@cyanheads/eurostat-mcp-server 0.11.1 → 0.11.3

package/dist/services/eurostat-catalogue/dimension-value-index.js

@@ -0,0 +1,410 @@
+/**
+ * @fileoverview The dimension value index search reads when a dataset's own text lacks a
+ * query word: the codes Eurostat's metabase lists for each dataset (`time` excluded), each
+ * labelled in English from the global codelists. Both downloads are gzip files read as
+ * streams, never held whole; the index keeps joined strings and typed arrays.
+ * @module services/eurostat-catalogue/dimension-value-index
+ */
+import { fetchWithTimeout } from '@cyanheads/mcp-ts-core/utils';
+import { getServerConfig } from '../../config/server-config.js';
+import { decodeXmlEntities } from '../xml-entities.js';
+import { continuesWord, foldSearchText } from './search-text.js';
+/** The period dimension, listed for every dataset; its codes are dates, not vocabulary. */
+const TIME_DIMENSION = 'time';
+/** Joins labels and codes. A search word holds no whitespace, so no match spans two labels. */
+const SEPARATOR = '\n';
+/**
+ * A codelist's opening tag, a code's opening tag, an English name, and a code's closing tag.
+ * `<s:Code\b` does not match `<s:Codelist`.
+ */
+const CODELIST_TOKEN = /<s:Codelist\b[^>]*?\sid="([^"]*)"|<s:Code\b[^>]*?\sid="([^"]*)"|<c:Name xml:lang="en">([^<]*)<\/c:Name>|<\/s:Code>/g;
+const CODE_CLOSE = '</s:Code>';
+/** The codelist dump's root element closes the document; a stream without it was cut short. */
+const DOCUMENT_CLOSE = '</m:Structure>';
+/**
+ * A copy of `text` in a new string. A string cut from a decoded chunk keeps the whole chunk
+ * alive, and on JavaScriptCore so does a concatenate-then-slice copy, which slices the
+ * original again; a join writes new characters on every runtime.
+ */
+function copyOf(text) {
+    return [text, ''].join(SEPARATOR).slice(0, -SEPARATOR.length);
+}
+/** Offsets of each part joined by {@link SEPARATOR}, plus one past the end. */
+function offsetsOf(parts) {
+    const starts = new Int32Array(parts.length + 1);
+    for (let i = 0; i < parts.length; i++) {
+        starts[i + 1] = starts[i] + parts[i].length + SEPARATOR.length;
+    }
+    return starts;
+}
+/** The labels of the codes Eurostat's metabase lists per dataset, searchable at word starts. */
+export class DimensionValueIndex {
+    /** Epoch ms when this index finished loading: its age, and the snapshot a cursor binds. */
+    loadedAt;
+    parts;
+    constructor(parts, loadedAt) {
+        this.parts = parts;
+        this.loadedAt = loadedAt;
+    }
+    /** Dataset, value, and label counts, for logging. */
+    get size() {
+        return {
+            datasets: this.parts.datasetIds.size,
+            labels: this.parts.labelStart.length - 1,
+            values: this.parts.valueLabel.length,
+        };
+    }
+    /** The id of a dataset the metabase lists, or undefined for one it does not. */
+    datasetOf(code) {
+        return this.parts.datasetIds.get(code);
+    }
+    /**
+     * Flags, per label id, whether `word` occurs in the folded label with no letter or digit
+     * right before it; undefined when no label carries it. One scan of the joined labels;
+     * past a label's first match the scan resumes at the next label.
+     */
+    labelsAtWordStart(word) {
+        const { searchStart, searchText } = this.parts;
+        let flags;
+        let at = searchText.indexOf(word);
+        while (at !== -1) {
+            if (continuesWord(searchText, at)) {
+                at = searchText.indexOf(word, at + 1);
+                continue;
+            }
+            const label = this.labelAt(at);
+            flags ??= new Uint8Array(searchStart.length - 1);
+            flags[label] = 1;
+            at = searchText.indexOf(word, searchStart[label + 1]);
+        }
+        return flags;
+    }
+    /** Whether some value of `dataset` has a label `flags` marks. */
+    hasValue(dataset, flags) {
+        return this.someValue(dataset, (label) => flags[label] === 1);
+    }
+    /** Whether one value of `dataset` has a label every one of `flagSets` marks. */
+    hasValueMarkedByAll(dataset, flagSets) {
+        return this.someValue(dataset, (label) => flagSets.every((flags) => flags[label] === 1));
+    }
+    /**
+     * The values of `dataset` whose label any of `flagSets` marks: those marked by the most
+     * flag sets first, so a value carrying every word leads the ones carrying one of them,
+     * then metabase order.
+     */
+    valuesMarked(dataset, flagSets) {
+        const { codeStart, codeText, datasetStart, datasetValues, dimensions, labelStart, labelText } = this.parts;
+        const values = [];
+        for (let at = datasetStart[dataset]; at < datasetStart[dataset + 1]; at++) {
+            const value = datasetValues[at];
+            const label = this.parts.valueLabel[value];
+            const marked = label < 0 ? 0 : flagSets.filter((flags) => flags[label] === 1).length;
+            if (marked === 0)
+                continue;
+            values.push({
+                marked,
+                value: {
+                    dimension: dimensions[this.parts.valueDimension[value]],
+                    code: codeText.slice(codeStart[value], codeStart[value + 1] - 1),
+                    label: labelText.slice(labelStart[label], labelStart[label + 1] - 1),
+                },
+            });
+        }
+        return values.sort((a, b) => b.marked - a.marked).map(({ value }) => value);
+    }
+    someValue(dataset, test) {
+        const { datasetStart, datasetValues, valueLabel } = this.parts;
+        for (let at = datasetStart[dataset]; at < datasetStart[dataset + 1]; at++) {
+            const label = valueLabel[datasetValues[at]];
+            if (label >= 0 && test(label))
+                return true;
+        }
+        return false;
+    }
+    /** The label whose span of the joined search text holds offset `at`. */
+    labelAt(at) {
+        const { searchStart } = this.parts;
+        let low = 0;
+        let high = searchStart.length - 2;
+        while (low < high) {
+            const mid = (low + high + 1) >> 1;
+            if (searchStart[mid] <= at)
+                low = mid;
+            else
+                high = mid - 1;
+        }
+        return low;
+    }
+}
+/**
+ * Text of a stream in pieces that each end at a boundary, so no token spans two pieces. Only
+ * the newest chunk is searched for a boundary, so a long run without one costs its length.
+ */
+class BoundaryPieces {
+    pending = [];
+    boundary;
+    constructor(boundary) {
+        this.boundary = boundary;
+    }
+    /** The text up to the last boundary in `chunk`, or undefined while `chunk` holds none. */
+    cut(chunk) {
+        const at = chunk.lastIndexOf(this.boundary);
+        if (at === -1) {
+            this.pending.push(chunk);
+            return;
+        }
+        const end = at + this.boundary.length;
+        this.pending.push(chunk.slice(0, end));
+        const piece = this.pending.join('');
+        this.pending = [chunk.slice(end)];
+        return piece;
+    }
+    /** The text after the last boundary. */
+    rest() {
+        const rest = this.pending.join('');
+        this.pending = [];
+        return rest;
+    }
+}
+/**
+ * Reads `dataset<TAB>dimension<TAB>code` lines as they stream in, `time` rows skipped. Every
+ * string it keeps is a copy, so no decoded chunk outlives its turn.
