promptopskit 0.7.1 → 0.8.0
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.
- package/README.md +39 -4
- package/dist/{chunk-OWMP5NMV.js → chunk-5XH2GROP.js} +2 -2
- package/dist/{chunk-FICB2ZSK.js → chunk-DI7Q4T7A.js} +3 -3
- package/dist/{chunk-ULVY473P.js → chunk-E4GEKWVP.js} +31 -3
- package/dist/chunk-E4GEKWVP.js.map +1 -0
- package/dist/{chunk-3G3EAV7H.js → chunk-PDVUVZAU.js} +2 -2
- package/dist/chunk-RVAFVWR2.js +1760 -0
- package/dist/chunk-RVAFVWR2.js.map +1 -0
- package/dist/{chunk-35NN5F6O.js → chunk-UQLAWU67.js} +2 -2
- package/dist/{chunk-6NOUETAC.js → chunk-Z3DPPJNU.js} +2 -2
- package/dist/{chunk-R65RAHAG.js → chunk-ZVCKP4EM.js} +3 -3
- package/dist/cli/index.js +86 -5
- package/dist/cli/index.js.map +1 -1
- package/dist/index.cjs +964 -44
- package/dist/index.cjs.map +1 -1
- package/dist/index.d.cts +45 -5
- package/dist/index.d.ts +45 -5
- package/dist/index.js +18 -10
- package/dist/index.js.map +1 -1
- package/dist/providers/anthropic.cjs +954 -42
- package/dist/providers/anthropic.cjs.map +1 -1
- package/dist/providers/anthropic.d.cts +2 -2
- package/dist/providers/anthropic.d.ts +2 -2
- package/dist/providers/anthropic.js +3 -3
- package/dist/providers/gemini.cjs +954 -42
- package/dist/providers/gemini.cjs.map +1 -1
- package/dist/providers/gemini.d.cts +2 -2
- package/dist/providers/gemini.d.ts +2 -2
- package/dist/providers/gemini.js +3 -3
- package/dist/providers/llmasaservice.cjs +954 -42
- package/dist/providers/llmasaservice.cjs.map +1 -1
- package/dist/providers/llmasaservice.d.cts +2 -2
- package/dist/providers/llmasaservice.d.ts +2 -2
- package/dist/providers/llmasaservice.js +4 -4
- package/dist/providers/openai-responses.cjs +954 -42
- package/dist/providers/openai-responses.cjs.map +1 -1
- package/dist/providers/openai-responses.d.cts +2 -2
- package/dist/providers/openai-responses.d.ts +2 -2
- package/dist/providers/openai-responses.js +3 -3
- package/dist/providers/openai.cjs +954 -42
- package/dist/providers/openai.cjs.map +1 -1
- package/dist/providers/openai.d.cts +2 -2
- package/dist/providers/openai.d.ts +2 -2
- package/dist/providers/openai.js +3 -3
- package/dist/providers/openrouter.cjs +954 -42
- package/dist/providers/openrouter.cjs.map +1 -1
- package/dist/providers/openrouter.d.cts +2 -2
- package/dist/providers/openrouter.d.ts +2 -2
- package/dist/providers/openrouter.js +4 -4
- package/dist/{schema-DBcSns_b.d.cts → schema-BgVLZ2u5.d.cts} +641 -0
- package/dist/{schema-DBcSns_b.d.ts → schema-BgVLZ2u5.d.ts} +641 -0
- package/dist/testing.cjs +27 -1
- package/dist/testing.cjs.map +1 -1
- package/dist/testing.d.cts +1 -1
- package/dist/testing.d.ts +1 -1
- package/dist/testing.js +1 -1
- package/dist/{types-BlLXT5Qb.d.ts → types-2hnZmCkH.d.cts} +16 -3
- package/dist/{types-8T2D5KB5.d.cts → types-DgAT4jSR.d.ts} +16 -3
- package/dist/usagetap/index.d.cts +2 -2
- package/dist/usagetap/index.d.ts +2 -2
- package/package.json +1 -1
- package/dist/chunk-F552OBUV.js +0 -875
- package/dist/chunk-F552OBUV.js.map +0 -1
- package/dist/chunk-ULVY473P.js.map +0 -1
- /package/dist/{chunk-OWMP5NMV.js.map → chunk-5XH2GROP.js.map} +0 -0
- /package/dist/{chunk-FICB2ZSK.js.map → chunk-DI7Q4T7A.js.map} +0 -0
- /package/dist/{chunk-3G3EAV7H.js.map → chunk-PDVUVZAU.js.map} +0 -0
- /package/dist/{chunk-35NN5F6O.js.map → chunk-UQLAWU67.js.map} +0 -0
- /package/dist/{chunk-6NOUETAC.js.map → chunk-Z3DPPJNU.js.map} +0 -0
- /package/dist/{chunk-R65RAHAG.js.map → chunk-ZVCKP4EM.js.map} +0 -0
|
@@ -35,16 +35,17 @@ __export(openai_exports, {
|
|
|
35
35
|
module.exports = __toCommonJS(openai_exports);
|
|
36
36
|
|
|
37
37
|
// src/renderer/interpolate.ts
|
|
38
|
-
var VARIABLE_RE = /\{\{\s*([a-zA-Z_][a-zA-Z0-9_]*)
|
|
38
|
+
var VARIABLE_RE = /\{\{\s*([a-zA-Z_][a-zA-Z0-9_]*)(?:\s*\|\s*(compress|toon|compact|code))?\s*\}\}/g;
|
|
39
39
|
var ESCAPED_OPEN = /\\\{\\\{/g;
|
|
40
40
|
var ESCAPE_PLACEHOLDER = "\0ESCAPED_OPEN\0";
|
|
41
41
|
function interpolate(template, variables, options = {}) {
|
|
42
42
|
const { strict = false } = options;
|
|
43
43
|
const optionalVariables = new Set(options.optionalVariables ?? []);
|
|
44
44
|
let result = template.replace(ESCAPED_OPEN, ESCAPE_PLACEHOLDER);
|
|
45
|
-
result = result.replace(VARIABLE_RE, (match, name) => {
|
|
45
|
+
result = result.replace(VARIABLE_RE, (match, name, modifier) => {
|
|
46
46
|
if (name in variables) {
|
|
47
|
-
|
|
47
|
+
const value = variables[name];
|
|
48
|
+
return options.transformVariable ? options.transformVariable({ name, value, modifier, match }) : value;
|
|
48
49
|
}
|
|
49
50
|
if (strict && !optionalVariables.has(name)) {
|
|
50
51
|
throw new Error(`Missing required variable: "${name}"`);
|
|
@@ -72,12 +73,56 @@ function normalizeContextInput(input) {
|
|
|
72
73
|
warnings: input.warnings,
|
|
73
74
|
max_size: input.max_size,
|
|
74
75
|
trim: input.trim,
|
|
76
|
+
compression: normalizeContextInputCompression(input.compression),
|
|
75
77
|
allow_regex: normalizeContextRegex(input.allow_regex),
|
|
76
78
|
deny_regex: normalizeContextRegex(input.deny_regex),
|
|
77
79
|
non_empty: normalizeBuiltInValidator(input.non_empty),
|
|
78
80
|
reject_secrets: normalizeBuiltInValidator(input.reject_secrets)
|
|
79
81
|
};
|
|
80
82
|
}
|
|
83
|
+
function normalizeContextInputCompression(value) {
|
|
84
|
+
if (value === void 0) {
|
|
85
|
+
return void 0;
|
|
86
|
+
}
|
|
87
|
+
if (value === "heuristic") {
|
|
88
|
+
return { heuristic: { enabled: true } };
|
|
89
|
+
}
|
|
90
|
+
if (value === "code") {
|
|
91
|
+
return { code: { enabled: true } };
|
|
92
|
+
}
|
|
93
|
+
if (value.heuristic === void 0) {
|
|
94
|
+
return normalizeCodeCompression(value.code);
|
|
95
|
+
}
|
|
96
|
+
const normalized = {
|
|
97
|
+
heuristic: {
|
|
98
|
+
enabled: value.heuristic.enabled ?? true,
|
|
99
|
+
min_tokens: value.heuristic.min_tokens,
|
|
100
|
+
