riksdagsmonitor 0.8.14 → 0.8.15

package/dist/scripts/pipeline/orchestrator.js

@@ -0,0 +1,147 @@
+/**
+ * @module pipeline/orchestrator
+ * @description Unified pipeline orchestrator.
+ *
+ * The `PipelineOrchestrator` runs a collection of `ContentPipeline` instances
+ * either sequentially (default) or in parallel, collects results, logs
+ * progress, and surfaces a structured `OrchestratorResult`.
+ *
+ * Usage (sequential):
+ * ```ts
+ * import { PipelineOrchestrator } from './pipeline/orchestrator.js';
+ * import { MotionsPipeline } from './pipeline/plugins/motions-pipeline.js';
+ *
+ * const orchestrator = new PipelineOrchestrator({
+ *   pipelines: [new MotionsPipeline()],
+ * });
+ * const result = await orchestrator.run();
+ * ```
+ *
+ * @author Hack23 AB
+ * @license Apache-2.0
+ */
+// ---------------------------------------------------------------------------
+// PipelineOrchestrator
+// ---------------------------------------------------------------------------
+/**
+ * Orchestrates the execution of one or more `ContentPipeline` instances.
+ *
+ * Key behaviours:
+ * - **Error isolation**: a failure in one pipeline does not abort others.
+ * - **Parallel mode**: when `config.parallel = true` all pipelines run via
+ *   `Promise.allSettled` for throughput and robust error handling; otherwise
+ *   they execute sequentially for predictable log output.
+ * - **Result aggregation**: all per-pipeline `PipelineResult` objects are
+ *   merged into a single `OrchestratorResult`.
+ */
+export class PipelineOrchestrator {
+    pipelines;
+    parallel;
+    defaultOptions;
+    constructor(config) {
+        this.pipelines = config.pipelines;
+        this.parallel = config.parallel ?? false;
+        this.defaultOptions = config.defaultOptions ?? {};
+    }
+    // -------------------------------------------------------------------------
+    // Public API
+    // -------------------------------------------------------------------------
+    /**
+     * Run all pipelines and return an aggregated `OrchestratorResult`.
+     *
+     * @param overrideOptions - Options forwarded to every pipeline, merged on
+     *                          top of `defaultOptions`.
+     */
+    async run(overrideOptions) {
+        const startTime = Date.now();
+        const effectiveOptions = {
+            ...this.defaultOptions,
+            ...(overrideOptions ?? {}),
+        };
+        const results = {};
+        if (this.parallel) {
+            const settled = await Promise.allSettled(this.pipelines.map(p => this._runSingle(p, effectiveOptions)));
+            for (let i = 0; i < this.pipelines.length; i++) {
+                const pipeline = this.pipelines[i];
+                const outcome = settled[i];
+                if (outcome.status === 'fulfilled') {
+                    results[pipeline.name] = outcome.value;
+                }
+                else {
+                    // Promise should never reject because _runSingle catches all errors,
+                    // but handle it defensively.
+                    results[pipeline.name] = {
+                        success: false,
+                        error: String(outcome.reason),
+                        warnings: [],
+                        degraded: false,
+                        files: 0,
+                    };
+                }
+            }
+        }
+        else {
+            for (const pipeline of this.pipelines) {
+                results[pipeline.name] = await this._runSingle(pipeline, effectiveOptions);
+            }
+        }
+        return this._aggregate(results, Date.now() - startTime);
+    }
+    // -------------------------------------------------------------------------
+    // Private helpers
+    // -------------------------------------------------------------------------
+    /**
+     * Run a single pipeline, wrapping any unexpected throws in a failed result.
+     */
+    async _runSingle(pipeline, options) {
+        console.log(`[Orchestrator] ▶ Starting pipeline: ${pipeline.name}`);
+        const t0 = Date.now();
+        try {
+            const result = await pipeline.run(options);
+            const durationMs = Date.now() - t0;
+            console.log(`[Orchestrator] ${result.success ? '✅' : '❌'} Pipeline "${pipeline.name}" completed in ${durationMs}ms`);
+            return { ...result, durationMs };
+        }
+        catch (err) {
+            const durationMs = Date.now() - t0;
+            const message = err instanceof Error ? err.message : String(err);
+            console.error(`[Orchestrator] 💥 Pipeline "${pipeline.name}" threw unexpectedly after ${durationMs}ms: ${message}`);
+            return {
+                success: false,
+                error: message,
+                durationMs,
+                warnings: [`Unexpected throw from pipeline "${pipeline.name}": ${message}`],
+                degraded: false,
+                files: 0,
+            };
+        }
+    }
+    /**
+     * Aggregate individual pipeline results into a single orchestrator result.
+     */
+    _aggregate(results, totalDurationMs) {
+        let totalFiles = 0;
+        let allSucceeded = true;
+        const warnings = [];
+        for (const [name, result] of Object.entries(results)) {
+            if (!result.success) {
+                allSucceeded = false;
+                console.warn(`[Orchestrator] ⚠ Pipeline "${name}" did not succeed: ${result.error ?? 'unknown error'}`);
+            }
+            totalFiles += result.files ?? 0;
+            if (result.warnings) {
+                warnings.push(...result.warnings);
+            }
+        }
+        console.log(`[Orchestrator] 🏁 All pipelines done. ` +
+            `success=${allSucceeded}, files=${totalFiles}, duration=${totalDurationMs}ms`);
+        return {
+            allSucceeded,
+            totalFiles,
+            results,
+            warnings,
+            durationMs: totalDurationMs,
+        };
+    }
+}
+//# sourceMappingURL=orchestrator.js.map