+ */
+class MetabaseReader {
+    line = 0;
+    lines = new BoundaryPieces('\n');
+    table = {
+        datasetIds: new Map(),
+        dimensionIds: new Map(),
+        dimensions: [],
+        rowDatasets: [],
+        rowValues: [],
+        valueCodes: [],
+        valueDimensions: [],
+        valueIds: new Map(),
+    };
+    read(chunk) {
+        const text = this.lines.cut(chunk);
+        if (text !== undefined)
+            for (const line of text.slice(0, -1).split('\n'))
+                this.add(line);
+    }
+    /** The table read. Throws when the metabase listed no dataset: a broken download, not an empty catalogue. */
+    finish() {
+        const rest = this.lines.rest();
+        if (rest)
+            this.add(rest);
+        if (this.table.datasetIds.size === 0)
+            throw new Error('The metabase listed no dataset dimension values');
+        return this.table;
+    }
+    add(text) {
+        this.line += 1;
+        const line = text.endsWith('\r') ? text.slice(0, -1) : text;
+        if (!line)
+            return;
+        const fields = line.split('\t');
+        const [dataset, dimension, code] = fields;
+        if (fields.length !== 3 || !dataset || !dimension || code === undefined) {
+            throw new Error(`Metabase line ${this.line} is not a dataset, dimension and code separated by tabs`);
+        }
+        if (dimension === TIME_DIMENSION)
+            return;
+        const table = this.table;
+        let datasetId = table.datasetIds.get(dataset);
+        if (datasetId === undefined) {
+            datasetId = table.datasetIds.size;
+            table.datasetIds.set(copyOf(dataset), datasetId);
+        }
+        let dimensionId = table.dimensionIds.get(dimension);
+        if (dimensionId === undefined) {
+            dimensionId = table.dimensions.length;
+            table.dimensions.push(copyOf(dimension));
+            table.dimensionIds.set(table.dimensions[dimensionId], dimensionId);
+        }
+        const key = `${dimensionId}\t${code}`;
+        let valueId = table.valueIds.get(key);
+        if (valueId === undefined) {
+            valueId = table.valueCodes.length;
+            table.valueIds.set(copyOf(key), valueId);
+            table.valueCodes.push(copyOf(code));
+            table.valueDimensions.push(dimensionId);
+        }
+        table.rowDatasets.push(datasetId);
+        table.rowValues.push(valueId);
+    }
+}
+/**
+ * Reads the codelist dump as it streams in, keeping the first English name of each code the
+ * metabase lists, from the codelist named after its dimension in upper case.
+ */
+class CodelistReader {
+    code;
+    codelistDimensions;
+    dimensionId;
+    labelled = 0;
+    labels;
+    /** Whether the open code, or the codelist before its first code, has had its name. */
+    named = true;
+    pieces = new BoundaryPieces(CODE_CLOSE);
+    table;
+    constructor(table) {
+        this.table = table;
+        this.labels = new Array(table.valueCodes.length);
+        this.codelistDimensions = new Map(table.dimensions.map((dimension, id) => [dimension.toUpperCase(), id]));
+    }
+    read(chunk) {
+        const piece = this.pieces.cut(chunk);
+        if (piece !== undefined)
+            this.scan(piece);
+    }
+    /**
+     * The index, from the metabase table and the labels read. Throws when the dump stops before
+     * its root element closes, or labels none of the metabase's codes.
+     */
+    finish() {
+        const rest = this.pieces.rest();
+        this.scan(rest);
+        if (!rest.includes(DOCUMENT_CLOSE)) {
+            throw new Error('The codelist dump ended before its root element closed');
+        }
+        if (this.labelled === 0)
+            throw new Error('The codelists labelled none of the metabase codes');
+        return assembleIndex(this.table, this.labels);
+    }
+    scan(piece) {
+        CODELIST_TOKEN.lastIndex = 0;
+        for (let token = CODELIST_TOKEN.exec(piece); token; token = CODELIST_TOKEN.exec(piece)) {
+            const [, codelistId, codeId, name] = token;
+            if (codelistId !== undefined) {
+                this.dimensionId = this.codelistDimensions.get(decodeXmlEntities(codelistId));
+                this.named = true;
+            }
+            else if (codeId !== undefined) {
+                this.code = decodeXmlEntities(codeId);
+                this.named = false;
+            }
+            else if (name !== undefined) {
+                // The codelist's own name precedes its first code; only a code's first name labels it.
+                if (this.named || this.dimensionId === undefined)
+                    continue;
+                this.named = true;
+                const valueId = this.table.valueIds.get(`${this.dimensionId}\t${this.code}`);
+                if (valueId === undefined || this.labels[valueId] !== undefined)
+                    continue;
+                this.labels[valueId] = copyOf(decodeXmlEntities(name).trim());
+                this.labelled += 1;
+            }
+            else {
+                this.named = true;
+            }
+        }
+    }
+}
+/** The index from the metabase table and each value's label. */
+function assembleIndex(table, valueLabels) {
+    const { datasetIds, dimensions, rowDatasets, rowValues, valueCodes, valueDimensions } = table;
+    const labelIds = new Map();
+    const labels = [];
+    const valueLabel = new Int32Array(valueCodes.length).fill(-1);
+    valueLabels.forEach((label, valueId) => {
+        if (label === undefined)
+            return;
+        let labelId = labelIds.get(label);
+        if (labelId === undefined) {
+            labelId = labels.length;
+            labels.push(label);
+            labelIds.set(label, labelId);
+        }
+        valueLabel[valueId] = labelId;
+    });
+    const folded = labels.map(foldSearchText);
+    const datasetStart = new Int32Array(datasetIds.size + 1);
+    for (const datasetId of rowDatasets)
+        datasetStart[datasetId + 1] = datasetStart[datasetId + 1] + 1;
+    for (let d = 0; d < datasetIds.size; d++) {
+        datasetStart[d + 1] = datasetStart[d + 1] + datasetStart[d];
+    }
+    const fill = datasetStart.slice(0, -1);
+    const datasetValues = new Int32Array(rowValues.length);
+    rowDatasets.forEach((datasetId, row) => {
+        datasetValues[fill[datasetId]++] = rowValues[row];
+    });
+    return new DimensionValueIndex({
+        codeStart: offsetsOf(valueCodes),
+        codeText: valueCodes.join(SEPARATOR),
+        datasetIds,
+        datasetStart,
+        datasetValues,
+        dimensions,
+        labelStart: offsetsOf(labels),
+        labelText: labels.join(SEPARATOR),
+        searchStart: offsetsOf(folded),
+        searchText: folded.join(SEPARATOR),
+        valueDimension: Int32Array.from(valueDimensions),
+        valueLabel,
+    }, Date.now());
+}
+/**
+ * Build the index from the metabase (`dataset<TAB>dimension<TAB>code` lines) and the codelist
+ * dump (SDMX 2.1 structure XML), each read as a stream of decoded text. The metabase is read
+ * first, so only the labels of codes it lists are kept from the codelist named, in upper case,
+ * after each dimension. A code's first English name is its label.
+ *
+ * Throws on a metabase line that is not three tab-separated fields, a metabase that lists no
+ * dataset, a codelist dump that stops before its root element closes, or codelists that label
+ * none of the metabase's codes: each is a broken download, not an empty catalogue.
+ */
+export async function buildDimensionValueIndex(metabase, codelists) {
+    const metabaseReader = new MetabaseReader();
+    for await (const chunk of metabase)
+        metabaseReader.read(chunk);
+    const codelistReader = new CodelistReader(metabaseReader.finish());
+    for await (const chunk of codelists)
+        codelistReader.read(chunk);
+    return codelistReader.finish();
+}
+/**
+ * The decoded text of a gzip file at `url`, chunk by chunk. The request starts when the
+ * first chunk is asked for; `timeoutMs` bounds the whole transfer.
+ */
+async function* gzipText(url, timeoutMs, context) {
+    const response = await fetchWithTimeout(url, timeoutMs, context);
+    if (!response.body)
+        throw new Error(`${new URL(url).pathname} answered with no body`);
+    /**
+     * `DecompressionStream`'s writable side is typed `BufferSource`, wider than the
+     * `Uint8Array` chunks the body carries, so the transform is narrowed to the types in play.
+     */
+    const reader = response.body
+        .pipeThrough(new DecompressionStream('gzip'))
+        .getReader();
+    const decoder = new TextDecoder('utf-8');
+    try {
+        while (true) {
+            const { done, value } = await reader.read();
+            if (done)
+                break;
+            const text = decoder.decode(value, { stream: true });
+            if (text)
+                yield text;
+        }
+        const tail = decoder.decode();
+        if (tail)
+            yield tail;
+    }
+    finally {
+        // Ends the transfer when parsing stops early; a no-op once the stream is done.
+        await reader.cancel().catch(() => undefined);
+        reader.releaseLock();
+    }
+}
+/**
+ * Download and build the index: the metabase first, then the codelist dump, each within
+ * `EUROSTAT_BULK_TIMEOUT_MS`. One attempt; the caller schedules retries.