max_sentences: value.heuristic.max_sentences,
|
|
101
|
+
target_reduction: value.heuristic.target_reduction,
|
|
102
|
+
query: value.heuristic.query,
|
|
103
|
+
query_variable: value.heuristic.query_variable,
|
|
104
|
+
json_to_toon: value.heuristic.json_to_toon
|
|
105
|
+
}
|
|
106
|
+
};
|
|
107
|
+
const code = normalizeCodeCompression(value.code);
|
|
108
|
+
if (code?.code) {
|
|
109
|
+
normalized.code = code.code;
|
|
110
|
+
}
|
|
111
|
+
return normalized;
|
|
112
|
+
}
|
|
113
|
+
function normalizeCodeCompression(value) {
|
|
114
|
+
if (value === void 0) {
|
|
115
|
+
return void 0;
|
|
116
|
+
}
|
|
117
|
+
return {
|
|
118
|
+
code: {
|
|
119
|
+
enabled: value.enabled ?? true,
|
|
120
|
+
remove_comments: value.remove_comments,
|
|
121
|
+
trim_indentation: value.trim_indentation,
|
|
122
|
+
collapse_blank_lines: value.collapse_blank_lines
|
|
123
|
+
}
|
|
124
|
+
};
|
|
125
|
+
}
|
|
81
126
|
function normalizeContextRegex(value) {
|
|
82
127
|
if (value === void 0) {
|
|
83
128
|
return void 0;
|
|
@@ -331,27 +376,703 @@ function measureContextValueSize(value) {
|
|
|
331
376
|
return textEncoder.encode(value).length;
|
|
332
377
|
}
|
|
333
378
|
|
|
379
|
+
// src/toon-encoding.ts
|
|
380
|
+
var DEFAULT_INDENT = 2;
|
|
381
|
+
var DEFAULT_DELIMITER = ",";
|
|
382
|
+
function tryJsonToToon(text, options = {}) {
|
|
383
|
+
const trimmed = text.trim();
|
|
384
|
+
if (!trimmed.startsWith("{") && !trimmed.startsWith("[")) {
|
|
385
|
+
return void 0;
|
|
386
|
+
}
|
|
387
|
+
let parsed;
|
|
388
|
+
try {
|
|
389
|
+
parsed = JSON.parse(trimmed);
|
|
390
|
+
} catch {
|
|
391
|
+
return void 0;
|
|
392
|
+
}
|
|
393
|
+
const value = normalizeJsonValue(parsed);
|
|
394
|
+
if (value === void 0) {
|
|
395
|
+
return void 0;
|
|
396
|
+
}
|
|
397
|
+
return {
|
|
398
|
+
output: encodeToon(value, options),
|
|
399
|
+
value
|
|
400
|
+
};
|
|
401
|
+
}
|
|
402
|
+
function encodeToon(value, options = {}) {
|
|
403
|
+
const resolved = {
|
|
404
|
+
indent: options.indent ?? DEFAULT_INDENT,
|
|
405
|
+
delimiter: options.delimiter ?? DEFAULT_DELIMITER
|
|
406
|
+
};
|
|
407
|
+
return encodeValue(value, void 0, 0, resolved).join("\n");
|
|
408
|
+
}
|
|
409
|
+
function normalizeJsonValue(value) {
|
|
410
|
+
if (value === null || typeof value === "string" || typeof value === "boolean") {
|
|
411
|
+
return value;
|
|
412
|
+
}
|
|
413
|
+
if (typeof value === "number") {
|
|
414
|
+
return Number.isFinite(value) ? value : null;
|
|
415
|
+
}
|
|
416
|
+
if (Array.isArray(value)) {
|
|
417
|
+
return value.map((item) => normalizeJsonValue(item) ?? null);
|
|
418
|
+
}
|
|
419
|
+
if (isPlainObject(value)) {
|
|
420
|
+
const normalized = {};
|
|
421
|
+
for (const [key, child] of Object.entries(value)) {
|
|
422
|
+
normalized[key] = normalizeJsonValue(child) ?? null;
|
|
423
|
+
}
|
|
424
|
+
return normalized;
|
|
425
|
+
}
|
|
426
|
+
return void 0;
|
|
427
|
+
}
|
|
428
|
+
function encodeValue(value, key, depth, options) {
|
|
429
|
+
if (isPrimitive(value)) {
|
|
430
|
+
return [line(depth, key === void 0 ? encodePrimitive(value, options.delimiter) : `${encodeKey(key)}: ${encodePrimitive(value, options.delimiter)}`, options.indent)];
|
|
431
|
+
}
|
|
432
|
+
if (Array.isArray(value)) {
|
|
433
|
+
return encodeArray(value, key, depth, options);
|
|
434
|
+
}
|
|
435
|
+
return encodeObject(value, key, depth, options);
|
|
436
|
+
}
|
|
437
|
+
function encodeObject(value, key, depth, options) {
|
|
438
|
+
const entries = Object.entries(value);
|
|
439
|
+
const lines = [];
|
|
440
|
+
if (key !== void 0) {
|
|
441
|
+
lines.push(line(depth, `${encodeKey(key)}:`, options.indent));
|
|
442
|
+
}
|
|
443
|
+
const childDepth = key === void 0 ? depth : depth + 1;
|
|
444
|
+
for (const [childKey, childValue] of entries) {
|
|
445
|
+
lines.push(...encodeValue(childValue, childKey, childDepth, options));
|
|
446
|
+
}
|
|
447
|
+
return lines;
|
|
448
|
+
}
|
|
449
|
+
function encodeArray(value, key, depth, options) {
|
|
450
|
+
const prefix = key === void 0 ? "" : encodeKey(key);
|
|
451
|
+
if (value.length === 0) {
|
|
452
|
+
return [line(depth, key === void 0 ? "[]" : `${prefix}: []`, options.indent)];
|
|
453
|
+
}
|
|
454
|
+
if (value.every(isPrimitive)) {
|
|
455
|
+
const joined = value.map((item) => encodePrimitive(item, options.delimiter)).join(options.delimiter);
|
|
456
|
+
return [line(depth, `${prefix}[${value.length}]: ${joined}`, options.indent)];
|
|
457
|
+
}
|
|
458
|
+
const tabularFields = getTabularFields(value);
|
|
459
|
+
if (tabularFields) {
|
|
460
|
+
const lines2 = [
|
|
461
|
+
line(depth, `${prefix}[${value.length}]{${tabularFields.map(encodeKey).join(options.delimiter)}}:`, options.indent)
|
|
462
|
+
];
|
|
463
|
+
for (const row of value) {
|
|
464
|
+
lines2.push(line(depth + 1, tabularFields.map((field) => encodePrimitive(row[field], options.delimiter)).join(options.delimiter), options.indent));
|
|
465
|
+
}
|
|
466
|
+
return lines2;
|
|
467
|
+
}
|
|
468
|
+
const lines = [line(depth, `${prefix}[${value.length}]:`, options.indent)];
|
|
469
|
+
for (const item of value) {
|
|
470
|
+
lines.push(...encodeListItem(item, depth + 1, options));
|
|
471
|
+
}
|
|
472
|
+
return lines;
|
|
473
|
+
}
|
|
474
|
+
function encodeListItem(value, depth, options) {
|
|
475
|
+
if (isPrimitive(value)) {
|
|
476
|
+
return [line(depth, `- ${encodePrimitive(value, options.delimiter)}`, options.indent)];
|
|
477
|
+
}
|
|
478
|
+
if (Array.isArray(value)) {
|
|
479
|
+
const encoded = encodeArray(value, void 0, depth, options);
|
|
480
|
+
return encoded.map((item, index) => {
|
|
481
|
+
if (index === 0) {
|
|
482
|
+
return line(depth, `- ${item.trimStart()}`, options.indent);
|
|
483
|
+
}
|
|
484
|
+
return item;
|
|
485
|
+
});
|
|
486
|
+
}
|
|
487
|
+
const entries = Object.entries(value);
|
|
488
|
+
if (entries.length === 0) {
