create-entity-app-server 0.0.3 → 0.1.13

package/src/app/plugins/ocr/providers/naver.ts

@@ -1,186 +0,0 @@
-/**
- * Naver CLOVA General OCR 드라이버
- * multipart/form-data 기반 API
- */
-
-import type {
-    OcrDriver,
-    OcrProviderConfig,
-    RawOcrResult,
-    RecognizeOptions,
-    TextBlock,
-    RawOcrPage,
-} from "../types/index.ts";
-
-const EXTENSION_MAP: Record<string, string> = {
-    "image/jpeg": "jpg",
-    "image/png": "png",
-    "image/tiff": "tiff",
-    "image/bmp": "bmp",
-    "image/webp": "webp",
-    "application/pdf": "pdf",
-};
-
-export class NaverDriver implements OcrDriver {
-    private readonly apiUrl: string;
-    private readonly secretKey: string;
-    private readonly timeout: number;
-
-    /** Naver CLOVA OCR 드라이버를 설정으로 초기화한다 */
-    constructor(cfg: OcrProviderConfig) {
-        this.apiUrl = cfg.apiUrl ?? "";
-        this.secretKey = cfg.secretKey ?? "";
-        this.timeout = (cfg.timeoutSec ?? 30) * 1000;
-    }
-
-    /** 드라이버 이름을 반환한다 */
-    name() {
-        return "naver";
-    }
-    /** 지원하는 파일 형식 목록을 반환한다 */
-    supportedFormats() {
-        return [
-            "image/jpeg",
-            "image/png",
-            "image/tiff",
-            "image/bmp",
-            "application/pdf",
-        ];
-    }
-    /** 드라이버 리소스를 해제한다 */
-    async close() {}
-
-    /** 이미지 데이터를 Naver CLOVA API로 인식한다 */
-    async recognize(
-        data: Buffer,
-        opts: RecognizeOptions,
-    ): Promise<RawOcrResult> {
-        const start = Date.now();
-        const ext = EXTENSION_MAP[opts.mimeType] ?? "jpg";
-
-        const message = {
-            version: "V2",
-            requestId: crypto.randomUUID(),
-            timestamp: Date.now(),
-            lang: opts.languages?.[0] ?? "ko",
-            images: [{ format: ext, name: "image" }],
-        };
-
-        const formData = new FormData();
-        formData.append("message", JSON.stringify(message));
-        formData.append(
-            "file",
-            new Blob([data], { type: opts.mimeType }),
-            `image.${ext}`,
-        );
-
-        const resp = await fetch(this.apiUrl, {
-            method: "POST",
-            headers: { "X-OCR-SECRET": this.secretKey },
-            body: formData,
-            signal: AbortSignal.timeout(this.timeout),
-        });
-
-        if (!resp.ok) {
-            const text = await resp.text();
-            throw new Error(`Naver OCR API error ${resp.status}: ${text}`);
-        }
-
-        const json = (await resp.json()) as any;
-        const images = json.images ?? [];
-        const resultPages: RawOcrPage[] = [];
-        let fullText = "";
-
-        for (let pi = 0; pi < images.length; pi++) {
-            const img = images[pi];
-            const blocks: TextBlock[] = [];
-            const lines: string[] = [];
-
-            for (const field of img.fields ?? []) {
-                const tb = parseNaverField(field);
-                if (tb) {
-                    blocks.push(tb);
-                    lines.push(tb.text);
-                }
-            }
-
-            const pageText = lines.join(" ");
-            fullText += (pi > 0 ? "\n" : "") + pageText;
-
-            const avgConf =
-                blocks.length > 0
-                    ? blocks.reduce((s, b) => s + b.confidence, 0) /
-                      blocks.length
-                    : 0;
-
-            // 이미지 크기 추정 (bounding box 기반)
-            let maxRight = 0,
-                maxBottom = 0;
-            for (const b of blocks) {
-                if (b.boundingBox.right > maxRight)
-                    maxRight = b.boundingBox.right;
-                if (b.boundingBox.bottom > maxBottom)
-                    maxBottom = b.boundingBox.bottom;
-            }
-
-            resultPages.push({
-                pageNum: pi + 1,
-                width: maxRight,
-                height: maxBottom,
-                text: pageText,
-                blocks,
-            });
-        }