package/dist/scripts/pipeline/orchestrator.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"orchestrator.js","sourceRoot":"","sources":["../../../scripts/pipeline/orchestrator.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;GAqBG;AAUH,8EAA8E;AAC9E,uBAAuB;AACvB,8EAA8E;AAE9E;;;;;;;;;;GAUG;AACH,MAAM,OAAO,oBAAoB;IACd,SAAS,CAAoB;IAC7B,QAAQ,CAAU;IAClB,cAAc,CAAkB;IAEjD,YAAY,MAA0B;QACpC,IAAI,CAAC,SAAS,GAAG,MAAM,CAAC,SAAS,CAAC;QAClC,IAAI,CAAC,QAAQ,GAAG,MAAM,CAAC,QAAQ,IAAI,KAAK,CAAC;QACzC,IAAI,CAAC,cAAc,GAAG,MAAM,CAAC,cAAc,IAAI,EAAE,CAAC;IACpD,CAAC;IAED,4EAA4E;IAC5E,aAAa;IACb,4EAA4E;IAE5E;;;;;OAKG;IACH,KAAK,CAAC,GAAG,CAAC,eAAiC;QACzC,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;QAC7B,MAAM,gBAAgB,GAAoB;YACxC,GAAG,IAAI,CAAC,cAAc;YACtB,GAAG,CAAC,eAAe,IAAI,EAAE,CAAC;SAC3B,CAAC;QAEF,MAAM,OAAO,GAAmC,EAAE,CAAC;QAEnD,IAAI,IAAI,CAAC,QAAQ,EAAE,CAAC;YAClB,MAAM,OAAO,GAAG,MAAM,OAAO,CAAC,UAAU,CACtC,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC,EAAE,gBAAgB,CAAC,CAAC,CAC9D,CAAC;YACF,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,SAAS,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;gBAC/C,MAAM,QAAQ,GAAG,IAAI,CAAC,SAAS,CAAC,CAAC,CAAE,CAAC;gBACpC,MAAM,OAAO,GAAG,OAAO,CAAC,CAAC,CAAE,CAAC;gBAC5B,IAAI,OAAO,CAAC,MAAM,KAAK,WAAW,EAAE,CAAC;oBACnC,OAAO,CAAC,QAAQ,CAAC,IAAI,CAAC,GAAG,OAAO,CAAC,KAAK,CAAC;gBACzC,CAAC;qBAAM,CAAC;oBACN,qEAAqE;oBACrE,6BAA6B;oBAC7B,OAAO,CAAC,QAAQ,CAAC,IAAI,CAAC,GAAG;wBACvB,OAAO,EAAE,KAAK;wBACd,KAAK,EAAE,MAAM,CAAE,OAAiC,CAAC,MAAM,CAAC;wBACxD,QAAQ,EAAE,EAAE;wBACZ,QAAQ,EAAE,KAAK;wBACf,KAAK,EAAE,CAAC;qBACT,CAAC;gBACJ,CAAC;YACH,CAAC;QACH,CAAC;aAAM,CAAC;YACN,KAAK,MAAM,QAAQ,IAAI,IAAI,CAAC,SAAS,EAAE,CAAC;gBACtC,OAAO,CAAC,QAAQ,CAAC,IAAI,CAAC,GAAG,MAAM,IAAI,CAAC,UAAU,CAAC,QAAQ,EAAE,gBAAgB,CAAC,CAAC;YAC7E,CAAC;QACH,CAAC;QAED,OAAO,IAAI,CAAC,UAAU,CAAC,OAAO,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,SAAS,CAAC,CAAC;IAC1D,CAAC;IAED,4EAA4E;IAC5E,kBAAkB;IAClB,4EAA4E;IAE5E;;OAEG;IACK,KAAK,CAAC,UAAU,CACtB,QAAyB,EACzB,OAAwB;QAExB,OAAO,CAAC,GAAG,CAAC,uCAAuC,QAAQ,CAAC,IAAI,EAAE,CAAC,CAAC;QACpE,MAAM,EAAE,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;QACtB,IAAI,CAAC;YACH,MAAM,MAAM,GAAG,MAAM,QAAQ,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC;YAC3C,MAAM,UAAU,GAAG,IAAI,CAAC,GAAG,EAAE,GAAG,EAAE,CAAC;YACnC,OAAO,CAAC,GAAG,CACT,kBAAkB,MAAM,CAAC,OAAO,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,GAAG,cAAc,QAAQ,CAAC,IAAI,kBAAkB,UAAU,IAAI,CACxG,CAAC;YACF,OAAO,EAAE,GAAG,MAAM,EAAE,UAAU,EAAE,CAAC;QACnC,CAAC;QAAC,OAAO,GAAY,EAAE,CAAC;YACtB,MAAM,UAAU,GAAG,IAAI,CAAC,GAAG,EAAE,GAAG,EAAE,CAAC;YACnC,MAAM,OAAO,GAAG,GAAG,YAAY,KAAK,CAAC,CAAC,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC;YACjE,OAAO,CAAC,KAAK,CACX,+BAA+B,QAAQ,CAAC,IAAI,8BAA8B,UAAU,OAAO,OAAO,EAAE,CACrG,CAAC;YACF,OAAO;gBACL,OAAO,EAAE,KAAK;gBACd,KAAK,EAAE,OAAO;gBACd,UAAU;gBACV,QAAQ,EAAE,CAAC,mCAAmC,QAAQ,CAAC,IAAI,MAAM,OAAO,EAAE,CAAC;gBAC3E,QAAQ,EAAE,KAAK;gBACf,KAAK,EAAE,CAAC;aACT,CAAC;QACJ,CAAC;IACH,CAAC;IAED;;OAEG;IACK,UAAU,CAChB,OAAuC,EACvC,eAAuB;QAEvB,IAAI,UAAU,GAAG,CAAC,CAAC;QACnB,IAAI,YAAY,GAAG,IAAI,CAAC;QACxB,MAAM,QAAQ,GAAa,EAAE,CAAC;QAE9B,KAAK,MAAM,CAAC,IAAI,EAAE,MAAM,CAAC,IAAI,MAAM,CAAC,OAAO,CAAC,OAAO,CAAC,EAAE,CAAC;YACrD,IAAI,CAAC,MAAM,CAAC,OAAO,EAAE,CAAC;gBACpB,YAAY,GAAG,KAAK,CAAC;gBACrB,OAAO,CAAC,IAAI,CAAC,8BAA8B,IAAI,sBAAsB,MAAM,CAAC,KAAK,IAAI,eAAe,EAAE,CAAC,CAAC;YAC1G,CAAC;YACD,UAAU,IAAI,MAAM,CAAC,KAAK,IAAI,CAAC,CAAC;YAChC,IAAI,MAAM,CAAC,QAAQ,EAAE,CAAC;gBACpB,QAAQ,CAAC,IAAI,CAAC,GAAG,MAAM,CAAC,QAAQ,CAAC,CAAC;YACpC,CAAC;QACH,CAAC;QAED,OAAO,CAAC,GAAG,CACT,wCAAwC;YACtC,WAAW,YAAY,WAAW,UAAU,cAAc,eAAe,IAAI,CAChF,CAAC;QAEF,OAAO;YACL,YAAY;YACZ,UAAU;YACV,OAAO;YACP,QAAQ;YACR,UAAU,EAAE,eAAe;SAC5B,CAAC;IACJ,CAAC;CACF"}

