wiki-graph-core 0.3.1 → 0.4.1

package/dist/external/wikimatch/surface/screening.js

@@ -0,0 +1,168 @@
+import { z } from "zod";
+import {
+  ParsedJsonError,
+  RESPONSE_INTENT_CLASSIFIER_PROMPT_TEMPLATE,
+  requestGuaranteedJson
+} from "../../guaranteed/index.js";
+const protectedSurfaceSchema = z.object({
+  note: z.string().max(24).optional(),
+  surfaceId: z.string().min(1)
+}).strict();
+const protectionResponseSchema = z.object({
+  protectedSurfaces: z.array(protectedSurfaceSchema)
+}).strict();
+async function judgeWikimatchSurfaceProtection(options) {
+  if (options.suspiciousSurfaces.length === 0) {
+    return { protectedSurfaces: [] };
+  }
+  try {
+    return await requestGuaranteedJson({
+      messages: buildProtectionMessages(options),
+      parse: (response) => parseSurfaceProtectionResponse(
+        options.suspiciousSurfaces,
+        normalizeProtectionResponse(response)
+      ),
+      request: options.request,
+      responseIntentClassifierPrompt: RESPONSE_INTENT_CLASSIFIER_PROMPT_TEMPLATE,
+      schema: protectionResponseSchema,
+      ...options.maxRetries === void 0 ? {} : { maxRetries: options.maxRetries }
+    });
+  } catch (error) {
+    return {
+      fallback: {
+        issues: [formatFallbackIssue(error)],
+        reason: "guaranteed_json_failed"
+      },
+      protectedSurfaces: []
+    };
+  }
+}
+function normalizeProtectionResponse(response) {
+  return {
+    protectedSurfaces: response.protectedSurfaces.map((surface) => ({
+      ...surface.note === void 0 ? {} : { note: surface.note },
+      surfaceId: surface.surfaceId
+    }))
+  };
+}
+function parseSurfaceProtectionResponse(surfaces, response) {
+  const issues = validateSurfaceProtectionResponse(surfaces, response);
+  if (issues.length > 0) {
+    throw new ParsedJsonError(issues);
+  }
+  const surfacesById = createSurfaceMap(surfaces);
+  const protectedSurfaces = [];
+  for (const surface of response.protectedSurfaces) {
+    const source = surfacesById.get(surface.surfaceId);
+    protectedSurfaces.push({
+      ...surface.note === void 0 ? {} : { note: surface.note },
+      surfaceId: surface.surfaceId,
+      text: source.text
+    });
+  }
+  return { protectedSurfaces };
+}
+function validateSurfaceProtectionResponse(surfaces, response) {
+  const issues = [];
+  const surfacesById = createSurfaceMap(surfaces);
+  const seenIds = /* @__PURE__ */ new Set();
+  for (const surface of response.protectedSurfaces) {
+    if (!surfacesById.has(surface.surfaceId)) {
+      issues.push(
+        `Unknown surfaceId "${surface.surfaceId}". Use only these surface IDs: ${[
+          ...surfacesById.keys()
+        ].join(", ")}.`
+      );
+      continue;
+    }
+    if (seenIds.has(surface.surfaceId)) {
+      issues.push(`Duplicate protected surface ${surface.surfaceId}.`);
+      continue;
+    }
+    seenIds.add(surface.surfaceId);
+  }
+  return issues;
+}
+function buildProtectionMessages(input) {
+  return [
+    {
+      role: "system",
+      content: formatProtectionSystemPrompt(input)
+    },
+    {
+      role: "user",
+      content: formatProtectionPrompt(input)
+    }
+  ];
+}
+function formatProtectionSystemPrompt(input) {
+  return [
+    "You protect useful high-frequency surface strings before Wikidata grounding.",
+    "",
+    "The input surfaces are suspicious only because they are frequent after range suppression.",
+    "Most high-frequency function words, fragments, punctuation, generic words, and discourse words should NOT be protected.",
+    "A surface that is not protected will be removed before grounding.",
+    "",
+    "User recall policy:",
+    input.policyPrompt,
+    "",
+    "Protect a surface only when it is likely to be a meaningful entity, named concept, named event, place, person, organization, work, or domain term under the policy.",
+    "",
+    "Rules:",
+    "- Return JSON only.",
+    "- Return only protected surface IDs.",
+    "- Do not return a result for every input surface.",
+    "- Do not protect generic grammar words merely because they appear often.",
+    "- Do not infer or choose Wikidata QIDs in this stage."
+  ].join("\n");
+}
+function formatProtectionPrompt(input) {
+  return [
+    "Suspicious high-frequency surfaces:",
+    JSON.stringify(
+      input.suspiciousSurfaces.map((surface) => ({
+        count: surface.count,
+        surfaceId: surface.id,
+        text: surface.text
+      })),
+      null,
+      2
+    ),
+    "",
+    "Return this JSON shape:",
+    JSON.stringify(
+      {
+        protectedSurfaces: [
+          {
+            note: "optional, <= 12 Chinese chars or 6 English words",
+            surfaceId: "surface id from the input"
+          }
+        ]
+      },
+      null,
+      2
+    )
+  ].join("\n");
+}
+function createSurfaceMap(surfaces) {
+  return new Map(surfaces.map((surface) => [surface.id, surface]));
+}
+function formatFallbackIssue(error) {
+  if (error instanceof ParsedJsonError) {
+    return error.issues.join("\n");
+  }
+  if (error instanceof Error) {
+    const issues = error.issues;
+    if (Array.isArray(issues) && issues.every((issue) => typeof issue === "string")) {
+      return issues.join("\n");
+    }
+    return error.message;
+  }
+  return String(error);
+}
+export {
+  judgeWikimatchSurfaceProtection,
+  parseSurfaceProtectionResponse,
+  validateSurfaceProtectionResponse
+};
+//# sourceMappingURL=screening.js.map