|
|
489
|
+
return [line(depth, "-", options.indent)];
|
|
490
|
+
}
|
|
491
|
+
const [firstKey, firstValue] = entries[0];
|
|
492
|
+
const rest = Object.fromEntries(entries.slice(1));
|
|
493
|
+
const firstLines = encodeValue(firstValue, firstKey, depth, options);
|
|
494
|
+
const lines = [line(depth, `- ${firstLines[0].trimStart()}`, options.indent)];
|
|
495
|
+
for (const continuation of firstLines.slice(1)) {
|
|
496
|
+
lines.push(continuation);
|
|
497
|
+
}
|
|
498
|
+
if (Object.keys(rest).length > 0) {
|
|
499
|
+
lines.push(...encodeObject(rest, void 0, depth + 1, options));
|
|
500
|
+
}
|
|
501
|
+
return lines;
|
|
502
|
+
}
|
|
503
|
+
function getTabularFields(value) {
|
|
504
|
+
if (value.length === 0 || !value.every(isObjectRecord)) {
|
|
505
|
+
return void 0;
|
|
506
|
+
}
|
|
507
|
+
const first = value[0];
|
|
508
|
+
const fields = Object.keys(first);
|
|
509
|
+
if (fields.length === 0) {
|
|
510
|
+
return void 0;
|
|
511
|
+
}
|
|
512
|
+
for (const row of value) {
|
|
513
|
+
const keys = Object.keys(row);
|
|
514
|
+
if (keys.length !== fields.length) {
|
|
515
|
+
return void 0;
|
|
516
|
+
}
|
|
517
|
+
for (const field of fields) {
|
|
518
|
+
if (!(field in row) || !isPrimitive(row[field])) {
|
|
519
|
+
return void 0;
|
|
520
|
+
}
|
|
521
|
+
}
|
|
522
|
+
}
|
|
523
|
+
return fields;
|
|
524
|
+
}
|
|
525
|
+
function encodePrimitive(value, delimiter) {
|
|
526
|
+
if (value === null) {
|
|
527
|
+
return "null";
|
|
528
|
+
}
|
|
529
|
+
if (typeof value === "number" || typeof value === "boolean") {
|
|
530
|
+
return String(value);
|
|
531
|
+
}
|
|
532
|
+
return encodeString(value, delimiter);
|
|
533
|
+
}
|
|
534
|
+
function encodeString(value, delimiter) {
|
|
535
|
+
if (isSafeUnquotedString(value, delimiter)) {
|
|
536
|
+
return value;
|
|
537
|
+
}
|
|
538
|
+
return JSON.stringify(value);
|
|
539
|
+
}
|
|
540
|
+
function encodeKey(key) {
|
|
541
|
+
if (/^[A-Z_][\w.]*$/i.test(key)) {
|
|
542
|
+
return key;
|
|
543
|
+
}
|
|
544
|
+
return JSON.stringify(key);
|
|
545
|
+
}
|
|
546
|
+
function isSafeUnquotedString(value, delimiter) {
|
|
547
|
+
if (value.length === 0 || value !== value.trim()) {
|
|
548
|
+
return false;
|
|
549
|
+
}
|
|
550
|
+
if (/^(?:true|false|null)$/i.test(value) || /^-?\d+(?:\.\d+)?(?:e[+-]?\d+)?$/i.test(value) || /^0\d+$/.test(value)) {
|
|
551
|
+
return false;
|
|
552
|
+
}
|
|
553
|
+
if (value.includes(":") || value.includes('"') || value.includes("\\") || value.includes(delimiter)) {
|
|
554
|
+
return false;
|
|
555
|
+
}
|
|
556
|
+
if (/[[\]{}]|[\u0000-\u001F]/.test(value) || value.startsWith("-")) {
|
|
557
|
+
return false;
|
|
558
|
+
}
|
|
559
|
+
return true;
|
|
560
|
+
}
|
|
561
|
+
function isPrimitive(value) {
|
|
562
|
+
return value === null || typeof value === "string" || typeof value === "number" || typeof value === "boolean";
|
|
563
|
+
}
|
|
564
|
+
function isObjectRecord(value) {
|
|
565
|
+
return value !== null && typeof value === "object" && !Array.isArray(value);
|
|
566
|
+
}
|
|
567
|
+
function isPlainObject(value) {
|
|
568
|
+
if (value === null || typeof value !== "object") {
|
|
569
|
+
return false;
|
|
570
|
+
}
|
|
571
|
+
const prototype = Object.getPrototypeOf(value);
|
|
572
|
+
return prototype === null || prototype === Object.prototype;
|
|
573
|
+
}
|
|
574
|
+
function line(depth, content, indent) {
|
|
575
|
+
return `${" ".repeat(depth * indent)}${content}`;
|
|
576
|
+
}
|
|
577
|
+
|
|
578
|
+
// src/token-compression.ts
|
|
579
|
+
var TOKEN_REGEX = /[\p{L}\p{N}]+|[^\s]/gu;
|
|
580
|
+
var TOKEN_NORMALIZE_REGEX = /[^\p{L}\p{N}]/gu;
|
|
581
|
+
var DEDUPE_NORMALIZE_REGEX = /[^\p{L}\p{N}\s]/gu;
|
|
582
|
+
var BOILERPLATE_REGEX = /copyright|all rights reserved|disclaimer|confidential/i;
|
|
583
|
+
var MAX_SEGMENT_TOKENS = 120;
|
|
584
|
+
var STOP_WORDS = /* @__PURE__ */ new Set([
|
|
585
|
+
"a",
|
|
586
|
+
"an",
|
|
587
|
+
"and",
|
|
588
|
+
"are",
|
|
589
|
+
"as",
|
|
590
|
+
"at",
|
|
591
|
+
"be",
|
|
592
|
+
"by",
|
|
593
|
+
"for",
|
|
594
|
+
"from",
|
|
595
|
+
"how",
|
|
596
|
+
"in",
|
|
597
|
+
"is",
|
|
598
|
+
"it",
|
|
599
|
+
"of",
|
|
600
|
+
"on",
|
|
601
|
+
"or",
|
|
602
|
+
"that",
|
|
603
|
+
"the",
|
|
604
|
+
"this",
|
|
605
|
+
"to",
|
|
606
|
+
"was",
|
|
607
|
+
"we",
|
|
608
|
+
"with",
|
|
609
|
+
"you",
|
|
610
|
+
"your",
|
|
611
|
+
"our"
|
|
612
|
+
]);
|
|
613
|
+
var DEFAULT_OPTIONS = {
|
|
614
|
+
min_tokens: 80,
|
|
615
|
+
max_sentences: 10,
|
|
616
|
+
target_reduction: 0.45
|
|
617
|
+
};
|
|
618
|
+
function tokenizeForHeuristicCompression(text) {
|
|
619
|
+
if (!text) {
|
|
620
|
+
return [];
|
|
621
|
+
}
|
|
622
|
+
return text.match(TOKEN_REGEX) ?? [];
|
|
623
|
+
}
|
|
624
|
+
function estimateHeuristicTokens(text) {
|
|
625
|
+
return tokenizeForHeuristicCompression(text).length;
|
|
626
|
+
}
|
|
627
|
+
function compressHeuristicText(input, options = {}) {
|
|
628
|
+
const inputText = input ?? "";
|
|
629
|
+
const inputTokens = estimateHeuristicTokens(inputText);
|
|
630
|
+
if (options.json_to_toon === true) {
|
|
631
|
+
const toonResult = tryJsonToToon(inputText);
|
|
632
|
+
if (toonResult) {
|
|
633
|
+
const toonOutputTokens = estimateHeuristicTokens(toonResult.output);
|
|
634
|
+
if (toonOutputTokens <= inputTokens) {
|
|
635
|
+
return {
|
|
636
|
+
...toCompressionOutput(toonResult.output, inputTokens),
|
|
637
|
+
outputFormat: "toon"
|
|
638
|
+
};
|
|
639
|
+
}
|
|
640
|
+
return {
|
|
641
|
+
...toCompressionOutput(inputText, inputTokens),
|
|
642
|
+
warnings: ["JSON-to-TOON skipped because TOON was not smaller than the original JSON."]
|
|
643
|
+
};
|
|
644
|
+
}
|
|
645
|
+
return {
|
|
646
|
+
...toCompressionOutput(inputText, inputTokens),
|
|
647
|
+
warnings: ["JSON-to-TOON skipped because the input is not a complete valid JSON object or array."]
