easyoref 1.14.2 → 1.15.0

package/dist/agent/graph.d.ts

@@ -1,79 +1,18 @@
 /**
  * LangGraph.js enrichment pipeline — phase-aware, time-validated.
  *
- * KEY DESIGN PRINCIPLES:
- *   1. TIME IS KING — every post is validated against the alert time window.
- *      LLM receives alert time + post time and scores time_relevance.
- *      Posts about previous/different attacks are rejected.
- *   2. PHASE-AWARE — each phase extracts only what's relevant:
- *      - early_warning: origin, ETA, rocket count, cassette
- *      - siren: carries early data + adds interception, impacts
- *      - resolved: carries all + adds casualties, injuries, final stats
- *   3. CARRY-FORWARD — results persist in Redis (EnrichmentData).
- *      Each phase inherits previous phase's findings.
- *   4. INLINE CITATIONS — no superscripts, no footer sources.
- *      Format: [[1]](url) right after each data point.
- *   5. DEDUP EDITS — hash-based check prevents "message not modified" spam.
+ * Lean orchestrator: connects filter → extract → vote → edit.
+ * All logic lives in dedicated modules:
+ *   - filters.ts:  deterministic noise filter, channel tracking
+ *   - extract.ts:  cheap LLM pre-filter, expensive extraction, post-filter
+ *   - vote.ts:     consensus voting (deterministic)
+ *   - message.ts:  message building, Telegram editing
+ *   - helpers.ts:  toIsraelTime, textHash
  *
  * Pipeline:
- *   preFilter → extractAndValidate → postFilter → vote → [clarify] → editMessage
+ *   collectAndFilter → extract → vote → [clarify → revote] → editMessage
  */
-import { ChatOpenAI } from "@langchain/openai";
-import type { ChannelPost } from "./store.js";
-import type { AlertType, CitedSource, EnrichmentData, InlineCite, ValidatedExtraction, VotedResult } from "./types.js";
-declare const AgentState: import("@langchain/langgraph").AnnotationRoot<{
-    alertId: import("@langchain/langgraph").BaseChannel<string, string | import("@langchain/langgraph").OverwriteValue<string>, unknown>;
-    alertTs: import("@langchain/langgraph").BaseChannel<number, number | import("@langchain/langgraph").OverwriteValue<number>, unknown>;
-    alertType: import("@langchain/langgraph").BaseChannel<AlertType, AlertType | import("@langchain/langgraph").OverwriteValue<AlertType>, unknown>;
-    alertAreas: import("@langchain/langgraph").BaseChannel<string[], string[] | import("@langchain/langgraph").OverwriteValue<string[]>, unknown>;
-    chatId: import("@langchain/langgraph").BaseChannel<string, string | import("@langchain/langgraph").OverwriteValue<string>, unknown>;
-    messageId: import("@langchain/langgraph").BaseChannel<number, number | import("@langchain/langgraph").OverwriteValue<number>, unknown>;
-    isCaption: import("@langchain/langgraph").BaseChannel<boolean, boolean | import("@langchain/langgraph").OverwriteValue<boolean>, unknown>;
-    currentText: import("@langchain/langgraph").BaseChannel<string, string | import("@langchain/langgraph").OverwriteValue<string>, unknown>;
-    channelPosts: import("@langchain/langgraph").BaseChannel<ChannelPost[], ChannelPost[] | import("@langchain/langgraph").OverwriteValue<ChannelPost[]>, unknown>;
-    filteredPosts: import("@langchain/langgraph").BaseChannel<ChannelPost[], ChannelPost[] | import("@langchain/langgraph").OverwriteValue<ChannelPost[]>, unknown>;
