@kk-irving/knowledge-mcp-server 1.0.0 → 1.0.1

package/dist/sources/confluence-sync.js

@@ -90,17 +90,41 @@ async function listAllSpaces() {
 function sleep(ms) {
     return new Promise((r) => setTimeout(r, ms));
 }
+/**
+ * 把 ISO 时间戳（含 T、Z）规范化为 CQL 可接受的 `YYYY-MM-DD HH:mm`。
+ * Atlassian CQL parser 要求 day-or-minute 精度，不支持秒、毫秒、Z 后缀。
+ *
+ * @example
+ *   "2026-06-22T10:30:45.123Z"  →  "2026-06-22 10:30"
+ *   "2026-06-22"                →  "2026-06-22 00:00"
+ */
+function toCqlDateTime(s) {
+    if (!s)
+        return "";
+    const m = s.match(/^(\d{4})-(\d{2})-(\d{2})(?:[T ](\d{2}):(\d{2}))?/);
+    if (!m)
+        return s;
+    const [, y, mo, d, hh = "00", mm = "00"] = m;
+    return `${y}-${mo}-${d} ${hh}:${mm}`;
+}
 /**
  * 全量/增量同步 Confluence pages。
  *
  * @param args.space  仅同步指定空间（可重复传入多个空间用 CSV，如 "TVENG,DOC"）
- * @param args.limit  最大同步条数（防一次性拉爆）。默认 1000
- * @param args.since  仅同步 lastmodified > since 的页面（YYYY-MM-DD HH:MM）
+ * @param args.limit  最大同步条数（防一次性拉爆）。默认 1000；传 0 / 负数 = 不限
+ * @param args.since  仅同步 lastmodified > since 的页面（YYYY-MM-DD HH:mm）
  */
 export async function syncConfluence(args = {}) {
     const db = getDb();
-    const limit = Math.max(1, Math.min(50000, args.limit ?? 1000));
-    const stateSince = args.since ?? getSyncState("confluence", "last_full_sync") ?? "";
+    // limit ≤ 0 → 无限拉
+    const userLimit = args.limit;
+    const limit = userLimit == null
+        ? 1000
+        : userLimit <= 0
+            ? Number.MAX_SAFE_INTEGER
+            : userLimit;
+    const stateSinceRaw = args.since ?? getSyncState("confluence", "last_full_sync") ?? "";
+    const stateSince = stateSinceRaw ? toCqlDateTime(stateSinceRaw) : "";
     // 决定要同步的空间集合
     let spaces;
     if (args.space) {
@@ -118,15 +142,17 @@ export async function syncConfluence(args = {}) {
     }
     let fetched = 0;
     let upserted = 0;
+    let maxUpdated = stateSinceRaw;
     const pageSize = 100;
     for (const spaceKey of spaces) {
         let start = 0;
+        let pageInSpace = 0;
         while (fetched < limit) {
             const remaining = limit - fetched;
             const n = Math.min(pageSize, remaining);
             let resp;
             if (stateSince) {
-                // 走 CQL 增量
+                // 走 CQL 增量（lastmodified 用 day/minute 精度格式）
                 const cql = `space.key = "${spaceKey}" AND type = page AND lastmodified > "${stateSince}"`;
                 resp = await confluenceGet("/rest/api/content/search", {
                     cql,
@@ -161,6 +187,13 @@ export async function syncConfluence(args = {}) {
             fetched += items.length;
             upserted += items.length;
             start += items.length;
+            pageInSpace++;
+            // 跟踪最大 updated（用于 watermark）
+            for (const r of rows) {
+                if (r.updated && r.updated > maxUpdated)
+                    maxUpdated = r.updated;
+            }
+            process.stderr.write(`[confluence-sync] space=${spaceKey} page=${pageInSpace}, got=${items.length}, total=${fetched}\n`);
             if (items.length < n)
                 break;
             await sleep(CONFLUENCE_REQUEST_DELAY_MS);
@@ -168,7 +201,8 @@ export async function syncConfluence(args = {}) {
         if (fetched >= limit)
             break;
     }
-    const watermark = new Date().toISOString();
+    // watermark：用已拉取的最大 updated 时间的 CQL 格式（YYYY-MM-DD HH:mm）
+    const watermark = maxUpdated ? toCqlDateTime(maxUpdated) : stateSince || toCqlDateTime(new Date().toISOString());
     setSyncState("confluence", "last_full_sync", watermark);
     return {
         source: "confluence",