package/dist/scripts/pipeline/types.d.ts

@@ -0,0 +1,117 @@
+/**
+ * @module pipeline/types
+ * @description ContentPipeline interface and related types for the standardised
+ * article-generation lifecycle: fetch → transform → generate → validate → write.
+ *
+ * Each article type implements `ContentPipeline` and is executed by the
+ * `PipelineOrchestrator` in `orchestrator.ts`.
+ *
+ * @author Hack23 AB
+ * @license Apache-2.0
+ */
+import type { Language } from '../types/language.js';
+import type { GenerationResult } from '../types/article.js';
+/**
+ * Named stages of the article-generation lifecycle.
+ * Stages execute sequentially for each language variant.
+ */
+export type PipelineStage = 'fetch' | 'transform' | 'generate' | 'validate' | 'write';
+/**
+ * Rich result returned by a completed pipeline run.
+ * Extends `GenerationResult` with per-stage timing and degradation metadata.
+ */
+export interface PipelineResult extends GenerationResult {
+    /** Wall-clock time in milliseconds for the complete pipeline run. */
+    durationMs?: number;
+    /**
+     * Per-stage duration breakdown (ms).
+     * Keys match `PipelineStage` values.
+     */
+    stageDurations?: Partial<Record<PipelineStage, number>>;
+    /**
+     * Warnings collected during the run (e.g. MCP queries that returned no data
+     * but were handled via graceful degradation).
+     */
+    warnings?: string[];
+    /**
+     * Whether the pipeline used cached / fallback data for any stage.
+     * `true` indicates at least one graceful-degradation path was taken.
+     */
+    degraded?: boolean;
+}
+/**
+ * Common options accepted by every `ContentPipeline.run()` implementation.
+ */
+export interface PipelineOptions {
+    /** Language variants to generate. Defaults to `['en', 'sv']`. */
+    languages?: Language[];
+    /**
+     * Callback invoked after each article HTML is generated.
+     * Used by the orchestrator to write files.  When `null` the article is
+     * generated in-memory only (useful for tests / dry-run mode).
+     */
+    writeArticle?: ((html: string, filename: string) => Promise<void>) | null;
+    /**
+     * When `true` articles are generated even when source data is sparse.
+     * Defaults to `false`.
+     */
+    allowDegradedContent?: boolean;
+}
+/**
+ * Contract that every article-type plugin must implement.
+ *
+ * Lifecycle:
+ * 1. **fetch**     – Retrieve raw data from MCP tools (may partially fail).
+ * 2. **transform** – Convert raw data into article payloads per language.
+ * 3. **generate**  – Render HTML via `generateArticleHTML`.
+ * 4. **validate**  – Verify HTML structure before writing.
+ * 5. **write**     – Persist files to `news/` directory.
+ *
+ * Implementations are responsible for graceful degradation: if the `fetch`
+ * stage partially fails they should log a warning and continue with whatever
+ * data is available rather than throwing.
+ */
+export interface ContentPipeline {
+    /**
+     * Human-readable name of this pipeline (e.g. `'motions'`).
+     * Used for logging and metrics.
+     */
+    readonly name: string;
