buf));
|
|
43
42
|
}
|
|
44
43
|
$.$mol_charset_ucf_encode = $mol_charset_ucf_encode;
|
|
44
|
+
const fast_char = `0123456789.,:;()?!-'" \n`;
|
|
45
|
+
const fast_map = new Array(0x80).fill(0);
|
|
46
|
+
for (let i = 0; i < fast_char.length; ++i)
|
|
47
|
+
fast_map[fast_char[i].charCodeAt(0)] = i | 0x80;
|
|
45
48
|
function $mol_charset_ucf_encode_to(str, buf, from = 0) {
|
|
46
49
|
let pos = from;
|
|
47
|
-
let mode =
|
|
50
|
+
let mode = 0x9C;
|
|
48
51
|
for (let i = 0; i < str.length; i++) {
|
|
49
52
|
let code = str.charCodeAt(i);
|
|
50
53
|
if (code >= 0xd800 && code < 0xe000)
|
|
51
54
|
code = ((code - 0xd800) << 10) + str.charCodeAt(++i) + 0x2400;
|
|
52
|
-
if (code <
|
|
53
|
-
if (
|
|
54
|
-
|
|
55
|
-
|
|
56
|
-
|
|
57
|
-
|
|
58
|
-
buf[pos++] = mode =
|
|
59
|
-
buf[pos++] = code | 0x80;
|
|
55
|
+
if (code < 0x80) {
|
|
56
|
+
if (mode !== 0x9C) {
|
|
57
|
+
const fast = fast_map[code];
|
|
58
|
+
if (fast)
|
|
59
|
+
code = fast;
|
|
60
|
+
else
|
|
61
|
+
buf[pos++] = mode = 0x9C;
|
|
60
62
|
}
|
|
63
|
+
buf[pos++] = code;
|
|
64
|
+
}
|
|
65
|
+
else if (code < 0x32_00) {
|
|
66
|
+
const page = (code >> 7) + 0x9C;
|
|
67
|
+
if (mode !== page)
|
|
68
|
+
buf[pos++] = mode = page;
|
|
69
|
+
buf[pos++] = code & 0x7F;
|
|
70
|
+
}
|
|
71
|
+
else if (code < 0x04_20_00) {
|
|
72
|
+
code -= 0x2000;
|
|
73
|
+
const page = (code >> 15) + 0x98;
|
|
74
|
+
if (mode !== page)
|
|
75
|
+
buf[pos++] = mode = page;
|
|
76
|
+
buf[pos++] = code & 0x7F;
|
|
77
|
+
buf[pos++] = code >> 7;
|
|
61
78
|
}
|
|
62
79
|
else {
|
|
63
|
-
if (
|
|
64
|
-
buf[pos++] =
|
|
65
|
-
|
|
66
|
-
|
|
67
|
-
|
|
68
|
-
if (mode !== page)
|
|
69
|
-
buf[pos++] = mode = page;
|
|
70
|
-
buf[pos++] = (code & 0x7F) | 0x80;
|
|
71
|
-
}
|
|
72
|
-
else if (code < 0x04_00_00) {
|
|
73
|
-
const page = code >> 15;
|
|
74
|
-
if (mode !== page)
|
|
75
|
-
buf[pos++] = mode = page;
|
|
76
|
-
buf[pos++] = (code & 0x7F) | 0x80;
|
|
77
|
-
buf[pos++] = code >> 7;
|
|
78
|
-
}
|
|
79
|
-
else {
|
|
80
|
-
if (mode !== 0x08)
|
|
81
|
-
buf[pos++] = mode = 0x08;
|
|
82
|
-
buf[pos++] = (code & 0x7F) | 0x80;
|
|
83
|
-
buf[pos++] = code >> 7;
|
|
84
|
-
buf[pos++] = code >> 15;
|
|
85
|
-
}
|
|
80
|
+
if (mode !== 0x97)
|
|
81
|
+
buf[pos++] = mode = 0x97;
|
|
82
|
+
buf[pos++] = code & 0x7F;
|
|
83
|
+
buf[pos++] = code >> 7;
|
|
84
|
+
buf[pos++] = code >> 15;
|
|
86
85
|
}
|
|
87
86
|
}
|
|
88
87
|
return pos - from;
|
|
89
88
|
}
|
|
89
|
+
$.$mol_charset_ucf_encode_to = $mol_charset_ucf_encode_to;
|
|
90
90
|
})($ || ($ = {}));
|
|
91
91
|
|
|
92
92
|
;
|
|
93
93
|
"use strict";
|
|
94
94
|
var $;
|
|
95
95
|
(function ($) {
|
|
96
|
-
|
|
96
|
+
const fast_char = `0123456789.,:;()?!-'" \n`;
|
|
97
|
+
