@cogitator-ai/evals 0.11.0 → 0.13.0
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 +159 -40
- package/dist/calibrate-judge.d.ts +53 -0
- package/dist/calibrate-judge.d.ts.map +1 -0
- package/dist/calibrate-judge.js +114 -0
- package/dist/calibrate-judge.js.map +1 -0
- package/dist/comparison-stats.d.ts +104 -0
- package/dist/comparison-stats.d.ts.map +1 -0
- package/dist/comparison-stats.js +179 -0
- package/dist/comparison-stats.js.map +1 -0
- package/dist/datasets/dataset.d.ts.map +1 -1
- package/dist/datasets/dataset.js +3 -27
- package/dist/datasets/dataset.js.map +1 -1
- package/dist/eval-comparison.d.ts +30 -10
- package/dist/eval-comparison.d.ts.map +1 -1
- package/dist/eval-comparison.js +73 -103
- package/dist/eval-comparison.js.map +1 -1
- package/dist/eval-suite.d.ts +33 -0
- package/dist/eval-suite.d.ts.map +1 -1
- package/dist/eval-suite.js +81 -2
- package/dist/eval-suite.js.map +1 -1
- package/dist/index.d.ts +21 -5
- package/dist/index.d.ts.map +1 -1
- package/dist/index.js +10 -1
- package/dist/index.js.map +1 -1
- package/dist/metrics/g-eval.d.ts +29 -0
- package/dist/metrics/g-eval.d.ts.map +1 -0
- package/dist/metrics/g-eval.js +89 -0
- package/dist/metrics/g-eval.js.map +1 -0
- package/dist/metrics/index.d.ts +3 -1
- package/dist/metrics/index.d.ts.map +1 -1
- package/dist/metrics/index.js +1 -0
- package/dist/metrics/index.js.map +1 -1
- package/dist/metrics/judge-prompt.d.ts +31 -0
- package/dist/metrics/judge-prompt.d.ts.map +1 -0
- package/dist/metrics/judge-prompt.js +90 -0
- package/dist/metrics/judge-prompt.js.map +1 -0
- package/dist/metrics/llm-judge.d.ts +96 -9
- package/dist/metrics/llm-judge.d.ts.map +1 -1
- package/dist/metrics/llm-judge.js +265 -120
- package/dist/metrics/llm-judge.js.map +1 -1
- package/dist/metrics/types.d.ts +25 -0
- package/dist/metrics/types.d.ts.map +1 -1
- package/dist/metrics/types.js.map +1 -1
- package/dist/options.d.ts +4 -0
- package/dist/options.d.ts.map +1 -0
- package/dist/options.js +9 -0
- package/dist/options.js.map +1 -0
- package/dist/pairwise.d.ts +53 -0
- package/dist/pairwise.d.ts.map +1 -0
- package/dist/pairwise.js +118 -0
- package/dist/pairwise.js.map +1 -0
- package/dist/reporters/comparison.d.ts +24 -0
- package/dist/reporters/comparison.d.ts.map +1 -0
- package/dist/reporters/comparison.js +290 -0
- package/dist/reporters/comparison.js.map +1 -0
- package/dist/reporters/format.d.ts +21 -0
- package/dist/reporters/format.d.ts.map +1 -1
- package/dist/reporters/format.js +31 -0
- package/dist/reporters/format.js.map +1 -1
- package/dist/reporters/github.d.ts.map +1 -1
- package/dist/reporters/github.js +1 -16
- package/dist/reporters/github.js.map +1 -1
- package/dist/reporters/html-format.d.ts +11 -0
- package/dist/reporters/html-format.d.ts.map +1 -0
- package/dist/reporters/html-format.js +111 -0
- package/dist/reporters/html-format.js.map +1 -0
- package/dist/reporters/html.d.ts +12 -0
- package/dist/reporters/html.d.ts.map +1 -0
- package/dist/reporters/html.js +119 -0
- package/dist/reporters/html.js.map +1 -0
- package/dist/reporters/index.d.ts +20 -6
- package/dist/reporters/index.d.ts.map +1 -1
- package/dist/reporters/index.js +15 -4
- package/dist/reporters/index.js.map +1 -1
- package/dist/reporters/markdown.d.ts +15 -0
- package/dist/reporters/markdown.d.ts.map +1 -0
- package/dist/reporters/markdown.js +67 -0
- package/dist/reporters/markdown.js.map +1 -0
- package/dist/runs/compare.d.ts +103 -0
- package/dist/runs/compare.d.ts.map +1 -0
- package/dist/runs/compare.js +208 -0
- package/dist/runs/compare.js.map +1 -0
- package/dist/runs/index.d.ts +9 -0
- package/dist/runs/index.d.ts.map +1 -0
- package/dist/runs/index.js +5 -0
- package/dist/runs/index.js.map +1 -0
- package/dist/runs/json.d.ts +11 -0
- package/dist/runs/json.d.ts.map +1 -0
- package/dist/runs/json.js +58 -0
- package/dist/runs/json.js.map +1 -0
- package/dist/runs/run.d.ts +44 -0
- package/dist/runs/run.d.ts.map +1 -0
- package/dist/runs/run.js +136 -0
- package/dist/runs/run.js.map +1 -0
- package/dist/runs/schema.d.ts +504 -0
- package/dist/runs/schema.d.ts.map +1 -0
- package/dist/runs/schema.js +171 -0
- package/dist/runs/schema.js.map +1 -0
- package/dist/runs/store.d.ts +57 -0
- package/dist/runs/store.d.ts.map +1 -0
- package/dist/runs/store.js +146 -0
- package/dist/runs/store.js.map +1 -0
- package/dist/schema.d.ts +16 -0
- package/dist/schema.d.ts.map +1 -1
- package/dist/schema.js +17 -0
- package/dist/schema.js.map +1 -1
- package/dist/stats/bootstrap.d.ts +22 -0
- package/dist/stats/bootstrap.d.ts.map +1 -0
- package/dist/stats/bootstrap.js +34 -0
- package/dist/stats/bootstrap.js.map +1 -0
- package/dist/stats/correlation.d.ts +11 -0
- package/dist/stats/correlation.d.ts.map +1 -0
- package/dist/stats/correlation.js +42 -0
- package/dist/stats/correlation.js.map +1 -0
- package/dist/stats/effect-size.d.ts +7 -0
- package/dist/stats/effect-size.d.ts.map +1 -0
- package/dist/stats/effect-size.js +17 -0
- package/dist/stats/effect-size.js.map +1 -0
- package/dist/stats/holm.d.ts +7 -0
- package/dist/stats/holm.d.ts.map +1 -0
- package/dist/stats/holm.js +21 -0
- package/dist/stats/holm.js.map +1 -0
- package/dist/stats/index.d.ts +10 -2
- package/dist/stats/index.d.ts.map +1 -1
- package/dist/stats/index.js +7 -1
- package/dist/stats/index.js.map +1 -1
- package/dist/stats/mcnemar.d.ts +9 -0
- package/dist/stats/mcnemar.d.ts.map +1 -1
- package/dist/stats/mcnemar.js +27 -13
- package/dist/stats/mcnemar.js.map +1 -1
- package/dist/stats/normal.d.ts +5 -0
- package/dist/stats/normal.d.ts.map +1 -0
- package/dist/stats/normal.js +19 -0
- package/dist/stats/normal.js.map +1 -0
- package/dist/stats/precision.d.ts +12 -0
- package/dist/stats/precision.d.ts.map +1 -0
- package/dist/stats/precision.js +29 -0
- package/dist/stats/precision.js.map +1 -0
- package/dist/stats/random.d.ts +6 -0
- package/dist/stats/random.d.ts.map +1 -0
- package/dist/stats/random.js +25 -0
- package/dist/stats/random.js.map +1 -0
- package/dist/stats/ranks.d.ts +5 -0
- package/dist/stats/ranks.d.ts.map +1 -0
- package/dist/stats/ranks.js +24 -0
- package/dist/stats/ranks.js.map +1 -0
- package/dist/stats/t-test.d.ts +4 -0
- package/dist/stats/t-test.d.ts.map +1 -1
- package/dist/stats/t-test.js +9 -3
- package/dist/stats/t-test.js.map +1 -1
- package/dist/stats/wilcoxon.d.ts +19 -0
- package/dist/stats/wilcoxon.d.ts.map +1 -0
- package/dist/stats/wilcoxon.js +70 -0
- package/dist/stats/wilcoxon.js.map +1 -0
- package/package.json +1 -1
package/dist/metrics/types.d.ts
CHANGED
|
@@ -39,11 +39,26 @@ export interface Passage {
|
|
|
39
39
|
score?: number;
|
|
40
40
|
source?: string;
|
|
41
41
|
}
|
|
42
|
+
/** The versioned instructions an agent run used, as its `RunResult.prompt` reports them */
|
|
43
|
+
export interface EvalPrompt {
|
|
44
|
+
/** The agent's `id` when it was set explicitly, else its `name` */
|
|
45
|
+
key: string;
|
|
46
|
+
versionId?: string;
|
|
47
|
+
version?: number;
|
|
48
|
+
abTest?: {
|
|
49
|
+
id: string;
|
|
50
|
+
variant: 'control' | 'treatment';
|
|
51
|
+
};
|
|
52
|
+
}
|
|
42
53
|
/** One user turn of a multi-turn case and the answer to it */
|
|
43
54
|
export interface EvalTurnResult {
|
|
44
55
|
input: string;
|
|
45
56
|
output: string;
|
|
46
57
|
toolCalls?: readonly EvalToolCall[];
|
|
58
|
+
/** The trace of the agent run that answered the turn */
|
|
59
|
+
traceId?: string;
|
|
60
|
+
/** The versioned instructions the agent run of the turn used */
|
|
61
|
+
prompt?: EvalPrompt;
|
|
47
62
|
/** Set on the turn that failed, the last one the case ran */
|
|
48
63
|
error?: string;
|
|
49
64
|
}
|
|
@@ -74,6 +89,16 @@ export interface EvalCaseResult {
|
|
|
74
89
|
retrieved?: readonly Passage[];
|
|
75
90
|
/** Each turn of a case with `turns`: `input` first, then the further turns */
|
|
76
91
|
turns?: readonly EvalTurnResult[];
|
|
92
|
+
/**
|
|
93
|
+
* The trace of the agent run, from its `RunResult.trace.traceId`, to find the run in a trace
|
|
94
|
+
* viewer. With `turns`, the last turn's. A `fn` target has none
|
|
95
|
+
*/
|
|
96
|
+
traceId?: string;
|
|
97
|
+
/**
|
|
98
|
+
* The versioned instructions the agent run used, from its `RunResult.prompt`. With `turns`, the
|
|
99
|
+
* last turn's. Set when the Cogitator versions instructions (`cogitator.prompts`)
|
|
100
|
+
*/
|
|
101
|
+
prompt?: EvalPrompt;
|
|
77
102
|
/**
|
|
78
103
|
* Every run of the case when the suite runs several `trials`, each with its own scores. The