-
-        const totalConf = resultPages.reduce(
-            (s, p) => s + p.blocks.reduce((ss, b) => ss + b.confidence, 0),
-            0,
-        );
-        const totalBlocks = resultPages.reduce(
-            (s, p) => s + p.blocks.length,
-            0,
-        );
-
-        return {
-            provider: "naver",
-            providerRaw: json,
-            fullText,
-            processingMs: Date.now() - start,
-            processedAt: new Date().toISOString(),
-            confidence: totalBlocks > 0 ? totalConf / totalBlocks : 0,
-            pages: resultPages,
-        };
-    }
-}
-
-/** Naver OCR 필드 데이터를 TextBlock으로 변환한다 */
-function parseNaverField(field: any): TextBlock | null {
-    const text = field.inferText ?? "";
-    if (!text) return null;
-
-    const conf = field.inferConfidence ?? 0;
-    const vertices = field.boundingPoly?.vertices ?? [];
-    if (vertices.length < 4) return null;
-
-    const xs = vertices.map((v: any) => v.x ?? 0);
-    const ys = vertices.map((v: any) => v.y ?? 0);
-    const left = Math.min(...xs);
-    const right = Math.max(...xs);
-    const top = Math.min(...ys);
-    const bottom = Math.max(...ys);
-
-    const slope =
-        right - left > 0
-            ? ((vertices[1]?.y ?? 0) - (vertices[0]?.y ?? 0)) / (right - left)
-            : 0;
-
-    return {
-        text,
-        boundingBox: { top, left, right, bottom },
-        centerX: (left + right) / 2,
-        centerY: (top + bottom) / 2,
-        slope,
-        confidence: conf,
-        lineBreak: field.lineBreak?.type === "SPACE" ? false : true,
-    };
-}

package/src/app/plugins/ocr/providers/tesseract.ts

@@ -1,198 +0,0 @@
-/**
- * 로컬 Tesseract CLI OCR 드라이버
- * tesseract CLI → TSV 파싱
- */
-
-import { execFile } from "node:child_process";
-import { writeFile, mkdtemp, rm } from "node:fs/promises";
-import { tmpdir } from "node:os";
-import { join } from "node:path";
-import { promisify } from "node:util";
-import type {
-    OcrDriver,
-    OcrProviderConfig,
-    RawOcrResult,
-    RecognizeOptions,
-    TextBlock,
-} from "../types/index.ts";
-
-const execFileAsync = promisify(execFile);
-
-// TSV 컬럼 인덱스 (Tesseract output)
-const TSV_LEVEL = 0;
-const TSV_PAGE_NUM = 1;
-const TSV_BLOCK_NUM = 2;
-const TSV_PAR_NUM = 3;
-const TSV_LINE_NUM = 4;
-const TSV_WORD_NUM = 5;
-const TSV_LEFT = 6;
-const TSV_TOP = 7;
-const TSV_WIDTH = 8;
-const TSV_HEIGHT = 9;
-const TSV_CONF = 10;
-const TSV_TEXT = 11;
-
-export class TesseractDriver implements OcrDriver {
-    private readonly lang: string;
-    private readonly dataPath: string;
-    private readonly psm: number;
-    private readonly oem: number;
-    private readonly timeout: number;
-
-    /** Tesseract CLI 드라이버를 설정으로 초기화한다 */
-    constructor(cfg: OcrProviderConfig) {
-        this.lang = cfg.defaultLang ?? "kor+eng";
-        this.dataPath = cfg.dataPath ?? "";
-        this.psm = cfg.psm ?? 3;
-        this.oem = cfg.oem ?? 3;
-        this.timeout = (cfg.timeoutSec ?? 30) * 1000;
-    }
-
-    /** 드라이버 이름을 반환한다 */
-    name() {
-        return "tesseract";
-    }
-    /** 지원하는 파일 형식 목록을 반환한다 */
-    supportedFormats() {
-        return [
-            "image/jpeg",
-            "image/png",
-            "image/tiff",
-            "image/bmp",
-            "image/webp",
-        ];
-    }
-    /** 드라이버 리소스를 해제한다 */
-    async close() {}
-
-    /** 이미지 데이터를 Tesseract CLI로 인식한다 */
-    async recognize(
-        data: Buffer,
-        opts: RecognizeOptions,
-    ): Promise<RawOcrResult> {
-        const start = Date.now();
-        const tmpDir = await mkdtemp(join(tmpdir(), "ocr-tess-"));
-
-        try {
-            const inputPath = join(tmpDir, "input");
-            const outputBase = join(tmpDir, "output");
-            await writeFile(inputPath, data);
-
-            const args = [
-                inputPath,
-                outputBase,
-                "-l",
-                opts.languages?.[0] ?? this.lang,
