@platforma-sdk/model 1.51.5 → 1.51.9
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/dist/package.json.cjs +1 -1
- package/dist/package.json.js +1 -1
- package/dist/render/util/label.cjs +32 -4
- package/dist/render/util/label.cjs.map +1 -1
- package/dist/render/util/label.d.ts.map +1 -1
- package/dist/render/util/label.js +32 -4
- package/dist/render/util/label.js.map +1 -1
- package/package.json +4 -4
- package/src/render/util/label.test.ts +81 -0
- package/src/render/util/label.ts +38 -3
package/dist/package.json.cjs
CHANGED
package/dist/package.json.js
CHANGED
|
@@ -110,6 +110,30 @@ function deriveLabels(values, specExtractor, ops = {}) {
|
|
|
110
110
|
}
|
|
111
111
|
return result;
|
|
112
112
|
};
|
|
113
|
+
const countUniqueLabels = (result) => result === undefined ? 0 : new Set(result.map((c) => c.label)).size;
|
|
114
|
+
// Post-processing: try removing types one by one (lowest importance first) to minimize the label set
|
|
115
|
+
// Accepts removal if it doesn't decrease the number of unique labels (cardinality)
|
|
116
|
+
const minimizeTypeSet = (typeSet) => {
|
|
117
|
+
const initialResult = calculate(typeSet);
|
|
118
|
+
if (initialResult === undefined) {
|
|
119
|
+
return typeSet;
|
|
120
|
+
}
|
|
121
|
+
const currentCardinality = countUniqueLabels(initialResult);
|
|
122
|
+
// Get types sorted by importance ascending (lowest first), excluding forced elements
|
|
123
|
+
const removableSorted = [...typeSet]
|
|
124
|
+
.filter((t) => !forceTraceElements?.has(t.split('@')[0])
|
|
125
|
+
&& !(ops.includeNativeLabel && t === LabelTypeFull))
|
|
126
|
+
.sort((a, b) => (importances.get(a) ?? 0) - (importances.get(b) ?? 0));
|
|
127
|
+
for (const typeToRemove of removableSorted) {
|
|
128
|
+
const reducedSet = new Set(typeSet);
|
|
129
|
+
reducedSet.delete(typeToRemove);
|
|
130
|
+
const candidateResult = calculate(reducedSet);
|
|
131
|
+
if (candidateResult !== undefined && countUniqueLabels(candidateResult) >= currentCardinality) {
|
|
132
|
+
typeSet.delete(typeToRemove);
|
|
133
|
+
}
|
|
134
|
+
}
|
|
135
|
+
return typeSet;
|
|
136
|
+
};
|
|
113
137
|
if (mainTypes.length === 0) {
|
|
114
138
|
if (secondaryTypes.length !== 0)
|
|
115
139
|
throw new Error('Non-empty secondary types list while main types list is empty.');
|
|
@@ -135,16 +159,20 @@ function deriveLabels(values, specExtractor, ops = {}) {
|
|
|
135
159
|
if (additionalType >= 0)
|
|
136
160
|
currentSet.add(mainTypes[additionalType]);
|
|
137
161
|
const candidateResult = calculate(currentSet);
|
|
138
|
-
|
|
139
|
-
|
|
140
|
-
return
|
|
162
|
+
if (candidateResult !== undefined && countUniqueLabels(candidateResult) === values.length) {
|
|
163
|
+
minimizeTypeSet(currentSet);
|
|
164
|
+
return calculate(currentSet);
|
|
165
|
+
}
|
|
141
166
|
additionalType++;
|
|
142
167
|
if (additionalType >= mainTypes.length) {
|
|
143
168
|
includedTypes++;
|
|
144
169
|
additionalType = includedTypes;
|
|
145
170
|
}
|
|
146
171
|
}
|
|
147
|
-
|
|
172
|
+
// Fallback: include all types, then try to minimize
|
|
173
|
+
const fallbackSet = new Set([...mainTypes, ...secondaryTypes]);
|
|
174
|
+
minimizeTypeSet(fallbackSet);
|
|
175
|
+
return calculate(fallbackSet, true);
|
|
148
176
|
}
|
|
149
177
|
|
|
150
178
|
exports.Trace = Trace;
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"label.cjs","sources":["../../../src/render/util/label.ts"],"sourcesContent":["import { Annotation, parseJson, readAnnotation, type PObjectSpec } from '@milaboratories/pl-model-common';\nimport { z } from 'zod';\n\nexport type RecordsWithLabel<T> = {\n value: T;\n label: string;\n};\n\nexport type LabelDerivationOps = {\n /** Force inclusion of native column label */\n includeNativeLabel?: boolean;\n /** Separator to use between label parts (\" / \" by default) */\n separator?: string;\n /** If true, label will be added as suffix (at the end of the generated label). By default label added as a prefix. */\n addLabelAsSuffix?: boolean;\n /** Trace elements list that will be forced to be included in the label. */\n forceTraceElements?: string[];\n};\n\nexport const TraceEntry = z.object({\n type: z.string(),\n importance: z.number().optional(),\n id: z.string().optional(),\n label: z.string(),\n});\nexport type TraceEntry = z.infer<typeof TraceEntry>;\ntype FullTraceEntry = TraceEntry & { fullType: string; occurrenceIndex: number };\n\nexport const Trace = z.array(TraceEntry);\nexport type Trace = z.infer<typeof Trace>;\ntype FullTrace = FullTraceEntry[];\n\n// Define the possible return types for the specExtractor function\ntype SpecExtractorResult = PObjectSpec | {\n spec: PObjectSpec;\n prefixTrace?: TraceEntry[];\n suffixTrace?: TraceEntry[];\n};\n\nconst DistancePenalty = 0.001;\n\nconst LabelType = '__LABEL__';\nconst LabelTypeFull = '__LABEL__@1';\n\nexport function deriveLabels<T>(\n values: T[],\n specExtractor: (obj: T) => SpecExtractorResult,\n ops: LabelDerivationOps = {},\n): RecordsWithLabel<T>[] {\n const importances = new Map<string, number>();\n\n const forceTraceElements = (ops.forceTraceElements !== undefined && ops.forceTraceElements.length > 0)\n ? new Set(ops.forceTraceElements)\n : undefined;\n\n // number of times certain type occurred among all of the\n const numberOfRecordsWithType = new Map<string, number>();\n\n const enrichedRecords = values.map((value) => {\n const extractorResult = specExtractor(value);\n let spec: PObjectSpec;\n let prefixTrace: TraceEntry[] | undefined;\n let suffixTrace: TraceEntry[] | undefined;\n\n // Check if the result is the new structure or just PObjectSpec\n if ('spec' in extractorResult && typeof extractorResult.spec === 'object') {\n // It's the new structure { spec, prefixTrace?, suffixTrace? }\n spec = extractorResult.spec;\n prefixTrace = extractorResult.prefixTrace;\n suffixTrace = extractorResult.suffixTrace;\n } else {\n // It's just PObjectSpec\n spec = extractorResult as PObjectSpec;\n }\n\n const label = readAnnotation(spec, Annotation.Label);\n const traceStr = readAnnotation(spec, Annotation.Trace);\n const baseTrace = (traceStr ? Trace.safeParse(parseJson(traceStr)).data : undefined) ?? [];\n\n const trace = [\n ...(prefixTrace ?? []),\n ...baseTrace,\n ...(suffixTrace ?? []),\n ];\n\n if (label !== undefined) {\n const labelEntry = { label, type: LabelType, importance: -2 };\n if (ops.addLabelAsSuffix) trace.push(labelEntry);\n else trace.splice(0, 0, labelEntry);\n }\n\n const fullTrace: FullTrace = [];\n\n const occurrences = new Map<string, number>();\n for (let i = trace.length - 1; i >= 0; --i) {\n const { type: typeName } = trace[i];\n const importance = trace[i].importance ?? 0;\n const occurrenceIndex = (occurrences.get(typeName) ?? 0) + 1;\n occurrences.set(typeName, occurrenceIndex);\n const fullType = `${typeName}@${occurrenceIndex}`;\n numberOfRecordsWithType.set(fullType, (numberOfRecordsWithType.get(fullType) ?? 0) + 1);\n importances.set(\n fullType,\n Math.max(\n importances.get(fullType) ?? Number.NEGATIVE_INFINITY,\n importance - (trace.length - i) * DistancePenalty,\n ),\n );\n fullTrace.push({ ...trace[i], fullType, occurrenceIndex: occurrenceIndex });\n }\n fullTrace.reverse();\n return {\n value,\n spec,\n label,\n fullTrace,\n };\n });\n\n // excluding repeated types (i.e. ..@2, ..@3, etc.) not found in some records\n const mainTypes: string[] = [];\n // repeated types (i.e. ..@2, ..@3, etc.) not found in some records\n const secondaryTypes: string[] = [];\n\n const allTypeRecords = [...importances];\n // sorting: most important types go first\n allTypeRecords.sort(([, i1], [, i2]) => i2 - i1);\n\n for (const [typeName] of allTypeRecords) {\n if (typeName.endsWith('@1') || numberOfRecordsWithType.get(typeName) === values.length)\n mainTypes.push(typeName);\n else secondaryTypes.push(typeName);\n }\n\n const calculate = (includedTypes: Set<string>, force: boolean = false) => {\n const result: RecordsWithLabel<T>[] = [];\n for (let i = 0; i < enrichedRecords.length; i++) {\n const r = enrichedRecords[i];\n const includedTrace = r.fullTrace\n .filter((fm) => includedTypes.has(fm.fullType)\n || (forceTraceElements && forceTraceElements.has(fm.type)));\n if (includedTrace.length === 0) {\n if (force)\n result.push({\n label: 'Unlabeled',\n value: r.value,\n } satisfies RecordsWithLabel<T>);\n else return undefined;\n }\n const labelSet = includedTrace\n .map((fm) => fm.label);\n const sep = ops.separator ?? ' / ';\n result.push({\n label: labelSet.join(sep),\n value: r.value,\n } satisfies RecordsWithLabel<T>);\n }\n return result;\n };\n\n if (mainTypes.length === 0) {\n if (secondaryTypes.length !