package/dist/external/wikimatch/surface/screening.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../../../../src/external/wikimatch/surface/screening.ts"],"sourcesContent":["import { z } from \"zod\";\n\nimport {\n  ParsedJsonError,\n  RESPONSE_INTENT_CLASSIFIER_PROMPT_TEMPLATE,\n  requestGuaranteedJson,\n  type GuaranteedRequest,\n} from \"../../guaranteed/index.js\";\nimport type { LLMessage } from \"../../llm/index.js\";\n\nimport type {\n  WikimatchProtectedSurface,\n  WikimatchSurface,\n  WikimatchSurfaceProtectionInput,\n  WikimatchSurfaceProtectionResponse,\n  WikimatchSurfaceProtectionResult,\n} from \"../types.js\";\n\nconst protectedSurfaceSchema = z\n  .object({\n    note: z.string().max(24).optional(),\n    surfaceId: z.string().min(1),\n  })\n  .strict();\n\nconst protectionResponseSchema = z\n  .object({\n    protectedSurfaces: z.array(protectedSurfaceSchema),\n  })\n  .strict();\n\nexport interface JudgeWikimatchSurfaceProtectionOptions extends WikimatchSurfaceProtectionInput {\n  readonly maxRetries?: number;\n  readonly request: GuaranteedRequest;\n}\n\nexport async function judgeWikimatchSurfaceProtection(\n  options: JudgeWikimatchSurfaceProtectionOptions,\n): Promise<WikimatchSurfaceProtectionResult> {\n  if (options.suspiciousSurfaces.length === 0) {\n    return { protectedSurfaces: [] };\n  }\n\n  try {\n    return await requestGuaranteedJson({\n      messages: buildProtectionMessages(options),\n      parse: (response) =>\n        parseSurfaceProtectionResponse(\n          options.suspiciousSurfaces,\n          normalizeProtectionResponse(response),\n        ),\n      request: options.request,\n      responseIntentClassifierPrompt:\n        RESPONSE_INTENT_CLASSIFIER_PROMPT_TEMPLATE,\n      schema: protectionResponseSchema,\n      ...(options.maxRetries === undefined\n        ? {}\n        : { maxRetries: options.maxRetries }),\n    });\n  } catch (error) {\n    return {\n      fallback: {\n        issues: [formatFallbackIssue(error)],\n        reason: \"guaranteed_json_failed\",\n      },\n      protectedSurfaces: [],\n    };\n  }\n}\n\nfunction normalizeProtectionResponse(\n  response: z.infer<typeof protectionResponseSchema>,\n): WikimatchSurfaceProtectionResponse {\n  return {\n    protectedSurfaces: response.protectedSurfaces.map((surface) => ({\n      ...(surface.note === undefined ? {} : { note: surface.note }),\n      surfaceId: surface.surfaceId,\n    })),\n  };\n}\n\nexport function parseSurfaceProtectionResponse(\n  surfaces: readonly WikimatchSurface[],\n  response: WikimatchSurfaceProtectionResponse,\n): WikimatchSurfaceProtectionResult {\n  const issues = validateSurfaceProtectionResponse(surfaces, response);\n\n  if (issues.length > 0) {\n    throw new ParsedJsonError(issues);\n  }\n\n  const surfacesById = createSurfaceMap(surfaces);\n  const protectedSurfaces: WikimatchProtectedSurface[] = [];\n\n  for (const surface of response.protectedSurfaces) {\n    const source = surfacesById.get(surface.surfaceId)!;\n\n    protectedSurfaces.push({\n      ...(surface.note === undefined ? {} : { note: surface.note }),\n      surfaceId: surface.surfaceId,\n      text: source.text,\n    });\n  }\n\n  return { protectedSurfaces };\n}\n\nexport function validateSurfaceProtectionResponse(\n  surfaces: readonly WikimatchSurface[],\n  response: WikimatchSurfaceProtectionResponse,\n): readonly string[] {\n  const issues: string[] = [];\n  const surfacesById = createSurfaceMap(surfaces);\n  const seenIds = new Set<string>();\n\n  for (const surface of response.protectedSurfaces) {\n    if (!surfacesById.has(surface.surfaceId)) {\n      issues.push(\n        `Unknown surfaceId \"${surface.surfaceId}\". Use only these surface IDs: ${[\n          ...surfacesById.keys(),\n        ].join(\", \")}.`,\n      );\n      continue;\n    }\n    if (seenIds.has(surface.surfaceId)) {\n      issues.push(`Duplicate protected surface ${surface.surfaceId}.`);\n      continue;\n    }\n\n    seenIds.add(surface.surfaceId);\n  }\n\n  return issues;\n}\n\nfunction buildProtectionMessages(\n  input: WikimatchSurfaceProtectionInput,\n): LLMessage[] {\n  return [\n    {\n      role: \"system\",\n      content: formatProtectionSystemPrompt(input),\n    },\n    {\n      role: \"user\",\n      content: formatProtectionPrompt(input),\n    },\n  ];\n}\n\nfunction formatProtectionSystemPrompt(\n  input: WikimatchSurfaceProtectionInput,\n): string {\n  return [\n    \"You protect useful high-frequency surface strings before Wikidata grounding.\",\n    \"\",\n    \"The input surfaces are suspicious only because they are frequent after range suppression.\",\n    \"Most high-frequency function words, fragments, punctuation, generic words, and discourse words should NOT be protected.\",\n    \"A surface that is not protected will be removed before grounding.\",\n    \"\",\n    \"User recall policy:\",\n    input.policyPrompt,\n    \"\",\n    \"Protect a surface only when it is likely to be a meaningful entity, named concept, named event, place, person, organization, work, or domain term under the policy.\",\n    \"\",\n    \"Rules:\",\n    \"- Return JSON only.\",\n    \"- Return only protected surface IDs.\",\n    \"- Do not return a result for every input surface.\",\n    \"- Do not protect generic grammar words merely because they appear often.\",\n    \"- Do not infer or choose Wikidata QIDs in this stage.\",\n  ].join(\"\\n\");\n}\n\nfunction formatProtectionPrompt(\n  input: WikimatchSurfaceProtectionInput,\n): string {\n  return [\n    \"Suspicious high-frequency surfaces:\",\n    JSON.stringify(\n      input.suspiciousSurfaces.map((surface) => ({\n        count: surface.count,\n        surfaceId: surface.id,\n        text: surface.text,\n      })),\n      null,\n      2,\n    ),\n    \"\",\n    \"Return this JSON shape:\",\n    JSON.stringify(\n      {\n        protectedSurfaces: [\n          {\n            note: \"optional, <= 12 Chinese chars or 6 English words\",\n            surfaceId: \"surface id from the input\",\n          },\n        ],\n      },\n      null,\n      2,\n    ),\n  ].join(\"\\n\");\n}\n\nfunction createSurfaceMap(\n  surfaces: readonly WikimatchSurface[],\n): ReadonlyMap<string, WikimatchSurface> {\n  return new Map(surfaces.map((surface) => [surface.id, surface]));\n}\n\nfunction formatFallbackIssue(error: unknown): string {\n  if (error instanceof ParsedJsonError) {\n    return error.issues.join(\"\\n\");\n  }\n  if (error instanceof Error) {\n    const issues = (error as { readonly issues?: unknown }).issues;\n\n    if (\n      Array.isArray(issues) &&\n      issues.every((issue) => typeof issue === \"string\")\n    ) {\n      return issues.join(\"\\n\");\n    }\n\n    return error.message;\n  }\n\n  return String(error);\n}\n"],"mappings":"AAAA,SAAS,SAAS;AAElB;AAAA,EACE;AAAA,EACA;AAAA,EACA;AAAA,OAEK;AAWP,MAAM,yBAAyB,EAC5B,OAAO;AAAA,EACN,MAAM,EAAE,OAAO,EAAE,IAAI,EAAE,EAAE,SAAS;AAAA,EAClC,WAAW,EAAE,OAAO,EAAE,IAAI,CAAC;AAC7B,CAAC,EACA,OAAO;AAEV,MAAM,2BAA2B,EAC9B,OAAO;AAAA,EACN,mBAAmB,EAAE,MAAM,sBAAsB;AACnD,CAAC,EACA,OAAO;AAOV,eAAsB,gCACpB,SAC2C;AAC3C,MAAI,QAAQ,mBAAmB,WAAW,GAAG;AAC3C,WAAO,EAAE,mBAAmB,CAAC,EAAE;AAAA,EACjC;AAEA,MAAI;AACF,WAAO,MAAM,sBAAsB;AAAA,MACjC,UAAU,wBAAwB,OAAO;AAAA,MACzC,OAAO,CAAC,aACN;AAAA,QACE,QAAQ;AAAA,QACR,4BAA4B,QAAQ;AAAA,MACtC;AAAA,MACF,SAAS,QAAQ;AAAA,MACjB,gCACE;AAAA,MACF,QAAQ;AAAA,MACR,GAAI,QAAQ,eAAe,SACvB,CAAC,IACD,EAAE,YAAY,QAAQ,WAAW;AAAA,IACvC,CAAC;AAAA,EACH,SAAS,OAAO;AACd,WAAO;AAAA,MACL,UAAU;AAAA,QACR,QAAQ,CAAC,oBAAoB,KAAK,CAAC;AAAA,QACnC,QAAQ;AAAA,MACV;AAAA,MACA,mBAAmB,CAAC;AAAA,IACtB;AAAA,EACF;AACF;AAEA,SAAS,4BACP,UACoC;AACpC,SAAO;AAAA,IACL,mBAAmB,SAAS,kBAAkB,IAAI,CAAC,aAAa;AAAA,MAC9D,GAAI,QAAQ,SAAS,SAAY,CAAC,IAAI,EAAE,MAAM,QAAQ,KAAK;AAAA,MAC3D,WAAW,QAAQ;AAAA,IACrB,EAAE;AAAA,EACJ;AACF;AAEO,SAAS,+BACd,UACA,UACkC;AAClC,QAAM,SAAS,kCAAkC,UAAU,QAAQ;AAEnE,MAAI,OAAO,SAAS,GAAG;AACrB,UAAM,IAAI,gBAAgB,MAAM;AAAA,EAClC;AAEA,QAAM,eAAe,iBAAiB,QAAQ;AAC9C,QAAM,oBAAiD,CAAC;AAExD,aAAW,WAAW,SAAS,mBAAmB;AAChD,UAAM,SAAS,aAAa,IAAI,QAAQ,SAAS;AAEjD,sBAAkB,KAAK;AAAA,MACrB,GAAI,QAAQ,SAAS,SAAY,CAAC,IAAI,EAAE,MAAM,QAAQ,KAAK;AAAA,MAC3D,WAAW,QAAQ;AAAA,MACnB,MAAM,OAAO;AAAA,IACf,CAAC;AAAA,EACH;AAEA,SAAO,EAAE,kBAAkB;AAC7B;AAEO,SAAS,kCACd,UACA,UACmB;AACnB,QAAM,SAAmB,CAAC;AAC1B,QAAM,eAAe,iBAAiB,QAAQ;AAC9C,QAAM,UAAU,oBAAI,IAAY;AAEhC,aAAW,WAAW,SAAS,mBAAmB;AAChD,QAAI,CAAC,aAAa,IAAI,QAAQ,SAAS,GAAG;AACxC,aAAO;AAAA,QACL,sBAAsB,QAAQ,SAAS,kCAAkC;AAAA,UACvE,GAAG,aAAa,KAAK;AAAA,QACvB,EAAE,KAAK,IAAI,CAAC;AAAA,MACd;AACA;AAAA,IACF;AACA,QAAI,QAAQ,IAAI,QAAQ,SAAS,GAAG;AAClC,aAAO,KAAK,+BAA+B,QAAQ,SAAS,GAAG;AAC/D;AAAA,IACF;AAEA,YAAQ,IAAI,QAAQ,SAAS;AAAA,EAC/B;AAEA,SAAO;AACT;AAEA,SAAS,wBACP,OACa;AACb,SAAO;AAAA,IACL;AAAA,MACE,MAAM;AAAA,MACN,SAAS,6BAA6B,KAAK;AAAA,IAC7C;AAAA,IACA;AAAA,MACE,MAAM;AAAA,MACN,SAAS,uBAAuB,KAAK;AAAA,IACvC;AAAA,EACF;AACF;AAEA,SAAS,6BACP,OACQ;AACR,SAAO;AAAA,IACL;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA,MAAM;AAAA,IACN;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,EACF,EAAE,KAAK,IAAI;AACb;AAEA,SAAS,uBACP,OACQ;AACR,SAAO;AAAA,IACL;AAAA,IACA,KAAK;AAAA,MACH,MAAM,mBAAmB,IAAI,CAAC,aAAa;AAAA,QACzC,OAAO,QAAQ;AAAA,QACf,WAAW,QAAQ;AAAA,QACnB,MAAM,QAAQ;AAAA,MAChB,EAAE;AAAA,MACF;AAAA,MACA;AAAA,IACF;AAAA,IACA;AAAA,IACA;AAAA,IACA,KAAK;AAAA,MACH;AAAA,QACE,mBAAmB;AAAA,UACjB;AAAA,YACE,MAAM;AAAA,YACN,WAAW;AAAA,UACb;AAAA,QACF;AAAA,MACF;AAAA,MACA;AAAA,MACA;AAAA,IACF;AAAA,EACF,EAAE,KAAK,IAAI;AACb;AAEA,SAAS,iBACP,UACuC;AACvC,SAAO,IAAI,IAAI,SAAS,IAAI,CAAC,YAAY,CAAC,QAAQ,IAAI,OAAO,CAAC,CAAC;AACjE;AAEA,SAAS,oBAAoB,OAAwB;AACnD,MAAI,iBAAiB,iBAAiB;AACpC,WAAO,MAAM,OAAO,KAAK,IAAI;AAAA,EAC/B;AACA,MAAI,iBAAiB,OAAO;AAC1B,UAAM,SAAU,MAAwC;AAExD,QACE,MAAM,QAAQ,MAAM,KACpB,OAAO,MAAM,CAAC,UAAU,OAAO,UAAU,QAAQ,GACjD;AACA,aAAO,OAAO,KAAK,IAAI;AAAA,IACzB;AAEA,WAAO,MAAM;AAAA,EACf;AAEA,SAAO,OAAO,KAAK;AACrB;","names":[]}

package/dist/external/wikimatch/surface/window.d.ts

@@ -0,0 +1,8 @@
+import { BuildWikimatchSurfaceProtectionInputOptions, WikimatchSurfaceProtectionBuildResult } from '../types.js';
+import '../../wikipage/types.js';
+import '../../wikipage/wikimedia-client/wiki.js';
+import '../../wikipage/fetch-log.js';
+
+declare function buildWikimatchSurfaceProtectionInput(options: BuildWikimatchSurfaceProtectionInputOptions): WikimatchSurfaceProtectionBuildResult;
+
+export { buildWikimatchSurfaceProtectionInput };

package/dist/external/wikimatch/surface/window.js

@@ -0,0 +1,44 @@
+import { suppressContainedRanges } from "../range-suppression.js";
+function buildWikimatchSurfaceProtectionInput(options) {
+  const suppressedCandidates = suppressContainedRanges(options.candidates);
+  const surfaces = listSurfaces(suppressedCandidates, options.text);
+  const suspiciousSurfaces = selectTopSurfaces(surfaces, options.percentile);
+  const suspiciousTexts = new Set(
+    suspiciousSurfaces.map((surface) => surface.text)
+  );
+  return {
+    candidates: suppressedCandidates.filter(
+      (candidate) => !suspiciousTexts.has(
+        options.text.slice(candidate.range.start, candidate.range.end)
+      )
+    ),
+    suppressedCandidates,
+    suspiciousSurfaces
+  };
+}
+function listSurfaces(candidates, text) {
+  const surfaces = /* @__PURE__ */ new Map();
+  for (const candidate of candidates) {
+    const surface = text.slice(candidate.range.start, candidate.range.end);
+    surfaces.set(surface, (surfaces.get(surface) ?? 0) + 1);
+  }
+  return [...surfaces.entries()].map(([surface, count], index) => ({
+    count,
+    id: `s${index + 1}`,
+    text: surface
+  })).sort(compareSurface);
+}
+function selectTopSurfaces(surfaces, percentile) {
+  if (surfaces.length === 0 || percentile <= 0) {
+    return [];
+  }
+  const count = Math.ceil(surfaces.length * Math.min(1, percentile));
+  return surfaces.slice(0, count);
+}
+function compareSurface(left, right) {
+  return right.count - left.count || left.text.localeCompare(right.text) || left.id.localeCompare(right.id);
+}
+export {
+  buildWikimatchSurfaceProtectionInput
+};
+//# sourceMappingURL=window.js.map