|
|
79
104
|
* case result itself is the first of them
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"types.d.ts","sourceRoot":"","sources":["../../src/metrics/types.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,WAAW,CAAC;AAE1C,iEAAiE;AACjE,MAAM,WAAW,WAAW;IAC1B,WAAW,EAAE,MAAM,CAAC;IACpB,YAAY,EAAE,MAAM,CAAC;IACrB,WAAW,EAAE,MAAM,CAAC;IACpB,IAAI,EAAE,MAAM,CAAC;CACd;AAED,MAAM,WAAW,WAAW;IAC1B,IAAI,EAAE,MAAM,CAAC;IACb,KAAK,EAAE,MAAM,CAAC;IACd,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB,QAAQ,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;IACnC;;;OAGG;IACH,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,yFAAyF;IACzF,KAAK,CAAC,EAAE,WAAW,CAAC;CACrB;AAED,iCAAiC;AACjC,MAAM,WAAW,YAAY;IAC3B,EAAE,EAAE,MAAM,CAAC;IACX,IAAI,EAAE,MAAM,CAAC;IACb,SAAS,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;CACpC;AAED,yEAAyE;AACzE,MAAM,WAAW,cAAc;IAC7B,MAAM,EAAE,MAAM,CAAC;IACf,IAAI,EAAE,MAAM,CAAC;IACb,MAAM,CAAC,EAAE,OAAO,CAAC;IACjB,KAAK,CAAC,EAAE,MAAM,CAAC;CAChB;AAED,0FAA0F;AAC1F,MAAM,WAAW,OAAO;IACtB,EAAE,CAAC,EAAE,MAAM,CAAC;IACZ,OAAO,EAAE,MAAM,CAAC;IAChB,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,MAAM,CAAC,EAAE,MAAM,CAAC;CACjB;AAED,8DAA8D;AAC9D,MAAM,WAAW,cAAc;IAC7B,KAAK,EAAE,MAAM,CAAC;IACd,MAAM,EAAE,MAAM,CAAC;IACf,SAAS,CAAC,EAAE,SAAS,YAAY,EAAE,CAAC;IACpC,6DAA6D;IAC7D,KAAK,CAAC,EAAE,MAAM,CAAC;CAChB;AAED,MAAM,WAAW,cAAc;IAC7B,IAAI,EAAE,QAAQ,CAAC;IACf,MAAM,EAAE,MAAM,CAAC;IACf,QAAQ,EAAE,MAAM,CAAC;IACjB,iFAAiF;IACjF,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,wDAAwD;IACxD,QAAQ,CAAC,EAAE,MAAM,CAAC;IAClB;;;OAGG;IACH,KAAK,CAAC,EAAE;QACN,WAAW,EAAE,MAAM,CAAC;QACpB,YAAY,EAAE,MAAM,CAAC;QACrB,WAAW,EAAE,MAAM,CAAC;QACpB,IAAI,EAAE,MAAM,CAAC;QACb,QAAQ,EAAE,MAAM,CAAC;KAClB,CAAC;IACF,iDAAiD;IACjD,SAAS,CAAC,EAAE,SAAS,YAAY,EAAE,CAAC;IACpC,yDAAyD;IACzD,WAAW,CAAC,EAAE,SAAS,cAAc,EAAE,CAAC;IACxC,sFAAsF;IACtF,SAAS,CAAC,EAAE,SAAS,OAAO,EAAE,CAAC;IAC/B,8EAA8E;IAC9E,KAAK,CAAC,EAAE,SAAS,cAAc,EAAE,CAAC;IAClC;;;OAGG;IACH,MAAM,CAAC,EAAE,aAAa,CAAC,cAAc,GAAG;QAAE,MAAM,EAAE,WAAW,EAAE,CAAA;KAAE,CAAC,CAAC;IACnE,2DAA2D;IAC3D,SAAS,CAAC,EAAE,OAAO,CAAC;IACpB,wEAAwE;IACxE,qBAAqB,CAAC,EAAE,OAAO,CAAC;IAChC,qEAAqE;IACrE,MAAM,CAAC,EAAE,OAAO,CAAC;CAClB;AAED,MAAM,MAAM,QAAQ,GAAG,CAAC,CAAC,MAAM,EAAE,cAAc,KAAK,OAAO,CAAC,WAAW,CAAC,CAAC,GAAG;IAC1E,UAAU,EAAE,MAAM,CAAC;CACpB,CAAC;AAEF,MAAM,WAAW,gBAAiB,SAAQ,WAAW;IACnD;;;OAGG;IACH,MAAM,CAAC,EAAE,MAAM,EAAE,CAAC;CACnB;AAED,MAAM,MAAM,mBAAmB,GAAG,CAAC,CAAC,OAAO,EAAE,cAAc,EAAE,KAAK,gBAAgB,CAAC,GAAG;IACpF,UAAU,EAAE,MAAM,CAAC;CACpB,CAAC;AAEF,8EAA8E;AAC9E,MAAM,MAAM,gBAAgB,GAAG,cAAc,GAAG;IAAE,MAAM,EAAE,WAAW,EAAE,CAAA;CAAE,CAAC;AAE1E,2DAA2D;AAC3D,MAAM,WAAW,aAAa;IAC5B;;;OAGG;IACH,IAAI,CAAC,EAAE,MAAM,CAAC;CACf;AAED,yEAAyE;AACzE,wBAAgB,mBAAmB,CAAC,KAAK,EAAE,OAAO,GAAG,MAAM,GAAG,SAAS,CAGtE"}
|
|
1
|
+
{"version":3,"file":"types.d.ts","sourceRoot":"","sources":["../../src/metrics/types.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,WAAW,CAAC;AAE1C,iEAAiE;AACjE,MAAM,WAAW,WAAW;IAC1B,WAAW,EAAE,MAAM,CAAC;IACpB,YAAY,EAAE,MAAM,CAAC;IACrB,WAAW,EAAE,MAAM,CAAC;IACpB,IAAI,EAAE,MAAM,CAAC;CACd;AAED,MAAM,WAAW,WAAW;IAC1B,IAAI,EAAE,MAAM,CAAC;IACb,KAAK,EAAE,MAAM,CAAC;IACd,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB,QAAQ,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;IACnC;;;OAGG;IACH,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,yFAAyF;IACzF,KAAK,CAAC,EAAE,WAAW,CAAC;CACrB;AAED,iCAAiC;AACjC,MAAM,WAAW,YAAY;IAC3B,EAAE,EAAE,MAAM,CAAC;IACX,IAAI,EAAE,MAAM,CAAC;IACb,SAAS,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;CACpC;AAED,yEAAyE;AACzE,MAAM,WAAW,cAAc;IAC7B,MAAM,EAAE,MAAM,CAAC;IACf,IAAI,EAAE,MAAM,CAAC;IACb,MAAM,CAAC,EAAE,OAAO,CAAC;IACjB,KAAK,CAAC,EAAE,MAAM,CAAC;CAChB;AAED,0FAA0F;AAC1F,MAAM,WAAW,OAAO;IACtB,EAAE,CAAC,EAAE,MAAM,CAAC;IACZ,OAAO,EAAE,MAAM,CAAC;IAChB,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,MAAM,CAAC,EAAE,MAAM,CAAC;CACjB;AAED,2FAA2F;AAC3F,MAAM,WAAW,UAAU;IACzB,mEAAmE;IACnE,GAAG,EAAE,MAAM,CAAC;IACZ,SAAS,CAAC,EAAE,MAAM,CAAC;IACnB,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB,MAAM,CAAC,EAAE;QAAE,EAAE,EAAE,MAAM,CAAC;QAAC,OAAO,EAAE,SAAS,GAAG,WAAW,CAAA;KAAE,CAAC;CAC3D;AAED,8DAA8D;AAC9D,MAAM,WAAW,cAAc;IAC7B,KAAK,EAAE,MAAM,CAAC;IACd,MAAM,EAAE,MAAM,CAAC;IACf,SAAS,CAAC,EAAE,SAAS,YAAY,EAAE,CAAC;IACpC,wDAAwD;IACxD,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB,gEAAgE;IAChE,MAAM,CAAC,EAAE,UAAU,CAAC;IACpB,6DAA6D;IAC7D,KAAK,CAAC,EAAE,MAAM,CAAC;CAChB;AAED,MAAM,WAAW,cAAc;IAC7B,IAAI,EAAE,QAAQ,CAAC;IACf,MAAM,EAAE,MAAM,CAAC;IACf,QAAQ,EAAE,MAAM,CAAC;IACjB,iFAAiF;IACjF,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,wDAAwD;IACxD,QAAQ,CAAC,EAAE,MAAM,CAAC;IAClB;;;OAGG;IACH,KAAK,CAAC,EAAE;QACN,WAAW,EAAE,MAAM,CAAC;QACpB,YAAY,EAAE,MAAM,CAAC;QACrB,WAAW,EAAE,MAAM,CAAC;QACpB,IAAI,EAAE,MAAM,CAAC;QACb,QAAQ,EAAE,MAAM,CAAC;KAClB,CAAC;IACF,iDAAiD;IACjD,SAAS,CAAC,EAAE,SAAS,YAAY,EAAE,CAAC;IACpC,yDAAyD;IACzD,WAAW,CAAC,EAAE,SAAS,cAAc,EAAE,CAAC;IACxC,sFAAsF;IACtF,SAAS,CAAC,EAAE,SAAS,OAAO,EAAE,CAAC;IAC/B,8EAA8E;IAC9E,KAAK,CAAC,EAAE,SAAS,cAAc,EAAE,CAAC;IAClC;;;OAGG;IACH,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB;;;OAGG;IACH,MAAM,CAAC,EAAE,UAAU,CAAC;IACpB;;;OAGG;IACH,MAAM,CAAC,EAAE,aAAa,CAAC,cAAc,GAAG;QAAE,MAAM,EAAE,WAAW,EAAE,CAAA;KAAE,CAAC,CAAC;IACnE,2DAA2D;IAC3D,SAAS,CAAC,EAAE,OAAO,CAAC;IACpB,wEAAwE;IACxE,qBAAqB,CAAC,EAAE,OAAO,CAAC;IAChC,qEAAqE;IACrE,MAAM,CAAC,EAAE,OAAO,CAAC;CAClB;AAED,MAAM,MAAM,QAAQ,GAAG,CAAC,CAAC,MAAM,EAAE,cAAc,KAAK,OAAO,CAAC,WAAW,CAAC,CAAC,GAAG;IAC1E,UAAU,EAAE,MAAM,CAAC;CACpB,CAAC;AAEF,MAAM,WAAW,gBAAiB,SAAQ,WAAW;IACnD;;;OAGG;IACH,MAAM,CAAC,EAAE,MAAM,EAAE,CAAC;CACnB;AAED,MAAM,MAAM,mBAAmB,GAAG,CAAC,CAAC,OAAO,EAAE,cAAc,EAAE,KAAK,gBAAgB,CAAC,GAAG;IACpF,UAAU,EAAE,MAAM,CAAC;CACpB,CAAC;AAEF,8EAA8E;AAC9E,MAAM,MAAM,gBAAgB,GAAG,cAAc,GAAG;IAAE,MAAM,EAAE,WAAW,EAAE,CAAA;CAAE,CAAC;AAE1E,2DAA2D;AAC3D,MAAM,WAAW,aAAa;IAC5B;;;OAGG;IACH,IAAI,CAAC,EAAE,MAAM,CAAC;CACf;AAED,yEAAyE;AACzE,wBAAgB,mBAAmB,CAAC,KAAK,EAAE,OAAO,GAAG,MAAM,GAAG,SAAS,CAGtE"}
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"types.js","sourceRoot":"","sources":["../../src/metrics/types.ts"],"names":[],"mappings":"
|
|
1
|
+
{"version":3,"file":"types.js","sourceRoot":"","sources":["../../src/metrics/types.ts"],"names":[],"mappings":"AAmJA,yEAAyE;AACzE,MAAM,UAAU,mBAAmB,CAAC,KAAc;IAChD,IAAI,OAAO,KAAK,KAAK,QAAQ,IAAI,MAAM,CAAC,QAAQ,CAAC,KAAK,CAAC;QAAE,OAAO,SAAS,CAAC;IAC1E,OAAO,SAAS,MAAM,CAAC,KAAK,CAAC,yBAAyB,CAAC;AACzD,CAAC"}
|
|
@@ -0,0 +1,4 @@
|
|
|
1
|
+
import type { z } from 'zod';
|
|
2
|
+
/** Parses options with a schema, throwing one readable error that names what is wrong */
|
|
3
|
+
export declare function parseOptions<Schema extends z.ZodType>(schema: Schema, value: unknown, label: string): z.output<Schema>;
|
|
4
|
+
//# sourceMappingURL=options.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"options.d.ts","sourceRoot":"","sources":["../src/options.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AAE7B,yFAAyF;AACzF,wBAAgB,YAAY,CAAC,MAAM,SAAS,CAAC,CAAC,OAAO,EACnD,MAAM,EAAE,MAAM,EACd,KAAK,EAAE,OAAO,EACd,KAAK,EAAE,MAAM,GACZ,CAAC,CAAC,MAAM,CAAC,MAAM,CAAC,CAOlB"}
|
package/dist/options.js
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
1
|
+
/** Parses options with a schema, throwing one readable error that names what is wrong */
|
|
2
|
+
export function parseOptions(schema, value, label) {
|
|
3
|
+
const parsed = schema.safeParse(value);
|
|
4
|
+
if (parsed.success)
|
|
5
|
+
return parsed.data;
|
|
6
|
+
const issues = parsed.error.issues.map((issue) => issue.path.length > 0 ? `${issue.path.join('.')}: ${issue.message}` : issue.message);
|
|
7
|
+
throw new Error(`${label}: ${issues.join(', ')}`);
|
|
8
|
+
}
|
|
9
|
+
//# sourceMappingURL=options.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"options.js","sourceRoot":"","sources":["../src/options.ts"],"names":[],"mappings":"AAEA,yFAAyF;AACzF,MAAM,UAAU,YAAY,CAC1B,MAAc,EACd,KAAc,EACd,KAAa;IAEb,MAAM,MAAM,GAAG,MAAM,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC;IACvC,IAAI,MAAM,CAAC,OAAO;QAAE,OAAO,MAAM,CAAC,IAAI,CAAC;IACvC,MAAM,MAAM,GAAG,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAC/C,KAAK,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,GAAG,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,KAAK,KAAK,CAAC,OAAO,EAAE,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CACpF,CAAC;IACF,MAAM,IAAI,KAAK,CAAC,GAAG,KAAK,KAAK,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;AACpD,CAAC"}
|
|
@@ -0,0 +1,53 @@
|
|
|
1
|
+
import type { EvalSuiteResult } from './eval-suite.js';
|
|
2
|
+