+ */
+export function fetchDimensionValueIndex(context) {
+    const { baseUrl, bulkTimeoutMs } = getServerConfig();
+    return buildDimensionValueIndex(gzipText(`${baseUrl}/catalogue/metabase.txt.gz`, bulkTimeoutMs, context), gzipText(`${baseUrl}/sdmx/2.1/codelist/ESTAT/all/latest?compressed=true`, bulkTimeoutMs, context));
+}
+//# sourceMappingURL=dimension-value-index.js.map

package/dist/services/eurostat-catalogue/dimension-value-index.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"dimension-value-index.js","sourceRoot":"","sources":["../../../src/services/eurostat-catalogue/dimension-value-index.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AAEH,OAAO,EAAE,gBAAgB,EAAuB,MAAM,8BAA8B,CAAC;AACrF,OAAO,EAAE,eAAe,EAAE,MAAM,2BAA2B,CAAC;AAC5D,OAAO,EAAE,iBAAiB,EAAE,MAAM,oBAAoB,CAAC;AACvD,OAAO,EAAE,aAAa,EAAE,cAAc,EAAE,MAAM,kBAAkB,CAAC;AAGjE,2FAA2F;AAC3F,MAAM,cAAc,GAAG,MAAM,CAAC;AAE9B,+FAA+F;AAC/F,MAAM,SAAS,GAAG,IAAI,CAAC;AAEvB;;;GAGG;AACH,MAAM,cAAc,GAClB,qHAAqH,CAAC;AAExH,MAAM,UAAU,GAAG,WAAW,CAAC;AAE/B,+FAA+F;AAC/F,MAAM,cAAc,GAAG,gBAAgB,CAAC;AAExC;;;;GAIG;AACH,SAAS,MAAM,CAAC,IAAY;IAC1B,OAAO,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC;AAChE,CAAC;AAuBD,+EAA+E;AAC/E,SAAS,SAAS,CAAC,KAAwB;IACzC,MAAM,MAAM,GAAG,IAAI,UAAU,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;IAChD,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QACtC,MAAM,CAAC,CAAC,GAAG,CAAC,CAAC,GAAI,MAAM,CAAC,CAAC,CAAY,GAAI,KAAK,CAAC,CAAC,CAAY,CAAC,MAAM,GAAG,SAAS,CAAC,MAAM,CAAC;IACzF,CAAC;IACD,OAAO,MAAM,CAAC;AAChB,CAAC;AAED,gGAAgG;AAChG,MAAM,OAAO,mBAAmB;IAC9B,2FAA2F;IAClF,QAAQ,CAAS;IACT,KAAK,CAAa;IAEnC,YAAY,KAAiB,EAAE,QAAgB;QAC7C,IAAI,CAAC,KAAK,GAAG,KAAK,CAAC;QACnB,IAAI,CAAC,QAAQ,GAAG,QAAQ,CAAC;IAC3B,CAAC;IAED,qDAAqD;IACrD,IAAI,IAAI;QACN,OAAO;YACL,QAAQ,EAAE,IAAI,CAAC,KAAK,CAAC,UAAU,CAAC,IAAI;YACpC,MAAM,EAAE,IAAI,CAAC,KAAK,CAAC,UAAU,CAAC,MAAM,GAAG,CAAC;YACxC,MAAM,EAAE,IAAI,CAAC,KAAK,CAAC,UAAU,CAAC,MAAM;SACrC,CAAC;IACJ,CAAC;IAED,gFAAgF;IAChF,SAAS,CAAC,IAAY;QACpB,OAAO,IAAI,CAAC,KAAK,CAAC,UAAU,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;IACzC,CAAC;IAED;;;;OAIG;IACH,iBAAiB,CAAC,IAAY;QAC5B,MAAM,EAAE,WAAW,EAAE,UAAU,EAAE,GAAG,IAAI,CAAC,KAAK,CAAC;QAC/C,IAAI,KAA6B,CAAC;QAClC,IAAI,EAAE,GAAG,UAAU,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC;QAClC,OAAO,EAAE,KAAK,CAAC,CAAC,EAAE,CAAC;YACjB,IAAI,aAAa,CAAC,UAAU,EAAE,EAAE,CAAC,EAAE,CAAC;gBAClC,EAAE,GAAG,UAAU,CAAC,OAAO,CAAC,IAAI,EAAE,EAAE,GAAG,CAAC,CAAC,CAAC;gBACtC,SAAS;YACX,CAAC;YACD,MAAM,KAAK,GAAG,IAAI,CAAC,OAAO,CAAC,EAAE,CAAC,CAAC;YAC/B,KAAK,KAAK,IAAI,UAAU,CAAC,WAAW,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;YACjD,KAAK,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;YACjB,EAAE,GAAG,UAAU,CAAC,OAAO,CAAC,IAAI,EAAE,WAAW,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC,CAAC;QACxD,CAAC;QACD,OAAO,KAAK,CAAC;IACf,CAAC;IAED,iEAAiE;IACjE,QAAQ,CAAC,OAAe,EAAE,KAAiB;QACzC,OAAO,IAAI,CAAC,SAAS,CAAC,OAAO,EAAE,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC;IAChE,CAAC;IAED,gFAAgF;IAChF,mBAAmB,CAAC,OAAe,EAAE,QAA+B;QAClE,OAAO,IAAI,CAAC,SAAS,CAAC,OAAO,EAAE,CAAC,KAAK,EAAE,EAAE,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;IAC3F,CAAC;IAED;;;;OAIG;IACH,YAAY,CAAC,OAAe,EAAE,QAA+B;QAC3D,MAAM,EAAE,SAAS,EAAE,QAAQ,EAAE,YAAY,EAAE,aAAa,EAAE,UAAU,EAAE,UAAU,EAAE,SAAS,EAAE,GAC3F,IAAI,CAAC,KAAK,CAAC;QACb,MAAM,MAAM,GAAmD,EAAE,CAAC;QAClE,KACE,IAAI,EAAE,GAAG,YAAY,CAAC,OAAO,CAAW,EACxC,EAAE,GAAI,YAAY,CAAC,OAAO,GAAG,CAAC,CAAY,EAC1C,EAAE,EAAE,EACJ,CAAC;YACD,MAAM,KAAK,GAAG,aAAa,CAAC,EAAE,CAAW,CAAC;YAC1C,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,CAAC,UAAU,CAAC,KAAK,CAAW,CAAC;YACrD,MAAM,MAAM,GAAG,KAAK,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,MAAM,CAAC;YACrF,IAAI,MAAM,KAAK,CAAC;gBAAE,SAAS;YAC3B,MAAM,CAAC,IAAI,CAAC;gBACV,MAAM;gBACN,KAAK,EAAE;oBACL,SAAS,EAAE,UAAU,CAAC,IAAI,CAAC,KAAK,CAAC,cAAc,CAAC,KAAK,CAAW,CAAW;oBAC3E,IAAI,EAAE,QAAQ,CAAC,KAAK,CAAC,SAAS,CAAC,KAAK,CAAC,EAAG,SAAS,CAAC,KAAK,GAAG,CAAC,CAAY,GAAG,CAAC,CAAC;oBAC5E,KAAK,EAAE,SAAS,CAAC,KAAK,CAAC,UAAU,CAAC,KAAK,CAAC,EAAG,UAAU,CAAC,KAAK,GAAG,CAAC,CAAY,GAAG,CAAC,CAAC;iBACjF;aACF,CAAC,CAAC;QACL,CAAC;QACD,OAAO,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC,MAAM,CAAC,CAAC,GAAG,CAAC,CAAC,EAAE,KAAK,EAAE,EAAE,EAAE,CAAC,KAAK,CAAC,CAAC;IAC9E,CAAC;IAEO,SAAS,CAAC,OAAe,EAAE,IAAgC;QACjE,MAAM,EAAE,YAAY,EAAE,aAAa,EAAE,UAAU,EAAE,GAAG,IAAI,CAAC,KAAK,CAAC;QAC/D,KACE,IAAI,EAAE,GAAG,YAAY,CAAC,OAAO,CAAW,EACxC,EAAE,GAAI,YAAY,CAAC,OAAO,GAAG,CAAC,CAAY,EAC1C,EAAE,EAAE,EACJ,CAAC;YACD,MAAM,KAAK,GAAG,UAAU,CAAC,aAAa,CAAC,EAAE,CAAW,CAAW,CAAC;YAChE,IAAI,KAAK,IAAI,CAAC,IAAI,IAAI,CAAC,KAAK,CAAC;gBAAE,OAAO,IAAI,CAAC;QAC7C,CAAC;QACD,OAAO,KAAK,CAAC;IACf,CAAC;IAED,wEAAwE;IAChE,OAAO,CAAC,EAAU;QACxB,MAAM,EAAE,WAAW,EAAE,GAAG,IAAI,CAAC,KAAK,CAAC;QACnC,IAAI,GAAG,GAAG,CAAC,CAAC;QACZ,IAAI,IAAI,GAAG,WAAW,CAAC,MAAM,GAAG,CAAC,CAAC;QAClC,OAAO,GAAG,GAAG,IAAI,EAAE,CAAC;YAClB,MAAM,GAAG,GAAG,CAAC,GAAG,GAAG,IAAI,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC;YAClC,IAAK,WAAW,CAAC,GAAG,CAAY,IAAI,EAAE;gBAAE,GAAG,GAAG,GAAG,CAAC;;gBAC7C,IAAI