package/dist/sources/gerrit-sync.js

@@ -106,22 +106,48 @@ ON CONFLICT(change_id) DO UPDATE SET
 `;
 function buildQuery(args) {
     const parts = [];
-    if (args.query)
+    if (args.query) {
+        // 用户传了自定义 query，完全用它，不再叠加默认作用域
         parts.push(args.query);
+    }
+    else {
+        // 没传自定义 query → 总是加默认作用域过滤（避免拉到所有人的 changes）
+        // 首次（无 since）时叠加 -age:365d 限制时间窗口；增量（有 since）时不需要 -age
+        // 用户可通过 KNOWLEDGE_GERRIT_SYNC_QUERY 环境变量自定义默认作用域
+        const customDefault = (process.env.KNOWLEDGE_GERRIT_SYNC_QUERY ?? "").trim();
+        if (customDefault) {
+            parts.push(customDefault);
+        }
+        else if (args.since) {
+            parts.push("(owner:self OR reviewer:self)");
+        }
+        else {
+            parts.push("(owner:self OR reviewer:self) -age:365d");
+        }
+    }
     if (args.project)
         parts.push(`project:${args.project}`);
     if (args.since)
         parts.push(`after:"${args.since}"`);
-    if (parts.length === 0)
-        parts.push("status:open OR -status:open"); // all
     return parts.join(" ");
 }
 /**
  * 全量/增量同步 Gerrit changes。
+ *
+ * @param args.query   自定义 query；缺省走 `KNOWLEDGE_GERRIT_SYNC_QUERY` 或 `(owner:self OR reviewer:self) -age:365d`
+ * @param args.project 仅同步特定 project
+ * @param args.since   仅同步 after:since 的 changes（YYYY-MM-DD）
+ * @param args.limit   最大同步条数（防一次性拉爆）。默认 1000；传 0 / 负数 = 不限
  */
 export async function syncGerrit(args = {}) {
     const db = getDb();
-    const limit = Math.max(1, Math.min(50000, args.limit ?? 1000));
+    // limit ≤ 0 → 无限拉
+    const userLimit = args.limit;
+    const limit = userLimit == null
+        ? 1000
+        : userLimit <= 0
+            ? Number.MAX_SAFE_INTEGER
+            : userLimit;
     const stateSince = args.since ?? getSyncState("gerrit", "last_full_sync_date") ?? "";
     const finalQuery = buildQuery({
         since: stateSince,
@@ -138,6 +164,7 @@ export async function syncGerrit(args = {}) {
     let offset = 0;
     let fetched = 0;
     let upserted = 0;
+    let maxUpdated = stateSince;
     const batchSize = 200;
     while (fetched < limit) {
         const remaining = limit - fetched;
@@ -169,12 +196,21 @@ export async function syncGerrit(args = {}) {
         fetched += changes.length;
         upserted += changes.length;
         offset += changes.length;
+        // 跟踪最大 updated（Gerrit 返回 ISO 格式："2026-06-22 10:30:45.123000000"）
+        for (const r of rows) {
+            if (r.updated && r.updated > maxUpdated)
+                maxUpdated = r.updated;
+        }
+        process.stderr.write(`[gerrit-sync] page offset=${offset - changes.length}, got=${changes.length}, total=${fetched}\n`);
         if (changes.length < n)
             break;
         if (!changes[changes.length - 1]?._more_changes)
             break;
     }
-    const today = new Date().toISOString().slice(0, 10);
-    setSyncState("gerrit", "last_full_sync_date", today);
-    return { source: "gerrit", fetched, upserted, watermark: today, query: finalQuery };
+    // watermark：用已拉取的最大 updated 的 day-only 形式（Gerrit `after:` 接受 YYYY-MM-DD）
+    const watermark = maxUpdated
+        ? maxUpdated.slice(0, 10)
+        : stateSince || new Date().toISOString().slice(0, 10);
+    setSyncState("gerrit", "last_full_sync_date", watermark);
+    return { source: "gerrit", fetched, upserted, watermark, query: finalQuery };
 }

