kakaotalk-chat-analyzer 0.2.19 → 0.2.21

package/dist/src/aggregator.d.ts

@@ -19,10 +19,17 @@ export declare class ReportAggregator {
     private readonly attachments;
     private readonly domains;
     private readonly keywordCounter;
+    private readonly repeatPhraseCounter;
+    private readonly shopSearchTopics;
     private readonly gapStats;
     private readonly dailySenderCounts;
     private readonly laughBySender;
     private readonly shortBySender;
+    private readonly dailyJoin;
+    private readonly dailyLeave;
+    private readonly dailyHidden;
+    private readonly dailyKick;
+    private readonly dailyNewSenders;
     private total;
     private totalCharacters;
     private messagesWithLinks;
@@ -38,6 +45,15 @@ export declare class ReportAggregator {
     private roomJoinMessages;
     private roomLeaveMessages;
     private roomDeletedMessages;
+    private roomHiddenMessages;
+    private roomKickMessages;
+    private roomSlowOnMessages;
+    private roomSlowOffMessages;
+    private roomSubManagerMessages;
+    private roomManagerMessages;
+    private roomShopSearchMessages;
+    private roomPhotoBundleMessages;
+    private pureLaughMessages;
     private prevMs;
     private prevSender;
     private runSender;
@@ -46,5 +62,6 @@ export declare class ReportAggregator {
     private lastDate;
     constructor(filePath: string, privacy: PrivacyMode, top: number);
     consume(record: ChatRecord): void;
+    private bumpSystemNotice;
     finalize(meta: FinalizeSourceMeta): ReportData;
 }

package/dist/src/aggregator.js

@@ -2,7 +2,9 @@ import { formatDate, formatDateTime, partsToUtcMs, weekdayIndex } from "./date.j
 import { maskPartialDisplayName, parseChatRoomNameFromExportPath, safeInputName } from "./analysis-labels.js";
 import { GapStreamStats } from "./gap-stats.js";
 import { KeywordCounter } from "./keyword-counter.js";
-import { detectSystemNotice, ROOM_EVENT_KEYWORD_STOP } from "./room-events.js";
+import { RepeatPhraseCounter } from "./repeat-phrase-counter.js";
+import { buildRoomPulse, computeActivityArc, computeBurstDays, computeConversationPace, } from "./report-enrichment.js";
+import { isOpenChatBoilerplate, splitMessageForAnalysis, SYSTEM_NOTICE_KEYWORD_STOP, } from "./system-notices.js";
 import { buildReportStory } from "./story.js";
 const ATTACHMENT_MARKERS = [
     "사진",
@@ -16,7 +18,9 @@ const ATTACHMENT_MARKERS = [
     "음성메시지",
     "삭제된 메시지",
 ];
-const KEYWORD_EXCLUDE = new Set([...ATTACHMENT_MARKERS, ...ROOM_EVENT_KEYWORD_STOP]);
+const KEYWORD_EXCLUDE = new Set([...ATTACHMENT_MARKERS, ...SYSTEM_NOTICE_KEYWORD_STOP]);
+const PHOTO_BUNDLE_RE = /^사진\s+\d+\s*장$/;
+const PURE_LAUGH_RE = /^[ㅋㅎㅠㅜ]+$/u;
 const WEEKDAY_LABELS_KO = ["일", "월", "화", "수", "목", "금", "토"];
 const URL_RE = /\bhttps?:\/\/[^\s<>"']+|www\.[^\s<>"']+/gi;
 const EMAIL_RE = /\b[A-Z0-9._%+-]+@[A-Z0-9.-]+\.[A-Z]{2,}\b/gi;
@@ -53,6 +57,13 @@ const STOPWORDS = new Set([
     "from",
     "http",
     "https",
+    "정치성향",
+    "강퇴됩니다",
+    "가려짐",
+    "초중반",
+    "비속어",
+    "반가워",
+    "닉네임",
 ]);
 const NIGHT_HOURS = new Set([23, 0, 1, 2, 3, 4, 5]);
 const EMOJI_RE = /\p{Extended_Pictographic}/u;
@@ -73,10 +84,17 @@ export class ReportAggregator {
     attachments = new Map();
     domains = new Map();
     keywordCounter = new KeywordCounter();
+    repeatPhraseCounter = new RepeatPhraseCounter();
+    shopSearchTopics = new Map();
     gapStats = new GapStreamStats();
     dailySenderCounts = new Map();
     laughBySender = new Map();
     shortBySender = new Map();
+    dailyJoin = new Map();
+    dailyLeave = new Map();
+    dailyHidden = new Map();
+    dailyKick = new Map();
+    dailyNewSenders = new Map();
     total = 0;
     totalCharacters = 0;
     messagesWithLinks = 0;
@@ -92,6 +110,15 @@ export class ReportAggregator {
     roomJoinMessages = 0;