import type { JudgeContext } from './metrics/llm-judge.js';
|
|
3
|
+
export interface PairwiseOptions {
|
|
4
|
+
/** What makes one answer better than the other. Default: correct, relevant, complete, clear */
|
|
5
|
+
criteria?: string;
|
|
6
|
+
/** The name the result reports. Default `pairwise` */
|
|
7
|
+
name?: string;
|
|
8
|
+
}
|
|
9
|
+
export type PairwiseSide = 'baseline' | 'challenger';
|
|
10
|
+
export type PairwiseOutcome = PairwiseSide | 'tie';
|
|
11
|
+
/** One of the two verdicts on a case */
|
|
12
|
+
export interface PairwiseVerdict {
|
|
13
|
+
/** The side shown to the judge first, as answer A */
|
|
14
|
+
first: PairwiseSide;
|
|
15
|
+
/** The side the judge preferred, or `tie` */
|
|
16
|
+
preferred: PairwiseOutcome;
|
|
17
|
+
reasoning?: string;
|
|
18
|
+
}
|
|
19
|
+
export interface PairwiseCase {
|
|
20
|
+
/** Position of the case in the dataset */
|
|
21
|
+
index: number;
|
|
22
|
+
input: string;
|
|
23
|
+
winner: PairwiseOutcome;
|
|
24
|
+
/** The verdict with the baseline first, then the one with the challenger first */
|
|
25
|
+
verdicts: PairwiseVerdict[];
|
|
26
|
+
/** The side that errored and so lost the case without a judge call */
|
|
27
|
+
errored?: PairwiseSide;
|
|
28
|
+
/** Why the judge gave no verdict in one of the orders, which makes the case a tie */
|
|
29
|
+
error?: string;
|
|
30
|
+
}
|
|
31
|
+
export interface PairwiseResult {
|
|
32
|
+
name: string;
|
|
33
|
+
wins: {
|
|
34
|
+
baseline: number;
|
|
35
|
+
challenger: number;
|
|
36
|
+
};
|
|
37
|
+
ties: number;
|
|
38
|
+
/** Cases compared: those at least one side answered */
|
|
39
|
+
cases: number;
|
|
40
|
+
/** Challenger wins over the decided cases (wins of either side), NaN when none was decided */
|
|
41
|
+
winRate: number;
|
|
42
|
+
/** Cases the judge failed on in at least one order, counted as ties */
|
|
43
|
+
judgeErrors: number;
|
|
44
|
+
/** USD the judge calls cost, as far as the judge reported usage */
|
|
45
|
+
cost: number;
|
|
46
|
+
perCase: PairwiseCase[];
|
|
47
|
+
}
|
|
48
|
+
/**
|
|
49
|
+
* Asks the judge which side answered each case better, see `EvalComparisonOptions.pairwise`.
|
|
50
|
+
* Cases both sides errored on are left out
|
|
51
|
+
*/
|
|
52
|
+
export declare function judgePairwise(baseline: EvalSuiteResult, challenger: EvalSuiteResult, judge: JudgeContext, options: PairwiseOptions, concurrency: number, signal?: AbortSignal): Promise<PairwiseResult>;
|
|
53
|
+
//# sourceMappingURL=pairwise.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"pairwise.d.ts","sourceRoot":"","sources":["../src/pairwise.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,cAAc,CAAC;AAEpD,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,qBAAqB,CAAC;AAUxD,MAAM,WAAW,eAAe;IAC9B,+FAA+F;IAC/F,QAAQ,CAAC,EAAE,MAAM,CAAC;IAClB,sDAAsD;IACtD,IAAI,CAAC,EAAE,MAAM,CAAC;CACf;AAED,MAAM,MAAM,YAAY,GAAG,UAAU,GAAG,YAAY,CAAC;AAErD,MAAM,MAAM,eAAe,GAAG,YAAY,GAAG,KAAK,CAAC;AAEnD,wCAAwC;AACxC,MAAM,WAAW,eAAe;IAC9B,qDAAqD;IACrD,KAAK,EAAE,YAAY,CAAC;IACpB,6CAA6C;IAC7C,SAAS,EAAE,eAAe,CAAC;IAC3B,SAAS,CAAC,EAAE,MAAM,CAAC;CACpB;AAED,MAAM,WAAW,YAAY;IAC3B,0CAA0C;IAC1C,KAAK,EAAE,MAAM,CAAC;IACd,KAAK,EAAE,MAAM,CAAC;IACd,MAAM,EAAE,eAAe,CAAC;IACxB,kFAAkF;IAClF,QAAQ,EAAE,eAAe,EAAE,CAAC;IAC5B,sEAAsE;IACtE,OAAO,CAAC,EAAE,YAAY,CAAC;IACvB,qFAAqF;IACrF,KAAK,CAAC,EAAE,MAAM,CAAC;CAChB;AAED,MAAM,WAAW,cAAc;IAC7B,IAAI,EAAE,MAAM,CAAC;IACb,IAAI,EAAE;QAAE,QAAQ,EAAE,MAAM,CAAC;QAAC,UAAU,EAAE,MAAM,CAAA;KAAE,CAAC;IAC/C,IAAI,EAAE,MAAM,CAAC;IACb,uDAAuD;IACvD,KAAK,EAAE,MAAM,CAAC;IACd,8FAA8F;IAC9F,OAAO,EAAE,MAAM,CAAC;IAChB,uEAAuE;IACvE,WAAW,EAAE,MAAM,CAAC;IACpB,mEAAmE;IACnE,IAAI,EAAE,MAAM,CAAC;IACb,OAAO,EAAE,YAAY,EAAE,CAAC;CACzB;AA4GD;;;GAGG;AACH,wBAAsB,aAAa,CACjC,QAAQ,EAAE,eAAe,EACzB,UAAU,EAAE,eAAe,EAC3B,KAAK,EAAE,YAAY,EACnB,OAAO,EAAE,eAAe,EACxB,WAAW,EAAE,MAAM,EACnB,MAAM,CAAC,EAAE,WAAW,GACnB,OAAO,CAAC,cAAc,CAAC,CAmCzB"}
|
package/dist/pairwise.js
ADDED
|
@@ -0,0 +1,118 @@
|
|
|
1
|
+
import { judgePrompt, parseJudgeOutput, sumUsage, untilAborted, usageOf, } from './metrics/llm-judge.js';
|
|
2
|
+
import { conversationBlock, evidenceBlock, tagged, wrapped } from './metrics/judge-prompt.js';
|
|
3
|
+
/** The instructions of a pairwise verdict */
|
|
4
|
+
function pairwiseInstructions(criteria) {
|
|
5
|
+
const basis = criteria
|
|
6
|
+
? `on these criteria:\n${criteria}`
|
|
7
|
+
: 'overall: which is more correct, relevant, complete and clear';
|
|
8
|
+
return `You are comparing two answers to the same request to decide which one is better ${basis}\n\nThe order of the answers says nothing about their quality, and a longer answer is better only for what its length adds.\n\nScore 0 when <answer_a> is better, 1 when <answer_b> is better, 0.5 when neither is better.\n\nRespond with JSON: {"score": <0, 0.5 or 1>, "reasoning": "<explanation>"}`;
|
|
9
|
+
}
|
|
10
|
+
/** The case and the two answers, A then B. A multi-turn case shows each side's whole conversation */
|
|
11
|
+
function pairwiseMessage(a, b) {
|
|
12
|
+
const evidence = evidenceBlock(a, false);
|
|
13
|
+
const multiTurn = (a.case.turns?.length ?? 0) > 0;
|
|
14
|
+
const shared = [
|
|
15
|
+
...(multiTurn
|
|
16
|
+
? ['Each answer is a whole conversation, the user messages being the same in both.']
|
|
17
|
+
: [tagged('input', a.case.input)]),
|
|
18
|
+
...(evidence === undefined ? [] : [evidence]),
|
|
19
|
+
...(a.case.expected ? [tagged('expected', a.case.expected)] : []),
|
|
20
|
+
];
|
|
21
|
+
const answer = (tag, result) => multiTurn ? wrapped(tag, conversationBlock(result)) : tagged(tag, result.output);
|
|
22
|
+
return [...shared, answer('answer_a', a), answer('answer_b', b)].join('\n');
|
|
23
|
+
}
|
|
24
|
+
const other = (side) => side === 'baseline' ? 'challenger' : 'baseline';
|
|
25
|
+
async function ask(judge, instructions, first, a, b, signal) {
|
|
26
|
+
try {
|
|
27
|
+
const answered = await untilAborted(judge.cogitator.run({
|
|
28
|
+
input: judgePrompt(instructions, pairwiseMessage(a, b)),
|
|
29
|
+
...(signal && { signal }),
|
|
30
|
+
}), signal);
|
|
31
|
+
const usage = usageOf(answered.usage);
|
|
32
|
+
const verdict = parseJudgeOutput(answered.output);
|
|
33
|
+
if (!verdict)
|
|
34
|
+
return { error: 'could not parse judge response', ...(usage && { usage }) };
|
|
35
|
+
const score = verdict.outOf === undefined ? verdict.score : verdict.score / verdict.outOf;
|
|
36
|
+
const preferred = score < 0.5 ? first : score > 0.5 ? other(first) : 'tie';
|
|
37
|
+
return {
|
|
38
|
+
verdict: {
|
|
39
|
+
first,
|
|
40
|
+
preferred,
|
|
41
|
+
...(verdict.reasoning !== undefined && { reasoning: verdict.reasoning }),
|
|
42
|
+
},
|
|
43
|
+
...(usage && { usage }),
|
|
44
|
+
};
|
|
45
|
+
}
|
|
46
|
+
catch (error) {
|
|
47
|
+
return { error: error instanceof Error ? error.message : String(error) };
|
|
48
|
+
}
|
|
49
|
+
}
|
|
50
|
+
/**
|
|
51
|
+
* A case judged both ways round: the baseline first, then the challenger first. Agreeing
|
|
52
|
+
* verdicts give the case to that side, disagreeing ones or a tie make it a tie
|
|
53
|
+
*/
|
|
54
|
+
async function judgeCase(judge, instructions, index, baseline, challenger, signal) {
|
|
55
|
+
const input = baseline.case.input;
|
|
56
|
+
if (baseline.error !== undefined || challenger.error !== undefined) {
|
|
57
|
+
const errored = baseline.error !== undefined ? 'baseline' : 'challenger';
|
|
58
|
+
return { result: { index, input, winner: other(errored), verdicts: [], errored } };
|
|
59
|
+
}
|
|
60
|
+
const asked = await Promise.all([
|
|
61
|
+
ask(judge, instructions, 'baseline', baseline, challenger, signal),
|
|
62
|
+
ask(judge, instructions, 'challenger', challenger, baseline, signal),
|
|
63
|
+
]);
|
|
64
|
+
const usage = sumUsage(asked.map((one) => one.usage));
|
|
65
|
+
const verdicts = asked.flatMap((one) => (one.verdict ? [one.verdict] : []));
|
|
66
|
+
const errors = asked.flatMap((one) => (one.error === undefined ? [] : [one.error]));
|
|
67
|
+
const [first, second] = verdicts;
|
|
68
|
+
const winner = errors.length === 0 && first.preferred === second.preferred ? first.preferred : 'tie';
|
|
69
|
+
return {
|
|
70
|
+
result: {
|
|
71
|
+
index,
|
|
72
|
+
input,
|
|
73
|
+
winner,
|
|
74
|
+
verdicts,
|
|
75
|
+
...(errors.length > 0 && { error: errors.join(', ') }),
|
|
76
|
+
},
|
|
77
|
+
...(usage && { usage }),
|
|
78
|
+
};
|
|
79
|
+
}
|
|
80
|
+
/**
|
|
81
|
+
* Asks the judge which side answered each case better, see `EvalComparisonOptions.pairwise`.