+    /**
+     * Execute the full pipeline lifecycle and return a `PipelineResult`.
+     *
+     * Implementations **must not throw** — all errors should be caught and
+     * returned via `PipelineResult.success = false` and `PipelineResult.error`.
+     */
+    run(options?: PipelineOptions): Promise<PipelineResult>;
+}
+/**
+ * Configuration passed to `PipelineOrchestrator`.
+ */
+export interface OrchestratorConfig {
+    /** Pipelines to run.  Order matters when `parallel = false`. */
+    pipelines: ContentPipeline[];
+    /**
+     * When `true` all pipelines run concurrently via `Promise.allSettled`.
+     * Defaults to `false` (sequential) for predictable logging.
+     */
+    parallel?: boolean;
+    /** Default options forwarded to each pipeline unless overridden. */
+    defaultOptions?: PipelineOptions;
+}
+/**
+ * Aggregate result produced by the orchestrator after running all pipelines.
+ */
+export interface OrchestratorResult {
+    /** `true` when every pipeline in the run reported success. */
+    allSucceeded: boolean;
+    /** Total number of files written across all pipelines. */
+    totalFiles: number;
+    /** Individual results keyed by pipeline name. */
+    results: Record<string, PipelineResult>;
+    /** Warnings collected across all pipelines. */
+    warnings: string[];
+    /** Total wall-clock time for the orchestrator run (ms). */
+    durationMs: number;
+}
+//# sourceMappingURL=types.d.ts.map

package/dist/scripts/pipeline/types.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"types.d.ts","sourceRoot":"","sources":["../../../scripts/pipeline/types.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;GAUG;AAEH,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,sBAAsB,CAAC;AACrD,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,qBAAqB,CAAC;AAM5D;;;GAGG;AACH,MAAM,MAAM,aAAa,GAAG,OAAO,GAAG,WAAW,GAAG,UAAU,GAAG,UAAU,GAAG,OAAO,CAAC;AAMtF;;;GAGG;AACH,MAAM,WAAW,cAAe,SAAQ,gBAAgB;IACtD,qEAAqE;IACrE,UAAU,CAAC,EAAE,MAAM,CAAC;IACpB;;;OAGG;IACH,cAAc,CAAC,EAAE,OAAO,CAAC,MAAM,CAAC,aAAa,EAAE,MAAM,CAAC,CAAC,CAAC;IACxD;;;OAGG;IACH,QAAQ,CAAC,EAAE,MAAM,EAAE,CAAC;IACpB;;;OAGG;IACH,QAAQ,CAAC,EAAE,OAAO,CAAC;CACpB;AAMD;;GAEG;AACH,MAAM,WAAW,eAAe;IAC9B,iEAAiE;IACjE,SAAS,CAAC,EAAE,QAAQ,EAAE,CAAC;IACvB;;;;OAIG;IACH,YAAY,CAAC,EAAE,CAAC,CAAC,IAAI,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,KAAK,OAAO,CAAC,IAAI,CAAC,CAAC,GAAG,IAAI,CAAC;IAC1E;;;OAGG;IACH,oBAAoB,CAAC,EAAE,OAAO,CAAC;CAChC;AAMD;;;;;;;;;;;;;GAaG;AACH,MAAM,WAAW,eAAe;IAC9B;;;OAGG;IACH,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAC;IAEtB;;;;;OAKG;IACH,GAAG,CAAC,OAAO,CAAC,EAAE,eAAe,GAAG,OAAO,CAAC,cAAc,CAAC,CAAC;CACzD;AAMD;;GAEG;AACH,MAAM,WAAW,kBAAkB;IACjC,gEAAgE;IAChE,SAAS,EAAE,eAAe,EAAE,CAAC;IAC7B;;;OAGG;IACH,QAAQ,CAAC,EAAE,OAAO,CAAC;IACnB,oEAAoE;IACpE,cAAc,CAAC,EAAE,eAAe,CAAC;CAClC;AAED;;GAEG;AACH,MAAM,WAAW,kBAAkB;IACjC,8DAA8D;IAC9D,YAAY,EAAE,OAAO,CAAC;IACtB,0DAA0D;IAC1D,UAAU,EAAE,MAAM,CAAC;IACnB,iDAAiD;IACjD,OAAO,EAAE,MAAM,CAAC,MAAM,EAAE,cAAc,CAAC,CAAC;IACxC,+CAA+C;IAC/C,QAAQ,EAAE,MAAM,EAAE,CAAC;IACnB,2DAA2D;IAC3D,UAAU,EAAE,MAAM,CAAC;CACpB"}