-                "--psm",
-                String(this.psm),
-                "--oem",
-                String(this.oem),
-                "tsv",
-            ];
-            if (this.dataPath) {
-                args.unshift("--tessdata-dir", this.dataPath);
-            }
-
-            await execFileAsync("tesseract", args, { timeout: this.timeout });
-
-            const { readFile } = await import("node:fs/promises");
-            const tsvData = await readFile(outputBase + ".tsv", "utf-8");
-            const { blocks, fullText } = parseTesseractTSV(tsvData);
-
-            const result: RawOcrResult = {
-                provider: "tesseract",
-                providerRaw: tsvData,
-                fullText,
-                processingMs: Date.now() - start,
-                processedAt: new Date().toISOString(),
-                confidence: 0,
-                pages: [],
-            };
-
-            if (blocks.length > 0) {
-                let maxRight = 0,
-                    maxBottom = 0,
-                    totalConf = 0;
-                for (const b of blocks) {
-                    if (b.boundingBox.right > maxRight)
-                        maxRight = b.boundingBox.right;
-                    if (b.boundingBox.bottom > maxBottom)
-                        maxBottom = b.boundingBox.bottom;
-                    totalConf += b.confidence;
-                }
-                result.pages = [
-                    {
-                        pageNum: 1,
-                        width: maxRight,
-                        height: maxBottom,
-                        text: fullText,
-                        blocks,
-                    },
-                ];
-                result.confidence = totalConf / blocks.length;
-            }
-
-            return result;
-        } finally {
-            await rm(tmpDir, { recursive: true, force: true }).catch(() => {});
-        }
-    }
-}
-
-/**
- * Tesseract TSV 출력을 파싱합니다.
- */
-function parseTesseractTSV(tsv: string): {
-    blocks: TextBlock[];
-    fullText: string;
-} {
-    const blocks: TextBlock[] = [];
-    const lines: string[] = [];
-    let curLine = "";
-    let prevLineNum = -1;
-
-    const rows = tsv.split("\n");
-    // 헤더 스킵
-    for (let i = 1; i < rows.length; i++) {
-        const fields = rows[i].split("\t");
-        if (fields.length <= TSV_TEXT) continue;
-
-        const text = fields[TSV_TEXT];
-        if (!text) continue;
-
-        let conf = parseFloat(fields[TSV_CONF]) || 0;
-        if (conf < 0) conf = 0;
-
-        const left = parseInt(fields[TSV_LEFT]) || 0;
-        const top = parseInt(fields[TSV_TOP]) || 0;
-        const width = parseInt(fields[TSV_WIDTH]) || 0;
-        const height = parseInt(fields[TSV_HEIGHT]) || 0;
-        const lineNum = parseInt(fields[TSV_LINE_NUM]) || 0;
-        const right = left + width;
-        const bottom = top + height;
-
-        if (lineNum !== prevLineNum && prevLineNum !== -1) {
-            const trimmed = curLine.trim();
-            if (trimmed) lines.push(trimmed);
-            curLine = "";
-        }
-        curLine += text + " ";
-        prevLineNum = lineNum;
-
-        blocks.push({
-            text,
-            confidence: conf / 100,
-            boundingBox: { top, left, right, bottom },
-            centerX: (left + right) / 2,
-            centerY: (top + bottom) / 2,
-            slope: 0,
-            lineBreak: false,
-        });
-    }
-
-    if (curLine.trim()) {
-        lines.push(curLine.trim());
-    }
-
-    return { blocks, fullText: lines.join("\n") };
-}

package/src/app/plugins/ocr/providers/upstage.ts

@@ -1,156 +0,0 @@
-/**
- * Upstage Document AI OCR 드라이버
- * multipart/form-data 기반 API
- */
-
-import type {
-    OcrDriver,
-    OcrProviderConfig,
-    RawOcrResult,
-    RawOcrPage,
-    RecognizeOptions,
-    TextBlock,
-} from "../types/index.ts";
-
-export class UpstageDriver implements OcrDriver {
-    private readonly apiKey: string;