,GAAG,GAAG,GAAG,CAAC,CAAC;QACtB,CAAC;QACD,OAAO,GAAG,CAAC;IACb,CAAC;CACF;AAED;;;GAGG;AACH,MAAM,cAAc;IACV,OAAO,GAAa,EAAE,CAAC;IACd,QAAQ,CAAS;IAElC,YAAY,QAAgB;QAC1B,IAAI,CAAC,QAAQ,GAAG,QAAQ,CAAC;IAC3B,CAAC;IAED,0FAA0F;IAC1F,GAAG,CAAC,KAAa;QACf,MAAM,EAAE,GAAG,KAAK,CAAC,WAAW,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;QAC5C,IAAI,EAAE,KAAK,CAAC,CAAC,EAAE,CAAC;YACd,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;YACzB,OAAO;QACT,CAAC;QACD,MAAM,GAAG,GAAG,EAAE,GAAG,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC;QACtC,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC;QACvC,MAAM,KAAK,GAAG,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;QACpC,IAAI,CAAC,OAAO,GAAG,CAAC,KAAK,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC;QAClC,OAAO,KAAK,CAAC;IACf,CAAC;IAED,wCAAwC;IACxC,IAAI;QACF,MAAM,IAAI,GAAG,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;QACnC,IAAI,CAAC,OAAO,GAAG,EAAE,CAAC;QAClB,OAAO,IAAI,CAAC;IACd,CAAC;CACF;AAeD;;;GAGG;AACH,MAAM,cAAc;IACV,IAAI,GAAG,CAAC,CAAC;IACA,KAAK,GAAG,IAAI,cAAc,CAAC,IAAI,CAAC,CAAC;IACjC,KAAK,GAAkB;QACtC,UAAU,EAAE,IAAI,GAAG,EAAE;QACrB,YAAY,EAAE,IAAI,GAAG,EAAE;QACvB,UAAU,EAAE,EAAE;QACd,WAAW,EAAE,EAAE;QACf,SAAS,EAAE,EAAE;QACb,UAAU,EAAE,EAAE;QACd,eAAe,EAAE,EAAE;QACnB,QAAQ,EAAE,IAAI,GAAG,EAAE;KACpB,CAAC;IAEF,IAAI,CAAC,KAAa;QAChB,MAAM,IAAI,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC;QACnC,IAAI,IAAI,KAAK,SAAS;YAAE,KAAK,MAAM,IAAI,IAAI,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,IAAI,CAAC;gBAAE,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;IAC3F,CAAC;IAED,6GAA6G;IAC7G,MAAM;QACJ,MAAM,IAAI,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC;QAC/B,IAAI,IAAI;YAAE,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;QACzB,IAAI,IAAI,CAAC,KAAK,CAAC,UAAU,CAAC,IAAI,KAAK,CAAC;YAClC,MAAM,IAAI,KAAK,CAAC,iDAAiD,CAAC,CAAC;QACrE,OAAO,IAAI,CAAC,KAAK,CAAC;IACpB,CAAC;IAEO,GAAG,CAAC,IAAY;QACtB,IAAI,CAAC,IAAI,IAAI,CAAC,CAAC;QACf,MAAM,IAAI,GAAG,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC;QAC5D,IAAI,CAAC,IAAI;YAAE,OAAO;QAClB,MAAM,MAAM,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;QAChC,MAAM,CAAC,OAAO,EAAE,SAAS,EAAE,IAAI,CAAC,GAAG,MAAM,CAAC;QAC1C,IAAI,MAAM,CAAC,MAAM,KAAK,CAAC,IAAI,CAAC,OAAO,IAAI,CAAC,SAAS,IAAI,IAAI,KAAK,SAAS,EAAE,CAAC;YACxE,MAAM,IAAI,KAAK,CACb,iBAAiB,IAAI,CAAC,IAAI,yDAAyD,CACpF,CAAC;QACJ,CAAC;QACD,IAAI,SAAS,KAAK,cAAc;YAAE,OAAO;QACzC,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,CAAC;QACzB,IAAI,SAAS,GAAG,KAAK,CAAC,UAAU,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC;QAC9C,IAAI,SAAS,KAAK,SAAS,EAAE,CAAC;YAC5B,SAAS,GAAG,KAAK,CAAC,UAAU,CAAC,IAAI,CAAC;YAClC,KAAK,CAAC,UAAU,CAAC,GAAG,CAAC,MAAM,CAAC,OAAO,CAAC,EAAE,SAAS,CAAC,CAAC;QACnD,CAAC;QACD,IAAI,WAAW,GAAG,KAAK,CAAC,YAAY,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC;QACpD,IAAI,WAAW,KAAK,SAAS,EAAE,CAAC;YAC9B,WAAW,GAAG,KAAK,CAAC,UAAU,CAAC,MAAM,CAAC;YACtC,KAAK,CAAC,UAAU,CAAC,IAAI,CAAC,MAAM,CAAC,SAAS,CAAC,CAAC,CAAC;YACzC,KAAK,CAAC,YAAY,CAAC,GAAG,CAAC,KAAK,CAAC,UAAU,CAAC,WAAW,CAAW,EAAE,WAAW,CAAC,CAAC;QAC/E,CAAC;QACD,MAAM,GAAG,GAAG,GAAG,WAAW,KAAK,IAAI,EAAE,CAAC;QACtC,IAAI,OAAO,GAAG,KAAK,CAAC,QAAQ,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;QACtC,IAAI,OAAO,KAAK,SAAS,EAAE,CAAC;YAC1B,OAAO,GAAG,KAAK,CAAC,UAAU,CAAC,MAAM,CAAC;YAClC,KAAK,CAAC,QAAQ,CAAC,GAAG,CAAC,MAAM,CAAC,GAAG,CAAC,EAAE,OAAO,CAAC,CAAC;YACzC,KAAK,CAAC,UAAU,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC;YACpC,KAAK,CAAC,eAAe,CAAC,IAAI,CAAC,WAAW,CAAC,CAAC;QAC1C,CAAC;QACD,KAAK,CAAC,WAAW,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC;QAClC,KAAK,CAAC,SAAS,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;IAChC,CAAC;CACF;AAED;;;GAGG;AACH,MAAM,cAAc;IACV,IAAI,CAAqB;IAChB,kBAAkB,CAAsB;IACjD,WAAW,CAAqB;IAChC,QAAQ,GAAG,CAAC,CAAC;IACJ,MAAM,CAAyB;IAChD,sFAAsF;IAC9E,KAAK,GAAG,IAAI,CAAC;IACJ,MAAM,GAAG,IAAI,cAAc,CAAC,UAAU,CAAC,CAAC;IACxC,KAAK,CAAgB;IAEtC,YAAY,KAAoB;QAC9B,IAAI,CAAC,KAAK,GAAG,KAAK,CAAC;QACnB,IAAI,CAAC,MAAM,GAAG,IAAI,KAAK,CAAC,KAAK,CAAC,UAAU,CAAC,MAAM,CAAC,CAAC;QACjD,IAAI,CAAC,kBAAkB,GAAG,IAAI,GAAG,CAC/B,KAAK,CAAC,UAAU,CAAC,GAAG,CAAC,CAAC,SAAS,EAAE,EAAE,EAAE,EAAE,CAAC,CAAC,SAAS,CAAC,WAAW,EAAE,EAAE,EAAE,CAAC,CAAC,CACvE,CAAC;IACJ,CAAC;IAED,IAAI,CAAC,KAAa;QAChB,MAAM,KAAK,GAAG,IAAI,CAAC,MAAM,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC;QACrC,IAAI,KAAK,KAAK,SAAS;YAAE,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;IAC5C,CAAC;IAED;;;OAGG;IACH,MAAM;QACJ,MAAM,IAAI,GAAG,IAAI,CAAC,MAAM,CAAC,IAAI,EAAE,CAAC;QAChC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QAChB,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,cAAc,CAAC,EAAE,CAAC;YACnC,MAAM,IAAI,KAAK,CAAC,wDAAwD,CAAC,CAAC;QAC5E,CAAC;QACD,IAAI,IAAI,CAAC,QAAQ,KAAK,CAAC;YAAE,MAAM,IAAI,KAAK,CAAC,mDAAmD,CAAC,CAAC;QAC9F,OAAO,aAAa,CAAC,IAAI,CAAC,KAAK,EAAE,IAAI,CAAC,MAAM,CAAC,CAAC;IAChD,CAAC;IAEO,IAAI,CAAC,KAAa;QACxB,cAAc,CAAC,SAAS,GAAG,CAAC,CAAC;QAC7B,KAAK,IAAI,KAAK,GAAG,cAAc,CAAC,IAAI,CAAC,KAAK,CAAC,EAAE,KAAK,EAAE,KAAK,GAAG,cAAc,CAAC,IAAI,CAAC,KAAK,CAAC,EAAE,CAAC;YACvF,MAAM,CAAC,EAAE,UAAU,EAAE,MAAM,EAAE,IAAI,CAAC,GAAG,KAAK,CAAC;YAC3C,IAAI,UAAU,KAAK,SAAS,EAAE,CAAC;gBAC7B,IAAI,CAAC,WAAW,GAAG,IAAI,CAAC,kBAAkB,CAAC,GAAG,CAAC,iBAAiB,CAAC,UAAU,CAAC,CAAC,CAAC;gBAC9E,IAAI,CAAC,KAAK,GAAG,IAAI,CAAC;YACpB,CAAC;iBAAM,IAAI,MAAM,KAAK,SAAS,EAAE,CAAC;gBAChC,IAAI,CAAC,IAAI,GAAG,iBAAiB,CAAC,MAAM,CAAC,CAAC;gBACtC,IAAI,CAAC,KAAK,GAAG,KAAK,CAAC;YACrB,CAAC;iBAAM,IAAI,IAAI,KAAK,SAAS,EAAE,CAAC;gBAC9B,uFAAuF;gBACvF,IAAI,IAAI,CAAC,KAAK,IAAI,IAAI,CAAC,WAAW,KAAK,SAAS;oBAAE,SAAS;gBAC3D,IAAI,CAAC,KAAK,GAAG,IAAI,CAAC;gBAClB,MAAM,OAAO,GAAG,IAAI,CAAC,KAAK,CAAC,QAAQ,CAAC,GAAG,CAAC,GAAG,IAAI,CAAC,WAAW,KAAK,IAAI,CAAC,IAAI,EAAE,CAAC,CAAC;gBAC7E,IAAI,OAAO,KAAK,SAAS,IAAI,IAAI,CAAC,MAAM,CAAC,OAAO,CAAC,KAAK,SAAS;oBAAE,SAAS;gBAC1E,IAAI,CAAC,MAAM,CAAC,OAAO,CAAC,GAAG,MAAM,CAAC,iBAAiB,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC;gBAC9D,IAAI,CAAC,QAAQ,IAAI,CAAC,CAAC;YACrB,CAAC;iBAAM,CAAC;gBACN,IAAI,CAAC,KAAK,GAAG,IAAI,CAAC;YACpB,CAAC;QACH,CAAC;IACH,CAAC;CACF;AAED,gEAAgE;AAChE,SAAS,aAAa,CACpB,KAAoB,EACpB,WAA4C;IAE5C,MAAM,EAAE,UAAU,EAAE,UAAU,EAAE,WAAW,EAAE,SAAS,EAAE,UAAU,EAAE,eAAe,EAAE,GAAG,KAAK,CAAC;IAC9F,MAAM,QAAQ,GAAG,IAAI,GAAG,EAAkB,CAAC;IAC3C,MAAM,MAAM,GAAa,EAAE,CAAC;IAC5B,MAAM,UAAU,GAAG,IAAI,UAAU,CAAC,UAAU,CAAC,MAAM,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC;IAC9D,WAAW,CAAC,OAAO,CAAC,CAAC,KAAK,EAAE,OAAO,EAAE,EAAE;QACrC,IAAI,KAAK,KAAK,SAAS;YAAE,OAAO;QAChC,IAAI,OAAO,GAAG,QAAQ,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC;QAClC,IAAI,OAAO,KAAK,SAAS,EAAE,CAAC;YAC1B,OAAO,GAAG,MAAM,CAAC,MAAM,CAAC;YACxB,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;YACnB,QAAQ,CAAC,GAAG,CAAC,KAAK,EAAE,OAAO,CAAC,CAAC;QAC/B,CAAC;QACD,UAAU,CAAC,OAAO,CAAC,GAAG,OAAO,CAAC;IAChC,CAAC,CAAC,CAAC;IACH,MAAM,MAAM,GAAG,MAAM,CAAC,GAAG,CAAC,cAAc,CAAC,CAAC;IAE1C,MAAM,YAAY,GAAG,IAAI,UAAU,CAAC,UAAU,CAAC,IAAI,GAAG,CAAC,CAAC,CAAC;IACzD,KAAK,MAAM,SAAS,IAAI,WAAW;QACjC,YAAY,CAAC,SAAS,GAAG,CAAC,CAAC,GAAI,YAAY,CAAC,SAAS,GAAG,CAAC,CAAY,GAAG,CAAC,CAAC;IAC5E,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,UAAU,CAAC,IAAI,EAAE,CAAC,EAAE,EAAE,CAAC;QACzC,YAAY,CAAC,CAAC,GAAG,CAAC,CAAC,GAAI,YAAY,CAAC,CAAC,GAAG,CAAC,CAAY,GAAI,YAAY,CAAC,CAAC,CAAY,CAAC;IACtF,CAAC;IACD,MAAM,IAAI,GAAG,YAAY,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC;IACvC,MAAM,aAAa,GAAG,IAAI,UAAU,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC;IACvD,WAAW,CAAC,OAAO,CAAC,CAAC,SAAS,EAAE,GAAG,EAAE,EAAE;QACrC,aAAa,CAAE,IAAI,CAAC,SAAS,CAAY,EAAE,CAAC,GAAG,SAAS,CAAC,GAAG,CAAW,CAAC;IAC1E,CAAC,CAAC,CAAC;IAEH,OAAO,IAAI,mBAAmB,CAC5B;QACE,SAAS,EAAE,SAAS,CAAC,UAAU,CAAC;QAChC,QAAQ,EAAE,UAAU,CAAC,IAAI,CAAC,SAAS,CAAC;QACpC,UAAU;QACV,YAAY;QACZ,aAAa;QACb,UAAU;QACV,UAAU,EAAE,SAAS,CAAC,MAAM,CAAC;QAC7B,SAAS,EAAE,MAAM,CAAC,IAAI,CAAC,SAAS,CAAC;QACjC,WAAW,EAAE,SAAS,CAAC,MAAM,CAAC;QAC9B,UAAU,EAAE,MAAM,CAAC,IAAI,CAAC,SAAS,CAAC;QAClC,cAAc,EAAE,UAAU,CAAC,IAAI,CAAC,eAAe,CAAC;QAChD,UAAU;KACX,EACD,IAAI,CAAC,GAAG,EAAE,CACX,CAAC;AACJ,CAAC;AAED;;;;;;;;;GASG;AACH,MAAM,CAAC,KAAK,UAAU,wBAAwB,CAC5C,QAA+B,EAC/B,SAAgC;IAEhC,MAAM,cAAc,GAAG,IAAI,cAAc,EAAE,CAAC;IAC5C,IAAI,KAAK,EAAE,MAAM,KAAK,IAAI,QAAQ;QAAE,cAAc,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;IAC/D,MAAM,cAAc,GAAG,IAAI,cAAc,CAAC,cAAc,CAAC,MAAM,EAAE,CAAC,CAAC;IACnE,IAAI,KAAK,EAAE,MAAM,KAAK,IAAI,SAAS;QAAE,cAAc,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;IAChE,OAAO,cAAc,CAAC,MAAM,EAAE,CAAC;AACjC,CAAC;AAED;;;GAGG;AACH,KAAK,SAAS,CAAC,CAAC,QAAQ,CACtB,GAAW,EACX,SAAiB,EACjB,OAAuB;IAEvB,MAAM,QAAQ,GAAG,MAAM,gBAAgB,CAAC,GAAG,EAAE,SAAS,EAAE,OAAO,CAAC,CAAC;IACjE,IAAI,CAAC,QAAQ,CAAC,IAAI;QAAE,MAAM,IAAI,KAAK,CAAC,GAAG,IAAI,GAAG,CAAC,GAAG,CAAC,CAAC,QAAQ,wBAAwB,CAAC,CAAC;IACtF;;;OAGG;IACH,MAAM,MAAM,GAAG,QAAQ,CAAC,IAAI;SACzB,WAAW,CACV,IAAI,mBAAmB,CAAC,MAAM,CAG7B,CACF;SACA,SAAS,EAAE,CAAC;IACf,MAAM,OAAO,GAAG,IAAI,WAAW,CAAC,OAAO,CAAC,CAAC;IACzC,IAAI,CAAC;QACH,OAAO,IAAI,EAAE,CAAC;YACZ,MAAM,EAAE,IAAI,EAAE,KAAK,EAAE,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;YAC5C,IAAI,IAAI;gBAAE,MAAM;YAChB,MAAM,IAAI,GAAG,OAAO,CAAC,MAAM,CAAC,KAAK,EAAE,EAAE,MAAM,EAAE,IAAI,EAAE,CAAC,CAAC;YACrD,IAAI,IAAI;gBAAE,MAAM,IAAI,CAAC;QACvB,CAAC;QACD,MAAM,IAAI,GAAG,OAAO,CAAC,MAAM,EAAE,CAAC;QAC9B,IAAI,IAAI;YAAE,MAAM,IAAI,CAAC;IACvB,CAAC;YAAS,CAAC;QACT,+EAA+E;QAC/E,MAAM,MAAM,CAAC,MAAM,EAAE,CAAC,KAAK,CAAC,GAAG,EAAE,CAAC,SAAS,CAAC,CAAC;QAC7C,MAAM,CAAC,WAAW,EAAE,CAAC;IACvB,CAAC;AACH,CAAC;AAED;;;GAGG;AACH,MAAM,UAAU,wBAAwB,CAAC,OAAuB;IAC9D,MAAM,EAAE,OAAO,EAAE,aAAa,EAAE,GAAG,eAAe,EAAE,CAAC;IACrD,OAAO,wBAAwB,CAC7B,QAAQ,CAAC,GAAG,OAAO,4BAA4B,EAAE,aAAa,EAAE,OAAO,CAAC,EACxE,QAAQ,CACN,GAAG,OAAO,qDAAqD,EAC/D,aAAa,EACb,OAAO,CACR,CACF,CAAC;AACJ,CAAC"}

package/dist/services/eurostat-catalogue/eurostat-catalogue-service.d.ts

@@ -7,6 +7,27 @@ import type { Context } from '@cyanheads/mcp-ts-core';
 import type { AppConfig } from '@cyanheads/mcp-ts-core/config';
 import type { StorageService } from '@cyanheads/mcp-ts-core/storage';
 import type { BrowseItem, DatasetResult } from './types.js';
+/** Most distinct words a search that matches no catalogue text looks for in dimension values. */