== 0) throw new Error('Non-empty secondary types list while main types list is empty.');\n return calculate(new Set(LabelTypeFull), true)!;\n }\n\n //\n // includedTypes = 2\n // * *\n // T0 T1 T2 T3 T4 T5\n // *\n // additionalType = 3\n //\n // Resulting set: T0, T1, T3\n //\n let includedTypes = 0;\n let additionalType = -1;\n while (includedTypes < mainTypes.length) {\n const currentSet = new Set<string>();\n if (ops.includeNativeLabel) currentSet.add(LabelTypeFull);\n for (let i = 0; i < includedTypes; ++i) currentSet.add(mainTypes[i]);\n if (additionalType >= 0)\n currentSet.add(mainTypes[additionalType]);\n\n const candidateResult = calculate(currentSet);\n\n // checking if labels uniquely separate our records\n if (candidateResult !== undefined && new Set(candidateResult.map((c) => c.label)).size === values.length) return candidateResult;\n\n additionalType++;\n if (additionalType >= mainTypes.length) {\n includedTypes++;\n additionalType = includedTypes;\n }\n }\n\n return calculate(new Set([...mainTypes, ...secondaryTypes]), true)!;\n}\n"],"names":["z","readAnnotation","Annotation","parseJson"],"mappings":";;;;;AAmBO,MAAM,UAAU,GAAGA,KAAC,CAAC,MAAM,CAAC;AACjC,IAAA,IAAI,EAAEA,KAAC,CAAC,MAAM,EAAE;AAChB,IAAA,UAAU,EAAEA,KAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;AACjC,IAAA,EAAE,EAAEA,KAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;AACzB,IAAA,KAAK,EAAEA,KAAC,CAAC,MAAM,EAAE;AAClB,CAAA;AAIM,MAAM,KAAK,GAAGA,KAAC,CAAC,KAAK,CAAC,UAAU;AAWvC,MAAM,eAAe,GAAG,KAAK;AAE7B,MAAM,SAAS,GAAG,WAAW;AAC7B,MAAM,aAAa,GAAG,aAAa;AAE7B,SAAU,YAAY,CAC1B,MAAW,EACX,aAA8C,EAC9C,MAA0B,EAAE,EAAA;AAE5B,IAAA,MAAM,WAAW,GAAG,IAAI,GAAG,EAAkB;AAE7C,IAAA,MAAM,kBAAkB,GAAG,CAAC,GAAG,CAAC,kBAAkB,KAAK,SAAS,IAAI,GAAG,CAAC,kBAAkB,CAAC,MAAM,GAAG,CAAC;AACnG,UAAE,IAAI,GAAG,CAAC,GAAG,CAAC,kBAAkB;UAC9B,SAAS;;AAGb,IAAA,MAAM,uBAAuB,GAAG,IAAI,GAAG,EAAkB;IAEzD,MAAM,eAAe,GAAG,MAAM,CAAC,GAAG,CAAC,CAAC,KAAK,KAAI;AAC3C,QAAA,MAAM,eAAe,GAAG,aAAa,CAAC,KAAK,CAAC;AAC5C,QAAA,IAAI,IAAiB;AACrB,QAAA,IAAI,WAAqC;AACzC,QAAA,IAAI,WAAqC;;QAGzC,IAAI,MAAM,IAAI,eAAe,IAAI,OAAO,eAAe,CAAC,IAAI,KAAK,QAAQ,EAAE;;AAEzE,YAAA,IAAI,GAAG,eAAe,CAAC,IAAI;AAC3B,YAAA,WAAW,GAAG,eAAe,CAAC,WAAW;AACzC,YAAA,WAAW,GAAG,eAAe,CAAC,WAAW;QAC3C;aAAO;;YAEL,IAAI,GAAG,eAA8B;QACvC;QAEA,MAAM,KAAK,GAAGC,4BAAc,CAAC,IAAI,EAAEC,wBAAU,CAAC,KAAK,CAAC;QACpD,MAAM,QAAQ,GAAGD,4BAAc,CAAC,IAAI,EAAEC,wBAAU,CAAC,KAAK,CAAC;QACvD,MAAM,SAAS,GAAG,CAAC,QAAQ,GAAG,KAAK,CAAC,SAAS,CAACC,uBAAS,CAAC,QAAQ,CAAC,CAAC,CAAC,IAAI,GAAG,SAAS,KAAK,EAAE;AAE1F,QAAA,MAAM,KAAK,GAAG;AACZ,YAAA,IAAI,WAAW,IAAI,EAAE,CAAC;AACtB,YAAA,GAAG,SAAS;AACZ,YAAA,IAAI,WAAW,IAAI,EAAE,CAAC;SACvB;AAED,QAAA,IAAI,KAAK,KAAK,SAAS,EAAE;AACvB,YAAA,MAAM,UAAU,GAAG,EAAE,KAAK,EAAE,IAAI,EAAE,SAAS,EAAE,UAAU,EAAE,EAAE,EAAE;YAC7D,IAAI,GAAG,CAAC,gBAAgB;AAAE,gBAAA,KAAK,CAAC,IAAI,CAAC,UAAU,CAAC;;gBAC3C,KAAK,CAAC,MAAM,CAAC,CAAC,EAAE,CAAC,EAAE,UAAU,CAAC;QACrC;QAEA,MAAM,SAAS,GAAc,EAAE;AAE/B,QAAA,MAAM,WAAW,GAAG,IAAI,GAAG,EAAkB;AAC7C,QAAA,KAAK,IAAI,CAAC,GAAG,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC,IAAI,CAAC,EAAE,EAAE,CAAC,EAAE;YAC1C,MAAM,EAAE,IAAI,EAAE,QAAQ,EAAE,GAAG,KAAK,CAAC,CAAC,CAAC;YACnC,MAAM,UAAU,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,UAAU,IAAI,CAAC;AAC3C,YAAA,MAAM,eAAe,GAAG,CAAC,WAAW,CAAC,GAAG,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC;AAC5D,YAAA,WAAW,CAAC,GAAG,CAAC,QAAQ,EAAE,eAAe,CAAC;AAC1C,YAAA,MAAM,QAAQ,GAAG,CAAA,EAAG,QAAQ,CAAA,CAAA,EAAI,eAAe,EAAE;AACjD,YAAA,uBAAuB,CAAC,GAAG,CAAC,QAAQ,EAAE,CAAC,uBAAuB,CAAC,GAAG,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;AACvF,YAAA,WAAW,CAAC,GAAG,CACb,QAAQ,EACR,IAAI,CAAC,GAAG,CACN,WAAW,CAAC,GAAG,CAAC,QAAQ,CAAC,IAAI,MAAM,CAAC,iBAAiB,EACrD,UAAU,GAAG,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,IAAI,eAAe,CAClD,CACF;AACD,YAAA,SAAS,CAAC,IAAI,CAAC,EAAE,GAAG,KAAK,CAAC,CAAC,CAAC,EAAE,QAAQ,EAAE,eAAe,EAAE,eAAe,EAAE,CAAC;QAC7E;QACA,SAAS,CAAC,OAAO,EAAE;QACnB,OAAO;YACL,KAAK;YACL,IAAI;YACJ,KAAK;YACL,SAAS;SACV;AACH,IAAA,CAAC,CAAC;;IAGF,MAAM,SAAS,GAAa,EAAE;;IAE9B,MAAM,cAAc,GAAa,EAAE;AAEnC,IAAA,MAAM,cAAc,GAAG,CAAC,GAAG,WAAW,CAAC;;IAEvC,cAAc,CAAC,IAAI,CAAC,CAAC,GAAG,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,KAAK,EAAE,GAAG,EAAE,CAAC;AAEhD,IAAA,KAAK,MAAM,CAAC,QAAQ,CAAC,IAAI,cAAc,EAAE;AACvC,QAAA,IAAI,QAAQ,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,uBAAuB,CAAC,GAAG,CAAC,QAAQ,CAAC,KAAK,MAAM,CAAC,MAAM;AACpF,YAAA,SAAS,CAAC,IAAI,CAAC,QAAQ,CAAC;;AACrB,YAAA,cAAc,CAAC,IAAI,CAAC,QAAQ,CAAC;IACpC;IAEA,MAAM,SAAS,GAAG,CAAC,aAA0B,EAAE,KAAA,GAAiB,KAAK,KAAI;QACvE,MAAM,MAAM,GAA0B,EAAE;AACxC,QAAA,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,eAAe,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;AAC/C,YAAA,MAAM,CAAC,GAAG,eAAe,CAAC,CAAC,CAAC;AAC5B,YAAA,MAAM,aAAa,GAAG,CAAC,CAAC;AACrB,iBAAA,MAAM,CAAC,CAAC,EAAE,KAAK,aAAa,CAAC,GAAG,CAAC,EAAE,CAAC,QAAQ;AACxC,oBAAC,kBAAkB,IAAI,kBAAkB,CAAC,GAAG,CAAC,EAAE,CAAC,IAAI,CAAC,CAAC,CAAC;AAC/D,YAAA,IAAI,aAAa,CAAC,MAAM,KAAK,CAAC,EAAE;AAC9B,gBAAA,IAAI,KAAK;oBACP,MAAM,CAAC,IAAI,CAAC;AACV,wBAAA,KAAK,EAAE,WAAW;wBAClB,KAAK,EAAE,CAAC,CAAC,KAAK;AACe,qBAAA,CAAC;;AAC7B,oBAAA,OAAO,SAAS;YACvB;YACA,MAAM,QAAQ,GAAG;iBACd,GAAG,CAAC,CAAC,EAAE,KAAK,EAAE,CAAC,KAAK,CAAC;AACxB,YAAA,MAAM,GAAG,GAAG,GAAG,CAAC,SAAS,IAAI,KAAK;YAClC,MAAM,CAAC,IAAI,CAAC;AACV,gBAAA,KAAK,EAAE,QAAQ,CAAC,IAAI,CAAC,GAAG,CAAC;gBACzB,KAAK,EAAE,CAAC,CAAC,KAAK;AACe,aAAA,CAAC;QAClC;AACA,QAAA,OAAO,MAAM;AACf,IAAA,CAAC;AAED,IAAA,IAAI,SAAS,CAAC,MAAM,KAAK,CAAC,EAAE;AAC1B,QAAA,IAAI,cAAc,CAAC,MAAM,KAAK,CAAC;AAAE,YAAA,MAAM,IAAI,KAAK,CAAC,gEAAgE,CAAC;QAClH,OAAO,SAAS,CAAC,IAAI,GAAG,CAAC,aAAa,CAAC,EAAE,IAAI,CAAE;IACjD;;;;;;;;;;IAWA,IAAI,aAAa,GAAG,CAAC;AACrB,IAAA,IAAI,cAAc,GAAG,EAAE;AACvB,IAAA,OAAO,aAAa,GAAG,SAAS,CAAC,MAAM,EAAE;AACvC,QAAA,MAAM,UAAU,GAAG,IAAI,GAAG,EAAU;QACpC,IAAI,GAAG,CAAC,kBAAkB;AAAE,YAAA,UAAU,CAAC,GAAG,CAAC,aAAa,CAAC;QACzD,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,aAAa,EAAE,EAAE,CAAC;YAAE,UAAU,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC;QACpE,IAAI,cAAc,IAAI,CAAC;YACrB,UAAU,CAAC,GAAG,CAAC,SAAS,CAAC,cAAc,CAAC,CAAC;AAE3C,QAAA,MAAM,eAAe,GAAG,SAAS,CAAC,UAAU,CAAC;;QAG7C,IAAI,eAAe,KAAK,SAAS,IAAI,IAAI,GAAG,CAAC,eAAe,CAAC,GAAG,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,IAAI,KAAK,MAAM,CAAC,MAAM;AAAE,YAAA,OAAO,eAAe;AAEhI,QAAA,cAAc,EAAE;AAChB,QAAA,IAAI,cAAc,IAAI,SAAS,CAAC,MAAM,EAAE;AACtC,YAAA,aAAa,EAAE;YACf,cAAc,GAAG,aAAa;QAChC;IACF;AAEA,IAAA,OAAO,SAAS,CAAC,IAAI,GAAG,CAAC,CAAC,GAAG,SAAS,EAAE,GAAG,cAAc,CAAC,CAAC,EAAE,IAAI,CAAE;AACrE;;;;;;"}
|
|
1
|
+
{"version":3,"file":"label.cjs","sources":["../../../src/render/util/label.ts"],"sourcesContent":["import { Annotation, parseJson, readAnnotation, type PObjectSpec } from '@milaboratories/pl-model-common';\nimport { z } from 'zod';\n\nexport type RecordsWithLabel<T> = {\n value: T;\n label: string;\n};\n\nexport type LabelDerivationOps = {\n /** Force inclusion of native column label */\n includeNativeLabel?: boolean;\n /** Separator to use between label parts (\" / \" by default) */\n separator?: string;\n /** If true, label will be added as suffix (at the end of the generated label). By default label added as a prefix. */\n addLabelAsSuffix?: boolean;\n /** Trace elements list that will be forced to be included in the label. */\n forceTraceElements?: string[];\n};\n\nexport const TraceEntry = z.object({\n type: z.string(),\n importance: z.number().optional(),\n id: z.string().optional(),\n label: z.string(),\n});\nexport type TraceEntry = z.infer<typeof TraceEntry>;\ntype FullTraceEntry = TraceEntry & { fullType: string; occurrenceIndex: number };\n\nexport const Trace = z.array(TraceEntry);\nexport type Trace = z.infer<typeof Trace>;\ntype FullTrace = FullTraceEntry[];\n\n// Define the possible return types for the specExtractor function\ntype SpecExtractorResult = PObjectSpec | {\n spec: PObjectSpec;\n prefixTrace?: TraceEntry[];\n suffixTrace?: TraceEntry[];\n};\n\nconst DistancePenalty = 0.001;\n\nconst LabelType = '__LABEL__';\nconst LabelTypeFull = '__LABEL__@1';\n\nexport function deriveLabels<T>(\n values: T[],\n specExtractor: (obj: T) => SpecExtractorResult,\n ops: LabelDerivationOps = {},\n): RecordsWithLabel<T>[] {\n const importances = new Map<string, number>();\n\n const forceTraceElements = (ops.forceTraceElements !