tm-extractor 0.0.1

package/dist/core/data-transformer.d.ts

@@ -0,0 +1,44 @@
+import { TalentData, Talent, CleanedTalent, CleanedTalentData, SinglePersonResult, TmaExtractorResult, ScmData, CleanedScmData } from '../types/tma-types';
+/**
+ * Data transformation class that implements filteredExtractedData logic
+ * Removes unnecessary fields and metadata from extracted TMA data
+ */
+export declare class DataTransformer {
+    /**
+     * Removes the 'sign' field from talents array
+     */
+    static removeTalentSign(talents: Talent[]): CleanedTalent[];
+    /**
+     * Removes SCM metadata while keeping activity data
+     * Filters out detectionConfidence and metadata according to filtered output spec
+     */
+    static removeScmMetadata(scmData: ScmData): CleanedScmData;
+    /**
+     * Cleans talent data by removing unnecessary fields
+     * Implements the filteredExtractedData logic from TMARawExtractor.vue
+     */
+    static cleanTalentData(rawData: TalentData): CleanedTalentData;
+    /**
+     * Transforms single person result to match filtered output format
+     */
+    static transformSinglePersonResult(result: SinglePersonResult): TmaExtractorResult;
+    /**
+     * Validates cleaned data structure
+     */
+    static validateCleanedData(data: CleanedTalentData): {
+        isValid: boolean;
+        errors: string[];
+    };
+    /**
+     * Gets summary statistics of cleaned data
+     */
+    static getDataSummary(data: CleanedTalentData): {
+        totalTalents: number;
+        totalStrengths: number;
+        totalTypologies: number;
+        totalPersonalBranding: number;
+        totalScmActivities: number | null;
+        hasScmData: boolean;
+    };
+}
+//# sourceMappingURL=data-transformer.d.ts.map

package/dist/core/data-transformer.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"data-transformer.d.ts","sourceRoot":"","sources":["../../src/core/data-transformer.ts"],"names":[],"mappings":"AAAA,OAAO,EACL,UAAU,EACV,MAAM,EACN,aAAa,EACb,iBAAiB,EACjB,kBAAkB,EAClB,kBAAkB,EAClB,OAAO,EACP,cAAc,EACf,MAAM,oBAAoB,CAAC;AAE5B;;;GAGG;AACH,qBAAa,eAAe;IAC1B;;OAEG;IACH,MAAM,CAAC,gBAAgB,CAAC,OAAO,EAAE,MAAM,EAAE,GAAG,aAAa,EAAE;IAO3D;;;OAGG;IACH,MAAM,CAAC,iBAAiB,CAAC,OAAO,EAAE,OAAO,GAAG,cAAc;IAoB1D;;;OAGG;IACH,MAAM,CAAC,eAAe,CAAC,OAAO,EAAE,UAAU,GAAG,iBAAiB;IAiB9D;;OAEG;IACH,MAAM,CAAC,2BAA2B,CAAC,MAAM,EAAE,kBAAkB,GAAG,kBAAkB;IASlF;;OAEG;IACH,MAAM,CAAC,mBAAmB,CAAC,IAAI,EAAE,iBAAiB,GAAG;QAAE,OAAO,EAAE,OAAO,CAAC;QAAC,MAAM,EAAE,MAAM,EAAE,CAAA;KAAE;IA2D3F;;OAEG;IACH,MAAM,CAAC,cAAc,CAAC,IAAI,EAAE,iBAAiB,GAAG;QAC9C,YAAY,EAAE,MAAM,CAAC;QACrB,cAAc,EAAE,MAAM,CAAC;QACvB,eAAe,EAAE,MAAM,CAAC;QACxB,qBAAqB,EAAE,MAAM,CAAC;QAC9B,kBAAkB,EAAE,MAAM,GAAG,IAAI,CAAC;QAClC,UAAU,EAAE,OAAO,CAAC;KACrB;CAUF"}

