@jingyi0605/codingns 0.8.5 → 0.9.0

package/dist/server/modules/affairs-indexer/core/src/parser/plain-text-parser-adapter.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"plain-text-parser-adapter.js","sourceRoot":"","sources":["../../../../../../../src/modules/affairs-indexer/core/src/parser/plain-text-parser-adapter.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,MAAM,SAAS,CAAC;AACzB,OAAO,IAAI,MAAM,WAAW,CAAC;AAQ7B,MAAM,yBAAyB,GAAG,IAAI,GAAG,CAAC;IACxC,KAAK;IACL,MAAM;IACN,MAAM;IACN,MAAM;IACN,OAAO;IACP,MAAM;IACN,MAAM;IACN,OAAO;IACP,OAAO;IACP,MAAM;IACN,MAAM;CACP,CAAC,CAAC;AAEH,SAAS,mBAAmB,CAAC,IAAY;IACvC,OAAO,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC;AAC1C,CAAC;AAED,SAAS,YAAY,CAAC,IAAY,EAAE,KAAK,GAAG,GAAG;IAC7C,MAAM,UAAU,GAAG,mBAAmB,CAAC,IAAI,CAAC,CAAC;IAC7C,IAAI,UAAU,CAAC,MAAM,IAAI,KAAK,EAAE,CAAC;QAC/B,OAAO,UAAU,CAAC;IACpB,CAAC;IACD,OAAO,GAAG,UAAU,CAAC,KAAK,CAAC,CAAC,EAAE,KAAK,GAAG,CAAC,CAAC,GAAG,CAAC;AAC9C,CAAC;AAED;;GAEG;AACH,MAAM,OAAO,sBAAsB;IACxB,IAAI,GAAG,YAAY,CAAC;IAE7B,QAAQ,CAAC,GAAW;QAClB,OAAO,yBAAyB,CAAC,GAAG,CAAC,GAAG,CAAC,WAAW,EAAE,CAAC,CAAC;IAC1D,CAAC;IAED,KAAK,CAAC,YAAY;QAChB,OAAO,WAAW,CAAC;IACrB,CAAC;IAED,KAAK,CAAC,KAAK,CAAC,KAAiB;QAC3B,MAAM,GAAG,GAAG,EAAE,CAAC,YAAY,CAAC,KAAK,CAAC,QAAQ,EAAE,OAAO,CAAC,CAAC;QACrD,MAAM,KAAK,GAAG,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,QAAQ,EAAE,KAAK,CAAC,SAAS,CAAC,CAAC;QAE7D,OAAO;YACL,KAAK;YACL,IAAI,EAAE,GAAG;YACT,OAAO,EAAE,YAAY,CAAC,GAAG,CAAC;YAC1B,MAAM,EAAE,KAAK,CAAC,SAAS,KAAK,OAAO,IAAI,KAAK,CAAC,SAAS,KAAK,MAAM;gBAC/D,CAAC,CAAC,eAAe;gBACjB,CAAC,CAAC,KAAK,CAAC,SAAS,KAAK,OAAO,IAAI,KAAK,CAAC,SAAS,KAAK,OAAO,IAAI,KAAK,CAAC,SAAS,KAAK,MAAM,IAAI,KAAK,CAAC,SAAS,KAAK,MAAM;oBACtH,CAAC,CAAC,0BAA0B;oBAC5B,CAAC,CAAC,YAAY;YAClB,QAAQ,EAAE;gBACR,OAAO,EAAE,IAAI,CAAC,IAAI;aACnB;SACF,CAAC;IACJ,CAAC;CACF"}

package/dist/server/modules/affairs-indexer/core/src/parser/plain-text-parser.d.ts

@@ -0,0 +1,9 @@
+import type { StructuredDocument } from "./parser-adapter.js";
+export interface ParsedDocument {
+    title: string;
+    text: string;
+    summary: string;
+    parser: string;
+    metadata?: Record<string, unknown>;
+    structured?: StructuredDocument;
+}

package/dist/server/modules/affairs-indexer/core/src/parser/plain-text-parser.js

@@ -0,0 +1,2 @@
+export {};
+//# sourceMappingURL=plain-text-parser.js.map

package/dist/server/modules/affairs-indexer/core/src/parser/plain-text-parser.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"plain-text-parser.js","sourceRoot":"","sources":["../../../../../../../src/modules/affairs-indexer/core/src/parser/plain-text-parser.ts"],"names":[],"mappings":""}

package/dist/server/modules/affairs-indexer/core/src/parser/pptx-parser-adapter.d.ts

@@ -0,0 +1,7 @@
+import { BaseComplexParserAdapter } from "./base-complex-parser-adapter.js";
+import type { ParseInput, ParsedDocumentPayload } from "./parser-adapter.js";
+export declare class PptxParserAdapter extends BaseComplexParserAdapter {
+    readonly name = "pptx_parser";
+    supports(ext: string): boolean;
+    protected parseComplex(input: ParseInput): Promise<ParsedDocumentPayload>;
+}

package/dist/server/modules/affairs-indexer/core/src/parser/pptx-parser-adapter.js