|
|
82
|
+
* Cases both sides errored on are left out
|
|
83
|
+
*/
|
|
84
|
+
export async function judgePairwise(baseline, challenger, judge, options, concurrency, signal) {
|
|
85
|
+
const instructions = pairwiseInstructions(options.criteria);
|
|
86
|
+
const pairs = baseline.results.flatMap((b, index) => {
|
|
87
|
+
const c = challenger.results[index];
|
|
88
|
+
return c && (b.error === undefined || c.error === undefined) ? [{ index, b, c }] : [];
|
|
89
|
+
});
|
|
90
|
+
const judged = new Array(pairs.length);
|
|
91
|
+
let next = 0;
|
|
92
|
+
const worker = async () => {
|
|
93
|
+
while (next < pairs.length && !signal?.aborted) {
|
|
94
|
+
const at = next++;
|
|
95
|
+
const { index, b, c } = pairs[at];
|
|
96
|
+
judged[at] = await judgeCase(judge, instructions, index, b, c, signal);
|
|
97
|
+
}
|
|
98
|
+
};
|
|
99
|
+
await Promise.all(Array.from({ length: Math.min(concurrency, pairs.length) }, worker));
|
|
100
|
+
signal?.throwIfAborted();
|
|
101
|
+
const perCase = judged.map((one) => one.result);
|
|
102
|
+
const wins = {
|
|
103
|
+
baseline: perCase.filter((one) => one.winner === 'baseline').length,
|
|
104
|
+
challenger: perCase.filter((one) => one.winner === 'challenger').length,
|
|
105
|
+
};
|
|
106
|
+
const decided = wins.baseline + wins.challenger;
|
|
107
|
+
return {
|
|
108
|
+
name: options.name ?? 'pairwise',
|
|
109
|
+
wins,
|
|
110
|
+
ties: perCase.length - decided,
|
|
111
|
+
cases: perCase.length,
|
|
112
|
+
winRate: decided === 0 ? Number.NaN : wins.challenger / decided,
|
|
113
|
+
judgeErrors: perCase.filter((one) => one.error !== undefined).length,
|
|
114
|
+
cost: judged.reduce((sum, one) => sum + (one.usage?.cost ?? 0), 0),
|
|
115
|
+
perCase,
|
|
116
|
+
};
|
|
117
|
+
}
|
|
118
|
+
//# sourceMappingURL=pairwise.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"pairwise.js","sourceRoot":"","sources":["../src/pairwise.ts"],"names":[],"mappings":"AAGA,OAAO,EACL,WAAW,EACX,gBAAgB,EAChB,QAAQ,EACR,YAAY,EACZ,OAAO,GACR,MAAM,qBAAqB,CAAC;AAC7B,OAAO,EAAE,iBAAiB,EAAE,aAAa,EAAE,MAAM,EAAE,OAAO,EAAE,MAAM,wBAAwB,CAAC;AAkD3F,6CAA6C;AAC7C,SAAS,oBAAoB,CAAC,QAA4B;IACxD,MAAM,KAAK,GAAG,QAAQ;QACpB,CAAC,CAAC,uBAAuB,QAAQ,EAAE;QACnC,CAAC,CAAC,8DAA8D,CAAC;IACnE,OAAO,mFAAmF,KAAK,ySAAyS,CAAC;AAC3Y,CAAC;AAED,qGAAqG;AACrG,SAAS,eAAe,CAAC,CAAmB,EAAE,CAAmB;IAC/D,MAAM,QAAQ,GAAG,aAAa,CAAC,CAAC,EAAE,KAAK,CAAC,CAAC;IACzC,MAAM,SAAS,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,EAAE,MAAM,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC;IAClD,MAAM,MAAM,GAAG;QACb,GAAG,CAAC,SAAS;YACX,CAAC,CAAC,CAAC,gFAAgF,CAAC;YACpF,CAAC,CAAC,CAAC,MAAM,CAAC,OAAO,EAAE,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC;QACpC,GAAG,CAAC,QAAQ,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC;QAC7C,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,UAAU,EAAE,CAAC,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;KAClE,CAAC;IACF,MAAM,MAAM,GAAG,CAAC,GAAW,EAAE,MAAwB,EAAE,EAAE,CACvD,SAAS,CAAC,CAAC,CAAC,OAAO,CAAC,GAAG,EAAE,iBAAiB,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,GAAG,EAAE,MAAM,CAAC,MAAM,CAAC,CAAC;IACnF,OAAO,CAAC,GAAG,MAAM,EAAE,MAAM,CAAC,UAAU,EAAE,CAAC,CAAC,EAAE,MAAM,CAAC,UAAU,EAAE,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;AAC9E,CAAC;AAQD,MAAM,KAAK,GAAG,CAAC,IAAkB,EAAgB,EAAE,CACjD,IAAI,KAAK,UAAU,CAAC,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,UAAU,CAAC;AAElD,KAAK,UAAU,GAAG,CAChB,KAAmB,EACnB,YAAoB,EACpB,KAAmB,EACnB,CAAmB,EACnB,CAAmB,EACnB,MAA+B;IAE/B,IAAI,CAAC;QACH,MAAM,QAAQ,GAAG,MAAM,YAAY,CACjC,KAAK,CAAC,SAAS,CAAC,GAAG,CAAC;YAClB,KAAK,EAAE,WAAW,CAAC,YAAY,EAAE,eAAe,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC;YACvD,GAAG,CAAC,MAAM,IAAI,EAAE,MAAM,EAAE,CAAC;SAC1B,CAAC,EACF,MAAM,CACP,CAAC;QACF,MAAM,KAAK,GAAG,OAAO,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC;QACtC,MAAM,OAAO,GAAG,gBAAgB,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC;QAClD,IAAI,CAAC,OAAO;YAAE,OAAO,EAAE,KAAK,EAAE,gCAAgC,EAAE,GAAG,CAAC,KAAK,IAAI,EAAE,KAAK,EAAE,CAAC,EAAE,CAAC;QAC1F,MAAM,KAAK,GAAG,OAAO,CAAC,KAAK,KAAK,SAAS,CAAC,CAAC,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC,CAAC,OAAO,CAAC,KAAK,GAAG,OAAO,CAAC,KAAK,CAAC;QAC1F,MAAM,SAAS,GAAoB,KAAK,GAAG,GAAG,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,KAAK,GAAG,GAAG,CAAC,CAAC,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC;QAC5F,OAAO;YACL,OAAO,EAAE;gBACP,KAAK;gBACL,SAAS;gBACT,GAAG,CAAC,OAAO,CAAC,SAAS,KAAK,SAAS,IAAI,EAAE,SAAS,EAAE,OAAO,CAAC,SAAS,EAAE,CAAC;aACzE;YACD,GAAG,CAAC,KAAK,IAAI,EAAE,KAAK,EAAE,CAAC;SACxB,CAAC;IACJ,CAAC;IAAC,OAAO,KAAK,EAAE,CAAC;QACf,OAAO,EAAE,KAAK,EAAE,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,EAAE,CAAC;IAC3E,CAAC;AACH,CAAC;AAED;;;GAGG;AACH,KAAK,UAAU,SAAS,CACtB,KAAmB,EACnB,YAAoB,EACpB,KAAa,EACb,QAA0B,EAC1B,UAA4B,EAC5B,MAA+B;IAE/B,MAAM,KAAK,GAAG,QAAQ,CAAC,IAAI,CAAC,KAAK,CAAC;IAClC,IAAI,QAAQ,CAAC,KAAK,KAAK,SAAS,IAAI,UAAU,CAAC,KAAK,KAAK,SAAS,EAAE,CAAC;QACnE,MAAM,OAAO,GAAiB,QAAQ,CAAC,KAAK,KAAK,SAAS,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,YAAY,CAAC;QACvF,OAAO,EAAE,MAAM,EAAE,EAAE,KAAK,EAAE,KAAK,EAAE,MAAM,EAAE,KAAK,CAAC,OAAO,CAAC,EAAE,QAAQ,EAAE,EAAE,EAAE,OAAO,EAAE,EAAE,CAAC;IACrF,CAAC;IACD,MAAM,KAAK,GAAG,MAAM,OAAO,CAAC,GAAG,CAAC;QAC9B,GAAG,CAAC,KAAK,EAAE,YAAY,EAAE,UAAU,EAAE,QAAQ,EAAE,UAAU,EAAE,MAAM,CAAC;QAClE,GAAG,CAAC,KAAK,EAAE,YAAY,EAAE,YAAY,EAAE,UAAU,EAAE,QAAQ,EAAE,MAAM,CAAC;KACrE,CAAC,CAAC;IACH,MAAM,KAAK,GAAG,QAAQ,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC,CAAC;IACtD,MAAM,QAAQ,GAAG,KAAK,CAAC,OAAO,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC;IAC5E,MAAM,MAAM,GAAG,KAAK,CAAC,OAAO,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,CAAC,GAAG,CAAC,KAAK,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;IACpF,MAAM,CAAC,KAAK,EAAE,MAAM,CAAC,GAAG,QAAQ,CAAC;IACjC,MAAM,MAAM,GACV,MAAM,CAAC,MAAM,KAAK,CAAC,IAAI,KAAK,CAAC,SAAS,KAAK,MAAM,CAAC,SAAS,CAAC,CAAC,CAAC,KAAK,CAAC,SAAS,CAAC,CAAC,CAAC,KAAK,CAAC;IACxF,OAAO;QACL,MAAM,EAAE;YACN,KAAK;YACL,KAAK;YACL,MAAM;YACN,QAAQ;YACR,GAAG,CAAC,MAAM,CAAC,MAAM,GAAG,CAAC,IAAI,EAAE,KAAK,EAAE,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;SACvD;QACD,GAAG,CAAC,KAAK,IAAI,EAAE,KAAK,EAAE,CAAC;KACxB,CAAC;AACJ,CAAC;AAED;;;GAGG;AACH,MAAM,CAAC,KAAK,UAAU,aAAa,CACjC,QAAyB,EACzB,UAA2B,EAC3B,KAAmB,EACnB,OAAwB,EACxB,WAAmB,EACnB,MAAoB;IAEpB,MAAM,YAAY,GAAG,oBAAoB,CAAC,OAAO,CAAC,QAAQ,CAAC,CAAC;IAC5D,MAAM,KAAK,GAAG,QAAQ,CAAC,OAAO,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,KAAK,EAAE,EAAE;QAClD,MAAM,CAAC,GAAG,UAAU,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC;QACpC,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,KAAK,KAAK,SAAS,IAAI,CAAC,CAAC,KAAK,KAAK,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,KAAK,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;IACxF,CAAC,CAAC,CAAC;IAEH,MAAM,MAAM,GAAG,IAAI,KAAK,CAAgD,KAAK,CAAC,MAAM,CAAC,CAAC;IACtF,IAAI,IAAI,GAAG,CAAC,CAAC;IACb,MAAM,MAAM,GAAG,KAAK,IAAI,EAAE;QACxB,OAAO,IAAI,GAAG,KAAK,CAAC,MAAM,IAAI,CAAC,MAAM,EAAE,OAAO,EAAE,CAAC;YAC/C,MAAM,EAAE,GAAG,IAAI,EAAE,CAAC;YAClB,MAAM,EAAE,KAAK,EAAE,CAAC,EAAE,CAAC,EAAE,GAAG,KAAK,CAAC,EAAE,CAAC,CAAC;YAClC,MAAM,CAAC,EAAE,CAAC,GAAG,MAAM,SAAS,CAAC,KAAK,EAAE,YAAY,EAAE,KAAK,EAAE,CAAC,EAAE,CAAC,EAAE,MAAM,CAAC,CAAC;QACzE,CAAC;IACH,CAAC,CAAC;IACF,MAAM,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,CAAC,EAAE,MAAM,EAAE,IAAI,CAAC,GAAG,CAAC,WAAW,EAAE,KAAK,CAAC,MAAM,CAAC,EAAE,EAAE,MAAM,CAAC,CAAC,CAAC;IACvF,MAAM,EAAE,cAAc,EAAE,CAAC;IAEzB,MAAM,OAAO,GAAG,MAAM,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC;IAChD,MAAM,IAAI,GAAG;QACX,QAAQ,EAAE,OAAO,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,GAAG,CAAC,MAAM,KAAK,UAAU,CAAC,CAAC,MAAM;QACnE,UAAU,EAAE,OAAO,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,GAAG,CAAC,MAAM,KAAK,YAAY,CAAC,CAAC,MAAM;KACxE,CAAC;IACF,MAAM,OAAO,GAAG,IAAI,CAAC,QAAQ,GAAG,IAAI,CAAC,UAAU,CAAC;IAChD,OAAO;QACL,IAAI,EAAE,OAAO,CAAC,IAAI,IAAI,UAAU;QAChC,IAAI;QACJ,IAAI,EAAE,OAAO,CAAC,MAAM,GAAG,OAAO;QAC9B,KAAK,EAAE,OAAO,CAAC,MAAM;QACrB,OAAO,EAAE,OAAO,KAAK,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,UAAU,GAAG,OAAO;QAC/D,WAAW,EAAE,OAAO,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,GAAG,CAAC,KAAK,KAAK,SAAS,CAAC,CAAC,MAAM;QACpE,IAAI,EAAE,MAAM,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,GAAG,EAAE,EAAE,CAAC,GAAG,GAAG,CAAC,GAAG,CAAC,KAAK,EAAE,IAAI,IAAI,CAAC,CAAC,EAAE,CAAC,CAAC;QAClE,OAAO;KACR,CAAC;AACJ,CAAC"}