|
|
648
|
+
};
|
|
649
|
+
}
|
|
650
|
+
const analysis = preprocessContext(inputText);
|
|
651
|
+
if (inputTokens === 0) {
|
|
652
|
+
return toCompressionOutput(inputText, inputTokens);
|
|
653
|
+
}
|
|
654
|
+
const minTokens = options.min_tokens ?? DEFAULT_OPTIONS.min_tokens;
|
|
655
|
+
if (inputTokens <= minTokens) {
|
|
656
|
+
return toCompressionOutput(inputText, inputTokens);
|
|
657
|
+
}
|
|
658
|
+
const targetReduction = options.target_reduction ?? DEFAULT_OPTIONS.target_reduction;
|
|
659
|
+
const targetTokens = Math.max(1, Math.max(minTokens, Math.floor(inputTokens * (1 - targetReduction))));
|
|
660
|
+
const maxSentences = options.max_sentences ?? DEFAULT_OPTIONS.max_sentences;
|
|
661
|
+
const query = options.query ?? "";
|
|
662
|
+
const terms = queryTerms(query);
|
|
663
|
+
const scoredSentences = analysis.candidates.map((candidate) => ({
|
|
664
|
+
sentence: candidate.sentence,
|
|
665
|
+
index: candidate.index,
|
|
666
|
+
...scoreSentenceCandidate(candidate, terms),
|
|
667
|
+
tokens: candidate.tokenCount
|
|
668
|
+
}));
|
|
669
|
+
const selected = [];
|
|
670
|
+
const selectedIndices = /* @__PURE__ */ new Set();
|
|
671
|
+
const coveredTerms = /* @__PURE__ */ new Set();
|
|
672
|
+
let selectedTokens = 0;
|
|
673
|
+
while (selected.length < maxSentences) {
|
|
674
|
+
let bestCandidate;
|
|
675
|
+
let bestAdjustedScore = Number.NEGATIVE_INFINITY;
|
|
676
|
+
for (const candidate of scoredSentences) {
|
|
677
|
+
if (selectedIndices.has(candidate.index)) {
|
|
678
|
+
continue;
|
|
679
|
+
}
|
|
680
|
+
const underBudget = selectedTokens + candidate.tokens <= targetTokens;
|
|
681
|
+
const forceTopCandidate = selected.length === 0;
|
|
682
|
+
const forceRelevantSecond = selected.length === 1 && candidate.overlapCount > 0;
|
|
683
|
+
if (!underBudget && !forceTopCandidate && !forceRelevantSecond) {
|
|
684
|
+
continue;
|
|
685
|
+
}
|
|
686
|
+
let uncoveredOverlapCount = 0;
|
|
687
|
+
for (const term of candidate.overlapTerms) {
|
|
688
|
+
if (!coveredTerms.has(term)) {
|
|
689
|
+
uncoveredOverlapCount += 1;
|
|
690
|
+
}
|
|
691
|
+
}
|
|
692
|
+
const adjustedScore = candidate.score + uncoveredOverlapCount * 1.2;
|
|
693
|
+
if (adjustedScore > bestAdjustedScore || adjustedScore === bestAdjustedScore && bestCandidate && isBetterTieBreak(candidate, bestCandidate)) {
|
|
694
|
+
bestAdjustedScore = adjustedScore;
|
|
695
|
+
bestCandidate = candidate;
|
|
696
|
+
}
|
|
697
|
+
}
|
|
698
|
+
if (!bestCandidate) {
|
|
699
|
+
break;
|
|
700
|
+
}
|
|
701
|
+
selected.push(bestCandidate);
|
|
702
|
+
selectedIndices.add(bestCandidate.index);
|
|
703
|
+
selectedTokens += bestCandidate.tokens;
|
|
704
|
+
for (const term of bestCandidate.overlapTerms) {
|
|
705
|
+
coveredTerms.add(term);
|
|
706
|
+
}
|
|
707
|
+
}
|
|
708
|
+
selected.sort((left, right) => left.index - right.index);
|
|
709
|
+
let output = selected.map((item) => item.sentence).join(" ");
|
|
710
|
+
if (!output) {
|
|
711
|
+
output = tokenizeForHeuristicCompression(inputText).slice(0, targetTokens).join(" ");
|
|
712
|
+
} else {
|
|
713
|
+
const outputTokens = tokenizeForHeuristicCompression(output);
|
|
714
|
+
if (outputTokens.length > targetTokens) {
|
|
715
|
+
output = outputTokens.slice(0, targetTokens).join(" ");
|
|
716
|
+
}
|
|
717
|
+
}
|
|
718
|
+
return toCompressionOutput(output, inputTokens);
|
|
719
|
+
}
|
|
720
|
+
function toCompressionOutput(output, inputTokens) {
|
|
721
|
+
const outputTokens = estimateHeuristicTokens(output);
|
|
722
|
+
const tokensSaved = Math.max(0, inputTokens - outputTokens);
|
|
723
|
+
return {
|
|
724
|
+
output,
|
|
725
|
+
inputTokens,
|
|
726
|
+
outputTokens,
|
|
727
|
+
tokensSaved,
|
|
728
|
+
compressionRatio: outputTokens === 0 ? 0 : inputTokens / outputTokens
|
|
729
|
+
};
|
|
730
|
+
}
|
|
731
|
+
function isBetterTieBreak(candidate, bestCandidate) {
|
|
732
|
+
if (candidate.overlapCount !== bestCandidate.overlapCount) {
|
|
733
|
+
return candidate.overlapCount > bestCandidate.overlapCount;
|
|
734
|
+
}
|
|
735
|
+
if (candidate.tokens !== bestCandidate.tokens) {
|
|
736
|
+
return candidate.tokens < bestCandidate.tokens;
|
|
737
|
+
}
|
|
738
|
+
return candidate.index < bestCandidate.index;
|
|
739
|
+
}
|
|
740
|
+
function normalizeForDeduping(sentence) {
|
|
741
|
+
return sentence.toLowerCase().replace(DEDUPE_NORMALIZE_REGEX, " ").replace(/\s+/g, " ").trim();
|
|
742
|
+
}
|
|
743
|
+
function splitIntoSentences(text) {
|
|
744
|
+
return text.split(/(?<=[.!?])\s+|\n+/g).map((sentence) => sentence.trim()).filter(Boolean);
|
|
745
|
+
}
|
|
746
|
+
function chunkTokens(tokens, chunkSize) {
|
|
747
|
+
const chunks = [];
|
|
748
|
+
for (let index = 0; index < tokens.length; index += chunkSize) {
|
|
749
|
+
chunks.push(tokens.slice(index, index + chunkSize));
|
|
750
|
+
}
|
|
751
|
+
return chunks;
|
|
752
|
+
}
|
|
753
|
+
function splitOversizedSentence(sentence, maxTokens = MAX_SEGMENT_TOKENS) {
|
|
754
|
+
const rawTokens = tokenizeForHeuristicCompression(sentence);
|
|
755
|
+
if (rawTokens.length <= maxTokens) {
|
|
756
|
+
return [sentence];
|
|
757
|
+
}
|
|
758
|
+
const commaSplit = sentence.split(/(?<=[,;:])\s+|\s+-\s+/g).map((segment) => segment.trim()).filter(Boolean);
|
|
759
|
+
if (commaSplit.length > 1) {
|
|
760
|
+
return commaSplit.flatMap((segment) => {
|
|
761
|
+
if (estimateHeuristicTokens(segment) <= maxTokens) {
|
|
762
|
+
return [segment];
|
|
763
|
+
}
|
|
764
|
+
return chunkTokens(tokenizeForHeuristicCompression(segment), maxTokens).map((chunk) => chunk.join(" "));
|
|
765
|
+
});
|
|
766
|
+
}
|
|
767
|
+
return chunkTokens(rawTokens, maxTokens).map((chunk) => chunk.join(" "));
|
|
768
|
+
}
|
|
769
|
+
function normalizeToken(token) {
|
|
770
|
+
return token.replace(TOKEN_NORMALIZE_REGEX, "").trim();
|
|
771
|
+
}
|
|
772
|
+
function tokenizeNormalized(text) {
|
|
773
|
+
return tokenizeForHeuristicCompression(text.toLowerCase()).map(normalizeToken).filter(Boolean);
|
|
774
|
+
}
|
|
775
|
+
function queryTerms(query) {
|
|
776
|
+
return new Set(tokenizeNormalized(query).filter((term) => term.length >= 3 && !STOP_WORDS.has(term)));
|
|
777
|
+
}
|
|
778
|
+
function preprocessContext(context) {
|
|
779
|
+
const originalTokens = estimateHeuristicTokens(context);
|
|
780
|
+
const dedupeSet = /* @__PURE__ */ new Set();
|
|
781
|
+
const candidates = [];
|
|
782
|
+
for (const sentence of splitIntoSentences(context)) {
|
|
783
|
+
for (const segment of splitOversizedSentence(sentence)) {
|
|
784
|
+
const normalized = normalizeForDeduping(segment);
|
|
785
|
+
if (!normalized || dedupeSet.has(normalized)) {
|
|
786
|
+
continue;
|
|
787
|
+
}
|
|
788
|
+
dedupeSet.add(normalized);
|
|
789
|
+
const rawTokens = tokenizeForHeuristicCompression(segment);
|
|
790
|
+
const normalizedTerms = rawTokens.map((token) => normalizeToken(token.toLowerCase())).filter(Boolean);
|
|
791
|
+
const lowered = segment.toLowerCase();
|
|
792
|
+
candidates.push({
|
|
793
|
+
sentence: segment,
|
|
794
|
+
index: candidates.length,
|
|
795
|
+
tokenCount: rawTokens.length,
|
|
796
|
+
uniqueTerms: new Set(normalizedTerms),
|
|
797
|
+