@@ -0,0 +1,130 @@
+import fs from "node:fs";
+import path from "node:path";
+import { AppError, APP_ERROR_CODES } from "../../../contracts/src/index.js";
+import { BaseComplexParserAdapter } from "./base-complex-parser-adapter.js";
+import { decodeXmlEntities, extractXmlAttributes, readZipEntries, readZipText, shortSummary, } from "./openxml-utils.js";
+const SUPPORTED_PPTX_EXTENSIONS = new Set([".pptx"]);
+function resolvePresentationTarget(target) {
+    const normalized = path.posix.normalize(path.posix.join("ppt", target));
+    return normalized.replace(/^\/+/, "");
+}
+function parsePresentationRelationships(relationshipXml) {
+    const relationships = new Map();
+    const relationshipPattern = /<Relationship\b([^>]*)\/>/g;
+    for (const match of relationshipXml.matchAll(relationshipPattern)) {
+        const attributes = extractXmlAttributes(match[1]);
+        const relationId = attributes.Id;
+        const target = attributes.Target;
+        if (!relationId || !target) {
+            continue;
+        }
+        relationships.set(relationId, resolvePresentationTarget(target));
+    }
+    return relationships;
+}
+function parsePresentationSlidePaths(presentationXml, relationshipXml) {
+    const relationships = parsePresentationRelationships(relationshipXml);
+    const slidePaths = [];
+    const pattern = /<p:sldId\b([^>]*)\/>/g;
+    for (const match of presentationXml.matchAll(pattern)) {
+        const attributes = extractXmlAttributes(match[1]);
+        const relationId = attributes["r:id"];
+        if (!relationId) {
+            continue;
+        }
+        const target = relationships.get(relationId);
+        if (!target) {
+            continue;
+        }
+        slidePaths.push(target);
+    }
+    return slidePaths;
+}
+function parseSlideText(slideXml) {
+    const texts = [];
+    const textPattern = /<a:t\b[^>]*>([\s\S]*?)<\/a:t>/g;
+    for (const match of slideXml.matchAll(textPattern)) {
+        const text = decodeXmlEntities(match[1]).trim();
+        if (text) {
+            texts.push(text);
+        }
+    }
+    return texts.join("\n").replace(/\n{3,}/g, "\n\n").trim();
+}
+function parseSlides(entries, slidePaths) {
+    const slides = [];
+    for (let index = 0; index < slidePaths.length; index += 1) {
+        const slidePath = slidePaths[index];
+        const slideXml = readZipText(entries, slidePath, "PPTX", false);
+        if (!slideXml) {
+            continue;
+        }
+        const text = parseSlideText(slideXml);
+        if (!text) {
+            continue;
+        }
+        slides.push({
+            slideIndex: index + 1,
+            text,
+        });
+    }
+    return slides;
+}
+export class PptxParserAdapter extends BaseComplexParserAdapter {
+    name = "pptx_parser";
+    supports(ext) {
+        return SUPPORTED_PPTX_EXTENSIONS.has(ext.toLowerCase());
+    }
+    async parseComplex(input) {
+        let fileBuffer;
+        try {
+            fileBuffer = fs.readFileSync(input.filePath);
+        }
+        catch (error) {
+            throw new AppError(`PPTX 文件读取失败：${path.basename(input.filePath)}`, APP_ERROR_CODES.PARSER_COMPLEX_CONTENT_UNREADABLE, { cause: error });
+        }
+        const entries = readZipEntries(fileBuffer, "PPTX");
+        const presentationXml = readZipText(entries, "ppt/presentation.xml", "PPTX");
+        const relationshipXml = readZipText(entries, "ppt/_rels/presentation.xml.rels", "PPTX");
+        const slidePaths = parsePresentationSlidePaths(presentationXml, relationshipXml);
+        if (slidePaths.length === 0) {
+            throw new AppError("PPTX 中未发现 slide 列表", APP_ERROR_CODES.PARSER_COMPLEX_CONTENT_UNREADABLE);
+        }
+        const slides = parseSlides(entries, slidePaths);
+        if (slides.length === 0) {
+            throw new AppError("PPTX 中没有可读取的 slide 文本", APP_ERROR_CODES.PARSER_COMPLEX_CONTENT_UNREADABLE);
+        }
+        const blocks = slides.map(slide => ({
+            kind: "slide",
+            slideIndex: slide.slideIndex,
+            text: slide.text,
+            metadata: {
+                slideIndex: slide.slideIndex,
+            },
+        }));
+        const text = slides
+            .map(slide => `Slide ${slide.slideIndex}\n${slide.text}`.trim())
+            .join("\n\n")
+            .trim();
+        const title = path.basename(input.filePath, input.extension);
+        return {
+            title,
+            text,
+            summary: shortSummary(text),
+            parser: "pptx",
+            metadata: {
+                adapter: this.name,
+                slideCount: slidePaths.length,
+                extractedSlideCount: slides.length,
+            },
+            structured: {
+                blocks,
+                stats: {
+                    slideCount: slidePaths.length,
+                    extractedSlideCount: slides.length,
+                },
+            },
+        };
+    }
+}
+//# sourceMappingURL=pptx-parser-adapter.js.map