     roomLeaveMessages = 0;
     roomDeletedMessages = 0;
+    roomHiddenMessages = 0;
+    roomKickMessages = 0;
+    roomSlowOnMessages = 0;
+    roomSlowOffMessages = 0;
+    roomSubManagerMessages = 0;
+    roomManagerMessages = 0;
+    roomShopSearchMessages = 0;
+    roomPhotoBundleMessages = 0;
+    pureLaughMessages = 0;
     prevMs = null;
     prevSender = null;
     runSender = null;
@@ -107,20 +134,21 @@ export class ReportAggregator {
         if (this.prevSender !== null && record.sender !== this.prevSender) {
             this.speakerSwitches += 1;
         }
+        const dayKey = formatDate(record.date);
         const stat = getParticipantStat(this.senderStats, record.sender);
         if (!this.sendersRegistered.has(record.sender)) {
             this.sendersRegistered.add(record.sender);
             this.senderNamesNormalized.add(normalizeToken(record.sender));
+            increment(this.dailyNewSenders, dayKey);
         }
-        const msg = record.message;
+        const split = splitMessageForAnalysis(record.message);
+        for (const kind of split.notices)
+            this.bumpSystemNotice(kind, dayKey);
+        for (const tag of split.shopSearchTags)
+            increment(this.shopSearchTopics, tag);
+        const msg = split.userText.length > 0 ? split.userText : record.message;
         const messageLength = msg.length;
-        const systemNotice = detectSystemNotice(msg);
-        if (systemNotice === "join")
-            this.roomJoinMessages += 1;
-        if (systemNotice === "leave")
-            this.roomLeaveMessages += 1;
-        if (systemNotice === "deleted")
-            this.roomDeletedMessages += 1;
+        const isPureSystem = split.notices.length > 0 && split.userText.length === 0;
         const foundAttachments = getAttachmentMarkers(msg);
         const foundDomains = LINK_HINT_RE.test(msg) ? getDomains(msg) : [];
         const ms = partsToUtcMs(record.date);
@@ -128,75 +156,81 @@ export class ReportAggregator {
             this.firstDate = record.date;
         this.lastDate = record.date;
         this.total += 1;
-        if (messageLength > 0 && EMOJI_RE.test(msg)) {
-            this.emojiMessages += 1;
-        }
-        if (messageLength > 0 && LAUGH_RE.test(msg)) {
-            this.laughMessages += 1;
-            increment(this.laughBySender, record.sender);
-        }
-        const trimmed = msg.trim();
-        if (trimmed.length > 0 && trimmed.length <= 3) {
-            this.shortMessages += 1;
-            increment(this.shortBySender, record.sender);
-        }
-        if (msg.includes("?") || msg.includes("？")) {
-            this.questionMessages += 1;
-        }
         const wi = weekdayIndex(record.date);
-        if (wi === 0 || wi === 6) {
-            this.weekendMessages += 1;
-        }
-        if (NIGHT_HOURS.has(record.date.hour)) {
-            this.nightMessages += 1;
-            stat.nightMessages += 1;
-        }
-        if (this.prevMs !== null) {
-            const delta = ms - this.prevMs;
-            this.gapStats.add(delta);
-        }
-        this.prevMs = ms;
-        if (record.sender === this.prevSender) {
-            this.runLen += 1;
-            if (this.runLen >= 3) {
-                this.monologueMessages += 1;
+        if (!isPureSystem) {
+            if (messageLength > 0 && EMOJI_RE.test(msg)) {
+                this.emojiMessages += 1;
             }
-        }
-        else {
-            if (this.prevSender !== null && this.runSender !== null) {
-                const prevStat = getParticipantStat(this.senderStats, this.prevSender);
-                prevStat.maxConsecutive = Math.max(prevStat.maxConsecutive, this.runLen);
+            if (messageLength > 0 && LAUGH_RE.test(msg)) {
+                this.laughMessages += 1;
+                increment(this.laughBySender, record.sender);