|
|
@@ -0,0 +1,24 @@
|
|
|
1
|
+
import type { ComparisonResult } from '../eval-comparison.js';
|
|
2
|
+
import type { RunComparison } from '../runs/compare.js';
|
|
3
|
+
import type { ReporterOptions } from './index.js';
|
|
4
|
+
/** What `reportComparison()` renders: an `EvalComparison` result or a `compareRuns()` result */
|
|
5
|
+
export type ReportableComparison = ComparisonResult | RunComparison;
|
|
6
|
+
export type ComparisonReporterType = 'markdown' | 'html' | 'console';
|
|
7
|
+
/**
|
|
8
|
+
* A compact Markdown summary of a comparison for a pull request comment: the winner, a table of
|
|
9
|
+
* the metrics with means, difference, interval, p-values and significance, the cases that
|
|
10
|
+
* regressed and improved, and the warnings. Every value is escaped
|
|
11
|
+
*/
|
|
12
|
+
export declare function comparisonMarkdown(comparison: ReportableComparison): string;
|
|
13
|
+
/**
|
|
14
|
+
* A self-contained HTML page of a comparison: inline CSS for light and dark color schemes, no
|
|
15
|
+
* scripts, no external assets, and every value escaped
|
|
16
|
+
*/
|
|
17
|
+
export declare function comparisonHtml(comparison: ReportableComparison): string;
|
|
18
|
+
/**
|
|
19
|
+
* Renders an `EvalComparison` result or a `compareRuns()` result: `console` prints it, `markdown`
|
|
20
|
+
* writes a compact summary for a pull request comment to `<path>.md`, and `html` a
|
|
21
|
+
* self-contained page to `<path>.html`. The default path is `eval-comparison`
|
|
22
|
+
*/
|
|
23
|
+
export declare function reportComparison(comparison: ReportableComparison, type: ComparisonReporterType | ComparisonReporterType[], options?: ReporterOptions): void;
|
|
24
|
+
//# sourceMappingURL=comparison.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"comparison.d.ts","sourceRoot":"","sources":["../../src/reporters/comparison.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,oBAAoB,CAAC;AAC3D,OAAO,KAAK,EAAE,aAAa,EAAU,MAAM,iBAAiB,CAAC;AAK7D,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,SAAS,CAAC;AAG/C,gGAAgG;AAChG,MAAM,MAAM,oBAAoB,GAAG,gBAAgB,GAAG,aAAa,CAAC;AAEpE,MAAM,MAAM,sBAAsB,GAAG,UAAU,GAAG,MAAM,GAAG,SAAS,CAAC;AAsJrE;;;;GAIG;AACH,wBAAgB,kBAAkB,CAAC,UAAU,EAAE,oBAAoB,GAAG,MAAM,CAqC3E;AAgBD;;;GAGG;AACH,wBAAgB,cAAc,CAAC,UAAU,EAAE,oBAAoB,GAAG,MAAM,CAgFvE;AAoED;;;;GAIG;AACH,wBAAgB,gBAAgB,CAC9B,UAAU,EAAE,oBAAoB,EAChC,IAAI,EAAE,sBAAsB,GAAG,sBAAsB,EAAE,EACvD,OAAO,CAAC,EAAE,eAAe,GACxB,IAAI,CAuBN"}
|
|
@@ -0,0 +1,290 @@
|
|
|
1
|
+
import { writeFileSync } from 'node:fs';
|
|
2
|
+
import { caseChanges } from '../comparison-stats.js';
|
|
3
|
+
import { cell, num, truncateInput } from './format.js';
|
|
4
|
+
import { prepare, reportPath } from './index.js';
|
|
5
|
+
import { escapeHtml, htmlDocument, htmlTable, htmlTiles } from './html-format.js';
|
|
6
|
+
/** Most case changes the Markdown summary lists per metric and direction */
|
|
7
|
+
const MARKDOWN_MAX_CHANGES = 10;
|
|
8
|
+
function isRunComparison(comparison) {
|
|
9
|
+
return !('summary' in comparison);
|
|
10
|
+
}
|
|
11
|
+
function runLabel(ref) {
|
|
12
|
+
return ref.name === undefined ? ref.runId : `${ref.name} (${ref.runId})`;
|
|
13
|
+
}
|
|
14
|
+
function viewOf(comparison) {
|
|
15
|
+
if (isRunComparison(comparison)) {
|
|
16
|
+
return {
|
|
17
|
+
title: 'Run comparison',
|
|
18
|
+
before: 'base',
|
|
19
|
+
after: 'head',
|
|
20
|
+
subtitle: `base ${runLabel(comparison.base)}, head ${runLabel(comparison.head)}`,
|
|
21
|
+
winner: comparison.winner,
|
|
22
|
+
metrics: Object.entries(comparison.metrics).map(([name, m]) => ({
|
|
23
|
+
name,
|
|
24
|
+
before: m.base,
|
|
25
|
+
after: m.head,
|
|
26
|
+
n: m.n,
|
|
27
|
+
difference: m.difference,
|
|
28
|
+
ci: m.ci,
|
|
29
|
+
test: m.test,
|
|
30
|
+
pValue: m.pValue,
|
|
31
|
+
adjustedPValue: m.adjustedPValue,
|
|
32
|
+
significant: m.significant,
|
|
33
|
+
winner: m.winner,
|
|
34
|
+
improved: m.improved.map((c) => ({ ...c, before: c.base, after: c.head })),
|
|
35
|
+
regressed: m.regressed.map((c) => ({ ...c, before: c.base, after: c.head })),
|
|
36
|
+
})),
|
|
37
|
+
cases: {
|
|
38
|
+
matched: comparison.matched,
|
|
39
|
+
added: comparison.added.map((c) => c.input),
|
|
40
|
+
removed: comparison.removed.map((c) => c.input),
|
|
41
|
+
},
|
|
42
|
+
warnings: comparison.warnings,
|
|
43
|
+
};
|
|
44
|
+
}
|
|
45
|
+
const { baseline, challenger, summary } = comparison;
|
|
46
|
+
const pairs = baseline.results.map((r, at) => [r, challenger.results[at]]);
|
|
47
|
+
const inputOf = (index) => baseline.results[index].case.input;
|
|
48
|
+
return {
|
|
49
|
+
title: 'A/B comparison',
|
|
50
|
+
before: 'baseline',
|
|
51
|
+
after: 'challenger',
|
|
52
|
+
winner: summary.winner,
|
|
53
|
+
metrics: Object.entries(summary.metrics).map(([name, m]) => ({
|
|
54
|
+
name,
|
|
55
|
+
before: m.baseline,
|
|
56
|
+
after: m.challenger,
|
|
57
|
+
n: m.n,
|
|
58
|
+
difference: m.difference,
|
|
59
|
+
ci: m.ci,
|
|
60
|
+
test: m.test,
|
|
61
|
+
pValue: m.pValue,
|
|
62
|
+
adjustedPValue: m.adjustedPValue,
|
|
63
|
+
significant: m.significant,
|
|
64
|
+
winner: m.winner,
|
|
65
|
+
...caseChanges(pairs, name, inputOf),
|
|
66
|
+
})),
|
|
67
|
+
warnings: [],
|
|
68
|
+
};
|
|
69
|
+
}
|
|
70
|
+
function signed(n) {
|
|
71
|
+
return n > 0 ? `+${num(n)}` : num(n);
|
|
72
|
+
}
|
|
73
|
+
function ciText(ci) {
|
|
74
|
+
return `[${num(ci.low)}, ${num(ci.high)}]`;
|
|
75
|
+
}
|
|
76
|
+
function pText(p) {
|
|
77
|
+
return p < 0.0001 ? '<0.0001' : num(p);
|
|
78
|
+
}
|
|
79
|
+
function changeLines(view, kind, heading) {
|
|
80
|
+
const rows = view.metrics.flatMap((m) => {
|
|
81
|
+
const changes = m[kind];
|
|
82
|
+
const shown = changes
|
|
83
|
+
.slice(0, MARKDOWN_MAX_CHANGES)
|
|
84
|
+
.map((c) => `| ${cell(m.name)} | ${cell(c.input)} | ${num(c.before)} | ${num(c.after)} | ${signed(c.delta)} |`);
|
|
85
|
+
const more = changes.length > MARKDOWN_MAX_CHANGES
|
|
86
|
+
? [`| ${cell(m.name)} | and ${changes.length - MARKDOWN_MAX_CHANGES} more | | | |`]
|
|
87
|
+
: [];
|
|
88
|
+
return [...shown, ...more];
|
|
89
|
+
});
|
|
90
|
+
if (rows.length === 0)
|
|
91
|
+
return [];
|
|
92
|
+
return [
|
|
93
|
+
`**${heading}**`,
|
|
94
|
+
'',
|
|
95
|
+
`| Metric | Case | ${view.before} | ${view.after} | Change |`,
|
|
96
|
+
'| --- | --- | ---: | ---: | ---: |',
|
|
97
|
+
...rows,
|
|
98
|
+
'',
|
|
99
|
+
];
|
|
100
|
+
}
|
|
101
|
+
/**
|
|
102
|
+
* A compact Markdown summary of a comparison for a pull request comment: the winner, a table of
|
|
103
|
+
* the metrics with means, difference, interval, p-values and significance, the cases that
|
|
104
|
+
* regressed and improved, and the warnings. Every value is escaped
|
|
105
|
+
*/
|
|
106
|
+
export function comparisonMarkdown(comparison) {
|
|
107
|
+
const view = viewOf(comparison);
|
|
108
|
+
const lines = [`### ${view.title}`, ''];
|
|
109
|
+
if (view.subtitle)
|
|
110
|
+
lines.push(cell(view.subtitle), '');
|
|
111
|
+
const totals = [`winner: **${view.winner}**`];
|
|
112
|
+
if (view.cases) {
|
|
113
|
+
totals.push(`**${view.cases.matched}** cases matched`, `**${view.cases.added.length}** added`, `**${view.cases.removed.length}** removed`);