package/dist/external/wikimatch/surface/window.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../../../../src/external/wikimatch/surface/window.ts"],"sourcesContent":["import type {\n  BuildWikimatchSurfaceProtectionInputOptions,\n  WikimatchCandidate,\n  WikimatchSurface,\n  WikimatchSurfaceProtectionBuildResult,\n} from \"../types.js\";\nimport { suppressContainedRanges } from \"../range-suppression.js\";\n\nexport function buildWikimatchSurfaceProtectionInput(\n  options: BuildWikimatchSurfaceProtectionInputOptions,\n): WikimatchSurfaceProtectionBuildResult {\n  const suppressedCandidates = suppressContainedRanges(options.candidates);\n  const surfaces = listSurfaces(suppressedCandidates, options.text);\n  const suspiciousSurfaces = selectTopSurfaces(surfaces, options.percentile);\n  const suspiciousTexts = new Set(\n    suspiciousSurfaces.map((surface) => surface.text),\n  );\n\n  return {\n    candidates: suppressedCandidates.filter(\n      (candidate) =>\n        !suspiciousTexts.has(\n          options.text.slice(candidate.range.start, candidate.range.end),\n        ),\n    ),\n    suppressedCandidates,\n    suspiciousSurfaces,\n  };\n}\n\nfunction listSurfaces(\n  candidates: readonly WikimatchCandidate[],\n  text: string,\n): readonly WikimatchSurface[] {\n  const surfaces = new Map<string, number>();\n\n  for (const candidate of candidates) {\n    const surface = text.slice(candidate.range.start, candidate.range.end);\n\n    surfaces.set(surface, (surfaces.get(surface) ?? 0) + 1);\n  }\n\n  return [...surfaces.entries()]\n    .map(([surface, count], index) => ({\n      count,\n      id: `s${index + 1}`,\n      text: surface,\n    }))\n    .sort(compareSurface);\n}\n\nfunction selectTopSurfaces(\n  surfaces: readonly WikimatchSurface[],\n  percentile: number,\n): readonly WikimatchSurface[] {\n  if (surfaces.length === 0 || percentile <= 0) {\n    return [];\n  }\n\n  const count = Math.ceil(surfaces.length * Math.min(1, percentile));\n\n  return surfaces.slice(0, count);\n}\n\nfunction compareSurface(\n  left: WikimatchSurface,\n  right: WikimatchSurface,\n): number {\n  return (\n    right.count - left.count ||\n    left.text.localeCompare(right.text) ||\n    left.id.localeCompare(right.id)\n  );\n}\n"],"mappings":"AAMA,SAAS,+BAA+B;AAEjC,SAAS,qCACd,SACuC;AACvC,QAAM,uBAAuB,wBAAwB,QAAQ,UAAU;AACvE,QAAM,WAAW,aAAa,sBAAsB,QAAQ,IAAI;AAChE,QAAM,qBAAqB,kBAAkB,UAAU,QAAQ,UAAU;AACzE,QAAM,kBAAkB,IAAI;AAAA,IAC1B,mBAAmB,IAAI,CAAC,YAAY,QAAQ,IAAI;AAAA,EAClD;AAEA,SAAO;AAAA,IACL,YAAY,qBAAqB;AAAA,MAC/B,CAAC,cACC,CAAC,gBAAgB;AAAA,QACf,QAAQ,KAAK,MAAM,UAAU,MAAM,OAAO,UAAU,MAAM,GAAG;AAAA,MAC/D;AAAA,IACJ;AAAA,IACA;AAAA,IACA;AAAA,EACF;AACF;AAEA,SAAS,aACP,YACA,MAC6B;AAC7B,QAAM,WAAW,oBAAI,IAAoB;AAEzC,aAAW,aAAa,YAAY;AAClC,UAAM,UAAU,KAAK,MAAM,UAAU,MAAM,OAAO,UAAU,MAAM,GAAG;AAErE,aAAS,IAAI,UAAU,SAAS,IAAI,OAAO,KAAK,KAAK,CAAC;AAAA,EACxD;AAEA,SAAO,CAAC,GAAG,SAAS,QAAQ,CAAC,EAC1B,IAAI,CAAC,CAAC,SAAS,KAAK,GAAG,WAAW;AAAA,IACjC;AAAA,IACA,IAAI,IAAI,QAAQ,CAAC;AAAA,IACjB,MAAM;AAAA,EACR,EAAE,EACD,KAAK,cAAc;AACxB;AAEA,SAAS,kBACP,UACA,YAC6B;AAC7B,MAAI,SAAS,WAAW,KAAK,cAAc,GAAG;AAC5C,WAAO,CAAC;AAAA,EACV;AAEA,QAAM,QAAQ,KAAK,KAAK,SAAS,SAAS,KAAK,IAAI,GAAG,UAAU,CAAC;AAEjE,SAAO,SAAS,MAAM,GAAG,KAAK;AAChC;AAEA,SAAS,eACP,MACA,OACQ;AACR,SACE,MAAM,QAAQ,KAAK,SACnB,KAAK,KAAK,cAAc,MAAM,IAAI,KAClC,KAAK,GAAG,cAAc,MAAM,EAAE;AAElC;","names":[]}

package/dist/external/wikimatch/types.d.ts

@@ -0,0 +1,153 @@
+import { DisambiguationExpansion } from '../wikipage/types.js';
+import '../wikipage/wikimedia-client/wiki.js';
+import '../wikipage/fetch-log.js';
+
+interface WikimatchCandidate {
+    readonly hasMoreOptions?: boolean;
+    readonly id: string;
+    readonly qidOptions: readonly WikimatchQidOption[];
+    readonly range: WikimatchTextRange;
+    readonly surface: string;
+}
+interface WikimatchSentence {
+    readonly id?: string;
+    readonly range: WikimatchTextRange;
+    readonly text: string;
+}
+interface WikimatchQidOption {
+    readonly description?: string;
+    readonly disambiguation?: DisambiguationExpansion;
+    readonly isDisambiguation?: boolean;
+    readonly label?: string;
+    readonly qid: string;
+}
+interface WikimatchTextRange {
+    readonly end: number;
+    readonly start: number;
+}
+interface WikimatchWindow {
+    readonly baseOffset: number;
+    readonly candidates: readonly WikimatchCandidate[];
+    readonly groups: readonly WikimatchConflictGroup[];
+    readonly text: string;
+}
+interface WikimatchSurface {
+    readonly id: string;
+    readonly count: number;
+    readonly text: string;
+}
+interface WikimatchConflictGroup {
+    readonly candidateIds: readonly string[];
+    readonly id: string;
+    readonly range: WikimatchTextRange;
+}
+interface BuildWikimatchWindowsOptions {
+    readonly candidates: readonly WikimatchCandidate[];
+    readonly contextWords: number;
+    readonly optionBudget: number;
+    readonly text: string;
+}
+interface NarrowWikimatchCandidateOptionsInput {
+    readonly candidate: WikimatchCandidate;
+    readonly policyPrompt: string;
+    readonly text: string;
+}
+interface WikimatchCandidateOptionNarrowingResult {
+    readonly candidate: WikimatchCandidate;
+    readonly fallback?: WikimatchCandidateOptionNarrowingFallback;
+}
+interface WikimatchCandidateOptionNarrowingFallback {
+    readonly issues: readonly string[];
+    readonly reason: "guaranteed_json_failed";
+}
+interface WikimatchCandidateOptionNarrowingResponse {
+    readonly qids: readonly WikimatchCandidateOptionNarrowingItemOutput[];
+}
+interface WikimatchCandidateOptionNarrowingItemOutput {
+    readonly decision: "keep" | "reject";
+    readonly qid: string;
+}
+interface BuildWikimatchSurfaceProtectionInputOptions {
+    readonly candidates: readonly WikimatchCandidate[];
+    readonly percentile: number;
+    readonly text: string;
+}
+interface WikimatchSurfaceProtectionInput {
+    readonly policyPrompt: string;
+    readonly suspiciousSurfaces: readonly WikimatchSurface[];
+}
+interface WikimatchSurfaceProtectionBuildResult {
+    readonly candidates: readonly WikimatchCandidate[];
+    readonly suppressedCandidates: readonly WikimatchCandidate[];
+    readonly suspiciousSurfaces: readonly WikimatchSurface[];
+}
+interface WikimatchSurfaceProtectionResult {
+    readonly fallback?: WikimatchSurfaceProtectionFallback;
+    readonly protectedSurfaces: readonly WikimatchProtectedSurface[];
+}
+interface WikimatchProtectedSurface {
+    readonly note?: string;
+    readonly surfaceId: string;
+    readonly text: string;
+}
+interface WikimatchSurfaceProtectionFallback {
+    readonly issues: readonly string[];
+    readonly reason: "guaranteed_json_failed";
+}
+interface WikimatchProtectedSurfaceOutput {
+    readonly note?: string;
+    readonly surfaceId: string;
+}
+interface WikimatchSurfaceProtectionResponse {
+    readonly protectedSurfaces: readonly WikimatchProtectedSurfaceOutput[];
+}
+type WikimatchPolicyDecision = "continue" | "never_recall" | "recall" | "skip_this_time";
+interface WikimatchPolicyJudgeInput {
+    readonly candidates: readonly WikimatchCandidate[];
+    readonly policyPrompt: string;
+    readonly window: WikimatchWindow;
+}
+interface WikimatchPolicyJudgeResult {
+    readonly continuations: readonly WikimatchPolicyContinuation[];
+    readonly fallback?: WikimatchPolicyFallback;
+    readonly mentions: readonly WikimatchAcceptedMention[];
+    readonly policyUpdates: readonly WikimatchPolicyUpdate[];
+}
+interface WikimatchAcceptedMention {
+    readonly candidateId: string;
+    readonly confidence?: number;
+    readonly qid: string;
+    readonly range: WikimatchTextRange;
+    readonly note?: string;
+    readonly surface: string;
+}
+interface WikimatchPolicyUpdate {
+    readonly candidateId: string;
+    readonly decision: Exclude<WikimatchPolicyDecision, "recall" | "continue">;
+    readonly note?: string;
+    readonly qid?: string;
+    readonly surface: string;
+}
+interface WikimatchPolicyContinuation {
+    readonly candidateIds: readonly string[];
+    readonly groupId: string;
+}
+interface WikimatchPolicyFallback {
+    readonly issues: readonly string[];
+    readonly reason: "guaranteed_json_failed";
+}
+interface WikimatchPolicyDecisionOutput {
+    readonly candidateId: string;
+    readonly confidence?: number;
+    readonly decision: WikimatchPolicyDecision;
+    readonly qid?: string;
+}
+interface WikimatchPolicyGroupOutput {
+    readonly decisions: readonly WikimatchPolicyDecisionOutput[];
+    readonly groupId: string;
+}
+interface WikimatchPolicyResponse {
+    readonly groups: readonly WikimatchPolicyGroupOutput[];
+}
+
+export type { BuildWikimatchSurfaceProtectionInputOptions, BuildWikimatchWindowsOptions, NarrowWikimatchCandidateOptionsInput, WikimatchAcceptedMention, WikimatchCandidate, WikimatchCandidateOptionNarrowingFallback, WikimatchCandidateOptionNarrowingItemOutput, WikimatchCandidateOptionNarrowingResponse, WikimatchCandidateOptionNarrowingResult, WikimatchConflictGroup, WikimatchPolicyContinuation, WikimatchPolicyDecision, WikimatchPolicyDecisionOutput, WikimatchPolicyFallback, WikimatchPolicyGroupOutput, WikimatchPolicyJudgeInput, WikimatchPolicyJudgeResult, WikimatchPolicyResponse, WikimatchPolicyUpdate, WikimatchProtectedSurface, WikimatchProtectedSurfaceOutput, WikimatchQidOption, WikimatchSentence, WikimatchSurface, WikimatchSurfaceProtectionBuildResult, WikimatchSurfaceProtectionFallback, WikimatchSurfaceProtectionInput, WikimatchSurfaceProtectionResponse, WikimatchSurfaceProtectionResult, WikimatchTextRange, WikimatchWindow };