package/dist/server/modules/affairs-indexer/core/src/parser/pptx-parser-adapter.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"pptx-parser-adapter.js","sourceRoot":"","sources":["../../../../../../../src/modules/affairs-indexer/core/src/parser/pptx-parser-adapter.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,MAAM,SAAS,CAAC;AACzB,OAAO,IAAI,MAAM,WAAW,CAAC;AAC7B,OAAO,EAAE,QAAQ,EAAE,eAAe,EAAE,MAAM,iCAAiC,CAAC;AAC5E,OAAO,EAAE,wBAAwB,EAAE,MAAM,kCAAkC,CAAC;AAC5E,OAAO,EACL,iBAAiB,EACjB,oBAAoB,EACpB,cAAc,EACd,WAAW,EACX,YAAY,GACb,MAAM,oBAAoB,CAAC;AAG5B,MAAM,yBAAyB,GAAG,IAAI,GAAG,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC;AAOrD,SAAS,yBAAyB,CAAC,MAAc;IAC/C,MAAM,UAAU,GAAG,IAAI,CAAC,KAAK,CAAC,SAAS,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,KAAK,EAAE,MAAM,CAAC,CAAC,CAAC;IACxE,OAAO,UAAU,CAAC,OAAO,CAAC,MAAM,EAAE,EAAE,CAAC,CAAC;AACxC,CAAC;AAED,SAAS,8BAA8B,CAAC,eAAuB;IAC7D,MAAM,aAAa,GAAG,IAAI,GAAG,EAAkB,CAAC;IAChD,MAAM,mBAAmB,GAAG,4BAA4B,CAAC;IACzD,KAAK,MAAM,KAAK,IAAI,eAAe,CAAC,QAAQ,CAAC,mBAAmB,CAAC,EAAE,CAAC;QAClE,MAAM,UAAU,GAAG,oBAAoB,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC;QAClD,MAAM,UAAU,GAAG,UAAU,CAAC,EAAE,CAAC;QACjC,MAAM,MAAM,GAAG,UAAU,CAAC,MAAM,CAAC;QACjC,IAAI,CAAC,UAAU,IAAI,CAAC,MAAM,EAAE,CAAC;YAC3B,SAAS;QACX,CAAC;QACD,aAAa,CAAC,GAAG,CAAC,UAAU,EAAE,yBAAyB,CAAC,MAAM,CAAC,CAAC,CAAC;IACnE,CAAC;IACD,OAAO,aAAa,CAAC;AACvB,CAAC;AAED,SAAS,2BAA2B,CAAC,eAAuB,EAAE,eAAuB;IACnF,MAAM,aAAa,GAAG,8BAA8B,CAAC,eAAe,CAAC,CAAC;IACtE,MAAM,UAAU,GAAa,EAAE,CAAC;IAChC,MAAM,OAAO,GAAG,uBAAuB,CAAC;IACxC,KAAK,MAAM,KAAK,IAAI,eAAe,CAAC,QAAQ,CAAC,OAAO,CAAC,EAAE,CAAC;QACtD,MAAM,UAAU,GAAG,oBAAoB,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC;QAClD,MAAM,UAAU,GAAG,UAAU,CAAC,MAAM,CAAC,CAAC;QACtC,IAAI,CAAC,UAAU,EAAE,CAAC;YAChB,SAAS;QACX,CAAC;QACD,MAAM,MAAM,GAAG,aAAa,CAAC,GAAG,CAAC,UAAU,CAAC,CAAC;QAC7C,IAAI,CAAC,MAAM,EAAE,CAAC;YACZ,SAAS;QACX,CAAC;QACD,UAAU,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;IAC1B,CAAC;IACD,OAAO,UAAU,CAAC;AACpB,CAAC;AAED,SAAS,cAAc,CAAC,QAAgB;IACtC,MAAM,KAAK,GAAa,EAAE,CAAC;IAC3B,MAAM,WAAW,GAAG,gCAAgC,CAAC;IACrD,KAAK,MAAM,KAAK,IAAI,QAAQ,CAAC,QAAQ,CAAC,WAAW,CAAC,EAAE,CAAC;QACnD,MAAM,IAAI,GAAG,iBAAiB,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;QAChD,IAAI,IAAI,EAAE,CAAC;YACT,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QACnB,CAAC;IACH,CAAC;IAED,OAAO,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,OAAO,CAAC,SAAS,EAAE,MAAM,CAAC,CAAC,IAAI,EAAE,CAAC;AAC5D,CAAC;AAED,SAAS,WAAW,CAAC,OAA4B,EAAE,UAAoB;IACrE,MAAM,MAAM,GAAkB,EAAE,CAAC;IACjC,KAAK,IAAI,KAAK,GAAG,CAAC,EAAE,KAAK,GAAG,UAAU,CAAC,MAAM,EAAE,KAAK,IAAI,CAAC,EAAE,CAAC;QAC1D,MAAM,SAAS,GAAG,UAAU,CAAC,KAAK,CAAC,CAAC;QACpC,MAAM,QAAQ,GAAG,WAAW,CAAC,OAAO,EAAE,SAAS,EAAE,MAAM,EAAE,KAAK,CAAC,CAAC;QAChE,IAAI,CAAC,QAAQ,EAAE,CAAC;YACd,SAAS;QACX,CAAC;QACD,MAAM,IAAI,GAAG,cAAc,CAAC,QAAQ,CAAC,CAAC;QACtC,IAAI,CAAC,IAAI,EAAE,CAAC;YACV,SAAS;QACX,CAAC;QACD,MAAM,CAAC,IAAI,CAAC;YACV,UAAU,EAAE,KAAK,GAAG,CAAC;YACrB,IAAI;SACL,CAAC,CAAC;IACL,CAAC;IACD,OAAO,MAAM,CAAC;AAChB,CAAC;AAED,MAAM,OAAO,iBAAkB,SAAQ,wBAAwB;IACpD,IAAI,GAAG,aAAa,CAAC;IAE9B,QAAQ,CAAC,GAAW;QAClB,OAAO,yBAAyB,CAAC,GAAG,CAAC,GAAG,CAAC,WAAW,EAAE,CAAC,CAAC;IAC1D,CAAC;IAES,KAAK,CAAC,YAAY,CAAC,KAAiB;QAC5C,IAAI,UAAkB,CAAC;QACvB,IAAI,CAAC;YACH,UAAU,GAAG,EAAE,CAAC,YAAY,CAAC,KAAK,CAAC,QAAQ,CAAC,CAAC;QAC/C,CAAC;QAAC,OAAO,KAAK,EAAE,CAAC;YACf,MAAM,IAAI,QAAQ,CAChB,eAAe,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,QAAQ,CAAC,EAAE,EAC9C,eAAe,CAAC,iCAAiC,EACjD,EAAE,KAAK,EAAE,KAAK,EAAE,CACjB,CAAC;QACJ,CAAC;QAED,MAAM,OAAO,GAAG,cAAc,CAAC,UAAU,EAAE,MAAM,CAAC,CAAC;QACnD,MAAM,eAAe,GAAG,WAAW,CAAC,OAAO,EAAE,sBAAsB,EAAE,MAAM,CAAC,CAAC;QAC7E,MAAM,eAAe,GAAG,WAAW,CAAC,OAAO,EAAE,iCAAiC,EAAE,MAAM,CAAC,CAAC;QACxF,MAAM,UAAU,GAAG,2BAA2B,CAAC,eAAe,EAAE,eAAe,CAAC,CAAC;QACjF,IAAI,UAAU,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YAC5B,MAAM,IAAI,QAAQ,CAChB,oBAAoB,EACpB,eAAe,CAAC,iCAAiC,CAClD,CAAC;QACJ,CAAC;QAED,MAAM,MAAM,GAAG,WAAW,CAAC,OAAO,EAAE,UAAU,CAAC,CAAC;QAChD,IAAI,MAAM,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YACxB,MAAM,IAAI,QAAQ,CAChB,uBAAuB,EACvB,eAAe,CAAC,iCAAiC,CAClD,CAAC;QACJ,CAAC;QAED,MAAM,MAAM,GAAsB,MAAM,CAAC,GAAG,CAAC,KAAK,CAAC,EAAE,CAAC,CAAC;YACrD,IAAI,EAAE,OAAO;YACb,UAAU,EAAE,KAAK,CAAC,UAAU;YAC5B,IAAI,EAAE,KAAK,CAAC,IAAI;YAChB,QAAQ,EAAE;gBACR,UAAU,EAAE,KAAK,CAAC,UAAU;aAC7B;SACF,CAAC,CAAC,CAAC;QAEJ,MAAM,IAAI,GAAG,MAAM;aAChB,GAAG,CAAC,KAAK,CAAC,EAAE,CAAC,SAAS,KAAK,CAAC,UAAU,KAAK,KAAK,CAAC,IAAI,EAAE,CAAC,IAAI,EAAE,CAAC;aAC/D,IAAI,CAAC,MAAM,CAAC;aACZ,IAAI,EAAE,CAAC;QACV,MAAM,KAAK,GAAG,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,QAAQ,EAAE,KAAK,CAAC,SAAS,CAAC,CAAC;QAE7D,OAAO;YACL,KAAK;YACL,IAAI;YACJ,OAAO,EAAE,YAAY,CAAC,IAAI,CAAC;YAC3B,MAAM,EAAE,MAAM;YACd,QAAQ,EAAE;gBACR,OAAO,EAAE,IAAI,CAAC,IAAI;gBAClB,UAAU,EAAE,UAAU,CAAC,MAAM;gBAC7B,mBAAmB,EAAE,MAAM,CAAC,MAAM;aACnC;YACD,UAAU,EAAE;gBACV,MAAM;gBACN,KAAK,EAAE;oBACL,UAAU,EAAE,UAAU,CAAC,MAAM;oBAC7B,mBAAmB,EAAE,MAAM,CAAC,MAAM;iBACnC;aACF;SACF,CAAC;IACJ,CAAC;CACF"}

package/dist/server/modules/affairs-indexer/core/src/parser/xlsx-parser-adapter.d.ts

@@ -0,0 +1,7 @@
+import { BaseComplexParserAdapter } from "./base-complex-parser-adapter.js";
+import type { ParseInput, ParsedDocumentPayload } from "./parser-adapter.js";
+export declare class XlsxParserAdapter extends BaseComplexParserAdapter {
+    readonly name = "xlsx_parser";
+    supports(ext: string): boolean;
+    protected parseComplex(input: ParseInput): Promise<ParsedDocumentPayload>;
+}

package/dist/server/modules/affairs-indexer/core/src/parser/xlsx-parser-adapter.js