             }
-            this.runSender = record.sender;
-            this.runLen = 1;
-        }
-        this.prevSender = record.sender;
-        stat.messages += 1;
-        stat.characters += messageLength;
-        this.totalCharacters += messageLength;
-        if (foundAttachments.length > 0) {
-            stat.attachmentMessages += 1;
-            this.messagesWithAttachments += 1;
-            for (const marker of foundAttachments)
-                increment(this.attachments, marker);
-        }
-        if (foundDomains.length > 0) {
-            stat.linkMessages += 1;
-            this.messagesWithLinks += 1;
-            for (const domain of foundDomains)
-                increment(this.domains, domain);
-        }
-        if (systemNotice === null &&
-            messageLength >= 2 &&
-            HAS_TOKEN_CHAR_RE.test(msg) &&
-            shouldExtractKeywords(msg, foundAttachments)) {
-            for (const keyword of extractKeywords(msg, this.senderNamesNormalized)) {
-                this.keywordCounter.add(keyword);
+            if (messageLength > 0 && PURE_LAUGH_RE.test(msg.trim())) {
+                this.pureLaughMessages += 1;
+            }
+            const trimmed = msg.trim();
+            if (trimmed.length > 0 && trimmed.length <= 3) {
+                this.shortMessages += 1;
+                increment(this.shortBySender, record.sender);
+            }
+            if (msg.includes("?") || msg.includes("？")) {
+                this.questionMessages += 1;
+            }
+            if (wi === 0 || wi === 6) {
+                this.weekendMessages += 1;
+            }
+            if (NIGHT_HOURS.has(record.date.hour)) {
+                this.nightMessages += 1;
+                stat.nightMessages += 1;
+            }
+            if (this.prevMs !== null) {
+                const delta = ms - this.prevMs;
+                this.gapStats.add(delta);
+            }
+            this.prevMs = ms;
+            if (record.sender === this.prevSender) {
+                this.runLen += 1;
+                if (this.runLen >= 3) {
+                    this.monologueMessages += 1;
+                }
+            }
+            else {
+                if (this.prevSender !== null && this.runSender !== null) {
+                    const prevStat = getParticipantStat(this.senderStats, this.prevSender);
+                    prevStat.maxConsecutive = Math.max(prevStat.maxConsecutive, this.runLen);
+                }
+                this.runSender = record.sender;
+                this.runLen = 1;
+            }
+            this.prevSender = record.sender;
+            stat.messages += 1;
+            stat.characters += messageLength;
+            this.totalCharacters += messageLength;
+            if (foundAttachments.length > 0) {
+                stat.attachmentMessages += 1;
+                this.messagesWithAttachments += 1;
+                for (const marker of foundAttachments)
+                    increment(this.attachments, marker);
+            }
+            if (foundDomains.length > 0) {
+                stat.linkMessages += 1;
+                this.messagesWithLinks += 1;
+                for (const domain of foundDomains)
+                    increment(this.domains, domain);
+            }
+            if (messageLength >= 2 &&
+                HAS_TOKEN_CHAR_RE.test(msg) &&
+                !isOpenChatBoilerplate(msg) &&
+                shouldExtractKeywords(msg, foundAttachments)) {
+                for (const keyword of extractKeywords(msg, this.senderNamesNormalized)) {
+                    this.keywordCounter.add(keyword);
+                }
+                if (messageLength >= 12 && !isOpenChatBoilerplate(msg))
+                    this.repeatPhraseCounter.add(msg);
             }
         }
-        const dayKey = formatDate(record.date);