package/dist/external/wikimatch/wikispine.d.ts

@@ -0,0 +1,45 @@
+import { WikimatchSentence, WikimatchCandidate } from './types.js';
+import '../wikipage/types.js';
+import '../wikipage/wikimedia-client/wiki.js';
+import '../wikipage/fetch-log.js';
+
+interface MatchWikispineSentenceCandidatesOptions {
+    readonly command?: string;
+    readonly dataDir?: string;
+    readonly endpoint?: string;
+    readonly fetch?: typeof fetch;
+    readonly includeDisambiguation?: boolean;
+    readonly maxCandidatesPerSurface?: number;
+    readonly onProgress?: (progress: WikispineMatchProgress) => Promise<void> | void;
+    readonly provider?: WikispineProvider;
+    readonly sentences: readonly WikimatchSentence[];
+}
+type WikispineProvider = "cli" | "fetch";
+interface WikispineMatchProgress {
+    readonly coveredRangeEnd: number;
+}
+interface TestWikispineRuntimeOptions {
+    readonly command?: string;
+    readonly dataDir?: string;
+    readonly endpoint?: string;
+    readonly fetch?: typeof fetch;
+    readonly provider?: WikispineProvider;
+}
+interface WikispineRuntimeTestResult {
+    readonly durationMs: number;
+    readonly metadata?: WikispineMetadata;
+    readonly ok: true;
+    readonly provider: WikispineProvider;
+}
+interface WikispineMetadata {
+    readonly automaton_shard_count: number;
+    readonly format: string;
+    readonly qid_count: number;
+    readonly surface_count: number;
+    readonly surface_normalization: string;
+}
+declare const DEFAULT_WIKISPINE_FETCH_ENDPOINT = "https://wikispi-service-cxbfjlteab.cn-hangzhou.fcapp.run";
+declare function matchWikispineSentenceCandidates(options: MatchWikispineSentenceCandidatesOptions): Promise<readonly WikimatchCandidate[]>;
+declare function testWikispineRuntime(options: TestWikispineRuntimeOptions): Promise<WikispineRuntimeTestResult>;
+
+export { DEFAULT_WIKISPINE_FETCH_ENDPOINT, type MatchWikispineSentenceCandidatesOptions, type TestWikispineRuntimeOptions, type WikispineMatchProgress, type WikispineProvider, type WikispineRuntimeTestResult, matchWikispineSentenceCandidates, testWikispineRuntime };