@@ -0,0 +1,228 @@
+import fs from "node:fs";
+import path from "node:path";
+import { AppError, APP_ERROR_CODES } from "../../../contracts/src/index.js";
+import { BaseComplexParserAdapter } from "./base-complex-parser-adapter.js";
+import { decodeXmlEntities, extractXmlAttributes, readZipEntries, readZipText, shortSummary, } from "./openxml-utils.js";
+const SUPPORTED_XLSX_EXTENSIONS = new Set([".xlsx"]);
+function resolveWorkbookTarget(target) {
+    const normalized = path.posix.normalize(path.posix.join("xl", target));
+    return normalized.replace(/^\/+/, "");
+}
+function parseWorkbookRelationships(xml) {
+    const relations = new Map();
+    const relationshipPattern = /<Relationship\b([^>]*)\/>/g;
+    for (const match of xml.matchAll(relationshipPattern)) {
+        const attributes = extractXmlAttributes(match[1]);
+        const relationId = attributes.Id;
+        const target = attributes.Target;
+        if (!relationId || !target) {
+            continue;
+        }
+        relations.set(relationId, resolveWorkbookTarget(target));
+    }
+    return relations;
+}
+function parseWorkbookSheets(workbookXml, relationshipXml) {
+    const relationships = parseWorkbookRelationships(relationshipXml);
+    const sheets = [];
+    const sheetPattern = /<sheet\b([^>]*)\/>/g;
+    for (const match of workbookXml.matchAll(sheetPattern)) {
+        const attributes = extractXmlAttributes(match[1]);
+        const relationId = attributes["r:id"];
+        const name = attributes.name?.trim();
+        if (!relationId || !name) {
+            continue;
+        }
+        const target = relationships.get(relationId);
+        if (!target) {
+            continue;
+        }
+        sheets.push({
+            name,
+            target,
+        });
+    }
+    if (sheets.length === 0) {
+        throw new AppError("XLSX 中未发现可读取的 sheet", APP_ERROR_CODES.PARSER_COMPLEX_CONTENT_UNREADABLE);
+    }
+    return sheets;
+}
+function parseSharedStrings(xml) {
+    if (!xml) {
+        return [];
+    }
+    const items = [];
+    const itemPattern = /<si\b[^>]*>([\s\S]*?)<\/si>/g;
+    for (const match of xml.matchAll(itemPattern)) {
+        const rawItem = match[1];
+        const textParts = [...rawItem.matchAll(/<t\b[^>]*>([\s\S]*?)<\/t>/g)]
+            .map(item => decodeXmlEntities(item[1]));
+        items.push(textParts.join(""));
+    }
+    return items;
+}
+function columnLettersToIndex(letters) {
+    let value = 0;
+    for (const char of letters.toUpperCase()) {
+        value = (value * 26) + (char.charCodeAt(0) - 64);
+    }
+    return Math.max(0, value - 1);
+}
+function getCellColumnIndex(cellRef, fallbackIndex) {
+    if (!cellRef) {
+        return fallbackIndex;
+    }
+    const match = cellRef.match(/[A-Z]+/i);
+    if (!match) {
+        return fallbackIndex;
+    }
+    return columnLettersToIndex(match[0]);
+}
+function readCellValue(rawCellXml, type, sharedStrings) {
+    if (type === "inlineStr") {
+        const inlineTexts = [...rawCellXml.matchAll(/<t\b[^>]*>([\s\S]*?)<\/t>/g)]
+            .map(match => decodeXmlEntities(match[1]));
+        return inlineTexts.join("").trim();
+    }
+    const rawValue = rawCellXml.match(/<v\b[^>]*>([\s\S]*?)<\/v>/)?.[1];
+    if (type === "s") {
+        const index = Number(rawValue ?? "-1");
+        return sharedStrings[index] ?? "";
+    }
+    if (type === "b") {
+        return rawValue === "1" ? "TRUE" : "FALSE";
+    }
+    if (type === "str" || type === "e") {
+        return decodeXmlEntities(rawValue ?? "").trim();
+    }
+    if (rawValue && rawValue.length > 0) {
+        return decodeXmlEntities(rawValue).trim();
+    }
+    const formulaValue = rawCellXml.match(/<f\b[^>]*>([\s\S]*?)<\/f>/)?.[1];
+    return formulaValue ? `=${decodeXmlEntities(formulaValue).trim()}` : "";
+}
+function trimTrailingEmptyCells(row) {
+    const result = [...row];
+    while (result.length > 0 && result[result.length - 1] === "") {
+        result.pop();
+    }
+    return result;
+}
+function parseWorksheet(sheetName, xml, sharedStrings) {
+    const rows = [];
+    const rowPattern = /<row\b[^>]*>([\s\S]*?)<\/row>/g;
+    for (const rowMatch of xml.matchAll(rowPattern)) {
+        const rawRow = rowMatch[1];
+        const row = [];
+        let fallbackColumnIndex = 0;
+        const cellPattern = /<c\b([^>]*)>([\s\S]*?)<\/c>|<c\b([^>]*)\/>/g;
+        for (const cellMatch of rawRow.matchAll(cellPattern)) {
+            const rawAttributes = cellMatch[1] ?? cellMatch[3] ?? "";
+            const rawCellXml = cellMatch[2] ?? "";
+            const attributes = extractXmlAttributes(rawAttributes);
+            const columnIndex = getCellColumnIndex(attributes.r, fallbackColumnIndex);
+            const cellValue = readCellValue(rawCellXml, attributes.t, sharedStrings);
+            while (row.length < columnIndex) {
+                row.push("");
+            }
+            row[columnIndex] = cellValue;
+            fallbackColumnIndex = columnIndex + 1;
+        }
+        const normalizedRow = trimTrailingEmptyCells(row);
+        if (normalizedRow.length > 0) {
+            rows.push(normalizedRow);
+        }
+    }
+    const text = rows.map(row => row.filter(cell => cell.length > 0).join(" ")).filter(Boolean).join("\n");
+    const columnCount = rows.reduce((max, row) => Math.max(max, row.length), 0);
+    return {
+        sheetName,
+        rows,
+        text,
+        rowCount: rows.length,
+        columnCount,
+    };
+}
+export class XlsxParserAdapter extends BaseComplexParserAdapter {
+    name = "xlsx_parser";
+    supports(ext) {
+        return SUPPORTED_XLSX_EXTENSIONS.has(ext.toLowerCase());
+    }
+    async parseComplex(input) {
+        let fileBuffer;
+        try {
+            fileBuffer = fs.readFileSync(input.filePath);
+        }
+        catch (error) {
+            throw new AppError(`XLSX 文件读取失败：${path.basename(input.filePath)}`, APP_ERROR_CODES.PARSER_COMPLEX_CONTENT_UNREADABLE, { cause: error });
+        }
+        const entries = readZipEntries(fileBuffer, "XLSX");
+        const workbookXml = readZipText(entries, "xl/workbook.xml", "XLSX");
+        const relationshipXml = readZipText(entries, "xl/_rels/workbook.xml.rels", "XLSX");
+        const sharedStrings = parseSharedStrings(readZipText(entries, "xl/sharedStrings.xml", "XLSX", false));
+        const workbookSheets = parseWorkbookSheets(workbookXml, relationshipXml);
+        const parsedSheets = workbookSheets
+            .map(sheet => ({
+            sheet,
+            worksheetXml: readZipText(entries, sheet.target, "XLSX", false),
+        }))
+            .filter(item => item.worksheetXml)
+            .map(item => parseWorksheet(item.sheet.name, item.worksheetXml ?? "", sharedStrings))
+            .filter(item => item.rowCount > 0 || item.text.length > 0);
+        if (parsedSheets.length === 0) {
+            throw new AppError("XLSX 中没有可读取的 sheet 内容", APP_ERROR_CODES.PARSER_COMPLEX_CONTENT_UNREADABLE);
+        }
+        const blocks = [];
+        for (const sheet of parsedSheets) {
+            blocks.push({
+                kind: "sheet",
+                sheetName: sheet.sheetName,
+                text: sheet.text,
+                metadata: {
+                    rowCount: sheet.rowCount,
+                    columnCount: sheet.columnCount,
+                },
+            });
+            blocks.push({
+                kind: "table",
+                sheetName: sheet.sheetName,
+                cells: sheet.rows,
+                metadata: {
+                    rowCount: sheet.rowCount,
+                    columnCount: sheet.columnCount,
+                },
+            });
+        }
+        const text = parsedSheets
+            .map(sheet => `${sheet.sheetName}\n${sheet.text}`.trim())
+            .join("\n\n")
+            .trim();
+        if (!text) {
+            throw new AppError("XLSX 文本内容为空", APP_ERROR_CODES.PARSER_COMPLEX_CONTENT_UNREADABLE);
+        }
+        const totalRows = parsedSheets.reduce((sum, sheet) => sum + sheet.rowCount, 0);
+        const maxColumnCount = parsedSheets.reduce((max, sheet) => Math.max(max, sheet.columnCount), 0);
+        const title = path.basename(input.filePath, input.extension);
+        return {
+            title,
+            text,
+            summary: shortSummary(text),
+            parser: "xlsx",
+            metadata: {
+                adapter: this.name,
+                sheetCount: parsedSheets.length,
+                rowCount: totalRows,
+                columnCount: maxColumnCount,
+            },
+            structured: {
+                blocks,
+                stats: {
+                    sheetCount: parsedSheets.length,
+                    rowCount: totalRows,
+                    columnCount: maxColumnCount,
+                },
+            },
+        };
+    }
+}
+//# sourceMappingURL=xlsx-parser-adapter.js.map