isBoilerplate: BOILERPLATE_REGEX.test(lowered)
|
|
798
|
+
});
|
|
799
|
+
}
|
|
800
|
+
}
|
|
801
|
+
return { originalTokens, candidates };
|
|
802
|
+
}
|
|
803
|
+
function scoreSentenceCandidate(candidate, terms) {
|
|
804
|
+
if (candidate.tokenCount === 0) {
|
|
805
|
+
return {
|
|
806
|
+
score: 0,
|
|
807
|
+
overlapCount: 0,
|
|
808
|
+
overlapTerms: []
|
|
809
|
+
};
|
|
810
|
+
}
|
|
811
|
+
const overlapTerms = [];
|
|
812
|
+
let overlap = 0;
|
|
813
|
+
for (const term of terms) {
|
|
814
|
+
if (candidate.uniqueTerms.has(term)) {
|
|
815
|
+
overlap += 1;
|
|
816
|
+
overlapTerms.push(term);
|
|
817
|
+
}
|
|
818
|
+
}
|
|
819
|
+
const overlapScore = overlap * 2;
|
|
820
|
+
const densityScore = terms.size > 0 ? overlap / terms.size : 0;
|
|
821
|
+
const lengthScore = Math.min(candidate.tokenCount, 40) / 40;
|
|
822
|
+
const boilerplatePenalty = candidate.isBoilerplate ? 0.5 : 0;
|
|
823
|
+
return {
|
|
824
|
+
score: overlapScore + densityScore + lengthScore - boilerplatePenalty,
|
|
825
|
+
overlapCount: overlap,
|
|
826
|
+
overlapTerms
|
|
827
|
+
};
|
|
828
|
+
}
|
|
829
|
+
|
|
830
|
+
// src/code-compaction.ts
|
|
831
|
+
function compactCode(input, options = {}) {
|
|
832
|
+
const inputText = input ?? "";
|
|
833
|
+
const removeComments = options.remove_comments ?? true;
|
|
834
|
+
const trimIndentation = options.trim_indentation ?? true;
|
|
835
|
+
const collapseBlankLines = options.collapse_blank_lines ?? true;
|
|
836
|
+
let output = inputText.replace(/\r\n?/g, "\n");
|
|
837
|
+
if (removeComments) {
|
|
838
|
+
output = stripCodeComments(output);
|
|
839
|
+
}
|
|
840
|
+
output = output.split("\n").map((line2) => line2.replace(/[ \t]+$/g, "")).join("\n");
|
|
841
|
+
if (trimIndentation) {
|
|
842
|
+
output = stripCommonIndent(output);
|
|
843
|
+
}
|
|
844
|
+
if (collapseBlankLines) {
|
|
845
|
+
output = output.split("\n").filter((line2) => line2.trim().length > 0).join("\n");
|
|
846
|
+
}
|
|
847
|
+
output = output.trim();
|
|
848
|
+
const inputTokens = estimateHeuristicTokens(inputText);
|
|
849
|
+
const outputTokens = estimateHeuristicTokens(output);
|
|
850
|
+
return {
|
|
851
|
+
output,
|
|
852
|
+
inputTokens,
|
|
853
|
+
outputTokens,
|
|
854
|
+
tokensSaved: Math.max(0, inputTokens - outputTokens),
|
|
855
|
+
compressionRatio: outputTokens === 0 ? 0 : inputTokens / outputTokens
|
|
856
|
+
};
|
|
857
|
+
}
|
|
858
|
+
function stripCommonIndent(input) {
|
|
859
|
+
const lines = input.split("\n");
|
|
860
|
+
const nonEmptyLines = lines.filter((line2) => line2.trim().length > 0);
|
|
861
|
+
if (nonEmptyLines.length === 0) {
|
|
862
|
+
return "";
|
|
863
|
+
}
|
|
864
|
+
const commonIndent = Math.min(
|
|
865
|
+
...nonEmptyLines.map((line2) => {
|
|
866
|
+
const match = line2.match(/^[ \t]*/);
|
|
867
|
+
return match?.[0].length ?? 0;
|
|
868
|
+
})
|
|
869
|
+
);
|
|
870
|
+
if (commonIndent === 0) {
|
|
871
|
+
return input;
|
|
872
|
+
}
|
|
873
|
+
return lines.map((line2) => line2.slice(commonIndent)).join("\n");
|
|
874
|
+
}
|
|
875
|
+
function stripCodeComments(input) {
|
|
876
|
+
let output = "";
|
|
877
|
+
let index = 0;
|
|
878
|
+
let quote;
|
|
879
|
+
let escaped = false;
|
|
880
|
+
let atLineStart = true;
|
|
881
|
+
while (index < input.length) {
|
|
882
|
+
const char = input[index];
|
|
883
|
+
const next = input[index + 1];
|
|
884
|
+
if (quote) {
|
|
885
|
+
output += char;
|
|
886
|
+
if (escaped) {
|
|
887
|
+
escaped = false;
|
|
888
|
+
} else if (char === "\\") {
|
|
889
|
+
escaped = true;
|
|
890
|
+
} else if (char === quote) {
|
|
891
|
+
quote = void 0;
|
|
892
|
+
}
|
|
893
|
+
atLineStart = char === "\n";
|
|
894
|
+
index += 1;
|
|
895
|
+
continue;
|
|
896
|
+
}
|
|
897
|
+
if (char === '"' || char === "'" || char === "`") {
|
|
898
|
+
quote = char;
|
|
899
|
+
output += char;
|
|
900
|
+
atLineStart = false;
|
|
901
|
+
index += 1;
|
|
902
|
+
continue;
|
|
903
|
+
}
|
|
904
|
+
if (atLineStart && char === "#" && next === "!") {
|
|
905
|
+
while (index < input.length && input[index] !== "\n") {
|
|
906
|
+
output += input[index];
|
|
907
|
+
index += 1;
|
|
908
|
+
}
|
|
909
|
+
continue;
|
|
910
|
+
}
|
|
911
|
+
if (char === "/" && next === "/") {
|
|
912
|
+
index += 2;
|
|
913
|
+
while (index < input.length && input[index] !== "\n") {
|
|
914
|
+
index += 1;
|
|
915
|
+
}
|
|
916
|
+
continue;
|
|
917
|
+
}
|
|
918
|
+
if (char === "/" && next === "*") {
|
|
919
|
+
index += 2;
|
|
920
|
+
while (index < input.length && !(input[index] === "*" && input[index + 1] === "/")) {
|
|
921
|
+
if (input[index] === "\n") {
|
|
922
|
+
output += "\n";
|
|
923
|
+
}
|
|
924
|
+
index += 1;
|
|
925
|
+
}
|
|
926
|
+
index += input[index] === "*" ? 2 : 0;
|
|
927
|
+
continue;
|
|
928
|
+
}
|
|
929
|
+
if (char === "#" && (next === " " || next === " ")) {
|
|
930
|
+
index += 1;
|
|
931
|
+
while (index < input.length && input[index] !== "\n") {
|
|
932
|
+
index += 1;
|
|
933
|
+
}
|
|
934
|
+
continue;
|
|
935
|
+
}
|
|
936
|
+
output += char;
|
|
937
|
+
atLineStart = char === "\n" || atLineStart && (char === " " || char === " ");
|
|
938
|
+
index += 1;
|
|
939
|
+
}
|
|
940
|
+
return output;
|
|
941
|
+
}
|
|
942
|
+
|
|
334
943
|
// src/renderer/renderer.ts
|
|
335
944
|
function renderSections(asset, options = {}) {
|
|
336
945
|
const { variables = {}, strict = false } = options;
|
|
337
|
-
const
|
|
946
|
+
const contextInputs = getContextInputs(asset);
|
|
947
|
+
const optionalVariables = contextInputs.filter((input) => input.optional === true).map((input) => input.name);
|
|
948
|
+
const contextInputsByName = new Map(contextInputs.map((input) => [input.name, input]));
|
|
949
|
+
const compressionCache = /* @__PURE__ */ new Map();
|
|
950
|
+
const createTransformVariable = (template) => {
|
|
951
|
+
return ({ name, value, modifier, match }) => {
|
|
952
|
+
const input = contextInputsByName.get(name);
|
|
953
|
+
const configuredHeuristic = input?.compression?.heuristic;
|
|
954
|
+
const configuredCode = input?.compression?.code;
|
|
955
|
+
const shouldCompactCode = modifier === "compact" || modifier === "code" || configuredCode?.enabled === true;
|
|
956
|
+
const shouldCompress = modifier === "compress" || configuredHeuristic?.enabled === true;
|
|
957
|
+
if (modifier === "toon") {
|
|
958
|
+
const result3 = toToonPlaceholder(value);
|
|
959
|
+
if (!result3) {
|
|
960
|
+
options.compression?.onCompressionWarning?.(
|
|
961
|
+
`POK031: JSON-to-TOON skipped for placeholder "${name}" because the value is not a complete valid JSON object or array.`
|
|
962
|
+
);
|
|
963
|
+
return value;
|
|
964
|
+
}
|
|
965
|
+
if (result3.tokensSaved > 0 || result3.output !== value) {
|
|
966
|
+
options.compression?.onHeuristicCompression?.({
|
|
967
|
+
scope: "placeholder",
|
|
968
|
+
variable: name,
|
|
969
|
+
...result3
|
|
970
|
+
});
|
|
971
|
+
}
|
|
972
|
+
return result3.output;
|
|
973
|
+
}
|
|
974
|
+
if (shouldCompactCode) {
|
|
975
|
+
const result3 = compactCode(value, toCodeCompactionOptions(configuredCode));
|
|
976
|
+
if (result3.tokensSaved > 0 || result3.output !== value) {
|
|
977
|
+