-    extractions: import("@langchain/langgraph").BaseChannel<ValidatedExtraction[], ValidatedExtraction[] | import("@langchain/langgraph").OverwriteValue<ValidatedExtraction[]>, unknown>;
-    votedResult: import("@langchain/langgraph").BaseChannel<VotedResult | null, VotedResult | import("@langchain/langgraph").OverwriteValue<VotedResult | null> | null, unknown>;
-    /** Tracks whether clarify has already run (prevents infinite loop) */
-    clarifyAttempted: import("@langchain/langgraph").BaseChannel<boolean, boolean | import("@langchain/langgraph").OverwriteValue<boolean>, unknown>;
-    /** Cross-phase enrichment data loaded at start */
-    previousEnrichment: import("@langchain/langgraph").BaseChannel<EnrichmentData, EnrichmentData | import("@langchain/langgraph").OverwriteValue<EnrichmentData>, unknown>;
-    /** Session start timestamp for time window calculations */
-    sessionStartTs: import("@langchain/langgraph").BaseChannel<number, number | import("@langchain/langgraph").OverwriteValue<number>, unknown>;
-    /** Phase start timestamp */
-    phaseStartTs: import("@langchain/langgraph").BaseChannel<number, number | import("@langchain/langgraph").OverwriteValue<number>, unknown>;
-}>;
-type AgentStateType = typeof AgentState.State;
-declare function getLLM(): ChatOpenAI;
-declare function buildRegionKeywords(): string[];
-/** Format timestamp as HH:MM Israel time */
-declare function toIsraelTime(ts: number): string;
-/** MD5 hash for edit dedup */
-declare function textHash(text: string): string;
-/** Phase-specific extraction instructions */
-declare function getPhaseInstructions(alertType: AlertType): string;
-declare function postFilter(state: AgentStateType): Partial<AgentStateType>;
-declare function vote(state: AgentStateType): Partial<AgentStateType>;
-/** Format inline citations: [[1]](url), [[2]](url) */
-declare function inlineCites(indices: number[], citedSources: CitedSource[]): string;
-/** Get InlineCite[] from citation indices */
-declare function extractCites(indices: number[], citedSources: CitedSource[]): InlineCite[];
-/** Format inline citations from InlineCite[] (for carry-forward data) */
-declare function inlineCitesFromData(cites: InlineCite[]): string;
-/**
- * Build enrichment data from current vote + previous enrichment (carry-forward).
- * Returns updated EnrichmentData for Redis persistence.
- */
-declare function buildEnrichmentFromVote(r: VotedResult, prev: EnrichmentData, alertType: AlertType, alertTs: number): EnrichmentData;
-/**
- * Build the enriched message text from current message + enrichment data.
- * Uses inline [[1]](url) citations. No superscripts. No footer sources.
- */
-declare function buildEnrichedMessage(currentText: string, alertType: AlertType, alertTs: number, enrichment: EnrichmentData): string;
-/**
- * Insert a line before the time line (last "Время" / "Time" / "שעת" line).
- */
-declare function insertBeforeTimeLine(text: string, line: string): string;
+import type { AlertType } from "./types.js";
 export interface RunEnrichmentInput {
     alertId: string;
     alertTs: number;
@@ -85,27 +24,4 @@ export interface RunEnrichmentInput {
     currentText: string;
 }
 export declare function runEnrichment(input: RunEnrichmentInput): Promise<void>;
-export declare const _test: {
-    readonly getLLM: typeof getLLM;
-    readonly buildRegionKeywords: typeof buildRegionKeywords;