package/dist/server/modules/affairs-indexer/core/src/parser/xlsx-parser-adapter.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"xlsx-parser-adapter.js","sourceRoot":"","sources":["../../../../../../../src/modules/affairs-indexer/core/src/parser/xlsx-parser-adapter.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,MAAM,SAAS,CAAC;AACzB,OAAO,IAAI,MAAM,WAAW,CAAC;AAC7B,OAAO,EAAE,QAAQ,EAAE,eAAe,EAAE,MAAM,iCAAiC,CAAC;AAC5E,OAAO,EAAE,wBAAwB,EAAE,MAAM,kCAAkC,CAAC;AAC5E,OAAO,EACL,iBAAiB,EACjB,oBAAoB,EACpB,cAAc,EACd,WAAW,EACX,YAAY,GACb,MAAM,oBAAoB,CAAC;AAG5B,MAAM,yBAAyB,GAAG,IAAI,GAAG,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC;AAerD,SAAS,qBAAqB,CAAC,MAAc;IAC3C,MAAM,UAAU,GAAG,IAAI,CAAC,KAAK,CAAC,SAAS,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,IAAI,EAAE,MAAM,CAAC,CAAC,CAAC;IACvE,OAAO,UAAU,CAAC,OAAO,CAAC,MAAM,EAAE,EAAE,CAAC,CAAC;AACxC,CAAC;AAED,SAAS,0BAA0B,CAAC,GAAW;IAC7C,MAAM,SAAS,GAAG,IAAI,GAAG,EAAkB,CAAC;IAC5C,MAAM,mBAAmB,GAAG,4BAA4B,CAAC;IACzD,KAAK,MAAM,KAAK,IAAI,GAAG,CAAC,QAAQ,CAAC,mBAAmB,CAAC,EAAE,CAAC;QACtD,MAAM,UAAU,GAAG,oBAAoB,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC;QAClD,MAAM,UAAU,GAAG,UAAU,CAAC,EAAE,CAAC;QACjC,MAAM,MAAM,GAAG,UAAU,CAAC,MAAM,CAAC;QACjC,IAAI,CAAC,UAAU,IAAI,CAAC,MAAM,EAAE,CAAC;YAC3B,SAAS;QACX,CAAC;QACD,SAAS,CAAC,GAAG,CAAC,UAAU,EAAE,qBAAqB,CAAC,MAAM,CAAC,CAAC,CAAC;IAC3D,CAAC;IACD,OAAO,SAAS,CAAC;AACnB,CAAC;AAED,SAAS,mBAAmB,CAAC,WAAmB,EAAE,eAAuB;IACvE,MAAM,aAAa,GAAG,0BAA0B,CAAC,eAAe,CAAC,CAAC;IAClE,MAAM,MAAM,GAAoB,EAAE,CAAC;IACnC,MAAM,YAAY,GAAG,qBAAqB,CAAC;IAC3C,KAAK,MAAM,KAAK,IAAI,WAAW,CAAC,QAAQ,CAAC,YAAY,CAAC,EAAE,CAAC;QACvD,MAAM,UAAU,GAAG,oBAAoB,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC;QAClD,MAAM,UAAU,GAAG,UAAU,CAAC,MAAM,CAAC,CAAC;QACtC,MAAM,IAAI,GAAG,UAAU,CAAC,IAAI,EAAE,IAAI,EAAE,CAAC;QACrC,IAAI,CAAC,UAAU,IAAI,CAAC,IAAI,EAAE,CAAC;YACzB,SAAS;QACX,CAAC;QACD,MAAM,MAAM,GAAG,aAAa,CAAC,GAAG,CAAC,UAAU,CAAC,CAAC;QAC7C,IAAI,CAAC,MAAM,EAAE,CAAC;YACZ,SAAS;QACX,CAAC;QACD,MAAM,CAAC,IAAI,CAAC;YACV,IAAI;YACJ,MAAM;SACP,CAAC,CAAC;IACL,CAAC;IAED,IAAI,MAAM,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QACxB,MAAM,IAAI,QAAQ,CAChB,qBAAqB,EACrB,eAAe,CAAC,iCAAiC,CAClD,CAAC;IACJ,CAAC;IAED,OAAO,MAAM,CAAC;AAChB,CAAC;AAED,SAAS,kBAAkB,CAAC,GAAkB;IAC5C,IAAI,CAAC,GAAG,EAAE,CAAC;QACT,OAAO,EAAE,CAAC;IACZ,CAAC;IAED,MAAM,KAAK,GAAa,EAAE,CAAC;IAC3B,MAAM,WAAW,GAAG,8BAA8B,CAAC;IACnD,KAAK,MAAM,KAAK,IAAI,GAAG,CAAC,QAAQ,CAAC,WAAW,CAAC,EAAE,CAAC;QAC9C,MAAM,OAAO,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC;QACzB,MAAM,SAAS,GAAG,CAAC,GAAG,OAAO,CAAC,QAAQ,CAAC,4BAA4B,CAAC,CAAC;aAClE,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC,iBAAiB,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;QAC3C,KAAK,CAAC,IAAI,CAAC,SAAS,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC,CAAC;IACjC,CAAC;IACD,OAAO,KAAK,CAAC;AACf,CAAC;AAED,SAAS,oBAAoB,CAAC,OAAe;IAC3C,IAAI,KAAK,GAAG,CAAC,CAAC;IACd,KAAK,MAAM,IAAI,IAAI,OAAO,CAAC,WAAW,EAAE,EAAE,CAAC;QACzC,KAAK,GAAG,CAAC,KAAK,GAAG,EAAE,CAAC,GAAG,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC,CAAC,GAAG,EAAE,CAAC,CAAC;IACnD,CAAC;IACD,OAAO,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,KAAK,GAAG,CAAC,CAAC,CAAC;AAChC,CAAC;AAED,SAAS,kBAAkB,CAAC,OAA2B,EAAE,aAAqB;IAC5E,IAAI,CAAC,OAAO,