options.compression?.onCodeCompaction?.({
|
|
978
|
+
scope: "placeholder",
|
|
979
|
+
variable: name,
|
|
980
|
+
...result3
|
|
981
|
+
});
|
|
982
|
+
}
|
|
983
|
+
return result3.output;
|
|
984
|
+
}
|
|
985
|
+
if (!shouldCompress) {
|
|
986
|
+
return value;
|
|
987
|
+
}
|
|
988
|
+
const query = resolveHeuristicQuery(configuredHeuristic, variables, template, match);
|
|
989
|
+
const compressionOptions = {
|
|
990
|
+
min_tokens: configuredHeuristic?.min_tokens,
|
|
991
|
+
max_sentences: configuredHeuristic?.max_sentences,
|
|
992
|
+
target_reduction: configuredHeuristic?.target_reduction,
|
|
993
|
+
query,
|
|
994
|
+
json_to_toon: configuredHeuristic?.json_to_toon
|
|
995
|
+
};
|
|
996
|
+
const cacheKey = JSON.stringify([name, value, compressionOptions]);
|
|
997
|
+
const cached = compressionCache.get(cacheKey);
|
|
998
|
+
if (cached !== void 0) {
|
|
999
|
+
return cached;
|
|
1000
|
+
}
|
|
1001
|
+
const result2 = compressHeuristicText(value, compressionOptions);
|
|
1002
|
+
compressionCache.set(cacheKey, result2.output);
|
|
1003
|
+
reportHeuristicCompressionWarnings(options, result2, `placeholder "${name}"`);
|
|
1004
|
+
if (result2.tokensSaved > 0 || result2.output !== value) {
|
|
1005
|
+
options.compression?.onHeuristicCompression?.({
|
|
1006
|
+
scope: "placeholder",
|
|
1007
|
+
variable: name,
|
|
1008
|
+
...result2
|
|
1009
|
+
});
|
|
1010
|
+
}
|
|
1011
|
+
return result2.output;
|
|
1012
|
+
};
|
|
1013
|
+
};
|
|
338
1014
|
const result = {};
|
|
339
1015
|
if (asset.sections.system_instructions) {
|
|
340
1016
|
result.system_instructions = interpolate(
|
|
341
1017
|
asset.sections.system_instructions,
|
|
342
1018
|
variables,
|
|
343
|
-
{
|
|
1019
|
+
{
|
|
1020
|
+
strict,
|
|
1021
|
+
optionalVariables,
|
|
1022
|
+
transformVariable: createTransformVariable(asset.sections.system_instructions)
|
|
1023
|
+
}
|
|
344
1024
|
);
|
|
345
1025
|
}
|
|
346
1026
|
if (asset.sections.prompt_template) {
|
|
347
1027
|
result.prompt_template = interpolate(
|
|
348
1028
|
asset.sections.prompt_template,
|
|
349
1029
|
variables,
|
|
350
|
-
{
|
|
1030
|
+
{
|
|
1031
|
+
strict,
|
|
1032
|
+
optionalVariables,
|
|
1033
|
+
transformVariable: createTransformVariable(asset.sections.prompt_template)
|
|
1034
|
+
}
|
|
351
1035
|
);
|
|
352
1036
|
}
|
|
353
1037
|
return result;
|
|
354
1038
|
}
|
|
1039
|
+
function resolveHeuristicQuery(options, variables, template, match) {
|
|
1040
|
+
if (options?.query !== void 0) {
|
|
1041
|
+
return options.query;
|
|
1042
|
+
}
|
|
1043
|
+
if (options?.query_variable && variables[options.query_variable] !== void 0) {
|
|
1044
|
+
return variables[options.query_variable];
|
|
1045
|
+
}
|
|
1046
|
+
return template.replace(match, " ");
|
|
1047
|
+
}
|
|
1048
|
+
function toToonPlaceholder(value) {
|
|
1049
|
+
const toon = tryJsonToToon(value);
|
|
1050
|
+
if (!toon) {
|
|
1051
|
+
return void 0;
|
|
1052
|
+
}
|
|
1053
|
+
const inputTokens = estimateHeuristicTokens(value);
|
|
1054
|
+
const outputTokens = estimateHeuristicTokens(toon.output);
|
|
1055
|
+
return {
|
|
1056
|
+
output: toon.output,
|
|
1057
|
+
inputTokens,
|
|
1058
|
+
outputTokens,
|
|
1059
|
+
tokensSaved: Math.max(0, inputTokens - outputTokens),
|
|
1060
|
+
compressionRatio: outputTokens === 0 ? 0 : inputTokens / outputTokens,
|
|
1061
|
+
outputFormat: "toon"
|
|
1062
|
+
};
|
|
1063
|
+
}
|
|
1064
|
+
function toCodeCompactionOptions(options) {
|
|
1065
|
+
return {
|
|
1066
|
+
remove_comments: options?.remove_comments,
|
|
1067
|
+
trim_indentation: options?.trim_indentation,
|
|
1068
|
+
collapse_blank_lines: options?.collapse_blank_lines
|
|
1069
|
+
};
|
|
1070
|
+
}
|
|
1071
|
+
function reportHeuristicCompressionWarnings(options, result, scope) {
|
|
1072
|
+
for (const warning of result.warnings ?? []) {
|
|
1073
|
+
options.compression?.onCompressionWarning?.(`POK031: ${warning} Scope: ${scope}.`);
|
|
1074
|
+
}
|
|
1075
|
+
}
|
|
355
1076
|
|
|
356
1077
|
// src/overrides/apply-overrides.ts
|
|
357
1078
|
function applyOverrides(asset, options = {}) {
|
|
@@ -388,6 +1109,14 @@ function mergeOverride(base, override) {
|
|
|
388
1109
|
thetokencompany: mergeRecordBlock(
|
|
389
1110
|
result.compression?.thetokencompany,
|
|
390
1111
|
override.compression.thetokencompany
|
|
1112
|
+
),
|
|
1113
|
+
heuristic: mergeRecordBlock(
|
|
1114
|
+
result.compression?.heuristic,
|
|
1115
|
+
override.compression.heuristic
|
|
1116
|
+
),
|
|
1117
|
+
code: mergeRecordBlock(
|
|
1118
|
+
result.compression?.code,
|
|
1119
|
+
override.compression.code
|
|
391
1120
|
)
|
|
392
1121
|
};
|
|
393
1122
|
}
|
|
@@ -621,8 +1350,25 @@ var TheTokenCompanyCompressionSchema = import_zod.z.object({
|
|
|
621
1350
|
model: import_zod.z.string().min(1).optional(),
|
|
622
1351
|
aggressiveness: import_zod.z.number().min(0).max(1).optional()
|
|
623
1352
|
});
|
|
1353
|
+
var HeuristicCompressionSchema = import_zod.z.object({
|
|
1354
|
+
enabled: import_zod.z.boolean().optional(),
|
|
1355
|
+
min_tokens: import_zod.z.number().int().positive().optional(),
|
|
1356
|
+
max_sentences: import_zod.z.number().int().positive().optional(),
|
|
1357
|
+
target_reduction: import_zod.z.number().min(0).max(1).optional(),
|
|
1358
|
+
query: import_zod.z.string().optional(),
|
|
1359
|
+
query_variable: import_zod.z.string().min(1).optional(),
|
|
1360
|
+
json_to_toon: import_zod.z.boolean().optional()
|
|
1361
|
+
});
|
|
1362
|
+
var CodeCompactionSchema = import_zod.z.object({
|
|
1363
|
+
enabled: import_zod.z.boolean().optional(),
|
|
1364
|
+
remove_comments: import_zod.z.boolean().optional(),
|
|
1365
|
+
trim_indentation: import_zod.z.boolean().optional(),
|
|
1366
|
+
collapse_blank_lines: import_zod.z.boolean().optional()
|
|
1367
|
+
});
|
|
624
1368
|
var CompressionSchema = import_zod.z.object({
|
|
625
|
-
thetokencompany: TheTokenCompanyCompressionSchema.optional()
|
|
1369
|
+
thetokencompany: TheTokenCompanyCompressionSchema.optional(),
|
|
1370
|
+
heuristic: HeuristicCompressionSchema.optional(),
|
|
1371
|
+
code: CodeCompactionSchema.optional()
|
|
626
1372
|
});
|
|
627
1373
|
var OpenAICacheSchema = import_zod.z.object({
|
|
628
1374
|
prompt_cache_key: import_zod.z.string().min(1).optional(),
|
|
@@ -662,12 +1408,21 @@ var ContextBuiltInValidatorSchema = import_zod.z.union([
|
|
|
662
1408
|
return_message: import_zod.z.string().optional()
|
|
663
1409
|
})
|
|
664
1410
|
]);
|
|
1411
|
+
var ContextInputCompressionSchema = import_zod.z.union([
|
|
1412
|
+
import_zod.z.literal("heuristic"),
|
|
1413
|
+
import_zod.z.literal("code"),
|
|
1414
|
+
import_zod.z.object({
|
|
1415
|
+
heuristic: HeuristicCompressionSchema.optional(),
|
|
1416
|
+
code: CodeCompactionSchema.optional()
|
|
1417
|
+
})
|
|
1418
|
+
]);
|
|
665
1419
|
var ContextInputDefinitionObjectSchema = import_zod.z.object({
|
|
666
1420
|
name: import_zod.z.string(),
|
|
667
1421
|
optional: import_zod.z.boolean().optional(),
|
|
668
1422
|
warnings: import_zod.z.boolean().optional(),
|
|
669
1423
|
max_size: import_zod.z.number().int().positive().optional(),
|
|
670
1424
|
trim: import_zod.z.union([import_zod.z.boolean(), import_zod.z.enum(["start", "end", "both"])]).optional(),