== undefined && ops.forceTraceElements.length > 0)\n ? new Set(ops.forceTraceElements)\n : undefined;\n\n // number of times certain type occurred among all of the\n const numberOfRecordsWithType = new Map<string, number>();\n\n const enrichedRecords = values.map((value) => {\n const extractorResult = specExtractor(value);\n let spec: PObjectSpec;\n let prefixTrace: TraceEntry[] | undefined;\n let suffixTrace: TraceEntry[] | undefined;\n\n // Check if the result is the new structure or just PObjectSpec\n if ('spec' in extractorResult && typeof extractorResult.spec === 'object') {\n // It's the new structure { spec, prefixTrace?, suffixTrace? }\n spec = extractorResult.spec;\n prefixTrace = extractorResult.prefixTrace;\n suffixTrace = extractorResult.suffixTrace;\n } else {\n // It's just PObjectSpec\n spec = extractorResult as PObjectSpec;\n }\n\n const label = readAnnotation(spec, Annotation.Label);\n const traceStr = readAnnotation(spec, Annotation.Trace);\n const baseTrace = (traceStr ? Trace.safeParse(parseJson(traceStr)).data : undefined) ?? [];\n\n const trace = [\n ...(prefixTrace ?? []),\n ...baseTrace,\n ...(suffixTrace ?? []),\n ];\n\n if (label !== undefined) {\n const labelEntry = { label, type: LabelType, importance: -2 };\n if (ops.addLabelAsSuffix) trace.push(labelEntry);\n else trace.splice(0, 0, labelEntry);\n }\n\n const fullTrace: FullTrace = [];\n\n const occurrences = new Map<string, number>();\n for (let i = trace.length - 1; i >= 0; --i) {\n const { type: typeName } = trace[i];\n const importance = trace[i].importance ?? 0;\n const occurrenceIndex = (occurrences.get(typeName) ?? 0) + 1;\n occurrences.set(typeName, occurrenceIndex);\n const fullType = `${typeName}@${occurrenceIndex}`;\n numberOfRecordsWithType.set(fullType, (numberOfRecordsWithType.get(fullType) ?? 0) + 1);\n importances.set(\n fullType,\n Math.max(\n importances.get(fullType) ?? Number.NEGATIVE_INFINITY,\n importance - (trace.length - i) * DistancePenalty,\n ),\n );\n fullTrace.push({ ...trace[i], fullType, occurrenceIndex: occurrenceIndex });\n }\n fullTrace.reverse();\n return {\n value,\n spec,\n label,\n fullTrace,\n };\n });\n\n // excluding repeated types (i.e. ..@2, ..@3, etc.) not found in some records\n const mainTypes: string[] = [];\n // repeated types (i.e. ..@2, ..@3, etc.) not found in some records\n const secondaryTypes: string[] = [];\n\n const allTypeRecords = [...importances];\n // sorting: most important types go first\n allTypeRecords.sort(([, i1], [, i2]) => i2 - i1);\n\n for (const [typeName] of allTypeRecords) {\n if (typeName.endsWith('@1') || numberOfRecordsWithType.get(typeName) === values.length)\n mainTypes.push(typeName);\n else secondaryTypes.push(typeName);\n }\n\n const calculate = (includedTypes: Set<string>, force: boolean = false) => {\n const result: RecordsWithLabel<T>[] = [];\n for (let i = 0; i < enrichedRecords.length; i++) {\n const r = enrichedRecords[i];\n const includedTrace = r.fullTrace\n .filter((fm) => includedTypes.has(fm.fullType)\n || (forceTraceElements && forceTraceElements.has(fm.type)));\n if (includedTrace.length === 0) {\n if (force)\n result.push({\n label: 'Unlabeled',\n value: r.value,\n } satisfies RecordsWithLabel<T>);\n else return undefined;\n }\n const labelSet = includedTrace\n .map((fm) => fm.label);\n const sep = ops.separator ?? ' / ';\n result.push({\n label: labelSet.join(sep),\n value: r.value,\n } satisfies RecordsWithLabel<T>);\n }\n return result;\n };\n\n const countUniqueLabels = (result: RecordsWithLabel<T>[] | undefined): number =>\n result === undefined ? 0 : new Set(result.map((c) => c.label)).size;\n\n // Post-processing: try removing types one by one (lowest importance first) to minimize the label set\n // Accepts removal if it doesn't decrease the number of unique labels (cardinality)\n const minimizeTypeSet = (typeSet: Set<string>): Set<string> => {\n const initialResult = calculate(typeSet);\n if (initialResult === undefined) {\n return typeSet;\n }\n const currentCardinality = countUniqueLabels(initialResult);\n\n // Get types sorted by importance ascending (lowest first), excluding forced elements\n const removableSorted = [...typeSet]\n .filter((t) =>\n !forceTraceElements?.has(t.split('@')[0])\n && !(ops.includeNativeLabel && t === LabelTypeFull))\n .sort((a, b) => (importances.get(a) ?? 0) - (importances.get(b) ?? 0));\n\n for (const typeToRemove of removableSorted) {\n const reducedSet = new Set(typeSet);\n reducedSet.delete(typeToRemove);\n const candidateResult = calculate(reducedSet);\n if (candidateResult !== undefined && countUniqueLabels(candidateResult) >= currentCardinality) {\n typeSet.delete(typeToRemove);\n }\n }\n return typeSet;\n };\n\n if (mainTypes.length === 0) {\n if (secondaryTypes.length !== 0) throw new Error('Non-empty secondary types list while main types list is empty.');\n return calculate(new Set(LabelTypeFull), true)!;\n }\n\n //\n // includedTypes = 2\n // * *\n // T0 T1 T2 T3 T4 T5\n // *\n // additionalType = 3\n //\n // Resulting set: T0, T1, T3\n //\n let includedTypes = 0;\n let additionalType = -1;\n while (includedTypes < mainTypes.length) {\n const currentSet = new Set<string>();\n if (ops.includeNativeLabel) currentSet.add(LabelTypeFull);\n for (let i = 0; i < includedTypes; ++i) currentSet.add(mainTypes[i]);\n if (additionalType >= 0)\n currentSet.add(mainTypes[additionalType]);\n\n const candidateResult = calculate(currentSet);\n\n if (candidateResult !== undefined && countUniqueLabels(candidateResult) === values.length) {\n minimizeTypeSet(currentSet);\n return calculate(currentSet)!;\n }\n\n additionalType++;\n if (additionalType >= mainTypes.length) {\n includedTypes++;\n additionalType = includedTypes;\n }\n }\n\n // Fallback: include all types, then try to minimize\n const fallbackSet = new Set([...mainTypes, ...secondaryTypes]);\n minimizeTypeSet(fallbackSet);\n return calculate(fallbackSet, true)!;\n}\n"],"names":["z","readAnnotation","Annotation","parseJson"],"mappings":";;;;;AAmBO,MAAM,UAAU,GAAGA,KAAC,CAAC,MAAM,CAAC;AACjC,IAAA,IAAI,EAAEA,KAAC,CAAC,MAAM,EAAE;AAChB,IAAA,UAAU,EAAEA,KAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;AACjC,IAAA,EAAE,EAAEA,KAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;AACzB,IAAA,KAAK,EAAEA,KAAC,CAAC,MAAM,EAAE;AAClB,CAAA;AAIM,MAAM,KAAK,GAAGA,KAAC,CAAC,KAAK,CAAC,UAAU;AAWvC,MAAM,eAAe,GAAG,KAAK;AAE7B,MAAM,SAAS,GAAG,WAAW;AAC7B,MAAM,aAAa,GAAG,aAAa;AAE7B,SAAU,YAAY,CAC1B,MAAW,EACX,aAA8C,EAC9C,MAA0B,EAAE,EAAA;AAE5B,IAAA,MAAM,WAAW,GAAG,IAAI,GAAG,EAAkB;AAE7C,IAAA,MAAM,kBAAkB,GAAG,CAAC,GAAG,CAAC,kBAAkB,KAAK,SAAS,IAAI,GAAG,CAAC,kBAAkB,CAAC,MAAM,GAAG,CAAC;AACnG,UAAE,IAAI,GAAG,CAAC,GAAG,CAAC,kBAAkB;UAC9B,SAAS;;AAGb,IAAA,MAAM,uBAAuB,GAAG,IAAI,GAAG,EAAkB;IAEzD,MAAM,eAAe,GAAG,MAAM,CAAC,GAAG,CAAC,CAAC,KAAK,KAAI;AAC3C,QAAA,MAAM,eAAe,GAAG,aAAa,CAAC,KAAK,CAAC;AAC5C,QAAA,IAAI,IAAiB;AACrB,QAAA,IAAI,WAAqC;AACzC,QAAA,IAAI,WAAqC;;QAGzC,IAAI,MAAM,IAAI,eAAe,IAAI,OAAO,eAAe,CAAC,IAAI,KAAK,QAAQ,EAAE;;AAEzE,YAAA,IAAI,GAAG,eAAe,CAAC,IAAI;AAC3B,YAAA,WAAW,GAAG,eAAe,CAAC,WAAW;AACzC,YAAA,WAAW,GAAG,eAAe,CAAC,WAAW;QAC3C;aAAO;;YAEL,IAAI,GAAG,eAA8B;QACvC;QAEA,MAAM,KAAK,GAAGC,4BAAc,CAAC,IAAI,EAAEC,wBAAU,CAAC,KAAK,CAAC;QACpD,MAAM,QAAQ,GAAGD,4BAAc,CAAC,IAAI,EAAEC,wBAAU,CAAC,KAAK,CAAC;QACvD,MAAM,SAAS,GAAG,CAAC,QAAQ,GAAG,KAAK,CAAC,SAAS,CAACC,uBAAS,CAAC,QAAQ,CAAC,CAAC,CAAC,IAAI,GAAG,SAAS,KAAK,EAAE;AAE1F,QAAA,MAAM,KAAK,GAAG;AACZ,YAAA,IAAI,WAAW,IAAI,EAAE,CAAC;AACtB,YAAA,GAAG,SAAS;AACZ,YAAA,IAAI,WAAW,IAAI,EAAE,CAAC;SACvB;AAED,QAAA,IAAI,KAAK,KAAK,SAAS,EAAE;AACvB,YAAA,MAAM,UAAU,GAAG,EAAE,KAAK,EAAE,IAAI,EAAE,SAAS,EAAE,UAAU,EAAE,EAAE,EAAE;YAC7D,IAAI,GAAG,CAAC,gBAAgB;AAAE,gBAAA,KAAK,CAAC,IAAI,CAAC,UAAU,CAAC;;gBAC3C,KAAK,CAAC,MAAM,CAAC,CAAC,EAAE,CAAC,EAAE,UAAU,CAAC;QACrC;QAEA,MAAM,SAAS,GAAc,EAAE;AAE/B,QAAA,MAAM,WAAW,GAAG,IAAI,GAAG,EAAkB;AAC7C,QAAA,KAAK,IAAI,CAAC,GAAG,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC,IAAI,CAAC,EAAE,EAAE,CAAC,EAAE;YAC1C,MAAM,EAAE,IAAI,EAAE,QAAQ,EAAE,GAAG,KAAK,CAAC,CAAC,CAAC;YACnC,MAAM,UAAU,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,UAAU,IAAI,CAAC;AAC3C,YAAA,MAAM,eAAe,GAAG,CAAC,WAAW,CAAC,GAAG,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC;AAC5D,YAAA,WAAW,CAAC,GAAG,CAAC,QAAQ,EAAE,eAAe,CAAC;AAC1C,YAAA,MAAM,QAAQ,GAAG,CAAA,EAAG,QAAQ,CAAA,CAAA,EAAI,eAAe,EAAE;AACjD,YAAA,uBAAuB,CAAC,GAAG,CAAC,QAAQ,EAAE,CAAC,uBAAuB,CAAC,GAAG,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;AACvF,YAAA,WAAW,CAAC,GAAG,CACb,QAAQ,EACR,IAAI,CAAC,GAAG,CACN,WAAW,CAAC,GAAG,CAAC,QAAQ,CAAC,IAAI,MAAM,CAAC,iBAAiB,EACrD,UAAU,GAAG,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,IAAI,eAAe,CAClD,CACF;AACD,YAAA,SAAS,CAAC,IAAI,CAAC,EAAE,GAAG,KAAK,CAAC,CAAC,CAAC,EAAE,QAAQ,EAAE,eAAe,EAAE,eAAe,EAAE,CAAC;QAC7E;QACA,SAAS,CAAC,OAAO,EAAE;QACnB,OAAO;YACL,KAAK;YACL,IAAI;YACJ,KAAK;YACL,SAAS;SACV;AACH,IAAA,CAAC,CAAC;;IAGF,MAAM,SAAS,GAAa,EAAE;;IAE9B,MAAM,cAAc,GAAa,EAAE;AAEnC,IAAA,MAAM,cAAc,GAAG,CAAC,GAAG,WAAW,CAAC;;IAEvC,cAAc,CAAC,IAAI,CAAC,CAAC,GAAG,