package/dist/core/pdf-processor.d.ts

@@ -0,0 +1,48 @@
+import { TmaFormat, PageTexts, TmaExtractorConfig } from "../types/tma-types";
+/**
+ * PDF Processing class that wraps PDF.js functionality for TMA extraction
+ */
+export declare class PdfProcessor {
+    private workerSrc?;
+    constructor(config?: Partial<TmaExtractorConfig>);
+    /**
+     * Configures PDF.js worker source
+     */
+    private configurePdfWorker;
+    /**
+     * Gets the total number of pages in a PDF file
+     */
+    getPageCount(file: File): Promise<number>;
+    /**
+     * Detects TMA format based on page count
+     */
+    detectFormat(pageCount: number): TmaFormat;
+    /**
+     * Gets page numbers for extraction based on page count
+     */
+    getPageNumbers(pageCount: number): number[];
+    /**
+     * Gets SCM page number based on page count
+     */
+    getScmPageNumber(pageCount: number): number;
+    /**
+     * Extracts text content from a single PDF page
+     */
+    private getPageText;
+    /**
+     * Extracts text content from specific PDF pages
+     */
+    extractPageTexts(file: File, pageNumbers: number[]): Promise<PageTexts>;
+    /**
+     * Gets a specific PDF page for external processing (e.g., SCM)
+     */
+    getPage(file: File, pageNumber: number): Promise<any>;
+    /**
+     * Validates if a file is a valid PDF
+     */
+    validatePdfFile(file: File): {
+        isValid: boolean;
+        error?: string;
+    };
+}
+//# sourceMappingURL=pdf-processor.d.ts.map

package/dist/core/pdf-processor.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"pdf-processor.d.ts","sourceRoot":"","sources":["../../src/core/pdf-processor.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,SAAS,EAAE,SAAS,EAAE,kBAAkB,EAAE,MAAM,oBAAoB,CAAC;AAK9E;;GAEG;AACH,qBAAa,YAAY;IACvB,OAAO,CAAC,SAAS,CAAC,CAAS;gBAEf,MAAM,CAAC,EAAE,OAAO,CAAC,kBAAkB,CAAC;IAKhD;;OAEG;IACH,OAAO,CAAC,kBAAkB;IAY1B;;OAEG;IACG,YAAY,CAAC,IAAI,EAAE,IAAI,GAAG,OAAO,CAAC,MAAM,CAAC;IAe/C;;OAEG;IACH,YAAY,CAAC,SAAS,EAAE,MAAM,GAAG,SAAS;IAI1C;;OAEG;IACH,cAAc,CAAC,SAAS,EAAE,MAAM,GAAG,MAAM,EAAE;IAc3C;;OAEG;IACH,gBAAgB,CAAC,SAAS,EAAE,MAAM,GAAG,MAAM;IAc3C;;OAEG;YACW,WAAW;IAazB;;OAEG;IACG,gBAAgB,CACpB,IAAI,EAAE,IAAI,EACV,WAAW,EAAE,MAAM,EAAE,GACpB,OAAO,CAAC,SAAS,CAAC;IA6CrB;;OAEG;IACG,OAAO,CAAC,IAAI,EAAE,IAAI,EAAE,UAAU,EAAE,MAAM,GAAG,OAAO,CAAC,GAAG,CAAC;IA2B3D;;OAEG;IACH,eAAe,CAAC,IAAI,EAAE,IAAI,GAAG;QAAE,OAAO,EAAE,OAAO,CAAC;QAAC,KAAK,CAAC,EAAE,MAAM,CAAA;KAAE;CAoBlE"}

package/dist/extractors/branding-extractor.d.ts

@@ -0,0 +1,21 @@
+import { PersonalBranding } from '../types/tma-types';
+/**
+ * Extracts personal branding data based on TMA format
+ */
+export declare function extractPersonalBranding(text: string, pageCount?: number): PersonalBranding[];
+/**
+ * Validates extracted personal branding data
+ */
+export declare function validatePersonalBrandingData(branding: PersonalBranding[]): {
+    isValid: boolean;
+    errors: string[];
+};
+/**
+ * Normalizes personal branding IDs for consistency
+ */
+export declare function normalizePersonalBrandingIds(branding: PersonalBranding[]): PersonalBranding[];
+/**
+ * Filters personal branding to top N items
+ */
+export declare function getTopPersonalBranding(branding: PersonalBranding[], count?: number): PersonalBranding[];
+//# sourceMappingURL=branding-extractor.d.ts.map

