bunnyquery 1.3.4 → 1.4.0

package/dist/engine.cjs

@@ -1,9 +1,57 @@
 'use strict';
 
+// src/engine/attachment_parsers.ts
+var MAX_PARSED_CONTENT_CHARS = 2e5;
+var _parsers = [];
+function registerAttachmentParser(parser) {
+  if (parser && typeof parser.match === "function" && typeof parser.parse === "function" && _parsers.indexOf(parser) === -1) {
+    _parsers.push(parser);
+  }
+}
+function clearAttachmentParsers() {
+  _parsers.length = 0;
+}
+function getAttachmentParsers() {
+  return _parsers.slice();
+}
+function findAttachmentParser(name, mime) {
+  for (let i = 0; i < _parsers.length; i++) {
+    try {
+      if (_parsers[i].match({ name, mime })) return _parsers[i];
+    } catch {
+    }
+  }
+  return void 0;
+}
+async function parseAttachmentContent(file, name, mime) {
+  const parser = findAttachmentParser(name, mime);
+  if (!parser) return null;
+  let raw;
+  try {
+    raw = await parser.parse(file);
+  } catch (err) {
+    console.error(
+      `[chat-engine] attachment parser ${parser.name || "(unnamed)"} failed for ${name}:`,
+      err
+    );
+    return null;
+  }
+  let text = (raw == null ? "" : String(raw)).trim();
+  if (!text) return null;
+  if (text.length > MAX_PARSED_CONTENT_CHARS) {
+    text = text.slice(0, MAX_PARSED_CONTENT_CHARS) + `
+...[truncated for length; original ${text.length} characters]`;
+  }
+  return text;
+}
+
 // src/engine/config.ts
 var _config = null;
 function configureChatEngine(config) {
   _config = config;
+  if (config.attachmentParsers) {
+    for (const parser of config.attachmentParsers) registerAttachmentParser(parser);
+  }
 }
 function chatEngineConfig() {
   if (!_config) {
@@ -30,13 +78,41 @@ var OFFICE_FILE_EXTENSIONS = /* @__PURE__ */ new Set([
   "pptx",
   "pptm",
   "hwp",
-  "hwpx"
+  "hwpx",
+  "ods",
+  "odt",
+  "odp",
+  "epub"
+]);
+var WEB_FETCHABLE_TEXT_EXTENSIONS = /* @__PURE__ */ new Set([
+  "csv",
+  "tsv",
+  "tab",
+  "txt",
+  "text",
+  "md",
+  "markdown",
+  "json",
+  "ndjson",
+  "jsonl",
+  "xml",
+  "yaml",
+  "yml",
+  "log",
+  // RTF is a TEXT format (not a binary zip), so web_fetch can read it and the
+  // model decodes its control words. Pin it here so a `.rtf` reported as
+  // `application/msword` isn't misrouted to server-side extraction (which has no
+  // .rtf extractor → "unsupported format" note).
+  "rtf",
+  "htm",
+  "html"
 ]);
 function isOfficeFile(name, mime) {
   const ext = (name || "").split(".").pop()?.toLowerCase() || "";
   if (OFFICE_FILE_EXTENSIONS.has(ext)) return true;
+  if (WEB_FETCHABLE_TEXT_EXTENSIONS.has(ext)) return false;
   const m = (mime || "").toLowerCase();
-  return m.includes("officedocument") || m.includes("hwp") || m === "application/msword" || m === "application/vnd.ms-excel" || m === "application/vnd.ms-powerpoint";
+  return m.includes("officedocument") || m.includes("opendocument") || m.includes("hwp") || m.includes("epub") || m === "application/msword" || m === "application/vnd.ms-excel" || m === "application/vnd.ms-powerpoint";
 }
 var _extractPlaceholderSeq = 0;
 function makeExtractPlaceholder(seed) {
@@ -157,6 +233,14 @@ File metadata:
 ` + (attachment.mime ? `- mime type: ${attachment.mime}
 ` : "") + (typeof attachment.size === "number" ? `- size (bytes): ${attachment.size}
 ` : "");
+  if (options?.inlineContent) {
+    return head + `
+The file's content was parsed by the client and is provided inline below. Read it directly \u2014 do NOT fetch any URL for this file. Use the storage path above (not this content) for the "src::" unique_id.
+
+----- BEGIN FILE CONTENT -----
+${options.inlineContent}
+----- END FILE CONTENT -----`;
+  }
   if (options?.inlineContentPlaceholder) {
     return head + `
 The file's text content was extracted on the server and is provided inline below. Read it directly \u2014 do NOT fetch any URL for this file. Use the storage path above (not this content) for the "src::" unique_id.