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,KAAK,EAAE,GAAG,EAAE,CAAC;AAEhD,IAAA,KAAK,MAAM,CAAC,QAAQ,CAAC,IAAI,cAAc,EAAE;AACvC,QAAA,IAAI,QAAQ,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,uBAAuB,CAAC,GAAG,CAAC,QAAQ,CAAC,KAAK,MAAM,CAAC,MAAM;AACpF,YAAA,SAAS,CAAC,IAAI,CAAC,QAAQ,CAAC;;AACrB,YAAA,cAAc,CAAC,IAAI,CAAC,QAAQ,CAAC;IACpC;IAEA,MAAM,SAAS,GAAG,CAAC,aAA0B,EAAE,KAAA,GAAiB,KAAK,KAAI;QACvE,MAAM,MAAM,GAA0B,EAAE;AACxC,QAAA,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,eAAe,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;AAC/C,YAAA,MAAM,CAAC,GAAG,eAAe,CAAC,CAAC,CAAC;AAC5B,YAAA,MAAM,aAAa,GAAG,CAAC,CAAC;AACrB,iBAAA,MAAM,CAAC,CAAC,EAAE,KAAK,aAAa,CAAC,GAAG,CAAC,EAAE,CAAC,QAAQ;AACxC,oBAAC,kBAAkB,IAAI,kBAAkB,CAAC,GAAG,CAAC,EAAE,CAAC,IAAI,CAAC,CAAC,CAAC;AAC/D,YAAA,IAAI,aAAa,CAAC,MAAM,KAAK,CAAC,EAAE;AAC9B,gBAAA,IAAI,KAAK;oBACP,MAAM,CAAC,IAAI,CAAC;AACV,wBAAA,KAAK,EAAE,WAAW;wBAClB,KAAK,EAAE,CAAC,CAAC,KAAK;AACe,qBAAA,CAAC;;AAC7B,oBAAA,OAAO,SAAS;YACvB;YACA,MAAM,QAAQ,GAAG;iBACd,GAAG,CAAC,CAAC,EAAE,KAAK,EAAE,CAAC,KAAK,CAAC;AACxB,YAAA,MAAM,GAAG,GAAG,GAAG,CAAC,SAAS,IAAI,KAAK;YAClC,MAAM,CAAC,IAAI,CAAC;AACV,gBAAA,KAAK,EAAE,QAAQ,CAAC,IAAI,CAAC,GAAG,CAAC;gBACzB,KAAK,EAAE,CAAC,CAAC,KAAK;AACe,aAAA,CAAC;QAClC;AACA,QAAA,OAAO,MAAM;AACf,IAAA,CAAC;AAED,IAAA,MAAM,iBAAiB,GAAG,CAAC,MAAyC,KAClE,MAAM,KAAK,SAAS,GAAG,CAAC,GAAG,IAAI,GAAG,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,IAAI;;;AAIrE,IAAA,MAAM,eAAe,GAAG,CAAC,OAAoB,KAAiB;AAC5D,QAAA,MAAM,aAAa,GAAG,SAAS,CAAC,OAAO,CAAC;AACxC,QAAA,IAAI,aAAa,KAAK,SAAS,EAAE;AAC/B,YAAA,OAAO,OAAO;QAChB;AACA,QAAA,MAAM,kBAAkB,GAAG,iBAAiB,CAAC,aAAa,CAAC;;AAG3D,QAAA,MAAM,eAAe,GAAG,CAAC,GAAG,OAAO;aAChC,MAAM,CAAC,CAAC,CAAC,KACR,CAAC,kBAAkB,EAAE,GAAG,CAAC,CAAC,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC;eACrC,EAAE,GAAG,CAAC,kBAAkB,IAAI,CAAC,KAAK,aAAa,CAAC;AACpD,aAAA,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,KAAK,CAAC,WAAW,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,WAAW,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC;AAExE,QAAA,KAAK,MAAM,YAAY,IAAI,eAAe,EAAE;AAC1C,YAAA,MAAM,UAAU,GAAG,IAAI,GAAG,CAAC,OAAO,CAAC;AACnC,YAAA,UAAU,CAAC,MAAM,CAAC,YAAY,CAAC;AAC/B,YAAA,MAAM,eAAe,GAAG,SAAS,CAAC,UAAU,CAAC;YAC7C,IAAI,eAAe,KAAK,SAAS,IAAI,iBAAiB,CAAC,eAAe,CAAC,IAAI,kBAAkB,EAAE;AAC7F,gBAAA,OAAO,CAAC,MAAM,CAAC,YAAY,CAAC;YAC9B;QACF;AACA,QAAA,OAAO,OAAO;AAChB,IAAA,CAAC;AAED,IAAA,IAAI,SAAS,CAAC,MAAM,KAAK,CAAC,EAAE;AAC1B,QAAA,IAAI,cAAc,CAAC,MAAM,KAAK,CAAC;AAAE,YAAA,MAAM,IAAI,KAAK,CAAC,gEAAgE,CAAC;QAClH,OAAO,SAAS,CAAC,IAAI,GAAG,CAAC,aAAa,CAAC,EAAE,IAAI,CAAE;IACjD;;;;;;;;;;IAWA,IAAI,aAAa,GAAG,CAAC;AACrB,IAAA,IAAI,cAAc,GAAG,EAAE;AACvB,IAAA,OAAO,aAAa,GAAG,SAAS,CAAC,MAAM,EAAE;AACvC,QAAA,MAAM,UAAU,GAAG,IAAI,GAAG,EAAU;QACpC,IAAI,GAAG,CAAC,kBAAkB;AAAE,YAAA,UAAU,CAAC,GAAG,CAAC,aAAa,CAAC;QACzD,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,aAAa,EAAE,EAAE,CAAC;YAAE,UAAU,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC;QACpE,IAAI,cAAc,IAAI,CAAC;YACrB,UAAU,CAAC,GAAG,CAAC,SAAS,CAAC,cAAc,CAAC,CAAC;AAE3C,QAAA,MAAM,eAAe,GAAG,SAAS,CAAC,UAAU,CAAC;AAE7C,QAAA,IAAI,eAAe,KAAK,SAAS,IAAI,iBAAiB,CAAC,eAAe,CAAC,KAAK,MAAM,CAAC,MAAM,EAAE;YACzF,eAAe,CAAC,UAAU,CAAC;AAC3B,YAAA,OAAO,SAAS,CAAC,UAAU,CAAE;QAC/B;AAEA,QAAA,cAAc,EAAE;AAChB,QAAA,IAAI,cAAc,IAAI,SAAS,CAAC,MAAM,EAAE;AACtC,YAAA,aAAa,EAAE;YACf,cAAc,GAAG,aAAa;QAChC;IACF;;AAGA,IAAA,MAAM,WAAW,GAAG,IAAI,GAAG,CAAC,CAAC,GAAG,SAAS,EAAE,GAAG,cAAc,CAAC,CAAC;IAC9D,eAAe,CAAC,WAAW,CAAC;AAC5B,IAAA,OAAO,SAAS,CAAC,WAAW,EAAE,IAAI,CAAE;AACtC;;;;;;"}
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"label.d.ts","sourceRoot":"","sources":["../../../src/render/util/label.ts"],"names":[],"mappings":"AAAA,OAAO,EAAyC,KAAK,WAAW,EAAE,MAAM,iCAAiC,CAAC;AAC1G,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AAExB,MAAM,MAAM,gBAAgB,CAAC,CAAC,IAAI;IAChC,KAAK,EAAE,CAAC,CAAC;IACT,KAAK,EAAE,MAAM,CAAC;CACf,CAAC;AAEF,MAAM,MAAM,kBAAkB,GAAG;IAC/B,6CAA6C;IAC7C,kBAAkB,CAAC,EAAE,OAAO,CAAC;IAC7B,8DAA8D;IAC9D,SAAS,CAAC,EAAE,MAAM,CAAC;IACnB,sHAAsH;IACtH,gBAAgB,CAAC,EAAE,OAAO,CAAC;IAC3B,2EAA2E;IAC3E,kBAAkB,CAAC,EAAE,MAAM,EAAE,CAAC;CAC/B,CAAC;AAEF,eAAO,MAAM,UAAU;;;;;;;;;;;;;;;EAKrB,CAAC;AACH,MAAM,MAAM,UAAU,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,UAAU,CAAC,CAAC;AAGpD,eAAO,MAAM,KAAK;;;;;;;;;;;;;;;WAAsB,CAAC;AACzC,MAAM,MAAM,KAAK,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,KAAK,CAAC,CAAC;AAI1C,KAAK,mBAAmB,GAAG,WAAW,GAAG;IACvC,IAAI,EAAE,WAAW,CAAC;IAClB,WAAW,CAAC,EAAE,UAAU,EAAE,CAAC;IAC3B,WAAW,CAAC,EAAE,UAAU,EAAE,CAAC;CAC5B,CAAC;AAOF,wBAAgB,YAAY,CAAC,CAAC,EAC5B,MAAM,EAAE,CAAC,EAAE,EACX,aAAa,EAAE,CAAC,GAAG,EAAE,CAAC,KAAK,mBAAmB,EAC9C,GAAG,GAAE,kBAAuB,GAC3B,gBAAgB,CAAC,CAAC,CAAC,EAAE,
|
|
1
|
+
{"version":3,"file":"label.d.ts","sourceRoot":"","sources":["../../../src/render/util/label.ts"],"names":[],"mappings":"AAAA,OAAO,EAAyC,KAAK,WAAW,EAAE,MAAM,iCAAiC,CAAC;AAC1G,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AAExB,MAAM,MAAM,gBAAgB,CAAC,CAAC,IAAI;IAChC,KAAK,EAAE,CAAC,CAAC;IACT,KAAK,EAAE,MAAM,CAAC;CACf,CAAC;AAEF,MAAM,MAAM,kBAAkB,GAAG;IAC/B,6CAA6C;IAC7C,kBAAkB,CAAC,EAAE,OAAO,CAAC;IAC7B,8DAA8D;IAC9D,SAAS,CAAC,EAAE,MAAM,CAAC;IACnB,sHAAsH;IACtH,gBAAgB,CAAC,EAAE,OAAO,CAAC;IAC3B,2EAA2E;IAC3E,kBAAkB,CAAC,EAAE,MAAM,EAAE,CAAC;CAC/B,CAAC;AAEF,eAAO,MAAM,UAAU;;;;;;;;;;;;;;;EAKrB,CAAC;AACH,MAAM,MAAM,UAAU,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,UAAU,CAAC,CAAC;AAGpD,eAAO,MAAM,KAAK;;;;;;;;;;;;;;;WAAsB,CAAC;AACzC,MAAM,MAAM,KAAK,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,KAAK,CAAC,CAAC;AAI1C,KAAK,mBAAmB,GAAG,WAAW,GAAG;IACvC,IAAI,EAAE,WAAW,CAAC;IAClB,WAAW,CAAC,EAAE,UAAU,EAAE,CAAC;IAC3B,WAAW,CAAC,EAAE,UAAU,EAAE,CAAC;CAC5B,CAAC;AAOF,wBAAgB,YAAY,CAAC,CAAC,EAC5B,MAAM,EAAE,CAAC,EAAE,EACX,aAAa,EAAE,CAAC,GAAG,EAAE,CAAC,KAAK,mBAAmB,EAC9C,GAAG,GAAE,kBAAuB,GAC3B,gBAAgB,CAAC,CAAC,CAAC,EAAE,CAuLvB"}
|
|
@@ -108,6 +108,30 @@ function deriveLabels(values, specExtractor, ops = {}) {
|
|
|
108
108
|
}
|
|
109
109
|
return result;
|
|
110
110
|
};
|
|
111
|
+
const countUniqueLabels = (result) => result === undefined ? 0 : new Set(result.map((c) => c.label)).size;
|
|
112
|
+
// Post-processing: try removing types one by one (lowest importance first) to minimize the label set
|
|
113
|
+
// Accepts removal if it doesn't decrease the number of unique labels (cardinality)
|
|
114
|
+
const minimizeTypeSet = (typeSet) => {
|
|
115
|
+
const initialResult = calculate(typeSet);
|
|
116
|
+
if (initialResult === undefined) {
|
|
117
|
+
return typeSet;
|
|
118
|
+
}
|
|
119
|
+
const currentCardinality = countUniqueLabels(initialResult);
|
|
120
|
+
// Get types sorted by importance ascending (lowest first), excluding forced elements
|
|
121
|
+
const removableSorted = [...typeSet]
|
|
122
|
+
.filter((t) => !forceTraceElements?.has(t.split('@')[0])
|
|
123
|
+
&& !(ops.includeNativeLabel && t === LabelTypeFull))
|
|
124
|
+
.sort((a, b) => (importances.get(a) ?? 0) - (importances.get(b) ?? 0));
|
|
125
|
+
for (const typeToRemove of removableSorted) {
|
|
126
|
+
const reducedSet = new Set(typeSet);
|
|
127
|
+
reducedSet.delete(typeToRemove);
|
|
128
|
+
const candidateResult = calculate(reducedSet);
|
|
129
|
+
if (candidateResult !== undefined && countUniqueLabels(candidateResult) >= currentCardinality) {
|
|
130
|
+
typeSet.delete(typeToRemove);
|
|
131
|
+
}
|
|
132
|
+
}
|
|
133
|
+
return typeSet;
|
|
134
|
+
};
|
|
111
135
|
if (mainTypes.length === 0) {
|
|
112
136
|
if (secondaryTypes.length !== 0)
|
|
113
137
|
throw new Error('Non-empty secondary types list while main types list is empty.');
|
|
@@ -133,16 +157,20 @@ function deriveLabels(values, specExtractor, ops = {}) {
|
|
|
133
157
|
if (additionalType >= 0)
|
|
134
158
|
currentSet.add(mainTypes[additionalType]);
|
|
135
159
|
const candidateResult = calculate(currentSet);
|
|
136
|
-
|
|
137
|
-
|
|
138
|
-
return
|
|
160
|
+
if (candidateResult !== undefined && countUniqueLabels(candidateResult) === values.length) {
|
|
161
|
+
minimizeTypeSet(currentSet);
|
|
162
|
+
return calculate(currentSet);
|
|
163
|
+
}
|
|
139
164
|
additionalType++;
|
|
140
165
|
if (additionalType >= mainTypes.length) {
|
|
141
166
|
includedTypes++;
|
|
142
167
|
additionalType = includedTypes;
|
|
143
168
|
}
|
|
144
169
|
}
|
|
145
|
-
|
|
170
|
+
// Fallback: include all types, then try to minimize
|
|
171
|
+
const fallbackSet = new Set([...mainTypes, ...secondaryTypes]);
|
|
172
|
+
minimizeTypeSet(fallbackSet);
|
|
173
|
+
return calculate(fallbackSet, true);
|
|
146
174
|
}
|
|
147
175
|
|
|
148
176
|
export { Trace, TraceEntry, deriveLabels };