+export declare const MAX_VALUE_SEARCH_WORDS = 10;
+/** Most distinct words, in query order, a search that matches nothing names as held by no dataset. */
+export declare const MAX_UNMATCHED_WORD_CHECKS = 32;
+/**
+ * How a dataset matched a search, best first: 0, its label holds the words as a phrase
+ * starting at a word start; 1, its label or code holds every word; 2, some word matched
+ * only in its theme breadcrumb.
+ */
+export type SearchTier = 0 | 1 | 2;
+/**
+ * The tier test for one search over `words`, whitespace-free tokens as `search()` splits a
+ * query. The returned function ranks a dataset from its label, folded by
+ * {@link foldSearchText}, and its code, never its breadcrumb, so every TOC placement of a
+ * code ranks alike. The phrase is the words joined by single spaces and folded the same way;
+ * it must start where no letter or digit precedes it, so "employment rate" is not found
+ * inside "Unemployment rate". The word test is the substring test matching applies.
+ */
+export declare function searchTierOf(words: readonly string[]): (foldedLabel: string, code: string) => SearchTier;
+/** Why a search that matched no catalogue text did not read dimension values. */
+export type ValuesNotSearched = 'index_unavailable' | 'too_many_words';
 export declare class EurostatCatalogueService {
     private cache;
     /** Shared in-flight refresh so concurrent callers past the TTL trigger one fetch. */
@@ -19,6 +40,11 @@ export declare class EurostatCatalogueService {
      * past its TTL that the refresh could not replace.
      */
     private refreshRetryAt;
+    /**
+     * The dimension value index a search that matches no catalogue text reads. Undefined until
+     * a TOC load starts the first index load.
+     */
+    private valueIndex;
     constructor(_config: AppConfig, _storage: StorageService);
     /**
      * Ensure a usable TOC is loaded. A cache younger than the configured TTL is
@@ -43,16 +69,28 @@ export declare class EurostatCatalogueService {
      * Only the TOC is required. A Comext list that fails to load leaves the catalogue
      * answering from the TOC alone: the failure is logged, and the merge is retried
      * after `REFRESH_RETRY_COOLDOWN_MS` without re-fetching the TOC.
+     *
+     * Every TOC load also starts the dimension value index load, which nothing awaits.
      */
     private fetchAndParseToc;
     /** Retry a failed Comext merge onto the TOC already held; concurrent callers share it. */
     private retryComextMerge;
     private logComextFailure;
+    /**
+     * Start a dimension value index load unless one is in flight, a failed load's backoff has
+     * not elapsed, or the index in service is younger than the TOC's TTL. Catalogue calls never
+     * wait for it: the load runs detached from the calling request and its trace, logs to the
+     * server log, and settles into the slot. A failure keeps the previous index in service and
+     * holds off the next attempt for `REFRESH_RETRY_COOLDOWN_MS`, doubling with each failure in
+     * a row up to the TOC's TTL.
+     */
+    private loadValueIndex;
     /**
      * Assemble a catalogue snapshot. `flows` undefined means the Comext list failed to
      * load, which arms the merge retry. `loadedAt` is the TOC's load time — the TTL and
-     * the cursor generation both key on it. A later merge only appends entries, so a
-     * cursor paged over the TOC alone still addresses the same leading matches.
+     * the cursor generation both key on it. A later merge only appends entries, and
+     * `search()` pages entries appended after a walk began behind the ones it began over, so
+     * a cursor paged over the TOC alone still pages every match once.
      */
     private buildCache;
     /** The Comext host's dataflow list — the `DS-*` collections the TOC does not carry. */
@@ -134,23 +172,46 @@ export declare class EurostatCatalogueService {
      * Decode a search cursor and confirm it belongs to this search. Malformed,
      * cross-query and superseded-snapshot cursors are rejected identically —
      * the caller's recovery is the same in all three cases, and the cursor is
-     * opaque, so telling them apart would only describe its internal shape.
+     * opaque, so telling them apart would only describe its internal shape. `entryCount`,
+     * the snapshot's length when the walk began, can be no more than its length now.
+     * `valueIndex`, present on a walk over rows found through dimension values, is checked
+     * once `search()` knows which rows it pages.
      */
     private decodeSearchCursor;
+    /** One search result row for the TOC placement `entries[i]`. */
+    private toDatasetResult;
     /**
-     * Each dataset's search text — label, breadcrumb, and code, lowercased, with whitespace
-     * beside punctuation removed — indexed like `entries`. Folders, which search never
+     * Each dataset's search text, indexed like `entries`. Folders, which search never
      * returns, are undefined.