@@ -655,14 +739,14 @@ async function callOpenAIWithPublicMcp(prompt, service, owner, messages, system,
   });
 }
 async function notifyAgentSaveAttachment(info) {
-  const { platform, service, owner, attachment } = info;
-  const office = isOfficeFile(attachment.name, attachment.mime);
+  const { platform, service, owner, attachment, parsedContent } = info;
+  const office = !parsedContent && isOfficeFile(attachment.name, attachment.mime);
   const placeholder = office ? makeExtractPlaceholder(attachment.storagePath) : void 0;
   const extractContent = office && placeholder ? [{ path: attachment.storagePath, placeholder, name: attachment.name, mime: attachment.mime }] : void 0;
   const skapiExtract = extractContent && extractContent.length ? { _skapi_extract: extractContent } : {};
   const userMessage = buildIndexingUserMessage(
     attachment,
-    placeholder ? { inlineContentPlaceholder: placeholder } : void 0
+    parsedContent ? { inlineContent: parsedContent } : placeholder ? { inlineContentPlaceholder: placeholder } : void 0
   );
   const systemPrompt = buildIndexingSystemPrompt({
     service,
@@ -1248,6 +1332,7 @@ var ChatSession = class {
         this.enqueueTypewrite(aiIdx, answer, lid);
       }
     }
+    this._removeStrayPendingAssistants();
     this.promoteNextQueuedToRunning();
     this.updateHistoryCache();
     this.host.notify();
@@ -1290,6 +1375,7 @@ var ChatSession = class {
         if (thPos2 !== -1) this.state.messages.splice(thPos2, 1);
       }
       if (serverId) this.cancelledServerIds.delete(serverId);
+      this._removeStrayPendingAssistants();
       this.promoteNextQueuedToRunning();
       this.updateHistoryCache();
       this.host.notify();
@@ -1303,6 +1389,7 @@ var ChatSession = class {
       return;
     }
     this.insertAtTarget({ role: "assistant", content: getErrorMessage(err), isError: true }, targetIdx);
+    this._removeStrayPendingAssistants();
     this.promoteNextQueuedToRunning();
     this.updateHistoryCache();
     this.host.notify();
@@ -1464,16 +1551,48 @@ var ChatSession = class {
     if (pendingIdx === -1) return Promise.resolve();
     if (latest.isError || !latest.content) {
       this.state.messages[pendingIdx] = { role: "assistant", content: latest.content || "", isError: !!latest.isError };
+      this._removeStrayPendingAssistants();
       this.host.notify();
       this.promoteNextQueuedToRunning();
       return Promise.resolve();
     }
     var lid = this._newLocalId();
     this.state.messages[pendingIdx] = { role: "assistant", content: "", isPending: false, _localId: lid };
+    this._removeStrayPendingAssistants();
     this.host.notify();
     this.promoteNextQueuedToRunning();
     return this.enqueueTypewrite(pendingIdx, latest.content, lid);
   }
+  // Remove any leftover non-background pending ("Thinking…") assistant bubbles.
+  // There is normally at most ONE such bubble at a time (promoteNext* refuses to
+  // add a second), so any extra is a duplicate — it appears when a concurrent
+  // history refetch re-maps the still-"running" turn into a pending placeholder
+  // (with a real _serverItemId) while the local pending bubble (no _serverItemId)
+  // is rescued and re-appended (see loadHistory rescue below). Each resolve path
+  // only replaces the FIRST pending bubble, so without this a stray "Thinking…"
+  // survives next to the reply/error. MUST run AFTER the resolved bubble has been
+  // made non-pending and BEFORE promoteNext*() (so a freshly-promoted Thinking,
+  // which is added only once no pending assistant remains, is preserved).
+  _removeStrayPendingAssistants() {
+    for (var k = this.state.messages.length - 1; k >= 0; k--) {
+      var m = this.state.messages[k];
+      if (m.isPending && m.role === "assistant" && !m.isBackgroundTask) this.state.messages.splice(k, 1);
+    }
+  }
+  // Drop the pending flags on the resolved turn's USER bubble (preserving its
+  // content + background-task marker). Needed because a bg "Indexing:" turn's user
+  // bubble carries isPendingInProcess; leaving it set keeps the bubble visually
+  // stuck and keeps its bgTaskQueue entry alive forever.