|
|
1425
|
+
compression: ContextInputCompressionSchema.optional(),
|
|
671
1426
|
allow_regex: ContextRegexSchema.optional(),
|
|
672
1427
|
deny_regex: ContextRegexSchema.optional(),
|
|
673
1428
|
non_empty: ContextBuiltInValidatorSchema.optional(),
|
|
@@ -770,9 +1525,9 @@ function assertRegexFrontMatterQuoting(content, filePath) {
|
|
|
770
1525
|
return;
|
|
771
1526
|
}
|
|
772
1527
|
let pendingRegex;
|
|
773
|
-
for (const [index,
|
|
1528
|
+
for (const [index, line2] of frontMatter.lines.entries()) {
|
|
774
1529
|
const lineNumber = frontMatter.startLine + index;
|
|
775
|
-
const regexMatch =
|
|
1530
|
+
const regexMatch = line2.match(REGEX_FIELD);
|
|
776
1531
|
if (regexMatch?.groups) {
|
|
777
1532
|
const field = regexMatch.groups.field;
|
|
778
1533
|
const indent2 = regexMatch.groups.indent.length;
|
|
@@ -785,15 +1540,15 @@ function assertRegexFrontMatterQuoting(content, filePath) {
|
|
|
785
1540
|
if (!pendingRegex) {
|
|
786
1541
|
continue;
|
|
787
1542
|
}
|
|
788
|
-
if (
|
|
1543
|
+
if (line2.trim() === "" || line2.trimStart().startsWith("#")) {
|
|
789
1544
|
continue;
|
|
790
1545
|
}
|
|
791
|
-
const indent =
|
|
1546
|
+
const indent = line2.match(/^\s*/)?.[0].length ?? 0;
|
|
792
1547
|
if (indent <= pendingRegex.indent) {
|
|
793
1548
|
pendingRegex = void 0;
|
|
794
1549
|
continue;
|
|
795
1550
|
}
|
|
796
|
-
const patternMatch =
|
|
1551
|
+
const patternMatch = line2.match(PATTERN_FIELD);
|
|
797
1552
|
if (!patternMatch?.groups) {
|
|
798
1553
|
continue;
|
|
799
1554
|
}
|
|
@@ -888,8 +1643,8 @@ function extractSections(body) {
|
|
|
888
1643
|
let currentKey = null;
|
|
889
1644
|
let currentLines = [];
|
|
890
1645
|
let foundAnyH1 = false;
|
|
891
|
-
for (const
|
|
892
|
-
const h1Match =
|
|
1646
|
+
for (const line2 of lines) {
|
|
1647
|
+
const h1Match = line2.match(/^#\s+(.+)$/);
|
|
893
1648
|
if (h1Match) {
|
|
894
1649
|
if (currentKey) {
|
|
895
1650
|
sections[currentKey] = currentLines.join("\n").trim();
|
|
@@ -899,7 +1654,7 @@ function extractSections(body) {
|
|
|
899
1654
|
currentKey = SECTION_MAP[heading] ?? null;
|
|
900
1655
|
currentLines = [];
|
|
901
1656
|
} else {
|
|
902
|
-
currentLines.push(
|
|
1657
|
+
currentLines.push(line2);
|
|
903
1658
|
}
|
|
904
1659
|
}
|
|
905
1660
|
if (currentKey) {
|
|
@@ -1232,7 +1987,9 @@ function mergeCompression(base, local) {
|
|
|
1232
1987
|
const merged = {
|
|
1233
1988
|
...base ?? {},
|
|
1234
1989
|
...local ?? {},
|
|
1235
|
-
thetokencompany: mergeRecordBlock2(base?.thetokencompany, local?.thetokencompany)
|
|
1990
|
+
thetokencompany: mergeRecordBlock2(base?.thetokencompany, local?.thetokencompany),
|
|
1991
|
+
heuristic: mergeRecordBlock2(base?.heuristic, local?.heuristic),
|
|
1992
|
+
code: mergeRecordBlock2(base?.code, local?.code)
|
|
1236
1993
|
};
|
|
1237
1994
|
removeEmptyProviderBlocks(merged);
|
|
1238
1995
|
return Object.keys(merged).length > 0 ? merged : void 0;
|
|
@@ -1486,33 +2243,152 @@ function compactHistoryForPrompt(asset, runtime) {
|
|
|
1486
2243
|
// src/compression.ts
|
|
1487
2244
|
var THETOKENCOMPANY_DEFAULT_MODEL = "bear-2";
|
|
1488
2245
|
var THETOKENCOMPANY_DEFAULT_BASE_URL = "https://api.thetokencompany.com";
|
|
2246
|
+
var HEURISTIC_COMPRESSION_MODEL = "local-heuristic-v1";
|
|
2247
|
+
var CODE_COMPACTION_MODEL = "local-code-compactor-v1";
|
|
2248
|
+
function summarizePromptCompression(compression = []) {
|
|
2249
|
+
const totals = compression.reduce(
|
|
2250
|
+
(accumulator, result) => {
|
|
2251
|
+
accumulator.inputTokens += result.inputTokens;
|
|
2252
|
+
accumulator.outputTokens += result.outputTokens;
|
|
2253
|
+
accumulator.tokensSaved += result.tokensSaved;
|
|
2254
|
+
return accumulator;
|
|
2255
|
+
},
|
|
2256
|
+
{
|
|
2257
|
+
inputTokens: 0,
|
|
2258
|
+
outputTokens: 0,
|
|
2259
|
+
tokensSaved: 0
|
|
2260
|
+
}
|
|
2261
|
+
);
|
|
2262
|
+
return {
|
|
2263
|
+
steps: compression.length,
|
|
2264
|
+
inputTokens: totals.inputTokens,
|
|
2265
|
+
outputTokens: totals.outputTokens,
|
|
2266
|
+
tokensSaved: totals.tokensSaved,
|
|
2267
|
+
reductionRatio: totals.inputTokens === 0 ? 0 : totals.tokensSaved / totals.inputTokens
|
|
2268
|
+
};
|
|
2269
|
+
}
|
|
1489
2270
|
async function applyPromptCompressionForRender(asset, runtime) {
|
|
1490
|
-
const
|
|
1491
|
-
|
|
1492
|
-
|
|
1493
|
-
|
|
2271
|
+
const theTokenCompanyConfig = asset.compression?.thetokencompany;
|
|
2272
|
+
const heuristicConfig = asset.compression?.heuristic;
|
|
2273
|
+
const codeConfig = asset.compression?.code;
|
|
2274
|
+
const hasTheTokenCompanyCompression = theTokenCompanyConfig?.enabled === true && Boolean(asset.sections.prompt_template);
|
|
2275
|
+
const hasHeuristicPromptCompression = heuristicConfig?.enabled === true && Boolean(asset.sections.prompt_template);
|
|
2276
|
+
const hasCodePromptCompaction = codeConfig?.enabled === true && Boolean(asset.sections.prompt_template);
|
|
2277
|
+
const hasPlaceholderCompression = usesPlaceholderCompression(asset);
|
|
2278
|
+
if (!hasTheTokenCompanyCompression && !hasHeuristicPromptCompression && !hasCodePromptCompaction && !hasPlaceholderCompression) {
|
|
2279
|
+
return { asset, runtime, compression: [], warnings: [] };
|
|
2280
|
+
}
|
|
2281
|
+
const compression = [];
|
|
2282
|
+
const warnings = [];
|
|
1494
2283
|
const sections = renderSections(asset, {
|
|
1495
2284
|
variables: runtime.variables,
|
|
1496
|
-
strict: runtime.strict
|
|
1497
|
-
|
|
1498
|
-
|
|
1499
|
-
|
|
1500
|
-
|
|
1501
|
-
|
|
1502
|
-
|
|
1503
|
-
|
|
1504
|
-
|
|
1505
|
-
|
|
1506
|
-
|
|
1507
|
-
|
|
2285
|
+
strict: runtime.strict,
|
|
2286
|
+
compression: {
|
|
2287
|
+
onHeuristicCompression: (event) => {
|
|
2288
|
+
compression.push({
|
|
2289
|
+
provider: "heuristic",
|
|
2290
|
+
model: HEURISTIC_COMPRESSION_MODEL,
|
|
2291
|
+
inputTokens: event.inputTokens,
|
|
2292
|
+
outputTokens: event.outputTokens,
|
|
2293
|
+
tokensSaved: event.tokensSaved,
|
|
2294
|
+
compressionRatio: event.compressionRatio,
|
|
2295
|
+
scope: event.scope,
|
|
2296
|
+
variable: event.variable,
|
|
2297
|
+
outputFormat: event.outputFormat
|
|
2298
|
+
});
|
|
2299
|
+
},
|
|
2300
|
+
onCodeCompaction: (event) => {
|
|
2301
|
+
compression.push({
|
|
2302
|
+
provider: "code",
|
|
2303
|
+
model: CODE_COMPACTION_MODEL,
|
|
2304
|
+
inputTokens: event.inputTokens,
|
|
2305
|
+
outputTokens: event.outputTokens,
|
|
2306
|
+
tokensSaved: event.tokensSaved,
|
|
2307
|
+
compressionRatio: event.compressionRatio,
|
|
2308
|
+
scope: event.scope,
|
|
2309
|
+
variable: event.variable,
|
|
2310
|
+
outputFormat: "code"
|
|
2311
|
+
});
|
|
2312
|
+
},
|
|
2313
|
+
onCompressionWarning: (warning) => {
|
|
2314
|
+
warnings.push(warning);
|
|
2315
|
+
}
|
|
2316
|
+
}
|
|
1508
2317
|
});
|
|
2318
|
+
let promptTemplate = sections.prompt_template;
|
|
2319
|
+
if (hasCodePromptCompaction && promptTemplate) {
|
|
2320
|
+
if (hasHeuristicPromptCompression) {
|
|
2321
|
+
warnings.push(
|
|
2322
|
+
"POK032: Local heuristic prompt compression skipped because compression.code is enabled for the prompt template."