package/dist/extractors/branding-extractor.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"branding-extractor.d.ts","sourceRoot":"","sources":["../../src/extractors/branding-extractor.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,gBAAgB,EAAE,MAAM,oBAAoB,CAAC;AAqDtD;;GAEG;AACH,wBAAgB,uBAAuB,CAAC,IAAI,EAAE,MAAM,EAAE,SAAS,GAAE,MAAU,GAAG,gBAAgB,EAAE,CAQ/F;AAED;;GAEG;AACH,wBAAgB,4BAA4B,CAAC,QAAQ,EAAE,gBAAgB,EAAE,GAAG;IAAE,OAAO,EAAE,OAAO,CAAC;IAAC,MAAM,EAAE,MAAM,EAAE,CAAA;CAAE,CA2BjH;AAED;;GAEG;AACH,wBAAgB,4BAA4B,CAAC,QAAQ,EAAE,gBAAgB,EAAE,GAAG,gBAAgB,EAAE,CAQ7F;AAED;;GAEG;AACH,wBAAgB,sBAAsB,CAAC,QAAQ,EAAE,gBAAgB,EAAE,EAAE,KAAK,GAAE,MAAgC,GAAG,gBAAgB,EAAE,CAKhI"}

package/dist/extractors/scm-extractor.d.ts

@@ -0,0 +1,96 @@
+/**
+ * SCM (Strength Cluster Map) Extractor
+ * Framework-independent extraction of 114 activities with PSP/PSS color detection
+ * Integrates PDF.js for rendering and Tesseract.js for OCR
+ */
+import { type ScmColor, type ScmArea } from '../constants/scm-activities';
+export interface ScmDetectedActivity {
+    activityName: string;
+    psp: ScmColor | null;
+    pss: ScmColor;
+    area: ScmArea;
+    position: number;
+    confidence: number;
+}
+export interface ScmMappedActivity {
+    id: string;
+    typology: string;
+    cluster: string;
+    psp: ScmColor | null;
+    pss: ScmColor;
+    detectionConfidence: number;
+}
+export interface ScmData {
+    activities: ScmMappedActivity[];
+    metadata: {
+        totalActivities: number;
+        detectedActivities: number;
+        unmappedActivities: string[];
+        averageConfidence: number;
+        extractionTime: number;
+    };
+}
+export declare class ScmExtractor {
+    private tesseractWorker;
+    /**
+     * Extract SCM data from a PDF page
+     */
+    extractScmData(pdfPage: any): Promise<ScmData | null>;
+    /**
+     * Detect if SCM page has ID below the name using PDF.js text extraction
+     */
+    private detectScmHasId;
+    /**
+     * Get area configurations based on whether ID is present
+     */
+    private getScmAreaConfigs;
+    /**
+     * Render PDF page to high-resolution canvas
+     */
+    private renderPageToCanvas;
+    /**
+     * Detect color at specific coordinates using Canvas API
+     */
+    private detectColorAt;
+    /**
+     * Map RGB values to SCM color using exact hex color matching
+     */
+    private mapRgbToScmColor;
+    /**
+     * Extract activities from a specific area using pre-defined positions
+     */
+    private extractAreaActivities;
+    /**
+     * Calculate activity coordinates based on area and position
+     */
+    private calculateActivityCoordinates;
+    /**
+     * Extract activity name using Tesseract OCR from canvas region
+     */
+    private extractActivityName;
+    /**
+     * Initialize Tesseract worker for OCR
+     */
+    private initializeTesseract;
+    /**
+     * Normalize activity name for matching
+     */
+    private normalizeActivityName;
+    /**
+     * Find best matching activity from pre-defined list
+     */
+    private findBestActivityMatch;
+    /**
+     * Calculate detection confidence score
+     */
+    private calculateDetectionConfidence;
+    /**
+     * Map detected activities to definitions
+     */
+    private mapDetectedToDefinitions;
+    /**
+     * Cleanup Tesseract worker
+     */
+    cleanup(): Promise<void>;
+}
+//# sourceMappingURL=scm-extractor.d.ts.map

