@onco-foundry/mask-port 0.1.2 → 0.1.4

package/dist/llm_sensitive_word_finder.js

@@ -33,6 +33,7 @@ const extractedSchema = z.array(z.object({
 /** 姓名类目标：形态校验只拦阿拉伯数字（单字符中文名是合法情形，不拦）。 */
 const NAME_TARGETS = new Set(['patient_name', 'doctor_signature']);
 const modelResponseSchema = z.object({
+    stop_reason: z.string().optional(),
     content: z.array(z.object({
         type: z.string(),
         text: z.string().optional(),
@@ -88,7 +89,8 @@ export const createLlmSensitiveWordFinder = (options) => {
                     system: SYSTEM_PROMPT,
                     messages: [{ role: 'user', content: text }],
                     temperature: 0,
-                    max_tokens: 4096,
+                    // 推理型模型的思考也吃 max_tokens：给小了会在正文输出前被截断，响应里一个 text 块都没有。
+                    max_tokens: 16384,
                     stream: false,
                 }),
                 signal: AbortSignal.timeout(timeoutMs),
@@ -107,7 +109,9 @@ export const createLlmSensitiveWordFinder = (options) => {
         // 带推理的模型会先回 thinking 块，取第一个 text 块。
         const textBlock = payload.data.content.find((block) => block.type === 'text' && block.text);
         if (textBlock?.text === undefined) {
-            throw new AppError('LLM 敏感词提取响应缺少文本输出', 502);
+            const blockTypes = payload.data.content.map((block) => block.type).join('、');
+            throw new AppError(`LLM 敏感词提取响应缺少文本输出（stop_reason=${payload.data.stop_reason ?? '未知'}，`
+                + `块类型=${blockTypes}；stop_reason=max_tokens 说明思考烧光了 token 上限）`, 502);
         }
         const parsed = extractedSchema.safeParse(extractJsonArray(textBlock.text));
         if (!parsed.success) {

package/dist/textin_masker.d.ts

@@ -23,6 +23,8 @@ export type TextInMaskerOptions = {
     readonly baseURL?: string;
     readonly timeoutMs?: number;
     readonly maxInputBytes?: number;
+    /** 自动处理链路使用：已发现的敏感词无法完整定位时拒绝交付。 */
+    readonly requireAllMatches?: boolean;
     /** 测试注入缝：替换掉真实的 HTTP 层，不碰网络。 */
     readonly fetch?: typeof fetch;
     /**

package/dist/textin_masker.js

@@ -156,17 +156,28 @@ export const createTextInMasker = (options) => {
             if (payload.data.code !== 200 || payload.data.result === undefined) {
                 throw new AppError(`TextIn 脱敏接口报错：${payload.data.code} ${payload.data.message}`, 502);
             }
+            if (options.requireAllMatches && payload.data.result.pages.length !== 1) {
+                throw new AppError('TextIn 脱敏需要单页图片识别结果', 502);
+            }
             const page = payload.data.result.pages[0];
             const lines = page.lines;
             const recognizedText = lines.map((line) => line.text).join('\n');
             const scaleX = page.width !== undefined && page.width > 0 ? imageWidth / page.width : 1;
             const scaleY = page.height !== undefined && page.height > 0 ? imageHeight / page.height : 1;
-            const requested = (await finder(recognizedText))
+            if (options.requireAllMatches && !recognizedText.trim()) {
+                throw new AppError('TextIn 脱敏未识别出文字，无法确认处理结果', 502);
+            }
+            const detected = await finder(recognizedText);
+            if (options.requireAllMatches && detected.some(word => !word.text || !(word.target in TARGET_LABELS) || !recognizedText.includes(word.text))) {
+                throw new AppError('TextIn 脱敏判定结果无法定位', 502);
+            }
+            const requested = detected
                 .filter((word) => word.text !== '' && request.targets.includes(word.target));
             const finalQuads = [];
             const mappings = [];
             const counters = new Map();
             for (const word of requested) {
+                let located = 0;
                 for (const line of lines) {
                     let fromIndex = 0;
                     while (true) {
@@ -174,9 +185,30 @@ export const createTextInMasker = (options) => {
                         if (hit < 0)
                             break;
                         fromIndex = hit + word.text.length;
-                        const quad = locateWordQuad(line, hit, word.text.length);
-                        if (quad === undefined)
+                        const locatedLine = options.requireAllMatches && line.char_positions?.length !== line.text.length
+                            ? { ...line, char_positions: undefined } : line;
+                        const quad = locateWordQuad(locatedLine, hit, word.text.length);
+                        if (quad === undefined) {
+                            if (options.requireAllMatches)
+                                throw new AppError('TextIn 脱敏敏感词缺少坐标', 502);
                             continue;
+                        }
+                        if (options.requireAllMatches) {
+                            const scaled = scaleQuadToImage(quad, scaleX, scaleY);
+                            const xs = scaled.map(point => point[0]);
+                            const ys = scaled.map(point => point[1]);
+                            const area = Math.abs(scaled.reduce((sum, [x, y], index) => {
+                                const next = scaled[(index + 1) % scaled.length];
+                                return sum + x * next[1] - next[0] * y;
+                            }, 0));
+                            if (area < 1 || scaled.flat().some(n => !Number.isFinite(n)) ||
+                                Math.min(...xs) < 0 || Math.min(...ys) < 0 ||
+                                Math.max(...xs) > imageWidth || Math.max(...ys) > imageHeight ||
+                                Math.max(...xs) <= Math.min(...xs) || Math.max(...ys) <= Math.min(...ys)) {
+                                throw new AppError('TextIn 脱敏坐标无效', 502);
+                            }
+                        }
+                        located++;
                         finalQuads.push(clampQuad(scaleQuad(scaleQuadToImage(quad, scaleX, scaleY), SAFETY_SCALE), imageWidth, imageHeight));
                         const count = (counters.get(word.target) ?? 0) + 1;
                         counters.set(word.target, count);
@@ -187,6 +219,9 @@ export const createTextInMasker = (options) => {
                         });
                     }
                 }
+                if (options.requireAllMatches && located === 0) {
+                    throw new AppError('TextIn 脱敏敏感词无法在单行定位', 502);
+                }
             }
             const overlay = polygonSvg(imageWidth, imageHeight, finalQuads);
             const uprightMasked = await sharp(upright.data)

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@onco-foundry/mask-port",
-  "version": "0.1.2",
+  "version": "0.1.4",
   "type": "module",
   "files": [
     "dist",
@@ -18,8 +18,8 @@
     "zod": "^4.4.3",
     "@onco-foundry/capability-registry": "0.1.1",
     "@onco-foundry/errors": "0.1.1",
-    "@onco-foundry/resource-versioning": "0.2.0",
-    "@onco-foundry/trace-port": "0.3.2"
+    "@onco-foundry/trace-port": "0.3.2",
+    "@onco-foundry/resource-versioning": "0.3.0"
   },
   "publishConfig": {
     "access": "public"