         increment(this.daily, dayKey);
-        {
+        if (!isPureSystem) {
             let perDay = this.dailySenderCounts.get(dayKey);
             if (!perDay) {
                 perDay = new Map();
@@ -208,6 +242,49 @@ export class ReportAggregator {
         this.hourly[record.date.hour] = (this.hourly[record.date.hour] ?? 0) + 1;
         this.weekdays[wi] = (this.weekdays[wi] ?? 0) + 1;
     }
+    bumpSystemNotice(kind, dayKey) {
+        switch (kind) {
+            case "join":
+                this.roomJoinMessages += 1;
+                increment(this.dailyJoin, dayKey);
+                break;
+            case "leave":
+                this.roomLeaveMessages += 1;
+                increment(this.dailyLeave, dayKey);
+                break;
+            case "deleted":
+                this.roomDeletedMessages += 1;
+                break;
+            case "hidden":
+                this.roomHiddenMessages += 1;
+                increment(this.dailyHidden, dayKey);
+                break;
+            case "kick":
+                this.roomKickMessages += 1;
+                increment(this.dailyKick, dayKey);
+                break;
+            case "slowModeOn":
+                this.roomSlowOnMessages += 1;
+                break;
+            case "slowModeOff":
+                this.roomSlowOffMessages += 1;
+                break;
+            case "subManager":
+                this.roomSubManagerMessages += 1;
+                break;
+            case "manager":
+                this.roomManagerMessages += 1;
+                break;
+            case "shopSearch":
+                this.roomShopSearchMessages += 1;
+                break;
+            case "photoBundle":
+                this.roomPhotoBundleMessages += 1;
+                break;
+            default:
+                break;
+        }
+    }
     finalize(meta) {
         if (this.prevSender !== null && this.runSender !== null) {
             const prevStat = getParticipantStat(this.senderStats, this.prevSender);
@@ -303,6 +380,7 @@ export class ReportAggregator {
         const uniqueDomainCount = this.domains.size;
         const replyGapCoeffVariation = this.gapStats.coeffVariation();
         const monologueMessagesPercent = total > 0 ? round((this.monologueMessages / total) * 100, 1) : 0;
+        const lexicalTypeRichnessPercent = this.keywordCounter.typeTokenRichnessPercent();
         const insights = {
             weekendSharePercent,
             participantGini,
@@ -327,8 +405,13 @@ export class ReportAggregator {
             peakDaySharePercent,
             uniqueDomainCount,
             replyGapCoeffVariation,
+            lexicalTypeRichnessPercent,
         };
         const dailySorted = [...this.daily.entries()].map(([date, count]) => ({ date, count })).sort((a, b) => a.date.localeCompare(b.date));
+        const burstDays = computeBurstDays(dailySorted);
+        const activityArc = computeActivityArc(dailySorted);
+        const conversationPace = computeConversationPace(insights);
+        const roomPulse = buildRoomPulse(sortedDays, this.dailyJoin, this.dailyLeave, this.dailyHidden, this.dailyKick, this.dailyNewSenders);
         const laughByAlias = new Map();
         const shortByAlias = new Map();
         for (const [raw, c] of this.laughBySender) {
@@ -361,6 +444,10 @@ export class ReportAggregator {
             shortMessages: this.shortMessages,
             laughBySender: laughByAlias,
             shortBySender: shortByAlias,
+            burstDays,
+            activityArc,
+            conversationPace,
+            roomPulse,
         });
         const highlights = buildHighlights({
             total,
@@ -383,6 +470,16 @@ export class ReportAggregator {
             roomJoinMessages: this.roomJoinMessages,
             roomLeaveMessages: this.roomLeaveMessages,