package/dist/extractors/scm-extractor.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"scm-extractor.d.ts","sourceRoot":"","sources":["../../src/extractors/scm-extractor.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAIL,KAAK,QAAQ,EACb,KAAK,OAAO,EACb,MAAM,6BAA6B,CAAC;AAGrC,MAAM,WAAW,mBAAmB;IAClC,YAAY,EAAE,MAAM,CAAC;IACrB,GAAG,EAAE,QAAQ,GAAG,IAAI,CAAC;IACrB,GAAG,EAAE,QAAQ,CAAC;IACd,IAAI,EAAE,OAAO,CAAC;IACd,QAAQ,EAAE,MAAM,CAAC;IACjB,UAAU,EAAE,MAAM,CAAC;CACpB;AAED,MAAM,WAAW,iBAAiB;IAChC,EAAE,EAAE,MAAM,CAAC;IACX,QAAQ,EAAE,MAAM,CAAC;IACjB,OAAO,EAAE,MAAM,CAAC;IAChB,GAAG,EAAE,QAAQ,GAAG,IAAI,CAAC;IACrB,GAAG,EAAE,QAAQ,CAAC;IACd,mBAAmB,EAAE,MAAM,CAAC;CAC7B;AAED,MAAM,WAAW,OAAO;IACtB,UAAU,EAAE,iBAAiB,EAAE,CAAC;IAChC,QAAQ,EAAE;QACR,eAAe,EAAE,MAAM,CAAC;QACxB,kBAAkB,EAAE,MAAM,CAAC;QAC3B,kBAAkB,EAAE,MAAM,EAAE,CAAC;QAC7B,iBAAiB,EAAE,MAAM,CAAC;QAC1B,cAAc,EAAE,MAAM,CAAC;KACxB,CAAC;CACH;AAED,qBAAa,YAAY;IACvB,OAAO,CAAC,eAAe,CAAa;IAEpC;;OAEG;IACG,cAAc,CAAC,OAAO,EAAE,GAAG,GAAG,OAAO,CAAC,OAAO,GAAG,IAAI,CAAC;IA+D3D;;OAEG;YACW,cAAc;IA6B5B;;OAEG;IACH,OAAO,CAAC,iBAAiB;IAgDzB;;OAEG;YACW,kBAAkB;IAuBhC;;OAEG;IACH,OAAO,CAAC,aAAa;IA8BrB;;OAEG;IACH,OAAO,CAAC,gBAAgB;IAyCxB;;OAEG;YACW,qBAAqB;IAsDnC;;OAEG;IACH,OAAO,CAAC,4BAA4B;IA4EpC;;OAEG;YACW,mBAAmB;IA8EjC;;OAEG;YACW,mBAAmB;IAcjC;;OAEG;IACH,OAAO,CAAC,qBAAqB;IAS7B;;OAEG;IACH,OAAO,CAAC,qBAAqB;IAsB7B;;OAEG;IACH,OAAO,CAAC,4BAA4B;IAoCpC;;OAEG;IACH,OAAO,CAAC,wBAAwB;IA8BhC;;OAEG;IACG,OAAO;CAMd"}

package/dist/extractors/strength-extractor.d.ts

@@ -0,0 +1,21 @@
+import { Strength } from '../types/tma-types';
+/**
+ * Extracts strength data from text based on TMA format
+ */
+export declare function extractStrength(text: string, pageCount?: number): Strength[];
+/**
+ * Validates extracted strength data
+ */
+export declare function validateStrengthData(strengths: Strength[]): {
+    isValid: boolean;
+    errors: string[];
+};
+/**
+ * Sorts strengths by their number in ascending order
+ */
+export declare function sortStrengthsByNumber(strengths: Strength[]): Strength[];
+/**
+ * Filters strengths to get top N strongest ones (lowest numbers)
+ */
+export declare function getTopStrengths(strengths: Strength[], count?: number): Strength[];
+//# sourceMappingURL=strength-extractor.d.ts.map

package/dist/extractors/strength-extractor.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"strength-extractor.d.ts","sourceRoot":"","sources":["../../src/extractors/strength-extractor.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,QAAQ,EAAE,MAAM,oBAAoB,CAAC;AAoC9C;;GAEG;AACH,wBAAgB,eAAe,CAAC,IAAI,EAAE,MAAM,EAAE,SAAS,GAAE,MAAU,GAAG,QAAQ,EAAE,CAQ/E;AAED;;GAEG;AACH,wBAAgB,oBAAoB,CAAC,SAAS,EAAE,QAAQ,EAAE,GAAG;IAAE,OAAO,EAAE,OAAO,CAAC;IAAC,MAAM,EAAE,MAAM,EAAE,CAAA;CAAE,CA4BlG;AAED;;GAEG;AACH,wBAAgB,qBAAqB,CAAC,SAAS,EAAE,QAAQ,EAAE,GAAG,QAAQ,EAAE,CAEvE;AAED;;GAEG;AACH,wBAAgB,eAAe,CAAC,SAAS,EAAE,QAAQ,EAAE,EAAE,KAAK,GAAE,MAAU,GAAG,QAAQ,EAAE,CAMpF"}