package/dist/scripts/pipeline/types.js

@@ -0,0 +1,13 @@
+/**
+ * @module pipeline/types
+ * @description ContentPipeline interface and related types for the standardised
+ * article-generation lifecycle: fetch → transform → generate → validate → write.
+ *
+ * Each article type implements `ContentPipeline` and is executed by the
+ * `PipelineOrchestrator` in `orchestrator.ts`.
+ *
+ * @author Hack23 AB
+ * @license Apache-2.0
+ */
+export {};
+//# sourceMappingURL=types.js.map

package/dist/scripts/pipeline/types.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"types.js","sourceRoot":"","sources":["../../../scripts/pipeline/types.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;GAUG"}

package/dist/scripts/pipeline/validation.d.ts

@@ -0,0 +1,66 @@
+/**
+ * @module pipeline/validation
+ * @description Post-generation HTML structure validation.
+ *
+ * Validates that generated article HTML meets minimum structural requirements
+ * before the file is written to disk.  Failures are non-fatal by default —
+ * the orchestrator collects validation warnings and continues.
+ *
+ * @author Hack23 AB
+ * @license Apache-2.0
+ */
+/**
+ * Result of validating a single HTML article string.
+ */
+export interface ArticleValidationResult {
+    /** `true` when the HTML passes all required checks. */
+    passed: boolean;
+    /** Informational messages about checks that passed. */
+    passedChecks: string[];
+    /** Error messages for checks that failed (non-empty means `passed = false`). */
+    errors: string[];
+    /** Warning messages for checks that are advisory only. */
+    warnings: string[];
+}
+/**
+ * Options controlling which checks are enforced.
+ */
+export interface ValidationOptions {
+    /** Require a `<h1>` element (default: `true`). */
+    requireH1?: boolean;
+    /** Require at least one `<h2>` section (default: `true`). */
+    requireSections?: boolean;
+    /** Require the sources footer block and fail when absent (default: `true`). */
+    requireSources?: boolean;
+    /** Minimum word count threshold (default: `50`). */
+    minWordCount?: number;
+    /** Require valid `<html lang="…">` attribute (default: `true`). */
+    requireLangAttr?: boolean;
+    /** Require `<!DOCTYPE html>` declaration (default: `true`). */
+    requireDoctype?: boolean;
+}
+/**
+ * Validate the structure of a generated article HTML string.
+ *
+ * This is a lightweight regex / string-based check, not a full DOM parse.
+ * It is intentionally fast and dependency-free.
+ *
+ * @param html   - The complete HTML string to validate.
+ * @param opts   - Optional configuration overrides.
+ * @returns      Structured validation result.
+ */
+export declare function validateArticleHTML(html: string, opts?: ValidationOptions): ArticleValidationResult;
+/**
+ * Validate a batch of articles and return a summary.
+ *
+ * @param articles - Array of `{ filename, html }` objects.
+ * @param opts     - Optional validation configuration.
+ * @returns        Array of per-article validation results.
+ */
+export declare function validateArticleBatch(articles: ReadonlyArray<{
+    filename: string;
+    html: string;
+}>, opts?: ValidationOptions): Array<ArticleValidationResult & {
+    filename: string;
+}>;
+//# sourceMappingURL=validation.d.ts.map

