ylrz_saas_pharmacy/java/fs-service/scripts/fix_customer_facing_splitter_utf8.mjs

import { writeFileSync } from 'node:fs';
import { dirname, join } from 'node:path';
import { fileURLToPath } from 'node:url';

const scriptDir = dirname(fileURLToPath(import.meta.url));
const root = join(scriptDir, '../../..');

function u(s) {
  return s.replace(/\\u([0-9a-fA-F]{4})/g, (_, hex) => String.fromCharCode(parseInt(hex, 16)));
}

const content = u(`package com.fs.company.service.workflow.deepdialogue;

import java.util.ArrayList;
import java.util.LinkedHashSet;
import java.util.List;
import java.util.regex.Pattern;

/**
 * Split customer-facing text into: main answer first, then each question as its own message.
 */
public final class CustomerFacingMessageSplitter {

    private static final Pattern SENTENCE_SPLIT = Pattern.compile("[\\u3002\\uff01\\uff1f?!\\uff1b\\\\n]+");

    private CustomerFacingMessageSplitter() {
    }

    /**
     * @return non-empty list: index 0 = answer body; rest = follow-up / rhetorical questions
     */
    public static List<String> split(String reply, String plannedProbe, boolean shouldProbe) {
        List<String> out = new ArrayList<>();
        if (reply == null || reply.isBlank()) {
            appendUnique(out, plannedProbe, shouldProbe);
            return out;
        }

        String body = UniversalDeepDialogueService.stripProbeFromReply(reply, plannedProbe);
        List<String> answerParts = new ArrayList<>();
        List<String> questionParts = new ArrayList<>();

        for (String sentence : SENTENCE_SPLIT.split(body)) {
            String part = sentence != null ? sentence.trim() : "";
            if (part.isEmpty()) {
                continue;
            }
            if (isQuestionSentence(part)) {
                questionParts.add(normalizeQuestion(part));
            } else {
                answerParts.add(part);
            }
        }

        String answer = String.join("\\u3002", answerParts).trim();
        if (!answer.isEmpty() && !answer.endsWith("\\u3002") && !answer.endsWith("\\uff01")
                && !answer.endsWith("!") && !answer.endsWith("~") && !answer.endsWith("\\u2026")) {
            answer = answer + "\\u3002";
        }
        appendUnique(out, answer.isEmpty() ? null : answer.replaceAll("\\u3002+$", "\\u3002"), false);

        LinkedHashSet<String> seen = new LinkedHashSet<>();
        if (!out.isEmpty() && out.get(0) != null) {
            seen.add(out.get(0));
        }
        for (String q : questionParts) {
            appendUnique(out, q, false, seen);
        }
        appendUnique(out, plannedProbe, shouldProbe, seen);
        return out;
    }

    /** Keep declarative sentences only; questions are sent as separate follow-up messages. */
    public static String extractAnswerOnly(String reply) {
        if (reply == null || reply.isBlank()) {
            return reply;
        }
        List<String> parts = split(reply, null, false);
        if (parts.isEmpty()) {
            return reply.trim();
        }
        String answer = parts.get(0);
        return answer != null && !answer.isBlank() ? answer.trim() : reply.trim();
    }

    public static List<String> followUps(String reply, String plannedProbe, boolean shouldProbe) {
        List<String> all = split(reply, plannedProbe, shouldProbe);
        if (all.size() <= 1) {
            return List.of();
        }
        return new ArrayList<>(all.subList(1, all.size()));
    }

    private static void appendUnique(List<String> out, String text, boolean required) {
        appendUnique(out, text, required, null);
    }

    private static void appendUnique(List<String> out, String text, boolean required, LinkedHashSet<String> seen) {
        if (!required && (text == null || text.isBlank())) {
            return;
        }
        if (text == null || text.isBlank()) {
            return;
        }
        String normalized = text.trim();
        if (seen != null) {
            String key = normalized.length() > 12 ? normalized.substring(0, 12) : normalized;
            if (seen.contains(key) || seen.stream().anyMatch(s -> s != null && s.contains(key))) {
                return;
            }
            seen.add(key);
        }
        out.add(normalized);
    }

    static boolean isQuestionSentence(String sentence) {
        if (sentence == null || sentence.isBlank()) {
            return false;
        }
        String t = sentence.trim();
        if (t.endsWith("?") || t.endsWith("\\uff1f")) {
            return true;
        }
        return t.matches(".*(\\u5417|\\u5462|\\u5417\\uff1f|\\u5462\\uff1f|\\u4e48|\\u554a|\\u5427|\\u561b|\\u662f\\u4e0d\\u662f|\\u80fd\\u4e0d\\u80fd|\\u53ef\\u4e0d\\u53ef\\u4ee5|\\u8981\\u4e0d\\u8981|\\u6709\\u6ca1\\u6709|\\u591a\\u5c11|\\u51e0|\\u54ea|\\u8c01)[\\uff1f?]?$");
    }

    private static String normalizeQuestion(String sentence) {
        String t = sentence.trim();
        if (!t.endsWith("?") && !t.endsWith("\\uff1f")) {
            t = t + "\\uff1f";
        }
        return t;
    }
}
`);

for (const rel of [
  'java/fs-service/scripts/lobster_utf8_sources/CustomerFacingMessageSplitter.java',
  'java/fs-service/src/main/java/com/fs/company/service/workflow/deepdialogue/CustomerFacingMessageSplitter.java',
]) {
  writeFileSync(join(root, rel), content.replace(/\r\n/g, '\n'), 'utf8');
  console.log('fixed', rel);
}