             roomDeletedMessages: this.roomDeletedMessages,
+            roomHiddenMessages: this.roomHiddenMessages,
+            roomKickMessages: this.roomKickMessages,
+            pureLaughMessages: this.pureLaughMessages,
+            repeatedPhraseCount: this.repeatPhraseCounter.top(1, 3)[0]?.count ?? 0,
+            burstDays,
+            activityArc,
+            conversationPace,
+            roomPulse,
+            lexicalTypeRichnessPercent,
+            speakerSwitchRatePer100,
         });
         return {
             generatedAt: new Date().toISOString(),
@@ -423,15 +520,26 @@ export class ReportAggregator {
             attachments: topCounts(this.attachments, this.top),
             domains: topCounts(this.domains, this.top),
             keywords: this.keywordCounter.topCounts(this.top),
-            roomEvents: {
-                joinCount: this.roomJoinMessages,
-                leaveCount: this.roomLeaveMessages,
-                deletedCount: this.roomDeletedMessages,
-                total: this.roomJoinMessages + this.roomLeaveMessages + this.roomDeletedMessages,
-                joinSharePercent: total > 0 ? round((this.roomJoinMessages / total) * 100, 2) : 0,
-                leaveSharePercent: total > 0 ? round((this.roomLeaveMessages / total) * 100, 2) : 0,
-                deletedSharePercent: total > 0 ? round((this.roomDeletedMessages / total) * 100, 2) : 0,
-            },
+            roomEvents: buildRoomEventStats(total, {
+                join: this.roomJoinMessages,
+                leave: this.roomLeaveMessages,
+                deleted: this.roomDeletedMessages,
+                hidden: this.roomHiddenMessages,
+                kick: this.roomKickMessages,
+                slowModeOn: this.roomSlowOnMessages,
+                slowModeOff: this.roomSlowOffMessages,
+                subManager: this.roomSubManagerMessages,
+                manager: this.roomManagerMessages,
+                shopSearch: this.roomShopSearchMessages,
+                photoBundle: this.roomPhotoBundleMessages,
+            }),
+            repeatedPhrases: this.repeatPhraseCounter.top(8, 3),
+            shopSearchTopics: topCounts(this.shopSearchTopics, 10),
+            pureLaughMessages: this.pureLaughMessages,
+            conversationPace,
+            burstDays,
+            activityArc,
+            roomPulse,
             highlights,
             story,
         };
@@ -486,7 +594,45 @@ function shouldExtractKeywords(message, attachmentMarkers) {
     return true;
 }
 function getAttachmentMarkers(message) {
-    return ATTACHMENT_MARKERS.filter((marker) => message.includes(marker));
+    const found = ATTACHMENT_MARKERS.filter((marker) => message.includes(marker));
+    const t = message.trim();
+    if (PHOTO_BUNDLE_RE.test(t) && !found.includes("사진")) {
+        found.push("사진");
+    }
+    return found;
+}
+function buildRoomEventStats(total, c) {
+    const sum = c.join +
+        c.leave +
+        c.deleted +
+        c.hidden +
+        c.kick +
+        c.slowModeOn +
+        c.slowModeOff +
+        c.subManager +
+        c.manager +
+        c.shopSearch +
+        c.photoBundle;
+    const pct = (n) => (total > 0 ? round((n / total) * 100, 2) : 0);
+    return {
+        joinCount: c.join,
+        leaveCount: c.leave,
+        deletedCount: c.deleted,
+        hiddenCount: c.hidden,
+        kickCount: c.kick,
+        slowModeOnCount: c.slowModeOn,
+        slowModeOffCount: c.slowModeOff,
+        subManagerCount: c.subManager,
+        managerCount: c.manager,
+        shopSearchCount: c.shopSearch,
+        photoBundleCount: c.photoBundle,
+        total: sum,
+        joinSharePercent: pct(c.join),
+        leaveSharePercent: pct(c.leave),
+        deletedSharePercent: pct(c.deleted),
+        hiddenSharePercent: pct(c.hidden),