|
|
114
|
+
}
|
|
115
|
+
lines.push(totals.join(' | '), '');
|
|
116
|
+
if (view.warnings.length > 0) {
|
|
117
|
+
lines.push('**Warnings**', '', ...view.warnings.map((w) => `- ${cell(w)}`), '');
|
|
118
|
+
}
|
|
119
|
+
if (view.metrics.length > 0) {
|
|
120
|
+
lines.push(`| Metric | ${view.before} | ${view.after} | Difference | CI | p | Adjusted p | Significant | Winner |`, '| --- | ---: | ---: | ---: | --- | ---: | ---: | --- | --- |');
|
|
121
|
+
for (const m of view.metrics) {
|
|
122
|
+
lines.push(`| ${cell(m.name)} | ${num(m.before)} | ${num(m.after)} | ${signed(m.difference)} | ${ciText(m.ci)} | ${pText(m.pValue)} | ${pText(m.adjustedPValue)} | ${m.significant ? '**yes**' : 'no'} | ${m.winner} |`);
|
|
123
|
+
}
|
|
124
|
+
lines.push('');
|
|
125
|
+
}
|
|
126
|
+
else {
|
|
127
|
+
lines.push('No metric to compare.', '');
|
|
128
|
+
}
|
|
129
|
+
lines.push(...changeLines(view, 'regressed', 'Regressions'));
|
|
130
|
+
lines.push(...changeLines(view, 'improved', 'Improvements'));
|
|
131
|
+
return lines.join('\n');
|
|
132
|
+
}
|
|
133
|
+
function changesTable(view, kind) {
|
|
134
|
+
const rows = view.metrics.flatMap((m) => m[kind].map((c) => [
|
|
135
|
+
`<code>${escapeHtml(m.name)}</code>`,
|
|
136
|
+
escapeHtml(c.input),
|
|
137
|
+
escapeHtml(num(c.before)),
|
|
138
|
+
escapeHtml(num(c.after)),
|
|
139
|
+
`<span class="${c.delta < 0 ? 'fail' : 'pass'}">${escapeHtml(signed(c.delta))}</span>`,
|
|
140
|
+
]));
|
|
141
|
+
if (rows.length === 0)
|
|
142
|
+
return undefined;
|
|
143
|
+
return htmlTable(['Metric', 'Case', view.before, view.after, 'Change'], rows, new Set([2, 3, 4]));
|
|
144
|
+
}
|
|
145
|
+
/**
|
|
146
|
+
* A self-contained HTML page of a comparison: inline CSS for light and dark color schemes, no
|
|
147
|
+
* scripts, no external assets, and every value escaped
|
|
148
|
+
*/
|
|
149
|
+
export function comparisonHtml(comparison) {
|
|
150
|
+
const view = viewOf(comparison);
|
|
151
|
+
const sections = [`<h1>${escapeHtml(view.title)}</h1>`];
|
|
152
|
+
if (view.subtitle)
|
|
153
|
+
sections.push(`<p class="sub">${escapeHtml(view.subtitle)}</p>`);
|
|
154
|
+
sections.push(htmlTiles([
|
|
155
|
+
[
|
|
156
|
+
'winner',
|
|
157
|
+
`<span class="${view.winner === 'tie' ? 'muted' : 'pass'}">${escapeHtml(view.winner)}</span>`,
|
|
158
|
+
],
|
|
159
|
+
...(view.cases
|
|
160
|
+
? [
|
|
161
|
+
['cases matched', escapeHtml(String(view.cases.matched))],
|
|
162
|
+
['added', escapeHtml(String(view.cases.added.length))],
|
|
163
|
+
['removed', escapeHtml(String(view.cases.removed.length))],
|
|
164
|
+
]
|
|
165
|
+
: []),
|
|
166
|
+
]));
|
|
167
|
+
if (view.warnings.length > 0) {
|
|
168
|
+
sections.push('<h2>Warnings</h2>', `<ul>${view.warnings.map((w) => `<li class="warn">${escapeHtml(w)}</li>`).join('')}</ul>`);
|
|
169
|
+
}
|
|
170
|
+
sections.push('<h2>Metrics</h2>', view.metrics.length === 0
|
|
171
|
+
? '<p class="muted">No metric to compare.</p>'
|
|
172
|
+
: htmlTable([
|
|
173
|
+
'Metric',
|
|
174
|
+
view.before,
|
|
175
|
+
view.after,
|
|
176
|
+
'Cases',
|
|
177
|
+
'Difference',
|
|
178
|
+
'CI',
|
|
179
|
+
'Test',
|
|
180
|
+
'p',
|
|
181
|
+
'Adjusted p',
|
|
182
|
+
'Significant',
|
|
183
|
+
'Winner',
|
|
184
|
+
], view.metrics.map((m) => [
|
|
185
|
+
`<code>${escapeHtml(m.name)}</code>`,
|
|
186
|
+
escapeHtml(num(m.before)),
|
|
187
|
+
escapeHtml(num(m.after)),
|
|
188
|
+
escapeHtml(String(m.n)),
|
|
189
|
+
escapeHtml(signed(m.difference)),
|
|
190
|
+
escapeHtml(ciText(m.ci)),
|
|
191
|
+
escapeHtml(m.test),
|
|
192
|
+
escapeHtml(pText(m.pValue)),
|
|
193
|
+
escapeHtml(pText(m.adjustedPValue)),
|
|
194
|
+
m.significant ? '<span class="pass">yes</span>' : '<span class="muted">no</span>',
|
|
195
|
+
escapeHtml(m.winner),
|
|
196
|
+
]), new Set([1, 2, 3, 4, 7, 8])));
|
|
197
|
+
const regressed = changesTable(view, 'regressed');
|
|
198
|
+
if (regressed)
|
|
199
|
+
sections.push('<h2>Regressions</h2>', regressed);
|
|
200
|
+
const improved = changesTable(view, 'improved');
|
|
201
|
+
if (improved)
|
|
202
|
+
sections.push('<h2>Improvements</h2>', improved);
|
|
203
|
+
if (view.cases && (view.cases.added.length > 0 || view.cases.removed.length > 0)) {
|
|
204
|
+
const list = (inputs) => `<ul>${inputs.map((input) => `<li>${escapeHtml(input)}</li>`).join('')}</ul>`;
|
|
205
|
+
if (view.cases.added.length > 0) {
|
|
206
|
+
sections.push(`<h2>Cases only in the ${view.after} run</h2>`, list(view.cases.added));
|
|
207
|
+
}
|
|
208
|
+
if (view.cases.removed.length > 0) {
|
|
209
|
+
sections.push(`<h2>Cases only in the ${view.before} run</h2>`, list(view.cases.removed));
|
|
210
|
+
}
|
|
211
|
+
}
|
|
212
|
+
return htmlDocument(view.title, sections.join('\n'));
|
|
213
|
+
}
|
|
214
|
+
const GREEN = '\x1b[32m';
|
|
215
|
+
const RED = '\x1b[31m';
|
|
216
|
+
const YELLOW = '\x1b[33m';
|
|
217
|
+
const BOLD = '\x1b[1m';
|
|
218
|
+
const DIM = '\x1b[2m';
|
|
219
|
+
const RESET = '\x1b[0m';
|
|
220
|
+
function pad(text, width) {
|
|
221
|
+
return text.length >= width ? `${text} ` : text + ' '.repeat(width - text.length);
|
|
222
|
+
}
|
|
223
|
+
function printConsole(comparison) {
|
|
224
|
+
const view = viewOf(comparison);
|
|
225
|
+
console.log(`\n${BOLD}${view.title}${RESET}${view.subtitle ? ` ${DIM}${view.subtitle}${RESET}` : ''}`);
|
|
226
|
+
const cases = view.cases
|
|
227
|
+
? ` | ${view.cases.matched} cases matched, ${view.cases.added.length} added, ${view.cases.removed.length} removed`
|
|
228
|
+
: '';
|
|
229
|
+
console.log(`Winner: ${BOLD}${view.winner}${RESET}${cases}`);
|
|
230
|
+
for (const warning of view.warnings)
|
|
231
|
+
console.log(`${YELLOW}warning${RESET} ${warning}`);
|
|
232
|
+
if (view.metrics.length > 0) {
|
|
233
|
+
const width = Math.max(14, ...view.metrics.map((m) => m.name.length + 2));
|
|
234
|
+
const header = pad('Metric', width) +
|
|
235
|
+
pad(view.before, 12) +
|
|
236
|
+
pad(view.after, 12) +
|
|
237
|
+
pad('Diff', 10) +
|
|
238
|
+
pad('CI', 22) +
|
|
239
|
+
pad('p', 10) +
|
|
240
|
+
pad('Adj. p', 10) +
|
|
241
|
+
'Winner';
|
|
242
|
+
console.log(`\n${BOLD}${header}${RESET}`);
|
|
243
|
+
console.log(DIM + '─'.repeat(header.length) + RESET);
|
|
244
|
+
for (const m of view.metrics) {
|
|
245
|
+
const color = !m.significant ? '' : m.winner === view.after ? GREEN : RED;
|
|
246
|
+
console.log(pad(m.name, width) +
|
|
247
|
+
pad(num(m.before), 12) +
|
|
248
|
+
pad(num(m.after), 12) +
|
|
249
|
+
pad(signed(m.difference), 10) +
|
|
250
|
+
pad(ciText(m.ci), 22) +
|
|
251
|
+
pad(pText(m.pValue), 10) +
|
|
252
|
+
pad(pText(m.adjustedPValue), 10) +
|
|
253
|
+
`${color}${m.winner}${m.significant ? ' *' : ''}${color ? RESET : ''}`);
|
|
254
|
+
}
|
|
255
|
+
}
|
|
256
|
+
for (const [kind, heading, color] of [
|
|
257
|
+
['regressed', 'Regressions', RED],
|
|
258
|
+
['improved', 'Improvements', GREEN],
|
|
259
|
+
]) {
|
|
260
|
+
const changes = view.metrics.flatMap((m) => m[kind].map((c) => ({ metric: m.name, ...c })));
|
|
261
|
+
if (changes.length === 0)
|
|
262
|
+
continue;
|
|
263
|
+
console.log(`\n${BOLD}${heading}${RESET}`);
|
|
264
|
+
for (const c of changes) {
|
|
265
|
+
console.log(` ${color}${signed(c.delta)}${RESET} ${c.metric} ${DIM}${num(c.before)} -> ${num(c.after)}${RESET} ${truncateInput(c.input)}`);
|
|
266
|
+
}
|
|
267
|
+
}
|
|
268
|
+
}
|
|
269
|
+
/**
|
|
270
|
+
* Renders an `EvalComparison` result or a `compareRuns()` result: `console` prints it, `markdown`
|
|
271
|
+
* writes a compact summary for a pull request comment to `<path>.md`, and `html` a
|
|
272
|
+
* self-contained page to `<path>.html`. The default path is `eval-comparison`
|
|
273
|
+
*/
|
|
274
|
+
export function reportComparison(comparison, type, options) {