package/dist/external/wikimatch/wikispine.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../../../src/external/wikimatch/wikispine.ts"],"sourcesContent":["import { spawn } from \"child_process\";\n\nimport type {\n  WikimatchCandidate,\n  WikimatchQidOption,\n  WikimatchSentence,\n} from \"./types.js\";\n\nexport interface MatchWikispineSentenceCandidatesOptions {\n  readonly command?: string;\n  readonly dataDir?: string;\n  readonly endpoint?: string;\n  readonly fetch?: typeof fetch;\n  readonly includeDisambiguation?: boolean;\n  readonly maxCandidatesPerSurface?: number;\n  readonly onProgress?: (\n    progress: WikispineMatchProgress,\n  ) => Promise<void> | void;\n  readonly provider?: WikispineProvider;\n  readonly sentences: readonly WikimatchSentence[];\n}\n\nexport type WikispineProvider = \"cli\" | \"fetch\";\n\nexport interface WikispineMatchProgress {\n  readonly coveredRangeEnd: number;\n}\n\nexport interface TestWikispineRuntimeOptions {\n  readonly command?: string;\n  readonly dataDir?: string;\n  readonly endpoint?: string;\n  readonly fetch?: typeof fetch;\n  readonly provider?: WikispineProvider;\n}\n\nexport interface WikispineRuntimeTestResult {\n  readonly durationMs: number;\n  readonly metadata?: WikispineMetadata;\n  readonly ok: true;\n  readonly provider: WikispineProvider;\n}\n\ninterface WikispineMatchEvent {\n  readonly match: WikispineMatchRecord;\n  readonly type: \"match\";\n}\n\ninterface WikispineDoneEvent {\n  readonly stats?: unknown;\n  readonly type: \"done\";\n}\n\ntype WikispineEvent = WikispineDoneEvent | WikispineMatchEvent;\n\ninterface WikispineMatchRecord {\n  readonly end: number;\n  readonly qids: readonly WikispineQidCandidate[];\n  readonly start: number;\n  readonly surface_id: number;\n}\n\ninterface WikispineQidCandidate {\n  readonly disambiguation?: boolean;\n  readonly qid: string;\n}\n\ninterface WikispineMetadata {\n  readonly automaton_shard_count: number;\n  readonly format: string;\n  readonly qid_count: number;\n  readonly surface_count: number;\n  readonly surface_normalization: string;\n}\n\nconst WIKISPINE_RUNTIME_GUIDE_URL =\n  \"https://raw.githubusercontent.com/oomol-lab/wiki-graph/refs/heads/main/docs/wikispine-runtime.md\";\nexport const DEFAULT_WIKISPINE_FETCH_ENDPOINT =\n  \"https://wikispi-service-cxbfjlteab.cn-hangzhou.fcapp.run\";\n\nexport async function matchWikispineSentenceCandidates(\n  options: MatchWikispineSentenceCandidatesOptions,\n): Promise<readonly WikimatchCandidate[]> {\n  const candidates: WikimatchCandidate[] = [];\n  let candidateIndex = 1;\n\n  for (const sentence of options.sentences) {\n    for (const matched of await matchSentence(sentence, options)) {\n      const surface = sentence.text.slice(matched.start, matched.end);\n\n      candidates.push({\n        id: `c${candidateIndex}`,\n        qidOptions: matched.qids.map(toQidOption),\n        range: {\n          end: sentence.range.start + matched.end,\n          start: sentence.range.start + matched.start,\n        },\n        surface,\n      });\n      candidateIndex += 1;\n    }\n    await options.onProgress?.({ coveredRangeEnd: sentence.range.end });\n  }\n\n  return candidates;\n}\n\nasync function matchSentence(\n  sentence: WikimatchSentence,\n  options: MatchWikispineSentenceCandidatesOptions,\n): Promise<readonly WikispineMatchRecord[]> {\n  return resolveProvider(options) === \"fetch\"\n    ? await fetchWikispineMatch(options, sentence)\n    : await runWikispineMatch(\n        options.command ?? \"wikispine\",\n        buildMatchArgs(options),\n        sentence,\n        options,\n      );\n}\n\nexport async function testWikispineRuntime(\n  options: TestWikispineRuntimeOptions,\n): Promise<WikispineRuntimeTestResult> {\n  const provider = resolveProvider(options);\n  const startedAt = Date.now();\n\n  if (provider === \"fetch\") {\n    const endpoint = requireEndpoint(options.endpoint);\n    const metadata = await fetchWikispineMetadata(endpoint, options.fetch);\n\n    await fetchWikispineMatch(\n      {\n        ...options,\n        endpoint,\n        maxCandidatesPerSurface: 1,\n      },\n      {\n        id: \"test\",\n        range: { end: 7, start: 0 },\n        text: \"北京大学位于北京。\",\n      },\n    );\n\n    return {\n      durationMs: Date.now() - startedAt,\n      metadata,\n      ok: true,\n      provider,\n    };\n  }\n\n  await runWikispineMatch(\n    options.command ?? \"wikispine\",\n    buildMatchArgs({\n      ...options,\n      maxCandidatesPerSurface: 1,\n    }),\n    {\n      id: \"test\",\n      range: { end: 7, start: 0 },\n      text: \"北京大学位于北京。\",\n    },\n    {},\n  );\n\n  return {\n    durationMs: Date.now() - startedAt,\n    ok: true,\n    provider,\n  };\n}\n\nasync function runWikispineMatch(\n  command: string,\n  args: readonly string[],\n  sentence: WikimatchSentence,\n  options: Pick<MatchWikispineSentenceCandidatesOptions, \"onProgress\">,\n): Promise<readonly WikispineMatchRecord[]> {\n  return await new Promise((resolve, reject) => {\n    const child = spawn(command, args, {\n      stdio: [\"pipe\", \"pipe\", \"pipe\"],\n    });\n    const progress = createWikispineProgressReporter(options.onProgress, {\n      onFailure: (error) => {\n        reject(error);\n        child.kill();\n      },\n    });\n    const parser = createWikispineNdjsonParser({\n      onMatch: (match) => {\n        progress.report({\n          coveredRangeEnd: sentence.range.start + match.end,\n        });\n      },\n    });\n    const stderr: Buffer[] = [];\n\n    child.stdout.setEncoding(\"utf8\");\n    child.stdout.on(\"data\", (chunk: string) => {\n      try {\n        parser.push(chunk);\n      } catch (error) {\n        reject(toError(error));\n        child.kill();\n      }\n    });\n    child.stderr.on(\"data\", (chunk: Buffer) => {\n      stderr.push(chunk);\n    });\n    child.on(\"error\", (error) => {\n      reject(\n        new Error(\n          formatWikispineRuntimeError(\n            `Failed to start wikispine command: ${error.message}`,\n          ),\n        ),\n      );\n    });\n    child.on(\"close\", (code) => {\n      if (code !== 0) {\n        reject(\n          new Error(\n            formatWikispineRuntimeError(\n              `wikispine match failed with exit code ${code}: ${Buffer.concat(stderr).toString(\"utf8\")}`,\n            ),\n          ),\n        );\n        return;\n      }\n\n      void (async () => {\n        try {\n          const matches = parser.finish();\n          await progress.wait();\n          resolve(matches);\n        } catch (error) {\n          reject(toError(error));\n        }\n      })();\n    });\n\n    child.stdin.end(sentence.text);\n  });\n}\n\nfunction buildMatchArgs(\n  options: Pick<\n    MatchWikispineSentenceCandidatesOptions,\n    \"dataDir\" | \"includeDisambiguation\" | \"maxCandidatesPerSurface\"\n  >,\n): string[] {\n  const args = [\"match\"];\n\n  if (options.dataDir !== undefined) {\n    args.push(\"--data-dir\", options.dataDir);\n  }\n  if (options.includeDisambiguation === false) {\n    args.push(\"--exclude-disambiguation\");\n  }\n  if (options.maxCandidatesPerSurface !== undefined) {\n    args.push(\n      \"--max-candidates-per-surface\",\n      String(options.maxCandidatesPerSurface),\n    );\n  }\n\n  return args;\n}\n\nasync function fetchWikispineMatch(\n  options: Pick<\n    MatchWikispineSentenceCandidatesOptions,\n    | \"endpoint\"\n    | \"fetch\"\n    | \"includeDisambiguation\"\n    | \"maxCandidatesPerSurface\"\n    | \"onProgress\"\n  >,\n  sentence: WikimatchSentence,\n): Promise<readonly WikispineMatchRecord[]> {\n  const endpoint = requireEndpoint(options.endpoint);\n  const response = await (options.fetch ?? fetch)(`${endpoint}/match`, {\n    body: JSON.stringify({\n      options: {\n        ...(options.includeDisambiguation === undefined\n          ? {}\n          : { include_disambiguation: options.includeDisambiguation }),\n        ...(options.maxCandidatesPerSurface === undefined\n          ? {}\n          : {\n              max_candidates_per_surface: options.maxCandidatesPerSurface,\n            }),\n      },\n      text: sentence.text,\n    }),\n    headers: {\n      accept: \"application/x-ndjson\",\n      \"content-type\": \"application/json\",\n    },\n    method: \"POST\",\n  });\n\n  if (!response.ok) {\n    throw new Error(\n      formatWikispineRuntimeError(\n        `WikiSpine fetch provider failed with HTTP ${response.status}: ${await response.text()}`,\n      ),\n    );\n  }\n\n  const progress = createWikispineProgressReporter(options.onProgress);\n  const parser = createWikispineNdjsonParser({\n    onMatch: (match) => {\n      progress.report({\n        coveredRangeEnd: sentence.range.start + match.end,\n      });\n    },\n  });\n\n  if (response.body === null) {\n    parser.push(await response.text());\n    const matches = parser.finish();\n    await progress.wait();\n    return matches;\n  }\n\n  const reader = response.body.getReader();\n  const decoder = new TextDecoder();\n\n  while (true) {\n    const { done, value } = await reader.read();\n    if (done) {\n      break;\n    }\n    parser.push(decoder.decode(value, { stream: true }));\n    progress.throwIfFailed();\n  }\n\n  parser.push(decoder.decode());\n  const matches = parser.finish();\n  await progress.wait();\n  return matches;\n}\n\nfunction createWikispineProgressReporter(\n  onProgress:\n    | ((progress: WikispineMatchProgress) => Promise<void> | void)\n    | undefined,\n  options?: {\n    readonly onFailure?: (error: Error) => void;\n  },\n): {\n  readonly report: (progress: WikispineMatchProgress) => void;\n  readonly throwIfFailed: () => void;\n  readonly wait: () => Promise<void>;\n} {\n  const tasks: Promise<void>[] = [];\n  let failure: Error | undefined;\n\n  function fail(error: unknown): void {\n    if (failure !== undefined) {\n      return;\n    }\n    failure = toError(error);\n    options?.onFailure?.(failure);\n  }\n\n  return {\n    report: (progress) => {\n      if (onProgress === undefined || failure !== undefined) {\n        return;\n      }\n\n      try {\n        tasks.push(Promise.resolve(onProgress(progress)).catch(fail));\n      } catch (error) {\n        fail(error);\n      }\n    },\n    throwIfFailed: () => {\n      if (failure !== undefined) {\n        throw failure;\n      }\n    },\n    wait: async () => {\n      await Promise.all(tasks);\n      if (failure !== undefined) {\n        throw failure;\n      }\n    },\n  };\n}\n\nasync function fetchWikispineMetadata(\n  endpoint: string,\n  fetchFn: typeof fetch = fetch,\n): Promise<WikispineMetadata> {\n  const ready = await fetchFn(`${endpoint}/readyz`);\n\n  if (!ready.ok) {\n    throw new Error(\n      formatWikispineRuntimeError(\n        `WikiSpine fetch provider is not ready: HTTP ${ready.status}.`,\n      ),\n    );\n  }\n\n  const response = await fetchFn(`${endpoint}/metadata`);\n\n  if (!response.ok) {\n    throw new Error(\n      formatWikispineRuntimeError(\n        `WikiSpine metadata request failed with HTTP ${response.status}.`,\n      ),\n    );\n  }\n\n  try {\n    return parseWikispineMetadata(await response.json());\n  } catch (error) {\n    if (error instanceof Error && error.message.includes(\"setup guide\")) {\n      throw error;\n    }\n\n    throw new Error(\n      formatWikispineRuntimeError(\n        `Invalid WikiSpine metadata response: ${error instanceof Error ? error.message : String(error)}`,\n      ),\n    );\n  }\n}\n\nfunction createWikispineNdjsonParser(input?: {\n  readonly onMatch?: (match: WikispineMatchRecord) => void;\n}): {\n  readonly finish: () => readonly WikispineMatchRecord[];\n  readonly push: (chunk: string) => void;\n} {\n  const matches: WikispineMatchRecord[] = [];\n  let buffer = \"\";\n\n  function parseLine(line: string): void {\n    if (line.trim() === \"\") {\n      return;\n    }\n\n    const event = parseWikispineEvent(JSON.parse(line));\n\n    if (event.type === \"match\") {\n      matches.push(event.match);\n      input?.onMatch?.