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"label.js","sources":["../../../src/render/util/label.ts"],"sourcesContent":["import { Annotation, parseJson, readAnnotation, type PObjectSpec } from '@milaboratories/pl-model-common';\nimport { z } from 'zod';\n\nexport type RecordsWithLabel<T> = {\n value: T;\n label: string;\n};\n\nexport type LabelDerivationOps = {\n /** Force inclusion of native column label */\n includeNativeLabel?: boolean;\n /** Separator to use between label parts (\" / \" by default) */\n separator?: string;\n /** If true, label will be added as suffix (at the end of the generated label). By default label added as a prefix. */\n addLabelAsSuffix?: boolean;\n /** Trace elements list that will be forced to be included in the label. */\n forceTraceElements?: string[];\n};\n\nexport const TraceEntry = z.object({\n type: z.string(),\n importance: z.number().optional(),\n id: z.string().optional(),\n label: z.string(),\n});\nexport type TraceEntry = z.infer<typeof TraceEntry>;\ntype FullTraceEntry = TraceEntry & { fullType: string; occurrenceIndex: number };\n\nexport const Trace = z.array(TraceEntry);\nexport type Trace = z.infer<typeof Trace>;\ntype FullTrace = FullTraceEntry[];\n\n// Define the possible return types for the specExtractor function\ntype SpecExtractorResult = PObjectSpec | {\n spec: PObjectSpec;\n prefixTrace?: TraceEntry[];\n suffixTrace?: TraceEntry[];\n};\n\nconst DistancePenalty = 0.001;\n\nconst LabelType = '__LABEL__';\nconst LabelTypeFull = '__LABEL__@1';\n\nexport function deriveLabels<T>(\n values: T[],\n specExtractor: (obj: T) => SpecExtractorResult,\n ops: LabelDerivationOps = {},\n): RecordsWithLabel<T>[] {\n const importances = new Map<string, number>();\n\n const forceTraceElements = (ops.forceTraceElements !== undefined && ops.forceTraceElements.length > 0)\n ? new Set(ops.forceTraceElements)\n : undefined;\n\n // number of times certain type occurred among all of the\n const numberOfRecordsWithType = new Map<string, number>();\n\n const enrichedRecords = values.map((value) => {\n const extractorResult = specExtractor(value);\n let spec: PObjectSpec;\n let prefixTrace: TraceEntry[] | undefined;\n let suffixTrace: TraceEntry[] | undefined;\n\n // Check if the result is the new structure or just PObjectSpec\n if ('spec' in extractorResult && typeof extractorResult.spec === 'object') {\n // It's the new structure { spec, prefixTrace?, suffixTrace? }\n spec = extractorResult.spec;\n prefixTrace = extractorResult.prefixTrace;\n suffixTrace = extractorResult.suffixTrace;\n } else {\n // It's just PObjectSpec\n spec = extractorResult as PObjectSpec;\n }\n\n const label = readAnnotation(spec, Annotation.Label);\n const traceStr = readAnnotation(spec, Annotation.Trace);\n const baseTrace = (traceStr ? Trace.safeParse(parseJson(traceStr)).data : undefined) ?? [];\n\n const trace = [\n ...(prefixTrace ?? []),\n ...baseTrace,\n ...(suffixTrace ?? []),\n ];\n\n if (label !== undefined) {\n const labelEntry = { label, type: LabelType, importance: -2 };\n if (ops.addLabelAsSuffix) trace.push(labelEntry);\n else trace.splice(0, 0, labelEntry);\n }\n\n const fullTrace: FullTrace = [];\n\n const occurrences = new Map<string, number>();\n for (let i = trace.length - 1; i >= 0; --i) {\n const { type: typeName } = trace[i];\n const importance = trace[i].importance ?? 0;\n const occurrenceIndex = (occurrences.get(typeName) ?? 0) + 1;\n occurrences.set(typeName, occurrenceIndex);\n const fullType = `${typeName}@${occurrenceIndex}`;\n numberOfRecordsWithType.set(fullType, (numberOfRecordsWithType.get(fullType) ?? 0) + 1);\n importances.set(\n fullType,\n Math.max(\n importances.get(fullType) ?? Number.NEGATIVE_INFINITY,\n importance - (trace.length - i) * DistancePenalty,\n ),\n );\n fullTrace.push({ ...trace[i], fullType, occurrenceIndex: occurrenceIndex });\n }\n fullTrace.reverse();\n return {\n value,\n spec,\n label,\n fullTrace,\n };\n });\n\n // excluding repeated types (i.e. ..@2, ..@3, etc.) not found in some records\n const mainTypes: string[] = [];\n // repeated types (i.e. ..@2, ..@3, etc.) not found in some records\n const secondaryTypes: string[] = [];\n\n const allTypeRecords = [...importances];\n // sorting: most important types go first\n allTypeRecords.sort(([, i1], [, i2]) => i2 - i1);\n\n for (const [typeName] of allTypeRecords) {\n if (typeName.endsWith('@1') || numberOfRecordsWithType.get(typeName) === values.length)\n mainTypes.push(typeName);\n else secondaryTypes.push(typeName);\n }\n\n const calculate = (includedTypes: Set<string>, force: boolean = false) => {\n const result: RecordsWithLabel<T>[] = [];\n for (let i = 0; i < enrichedRecords.length; i++) {\n const r = enrichedRecords[i];\n const includedTrace = r.fullTrace\n .filter((fm) => includedTypes.has(fm.fullType)\n || (forceTraceElements && forceTraceElements.has(fm.type)));\n if (includedTrace.length === 0) {\n if (force)\n result.push({\n label: 'Unlabeled',\n value: r.value,\n } satisfies RecordsWithLabel<T>);\n else return undefined;\n }\n const labelSet = includedTrace\n .map((fm) => fm.label);\n const sep = ops.separator ?? ' / ';\n result.push({\n label: labelSet.join(sep),\n value: r.value,\n } satisfies RecordsWithLabel<T>);\n }\n return result;\n };\n\n if (mainTypes.length === 0) {\n if (secondaryTypes.length !== 0) throw new Error('Non-empty secondary types list while main types list is empty.');\n return calculate(new Set(LabelTypeFull), true)!;\n }\n\n //\n // includedTypes = 2\n // * *\n // T0 T1 T2 T3 T4 T5\n // *\n // additionalType = 3\n //\n // Resulting set: T0, T1, T3\n //\n let includedTypes = 0;\n let additionalType = -1;\n while (includedTypes < mainTypes.length) {\n const currentSet = new Set<string>();\n if (ops.includeNativeLabel) currentSet.add(LabelTypeFull);\n for (let i = 0; i < includedTypes; ++i) currentSet.add(mainTypes[i]);\n if (additionalType >= 0)\n currentSet.add(mainTypes[additionalType]);\n\n const candidateResult = calculate(currentSet);\n\n // checking if labels uniquely separate our records\n if (candidateResult !== undefined && new Set(candidateResult.map((c) => c.label)).size === values.length) return candidateResult;\n\n additionalType++;\n if (additionalType >= mainTypes.length) {\n includedTypes++;\n additionalType = includedTypes;\n }\n }\n\n return calculate(new Set([...mainTypes, ...secondaryTypes]), true)!;\n}\n"],"names":[],"mappings":";;;AAmBO,MAAM,UAAU,GAAG,CAAC,CAAC,MAAM,CAAC;AACjC,IAAA,IAAI,EAAE,CAAC,CAAC,MAAM,EAAE;AAChB,IAAA,UAAU,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;AACjC,IAAA,EAAE,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;AACzB,IAAA,KAAK,EAAE,CAAC,CAAC,MAAM,EAAE;AAClB,CAAA;AAIM,MAAM,KAAK,GAAG,CAAC,CAAC,KAAK,CAAC,UAAU;AAWvC,MAAM,eAAe,GAAG,KAAK;AAE7B,MAAM,SAAS,GAAG,WAAW;AAC7B,MAAM,aAAa,GAAG,aAAa;AAE7B,SAAU,YAAY,CAC1B,MAAW,EACX,aAA8C,EAC9C,MAA0B,EAAE,EAAA;AAE5B,IAAA,MAAM,WAAW,GAAG,IAAI,GAAG,EAAkB;AAE7C,IAAA,MAAM,kBAAkB,GAAG,CAAC,GAAG,CAAC,kBAAkB,KAAK,SAAS,IAAI,GAAG,CAAC,kBAAkB,CAAC,MAAM,GAAG,CAAC;AACnG,UAAE,IAAI,GAAG,CAAC,GAAG,CAAC,kBAAkB;UAC9B,SAAS;;AAGb,IAAA,MAAM,uBAAuB,GAAG,IAAI,GAAG,EAAkB;IAEzD,MAAM,eAAe,GAAG,MAAM,CAAC,GAAG,CAAC,CAAC,KAAK,KAAI;AAC3C,QAAA,MAAM,eAAe,GAAG,aAAa,CAAC,KAAK,CAAC;AAC5C,QAAA,IAAI,IAAiB;AACrB,QAAA,IAAI,WAAqC;AACzC,QAAA,IAAI,WAAqC;;QAGzC,IAAI,MAAM,IAAI,eAAe,IAAI,OAAO,eAAe,CAAC,IAAI,KAAK,QAAQ,EAAE;;AAEzE,YAAA,IAAI,GAAG,eAAe,CAAC,IAAI;AAC3B,YAAA,WAAW,GAAG,eAAe,CAAC,WAAW;AACzC,YAAA,WAAW,GAAG,eAAe,CAAC,WAAW;QAC3C;aAAO;;YAEL,IAAI,GAAG,eAA8B;QACvC;QAEA,MAAM,KAAK,GAAG,cAAc,CAAC,IAAI,EAAE,UAAU,CAAC,KAAK,CAAC;QACpD,MAAM,QAAQ,GAAG,cAAc,CAAC,IAAI,EAAE,UAAU,CAAC,KAAK,CAAC;QACvD,MAAM,SAAS,GAAG,CAAC,QAAQ,GAAG,KAAK,CAAC,SAAS,CAAC,SAAS,CAAC,QAAQ,CAAC,CAAC,CAAC,IAAI,GAAG,SAAS,KAAK,EAAE;AAE1F,QAAA,MAAM,KAAK,GAAG;AACZ,YAAA,IAAI,WAAW,IAAI,EAAE,CAAC;AACtB,YAAA,GAAG,SAAS;AACZ,YAAA,IAAI,WAAW,IAAI,EAAE,CAAC;SACvB;AAED,QAAA,IAAI,KAAK,KAAK,SAAS,EAAE;AACvB,YAAA,MAAM,UAAU,GAAG,EAAE,KAAK,EAAE,IAAI,EAAE,SAAS,EAAE,UAAU,EAAE,EAAE,EAAE;YAC7D,IAAI,GAAG,CAAC,gBAAgB;AAAE,gBAAA,KAAK,CAAC,IAAI,CAAC,UAAU,CAAC;;gBAC3C,KAAK,CAAC,MAAM,CAAC,CAAC,EAAE,CAAC,EAAE,UAAU,CAAC;QACrC;QAEA,MAAM,SAAS,GAAc,EAAE;AAE/B,QAAA,MAAM,WAAW,GAAG,IAAI,GAAG,EAAkB;AAC7C,QAAA,KAAK,IAAI,CAAC,GAAG,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC,IAAI,CAAC,EAAE,EAAE,CAAC,EAAE;YAC1C,MAAM,EAAE,IAAI,EAAE,QAAQ,EAAE,GAAG,KAAK,CAAC,CAAC,CAAC;YACnC,MAAM,UAAU,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,UAAU,IAAI,CAAC;AAC3C,YAAA,MAAM,eAAe,GAAG,CAAC,WAAW,CAAC,GAAG,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC;AAC5D,YAAA,WAAW,CAAC,GAAG,CAAC,QAAQ,EAAE,eAAe,CAAC;AAC1C,YAAA,MAAM,QAAQ,GAAG,CAAA,EAAG,QAAQ,CAAA,CAAA,EAAI,eAAe,EAAE;AACjD,YAAA,uBAAuB,CAAC,GAAG,CAAC,QAAQ,EAAE,CAAC,uBAAuB,CAAC,GAAG,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;AACvF,YAAA,WAAW,CAAC,GAAG,CACb,QAAQ,EACR,IAAI,CAAC,GAAG,CACN,WAAW,CAAC,GAAG,CAAC,QAAQ,CAAC,IAAI,MAAM,CAAC,iBAAiB,EACrD,UAAU,GAAG,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,IAAI,eAAe,CAClD,CACF;AACD,YAAA,SAAS,CAAC,IAAI,CAAC,EAAE,GAAG,KAAK,CAAC,CAAC,CAAC,EAAE,QAAQ,EAAE,eAAe,EAAE,eAAe,EAAE,CAAC;QAC7E;QACA,SAAS,CAAC,OAAO,EAAE;QACnB,OAAO;YACL,KAAK;YACL,IAAI;YACJ,KAAK;YACL,SAAS;SACV;AACH,IAAA,CAAC,CAAC;;IAGF,MAAM,SAAS,GAAa,EAAE;;IAE9B,MAAM,cAAc,GAAa,EAAE;AAEnC,IAAA,MAAM,cAAc,GAAG,CAAC,GAAG,WAAW,CAAC;;IAEvC,cAAc,CAAC,IAAI,CAAC,CAAC,GAAG,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,KAAK,EAAE,GAAG,EAAE,CAAC;AAEhD,IAAA,KAAK,MAAM,CAAC,QAAQ,CAAC,IAAI,cAAc,EAAE;AACvC,QAAA,IAAI,QAAQ,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,uBAAuB,CAAC,GAAG,CAAC,QAAQ,CAAC,KAAK,MAAM,CAAC,MAAM;AACpF,YAAA,SAAS,CAAC,IAAI,CAAC,QAAQ,CAAC;;AACrB,YAAA,cAAc,CAAC,IAAI,CAAC,QAAQ,CAAC;IACpC;IAEA,MAAM,SAAS,GAAG,CAAC,aAA0B,EAAE,KAAA,GAAiB,KAAK,KAAI;QACvE,MAAM,MAAM,GAA0B,EAAE;AACxC,QAAA,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,eAAe,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;AAC/C,YAAA,MAAM,CAAC,GAAG,eAAe,CAAC,CAAC,CAAC;AAC5B,YAAA,MAAM,aAAa,GAAG,CAAC,CAAC;AACrB,iBAAA,MAAM,CAAC,CAAC,EAAE,KAAK,aAAa,CAAC,GAAG,CAAC,EAAE,CAAC,QAAQ;AACxC,oBAAC,kBAAkB,IAAI,kBAAkB,CAAC,GAAG,CAAC,EAAE,CAAC,IAAI,CAAC,CAAC,CAAC;AAC/D,YAAA,IAAI,aAAa,CAAC,MAAM,KAAK,CAAC,EAAE;AAC9B,gBAAA,IAAI,KAAK;oBACP,MAAM,CAAC,IAAI,CAAC;AACV,wBAAA,KAAK,EAAE,WAAW;wBAClB,KAAK,EAAE,CAAC,CAAC,KAAK;AACe,qBAAA,CAAC;;AAC7B,oBAAA,OAAO,SAAS;YACvB;YACA,MAAM,QAAQ,GAAG;iBACd,GAAG,CAAC,CAAC,EAAE,KAAK,EAAE,CAAC,KAAK,CAAC;AACxB,YAAA,MAAM,GAAG,GAAG,GAAG,CAAC,SAAS,IAAI,KAAK;YAClC,MAAM,CAAC,IAAI,CAAC;AACV,gBAAA,KAAK,EAAE,QAAQ,CAAC,IAAI,CAAC,GAAG,CAAC;gBACzB,KAAK,EAAE,CAAC,CAAC,KAAK;AACe,aAAA,CAAC;QAClC;AACA,QAAA,OAAO,MAAM;AACf,IAAA,CAAC;AAED,IAAA,IAAI,SAAS,CAAC,MAAM,KAAK,CAAC,EAAE;AAC1B,QAAA,IAAI,cAAc,CAAC,MAAM,KAAK,CAAC;AAAE,YAAA,MAAM,IAAI,KAAK,CAAC,gEAAgE,CAAC;QAClH,OAAO,SAAS,CAAC,IAAI,GAAG,CAAC,aAAa,CAAC,EAAE,IAAI,CAAE;IACjD;;;;;;;;;;IAWA,IAAI,aAAa,GAAG,CAAC;AACrB,IAAA,IAAI,cAAc,GAAG,EAAE;AACvB,IAAA,OAAO,aAAa,GAAG,SAAS,CAAC,MAAM,EAAE;AACvC,QAAA,MAAM,UAAU,GAAG,IAAI,GAAG,EAAU;QACpC,IAAI,GAAG,CAAC,kBAAkB;AAAE,YAAA,UAAU,CAAC,GAAG,CAAC,aAAa,CAAC;QACzD,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,aAAa,EAAE,EAAE,CAAC;YAAE,UAAU,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC;QACpE,IAAI,cAAc,IAAI,CAAC;YACrB,UAAU,CAAC,GAAG,CAAC,SAAS,CAAC,cAAc,CAAC,CAAC;AAE3C,QAAA,MAAM,eAAe,GAAG,SAAS,CAAC,UAAU,CAAC;;QAG7C,IAAI,eAAe,KAAK,SAAS,IAAI,IAAI,GAAG,CAAC,eAAe,CAAC,GAAG,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,IAAI,KAAK,MAAM,CAAC,MAAM;AAAE,YAAA,OAAO,eAAe;AAEhI,QAAA,cAAc,EAAE;AAChB,QAAA,IAAI,cAAc,IAAI,SAAS,CAAC,MAAM,EAAE;AACtC,YAAA,aAAa,EAAE;YACf,cAAc,GAAG,aAAa;QAChC;IACF;AAEA,IAAA,OAAO,SAAS,CAAC,IAAI,GAAG,CAAC,CAAC,GAAG,SAAS,EAAE,GAAG,cAAc,CAAC,CAAC,EAAE,IAAI,CAAE;AACrE;;;;"}
|
|
1
|
+
{"version":3,"file":"label.js","sources":["../../../src/render/util/label.ts"],"sourcesContent":["import { Annotation, parseJson, readAnnotation, type PObjectSpec } from '@milaboratories/pl-model-common';\nimport { z } from 'zod';\n\nexport type RecordsWithLabel<T> = {\n value: T;\n label: string;\n};\n\nexport type LabelDerivationOps = {\n /** Force inclusion of native column label */\n includeNativeLabel?: boolean;\n /** Separator to use between label parts (\" / \" by default) */\n separator?: string;\n /** If true, label will be added as suffix (at the end of the generated label). By default label added as a prefix. */\n addLabelAsSuffix?: boolean;\n /** Trace elements list that will be forced to be included in the label. */\n forceTraceElements?: string[];\n};\n\nexport const TraceEntry = z.object({\n type: z.string(),\n importance: z.number().optional(),\n id: z.string().optional(),\n label: z.string(),\n});\nexport type TraceEntry = z.infer<typeof TraceEntry>;\ntype FullTraceEntry = TraceEntry & { fullType: string; occurrenceIndex: number };\n\nexport const Trace = z.array(TraceEntry);\nexport type Trace = z.infer<typeof Trace>;\ntype FullTrace = FullTraceEntry[];\n\n// Define the possible return types for the specExtractor function\ntype SpecExtractorResult = PObjectSpec | {\n spec: PObjectSpec;\n prefixTrace?: TraceEntry[];\n suffixTrace?: TraceEntry[];\n};\n\nconst DistancePenalty = 0.001;\n\nconst LabelType = '__LABEL__';\nconst LabelTypeFull = '__LABEL__@1';\n\nexport function deriveLabels<T>(\n values: T[],\n specExtractor: (obj: T) => SpecExtractorResult,\n ops: LabelDerivationOps = {},\n): RecordsWithLabel<T>[] {\n const importances = new Map<string, number>();\n\n const forceTraceElements = (ops.forceTraceElements !== undefined && ops.forceTraceElements.length > 0)\n ? new Set(ops.forceTraceElements)\n : undefined;\n\n // number of times certain type occurred among all of the\n const numberOfRecordsWithType = new Map<string, number>();\n\n const enrichedRecords = values.map((value) => {\n const extractorResult = specExtractor(value);\n let spec: PObjectSpec;\n let prefixTrace: TraceEntry[] | undefined;\n let suffixTrace: TraceEntry[] | undefined;\n\n // Check if the result is the new structure or just PObjectSpec\n if ('spec' in extractorResult && typeof extractorResult.spec === 'object') {\n // It's the new structure { spec, prefixTrace?, suffixTrace? }\n spec = extractorResult.spec;\n prefixTrace = extractorResult.prefixTrace;\n suffixTrace = extractorResult.suffixTrace;\n } else {\n // It's just PObjectSpec\n spec = extractorResult as PObjectSpec;\n }\n\n const label = readAnnotation(spec, Annotation.Label);\n const traceStr = readAnnotation(spec, Annotation.Trace);\n const baseTrace = (traceStr ? Trace.safeParse(parseJson(traceStr)).data : undefined) ?? [];\n\n const trace = [\n ...(prefixTrace ?? []),\n ...baseTrace,\n ...(suffixTrace ?? []),\n ];\n\n if (label !== undefined) {\n const labelEntry = { label, type: LabelType, importance: -2 };\n if (ops.addLabelAsSuffix) trace.push(labelEntry);\n else trace.splice(0, 0, labelEntry);\n }\n\n const fullTrace: FullTrace = [];\n\n const occurrences = new Map<string, number>();\n for (let i = trace.length - 1; i >= 0; --i) {\n const { type: typeName } = trace[i];\n const importance = trace[i].importance ?? 0;\n const occurrenceIndex = (occurrences.get(typeName) ?? 0) + 1;\n occurrences.set(typeName, occurrenceIndex);\n const fullType = `${typeName}@${occurrenceIndex}`;\n numberOfRecordsWithType.set(fullType, (numberOfRecordsWithType.get(fullType) ?? 0) + 1);\n importances.set(\n fullType,\n Math.max(\n importances.get(fullType) ?? Number.NEGATIVE_INFINITY,\n importance - (trace.length - i) * DistancePenalty,\n ),\n );\n fullTrace.push({ ...trace[i], fullType, occurrenceIndex: occurrenceIndex });\n }\n fullTrace.reverse();\n return {\n value,\n spec,\n label,\n fullTrace,\n };\n });\n\n // excluding repeated types (i.e. ..@2, ..@3, etc.) not found in some records\n const mainTypes: string[] = [];\n // repeated types (i.e. ..@2, ..@3, etc.) not found in some records\n const secondaryTypes: string[] = [];\n\n const allTypeRecords = [...importances];\n // sorting: most important types go first\n allTypeRecords.sort(([, i1], [, i2]) => i2 - i1);\n\n for (const [typeName] of allTypeRecords) {\n if (typeName.endsWith('@1') || numberOfRecordsWithType.get(typeName) === values.length)\n mainTypes.push(typeName);\n else secondaryTypes.push(typeName);\n }\n\n const calculate = (includedTypes: Set<string>, force: boolean = false) => {\n const result: RecordsWithLabel<T>[] = [];\n for (let i = 0; i < enrichedRecords.length; i++) {\n const r = enrichedRecords[i];\n const includedTrace = r.fullTrace\n .filter((fm) => includedTypes.has(fm.fullType)\n || (forceTraceElements && forceTraceElements.has(fm.type)));\n if (includedTrace.length === 0) {\n if (force)\n result.push({\n label: 'Unlabeled',\n value: r.value,\n } satisfies RecordsWithLabel<T>);\n else return undefined;\n }\n const labelSet = includedTrace\n .map((fm) => fm.label);\n const sep = ops.separator ?? ' / ';\n result.push({\n label: labelSet.join(sep),\n value: r.value,\n } satisfies RecordsWithLabel<T>);\n }\n return result;\n };\n\n const countUniqueLabels = (result: RecordsWithLabel<T>[] | undefined): number =>\n result === undefined ? 