package/dist/scripts/pipeline/validation.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"validation.d.ts","sourceRoot":"","sources":["../../../scripts/pipeline/validation.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;GAUG;AAMH;;GAEG;AACH,MAAM,WAAW,uBAAuB;IACtC,uDAAuD;IACvD,MAAM,EAAE,OAAO,CAAC;IAChB,uDAAuD;IACvD,YAAY,EAAE,MAAM,EAAE,CAAC;IACvB,gFAAgF;IAChF,MAAM,EAAE,MAAM,EAAE,CAAC;IACjB,0DAA0D;IAC1D,QAAQ,EAAE,MAAM,EAAE,CAAC;CACpB;AAED;;GAEG;AACH,MAAM,WAAW,iBAAiB;IAChC,kDAAkD;IAClD,SAAS,CAAC,EAAE,OAAO,CAAC;IACpB,6DAA6D;IAC7D,eAAe,CAAC,EAAE,OAAO,CAAC;IAC1B,+EAA+E;IAC/E,cAAc,CAAC,EAAE,OAAO,CAAC;IACzB,oDAAoD;IACpD,YAAY,CAAC,EAAE,MAAM,CAAC;IACtB,mEAAmE;IACnE,eAAe,CAAC,EAAE,OAAO,CAAC;IAC1B,+DAA+D;IAC/D,cAAc,CAAC,EAAE,OAAO,CAAC;CAC1B;AAmBD;;;;;;;;;GASG;AACH,wBAAgB,mBAAmB,CACjC,IAAI,EAAE,MAAM,EACZ,IAAI,GAAE,iBAAsB,GAC3B,uBAAuB,CAoFzB;AAED;;;;;;GAMG;AACH,wBAAgB,oBAAoB,CAClC,QAAQ,EAAE,aAAa,CAAC;IAAE,QAAQ,EAAE,MAAM,CAAC;IAAC,IAAI,EAAE,MAAM,CAAA;CAAE,CAAC,EAC3D,IAAI,GAAE,iBAAsB,GAC3B,KAAK,CAAC,uBAAuB,GAAG;IAAE,QAAQ,EAAE,MAAM,CAAA;CAAE,CAAC,CAKvD"}

package/dist/scripts/pipeline/validation.js

@@ -0,0 +1,129 @@
+/**
+ * @module pipeline/validation
+ * @description Post-generation HTML structure validation.
+ *
+ * Validates that generated article HTML meets minimum structural requirements
+ * before the file is written to disk.  Failures are non-fatal by default —
+ * the orchestrator collects validation warnings and continues.
+ *
+ * @author Hack23 AB
+ * @license Apache-2.0
+ */
+// ---------------------------------------------------------------------------
+// Default options
+// ---------------------------------------------------------------------------
+const DEFAULT_OPTIONS = {
+    requireH1: true,
+    requireSections: true,
+    requireSources: true,
+    minWordCount: 50,
+    requireLangAttr: true,
+    requireDoctype: true,
+};
+// ---------------------------------------------------------------------------
+// HTML structure validation
+// ---------------------------------------------------------------------------
+/**
+ * Validate the structure of a generated article HTML string.
+ *
+ * This is a lightweight regex / string-based check, not a full DOM parse.
+ * It is intentionally fast and dependency-free.
+ *
+ * @param html   - The complete HTML string to validate.
+ * @param opts   - Optional configuration overrides.
+ * @returns      Structured validation result.
+ */
+export function validateArticleHTML(html, opts = {}) {
+    const options = { ...DEFAULT_OPTIONS, ...opts };
+    const errors = [];
+    const warnings = [];
+    const passedChecks = [];
+    if (!html || typeof html !== 'string') {
+        return {
+            passed: false,
+            passedChecks,
+            errors: ['HTML is empty or not a string'],
+            warnings,
+        };
+    }
+    // --- DOCTYPE ---
+    if (options.requireDoctype) {
+        if (/<!DOCTYPE\s+html>/i.test(html)) {
+            passedChecks.push('DOCTYPE present');
+        }
+        else {
+            errors.push('Missing <!DOCTYPE html> declaration');
+        }
+    }
+    // --- lang attribute ---
+    if (options.requireLangAttr) {
+        if (/<html[^>]+lang=["'][a-z]{2,5}["']/i.test(html)) {
+            passedChecks.push('lang attribute present');
+        }
+        else {
+            errors.push('Missing valid lang attribute on <html> element');
+        }
+    }
+    // --- H1 ---
+    if (options.requireH1) {
+        if (/<h1[^>]*>[\s\S]+?<\/h1>/i.test(html)) {
+            passedChecks.push('H1 heading present');
+        }
+        else {
+            errors.push('Missing <h1> heading');
+        }
+    }
+    // --- Sections (H2) ---
+    if (options.requireSections) {
+        const h2Matches = html.match(/<h2[^>]*>/gi);
+        const h2Count = h2Matches ? h2Matches.length : 0;
+        if (h2Count >= 1) {
+            passedChecks.push(`${h2Count} <h2> section(s) present`);
+        }
+        else {
+            errors.push('No <h2> sections found — article content may be missing');
+        }
+    }
+    // --- Sources footer ---
+    if (options.requireSources) {
+        if (/article-sources|data-sources|riksdag-regering-mcp/i.test(html)) {
+            passedChecks.push('Sources block present');
+        }
+        else {
+            errors.push('Sources footer block not detected — article may lack attribution');
+        }
+    }
+    // --- Word count (article body only: strip <head>, <script>, <style> blocks first) ---
+    const bodyOnly = html
+        .replace(/<head[\s\S]*?<\/head>/gi, '')
+        .replace(/<script[\s\S]*?<\/script>/gi, '')
+        .replace(/<style[\s\S]*?<\/style>/gi, '');
+    const textContent = bodyOnly.replace(/<[^>]+>/g, ' ').replace(/\s+/g, ' ').trim();
+    const wordCount = textContent.split(' ').filter(w => w.length > 0).length;
+    if (wordCount >= options.minWordCount) {
+        passedChecks.push(`Word count ${wordCount} meets minimum ${options.minWordCount}`);
+    }
+    else {
+        errors.push(`Word count ${wordCount} is below minimum ${options.minWordCount} — article content may be too thin`);
+    }
+    return {
+        passed: errors.length === 0,
+        passedChecks,
+        errors,
+        warnings,
+    };
+}
+/**
+ * Validate a batch of articles and return a summary.
+ *
+ * @param articles - Array of `{ filename, html }` objects.
+ * @param opts     - Optional validation configuration.
+ * @returns        Array of per-article validation results.
+ */
+export function validateArticleBatch(articles, opts = {}) {
+    return articles.map(({ filename, html }) => ({
+        filename,
+        ...validateArticleHTML(html, opts),
+    }));
+}
+//# sourceMappingURL=validation.js.map