EAAE,CAAC;QACb,OAAO,aAAa,CAAC;IACvB,CAAC;IACD,MAAM,KAAK,GAAG,OAAO,CAAC,KAAK,CAAC,SAAS,CAAC,CAAC;IACvC,IAAI,CAAC,KAAK,EAAE,CAAC;QACX,OAAO,aAAa,CAAC;IACvB,CAAC;IACD,OAAO,oBAAoB,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC;AACxC,CAAC;AAED,SAAS,aAAa,CAAC,UAAkB,EAAE,IAAwB,EAAE,aAAuB;IAC1F,IAAI,IAAI,KAAK,WAAW,EAAE,CAAC;QACzB,MAAM,WAAW,GAAG,CAAC,GAAG,UAAU,CAAC,QAAQ,CAAC,4BAA4B,CAAC,CAAC;aACvE,GAAG,CAAC,KAAK,CAAC,EAAE,CAAC,iBAAiB,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;QAC7C,OAAO,WAAW,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC;IACrC,CAAC;IAED,MAAM,QAAQ,GAAG,UAAU,CAAC,KAAK,CAAC,2BAA2B,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC;IACpE,IAAI,IAAI,KAAK,GAAG,EAAE,CAAC;QACjB,MAAM,KAAK,GAAG,MAAM,CAAC,QAAQ,IAAI,IAAI,CAAC,CAAC;QACvC,OAAO,aAAa,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC;IACpC,CAAC;IACD,IAAI,IAAI,KAAK,GAAG,EAAE,CAAC;QACjB,OAAO,QAAQ,KAAK,GAAG,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,OAAO,CAAC;IAC7C,CAAC;IACD,IAAI,IAAI,KAAK,KAAK,IAAI,IAAI,KAAK,GAAG,EAAE,CAAC;QACnC,OAAO,iBAAiB,CAAC,QAAQ,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC;IAClD,CAAC;IACD,IAAI,QAAQ,IAAI,QAAQ,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QACpC,OAAO,iBAAiB,CAAC,QAAQ,CAAC,CAAC,IAAI,EAAE,CAAC;IAC5C,CAAC;IAED,MAAM,YAAY,GAAG,UAAU,CAAC,KAAK,CAAC,2BAA2B,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC;IACxE,OAAO,YAAY,CAAC,CAAC,CAAC,IAAI,iBAAiB,CAAC,YAAY,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;AAC1E,CAAC;AAED,SAAS,sBAAsB,CAAC,GAAa;IAC3C,MAAM,MAAM,GAAG,CAAC,GAAG,GAAG,CAAC,CAAC;IACxB,OAAO,MAAM,CAAC,MAAM,GAAG,CAAC,IAAI,MAAM,CAAC,MAAM,CAAC,MAAM,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC;QAC7D,MAAM,CAAC,GAAG,EAAE,CAAC;IACf,CAAC;IACD,OAAO,MAAM,CAAC;AAChB,CAAC;AAED,SAAS,cAAc,CAAC,SAAiB,EAAE,GAAW,EAAE,aAAuB;IAC7E,MAAM,IAAI,GAAe,EAAE,CAAC;IAC5B,MAAM,UAAU,GAAG,gCAAgC,CAAC;IAEpD,KAAK,MAAM,QAAQ,IAAI,GAAG,CAAC,QAAQ,CAAC,UAAU,CAAC,EAAE,CAAC;QAChD,MAAM,MAAM,GAAG,QAAQ,CAAC,CAAC,CAAC,CAAC;QAC3B,MAAM,GAAG,GAAa,EAAE,CAAC;QACzB,IAAI,mBAAmB,GAAG,CAAC,CAAC;QAC5B,MAAM,WAAW,GAAG,6CAA6C,CAAC;QAElE,KAAK,MAAM,SAAS,IAAI,MAAM,CAAC,QAAQ,CAAC,WAAW,CAAC,EAAE,CAAC;YACrD,MAAM,aAAa,GAAG,SAAS,CAAC,CAAC,CAAC,IAAI,SAAS,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;YACzD,MAAM,UAAU,GAAG,SAAS,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;YACtC,MAAM,UAAU,GAAG,oBAAoB,CAAC,aAAa,CAAC,CAAC;YACvD,MAAM,WAAW,GAAG,kBAAkB,CAAC,UAAU,CAAC,CAAC,EAAE,mBAAmB,CAAC,CAAC;YAC1E,MAAM,SAAS,GAAG,aAAa,CAAC,UAAU,EAAE,UAAU,CAAC,CAAC,EAAE,aAAa,CAAC,CAAC;YACzE,OAAO,GAAG,CAAC,MAAM,GAAG,WAAW,EAAE,CAAC;gBAChC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;YACf,CAAC;YACD,GAAG,CAAC,WAAW,CAAC,GAAG,SAAS,CAAC;YAC7B,mBAAmB,GAAG,WAAW,GAAG,CAAC,CAAC;QACxC,CAAC;QAED,MAAM,aAAa,GAAG,sBAAsB,CAAC,GAAG,CAAC,CAAC;QAClD,IAAI,aAAa,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YAC7B,IAAI,CAAC,IAAI,CAAC,aAAa,CAAC,CAAC;QAC3B,CAAC;IACH,CAAC;IAED,MAAM,IAAI,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,MAAM,CAAC,IAAI,CAAC,EAAE,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;IACvG,MAAM,WAAW,GAAG,IAAI,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,GAAG,EAAE,EAAE,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,EAAE,GAAG,CAAC,MAAM,CAAC,EAAE,CAAC,CAAC,CAAC;IAE5E,OAAO;QACL,SAAS;QACT,IAAI;QACJ,IAAI;QACJ,QAAQ,EAAE,IAAI,CAAC,MAAM;QACrB,WAAW;KACZ,CAAC;AACJ,CAAC;AAED,MAAM,OAAO,iBAAkB,SAAQ,wBAAwB;IACpD,IAAI,GAAG,aAAa,CAAC;IAE9B,QAAQ,CAAC,GAAW;QAClB,OAAO,yBAAyB,CAAC,GAAG,CAAC,GAAG,CAAC,WAAW,EAAE,CAAC,CAAC;IAC1D,CAAC;IAES,KAAK,CAAC,YAAY,CAAC,KAAiB;QAC5C,IAAI,UAAkB,CAAC;QACvB,IAAI,CAAC;YACH,UAAU,GAAG,EAAE,CAAC,YAAY,CAAC,KAAK,CAAC,QAAQ,CAAC,CAAC;QAC/C,CAAC;QAAC,OAAO,KAAK,EAAE,CAAC;YACf,MAAM,IAAI,QAAQ,CAChB,eAAe,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,QAAQ,CAAC,EAAE,EAC9C,eAAe,CAAC,iCAAiC,EACjD,EAAE,KAAK,EAAE,KAAK,EAAE,CACjB,CAAC;QACJ,CAAC;QAED,MAAM,OAAO,GAAG,cAAc,CAAC,UAAU,EAAE,MAAM,CAAC,CAAC;QACnD,MAAM,WAAW,GAAG,WAAW,CAAC,OAAO,EAAE,iBAAiB,EAAE,MAAM,CAAC,CAAC;QACpE,MAAM,eAAe,GAAG,WAAW,CAAC,OAAO,EAAE,4BAA4B,EAAE,MAAM,CAAC,CAAC;QACnF,MAAM,aAAa,GAAG,kBAAkB,CAAC,WAAW,CAAC,OAAO,EAAE,sBAAsB,EAAE,MAAM,EAAE,KAAK,CAAC,CAAC,CAAC;QACtG,MAAM,cAAc,GAAG,mBAAmB,CAAC,WAAW,EAAE,eAAe,CAAC,CAAC;QACzE,MAAM,YAAY,GAAG,cAAc;aAChC,GAAG,CAAC,KAAK,CAAC,EAAE,CAAC,CAAC;YACb,KAAK;YACL,YAAY,EAAE,WAAW,CAAC,OAAO,EAAE,KAAK,CAAC,MAAM,EAAE,MAAM,EAAE,KAAK,CAAC;SAChE,CAAC,CAAC;aACF,MAAM,CAAC,IAAI,CAAC,EAAE,CAAC,IAAI,CAAC,YAAY,CAAC;aACjC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC,cAAc,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,EAAE,IAAI,CAAC,YAAY,IAAI,EAAE,EAAE,aAAa,CAAC,CAAC;aACpF,MAAM,CAAC,IAAI,CAAC,EAAE,CAAC,IAAI,CAAC,QAAQ,GAAG,CAAC,IAAI,IAAI,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;QAE7D,IAAI,YAAY,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YAC9B,MAAM,IAAI,QAAQ,CAChB,uBAAuB,EACvB,eAAe,CAAC,iCAAiC,CAClD,CAAC;QACJ,CAAC;QAED,MAAM,MAAM,GAAsB,EAAE,CAAC;QACrC,KAAK,MAAM,KAAK,IAAI,YAAY,EAAE,CAAC;YACjC,MAAM,CAAC,IAAI,CAAC;gBACV,IAAI,EAAE,OAAO;gBACb,SAAS,EAAE,KAAK,CAAC,SAAS;gBAC1B,IAAI,EAAE,KAAK,CAAC,IAAI;gBAChB,QAAQ,EAAE;oBACR,QAAQ,EAAE,KAAK,CAAC,QAAQ;oBACxB,WAAW,EAAE,KAAK,CAAC,WAAW;iBAC/B;aACF,CAAC,CAAC;YACH,MAAM,CAAC,IAAI,CAAC;gBACV,IAAI,EAAE,OAAO;gBACb,SAAS,EAAE,KAAK,CAAC,SAAS;gBAC1B,KAAK,EAAE,KAAK,CAAC,IAAI;gBACjB,QAAQ,EAAE;oBACR,QAAQ,EAAE,KAAK,CAAC,QAAQ;oBACxB,WAAW,EAAE,KAAK,CAAC,WAAW;iBAC/B;aACF,CAAC,CAAC;QACL,CAAC;QAED,MAAM,IAAI,GAAG,YAAY;aACtB,GAAG,CAAC,KAAK,CAAC,EAAE,CAAC,GAAG,KAAK,CAAC,SAAS,KAAK,KAAK,CAAC,IAAI,EAAE,CAAC,IAAI,EAAE,CAAC;aACxD,IAAI,CAAC,MAAM,CAAC;aACZ,IAAI,EAAE,CAAC;QAEV,IAAI,CAAC,IAAI,EAAE,CAAC;YACV,MAAM,IAAI,QAAQ,CAChB,aAAa,EACb,eAAe,CAAC,iCAAiC,CAClD,CAAC;QACJ,CAAC;QAED,MAAM,SAAS,GAAG,YAAY,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,KAAK,EAAE,EAAE,CAAC,GAAG,GAAG,KAAK,CAAC,QAAQ,EAAE,CAAC,CAAC,CAAC;QAC/E,MAAM,cAAc,GAAG,YAAY,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,KAAK,EAAE,EAAE,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,EAAE,KAAK,CAAC,WAAW,CAAC,EAAE,CAAC,CAAC,CAAC;QAChG,MAAM,KAAK,GAAG,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,QAAQ,EAAE,KAAK,CAAC,SAAS,CAAC,CAAC;QAE7D,OAAO;YACL,KAAK;YACL,IAAI;YACJ,OAAO,EAAE,YAAY,CAAC,IAAI,CAAC;YAC3B,MAAM,EAAE,MAAM;YACd,QAAQ,EAAE;gBACR,OAAO,EAAE,IAAI,CAAC,IAAI;gBAClB,UAAU,EAAE,YAAY,CAAC,MAAM;gBAC/B,QAAQ,EAAE,SAAS;gBACnB,WAAW,EAAE,cAAc;aAC5B;YACD,UAAU,EAAE;gBACV,MAAM;gBACN,KAAK,EAAE;oBACL,UAAU,EAAE,YAAY,CAAC,MAAM;oBAC/B,QAAQ,EAAE,SAAS;oBACnB,WAAW,EAAE,cAAc;iBAC5B;aACF;SACF,CAAC;IACJ,CAAC;CACF"}