-    readonly LAUNCH_KEYWORDS: string[];
-    readonly TIME_WINDOW_MS: Record<AlertType, number>;
-    readonly toIsraelTime: typeof toIsraelTime;
-    readonly textHash: typeof textHash;
-    readonly postFilter: typeof postFilter;
-    readonly vote: typeof vote;
-    readonly buildEnrichmentFromVote: typeof buildEnrichmentFromVote;
-    readonly buildEnrichedMessage: typeof buildEnrichedMessage;
-    readonly insertBeforeTimeLine: typeof insertBeforeTimeLine;
-    readonly inlineCites: typeof inlineCites;
-    readonly inlineCitesFromData: typeof inlineCitesFromData;
-    readonly extractCites: typeof extractCites;
-    readonly COUNTRY_RU: Record<string, string>;
-    readonly SYSTEM_PROMPT_BASE: "You analyze Telegram channel messages about a missile/rocket attack on Israel.\nYour job: extract factual data, assess quality, AND validate temporal relevance.\n\nCRITICAL — TIME VALIDATION:\nYou will receive the alert time and the post time. You MUST determine if this post\nis about the CURRENT attack or about a previous/different event.\n- If post discusses events clearly BEFORE the alert time → time_relevance=0\n- If post is generic military news not specific to this attack → time_relevance=0.2\n- If post discusses the current attack → time_relevance=1.0\n- If uncertain → time_relevance=0.5 (the system will use alert_history to verify)\n\nReturn ONLY valid JSON (no markdown, no explanation):\n{\n  \"region_relevance\": float,       // 0–1: does this message discuss the specified alert region?\n  \"source_trust\": float,           // 0–1: factual reporting (1.0) vs unverified rumors/panic (0.0)\n  \"tone\": \"calm\"|\"neutral\"|\"alarmist\",\n  \"time_relevance\": float,         // 0–1: is this post about the CURRENT attack? (see rules above)\n  \"country_origin\": string|null,   // \"Iran\",\"Yemen\",\"Lebanon\",\"Gaza\",\"Iraq\",\"Syria\" or null\n  \"rocket_count\": int|null,\n  \"is_cassette\": bool|null,\n  \"intercepted\": int|null,\n  \"intercepted_qual\": \"all\"|\"most\"|\"many\"|\"few\"|\"exists\"|\"none\"|\"more_than\"|\"less_than\"|null,\n  \"intercepted_qual_num\": int|null,\n  \"sea_impact\": int|null,\n  \"sea_impact_qual\": \"all\"|\"most\"|\"many\"|\"few\"|\"exists\"|\"none\"|\"more_than\"|\"less_than\"|null,\n  \"sea_impact_qual_num\": int|null,\n  \"open_area_impact\": int|null,\n  \"open_area_impact_qual\": \"all\"|\"most\"|\"many\"|\"few\"|\"exists\"|\"none\"|\"more_than\"|\"less_than\"|null,\n  \"open_area_impact_qual_num\": int|null,\n  \"hits_confirmed\": int|null,\n  \"casualties\": int|null,\n  \"injuries\": int|null,\n  \"eta_refined_minutes\": int|null,\n  \"confidence\": float\n}\n\nRules:\n- If unrelated to the alert region, set region_relevance=0 and all data fields to null.\n- If message is speculative/unconfirmed rumor, set source_trust < 0.4.\n- If message uses excessive caps, exclamation marks, panic language → tone=\"alarmist\".\n- Only extract concrete numbers explicitly stated in the text. Never guess.\n- *_qual fields: use ONLY when NO exact count is given. If exact number present, set *_qual=null.\n- \"none\" qual is only valid if explicitly stated (e.g., \"все перехвачены\", \"не упало в море\").\n- For IDF (@idf_telegram) posts about ongoing operations (not this specific attack) → time_relevance=0.