package/dist/sources/zmind-sync.js

@@ -48,12 +48,18 @@ ON CONFLICT(id) DO UPDATE SET
  * 全量/增量同步 Zmind issues。
  *
  * @param args.since  仅同步 updated_on >= since 的 issues（YYYY-MM-DD）。未传则用 sync_state 水位。
- * @param args.limit  最大同步条数（防一次性拉爆）。默认 1000。
+ * @param args.limit  最大同步条数（防一次性拉爆）。默认 1000；传 0 / 负数 = 不限（拉到 API 返回空）
  * @param args.statusId 状态过滤；默认 "*" 全部
  */
 export async function syncZmind(args = {}) {
     const db = getDb();
-    const limit = Math.max(1, Math.min(50000, args.limit ?? 1000));
+    // limit ≤ 0 → 无限拉（直到 API 返回空）
+    const userLimit = args.limit;
+    const limit = userLimit == null
+        ? 1000
+        : userLimit <= 0
+            ? Number.MAX_SAFE_INTEGER
+            : userLimit;
     const stateSince = args.since ?? getSyncState("zmind", "last_full_sync") ?? "";
     const upsert = db.prepare(UPSERT_SQL);
     function upsertMany(rows) {
@@ -65,6 +71,7 @@ export async function syncZmind(args = {}) {
     let offset = 0;
     let fetched = 0;
     let upserted = 0;
+    let maxUpdatedOn = stateSince; // 用于设置 watermark = 已拉取的最大 updated_on
     const pageSize = 100;
     while (fetched < limit) {
         const remaining = limit - fetched;
@@ -75,8 +82,9 @@ export async function syncZmind(args = {}) {
             offset: String(offset),
             limit: String(batch),
         };
+        // Redmine 接受 `>=YYYY-MM-DD` 格式；ISO 时间串会解析失败 → 强制截到 day
         if (stateSince)
-            params.updated_on = `>=${stateSince}`;
+            params.updated_on = `>=${stateSince.slice(0, 10)}`;
         const data = await zmindGet("/issues.json", params);
         const issues = data.issues ?? [];
         if (issues.length === 0)
@@ -97,10 +105,21 @@ export async function syncZmind(args = {}) {
         fetched += issues.length;
         upserted += issues.length;
         offset += issues.length;
+        // 跟踪最大 updated_on（Zmind 返回 ISO 格式，截到 day 即可比较）
+        for (const r of rows) {
+            if (r.updated_on && r.updated_on > maxUpdatedOn) {
+                maxUpdatedOn = r.updated_on;
+            }
+        }
+        process.stderr.write(`[zmind-sync] page offset=${offset - issues.length}, got=${issues.length}, total=${fetched}\n`);
         if (issues.length < batch)
             break;
     }
-    const watermark = new Date().toISOString();
+    // watermark：用已拉取的最大 updated_on 的 day-only 形式（Redmine 接受 YYYY-MM-DD）
+    // 若没拉到任何数据，保留 stateSince（不前推）
+    const watermark = maxUpdatedOn
+        ? maxUpdatedOn.slice(0, 10)
+        : stateSince || new Date().toISOString().slice(0, 10);
     setSyncState("zmind", "last_full_sync", watermark);
     return { source: "zmind", fetched, upserted, watermark };
 }

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@kk-irving/knowledge-mcp-server",
-  "version": "1.0.0",
-  "description": "Local knowledge base MCP Server: vector (BGE-small-zh ONNX) + FTS5 hybrid retrieval across Zmind PR / Gerrit changes / Confluence pages",
+  "version": "1.0.1",
+  "description": "Local knowledge base MCP Server: vector (BGE-small-zh ONNX) + FTS5 hybrid retrieval across Zmind PR / Gerrit changes / Confluence pages; v1.0.1 fixes sync_gerrit default query (was 0-hit due to OR precedence)",
   "type": "module",
   "bin": {
     "knowledge-mcp-server": "dist/index.js"