      */
     private buildSearchText;
     /**
      * For a query that matched nothing, the words whose removal alone finds matches, with
-     * that count, in query order. `words` maps each distinct token to the word as typed.
-     * One pass: a placement missing exactly one word rescues its code for that word, and a
-     * code counts once per word whichever placement rescued it — so each count equals the
-     * `totalMatches` of the same search without that word.
+     * that count (`wordRemovalCounts`), and the words no searched text contains at all
+     * (`unmatchedWords`, with `allWordsUnmatched` when that is every word), each in query
+     * order and each omitted when empty. `words` maps each distinct token to the word as
+     * typed. One pass: a placement missing exactly one word rescues its code for that
+     * word, and a code counts once per word whichever placement rescued it — so each count
+     * equals the `totalMatches` of the same search without that word.
+     *
+     * `unmatchedWords` is drawn from the first {@link MAX_UNMATCHED_WORD_CHECKS} distinct
+     * words. `unmatchedWordsPartial` is set when a word past them is held by no text the pass
+     * read, so it may be one no dataset carries.
      */
     private countWithoutEachWord;
+    /**
+     * The datasets a query that matched no catalogue text finds when a word may also occur
+     * at a word start of the English label of a value the metabase lists for the dataset.
+     * Every word must still be found, in the dataset's own text or in a value's label. Each
+     * code keeps its placement whose text holds the most words, the first such in TOC order.
+     *
+     * Ordered by fewer words found only through values; then rows where one value's label
+     * carries all of those words; then {@link searchTierOf} over the tokens the dataset's text
+     * holds, in query order; then TOC order. `marks` holds, for each word found through
+     * values, which labels carry it. `words` are the query's distinct tokens, at most
+     * {@link MAX_VALUE_SEARCH_WORDS}.
+     */
+    private searchValues;
     /**
      * Search datasets by keyword. The query is tokenized on whitespace and every
      * token must match (AND), case-insensitively, somewhere in a combined
@@ -165,27 +226,52 @@ export declare class EurostatCatalogueService {
      * several TOC branches keeps its first matching placement as the canonical
      * `themePath`, so totals and page slots count unique query targets.
      *
+     * Matches are ordered by {@link searchTierOf} over the tokens — label phrase, then
+     * every word in label or code, then breadcrumb-only — and by that first matching
+     * placement within a tier. A tier reads the label and code alone, so it is the same
+     * whichever placement supplied the `themePath`.
+     *
      * Results are paginated with an opaque cursor (`limit` = page size, capped at
      * 100) bound to the normalized query and the catalogue snapshot that produced
      * it. An unusable cursor — malformed, from another query, or from a catalogue
      * since refreshed — is rejected as `invalid_cursor` instead of silently paging
-     * a different result set.
+     * a different result set. A Comext merge that lands mid-walk keeps the snapshot,
+     * so the cursor also carries the snapshot's length when the walk began: the
+     * datasets the merge appended page after every match the walk began over, each
+     * match still once.
+     *
+     * When no dataset's text holds every token, {@link searchValues} looks for the tokens
+     * the text lacks in the English labels of the dataset's dimension values. Its rows
+     * carry `matchedValues` and the result `matchedThroughValues`, and their cursor also
+     * binds the value index snapshot (`valueIndex`), so an index refresh mid-walk rejects it
+     * like a catalogue refresh. No search waits for the index: until one has loaded,
+     * `valuesNotSearched` is `index_unavailable`; past {@link MAX_VALUE_SEARCH_WORDS}
+     * distinct tokens it is `too_many_words`.
      *
      * An empty match reports the `DS-*` codes the query names that neither source lists
      * (`undisseminatedCodes`), so the caller can be told the collection is not
      * disseminated rather than to broaden the search, and — for two or more distinct
-     * words — each word whose removal alone finds matches (`wordRemovalCounts`).
-     * `comextListMissing` flags that the Comext list is not merged right now, so the
-     * `DS-*` collections were not searched and absence from that list proves nothing;
-     * `catalogueLoadedAt` marks a catalogue a failed refresh could not replace.
+     * words — each word whose removal alone finds matches (`wordRemovalCounts`) and each
+     * word no dataset's search text contains (`unmatchedWords`, as first typed), with
+     * `allWordsUnmatched` when that is every word. `unmatchedWords` reads the first
+     * {@link MAX_UNMATCHED_WORD_CHECKS} distinct words; `unmatchedWordsPartial` flags a query
+     * with a later word that may be one no dataset carries. `comextListMissing` flags that the
+     * Comext list is not merged right now, so the `DS-*` collections were not searched and
+     * absence from that list proves nothing; `catalogueLoadedAt` marks a catalogue a
+     * failed refresh could not replace.
      */
     search(query: string, limit: number, cursor: string | undefined, ctx: Context): Promise<{
+        allWordsUnmatched?: true;
         catalogueLoadedAt?: string;
         comextListMissing?: boolean;
         datasets: DatasetResult[];