package/dist/extractors/talent-extractor.d.ts

@@ -0,0 +1,25 @@
+import { Talent } from '../types/tma-types';
+/**
+ * Extracts talent order from text based on TMA format
+ */
+export declare function extractTalentOrder(text: string, isPdfParse?: boolean, pageCount?: number): Talent[];
+/**
+ * Extracts name from talent order text based on format
+ */
+export declare function extractName(text: string, pageCount: number): string;
+/**
+ * Validates extracted talent data
+ */
+export declare function validateTalentData(talents: Talent[]): {
+    isValid: boolean;
+    errors: string[];
+};
+/**
+ * Calculates talent groups for analysis
+ */
+export declare function calculateTalentGroups(talents: Talent[]): {
+    top14Talents: string[];
+    top7Talents: string[];
+    low14Talents: string[];
+};
+//# sourceMappingURL=talent-extractor.d.ts.map

package/dist/extractors/talent-extractor.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"talent-extractor.d.ts","sourceRoot":"","sources":["../../src/extractors/talent-extractor.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,MAAM,EAAE,MAAM,oBAAoB,CAAC;AA6C5C;;GAEG;AACH,wBAAgB,kBAAkB,CAChC,IAAI,EAAE,MAAM,EACZ,UAAU,GAAE,OAAe,EAC3B,SAAS,GAAE,MAAU,GACpB,MAAM,EAAE,CAaV;AAED;;GAEG;AACH,wBAAgB,WAAW,CAAC,IAAI,EAAE,MAAM,EAAE,SAAS,EAAE,MAAM,GAAG,MAAM,CAYnE;AAED;;GAEG;AACH,wBAAgB,kBAAkB,CAAC,OAAO,EAAE,MAAM,EAAE,GAAG;IAAE,OAAO,EAAE,OAAO,CAAC;IAAC,MAAM,EAAE,MAAM,EAAE,CAAA;CAAE,CA2B5F;AAED;;GAEG;AACH,wBAAgB,qBAAqB,CAAC,OAAO,EAAE,MAAM,EAAE;;;;EAYtD"}

package/dist/extractors/typology-extractor.d.ts

@@ -0,0 +1,25 @@
+import { Typology } from '../types/tma-types';
+/**
+ * Extracts all 30 typologies with their scores from text
+ */
+export declare function extractTypology(text: string): Typology[];
+/**
+ * Validates extracted typology data
+ */
+export declare function validateTypologyData(typologies: Typology[]): {
+    isValid: boolean;
+    errors: string[];
+};
+/**
+ * Sorts typologies by score in descending order
+ */
+export declare function sortTypologiesByScore(typologies: Typology[]): Typology[];
+/**
+ * Groups typologies by category
+ */
+export declare function groupTypologiesByCategory(typologies: Typology[]): Record<string, Typology[]>;
+/**
+ * Filters typologies to get top N highest scored ones
+ */
+export declare function getTopTypologies(typologies: Typology[], count?: number): Typology[];
+//# sourceMappingURL=typology-extractor.d.ts.map

package/dist/extractors/typology-extractor.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"typology-extractor.d.ts","sourceRoot":"","sources":["../../src/extractors/typology-extractor.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,QAAQ,EAAE,MAAM,oBAAoB,CAAC;AA6B9C;;GAEG;AACH,wBAAgB,eAAe,CAAC,IAAI,EAAE,MAAM,GAAG,QAAQ,EAAE,CAqBxD;AAED;;GAEG;AACH,wBAAgB,oBAAoB,CAAC,UAAU,EAAE,QAAQ,EAAE,GAAG;IAAE,OAAO,EAAE,OAAO,CAAC;IAAC,MAAM,EAAE,MAAM,EAAE,CAAA;CAAE,CA4BnG;AAED;;GAEG;AACH,wBAAgB,qBAAqB,CAAC,UAAU,EAAE,QAAQ,EAAE,GAAG,QAAQ,EAAE,CAExE;AAED;;GAEG;AACH,wBAAgB,yBAAyB,CAAC,UAAU,EAAE,QAAQ,EAAE,GAAG,MAAM,CAAC,MAAM,EAAE,QAAQ,EAAE,CAAC,CAS5F;AAED;;GAEG;AACH,wBAAgB,gBAAgB,CAAC,UAAU,EAAE,QAAQ,EAAE,EAAE,KAAK,GAAE,MAAU,GAAG,QAAQ,EAAE,CAMtF"}