0 : new Set(result.map((c) => c.label)).size;\n\n // Post-processing: try removing types one by one (lowest importance first) to minimize the label set\n // Accepts removal if it doesn't decrease the number of unique labels (cardinality)\n const minimizeTypeSet = (typeSet: Set<string>): Set<string> => {\n const initialResult = calculate(typeSet);\n if (initialResult === undefined) {\n return typeSet;\n }\n const currentCardinality = countUniqueLabels(initialResult);\n\n // Get types sorted by importance ascending (lowest first), excluding forced elements\n const removableSorted = [...typeSet]\n .filter((t) =>\n !forceTraceElements?.has(t.split('@')[0])\n && !(ops.includeNativeLabel && t === LabelTypeFull))\n .sort((a, b) => (importances.get(a) ?? 0) - (importances.get(b) ?? 0));\n\n for (const typeToRemove of removableSorted) {\n const reducedSet = new Set(typeSet);\n reducedSet.delete(typeToRemove);\n const candidateResult = calculate(reducedSet);\n if (candidateResult !== undefined && countUniqueLabels(candidateResult) >= currentCardinality) {\n typeSet.delete(typeToRemove);\n }\n }\n return typeSet;\n };\n\n if (mainTypes.length === 0) {\n if (secondaryTypes.length !== 0) throw new Error('Non-empty secondary types list while main types list is empty.');\n return calculate(new Set(LabelTypeFull), true)!;\n }\n\n //\n // includedTypes = 2\n // * *\n // T0 T1 T2 T3 T4 T5\n // *\n // additionalType = 3\n //\n // Resulting set: T0, T1, T3\n //\n let includedTypes = 0;\n let additionalType = -1;\n while (includedTypes < mainTypes.length) {\n const currentSet = new Set<string>();\n if (ops.includeNativeLabel) currentSet.add(LabelTypeFull);\n for (let i = 0; i < includedTypes; ++i) currentSet.add(mainTypes[i]);\n if (additionalType >= 0)\n currentSet.add(mainTypes[additionalType]);\n\n const candidateResult = calculate(currentSet);\n\n if (candidateResult !== undefined && countUniqueLabels(candidateResult) === values.length) {\n minimizeTypeSet(currentSet);\n return calculate(currentSet)!;\n }\n\n additionalType++;\n if (additionalType >= mainTypes.length) {\n includedTypes++;\n additionalType = includedTypes;\n }\n }\n\n // Fallback: include all types, then try to minimize\n const fallbackSet = new Set([...mainTypes, ...secondaryTypes]);\n minimizeTypeSet(fallbackSet);\n return calculate(fallbackSet, true)!;\n}\n"],"names":[],"mappings":";;;AAmBO,MAAM,UAAU,GAAG,CAAC,CAAC,MAAM,CAAC;AACjC,IAAA,IAAI,EAAE,CAAC,CAAC,MAAM,EAAE;AAChB,IAAA,UAAU,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;AACjC,IAAA,EAAE,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;AACzB,IAAA,KAAK,EAAE,CAAC,CAAC,MAAM,EAAE;AAClB,CAAA;AAIM,MAAM,KAAK,GAAG,CAAC,CAAC,KAAK,CAAC,UAAU;AAWvC,MAAM,eAAe,GAAG,KAAK;AAE7B,MAAM,SAAS,GAAG,WAAW;AAC7B,MAAM,aAAa,GAAG,aAAa;AAE7B,SAAU,YAAY,CAC1B,MAAW,EACX,aAA8C,EAC9C,MAA0B,EAAE,EAAA;AAE5B,IAAA,MAAM,WAAW,GAAG,IAAI,GAAG,EAAkB;AAE7C,IAAA,MAAM,kBAAkB,GAAG,CAAC,GAAG,CAAC,kBAAkB,KAAK,SAAS,IAAI,GAAG,CAAC,kBAAkB,CAAC,MAAM,GAAG,CAAC;AACnG,UAAE,IAAI,GAAG,CAAC,GAAG,CAAC,kBAAkB;UAC9B,SAAS;;AAGb,IAAA,MAAM,uBAAuB,GAAG,IAAI,GAAG,EAAkB;IAEzD,MAAM,eAAe,GAAG,MAAM,CAAC,GAAG,CAAC,CAAC,KAAK,KAAI;AAC3C,QAAA,MAAM,eAAe,GAAG,aAAa,CAAC,KAAK,CAAC;AAC5C,QAAA,IAAI,IAAiB;AACrB,QAAA,IAAI,WAAqC;AACzC,QAAA,IAAI,WAAqC;;QAGzC,IAAI,MAAM,IAAI,eAAe,IAAI,OAAO,eAAe,CAAC,IAAI,KAAK,QAAQ,EAAE;;AAEzE,YAAA,IAAI,GAAG,eAAe,CAAC,IAAI;AAC3B,YAAA,WAAW,GAAG,eAAe,CAAC,WAAW;AACzC,YAAA,WAAW,GAAG,eAAe,CAAC,WAAW;QAC3C;aAAO;;YAEL,IAAI,GAAG,eAA8B;QACvC;QAEA,MAAM,KAAK,GAAG,cAAc,CAAC,IAAI,EAAE,UAAU,CAAC,KAAK,CAAC;QACpD,MAAM,QAAQ,GAAG,cAAc,CAAC,IAAI,EAAE,UAAU,CAAC,KAAK,CAAC;QACvD,MAAM,SAAS,GAAG,CAAC,QAAQ,GAAG,KAAK,CAAC,SAAS,CAAC,SAAS,CAAC,QAAQ,CAAC,CAAC,CAAC,IAAI,GAAG,SAAS,KAAK,EAAE;AAE1F,QAAA,MAAM,KAAK,GAAG;AACZ,YAAA,IAAI,WAAW,IAAI,EAAE,CAAC;AACtB,YAAA,GAAG,SAAS;AACZ,YAAA,IAAI,WAAW,IAAI,EAAE,CAAC;SACvB;AAED,QAAA,IAAI,KAAK,KAAK,SAAS,EAAE;AACvB,YAAA,MAAM,UAAU,GAAG,EAAE,KAAK,EAAE,IAAI,EAAE,SAAS,EAAE,UAAU,EAAE,EAAE,EAAE;YAC7D,IAAI,GAAG,CAAC,gBAAgB;AAAE,gBAAA,KAAK,CAAC,IAAI,CAAC,UAAU,CAAC;;gBAC3C,KAAK,CAAC,MAAM,CAAC,CAAC,EAAE,CAAC,EAAE,UAAU,CAAC;QACrC;QAEA,MAAM,SAAS,GAAc,EAAE;AAE/B,QAAA,MAAM,WAAW,GAAG,IAAI,GAAG,EAAkB;AAC7C,QAAA,KAAK,IAAI,CAAC,GAAG,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC,IAAI,CAAC,EAAE,EAAE,CAAC,EAAE;YAC1C,MAAM,EAAE,IAAI,EAAE,QAAQ,EAAE,GAAG,KAAK,CAAC,CAAC,CAAC;YACnC,MAAM,UAAU,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,UAAU,IAAI,CAAC;AAC3C,YAAA,MAAM,eAAe,GAAG,CAAC,WAAW,CAAC,GAAG,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC;AAC5D,YAAA,WAAW,CAAC,GAAG,CAAC,QAAQ,EAAE,eAAe,CAAC;AAC1C,YAAA,MAAM,QAAQ,GAAG,CAAA,EAAG,QAAQ,CAAA,CAAA,EAAI,eAAe,EAAE;AACjD,YAAA,uBAAuB,CAAC,GAAG,CAAC,QAAQ,EAAE,CAAC,uBAAuB,CAAC,GAAG,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;AACvF,YAAA,WAAW,CAAC,GAAG,CACb,QAAQ,EACR,IAAI,CAAC,GAAG,CACN,WAAW,CAAC,GAAG,CAAC,QAAQ,CAAC,IAAI,MAAM,CAAC,iBAAiB,EACrD,UAAU,GAAG,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,IAAI,eAAe,CAClD,CACF;AACD,YAAA,SAAS,CAAC,IAAI,CAAC,EAAE,GAAG,KAAK,CAAC,CAAC,CAAC,EAAE,QAAQ,EAAE,eAAe,EAAE,eAAe,EAAE,CAAC;QAC7E;QACA,SAAS,CAAC,OAAO,EAAE;QACnB,OAAO;YACL,KAAK;YACL,IAAI;YACJ,KAAK;YACL,SAAS;SACV;AACH,IAAA,CAAC,CAAC;;IAGF,MAAM,SAAS,GAAa,EAAE;;IAE9B,MAAM,cAAc,GAAa,EAAE;AAEnC,IAAA,MAAM,cAAc,GAAG,CAAC,GAAG,WAAW,CAAC;;IAEvC,cAAc,CAAC,IAAI,CAAC,CAAC,GAAG,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,KAAK,EAAE,GAAG,EAAE,CAAC;AAEhD,IAAA,KAAK,MAAM,CAAC,QAAQ,CAAC,IAAI,cAAc,EAAE;AACvC,QAAA,IAAI,QAAQ,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,uBAAuB,CAAC,GAAG,CAAC,QAAQ,CAAC,KAAK,MAAM,CAAC,MAAM;AACpF,YAAA,SAAS,CAAC,IAAI,CAAC,QAAQ,CAAC;;AACrB,YAAA,cAAc,CAAC,IAAI,CAAC,QAAQ,CAAC;IACpC;IAEA,MAAM,SAAS,GAAG,CAAC,aAA0B,EAAE,KAAA,GAAiB,KAAK,KAAI;QACvE,MAAM,MAAM,GAA0B,EAAE;AACxC,QAAA,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,eAAe,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;AAC/C,YAAA,MAAM,CAAC,GAAG,eAAe,CAAC,CAAC,CAAC;AAC5B,YAAA,MAAM,aAAa,GAAG,CAAC,CAAC;AACrB,iBAAA,MAAM,CAAC,CAAC,EAAE,KAAK,aAAa,CAAC,GAAG,CAAC,EAAE,CAAC,QAAQ;AACxC,oBAAC,kBAAkB,IAAI,kBAAkB,CAAC,GAAG,CAAC,EAAE,CAAC,IAAI,CAAC,CAAC,CAAC;AAC/D,YAAA,IAAI,aAAa,CAAC,MAAM,KAAK,CAAC,EAAE;AAC9B,gBAAA,IAAI,KAAK;oBACP,MAAM,CAAC,IAAI,CAAC;AACV,wBAAA,KAAK,EAAE,WAAW;wBAClB,KAAK,EAAE,CAAC,CAAC,KAAK;AACe,qBAAA,CAAC;;AAC7B,oBAAA,OAAO,SAAS;YACvB;YACA,MAAM,QAAQ,GAAG;iBACd,GAAG,CAAC,CAAC,EAAE,KAAK,EAAE,CAAC,KAAK,CAAC;AACxB,YAAA,MAAM,GAAG,GAAG,GAAG,CAAC,SAAS,IAAI,KAAK;YAClC,MAAM,CAAC,IAAI,CAAC;AACV,gBAAA,KAAK,EAAE,QAAQ,CAAC,IAAI,CAAC,GAAG,CAAC;gBACzB,KAAK,EAAE,CAAC,CAAC,KAAK;AACe,aAAA,CAAC;QAClC;AACA,QAAA,OAAO,MAAM;AACf,IAAA,CAAC;AAED,IAAA,MAAM,iBAAiB,GAAG,CAAC,MAAyC,KAClE,MAAM,KAAK,SAAS,GAAG,CAAC,GAAG,IAAI,GAAG,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,IAAI;;;AAIrE,IAAA,MAAM,eAAe,GAAG,CAAC,OAAoB,KAAiB;AAC5D,QAAA,MAAM,aAAa,GAAG,SAAS,CAAC,OAAO,CAAC;AACxC,QAAA,IAAI,aAAa,KAAK,SAAS,EAAE;AAC/B,YAAA,OAAO,OAAO;QAChB;AACA,QAAA,MAAM,kBAAkB,GAAG,iBAAiB,CAAC,aAAa,CAAC;;AAG3D,QAAA,MAAM,eAAe,GAAG,CAAC,GAAG,OAAO;aAChC,MAAM,CAAC,CAAC,CAAC,KACR,CAAC,kBAAkB,EAAE,GAAG,CAAC,CAAC,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC;eACrC,EAAE,GAAG,CAAC,kBAAkB,IAAI,CAAC,KAAK,aAAa,CAAC;AACpD,aAAA,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,KAAK,CAAC,WAAW,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,WAAW,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC;AAExE,QAAA,KAAK,MAAM,YAAY,IAAI,eAAe,EAAE;AAC1C,YAAA,MAAM,UAAU,GAAG,IAAI,GAAG,CAAC,OAAO,CAAC;AACnC,YAAA,UAAU,CAAC,MAAM,CAAC,YAAY,CAAC;AAC/B,YAAA,MAAM,eAAe,GAAG,SAAS,CAAC,UAAU,CAAC;YAC7C,IAAI,eAAe,KAAK,SAAS,IAAI,iBAAiB,CAAC,eAAe,CAAC,IAAI,kBAAkB,EAAE;AAC7F,gBAAA,OAAO,CAAC,MAAM,CAAC,YAAY,CAAC;YAC9B;QACF;AACA,QAAA,OAAO,OAAO;AAChB,IAAA,CAAC;AAED,IAAA,IAAI,SAAS,CAAC,MAAM,KAAK,CAAC,EAAE;AAC1B,QAAA,IAAI,cAAc,CAAC,MAAM,KAAK,CAAC;AAAE,YAAA,MAAM,IAAI,KAAK,CAAC,gEAAgE,CAAC;QAClH,OAAO,SAAS,CAAC,IAAI,GAAG,CAAC,aAAa,CAAC,EAAE,IAAI,CAAE;IACjD;;;;;;;;;;IAWA,IAAI,aAAa,GAAG,CAAC;AACrB,IAAA,IAAI,cAAc,GAAG,EAAE;AACvB,IAAA,OAAO,aAAa,GAAG,SAAS,CAAC,MAAM,EAAE;AACvC,QAAA,MAAM,UAAU,GAAG,IAAI,GAAG,EAAU;QACpC,IAAI,GAAG,CAAC,kBAAkB;AAAE,YAAA,UAAU,CAAC,GAAG,CAAC,aAAa,CAAC;QACzD,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,aAAa,EAAE,EAAE,CAAC;YAAE,UAAU,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC;QACpE,IAAI,cAAc,IAAI,CAAC;YACrB,UAAU,CAAC,GAAG,CAAC,SAAS,CAAC,cAAc,CAAC,CAAC;AAE3C,QAAA,MAAM,eAAe,GAAG,SAAS,CAAC,UAAU,CAAC;AAE7C,QAAA,IAAI,eAAe,KAAK,SAAS,IAAI,iBAAiB,CAAC,eAAe,CAAC,KAAK,MAAM,CAAC,MAAM,EAAE;YACzF,eAAe,CAAC,UAAU,CAAC;AAC3B,YAAA,OAAO,SAAS,CAAC,UAAU,CAAE;QAC/B;AAEA,QAAA,cAAc,EAAE;AAChB,QAAA,IAAI,cAAc,IAAI,SAAS,CAAC,MAAM,EAAE;AACtC,YAAA,aAAa,EAAE;YACf,cAAc,GAAG,aAAa;QAChC;IACF;;AAGA,IAAA,MAAM,WAAW,GAAG,IAAI,GAAG,CAAC,CAAC,GAAG,SAAS,EAAE,GAAG,cAAc,CAAC,CAAC;IAC9D,eAAe,CAAC,WAAW,CAAC;AAC5B,IAAA,OAAO,SAAS,CAAC,WAAW,EAAE,IAAI,CAAE;AACtC;;;;"}
|
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "@platforma-sdk/model",
|
|
3
|
-
"version": "1.51.