\n- LANGUAGE NEUTRALITY: Posts may be in Hebrew, Russian, Arabic, or English. The language of the post\n  MUST NOT affect source_trust or confidence. Russian-language Israeli channels are equally reliable\n  and often break news faster than Hebrew ones. Judge ONLY by factual content and tone.\n- TRUST INTERCEPTION & IMPACT REPORTS: When a channel explicitly states interception results\n  (e.g., \"перехвачены\", \"intercepted\", \"יירוט\", \"упали в море\", \"fell in the sea\", \"נפלו בים\",\n  \"open area impact\", \"שטח פתוח\"), trust these claims with source_trust >= 0.7 and confidence >= 0.7.\n  Israeli Telegram channels often report interception results before official confirmation,\n  and these reports are typically accurate. Do NOT downgrade these just because they lack official source.";
-    readonly getPhaseInstructions: typeof getPhaseInstructions;
-    readonly SKIP: 0.6;
-    readonly UNCERTAIN: 0.75;
-    readonly CERTAIN: 0.95;
-};
-export {};
 //# sourceMappingURL=graph.d.ts.map

package/dist/agent/graph.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"graph.d.ts","sourceRoot":"","sources":["../../src/agent/graph.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;GAmBG;AAGH,OAAO,EAAE,UAAU,EAAE,MAAM,mBAAmB,CAAC;AAM/C,OAAO,KAAK,EAAE,WAAW,EAAE,MAAM,YAAY,CAAC;AAS9C,OAAO,KAAK,EACV,SAAS,EACT,WAAW,EACX,cAAc,EAEd,UAAU,EAEV,mBAAmB,EACnB,WAAW,EACZ,MAAM,YAAY,CAAC;AAKpB,QAAA,MAAM,UAAU;;;;;;;;;;;;;IAad,sEAAsE;;IAEtE,kDAAkD;;IAElD,2DAA2D;;IAE3D,4BAA4B;;EAE5B,CAAC;AAEH,KAAK,cAAc,GAAG,OAAO,UAAU,CAAC,KAAK,CAAC;AAI9C,iBAAS,MAAM,IAAI,UAAU,CAc5B;AAID,iBAAS,mBAAmB,IAAI,MAAM,EAAE,CAiCvC;AAsDD,4CAA4C;AAC5C,iBAAS,YAAY,CAAC,EAAE,EAAE,MAAM,GAAG,MAAM,CAMxC;AAED,8BAA8B;AAC9B,iBAAS,QAAQ,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,CAEtC;AAmID,6CAA6C;AAC7C,iBAAS,oBAAoB,CAAC,SAAS,EAAE,SAAS,GAAG,MAAM,CAmB1D;AAkPD,iBAAS,UAAU,CAAC,KAAK,EAAE,cAAc,GAAG,OAAO,CAAC,cAAc,CAAC,CAqDlE;AAMD,iBAAS,IAAI,CAAC,KAAK,EAAE,cAAc,GAAG,OAAO,CAAC,cAAc,CAAC,CAiP5D;AAiBD,sDAAsD;AACtD,iBAAS,WAAW,CAAC,OAAO,EAAE,MAAM,EAAE,EAAE,YAAY,EAAE,WAAW,EAAE,GAAG,MAAM,CAS3E;AAED,6CAA6C;AAC7C,iBAAS,YAAY,CACnB,OAAO,EAAE,MAAM,EAAE,EACjB,YAAY,EAAE,WAAW,EAAE,GAC1B,UAAU,EAAE,CASd;AAED,yEAAyE;AACzE,iBAAS,mBAAmB,CAAC,KAAK,EAAE,UAAU,EAAE,GAAG,MAAM,CAKxD;AA0CD;;;GAGG;AACH,iBAAS,uBAAuB,CAC9B,CAAC,EAAE,WAAW,EACd,IAAI,EAAE,cAAc,EACpB,SAAS,EAAE,SAAS,EACpB,OAAO,EAAE,MAAM,GACd,cAAc,CAqGhB;AAED;;;GAGG;AACH,iBAAS,oBAAoB,CAC3B,WAAW,EAAE,MAAM,EACnB,SAAS,EAAE,SAAS,EACpB,OAAO,EAAE,MAAM,EACf,UAAU,EAAE,cAAc,GACzB,MAAM,CAyGR;AAED;;GAEG;AACH,iBAAS,oBAAoB,CAAC,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,GAAG,MAAM,CAUhE;AA2SD,MAAM,WAAW,kBAAkB;IACjC,OAAO,EAAE,MAAM,CAAC;IAChB,OAAO,EAAE,MAAM,CAAC;IAChB,SAAS,EAAE,SAAS,CAAC;IACrB,UAAU,EAAE,MAAM,EAAE,CAAC;IACrB,MAAM,EAAE,MAAM,CAAC;IACf,SAAS,EAAE,MAAM,CAAC;IAClB,SAAS,EAAE,OAAO,CAAC;IACnB,WAAW,EAAE,MAAM,CAAC;CACrB;AAED,wBAAsB,aAAa,CAAC,KAAK,EAAE,kBAAkB,GAAG,OAAO,CAAC,IAAI,CAAC,CAwB5E;AAID,eAAO,MAAM,KAAK;;;;;;;;;;;;;;;;;;;;;CAqBR,CAAC"}
+{"version":3,"file":"graph.d.ts","sourceRoot":"","sources":["../../src/agent/graph.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;GAaG;AAqBH,OAAO,KAAK,EACV,SAAS,EAMV,MAAM,YAAY,CAAC;AA8PpB,MAAM,WAAW,kBAAkB;IACjC,OAAO,EAAE,MAAM,CAAC;IAChB,OAAO,EAAE,MAAM,CAAC;IAChB,SAAS,EAAE,SAAS,CAAC;IACrB,UAAU,EAAE,MAAM,EAAE,CAAC;IACrB,MAAM,EAAE,MAAM,CAAC;IACf,SAAS,EAAE,MAAM,CAAC;IAClB,SAAS,EAAE,OAAO,CAAC;IACnB,WAAW,EAAE,MAAM,CAAC;CACrB;AAED,wBAAsB,aAAa,CAAC,KAAK,EAAE,kBAAkB,GAAG,OAAO,CAAC,IAAI,CAAC,CAmB5E"}