|
|
275
|
+
const types = Array.isArray(type) ? type : [type];
|
|
276
|
+
for (const t of types) {
|
|
277
|
+
switch (t) {
|
|
278
|
+
case 'console':
|
|
279
|
+
printConsole(comparison);
|
|
280
|
+
break;
|
|
281
|
+
case 'markdown':
|
|
282
|
+
writeFileSync(prepare(reportPath(options?.path, 'md', 'eval-comparison')), comparisonMarkdown(comparison), 'utf-8');
|
|
283
|
+
break;
|
|
284
|
+
case 'html':
|
|
285
|
+
writeFileSync(prepare(reportPath(options?.path, 'html', 'eval-comparison')), comparisonHtml(comparison), 'utf-8');
|
|
286
|
+
break;
|
|
287
|
+
}
|
|
288
|
+
}
|
|
289
|
+
}
|
|
290
|
+
//# sourceMappingURL=comparison.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"comparison.js","sourceRoot":"","sources":["../../src/reporters/comparison.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,aAAa,EAAE,MAAM,SAAS,CAAC;AAIxC,OAAO,EAAE,WAAW,EAAE,MAAM,qBAAqB,CAAC;AAClD,OAAO,EAAE,IAAI,EAAE,GAAG,EAAE,aAAa,EAAE,MAAM,UAAU,CAAC;AACpD,OAAO,EAAE,OAAO,EAAE,UAAU,EAAE,MAAM,SAAS,CAAC;AAE9C,OAAO,EAAE,UAAU,EAAE,YAAY,EAAE,SAAS,EAAE,SAAS,EAAE,MAAM,eAAe,CAAC;AA4C/E,4EAA4E;AAC5E,MAAM,oBAAoB,GAAG,EAAE,CAAC;AAEhC,SAAS,eAAe,CAAC,UAAgC;IACvD,OAAO,CAAC,CAAC,SAAS,IAAI,UAAU,CAAC,CAAC;AACpC,CAAC;AAED,SAAS,QAAQ,CAAC,GAAW;IAC3B,OAAO,GAAG,CAAC,IAAI,KAAK,SAAS,CAAC,CAAC,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC,CAAC,GAAG,GAAG,CAAC,IAAI,KAAK,GAAG,CAAC,KAAK,GAAG,CAAC;AAC3E,CAAC;AAED,SAAS,MAAM,CAAC,UAAgC;IAC9C,IAAI,eAAe,CAAC,UAAU,CAAC,EAAE,CAAC;QAChC,OAAO;YACL,KAAK,EAAE,gBAAgB;YACvB,MAAM,EAAE,MAAM;YACd,KAAK,EAAE,MAAM;YACb,QAAQ,EAAE,QAAQ,QAAQ,CAAC,UAAU,CAAC,IAAI,CAAC,UAAU,QAAQ,CAAC,UAAU,CAAC,IAAI,CAAC,EAAE;YAChF,MAAM,EAAE,UAAU,CAAC,MAAM;YACzB,OAAO,EAAE,MAAM,CAAC,OAAO,CAAC,UAAU,CAAC,OAAO,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC;gBAC9D,IAAI;gBACJ,MAAM,EAAE,CAAC,CAAC,IAAI;gBACd,KAAK,EAAE,CAAC,CAAC,IAAI;gBACb,CAAC,EAAE,CAAC,CAAC,CAAC;gBACN,UAAU,EAAE,CAAC,CAAC,UAAU;gBACxB,EAAE,EAAE,CAAC,CAAC,EAAE;gBACR,IAAI,EAAE,CAAC,CAAC,IAAI;gBACZ,MAAM,EAAE,CAAC,CAAC,MAAM;gBAChB,cAAc,EAAE,CAAC,CAAC,cAAc;gBAChC,WAAW,EAAE,CAAC,CAAC,WAAW;gBAC1B,MAAM,EAAE,CAAC,CAAC,MAAM;gBAChB,QAAQ,EAAE,CAAC,CAAC,QAAQ,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,EAAE,GAAG,CAAC,EAAE,MAAM,EAAE,CAAC,CAAC,IAAI,EAAE,KAAK,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC;gBAC1E,SAAS,EAAE,CAAC,CAAC,SAAS,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,EAAE,GAAG,CAAC,EAAE,MAAM,EAAE,CAAC,CAAC,IAAI,EAAE,KAAK,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC;aAC7E,CAAC,CAAC;YACH,KAAK,EAAE;gBACL,OAAO,EAAE,UAAU,CAAC,OAAO;gBAC3B,KAAK,EAAE,UAAU,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,KAAK,CAAC;gBAC3C,OAAO,EAAE,UAAU,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,KAAK,CAAC;aAChD;YACD,QAAQ,EAAE,UAAU,CAAC,QAAQ;SAC9B,CAAC;IACJ,CAAC;IAED,MAAM,EAAE,QAAQ,EAAE,UAAU,EAAE,OAAO,EAAE,GAAG,UAAU,CAAC;IACrD,MAAM,KAAK,GAAG,QAAQ,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,EAAE,EAAE,CAAC,CAAC,CAAC,EAAE,UAAU,CAAC,OAAO,CAAC,EAAE,CAAC,CAAU,CAAC,CAAC;IACpF,MAAM,OAAO,GAAG,CAAC,KAAa,EAAE,EAAE,CAAC,QAAQ,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC;IACtE,OAAO;QACL,KAAK,EAAE,gBAAgB;QACvB,MAAM,EAAE,UAAU;QAClB,KAAK,EAAE,YAAY;QACnB,MAAM,EAAE,OAAO,CAAC,MAAM;QACtB,OAAO,EAAE,MAAM,CAAC,OAAO,CAAC,OAAO,CAAC,OAAO,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC;YAC3D,IAAI;YACJ,MAAM,EAAE,CAAC,CAAC,QAAQ;YAClB,KAAK,EAAE,CAAC,CAAC,UAAU;YACnB,CAAC,EAAE,CAAC,CAAC,CAAC;YACN,UAAU,EAAE,CAAC,CAAC,UAAU;YACxB,EAAE,EAAE,CAAC,CAAC,EAAE;YACR,IAAI,EAAE,CAAC,CAAC,IAAI;YACZ,MAAM,EAAE,CAAC,CAAC,MAAM;YAChB,cAAc,EAAE,CAAC,CAAC,cAAc;YAChC,WAAW,EAAE,CAAC,CAAC,WAAW;YAC1B,MAAM,EAAE,CAAC,CAAC,MAAM;YAChB,GAAG,WAAW,CAAC,KAAK,EAAE,IAAI,EAAE,OAAO,CAAC;SACrC,CAAC,CAAC;QACH,QAAQ,EAAE,EAAE;KACb,CAAC;AACJ,CAAC;AAED,SAAS,MAAM,CAAC,CAAS;IACvB,OAAO,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,IAAI,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC;AACvC,CAAC;AAED,SAAS,MAAM,CAAC,EAAiC;IAC/C,OAAO,IAAI,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,KAAK,GAAG,CAAC,EAAE,CAAC,IAAI,CAAC,GAAG,CAAC;AAC7C,CAAC;AAED,SAAS,KAAK,CAAC,CAAS;IACtB,OAAO,CAAC,GAAG,MAAM,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC;AACzC,CAAC;AAED,SAAS,WAAW,CAClB,IAAoB,EACpB,IAA8B,EAC9B,OAAe;IAEf,MAAM,IAAI,GAAG,IAAI,CAAC,OAAO,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,EAAE;QACtC,MAAM,OAAO,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC;QACxB,MAAM,KAAK,GAAG,OAAO;aAClB,KAAK,CAAC,CAAC,EAAE,oBAAoB,CAAC;aAC9B,GAAG,CACF,CAAC,CAAC,EAAE,EAAE,CACJ,KAAK,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,MAAM,IAAI,CAAC,CAAC,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,MAAM,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,KAAK,CAAC,MAAM,MAAM,CAAC,CAAC,CAAC,KAAK,CAAC,IAAI,CACrG,CAAC;QACJ,MAAM,IAAI,GACR,OAAO,CAAC,MAAM,GAAG,oBAAoB;YACnC,CAAC,CAAC,CAAC,KAAK,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,UAAU,OAAO,CAAC,MAAM,GAAG,oBAAoB,eAAe,CAAC;YACnF,CAAC,CAAC,EAAE,CAAC;QACT,OAAO,CAAC,GAAG,KAAK,EAAE,GAAG,IAAI,CAAC,CAAC;IAC7B,CAAC,CAAC,CAAC;IACH,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,EAAE,CAAC;IACjC,OAAO;QACL,KAAK,OAAO,IAAI;QAChB,EAAE;QACF,qBAAqB,IAAI,CAAC,MAAM,MAAM,IAAI,CAAC,KAAK,aAAa;QAC7D,oCAAoC;QACpC,GAAG,IAAI;QACP,EAAE;KACH,CAAC;AACJ,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,kBAAkB,CAAC,UAAgC;IACjE,MAAM,IAAI,GAAG,MAAM,CAAC,UAAU,CAAC,CAAC;IAChC,MAAM,KAAK,GAAG,CAAC,OAAO,IAAI,CAAC,KAAK,EAAE,EAAE,EAAE,CAAC,CAAC;IACxC,IAAI,IAAI,CAAC,QAAQ;QAAE,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,EAAE,EAAE,CAAC,CAAC;IAEvD,MAAM,MAAM,GAAG,CAAC,aAAa,IAAI,CAAC,MAAM,IAAI,CAAC,CAAC;IAC9C,IAAI,IAAI,CAAC,KAAK,EAAE,CAAC;QACf,MAAM,CAAC,IAAI,CACT,KAAK,IAAI,CAAC,KAAK,CAAC,OAAO,kBAAkB,EACzC,KAAK,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,MAAM,UAAU,EACtC,KAAK,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,MAAM,YAAY,CAC3C,CAAC;IACJ,CAAC;IACD,KAAK,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,EAAE,EAAE,CAAC,CAAC;IAEnC,IAAI,IAAI,CAAC,QAAQ,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QAC7B,KAAK,CAAC,IAAI,CAAC,cAAc,EAAE,EAAE,EAAE,GAAG,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,KAAK,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC;IAClF,CAAC;IAED,IAAI,IAAI,CAAC,OAAO,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QAC5B,KAAK,CAAC,IAAI,CACR,cAAc,IAAI,CAAC,MAAM,MAAM,IAAI,CAAC,KAAK,8DAA8D,EACvG,8DAA8D,CAC/D,CAAC;QACF,KAAK,MAAM,CAAC,IAAI,IAAI,CAAC,OAAO,EAAE,CAAC;YAC7B,KAAK,CAAC,IAAI,CACR,KAAK,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,MAAM,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,KAAK,CAAC,MAAM,MAAM,CAAC,CAAC,CAAC,UAAU,CAAC,MAAM,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,MAAM,KAAK,CAAC,CAAC,CAAC,MAAM,CAAC,MAAM,KAAK,CAAC,CAAC,CAAC,cAAc,CAAC,MAAM,CAAC,CAAC,WAAW,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,IAAI,MAAM,CAAC,CAAC,MAAM,IAAI,CAC7M,CAAC;QACJ,CAAC;QACD,KAAK,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;IACjB,CAAC;SAAM,CAAC;QACN,KAAK,CAAC,IAAI,CAAC,uBAAuB,EAAE,EAAE,CAAC,CAAC;IAC1C,CAAC;IAED,KAAK,CAAC,IAAI,CAAC,GAAG,WAAW,CAAC,IAAI,EAAE,WAAW,EAAE,aAAa,CAAC,CAAC,CAAC;IAC7D,KAAK,CAAC,IAAI,CAAC,GAAG,WAAW,CAAC,IAAI,EAAE,UAAU,EAAE,cAAc,CAAC,CAAC,CAAC;IAC7D,OAAO,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;AAC1B,CAAC;AAED,SAAS,YAAY,CAAC,IAAoB,EAAE,IAA8B;IACxE,MAAM,IAAI,GAAG,IAAI,CAAC,OAAO,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,EAAE,CACtC,CAAC,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC;QACjB,SAAS,UAAU,CAAC,CAAC,CAAC,IAAI,CAAC,SAAS;QACpC,UAAU,CAAC,CAAC,CAAC,KAAK,CAAC;QACnB,UAAU,CAAC,GAAG,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC;QACzB,UAAU,CAAC,GAAG,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC;QACxB,gBAAgB,CAAC,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,MAAM,KAAK,UAAU,CAAC