|
|
3
|
+
"version": "1.51.9",
|
|
4
4
|
"description": "Platforma.bio SDK / Block Model",
|
|
5
5
|
"type": "module",
|
|
6
6
|
"types": "./dist/index.d.ts",
|
|
@@ -23,8 +23,8 @@
|
|
|
23
23
|
"canonicalize": "~2.1.0",
|
|
24
24
|
"es-toolkit": "^1.39.10",
|
|
25
25
|
"zod": "~3.23.8",
|
|
26
|
-
"@milaboratories/ptabler-expression-js": "1.1.9",
|
|
27
26
|
"@milaboratories/pl-model-common": "1.23.0",
|
|
27
|
+
"@milaboratories/ptabler-expression-js": "1.1.9",
|
|
28
28
|
"@milaboratories/pl-error-like": "1.12.5"
|
|
29
29
|
},
|
|
30
30
|
"devDependencies": {
|
|
@@ -32,9 +32,9 @@
|
|
|
32
32
|
"vitest": "^4.0.16",
|
|
33
33
|
"@vitest/coverage-istanbul": "^4.0.16",
|
|
34
34
|
"fast-json-patch": "^3.1.1",
|
|
35
|
-
"@milaboratories/build-configs": "1.2.
|
|
35
|
+
"@milaboratories/build-configs": "1.2.2",
|
|
36
36
|
"@milaboratories/eslint-config": "1.0.5",
|
|
37
|
-
"@milaboratories/ts-builder": "1.2.
|
|
37
|
+
"@milaboratories/ts-builder": "1.2.2",
|
|
38
38
|
"@milaboratories/helpers": "1.13.0",
|
|
39
39
|
"@milaboratories/ts-configs": "1.2.0"
|
|
40
40
|
},
|
|
@@ -119,6 +119,87 @@ test('test fallback to native labels in label derivation', () => {
|
|
|
119
119
|
]);
|
|
120
120
|
});
|
|
121
121
|
|
|
122
|
+
test.each<{ name: string; traces: Trace[]; labels: string[] }>([
|
|
123
|
+
{
|
|
124
|
+
name: 'removes redundant low-importance type when high-importance alone suffices',
|
|
125
|
+
traces: [
|
|
126
|
+
[
|
|
127
|
+
{ type: 't1', importance: 10, label: 'High1' },
|
|
128
|
+
{ type: 't2', importance: 1, label: 'Low1' }
|
|
129
|
+
],
|
|
130
|
+
[
|
|
131
|
+
{ type: 't1', importance: 10, label: 'High2' },
|
|
132
|
+
{ type: 't2', importance: 1, label: 'Low2' }
|
|
133
|
+
]
|
|
134
|
+
],
|
|
135
|
+
// Both t1 and t2 distinguish, but t2 (low importance) should be removed since t1 alone suffices
|
|
136
|
+
labels: ['High1', 'High2']
|
|
137
|
+
},
|
|
138
|
+
{
|
|
139
|
+
name: 'keeps both types when both are needed for uniqueness',
|
|
140
|
+
traces: [
|
|
141
|
+
[
|
|
142
|
+
{ type: 't1', importance: 10, label: 'A' },
|
|
143
|
+
{ type: 't2', importance: 1, label: 'X' }
|
|
144
|
+
],
|
|
145
|
+
[
|
|
146
|
+
{ type: 't1', importance: 10, label: 'A' },
|
|
147
|
+
{ type: 't2', importance: 1, label: 'Y' }
|
|
148
|
+
],
|
|
149
|
+
[
|
|
150
|
+
{ type: 't1', importance: 10, label: 'B' },
|
|
151
|
+
{ type: 't2', importance: 1, label: 'Y' }
|
|
152
|
+
]
|
|
153
|
+
],
|
|
154
|
+
// Neither t1 nor t2 alone can distinguish all three, need both
|
|
155
|
+
labels: ['A / X', 'A / Y', 'B / Y']
|
|
156
|
+
},
|
|
157
|
+
{
|
|
158
|
+
name: 'removes multiple redundant types greedily',
|
|
159
|
+
traces: [
|
|
160
|
+
[
|
|
161
|
+
{ type: 't1', importance: 100, label: 'Unique1' },
|
|
162
|
+
{ type: 't2', importance: 10, label: 'Same' },
|
|
163
|
+
{ type: 't3', importance: 1, label: 'Same' }
|
|
164
|
+
],
|
|
165
|
+
[
|
|
166
|
+
{ type: 't1', importance: 100, label: 'Unique2' },
|
|
167
|
+
{ type: 't2', importance: 10, label: 'Same' },
|
|
168
|
+
{ type: 't3', importance: 1, label: 'Same' }
|
|
169
|
+
]
|
|
170
|
+
],
|
|
171
|
+
// t1 alone distinguishes; t2 and t3 are redundant and should be removed
|
|
172
|
+
labels: ['Unique1', 'Unique2']
|
|
173
|
+
},
|
|
174
|
+
{
|
|
175
|
+
name: 'fallback case: removes types that do not reduce cardinality',
|
|
176
|
+
traces: [
|
|
177
|
+
// Two columns with identical traces - cannot be distinguished
|
|
178
|
+
[
|
|
179
|
+
{ type: 't1', importance: 100, label: 'A' },
|
|
180
|
+
{ type: 't2', importance: 10, label: 'X' },
|
|
181
|
+
{ type: 't3', importance: 1, label: 'Same' }
|
|
182
|
+
],
|
|
183
|
+
[
|
|
184
|
+
{ type: 't1', importance: 100, label: 'A' },
|
|
185
|
+
{ type: 't2', importance: 10, label: 'X' },
|
|
186
|
+
{ type: 't3', importance: 1, label: 'Same' }
|
|
187
|
+
],
|
|
188
|
+
// Third column is different
|
|
189
|
+
[
|
|
190
|
+
{ type: 't1', importance: 100, label: 'B' },
|
|
191
|
+
{ type: 't2', importance: 10, label: 'Y' },
|
|
192
|
+
{ type: 't3', importance: 1, label: 'Same' }
|
|
193
|
+
]
|
|
194
|
+
],
|
|
195
|
+
// Cannot achieve full uniqueness (2 columns are identical), but t3 (Same) can be removed
|
|
196
|
+
// since it doesn't help distinguish anything. t1 alone gives cardinality 2.
|
|
197
|
+
labels: ['A', 'A', 'B']
|
|
198
|
+
}
|
|
199
|
+
])('test label minimization: $name', ({ traces, labels }) => {
|
|
200
|
+
expect(deriveLabels(tracesToSpecs(traces), (s) => s).map((r) => r.label)).toEqual(labels);
|
|
201
|
+
});
|
|
202
|
+
|
|
122
203
|
test.each<{ name: string; traces: Trace[]; labels: string[]; forceTraceElements: string[] }>([
|
|
123
204
|
{
|
|
124
205
|
name: 'force one element',
|
package/src/render/util/label.ts
CHANGED
|
@@ -158,6 +158,36 @@ export function deriveLabels<T>(
|
|
|
158
158
|
return result;
|
|
159
159
|
};
|
|
160
160
|
|
|
161
|
+
const countUniqueLabels = (result: RecordsWithLabel<T>[] | undefined): number =>
|
|
162
|
+
result === undefined ? 0 : new Set(result.map((c) => c.label)).size;
|
|
163
|
+
|
|
164
|
+
// Post-processing: try removing types one by one (lowest importance first) to minimize the label set
|
|
165
|
+
// Accepts removal if it doesn't decrease the number of unique labels (cardinality)
|
|
166
|
+
const minimizeTypeSet = (typeSet: Set<string>): Set<string> => {
|
|
167
|
+
const initialResult = calculate(typeSet);
|
|
168
|
+
if (initialResult === undefined) {
|
|
169
|
+
return typeSet;
|
|
170
|
+
}
|
|
171
|
+
const currentCardinality = countUniqueLabels(initialResult);
|
|
172
|
+
|
|
173
|
+
// Get types sorted by importance ascending (lowest first), excluding forced elements
|
|
174
|
+
const removableSorted = [...typeSet]
|
|
175
|
+
.filter((t) =>
|
|
176
|
+
!forceTraceElements?.has(t.split('@')[0])
|
|
177
|
+
&& !(ops.includeNativeLabel && t === LabelTypeFull))
|
|
178
|
+
.sort((a, b) => (importances.get(a) ?? 0) - (importances.get(b) ?? 0));
|
|
179
|
+
|
|
180
|
+
for (const typeToRemove of removableSorted) {
|
|
181
|
+
const reducedSet = new Set(typeSet);
|
|
182
|
+
reducedSet.delete(typeToRemove);
|
|
183
|
+
const candidateResult = calculate(reducedSet);
|
|
184
|
+
if (candidateResult !== undefined && countUniqueLabels(candidateResult) >= currentCardinality) {
|
|
185
|
+
typeSet.delete(typeToRemove);
|
|
186
|
+
}
|
|
187
|
+
}
|
|
188
|
+
return typeSet;
|
|
189
|
+
};
|
|
190
|
+
|
|
161
191
|
if (mainTypes.length === 0) {
|
|
162
192
|
if (secondaryTypes.length !== 0) throw new Error('Non-empty secondary types list while main types list is empty.');
|
|
163
193
|
return calculate(new Set(LabelTypeFull), true)!;
|
|
@@ -183,8 +213,10 @@ export function deriveLabels<T>(
|
|
|
183
213
|
|
|
184
214
|
const candidateResult = calculate(currentSet);
|
|
185
215
|
|
|
186
|
-
|
|
187
|
-
|
|
216
|
+
if (candidateResult !== undefined && countUniqueLabels(candidateResult) === values.length) {
|
|
217
|
+
minimizeTypeSet(currentSet);
|
|
218
|
+
return calculate(currentSet)!;
|
|
219
|
+
}
|
|
188
220
|
|
|
189
221
|
additionalType++;
|
|
190
222
|
if (additionalType >= mainTypes.length) {
|
|
@@ -193,5 +225,8 @@ export function deriveLabels<T>(
|
|
|
193
225
|
}
|
|
194
226
|
}
|
|
195
227
|
|
|
196
|
-
|
|
228
|
+
// Fallback: include all types, then try to minimize
|
|
229
|
+
const fallbackSet = new Set([...mainTypes, ...secondaryTypes]);
|
|
230
|
+
minimizeTypeSet(fallbackSet);
|
|
231
|
+
return calculate(fallbackSet, true)!;
|
|
197
232
|
}
|