(event.match);\n    }\n  }\n\n  function wrapParseError(error: unknown): Error {\n    return new Error(\n      formatWikispineRuntimeError(\n        `Invalid WikiSpine match response: ${error instanceof Error ? error.message : String(error)}`,\n      ),\n    );\n  }\n\n  return {\n    finish: () => {\n      try {\n        parseLine(buffer);\n        buffer = \"\";\n        return matches;\n      } catch (error) {\n        throw wrapParseError(error);\n      }\n    },\n    push: (chunk) => {\n      try {\n        buffer += chunk;\n        const lines = buffer.split(/\\r?\\n/u);\n        buffer = lines.pop() ?? \"\";\n\n        for (const line of lines) {\n          parseLine(line);\n        }\n      } catch (error) {\n        throw wrapParseError(error);\n      }\n    },\n  };\n}\n\nfunction parseWikispineEvent(value: unknown): WikispineEvent {\n  if (typeof value !== \"object\" || value === null) {\n    throw new Error(\"Expected wikispine event to be an object.\");\n  }\n\n  const record = value as Record<string, unknown>;\n\n  if (record.type === \"done\") {\n    return { type: \"done\" };\n  }\n  if (record.type === \"match\" && isWikispineMatchRecord(record.match)) {\n    return {\n      match: record.match,\n      type: \"match\",\n    };\n  }\n\n  throw new Error(`Unexpected wikispine event: ${JSON.stringify(value)}`);\n}\n\nfunction toError(error: unknown): Error {\n  return error instanceof Error ? error : new Error(String(error));\n}\n\nfunction parseWikispineMetadata(value: unknown): WikispineMetadata {\n  if (typeof value !== \"object\" || value === null) {\n    throw new Error(\n      formatWikispineRuntimeError(\"Expected WikiSpine metadata.\"),\n    );\n  }\n\n  const record = value as Record<string, unknown>;\n\n  if (\n    typeof record.automaton_shard_count !== \"number\" ||\n    typeof record.format !== \"string\" ||\n    typeof record.qid_count !== \"number\" ||\n    typeof record.surface_count !== \"number\" ||\n    typeof record.surface_normalization !== \"string\"\n  ) {\n    throw new Error(formatWikispineRuntimeError(\"Invalid WikiSpine metadata.\"));\n  }\n\n  return {\n    automaton_shard_count: record.automaton_shard_count,\n    format: record.format,\n    qid_count: record.qid_count,\n    surface_count: record.surface_count,\n    surface_normalization: record.surface_normalization,\n  };\n}\n\nfunction isWikispineMatchRecord(value: unknown): value is WikispineMatchRecord {\n  if (typeof value !== \"object\" || value === null) {\n    return false;\n  }\n\n  const record = value as Record<string, unknown>;\n\n  return (\n    typeof record.end === \"number\" &&\n    Array.isArray(record.qids) &&\n    record.qids.every(isWikispineQidCandidate) &&\n    typeof record.start === \"number\" &&\n    typeof record.surface_id === \"number\"\n  );\n}\n\nfunction isWikispineQidCandidate(\n  value: unknown,\n): value is WikispineQidCandidate {\n  if (typeof value !== \"object\" || value === null) {\n    return false;\n  }\n\n  const record = value as Record<string, unknown>;\n\n  return (\n    (!(\"disambiguation\" in record) ||\n      typeof record.disambiguation === \"boolean\") &&\n    typeof record.qid === \"string\" &&\n    /^Q[1-9]\\d*$/u.test(record.qid)\n  );\n}\n\nfunction toQidOption(candidate: WikispineQidCandidate): WikimatchQidOption {\n  return {\n    isDisambiguation: candidate.disambiguation === true,\n    qid: candidate.qid,\n  };\n}\n\nfunction resolveProvider(options: {\n  readonly provider?: WikispineProvider;\n}): WikispineProvider {\n  return options.provider ?? \"cli\";\n}\n\nfunction requireEndpoint(endpoint: string | undefined): string {\n  const normalized = (endpoint ?? DEFAULT_WIKISPINE_FETCH_ENDPOINT)\n    .trim()\n    .replace(/\\/+$/u, \"\");\n\n  if (normalized === \"\") {\n    throw new Error(\n      formatWikispineRuntimeError(\n        \"WikiSpine fetch provider has no default endpoint.\",\n      ),\n    );\n  }\n\n  return normalized;\n}\n\nfunction formatWikispineRuntimeError(message: string): string {\n  return `${message}\\nWikiSpine setup guide: ${WIKISPINE_RUNTIME_GUIDE_URL}`;\n}\n"],"mappings":"AAAA,SAAS,aAAa;AA2EtB,MAAM,8BACJ;AACK,MAAM,mCACX;AAEF,eAAsB,iCACpB,SACwC;AACxC,QAAM,aAAmC,CAAC;AAC1C,MAAI,iBAAiB;AAErB,aAAW,YAAY,QAAQ,WAAW;AACxC,eAAW,WAAW,MAAM,cAAc,UAAU,OAAO,GAAG;AAC5D,YAAM,UAAU,SAAS,KAAK,MAAM,QAAQ,OAAO,QAAQ,GAAG;AAE9D,iBAAW,KAAK;AAAA,QACd,IAAI,IAAI,cAAc;AAAA,QACtB,YAAY,QAAQ,KAAK,IAAI,WAAW;AAAA,QACxC,OAAO;AAAA,UACL,KAAK,SAAS,MAAM,QAAQ,QAAQ;AAAA,UACpC,OAAO,SAAS,MAAM,QAAQ,QAAQ;AAAA,QACxC;AAAA,QACA;AAAA,MACF,CAAC;AACD,wBAAkB;AAAA,IACpB;AACA,UAAM,QAAQ,aAAa,EAAE,iBAAiB,SAAS,MAAM,IAAI,CAAC;AAAA,EACpE;AAEA,SAAO;AACT;AAEA,eAAe,cACb,UACA,SAC0C;AAC1C,SAAO,gBAAgB,OAAO,MAAM,UAChC,MAAM,oBAAoB,SAAS,QAAQ,IAC3C,MAAM;AAAA,IACJ,QAAQ,WAAW;AAAA,IACnB,eAAe,OAAO;AAAA,IACtB;AAAA,IACA;AAAA,EACF;AACN;AAEA,eAAsB,qBACpB,SACqC;AACrC,QAAM,WAAW,gBAAgB,OAAO;AACxC,QAAM,YAAY,KAAK,IAAI;AAE3B,MAAI,aAAa,SAAS;AACxB,UAAM,WAAW,gBAAgB,QAAQ,QAAQ;AACjD,UAAM,WAAW,MAAM,uBAAuB,UAAU,QAAQ,KAAK;AAErE,UAAM;AAAA,MACJ;AAAA,QACE,GAAG;AAAA,QACH;AAAA,QACA,yBAAyB;AAAA,MAC3B;AAAA,MACA;AAAA,QACE,IAAI;AAAA,QACJ,OAAO,EAAE,KAAK,GAAG,OAAO,EAAE;AAAA,QAC1B,MAAM;AAAA,MACR;AAAA,IACF;AAEA,WAAO;AAAA,MACL,YAAY,KAAK,IAAI,IAAI;AAAA,MACzB;AAAA,MACA,IAAI;AAAA,MACJ;AAAA,IACF;AAAA,EACF;AAEA,QAAM;AAAA,IACJ,QAAQ,WAAW;AAAA,IACnB,eAAe;AAAA,MACb,GAAG;AAAA,MACH,yBAAyB;AAAA,IAC3B,CAAC;AAAA,IACD;AAAA,MACE,IAAI;AAAA,MACJ,OAAO,EAAE,KAAK,GAAG,OAAO,EAAE;AAAA,MAC1B,MAAM;AAAA,IACR;AAAA,IACA,CAAC;AAAA,EACH;AAEA,SAAO;AAAA,IACL,YAAY,KAAK,IAAI,IAAI;AAAA,IACzB,IAAI;AAAA,IACJ;AAAA,EACF;AACF;AAEA,eAAe,kBACb,SACA,MACA,UACA,SAC0C;AAC1C,SAAO,MAAM,IAAI,QAAQ,CAAC,SAAS,WAAW;AAC5C,UAAM,QAAQ,MAAM,SAAS,MAAM;AAAA,MACjC,OAAO,CAAC,QAAQ,QAAQ,MAAM;AAAA,IAChC,CAAC;AACD,UAAM,WAAW,gCAAgC,QAAQ,YAAY;AAAA,MACnE,WAAW,CAAC,UAAU;AACpB,eAAO,KAAK;AACZ,cAAM,KAAK;AAAA,MACb;AAAA,IACF,CAAC;AACD,UAAM,SAAS,4BAA4B;AAAA,MACzC,SAAS,CAAC,UAAU;AAClB,iBAAS,OAAO;AAAA,UACd,iBAAiB,SAAS,MAAM,QAAQ,MAAM;AAAA,QAChD,CAAC;AAAA,MACH;AAAA,IACF,CAAC;AACD,UAAM,SAAmB,CAAC;AAE1B,UAAM,OAAO,YAAY,MAAM;AAC/B,UAAM,OAAO,GAAG,QAAQ,CAAC,UAAkB;AACzC,UAAI;AACF,eAAO,KAAK,KAAK;AAAA,MACnB,SAAS,OAAO;AACd,eAAO,QAAQ,KAAK,CAAC;AACrB,cAAM,KAAK;AAAA,MACb;AAAA,IACF,CAAC;AACD,UAAM,OAAO,GAAG,QAAQ,CAAC,UAAkB;AACzC,aAAO,KAAK,KAAK;AAAA,IACnB,CAAC;AACD,UAAM,GAAG,SAAS,CAAC,UAAU;AAC3B;AAAA,QACE,IAAI;AAAA,UACF;AAAA,YACE,sCAAsC,MAAM,OAAO;AAAA,UACrD;AAAA,QACF;AAAA,MACF;AAAA,IACF,CAAC;AACD,UAAM,GAAG,SAAS,CAAC,SAAS;AAC1B,UAAI,SAAS,GAAG;AACd;AAAA,UACE,IAAI;AAAA,YACF;AAAA,cACE,yCAAyC,IAAI,KAAK,OAAO,OAAO,MAAM,EAAE,SAAS,MAAM,CAAC;AAAA,YAC1F;AAAA,UACF;AAAA,QACF;AACA;AAAA,MACF;AAEA,YAAM,YAAY;AAChB,YAAI;AACF,gBAAM,UAAU,OAAO,OAAO;AAC9B,gBAAM,SAAS,KAAK;AACpB,kBAAQ,OAAO;AAAA,QACjB,SAAS,OAAO;AACd,iBAAO,QAAQ,KAAK,CAAC;AAAA,QACvB;AAAA,MACF,GAAG;AAAA,IACL,CAAC;AAED,UAAM,MAAM,IAAI,SAAS,IAAI;AAAA,EAC/B,CAAC;AACH;AAEA,SAAS,eACP,SAIU;AACV,QAAM,OAAO,CAAC,OAAO;AAErB,MAAI,QAAQ,YAAY,QAAW;AACjC,SAAK,KAAK,cAAc,QAAQ,OAAO;AAAA,EACzC;AACA,MAAI,QAAQ,0BAA0B,OAAO;AAC3C,SAAK,KAAK,0BAA0B;AAAA,EACtC;AACA,MAAI,QAAQ,4BAA4B,QAAW;AACjD,SAAK;AAAA,MACH;AAAA,MACA,OAAO,QAAQ,uBAAuB;AAAA,IACxC;AAAA,EACF;AAEA,SAAO;AACT;AAEA,eAAe,oBACb,SAQA,UAC0C;AAC1C,QAAM,WAAW,gBAAgB,QAAQ,QAAQ;AACjD,QAAM,WAAW,OAAO,QAAQ,SAAS,OAAO,GAAG,QAAQ,UAAU;AAAA,IACnE,MAAM,KAAK,UAAU;AAAA,MACnB,SAAS;AAAA,QACP,GAAI,QAAQ,0BAA0B,SAClC,CAAC,IACD,EAAE,wBAAwB,QAAQ,sBAAsB;AAAA,QAC5D,GAAI,QAAQ,4BAA4B,SACpC,CAAC,IACD;AAAA,UACE,4BAA4B,QAAQ;AAAA,QACtC;AAAA,MACN;AAAA,MACA,MAAM,SAAS;AAAA,IACjB,CAAC;AAAA,IACD,SAAS;AAAA,MACP,QAAQ;AAAA,MACR,gBAAgB;AAAA,IAClB;AAAA,IACA,QAAQ;AAAA,EACV,CAAC;AAED,MAAI,CAAC,SAAS,IAAI;AAChB,UAAM,IAAI;AAAA,MACR;AAAA,QACE,6CAA6C,SAAS,MAAM,KAAK,MAAM,SAAS,KAAK,CAAC;AAAA,MACxF;AAAA,IACF;AAAA,EACF;AAEA,QAAM,WAAW,gCAAgC,QAAQ,UAAU;AACnE,QAAM,SAAS,4BAA4B;AAAA,IACzC,SAAS,CAAC,UAAU;AAClB,eAAS,OAAO;AAAA,QACd,iBAAiB,SAAS,MAAM,QAAQ,MAAM;AAAA,MAChD,CAAC;AAAA,IACH;AAAA,EACF,CAAC;AAED,MAAI,SAAS,SAAS,MAAM;AAC1B,WAAO,KAAK,MAAM,SAAS,KAAK,CAAC;AACjC,UAAMA,WAAU,OAAO,OAAO;AAC9B,UAAM,SAAS,KAAK;AACpB,WAAOA;AAAA,EACT;AAEA,QAAM,SAAS,SAAS,KAAK,UAAU;AACvC,QAAM,UAAU,IAAI,YAAY;AAEhC,SAAO,MAAM;AACX,UAAM,EAAE,MAAM,MAAM,IAAI,MAAM,OAAO,KAAK;AAC1C,QAAI,MAAM;AACR;AAAA,IACF;AACA,WAAO,KAAK,QAAQ,OAAO,OAAO,EAAE,QAAQ,KAAK,CAAC,CAAC;AACnD,aAAS,cAAc;AAAA,EACzB;AAEA,SAAO,KAAK,QAAQ,OAAO,CAAC;AAC5B,QAAM,UAAU,OAAO,OAAO;AAC9B,QAAM,SAAS,KAAK;AACpB,SAAO;AACT;AAEA,SAAS,gCACP,YAGA,SAOA;AACA,QAAM,QAAyB,CAAC;AAChC,MAAI;AAEJ,WAAS,KAAK,OAAsB;AAClC,QAAI,YAAY,QAAW;AACzB;AAAA,IACF;AACA,cAAU,QAAQ,KAAK;AACvB,aAAS,YAAY,OAAO;AAAA,EAC9B;AAEA,SAAO;AAAA,IACL,QAAQ,CAAC,aAAa;AACpB,UAAI,eAAe,UAAa,YAAY,QAAW;AACrD;AAAA,MACF;AAEA,UAAI;AACF,cAAM,KAAK,QAAQ,QAAQ,WAAW,QAAQ,CAAC,EAAE,MAAM,IAAI,CAAC;AAAA,MAC9D,SAAS,OAAO;AACd,aAAK,KAAK;AAAA,MACZ;AAAA,IACF;AAAA,IACA,eAAe,MAAM;AACnB,UAAI,YAAY,QAAW;AACzB,cAAM;AAAA,MACR;AAAA,IACF;AAAA,IACA,MAAM,YAAY;AAChB,YAAM,QAAQ,IAAI,KAAK;AACvB,UAAI,YAAY,QAAW;AACzB,cAAM;AAAA,MACR;AAAA,IACF;AAAA,EACF;AACF;AAEA,eAAe,uBACb,UACA,UAAwB,OACI;AAC5B,QAAM,QAAQ,MAAM,QAAQ,GAAG,QAAQ,SAAS;AAEhD,MAAI,CAAC,MAAM,IAAI;AACb,UAAM,IAAI;AAAA,MACR;AAAA,QACE,+CAA+C,MAAM,MAAM;AAAA,MAC7D;AAAA,IACF;AAAA,EACF;AAEA,QAAM,WAAW,MAAM,QAAQ,GAAG,QAAQ,WAAW;AAErD,MAAI,CAAC,SAAS,IAAI;AAChB,UAAM,IAAI;AAAA,MACR;AAAA,QACE,+CAA+C,SAAS,MAAM;AAAA,MAChE;AAAA,IACF;AAAA,EACF;AAEA,MAAI;AACF,WAAO,uBAAuB,MAAM,SAAS,KAAK,CAAC;AAAA,EACrD,SAAS,OAAO;AACd,QAAI,iBAAiB,SAAS,MAAM,QAAQ,SAAS,aAAa,GAAG;AACnE,YAAM;AAAA,IACR;AAEA,UAAM,IAAI;AAAA,MACR;AAAA,QACE,wCAAwC,iBAAiB,QAAQ,MAAM,UAAU,OAAO,KAAK,CAAC;AAAA,MAChG;AAAA,IACF;AAAA,EACF;AACF;AAEA,SAAS,4BAA4B,OAKnC;AACA,QAAM,UAAkC,CAAC;AACzC,MAAI,SAAS;AAEb,WAAS,UAAU,MAAoB;AACrC,QAAI,KAAK,KAAK,MAAM,IAAI;AACtB;AAAA,IACF;AAEA,UAAM,QAAQ,oBAAoB,KAAK,MAAM,IAAI,CAAC;AAElD,QAAI,MAAM,SAAS,SAAS;AAC1B,cAAQ,KAAK,MAAM,KAAK;AACxB,aAAO,UAAU,MAAM,KAAK;AAAA,IAC9B;AAAA,EACF;AAEA,WAAS,eAAe,OAAuB;AAC7C,WAAO,IAAI;AAAA,MACT;AAAA,QACE,qCAAqC,iBAAiB,QAAQ,MAAM,UAAU,OAAO,KAAK,CAAC;AAAA,MAC7F;AAAA,IACF;AAAA,EACF;AAEA,SAAO;AAAA,IACL,QAAQ,MAAM;AACZ,UAAI;AACF,kBAAU,MAAM;AAChB,iBAAS;AACT,eAAO;AAAA,MACT,SAAS,OAAO;AACd,cAAM,eAAe,KAAK;AAAA,MAC5B;AAAA,IACF;AAAA,IACA,MAAM,CAAC,UAAU;AACf,UAAI;AACF,kBAAU;AACV,cAAM,QAAQ,OAAO,MAAM,QAAQ;AACnC,iBAAS,MAAM,IAAI,KAAK;AAExB,mBAAW,QAAQ,OAAO;AACxB,oBAAU,IAAI;AAAA,QAChB;AAAA,MACF,SAAS,OAAO;AACd,cAAM,eAAe,KAAK;AAAA,MAC5B;AAAA,IACF;AAAA,EACF;AACF;AAEA,SAAS,oBAAoB,OAAgC;AAC3D,MAAI,OAAO,UAAU,YAAY,UAAU,MAAM;AAC/C,UAAM,IAAI,MAAM,2CAA2C;AAAA,EAC7D;AAEA,QAAM,SAAS;AAEf,MAAI,OAAO,SAAS,QAAQ;AAC1B,WAAO,EAAE,MAAM,OAAO;AAAA,EACxB;AACA,MAAI,OAAO,SAAS,WAAW,uBAAuB,OAAO,KAAK,GAAG;AACnE,WAAO;AAAA,MACL,OAAO,OAAO;AAAA,MACd,MAAM;AAAA,IACR;AAAA,EACF;AAEA,QAAM,IAAI,MAAM,+BAA+B,KAAK,UAAU,KAAK,CAAC,EAAE;AACxE;AAEA,SAAS,QAAQ,OAAuB;AACtC,SAAO,iBAAiB,QAAQ,QAAQ,IAAI,MAAM,OAAO,KAAK,CAAC;AACjE;AAEA,SAAS,uBAAuB,OAAmC;AACjE,MAAI,OAAO,UAAU,YAAY,UAAU,MAAM;AAC/C,UAAM,IAAI;AAAA,MACR,4BAA4B,8BAA8B;AAAA,IAC5D;AAAA,EACF;AAEA,QAAM,SAAS;AAEf,MACE,OAAO,OAAO,0BAA0B,YACxC,OAAO,OAAO,WAAW,YACzB,OAAO,OAAO,cAAc,YAC5B,OAAO,OAAO,kBAAkB,YAChC,OAAO,OAAO,0BAA0B,UACxC;AACA,UAAM,IAAI,MAAM,4BAA4B,6BAA6B,CAAC;AAAA,EAC5E;AAEA,SAAO;AAAA,IACL,uBAAuB,OAAO;AAAA,IAC9B,QAAQ,OAAO;AAAA,IACf,WAAW,OAAO;AAAA,IAClB,eAAe,OAAO;AAAA,IACtB,uBAAuB,OAAO;AAAA,EAChC;AACF;AAEA,SAAS,uBAAuB,OAA+C;AAC7E,MAAI,OAAO,UAAU,YAAY,UAAU,MAAM;AAC/C,WAAO;AAAA,EACT;AAEA,QAAM,SAAS;AAEf,SACE,OAAO,OAAO,QAAQ,YACtB,MAAM,QAAQ,OAAO,IAAI,KACzB,OAAO,KAAK,MAAM,uBAAuB,KACzC,OAAO,OAAO,UAAU,YACxB,OAAO,OAAO,eAAe;AAEjC;AAEA,SAAS,wBACP,OACgC;AAChC,MAAI,OAAO,UAAU,YAAY,UAAU,MAAM;AAC/C,WAAO;AAAA,EACT;AAEA,QAAM,SAAS;AAEf,UACG,EAAE,oBAAoB,WACrB,OAAO,OAAO,mBAAmB,cACnC,OAAO,OAAO,QAAQ,YACtB,eAAe,KAAK,OAAO,GAAG;AAElC;AAEA,SAAS,YAAY,WAAsD;AACzE,SAAO;AAAA,IACL,kBAAkB,UAAU,mBAAmB;AAAA,IAC/C,KAAK,UAAU;AAAA,EACjB;AACF;AAEA,SAAS,gBAAgB,SAEH;AACpB,SAAO,QAAQ,YAAY;AAC7B;AAEA,SAAS,gBAAgB,UAAsC;AAC7D,QAAM,cAAc,YAAY,kCAC7B,KAAK,EACL,QAAQ,SAAS,EAAE;AAEtB,MAAI,eAAe,IAAI;AACrB,UAAM,IAAI;AAAA,MACR;AAAA,QACE;AAAA,MACF;AAAA,IACF;AAAA,EACF;AAEA,SAAO;AACT;AAEA,SAAS,4BAA4B,SAAyB;AAC5D,SAAO,GAAG,OAAO;AAAA,yBAA4B,2BAA2B;AAC1E;","names":["matches"]}