function $mol_charset_ucf_decode(buffer, mode = 0x9C) {
|
|
97
98
|
let text = '';
|
|
98
99
|
let pos = 0;
|
|
99
100
|
let page_offset = 0;
|
|
100
101
|
while (pos < buffer.length) {
|
|
101
102
|
let code = buffer[pos++];
|
|
102
|
-
if (code <
|
|
103
|
-
if (
|
|
104
|
-
|
|
105
|
-
|
|
106
|
-
|
|
107
|
-
|
|
108
|
-
|
|
109
|
-
|
|
110
|
-
|
|
111
|
-
|
|
112
|
-
|
|
113
|
-
|
|
114
|
-
|
|
115
|
-
mode = code;
|
|
116
|
-
page_offset = mode << 15;
|
|
117
|
-
}
|
|
103
|
+
if (code < 0x80) {
|
|
104
|
+
if (mode < 0x9C)
|
|
105
|
+
code |= buffer[pos++] << 7;
|
|
106
|
+
if (mode === 0x97)
|
|
107
|
+
code |= buffer[pos++] << 15;
|
|
108
|
+
text += String.fromCodePoint(page_offset + code);
|
|
109
|
+
}
|
|
110
|
+
else if (code < 0x97) {
|
|
111
|
+
text += fast_char[code - 0x80];
|
|
112
|
+
}
|
|
113
|
+
else if (code >= 0x9C) {
|
|
114
|
+
mode = code;
|
|
115
|
+
page_offset = (mode - 0x9C) << 7;
|
|
118
116
|
}
|
|
119
|
-
else if (code
|
|
120
|
-
|
|
117
|
+
else if (code === 0x97) {
|
|
118
|
+
mode = code;
|
|
119
|
+
page_offset = 0;
|
|
121
120
|
}
|
|
122
121
|
else {
|
|
123
|
-
|
|
124
|
-
|
|
125
|
-
code |= buffer[pos++] << 7;
|
|
126
|
-
if (mode === 0x08)
|
|
127
|
-
code |= buffer[pos++] << 15;
|
|
128
|
-
text += String.fromCodePoint(page_offset | code);
|
|
122
|
+
mode = code;
|
|
123
|
+
page_offset = ((mode - 0x98) << 15) + 0x20_00;
|
|
129
124
|
}
|
|
130
125
|
}
|
|
131
126
|
return text;
|
|
@@ -4140,11 +4135,12 @@ var $;
|
|
|
4140
4135
|
(function ($$) {
|
|
4141
4136
|
$mol_test({
|
|
4142
4137
|
"Complex UCF encoding"($) {
|
|
4143
|
-
$mol_assert_equal($mol_charset_ucf_encode('hi мир,
|
|
4138
|
+
$mol_assert_equal($mol_charset_ucf_encode('hi мир, 美しい 世界 🏴☠\t\n'), new Uint8Array([
|
|
4144
4139
|
0x68, 0x69, 0x20,
|
|
4145
|
-
|
|
4146
|
-
|
|
4147
|
-
|
|
4140
|
+
0xA4, 0x3C, 0x38, 0x40, 0x8B, 0x95,
|
|
4141
|
+
0x98, 0x0E, 0xBF, 0xFC, 0x57, 0x44, 0x95,
|
|
4142
|
+
0x98, 0x16, 0x5C, 0x4C, 0xAA, 0x95,
|
|
4143
|
+
0x9B, 0x74, 0xA7, 0xDC, 0x0D, 0xE8, 0x20, 0x9C, 0x09, 0x0A,
|
|
4148
4144
|
]));
|
|
4149
4145
|
},
|
|
4150
4146
|
});
|
|
@@ -4159,11 +4155,12 @@ var $;
|
|
|
4159
4155
|
(function ($$) {
|
|
4160
4156
|
$mol_test({
|
|
4161
4157
|
"Complex UCF eecoding"($) {
|
|
4162
|
-
$mol_assert_equal('hi мир,
|
|
4158
|
+
$mol_assert_equal('hi мир, 美しい 世界 🏴☠\t\n', $mol_charset_ucf_decode(new Uint8Array([
|
|
4163
4159
|
0x68, 0x69, 0x20,
|
|
4164
|
-
|
|
4165
|
-
|
|
4166
|
-
|
|
4160
|
+
0xA4, 0x3C, 0x38, 0x40, 0x8B, 0x95,
|
|
4161
|
+
0x98, 0x0E, 0xBF, 0xFC, 0x57, 0x44, 0x95,
|
|
4162
|
+
0x98, 0x16, 0x5C, 0x4C, 0xAA, 0x95,
|
|
4163
|
+
0x9B, 0x74, 0xA7, 0xDC, 0x0D, 0xE8, 0x20, 0x9C, 0x09, 0x0A,
|
|
4167
4164
|
])));
|
|
4168
4165
|
},
|
|
4169
4166
|
});
|