|
|
2323
|
+
);
|
|
2324
|
+
}
|
|
2325
|
+
const result = compactCode(promptTemplate, toCodeCompactionOptions2(codeConfig));
|
|
2326
|
+
promptTemplate = result.output;
|
|
2327
|
+
if (result.tokensSaved > 0 || result.output !== sections.prompt_template) {
|
|
2328
|
+
compression.push({
|
|
2329
|
+
provider: "code",
|
|
2330
|
+
model: CODE_COMPACTION_MODEL,
|
|
2331
|
+
inputTokens: result.inputTokens,
|
|
2332
|
+
outputTokens: result.outputTokens,
|
|
2333
|
+
tokensSaved: result.tokensSaved,
|
|
2334
|
+
compressionRatio: result.compressionRatio,
|
|
2335
|
+
scope: "prompt_template",
|
|
2336
|
+
outputFormat: "code"
|
|
2337
|
+
});
|
|
2338
|
+
}
|
|
2339
|
+
} else if (hasHeuristicPromptCompression && promptTemplate) {
|
|
2340
|
+
const result = compressHeuristicText(promptTemplate, {
|
|
2341
|
+
min_tokens: heuristicConfig?.min_tokens,
|
|
2342
|
+
max_sentences: heuristicConfig?.max_sentences,
|
|
2343
|
+
target_reduction: heuristicConfig?.target_reduction,
|
|
2344
|
+
query: resolveHeuristicPromptQuery(heuristicConfig, runtime.variables, sections.system_instructions),
|
|
2345
|
+
json_to_toon: heuristicConfig?.json_to_toon
|
|
2346
|
+
});
|
|
2347
|
+
promptTemplate = result.output;
|
|
2348
|
+
reportHeuristicCompressionWarnings2(warnings, result.warnings, "prompt template");
|
|
2349
|
+
if (result.tokensSaved > 0 || result.output !== sections.prompt_template) {
|
|
2350
|
+
compression.push({
|
|
2351
|
+
provider: "heuristic",
|
|
2352
|
+
model: HEURISTIC_COMPRESSION_MODEL,
|
|
2353
|
+
inputTokens: result.inputTokens,
|
|
2354
|
+
outputTokens: result.outputTokens,
|
|
2355
|
+
tokensSaved: result.tokensSaved,
|
|
2356
|
+
compressionRatio: result.compressionRatio,
|
|
2357
|
+
scope: "prompt_template",
|
|
2358
|
+
outputFormat: result.outputFormat
|
|
2359
|
+
});
|
|
2360
|
+
}
|
|
2361
|
+
}
|
|
2362
|
+
if (hasTheTokenCompanyCompression && hasCodePromptCompaction && promptTemplate) {
|
|
2363
|
+
warnings.push(
|
|
2364
|
+
"POK033: TheTokenCompany compression skipped because compression.code is enabled; code is compacted locally and not text-compressed."
|
|
2365
|
+
);
|
|
2366
|
+
} else if (hasTheTokenCompanyCompression && promptTemplate) {
|
|
2367
|
+
const model = theTokenCompanyConfig?.model ?? THETOKENCOMPANY_DEFAULT_MODEL;
|
|
2368
|
+
const result = await compressWithTheTokenCompany(promptTemplate, {
|
|
2369
|
+
apiKey: runtime.theTokenCompany?.apiKey,
|
|
2370
|
+
baseURL: runtime.theTokenCompany?.baseURL,
|
|
2371
|
+
fetch: runtime.theTokenCompany?.fetch,
|
|
2372
|
+
model,
|
|
2373
|
+
aggressiveness: theTokenCompanyConfig?.aggressiveness
|
|
2374
|
+
});
|
|
2375
|
+
promptTemplate = result.output;
|
|
2376
|
+
compression.push({
|
|
2377
|
+
provider: "thetokencompany",
|
|
2378
|
+
model,
|
|
2379
|
+
inputTokens: result.input_tokens,
|
|
2380
|
+
outputTokens: result.output_tokens,
|
|
2381
|
+
tokensSaved: result.tokens_saved,
|
|
2382
|
+
compressionRatio: result.compression_ratio
|
|
2383
|
+
});
|
|
2384
|
+
}
|
|
1509
2385
|
return {
|
|
1510
2386
|
asset: {
|
|
1511
2387
|
...asset,
|
|
1512
2388
|
sections: {
|
|
1513
2389
|
...asset.sections,
|
|
1514
2390
|
...sections,
|
|
1515
|
-
prompt_template:
|
|
2391
|
+
...promptTemplate !== void 0 ? { prompt_template: promptTemplate } : {}
|
|
1516
2392
|
}
|
|
1517
2393
|
},
|
|
1518
2394
|
runtime: {
|
|
@@ -1520,16 +2396,45 @@ async function applyPromptCompressionForRender(asset, runtime) {
|
|
|
1520
2396
|
variables: {},
|
|
1521
2397
|
strict: false
|
|
1522
2398
|
},
|
|
1523
|
-
compression
|
|
1524
|
-
|
|
1525
|
-
|
|
1526
|
-
|
|
1527
|
-
|
|
1528
|
-
|
|
1529
|
-
|
|
1530
|
-
|
|
2399
|
+
compression,
|
|
2400
|
+
warnings
|
|
2401
|
+
};
|
|
2402
|
+
}
|
|
2403
|
+
function usesPlaceholderCompression(asset) {
|
|
2404
|
+
const hasContextInputCompression = getContextInputs(asset).some(
|
|
2405
|
+
(input) => input.compression?.heuristic?.enabled === true || input.compression?.code?.enabled === true
|
|
2406
|
+
);
|
|
2407
|
+
if (hasContextInputCompression) {
|
|
2408
|
+
return true;
|
|
2409
|
+
}
|
|
2410
|
+
return Boolean(
|
|
2411
|
+
asset.sections.system_instructions && hasCompressionModifier(asset.sections.system_instructions) || asset.sections.prompt_template && hasCompressionModifier(asset.sections.prompt_template)
|
|
2412
|
+
);
|
|
2413
|
+
}
|
|
2414
|
+
function hasCompressionModifier(template) {
|
|
2415
|
+
return /\{\{\s*[a-zA-Z_][a-zA-Z0-9_]*\s*\|\s*(?:compress|toon|compact|code)\s*\}\}/.test(template);
|
|
2416
|
+
}
|
|
2417
|
+
function resolveHeuristicPromptQuery(options, variables, systemInstructions) {
|
|
2418
|
+
if (options?.query !== void 0) {
|
|
2419
|
+
return options.query;
|
|
2420
|
+
}
|
|
2421
|
+
if (options?.query_variable && variables?.[options.query_variable] !== void 0) {
|
|
2422
|
+
return variables[options.query_variable];
|
|
2423
|
+
}
|
|
2424
|
+
return systemInstructions;
|
|
2425
|
+
}
|
|
2426
|
+
function toCodeCompactionOptions2(options) {
|
|
2427
|
+
return {
|
|
2428
|
+
remove_comments: options?.remove_comments,
|
|
2429
|
+
trim_indentation: options?.trim_indentation,
|
|
2430
|
+
collapse_blank_lines: options?.collapse_blank_lines
|
|
1531
2431
|
};
|
|
1532
2432
|
}
|
|
2433
|
+
function reportHeuristicCompressionWarnings2(warnings, compressionWarnings, scope) {
|
|
2434
|
+
for (const warning of compressionWarnings ?? []) {
|
|
2435
|
+
warnings.push(`POK031: ${warning} Scope: ${scope}.`);
|
|
2436
|
+
}
|
|
2437
|
+
}
|
|
1533
2438
|
async function compressWithTheTokenCompany(input, options) {
|
|
1534
2439
|
const apiKey = options.apiKey ?? getEnv("THETOKENCOMPANY_API_KEY") ?? getEnv("TTC_API_KEY");
|
|
1535
2440
|
if (!apiKey) {
|
|
@@ -1617,7 +2522,14 @@ function withPromptInputSupport(adapter) {
|
|
|
1617
2522
|
history: compactHistoryForPrompt(resolvedForRender, runtime)
|
|
1618
2523
|
});
|
|
1619
2524
|
const request = adapter.render(prepared.asset, prepared.runtime);
|
|
1620
|
-
return prepared.compression.length > 0
|
|
2525
|
+
return prepared.compression.length > 0 || prepared.warnings.length > 0 ? {
|
|
2526
|
+
...request,
|
|
2527
|
+
...prepared.compression.length > 0 ? {
|
|
2528
|
+
compression: prepared.compression,
|
|
2529
|
+
compressionSummary: summarizePromptCompression(prepared.compression)
|
|
2530
|
+
} : {},
|
|
2531
|
+
...prepared.warnings.length > 0 ? { warnings: prepared.warnings } : {}
|
|
2532
|
+
} : request;
|
|
1621
2533
|
};
|
|
1622
2534
|
return {
|
|
1623
2535
|
...adapter,
|