-    private readonly apiUrl: string;
-    private readonly timeout: number;
-
-    /** Upstage Document AI 드라이버를 설정으로 초기화한다 */
-    constructor(cfg: OcrProviderConfig) {
-        this.apiKey = cfg.apiKey ?? "";
-        this.apiUrl = cfg.apiUrl ?? "https://api.upstage.ai/v1/document-ai/ocr";
-        this.timeout = (cfg.timeoutSec ?? 60) * 1000;
-    }
-
-    /** 드라이버 이름을 반환한다 */
-    name() {
-        return "upstage";
-    }
-    /** 지원하는 파일 형식 목록을 반환한다 */
-    supportedFormats() {
-        return ["image/jpeg", "image/png", "image/tiff", "application/pdf"];
-    }
-    /** 드라이버 리소스를 해제한다 */
-    async close() {}
-
-    /** 이미지 데이터를 Upstage API로 인식한다 */
-    async recognize(
-        data: Buffer,
-        opts: RecognizeOptions,
-    ): Promise<RawOcrResult> {
-        const start = Date.now();
-
-        const formData = new FormData();
-        formData.append(
-            "document",
-            new Blob([data], { type: opts.mimeType }),
-            "document",
-        );
-
-        const resp = await fetch(this.apiUrl, {
-            method: "POST",
-            headers: { Authorization: `Bearer ${this.apiKey}` },
-            body: formData,
-            signal: AbortSignal.timeout(this.timeout),
-        });
-
-        if (!resp.ok) {
-            const text = await resp.text();
-            throw new Error(`Upstage OCR API error ${resp.status}: ${text}`);
-        }
-
-        const json = (await resp.json()) as any;
-        const blocks: TextBlock[] = [];
-        let fullText = "";
-        const resultPages: RawOcrPage[] = [];
-
-        for (let pi = 0; pi < (json.pages ?? []).length; pi++) {
-            const page = json.pages[pi];
-            const pageBlocks: TextBlock[] = [];
-            for (const word of page.words ?? []) {
-                // Upstage: bounding_box는 [{x,y}, {x,y}, {x,y}, {x,y}] 형태
-                const bbox = word.bounding_box ?? word.boundingBox ?? [];
-                let left = 0,
-                    right = 0,
-                    top = 0,
-                    bottom = 0;
-
-                if (
-                    bbox.length > 0 &&
-                    typeof bbox[0] === "object" &&
-                    bbox[0] !== null
-                ) {
-                    // [{x,y}, ...] 형태
-                    const xs = bbox.map((p: any) => p.x ?? 0);
-                    const ys = bbox.map((p: any) => p.y ?? 0);
-                    left = Math.min(...xs);
-                    right = Math.max(...xs);
-                    top = Math.min(...ys);
-                    bottom = Math.max(...ys);
-                } else if (bbox.length > 0) {
-                    // flat [x1,y1,x2,y2,...] 형태 (폴백)
-                    const xs = bbox.filter(
-                        (_: number, i: number) => i % 2 === 0,
-                    );
-                    const ys = bbox.filter(
-                        (_: number, i: number) => i % 2 === 1,
-                    );
-                    left = xs.length ? Math.min(...xs) : 0;
-                    right = xs.length ? Math.max(...xs) : 0;
-                    top = ys.length ? Math.min(...ys) : 0;
-                    bottom = ys.length ? Math.max(...ys) : 0;
-                }
-
-                const tb: TextBlock = {
-                    text: word.text ?? "",
-                    boundingBox: { top, left, right, bottom },
-                    centerX: (left + right) / 2,
-                    centerY: (top + bottom) / 2,
-                    slope: 0,
-                    confidence: word.confidence ?? 0,
-                    lineBreak: false,
-                };
-                pageBlocks.push(tb);
-                blocks.push(tb);
-            }
-
-            const pageWidth =
-                page.width ??
-                (pageBlocks.length > 0
-                    ? Math.max(...pageBlocks.map((b) => b.boundingBox.right))
-                    : 0);
-            const pageHeight =
-                page.height ??