package/dist/scripts/pipeline/validation.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"validation.js","sourceRoot":"","sources":["../../../scripts/pipeline/validation.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;GAUG;AAsCH,8EAA8E;AAC9E,kBAAkB;AAClB,8EAA8E;AAE9E,MAAM,eAAe,GAAgC;IACnD,SAAS,EAAE,IAAI;IACf,eAAe,EAAE,IAAI;IACrB,cAAc,EAAE,IAAI;IACpB,YAAY,EAAE,EAAE;IAChB,eAAe,EAAE,IAAI;IACrB,cAAc,EAAE,IAAI;CACrB,CAAC;AAEF,8EAA8E;AAC9E,4BAA4B;AAC5B,8EAA8E;AAE9E;;;;;;;;;GASG;AACH,MAAM,UAAU,mBAAmB,CACjC,IAAY,EACZ,OAA0B,EAAE;IAE5B,MAAM,OAAO,GAAgC,EAAE,GAAG,eAAe,EAAE,GAAG,IAAI,EAAE,CAAC;IAE7E,MAAM,MAAM,GAAa,EAAE,CAAC;IAC5B,MAAM,QAAQ,GAAa,EAAE,CAAC;IAC9B,MAAM,YAAY,GAAa,EAAE,CAAC;IAElC,IAAI,CAAC,IAAI,IAAI,OAAO,IAAI,KAAK,QAAQ,EAAE,CAAC;QACtC,OAAO;YACL,MAAM,EAAE,KAAK;YACb,YAAY;YACZ,MAAM,EAAE,CAAC,+BAA+B,CAAC;YACzC,QAAQ;SACT,CAAC;IACJ,CAAC;IAED,kBAAkB;IAClB,IAAI,OAAO,CAAC,cAAc,EAAE,CAAC;QAC3B,IAAI,oBAAoB,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;YACpC,YAAY,CAAC,IAAI,CAAC,iBAAiB,CAAC,CAAC;QACvC,CAAC;aAAM,CAAC;YACN,MAAM,CAAC,IAAI,CAAC,qCAAqC,CAAC,CAAC;QACrD,CAAC;IACH,CAAC;IAED,yBAAyB;IACzB,IAAI,OAAO,CAAC,eAAe,EAAE,CAAC;QAC5B,IAAI,oCAAoC,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;YACpD,YAAY,CAAC,IAAI,CAAC,wBAAwB,CAAC,CAAC;QAC9C,CAAC;aAAM,CAAC;YACN,MAAM,CAAC,IAAI,CAAC,gDAAgD,CAAC,CAAC;QAChE,CAAC;IACH,CAAC;IAED,aAAa;IACb,IAAI,OAAO,CAAC,SAAS,EAAE,CAAC;QACtB,IAAI,0BAA0B,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;YAC1C,YAAY,CAAC,IAAI,CAAC,oBAAoB,CAAC,CAAC;QAC1C,CAAC;aAAM,CAAC;YACN,MAAM,CAAC,IAAI,CAAC,sBAAsB,CAAC,CAAC;QACtC,CAAC;IACH,CAAC;IAED,wBAAwB;IACxB,IAAI,OAAO,CAAC,eAAe,EAAE,CAAC;QAC5B,MAAM,SAAS,GAAG,IAAI,CAAC,KAAK,CAAC,aAAa,CAAC,CAAC;QAC5C,MAAM,OAAO,GAAG,SAAS,CAAC,CAAC,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC;QACjD,IAAI,OAAO,IAAI,CAAC,EAAE,CAAC;YACjB,YAAY,CAAC,IAAI,CAAC,GAAG,OAAO,0BAA0B,CAAC,CAAC;QAC1D,CAAC;aAAM,CAAC;YACN,MAAM,CAAC,IAAI,CAAC,yDAAyD,CAAC,CAAC;QACzE,CAAC;IACH,CAAC;IAED,yBAAyB;IACzB,IAAI,OAAO,CAAC,cAAc,EAAE,CAAC;QAC3B,IAAI,oDAAoD,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;YACpE,YAAY,CAAC,IAAI,CAAC,uBAAuB,CAAC,CAAC;QAC7C,CAAC;aAAM,CAAC;YACN,MAAM,CAAC,IAAI,CAAC,kEAAkE,CAAC,CAAC;QAClF,CAAC;IACH,CAAC;IAED,uFAAuF;IACvF,MAAM,QAAQ,GAAG,IAAI;SAClB,OAAO,CAAC,yBAAyB,EAAE,EAAE,CAAC;SACtC,OAAO,CAAC,6BAA6B,EAAE,EAAE,CAAC;SAC1C,OAAO,CAAC,2BAA2B,EAAE,EAAE,CAAC,CAAC;IAC5C,MAAM,WAAW,GAAG,QAAQ,CAAC,OAAO,CAAC,UAAU,EAAE,GAAG,CAAC,CAAC,OAAO,CAAC,MAAM,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC;IAClF,MAAM,SAAS,GAAG,WAAW,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,MAAM,CAAC;IAC1E,IAAI,SAAS,IAAI,OAAO,CAAC,YAAY,EAAE,CAAC;QACtC,YAAY,CAAC,IAAI,CAAC,cAAc,SAAS,kBAAkB,OAAO,CAAC,YAAY,EAAE,CAAC,CAAC;IACrF,CAAC;SAAM,CAAC;QACN,MAAM,CAAC,IAAI,CACT,cAAc,SAAS,qBAAqB,OAAO,CAAC,YAAY,oCAAoC,CACrG,CAAC;IACJ,CAAC;IAED,OAAO;QACL,MAAM,EAAE,MAAM,CAAC,MAAM,KAAK,CAAC;QAC3B,YAAY;QACZ,MAAM;QACN,QAAQ;KACT,CAAC;AACJ,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,oBAAoB,CAClC,QAA2D,EAC3D,OAA0B,EAAE;IAE5B,OAAO,QAAQ,CAAC,GAAG,CAAC,CAAC,EAAE,QAAQ,EAAE,IAAI,EAAE,EAAE,EAAE,CAAC,CAAC;QAC3C,QAAQ;QACR,GAAG,mBAAmB,CAAC,IAAI,EAAE,IAAI,CAAC;KACnC,CAAC,CAAC,CAAC;AACN,CAAC"}