+        matchedThroughValues?: true;
         nextCursor?: string;
         totalMatches: number;
         undisseminatedCodes?: string[];
+        unmatchedWords?: string[];
+        unmatchedWordsPartial?: true;
+        valuesNotSearched?: ValuesNotSearched;
         wordRemovalCounts?: Array<{
             totalMatches: number;
             word: string;

package/dist/services/eurostat-catalogue/eurostat-catalogue-service.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"eurostat-catalogue-service.d.ts","sourceRoot":"","sources":["../../../src/services/eurostat-catalogue/eurostat-catalogue-service.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,KAAK,EAAE,OAAO,EAAE,MAAM,wBAAwB,CAAC;AACtD,OAAO,KAAK,EAAE,SAAS,EAAE,MAAM,+BAA+B,CAAC;AAO/D,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,gCAAgC,CAAC;AAgBrE,OAAO,KAAK,EAAE,UAAU,EAAE,aAAa,EAAY,MAAM,YAAY,CAAC;AAoEtE,qBAAa,wBAAwB;IACnC,OAAO,CAAC,KAAK,CAAuB;IACpC,qFAAqF;IACrF,OAAO,CAAC,OAAO,CAAgC;IAC/C,uDAAuD;IACvD,OAAO,CAAC,OAAO,CAAgC;IAC/C;;;;OAIG;IACH,OAAO,CAAC,cAAc,CAAK;IAK3B,YAAY,OAAO,EAAE,SAAS,EAAE,QAAQ,EAAE,cAAc,EAAI;IAE5D;;;;;;;;;;;;;;OAcG;IACH,OAAO,CAAC,YAAY;YAIN,UAAU;IAyCxB;;;;;;OAMG;YACW,gBAAgB;IAY9B,0FAA0F;IAC1F,OAAO,CAAC,gBAAgB;IAoBxB,OAAO,CAAC,gBAAgB;IAUxB;;;;;OAKG;IACH,OAAO,CAAC,UAAU;IAclB,uFAAuF;YACzE,gBAAgB;IA4B9B;;;;;;;OAOG;IACH,OAAO,CAAC,gBAAgB;YA0CV,eAAe;IAmC7B;;;;;;;;;OASG;IACH,OAAO,CAAC,cAAc;IAQtB;;;;;OAKG;IACH,OAAO,CAAC,QAAQ;IAwEhB,OAAO,CAAC,OAAO;IAWf,OAAO,CAAC,YAAY;IASpB,OAAO,CAAC,SAAS;IAMjB;;OAEG;IACH,OAAO,CAAC,SAAS;IAYjB,wDAAwD;IACxD,OAAO,CAAC,YAAY;IAOpB;;;;;;OAMG;IACH,OAAO,CAAC,gBAAgB;IAUxB;;;OAGG;IACH,OAAO,CAAC,SAAS;IAIjB;;;;;;;;;;;;;OAaG;IACG,MAAM,CACV,SAAS,EAAE,MAAM,GAAG,SAAS,EAC7B,GAAG,EAAE,OAAO,GACX,OAAO,CAAC;QACT,iBAAiB,CAAC,EAAE,MAAM,CAAC;QAC3B,iBAAiB,CAAC,EAAE,OAAO,CAAC;QAC5B,KAAK,EAAE,UAAU,EAAE,CAAC;QACpB,eAAe,CAAC,EAAE,MAAM,EAAE,EAAE,CAAC;QAC7B,UAAU,EAAE,MAAM,EAAE,CAAC;QACrB,SAAS,CAAC,EAAE,MAAM,CAAC;KACpB,CAAC,CA4ED;IAED,OAAO,CAAC,YAAY;IAepB;;;;;OAKG;IACH,OAAO,CAAC,kBAAkB;IAkB1B;;;;OAIG;IACH,OAAO,CAAC,eAAe;IAUvB;;;;;;OAMG;IACH,OAAO,CAAC,oBAAoB;IAyB5B;;;;;;;;;;;;;;;;;;;;;;;;;;;OA2BG;IACG,MAAM,CACV,KAAK,EAAE,MAAM,EACb,KAAK,EAAE,MAAM,EACb,MAAM,EAAE,MAAM,GAAG,SAAS,EAC1B,GAAG,EAAE,OAAO,GACX,OAAO,CAAC;QACT,iBAAiB,CAAC,EAAE,MAAM,CAAC;QAC3B,iBAAiB,CAAC,EAAE,OAAO,CAAC;QAC5B,QAAQ,EAAE,aAAa,EAAE,CAAC;QAC1B,UAAU,CAAC,EAAE,MAAM,CAAC;QACpB,YAAY,EAAE,MAAM,CAAC;QACrB,mBAAmB,CAAC,EAAE,MAAM,EAAE,CAAC;QAC/B,iBAAiB,CAAC,EAAE,KAAK,CAAC;YAAE,YAAY,EAAE,MAAM,CAAC;YAAC,IAAI,EAAE,MAAM,CAAA;SAAE,CAAC,CAAC;KACnE,CAAC,CAuED;CACF;AAgBD,wBAAgB,4BAA4B,CAAC,MAAM,EAAE,SAAS,EAAE,OAAO,EAAE,cAAc,GAAG,IAAI,CAE7F;AAED,wBAAgB,2BAA2B,IAAI,wBAAwB,CAOtE"}
+{"version":3,"file":"eurostat-catalogue-service.d.ts","sourceRoot":"","sources":["../../../src/services/eurostat-catalogue/eurostat-catalogue-service.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,KAAK,EAAE,OAAO,EAAE,MAAM,wBAAwB,CAAC;AACtD,OAAO,KAAK,EAAE,SAAS,EAAE,MAAM,+BAA+B,CAAC;AAQ/D,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,gCAAgC,CAAC;AAqBrE,OAAO,KAAK,EAAE,UAAU,EAAE,aAAa,EAAY,MAAM,YAAY,CAAC;AAKtE,iGAAiG;AACjG,eAAO,MAAM,sBAAsB,KAAK,CAAC;AAEzC,sGAAsG;AACtG,eAAO,MAAM,yBAAyB,KAAK,CAAC;AAwC5C;;;;GAIG;AACH,MAAM,MAAM,UAAU,GAAG,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;AAEnC;;;;;;;GAOG;AACH,wBAAgB,YAAY,CAC1B,KAAK,EAAE,SAAS,MAAM,EAAE,GACvB,CAAC,WAAW,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,KAAK,UAAU,CAQnD;AAwDD,iFAAiF;AACjF,MAAM,MAAM,iBAAiB,GAAG,mBAAmB,GAAG,gBAAgB,CAAC;AASvE,qBAAa,wBAAwB;IACnC,OAAO,CAAC,KAAK,CAAuB;IACpC,qFAAqF;IACrF,OAAO,CAAC,OAAO,CAAgC;IAC/C,uDAAuD;IACvD,OAAO,CAAC,OAAO,CAAgC;IAC/C;;;;OAIG;IACH,OAAO,CAAC,cAAc,CAAK;IAC3B;;;OAGG;IACH,OAAO,CAAC,UAAU,CAA6B;IAK/C,YAAY,OAAO,EAAE,SAAS,EAAE,QAAQ,EAAE,cAAc,EAAI;IAE5D;;;;;;;;;;;;;;OAcG;IACH,OAAO,CAAC,YAAY;YAIN,UAAU;IA0CxB;;;;;;;;OAQG;YACW,gBAAgB;IAc9B,0FAA0F;IAC1F,OAAO,CAAC,gBAAgB;IAoBxB,OAAO,CAAC,gBAAgB;IAUxB;;;;;;;OAOG;IACH,OAAO,CAAC,cAAc;IAyCtB;;;;;;OAMG;IACH,OAAO,CAAC,UAAU;IAclB,uFAAuF;YACzE,gBAAgB;IA4B9B;;;;;;;OAOG;IACH,OAAO,CAAC,gBAAgB;YA0CV,eAAe;IAiD7B;;;;;;;;;OASG;IACH,OAAO,CAAC,cAAc;IAQtB;;;;;OAKG;IACH,OAAO,CAAC,QAAQ;IAwEhB,OAAO,CAAC,OAAO;IAWf,OAAO,CAAC,YAAY;IASpB,OAAO,CAAC,SAAS;IAMjB;;OAEG;IACH,OAAO,CAAC,SAAS;IAYjB,wDAAwD;IACxD,OAAO,CAAC,YAAY;IAOpB;;;;;;OAMG;IACH,OAAO,CAAC,gBAAgB;IAUxB;;;OAGG;IACH,OAAO,CAAC,SAAS;IAIjB;;;;;;;;;;;;;OAaG;IACG,MAAM,CACV,SAAS,EAAE,MAAM,GAAG,SAAS,EAC7B,GAAG,EAAE,OAAO,GACX,OAAO,CAAC;QACT,iBAAiB,CAAC,EAAE,MAAM,CAAC;QAC3B,iBAAiB,CAAC,EAAE,OAAO,CAAC;QAC5B,KAAK,EAAE,UAAU,EAAE,CAAC;QACpB,eAAe,CAAC,EAAE,MAAM,EAAE,EAAE,CAAC;QAC7B,UAAU,EAAE,MAAM,EAAE,CAAC;QACrB,SAAS,CAAC,EAAE,MAAM,CAAC;KACpB,CAAC,CA4ED;IAED,OAAO,CAAC,YAAY;IAepB;;;;;;;;OAQG;IACH,OAAO,CAAC,kBAAkB;IA0B1B,gEAAgE;IAChE,OAAO,CAAC,eAAe;IAavB;;;OAGG;IACH,OAAO,CAAC,eAAe;IAWvB;;;;;;;;;;;;OAYG;IACH,OAAO,CAAC,oBAAoB;IAiD5B;;;;;;;;;;;OAWG;IACH,OAAO,CAAC,YAAY;IAoEpB;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;OA+CG;IACG,MAAM,CACV,KAAK,EAAE,MAAM,EACb,KAAK,EAAE,MAAM,EACb,MAAM,EAAE,MAAM,GAAG,SAAS,EAC1B,GAAG,EAAE,OAAO,GACX,OAAO,CAAC;QACT,iBAAiB,CAAC,EAAE,IAAI,CAAC;QACzB,iBAAiB,CAAC,EAAE,MAAM,CAAC;QAC3B,iBAAiB,CAAC,EAAE,OAAO,CAAC;QAC5B,QAAQ,EAAE,aAAa,EAAE,CAAC;QAC1B,oBAAoB,CAAC,EAAE,IAAI,CAAC;QAC5B,UAAU,CAAC,EAAE,MAAM,CAAC;QACpB,YAAY,EAAE,MAAM,CAAC;QACrB,mBAAmB,CAAC,EAAE,MAAM,EAAE,CAAC;QAC/B,cAAc,CAAC,EAAE,MAAM,EAAE,CAAC;QAC1B,qBAAqB,CAAC,EAAE,IAAI,CAAC;QAC7B,iBAAiB,CAAC,EAAE,iBAAiB,CAAC;QACtC,iBAAiB,CAAC,EAAE,KAAK,CAAC;YAAE,YAAY,EAAE,MAAM,CAAC;YAAC,IAAI,EAAE,MAAM,CAAA;SAAE,CAAC,CAAC;KACnE,CAAC,CAkHD;CACF;AAwBD,wBAAgB,4BAA4B,CAAC,MAAM,EAAE,SAAS,EAAE,OAAO,EAAE,cAAc,GAAG,IAAI,CAE7F;AAED,wBAAgB,2BAA2B,IAAI,wBAAwB,CAOtE"}