-                (pageBlocks.length > 0
-                    ? Math.max(...pageBlocks.map((b) => b.boundingBox.bottom))
-                    : 0);
-
-            resultPages.push({
-                pageNum: pi + 1,
-                width: pageWidth,
-                height: pageHeight,
-                text: pageBlocks.map((b) => b.text).join(" "),
-                blocks: pageBlocks,
-            });
-        }
-
-        fullText = json.text ?? blocks.map((b) => b.text).join(" ");
-
-        const avgConf =
-            blocks.length > 0
-                ? blocks.reduce((s, b) => s + b.confidence, 0) / blocks.length
-                : 0;
-
-        return {
-            provider: "upstage",
-            providerRaw: json,
-            fullText,
-            processingMs: Date.now() - start,
-            processedAt: new Date().toISOString(),
-            confidence: avgConf,
-            pages: resultPages,
-        };
-    }
-}

package/src/app/plugins/ocr/quota.ts

@@ -1,108 +0,0 @@
-/**
- * QuotaManager — 일/월 단위 OCR 사용 쿼터 관리
- */
-
-import type {
-    OcrQuotaConfig,
-    OcrEntityStore,
-    OcrUsageRecord,
-    QuotaStatus,
-} from "./types/index.ts";
-import { OcrError, ERR_QUOTA_EXCEEDED } from "./errors.ts";
-
-export class QuotaManager {
-    /** 쿼터 매니저를 설정과 엔티티 스토어로 초기화한다 */
-    constructor(
-        private readonly config: OcrQuotaConfig,
-        private readonly entityStore: OcrEntityStore,
-    ) {}
-
-    /** 현재 사용량이 쿼터를 초과하는지 확인 (80% 경고, 100% 에러) */
-    async check(): Promise<void> {
-        if (this.config.dailyLimit <= 0 && this.config.monthlyLimit <= 0)
-            return;
-
-        const today = new Date().toISOString().slice(0, 10);
-        const month = today.slice(0, 7);
-
-        let dailyUsed = 0;
-        let monthlyUsed = 0;
-
-        try {
-            dailyUsed = await this.entityStore.getDailyQuotaUsage(today);
-        } catch {
-            return; // 조회 실패 시 통과
-        }
-        try {
-            monthlyUsed = await this.entityStore.getMonthlyQuotaUsage(month);
-        } catch {
-            return;
-        }
-
-        if (this.config.dailyLimit > 0) {
-            const pct = (dailyUsed / this.config.dailyLimit) * 100;
-            if (dailyUsed >= this.config.dailyLimit) {
-                throw new OcrError(
-                    `daily limit reached (${dailyUsed}/${this.config.dailyLimit})`,
-                    ERR_QUOTA_EXCEEDED,
-                );
-            }
-            if (pct >= 80) {
-                console.log(
-                    `[WARN] OCR daily quota at ${pct.toFixed(0)}% (${dailyUsed}/${this.config.dailyLimit})`,
-                );
-            }
-        }
-
-        if (this.config.monthlyLimit > 0) {
-            const pct = (monthlyUsed / this.config.monthlyLimit) * 100;
-            if (monthlyUsed >= this.config.monthlyLimit) {
-                throw new OcrError(
-                    `monthly limit reached (${monthlyUsed}/${this.config.monthlyLimit})`,
-                    ERR_QUOTA_EXCEEDED,
-                );
-            }
-            if (pct >= 80) {
-                console.log(
-                    `[WARN] OCR monthly quota at ${pct.toFixed(0)}% (${monthlyUsed}/${this.config.monthlyLimit})`,
-                );
-            }
-        }
-    }
-
-    /** 사용량 카운터 1 증가 */
-    async increment(
-        provider: string,
-        docType: string,
-        pageCount: number,
-    ): Promise<void> {
-        const today = new Date().toISOString().slice(0, 10);
-        await this.entityStore.incrementUsage({
-            usageDate: today,
-            provider,
-            docType,
-            requestCount: 1,
-            pageCount,
-        });
-    }
-
-    /** 현재 쿼터 상태 조회 */
-    async getStatus(): Promise<QuotaStatus> {
-        const today = new Date().toISOString().slice(0, 10);
-        const month = today.slice(0, 7);
-
-        const dailyUsed = await this.entityStore
-            .getDailyQuotaUsage(today)
-            .catch(() => 0);
-        const monthlyUsed = await this.entityStore
-            .getMonthlyQuotaUsage(month)
-            .catch(() => 0);
-
-        return {
-            dailyUsed,
-            dailyLimit: this.config.dailyLimit,
-            monthlyUsed,
-            monthlyLimit: this.config.monthlyLimit,
-        };
-    }
-}