+        kickSharePercent: pct(c.kick),
+    };
 }
 function getDomains(message) {
     const matches = message.match(URL_RE) ?? [];
@@ -714,15 +860,72 @@ function buildHighlights(input) {
     if (input.monologueMessagesPercent >= 25) {
         out.push(`같은 사람 **3연속 이상** 메시지가 전체의 **${input.monologueMessagesPercent}%** — 긴 설명·정리 구간이 잦을 수 있어요.`);
     }
-    const sysTotal = input.roomJoinMessages + input.roomLeaveMessages + input.roomDeletedMessages;
+    const sysTotal = input.roomJoinMessages +
+        input.roomLeaveMessages +
+        input.roomDeletedMessages +
+        input.roomHiddenMessages +
+        input.roomKickMessages;
     if (sysTotal > 0) {
         const parts = [
             input.roomJoinMessages > 0 ? `들어옴 ${input.roomJoinMessages}` : "",
             input.roomLeaveMessages > 0 ? `나감 ${input.roomLeaveMessages}` : "",
-            input.roomDeletedMessages > 0 ? `삭제 알림 ${input.roomDeletedMessages}` : "",
+            input.roomDeletedMessages > 0 ? `삭제 ${input.roomDeletedMessages}` : "",
+            input.roomHiddenMessages > 0 ? `가림 ${input.roomHiddenMessages}` : "",
+            input.roomKickMessages > 0 ? `강퇴 ${input.roomKickMessages}` : "",
         ].filter(Boolean);
-        out.push(`카카오톡 **시스템 알림** **${sysTotal}**건(${parts.join(" · ")}) — 키워드와는 따로 집계했어요.`);
+        out.push(`카카오톡 **시스템·운영 알림** **${sysTotal}**건(${parts.join(" · ")}) — 본문 키워드와 분리했어요.`);
+    }
+    if (input.pureLaughMessages > 0) {
+        out.push(`**ㅋㅋ만** 보낸 리액션 메시지는 **${input.pureLaughMessages}**건이에요.`);
     }
-    return out.slice(0, 12);
+    if (input.repeatedPhraseCount >= 10) {
+        out.push(`같은 문장이 **${input.repeatedPhraseCount}회** 반복된 카피페asta급 문구도 있어요.`);
+    }
+    if (input.burstDays.length > 0) {
+        const top = input.burstDays[0];
+        const labels = input.burstDays
+            .slice(0, 3)
+            .map((d) => formatDayMdHighlight(d.date))
+            .join(" · ");
+        out.push(`메시지가 평소보다 몰린 날 **${input.burstDays.length}일** — 최고는 **${formatDayMdHighlight(top.date)}**(${formatCompactCount(top.count)}건). ${labels}`);
+    }
+    const head = input.activityArc.find((a) => a.id === "head");
+    const tail = input.activityArc.find((a) => a.id === "tail");
+    if (head && tail && head.messages > 0 && tail.messages > 0) {
+        const ratio = round(tail.messages / head.messages, 2);
+        if (ratio >= 1.25) {
+            out.push(`마지막 7일이 처음 7일보다 **${ratio}배** 활발 — 대화가 뜨거워지는 구간이 있었어요.`);
+        }
+        else if (ratio <= 0.8) {
+            out.push(`처음 7일이 마지막보다 더 붐볐어요(후반은 처음의 **${Math.round(ratio * 100)}%** 수준).`);
+        }
+    }
+    if (input.lexicalTypeRichnessPercent !== null && input.lexicalTypeRichnessPercent >= 18) {
+        out.push(`본문 단어는 **${input.lexicalTypeRichnessPercent}%** 정도로 서로 다른 표현이 많이 섞였어요.`);
+    }
+    const pace = input.conversationPace;
+    out.push(`대화 템포는 **${pace.emoji} ${pace.label}** — ${pace.detail}`);
+    const peakHidden = [...input.roomPulse].sort((a, b) => b.hidden - a.hidden)[0];
+    if (peakHidden && peakHidden.hidden >= 5) {
+        out.push(`가림 알림이 가장 많았던 날은 **${formatDayMdHighlight(peakHidden.date)}**(${peakHidden.hidden}건)이에요.`);
+    }
+    const peakJoin = [...input.roomPulse].sort((a, b) => b.join - a.join)[0];
+    if (peakJoin && peakJoin.join >= 20) {
+        out.push(`입장이 가장 몰린 날은 **${formatDayMdHighlight(peakJoin.date)}** — **${peakJoin.join}**명 들어왔어요.`);
+    }
+    return out.slice(0, 14);
+}
+function formatDayMdHighlight(ymd) {
+    const p = ymd.split("-");
+    if (p.length === 3)
+        return `${Number(p[1])}/${Number(p[2])}`;
+    return ymd;
+}
+function formatCompactCount(n) {
+    if (n >= 10_000)
+        return `${Math.round(n / 1000) / 10}만`;
+    if (n >= 1000)
+        return `${(n / 1000).toFixed(1)}k`;
+    return String(n);
 }
 //# sourceMappingURL=aggregator.js.map