package/dist/server/modules/affairs-indexer/core/src/repositories/catalog-repository.d.ts

@@ -0,0 +1,205 @@
+import { type OpenDatabaseOptions } from "../sqlite/open-database.js";
+export interface DocumentContextResult {
+    documentId: string;
+    path: string;
+    title: string;
+    summary: string;
+    modifiedAt: string;
+    inodeKey: string | null;
+    contentHash: string | null;
+    size: number;
+    extension: string;
+    tags: string[];
+}
+export interface BrowseTagNodeResult {
+    path: string;
+    name: string;
+    rootType: string;
+    parentPath: string | null;
+    depth: number;
+}
+export interface SearchDocumentResult {
+    documentId: string;
+    path: string;
+    title: string;
+    summary: string;
+    modifiedAt: string;
+}
+export interface ExportDocumentRecord {
+    documentId: string;
+    path: string;
+    title: string;
+    summary: string;
+    tags: string[];
+    derivedTags: string[];
+    mtime: string;
+}
+export interface ExportTagRecord {
+    path: string;
+    name: string;
+    rootType: string;
+    parentPath: string | null;
+    depth: number;
+}
+export interface ExportDocumentRow {
+    documentId: string;
+    path: string;
+    title: string;
+    summary: string;
+    mtime: string;
+}
+export interface ExportDocumentTagRow {
+    documentId: string;
+    tagPath: string;
+    derived: boolean;
+}
+export interface TagRecomputeDocumentRow {
+    documentId: string;
+    path: string;
+    title: string;
+    summary: string;
+    contentText: string;
+    mtime: string;
+    ctime: string;
+    extension: string;
+}
+export type TagResolvedSourceType = "manual_document" | "folder_binding" | "smart_rule" | "system_derived";
+export type TagRuleRelation = "and" | "or" | "not";
+export type TagRuleType = "file_name_contains" | "file_content_contains" | "file_extension_in" | "modified_time_between" | "document_path_in_folder";
+export interface TagRuleMatcherFileNameContains {
+    keyword: string;
+}
+export interface TagRuleMatcherFileContentContains {
+    keyword: string;
+}
+export interface TagRuleMatcherFileExtensionIn {
+    extensions: string[];
+}
+export interface TagRuleMatcherModifiedTimeBetween {
+    start?: string | null;
+    end?: string | null;
+}
+export interface TagRuleMatcherDocumentPathInFolder {
+    folderPath?: string | null;
+}
+export type TagRuleMatcher = TagRuleMatcherFileNameContains | TagRuleMatcherFileContentContains | TagRuleMatcherFileExtensionIn | TagRuleMatcherModifiedTimeBetween | TagRuleMatcherDocumentPathInFolder;
+export interface TagRuleRow {
+    id: string;
+    tagId: string;
+    tagPath: string;
+    enabled: boolean;
+    relation: TagRuleRelation;
+    ruleType: TagRuleType;
+    matcher: TagRuleMatcher;
+    minScore: number | null;
+    priority: number;
+    source: string;
+    createdAt: string;
+    updatedAt: string;
+}
+export interface ManualDocumentTagBindingRow {
+    id: string;
+    documentId: string;
+    tagId: string;
+    tagPath: string;
+    source: string;
+    createdAt: string;
+    updatedAt: string;
+}
+export interface ManualTagBindingStats {
+    identityBindingCount: number;
+    legacyBindingCount: number;
+    legacyFallbackBindingCount: number;
+    legacyFallbackDocumentCount: number;
+}
+export interface FolderTagBindingRow {
+    id: string;
+    folderPath: string;
+    tagId: string;
+    tagPath: string;
+    applyMode: string;
+    createdAt: string;
+    updatedAt: string;
+}
+export interface EffectiveFolderTagBindingRow extends FolderTagBindingRow {
+    documentPath: string;
+    documentId: string;
+}
+export interface TagDefinitionRow {
+    id: string;
+    rootType: string;
+    path: string;
+    name: string;
+    parentId: string | null;
+    canonicalName: string;
+    description: string | null;
+    status: string;
+    createdBy: string;
+    createdAt: string;
+    updatedAt: string;
+    disabledAt: string | null;
+}
+export interface ResolvedDocumentTagRow {
+    documentId: string;
+    path: string;
+    tagId: string;
+    sourceType: TagResolvedSourceType;
+    sourceRef: string | null;
+    evidence: string | null;
+    confidence: number;
+    updatedAt: string;
+}
+export interface RecomputeScope {
+    kind: "full" | "document" | "folder" | "tag";
+    documentId?: string;
+    folderPath?: string;
+    tagId?: string;
+    mode?: "full" | "folder_bindings_only";
+}
+export interface ExportTagPostingRow {
+    rootType: string;
+    tagPath: string;
+    documentId: string;
+    path: string;
+    title: string;
+    derived: boolean;
+}
+/**
+ * 最小 SQLite 查询仓库。
+ * 第二阶段继续承接只读查询，并补最小增量导出所需查询。
+ */
+export declare class CatalogRepository {
+    private readonly dbPath;
+    private readonly dbOptions;
+    constructor(dbPath: string, dbOptions?: OpenDatabaseOptions);
+    listTagDefinitions(includeDisabled?: boolean): TagDefinitionRow[];
+    getTagDefinitionById(tagId: string): TagDefinitionRow | null;
+    listTagRulesByTagIds(tagIds: string[]): TagRuleRow[];
+    listAllEnabledTagRules(): TagRuleRow[];
+    listManualDocumentTagBindingsByDocumentIds(documentIds: string[]): ManualDocumentTagBindingRow[];
+    getManualTagBindingStats(): ManualTagBindingStats;
+    private resolveManualFileTagBindingsForDocuments;
+    private listActiveDocumentIdentitiesByKeys;
+    listFolderTagBindingsByPaths(paths: string[]): FolderTagBindingRow[];
+    listEffectiveFolderTagBindingsForDocumentPaths(paths: string[]): EffectiveFolderTagBindingRow[];
+    listEffectiveFolderTagBindingsForFolderScope(folderPath: string): EffectiveFolderTagBindingRow[];
+    listResolvedDocumentTagsByDocumentIds(documentIds: string[]): ResolvedDocumentTagRow[];
+    listRecomputeCandidateDocuments(scope: RecomputeScope): TagRecomputeDocumentRow[];
+    getDocumentContext(documentId?: string, filePath?: string): DocumentContextResult | null;
+    browseTags(rootType?: string, parentPath?: string): BrowseTagNodeResult[];
+    searchDocuments(query: string, limit?: number): SearchDocumentResult[];
+    listExportDocumentsByDocumentIds(documentIds: string[]): ExportDocumentRecord[];
+    listExportDocuments(): ExportDocumentRecord[];
+    listActiveFileExtensions(): string[];
+    listExportDocumentsByPaths(paths: string[]): ExportDocumentRecord[];
+    listExportDocumentsByExtensions(extensions: string[]): ExportDocumentRecord[];
+    listExportTags(): ExportTagRecord[];
+    iterateExportDocuments(batchSize?: number): Generator<ExportDocumentRow[]>;
+    iterateExportDocumentRecords(batchSize?: number): Generator<ExportDocumentRecord[]>;
+    iterateDocumentTagRows(batchSize?: number): Generator<ExportDocumentTagRow[]>;
+    iterateTagPostingRows(batchSize?: number): Generator<ExportTagPostingRow[]>;
+    iterateDirectTagPostingRows(batchSize?: number): Generator<ExportTagPostingRow[]>;
+    iterateDerivedTagPostingRows(batchSize?: number): Generator<ExportTagPostingRow[]>;
+    iterateTagRecomputeDocuments(batchSize?: number): Generator<TagRecomputeDocumentRow[]>;
+    private listAllRecomputeCandidateDocuments;
+}