,MAAM,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,SAAS;KACvF,CAAC,CACH,CAAC;IACF,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,SAAS,CAAC;IACxC,OAAO,SAAS,CAAC,CAAC,QAAQ,EAAE,MAAM,EAAE,IAAI,CAAC,MAAM,EAAE,IAAI,CAAC,KAAK,EAAE,QAAQ,CAAC,EAAE,IAAI,EAAE,IAAI,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC;AACpG,CAAC;AAED;;;GAGG;AACH,MAAM,UAAU,cAAc,CAAC,UAAgC;IAC7D,MAAM,IAAI,GAAG,MAAM,CAAC,UAAU,CAAC,CAAC;IAChC,MAAM,QAAQ,GAAG,CAAC,OAAO,UAAU,CAAC,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC;IACxD,IAAI,IAAI,CAAC,QAAQ;QAAE,QAAQ,CAAC,IAAI,CAAC,kBAAkB,UAAU,CAAC,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC;IAEpF,QAAQ,CAAC,IAAI,CACX,SAAS,CAAC;QACR;YACE,QAAQ;YACR,gBAAgB,IAAI,CAAC,MAAM,KAAK,KAAK,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,KAAK,UAAU,CAAC,IAAI,CAAC,MAAM,CAAC,SAAS;SAC9F;QACD,GAAG,CAAC,IAAI,CAAC,KAAK;YACZ,CAAC,CAAE;gBACC,CAAC,eAAe,EAAE,UAAU,CAAC,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC;gBACzD,CAAC,OAAO,EAAE,UAAU,CAAC,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC;gBACtD,CAAC,SAAS,EAAE,UAAU,CAAC,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC;aACjD;YACb,CAAC,CAAC,EAAE,CAAC;KACR,CAAC,CACH,CAAC;IAEF,IAAI,IAAI,CAAC,QAAQ,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QAC7B,QAAQ,CAAC,IAAI,CACX,mBAAmB,EACnB,OAAO,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,oBAAoB,UAAU,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,OAAO,CAC1F,CAAC;IACJ,CAAC;IAED,QAAQ,CAAC,IAAI,CACX,kBAAkB,EAClB,IAAI,CAAC,OAAO,CAAC,MAAM,KAAK,CAAC;QACvB,CAAC,CAAC,4CAA4C;QAC9C,CAAC,CAAC,SAAS,CACP;YACE,QAAQ;YACR,IAAI,CAAC,MAAM;YACX,IAAI,CAAC,KAAK;YACV,OAAO;YACP,YAAY;YACZ,IAAI;YACJ,MAAM;YACN,GAAG;YACH,YAAY;YACZ,aAAa;YACb,QAAQ;SACT,EACD,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC;YACtB,SAAS,UAAU,CAAC,CAAC,CAAC,IAAI,CAAC,SAAS;YACpC,UAAU,CAAC,GAAG,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC;YACzB,UAAU,CAAC,GAAG,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC;YACxB,UAAU,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;YACvB,UAAU,CAAC,MAAM,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC;YAChC,UAAU,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC;YACxB,UAAU,CAAC,CAAC,CAAC,IAAI,CAAC;YAClB,UAAU,CAAC,KAAK,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC;YAC3B,UAAU,CAAC,KAAK,CAAC,CAAC,CAAC,cAAc,CAAC,CAAC;YACnC,CAAC,CAAC,WAAW,CAAC,CAAC,CAAC,+BAA+B,CAAC,CAAC,CAAC,+BAA+B;YACjF,UAAU,CAAC,CAAC,CAAC,MAAM,CAAC;SACrB,CAAC,EACF,IAAI,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC,CAC5B,CACN,CAAC;IAEF,MAAM,SAAS,GAAG,YAAY,CAAC,IAAI,EAAE,WAAW,CAAC,CAAC;IAClD,IAAI,SAAS;QAAE,QAAQ,CAAC,IAAI,CAAC,sBAAsB,EAAE,SAAS,CAAC,CAAC;IAChE,MAAM,QAAQ,GAAG,YAAY,CAAC,IAAI,EAAE,UAAU,CAAC,CAAC;IAChD,IAAI,QAAQ;QAAE,QAAQ,CAAC,IAAI,CAAC,uBAAuB,EAAE,QAAQ,CAAC,CAAC;IAE/D,IAAI,IAAI,CAAC,KAAK,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,IAAI,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,MAAM,GAAG,CAAC,CAAC,EAAE,CAAC;QACjF,MAAM,IAAI,GAAG,CAAC,MAAgB,EAAE,EAAE,CAChC,OAAO,MAAM,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,OAAO,UAAU,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,OAAO,CAAC;QAChF,IAAI,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YAChC,QAAQ,CAAC,IAAI,CAAC,yBAAyB,IAAI,CAAC,KAAK,WAAW,EAAE,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC;QACxF,CAAC;QACD,IAAI,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YAClC,QAAQ,CAAC,IAAI,CAAC,yBAAyB,IAAI,CAAC,MAAM,WAAW,EAAE,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC;QAC3F,CAAC;IACH,CAAC;IAED,OAAO,YAAY,CAAC,IAAI,CAAC,KAAK,EAAE,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC;AACvD,CAAC;AAED,MAAM,KAAK,GAAG,UAAU,CAAC;AACzB,MAAM,GAAG,GAAG,UAAU,CAAC;AACvB,MAAM,MAAM,GAAG,UAAU,CAAC;AAC1B,MAAM,IAAI,GAAG,SAAS,CAAC;AACvB,MAAM,GAAG,GAAG,SAAS,CAAC;AACtB,MAAM,KAAK,GAAG,SAAS,CAAC;AAExB,SAAS,GAAG,CAAC,IAAY,EAAE,KAAa;IACtC,OAAO,IAAI,CAAC,MAAM,IAAI,KAAK,CAAC,CAAC,CAAC,GAAG,IAAI,GAAG,CAAC,CAAC,CAAC,IAAI,GAAG,GAAG,CAAC,MAAM,CAAC,KAAK,GAAG,IAAI,CAAC,MAAM,CAAC,CAAC;AACpF,CAAC;AAED,SAAS,YAAY,CAAC,UAAgC;IACpD,MAAM,IAAI,GAAG,MAAM,CAAC,UAAU,CAAC,CAAC;IAChC,OAAO,CAAC,GAAG,CACT,KAAK,IAAI,GAAG,IAAI,CAAC,KAAK,GAAG,KAAK,GAAG,IAAI,CAAC,QAAQ,CAAC,CAAC,CAAC,IAAI,GAAG,GAAG,IAAI,CAAC,QAAQ,GAAG,KAAK,EAAE,CAAC,CAAC,CAAC,EAAE,EAAE,CAC1F,CAAC;IACF,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK;QACtB,CAAC,CAAC,MAAM,IAAI,CAAC,KAAK,CAAC,OAAO,mBAAmB,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,MAAM,WAAW,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,MAAM,UAAU;QAClH,CAAC,CAAC,EAAE,CAAC;IACP,OAAO,CAAC,GAAG,CAAC,WAAW,IAAI,GAAG,IAAI,CAAC,MAAM,GAAG,KAAK,GAAG,KAAK,EAAE,CAAC,CAAC;IAE7D,KAAK,MAAM,OAAO,IAAI,IAAI,CAAC,QAAQ;QAAE,OAAO,CAAC,GAAG,CAAC,GAAG,MAAM,UAAU,KAAK,IAAI,OAAO,EAAE,CAAC,CAAC;IAExF,IAAI,IAAI,CAAC,OAAO,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QAC5B,MAAM,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,EAAE,EAAE,GAAG,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC;QAC1E,MAAM,MAAM,GACV,GAAG,CAAC,QAAQ,EAAE,KAAK,CAAC;YACpB,GAAG,CAAC,IAAI,CAAC,MAAM,EAAE,EAAE,CAAC;YACpB,GAAG,CAAC,IAAI,CAAC,KAAK,EAAE,EAAE,CAAC;YACnB,GAAG,CAAC,MAAM,EAAE,EAAE,CAAC;YACf,GAAG,CAAC,IAAI,EAAE,EAAE,CAAC;YACb,GAAG,CAAC,GAAG,EAAE,EAAE,CAAC;YACZ,GAAG,CAAC,QAAQ,EAAE,EAAE,CAAC;YACjB,QAAQ,CAAC;QACX,OAAO,CAAC,GAAG,CAAC,KAAK,IAAI,GAAG,MAAM,GAAG,KAAK,EAAE,CAAC,CAAC;QAC1C,OAAO,CAAC,GAAG,CAAC,GAAG,GAAG,GAAG,CAAC,MAAM,CAAC,MAAM,CAAC,MAAM,CAAC,GAAG,KAAK,CAAC,CAAC;QACrD,KAAK,MAAM,CAAC,IAAI,IAAI,CAAC,OAAO,EAAE,CAAC;YAC7B,MAAM,KAAK,GAAG,CAAC,CAAC,CAAC,WAAW,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,MAAM,KAAK,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,GAAG,CAAC;YAC1E,OAAO,CAAC,GAAG,CACT,GAAG,CAAC,CAAC,CAAC,IAAI,EAAE,KAAK,CAAC;gBAChB,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,MAAM,CAAC,EAAE,EAAE,CAAC;gBACtB,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,KAAK,CAAC,EAAE,EAAE,CAAC;gBACrB,GAAG,CAAC,MAAM,CAAC,CAAC,CAAC,UAAU,CAAC,EAAE,EAAE,CAAC;gBAC7B,GAAG,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC;gBACrB,GAAG,CAAC,KAAK,CAAC,CAAC,CAAC,MAAM,CAAC,EAAE,EAAE,CAAC;gBACxB,GAAG,CAAC,KAAK,CAAC,CAAC,CAAC,cAAc,CAAC,EAAE,EAAE,CAAC;gBAChC,GAAG,KAAK,GAAG,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC,WAAW,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,GAAG,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,EAAE,CACzE,CAAC;QACJ,CAAC;IACH,CAAC;IAED,KAAK,MAAM,CAAC,IAAI,EAAE,OAAO,EAAE,KAAK,CAAC,IAAI;QACnC,CAAC,WAAW,EAAE,aAAa,EAAE,GAAG,CAAC;QACjC,CAAC,UAAU,EAAE,cAAc,EAAE,KAAK,CAAC;KAC3B,EAAE,CAAC;QACX,MAAM,OAAO,GAAG,IAAI,CAAC,OAAO,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,EAAE,MAAM,EAAE,CAAC,CAAC,IAAI,EAAE,GAAG,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC;QAC5F,IAAI,OAAO,CAAC,MAAM,KAAK,CAAC;YAAE,SAAS;QACnC,OAAO,CAAC,GAAG,CAAC,KAAK,IAAI,GAAG,OAAO,GAAG,KAAK,EAAE,CAAC,CAAC;QAC3C,KAAK,MAAM,CAAC,IAAI,OAAO,EAAE,CAAC;YACxB,OAAO,CAAC,GAAG,CACT,KAAK,KAAK,GAAG,MAAM,CAAC,CAAC,CAAC,KAAK,CAAC,GAAG,KAAK,IAAI,CAAC,CAAC,MAAM,IAAI,GAAG,GAAG,GAAG,CAAC,CAAC,CAAC,MAAM,CAAC,OAAO,GAAG,CAAC,CAAC,CAAC,KAAK,CAAC,GAAG,KAAK,IAAI,aAAa,CAAC,CAAC,CAAC,KAAK,CAAC,EAAE,CAC/H,CAAC;QACJ,CAAC;IACH,CAAC;AACH,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,gBAAgB,CAC9B,UAAgC,EAChC,IAAuD,EACvD,OAAyB;IAEzB,MAAM,KAAK,GAAG,KAAK,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC;IAClD,KAAK,MAAM,CAAC,IAAI,KAAK,EAAE,CAAC;QACtB,QAAQ,CAAC,EAAE,CAAC;YACV,KAAK,SAAS;gBACZ,YAAY,CAAC,UAAU,CAAC,CAAC;gBACzB,MAAM;YACR,KAAK,UAAU;gBACb,aAAa,CACX,OAAO,CAAC,UAAU,CAAC,OAAO,EAAE,IAAI,EAAE,IAAI,EAAE,iBAAiB,CAAC,CAAC,EAC3D,kBAAkB,CAAC,UAAU,CAAC,EAC9B,OAAO,CACR,CAAC;gBACF,MAAM;YACR,KAAK,MAAM;gBACT,aAAa,CACX,OAAO,CAAC,UAAU,CAAC,OAAO,EAAE,IAAI,EAAE,MAAM,EAAE,iBAAiB,CAAC,CAAC,EAC7D,cAAc,CAAC,UAAU,CAAC,EAC1B,OAAO,CACR,CAAC;gBACF,MAAM;QACV,CAAC;IACH,CAAC;AACH,CAAC"}
|