package/dist/scripts/populate-analysis-data.d.ts

@@ -0,0 +1,41 @@
+#!/usr/bin/env tsx
+/**
+ * @module populate-analysis-data
+ * @description Standalone script that fetches recent data from **all** MCP sources
+ * (riksdag-regering, SCB, World Bank) and persists it to `analysis/data/`.
+ *
+ * This populates the analysis folder with **all** supported data types:
+ * - **Documents**: propositions, motions, committeeReports, votes, speeches,
+ *   questions, interpellations
+ * - **Calendar events**: upcoming parliamentary events
+ * - **MPs**: current member profiles
+ * - **Government documents**: recent government publications (regeringen.se)
+ * - **Voting groups**: party-level voting patterns
+ * - **World Bank indicators**: Swedish economic context (GDP, unemployment, etc.)
+ * - **SCB statistics**: Swedish official statistics for key policy domains
+ *
+ * The script reuses the existing MCP client and data-persistence modules,
+ * ensuring the collision-free sidecar design (data + .meta.json) is
+ * consistently applied.
+ *
+ * Usage:
+ *   npx tsx scripts/populate-analysis-data.ts [--limit N] [--date YYYY-MM-DD]
+ *
+ * Options:
+ *   --limit N           Max documents per type (default: 20)
+ *   --date YYYY-MM-DD   Target date for analysis (default: today UTC)
+ *
+ * @author Hack23 AB
+ * @license Apache-2.0
+ */
+/**
+ * Compute riksmöte (parliamentary session year) from an ISO date string.
+ * Swedish parliamentary year runs Oct → Sep:
+ *   - 2025-10-01 to 2026-09-30 → "2025/26"
+ *   - 2026-10-01 to 2027-09-30 → "2026/27"
+ *
+ * @param dateStr - ISO date string (YYYY-MM-DD)
+ * @returns Riksmöte identifier (e.g. "2025/26")
+ */
+export declare function riksMoteFromDate(dateStr: string): string;
+//# sourceMappingURL=populate-analysis-data.d.ts.map

package/dist/scripts/populate-analysis-data.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"populate-analysis-data.d.ts","sourceRoot":"","sources":["../../scripts/populate-analysis-data.ts"],"names":[],"mappings":";AACA;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA4BG;AAgCH;;;;;;;;GAQG;AACH,wBAAgB,gBAAgB,CAAC,OAAO,EAAE,MAAM,GAAG,MAAM,CAMxD"}