package/dist/external/wikimatch/window.d.ts

@@ -0,0 +1,8 @@
+import { BuildWikimatchWindowsOptions, WikimatchWindow } from './types.js';
+import '../wikipage/types.js';
+import '../wikipage/wikimedia-client/wiki.js';
+import '../wikipage/fetch-log.js';
+
+declare function buildWikimatchWindows(options: BuildWikimatchWindowsOptions): readonly WikimatchWindow[];
+
+export { buildWikimatchWindows };

package/dist/external/wikimatch/window.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../../../src/external/wikimatch/window.ts"],"sourcesContent":["import { expandRangeByWords } from \"./words.js\";\nimport { countWikimatchCandidateOptions } from \"./options.js\";\nimport type {\n  BuildWikimatchWindowsOptions,\n  WikimatchCandidate,\n  WikimatchWindow,\n} from \"./types.js\";\n\ninterface CandidateGroup {\n  readonly candidates: readonly WikimatchCandidate[];\n  readonly range: {\n    readonly end: number;\n    readonly start: number;\n  };\n}\n\nexport function buildWikimatchWindows(\n  options: BuildWikimatchWindowsOptions,\n): readonly WikimatchWindow[] {\n  const groups = groupOverlappingCandidates(\n    normalizeCandidates(options.candidates ?? []),\n  );\n  const windows: WikimatchWindow[] = [];\n  let pendingGroups: CandidateGroup[] = [];\n  let pendingOptionCount = 0;\n\n  for (const group of groups) {\n    const groupOptionCount = countGroupOptions(group);\n\n    if (\n      pendingGroups.length > 0 &&\n      pendingOptionCount + groupOptionCount > options.optionBudget\n    ) {\n      windows.push(createWindow(options.text, pendingGroups, options));\n      pendingGroups = [];\n      pendingOptionCount = 0;\n    }\n\n    pendingGroups.push(group);\n    pendingOptionCount += groupOptionCount;\n  }\n\n  if (pendingGroups.length > 0) {\n    windows.push(createWindow(options.text, pendingGroups, options));\n  }\n\n  return windows;\n}\n\nfunction countGroupOptions(group: CandidateGroup): number {\n  return group.candidates.reduce(\n    (total, candidate) => total + countWikimatchCandidateOptions(candidate),\n    0,\n  );\n}\n\nfunction createWindow(\n  text: string,\n  groups: readonly CandidateGroup[],\n  options: BuildWikimatchWindowsOptions,\n): WikimatchWindow {\n  const groupStart = Math.min(...groups.map((group) => group.range.start));\n  const groupEnd = Math.max(...groups.map((group) => group.range.end));\n  const expandedRange = expandRangeByWords({\n    rangeEnd: groupEnd,\n    rangeStart: groupStart,\n    text,\n    words: options.contextWords,\n  });\n\n  return {\n    baseOffset: expandedRange.start,\n    candidates: groups.flatMap((group) => group.candidates),\n    groups: groups.map((group, index) => ({\n      candidateIds: group.candidates.map((candidate) => candidate.id),\n      id: `g${index + 1}`,\n      range: group.range,\n    })),\n    text: text.slice(expandedRange.start, expandedRange.end),\n  };\n}\n\nfunction groupOverlappingCandidates(\n  candidates: readonly WikimatchCandidate[],\n): readonly CandidateGroup[] {\n  const groups: CandidateGroup[] = [];\n  let current: CandidateGroup | undefined;\n\n  for (const candidate of candidates) {\n    if (current === undefined || candidate.range.start >= current.range.end) {\n      current = {\n        candidates: [candidate],\n        range: candidate.range,\n      };\n      groups.push(current);\n      continue;\n    }\n\n    current = {\n      candidates: [...current.candidates, candidate],\n      range: {\n        end: Math.max(current.range.end, candidate.range.end),\n        start: Math.min(current.range.start, candidate.range.start),\n      },\n    };\n    groups[groups.length - 1] = current;\n  }\n\n  return groups;\n}\n\nfunction normalizeCandidates(\n  candidates: readonly WikimatchCandidate[],\n): readonly WikimatchCandidate[] {\n  return candidates\n    .filter((candidate) => candidate.range.start < candidate.range.end)\n    .sort(compareCandidate);\n}\n\nfunction compareCandidate(\n  left: WikimatchCandidate,\n  right: WikimatchCandidate,\n): number {\n  return (\n    left.range.start - right.range.start ||\n    right.range.end - left.range.end ||\n    left.id.localeCompare(right.id)\n  );\n}\n"],"mappings":"AAAA,SAAS,0BAA0B;AACnC,SAAS,sCAAsC;AAexC,SAAS,sBACd,SAC4B;AAC5B,QAAM,SAAS;AAAA,IACb,oBAAoB,QAAQ,cAAc,CAAC,CAAC;AAAA,EAC9C;AACA,QAAM,UAA6B,CAAC;AACpC,MAAI,gBAAkC,CAAC;AACvC,MAAI,qBAAqB;AAEzB,aAAW,SAAS,QAAQ;AAC1B,UAAM,mBAAmB,kBAAkB,KAAK;AAEhD,QACE,cAAc,SAAS,KACvB,qBAAqB,mBAAmB,QAAQ,cAChD;AACA,cAAQ,KAAK,aAAa,QAAQ,MAAM,eAAe,OAAO,CAAC;AAC/D,sBAAgB,CAAC;AACjB,2BAAqB;AAAA,IACvB;AAEA,kBAAc,KAAK,KAAK;AACxB,0BAAsB;AAAA,EACxB;AAEA,MAAI,cAAc,SAAS,GAAG;AAC5B,YAAQ,KAAK,aAAa,QAAQ,MAAM,eAAe,OAAO,CAAC;AAAA,EACjE;AAEA,SAAO;AACT;AAEA,SAAS,kBAAkB,OAA+B;AACxD,SAAO,MAAM,WAAW;AAAA,IACtB,CAAC,OAAO,cAAc,QAAQ,+BAA+B,SAAS;AAAA,IACtE;AAAA,EACF;AACF;AAEA,SAAS,aACP,MACA,QACA,SACiB;AACjB,QAAM,aAAa,KAAK,IAAI,GAAG,OAAO,IAAI,CAAC,UAAU,MAAM,MAAM,KAAK,CAAC;AACvE,QAAM,WAAW,KAAK,IAAI,GAAG,OAAO,IAAI,CAAC,UAAU,MAAM,MAAM,GAAG,CAAC;AACnE,QAAM,gBAAgB,mBAAmB;AAAA,IACvC,UAAU;AAAA,IACV,YAAY;AAAA,IACZ;AAAA,IACA,OAAO,QAAQ;AAAA,EACjB,CAAC;AAED,SAAO;AAAA,IACL,YAAY,cAAc;AAAA,IAC1B,YAAY,OAAO,QAAQ,CAAC,UAAU,MAAM,UAAU;AAAA,IACtD,QAAQ,OAAO,IAAI,CAAC,OAAO,WAAW;AAAA,MACpC,cAAc,MAAM,WAAW,IAAI,CAAC,cAAc,UAAU,EAAE;AAAA,MAC9D,IAAI,IAAI,QAAQ,CAAC;AAAA,MACjB,OAAO,MAAM;AAAA,IACf,EAAE;AAAA,IACF,MAAM,KAAK,MAAM,cAAc,OAAO,cAAc,GAAG;AAAA,EACzD;AACF;AAEA,SAAS,2BACP,YAC2B;AAC3B,QAAM,SAA2B,CAAC;AAClC,MAAI;AAEJ,aAAW,aAAa,YAAY;AAClC,QAAI,YAAY,UAAa,UAAU,MAAM,SAAS,QAAQ,MAAM,KAAK;AACvE,gBAAU;AAAA,QACR,YAAY,CAAC,SAAS;AAAA,QACtB,OAAO,UAAU;AAAA,MACnB;AACA,aAAO,KAAK,OAAO;AACnB;AAAA,IACF;AAEA,cAAU;AAAA,MACR,YAAY,CAAC,GAAG,QAAQ,YAAY,SAAS;AAAA,MAC7C,OAAO;AAAA,QACL,KAAK,KAAK,IAAI,QAAQ,MAAM,KAAK,UAAU,MAAM,GAAG;AAAA,QACpD,OAAO,KAAK,IAAI,QAAQ,MAAM,OAAO,UAAU,MAAM,KAAK;AAAA,MAC5D;AAAA,IACF;AACA,WAAO,OAAO,SAAS,CAAC,IAAI;AAAA,EAC9B;AAEA,SAAO;AACT;AAEA,SAAS,oBACP,YAC+B;AAC/B,SAAO,WACJ,OAAO,CAAC,cAAc,UAAU,MAAM,QAAQ,UAAU,MAAM,GAAG,EACjE,KAAK,gBAAgB;AAC1B;AAEA,SAAS,iBACP,MACA,OACQ;AACR,SACE,KAAK,MAAM,QAAQ,MAAM,MAAM,SAC/B,MAAM,MAAM,MAAM,KAAK,MAAM,OAC7B,KAAK,GAAG,cAAc,MAAM,EAAE;AAElC;","names":[]}

package/dist/external/wikimatch/words.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../../../src/external/wikimatch/words.ts"],"sourcesContent":["export interface WordBoundary {\n  readonly end: number;\n  readonly start: number;\n}\n\nexport function listWordBoundaries(text: string): readonly WordBoundary[] {\n  const segmenter = new Intl.Segmenter(undefined, {\n    granularity: \"word\",\n  });\n  const graphemeSegmenter = new Intl.Segmenter(undefined, {\n    granularity: \"grapheme\",\n  });\n  const words: WordBoundary[] = [];\n\n  for (const segment of segmenter.segment(text)) {\n    if (\"isWordLike\" in segment && segment.isWordLike) {\n      words.push({\n        end: segment.index + segment.segment.length,\n        start: segment.index,\n      });\n    }\n  }\n\n  if (words.length > 0) {\n    return words;\n  }\n\n  return [...graphemeSegmenter.segment(text)]\n    .filter((segment) => segment.segment.trim() !== \"\")\n    .map((segment) => ({\n      end: segment.index + segment.segment.length,\n      start: segment.index,\n    }));\n}\n\nexport function expandRangeByWords(input: {\n  readonly rangeEnd: number;\n  readonly rangeStart: number;\n  readonly text: string;\n  readonly words: number;\n}): { readonly end: number; readonly start: number } {\n  const wordBoundaries = listWordBoundaries(input.text);\n\n  if (wordBoundaries.length === 0) {\n    return {\n      end: input.text.length,\n      start: 0,\n    };\n  }\n\n  const firstWordIndex = findWordIndexAtOrAfter(\n    wordBoundaries,\n    input.rangeStart,\n  );\n  const lastWordIndex = findWordIndexBeforeOrAt(wordBoundaries, input.rangeEnd);\n  const halfBudget = Math.max(0, Math.floor(input.words / 2));\n  const startIndex = Math.max(0, firstWordIndex - halfBudget);\n  const endIndex = Math.min(\n    wordBoundaries.length - 1,\n    lastWordIndex + halfBudget,\n  );\n\n  return {\n    end: wordBoundaries[endIndex]?.end ?? input.text.length,\n    start: wordBoundaries[startIndex]?.start ?? 0,\n  };\n}\n\nfunction findWordIndexAtOrAfter(\n  words: readonly WordBoundary[],\n  offset: number,\n): number {\n  const matched = words.findIndex((word) => word.end > offset);\n\n  return matched === -1 ? words.length - 1 : matched;\n}\n\nfunction findWordIndexBeforeOrAt(\n  words: readonly WordBoundary[],\n  offset: number,\n): number {\n  for (let index = words.length - 1; index >= 0; index -= 1) {\n    const word = words[index];\n\n    if (word !== undefined && word.start < offset) {\n      return index;\n    }\n  }\n\n  return 0;\n}\n"],"mappings":"AAKO,SAAS,mBAAmB,MAAuC;AACxE,QAAM,YAAY,IAAI,KAAK,UAAU,QAAW;AAAA,IAC9C,aAAa;AAAA,EACf,CAAC;AACD,QAAM,oBAAoB,IAAI,KAAK,UAAU,QAAW;AAAA,IACtD,aAAa;AAAA,EACf,CAAC;AACD,QAAM,QAAwB,CAAC;AAE/B,aAAW,WAAW,UAAU,QAAQ,IAAI,GAAG;AAC7C,QAAI,gBAAgB,WAAW,QAAQ,YAAY;AACjD,YAAM,KAAK;AAAA,QACT,KAAK,QAAQ,QAAQ,QAAQ,QAAQ;AAAA,QACrC,OAAO,QAAQ;AAAA,MACjB,CAAC;AAAA,IACH;AAAA,EACF;AAEA,MAAI,MAAM,SAAS,GAAG;AACpB,WAAO;AAAA,EACT;AAEA,SAAO,CAAC,GAAG,kBAAkB,QAAQ,IAAI,CAAC,EACvC,OAAO,CAAC,YAAY,QAAQ,QAAQ,KAAK,MAAM,EAAE,EACjD,IAAI,CAAC,aAAa;AAAA,IACjB,KAAK,QAAQ,QAAQ,QAAQ,QAAQ;AAAA,IACrC,OAAO,QAAQ;AAAA,EACjB,EAAE;AACN;AAEO,SAAS,mBAAmB,OAKkB;AACnD,QAAM,iBAAiB,mBAAmB,MAAM,IAAI;AAEpD,MAAI,eAAe,WAAW,GAAG;AAC/B,WAAO;AAAA,MACL,KAAK,MAAM,KAAK;AAAA,MAChB,OAAO;AAAA,IACT;AAAA,EACF;AAEA,QAAM,iBAAiB;AAAA,IACrB;AAAA,IACA,MAAM;AAAA,EACR;AACA,QAAM,gBAAgB,wBAAwB,gBAAgB,MAAM,QAAQ;AAC5E,QAAM,aAAa,KAAK,IAAI,GAAG,KAAK,MAAM,MAAM,QAAQ,CAAC,CAAC;AAC1D,QAAM,aAAa,KAAK,IAAI,GAAG,iBAAiB,UAAU;AAC1D,QAAM,WAAW,KAAK;AAAA,IACpB,eAAe,SAAS;AAAA,IACxB,gBAAgB;AAAA,EAClB;AAEA,SAAO;AAAA,IACL,KAAK,eAAe,QAAQ,GAAG,OAAO,MAAM,KAAK;AAAA,IACjD,OAAO,eAAe,UAAU,GAAG,SAAS;AAAA,EAC9C;AACF;AAEA,SAAS,uBACP,OACA,QACQ;AACR,QAAM,UAAU,MAAM,UAAU,CAAC,SAAS,KAAK,MAAM,MAAM;AAE3D,SAAO,YAAY,KAAK,MAAM,SAAS,IAAI;AAC7C;AAEA,SAAS,wBACP,OACA,QACQ;AACR,WAAS,QAAQ,MAAM,SAAS,GAAG,SAAS,GAAG,SAAS,GAAG;AACzD,UAAM,OAAO,MAAM,KAAK;AAExB,QAAI,SAAS,UAAa,KAAK,QAAQ,QAAQ;AAC7C,aAAO;AAAA,IACT;AAAA,EACF;AAEA,SAAO;AACT;","names":[]}

package/dist/external/wikipage/cache.d.ts

@@ -0,0 +1,20 @@
+import { Database } from '../../document/database.js';
+import { GcContext, GcJobResult } from '../../runtime/gc/types.js';
+import { EnrichmentStore, CachedQidRecord, CachedDisambiguationRecord } from './types.js';
+import 'sqlite3';
+import './wikimedia-client/wiki.js';
+import './fetch-log.js';
+
+declare class WikipageCache implements EnrichmentStore {
+    #private;
+    constructor(database: Database);
+    static open(path?: string): Promise<WikipageCache>;
+    close(): Promise<void>;
+    getQids(qids: readonly string[], language: string): Promise<ReadonlyMap<string, CachedQidRecord>>;
+    putQids(records: readonly CachedQidRecord[], language: string): Promise<void>;
+    getDisambiguations(qids: readonly string[], wiki: string): Promise<ReadonlyMap<string, CachedDisambiguationRecord>>;
+    putDisambiguations(records: readonly CachedDisambiguationRecord[], wiki: string): Promise<void>;
+}
+declare function runWikipageCacheGc(context: GcContext): Promise<GcJobResult>;
+
+export { WikipageCache, runWikipageCacheGc };