+  _clearPendingUserBubble(itemId) {
+    var uIdx = this.state.messages.findIndex(function(m) {
+      return m.role === "user" && m._serverItemId === itemId && (m.isPendingInProcess || m.isPendingQueued || m.isSendingToServer);
+    });
+    if (uIdx === -1) return;
+    var u = this.state.messages[uIdx];
+    var cleaned = { role: "user", content: u.content, _serverItemId: itemId };
+    if (u.isBackgroundTask) cleaned.isBackgroundTask = true;
+    this.state.messages[uIdx] = cleaned;
+  }
   // If an immediate-send request for the current cache key is still in flight
   // (e.g. the view unmounted then remounted mid-request), show the sending
   // state, await it, then render the reply from the cache. Skipped when the
@@ -1512,6 +1631,7 @@ var ChatSession = class {
       return m.isPending && m._serverItemId === itemId;
     });
     if (idx !== -1) {
+      this._clearPendingUserBubble(itemId);
       if (isErr) {
         this.state.messages[idx] = { role: "assistant", content: answer, isError: true, _serverItemId: itemId };
         this.host.notify();
@@ -1661,12 +1781,16 @@ var ChatSession = class {
         self.state.messages.forEach(function(m) {
           if (m.isCancelled && m._serverItemId) locallyCancelled[m._serverItemId] = 1;
         });
+        var mappedHasPendingAssistant = mapped.some(function(m) {
+          return m.isPending && m.role === "assistant" && !m.isBackgroundTask;
+        });
         var rescued = [];
         for (var ri = 0; ri < self.state.messages.length; ri++) {
           var mm = self.state.messages[ri];
           if (mm.isBackgroundTask) continue;
           if (mm._serverItemId && serverIds[mm._serverItemId]) continue;
           if (!mm._serverItemId) {
+            if (mappedHasPendingAssistant) continue;
             if (mm.isSendingToServer || mm.isPendingQueued || mm.isPendingInProcess || mm.isPending) rescued.push(mm);
             else if (self.state.sending && mm.role === "user") {
               var next = self.state.messages[ri + 1];
@@ -1824,44 +1948,49 @@ var ChatSession = class {
             att.storagePath = member.storagePath;
           }
           var mime = member.file.type || self.host.getMimeType(member.file.name);
-          return notifyAgentSaveAttachment({
-            platform: id.platform,
-            model: id.model,
-            service: id.serviceId,
-            owner: id.owner,
-            userId: id.userId || id.serviceId,
-            serviceName: id.serviceName,
-            serviceDescription: id.serviceDescription,
-            attachment: {
-              name: member.file.name,
-              storagePath: member.storagePath,
-              mime: mime || void 0,
-              size: member.file.size,
-              url
-            }
-          }).then(function(ack) {
-            if (ack && typeof ack.id === "string") {
-              self.bgTaskQueue.push({
-                serviceId: id.serviceId,
-                platform: id.platform,
-                id: ack.id,
-                filename: member.file.name,
+          return Promise.resolve(
+            parseAttachmentContent(member.file, member.file.name, mime || void 0)
+          ).then(function(parsedContent) {
+            return notifyAgentSaveAttachment({
+              platform: id.platform,
+              model: id.model,
+              service: id.serviceId,
+              owner: id.owner,
+              userId: id.userId || id.serviceId,
+              serviceName: id.serviceName,
+              serviceDescription: id.serviceDescription,
+              attachment: {
+                name: member.file.name,
                 storagePath: member.storagePath,
-                isReindex: hadExists,
                 mime: mime || void 0,
                 size: member.file.size,
-                status: ack.status === "running" ? "running" : "pending",
-                poll: ack.poll
-              });
-              self.drainBgTaskQueue();
-            }
-          }, function(e) {
-            console.error("[chat-engine] indexing request failed", e);
-            anyIndexFailed = true;
-            if (!att.errorCode && !att.errorDetail) {
-              att.errorCode = e && (e.code || e.body && e.body.code) || "";
-              att.errorDetail = e && (e.message || e.body && e.body.message) || (typeof e === "string" ? e : "");
-            }
+                url
+              },
+              parsedContent: parsedContent || void 0
+            }).then(function(ack) {
+              if (ack && typeof ack.id === "string") {
+                self.bgTaskQueue.push({
+                  serviceId: id.serviceId,
+                  platform: id.platform,
+                  id: ack.id,
+                  filename: member.file.name,
+                  storagePath: member.storagePath,
+                  isReindex: hadExists,
+                  mime: mime || void 0,
+                  size: member.file.size,
+                  status: ack.status === "running" ? "running" : "pending",
+                  poll: ack.poll
+                });
+                self.drainBgTaskQueue();
+              }
+            }, function(e) {
+              console.error("[chat-engine] indexing request failed", e);
+              anyIndexFailed = true;
+              if (!att.errorCode && !att.errorDetail) {
+                att.errorCode = e && (e.code || e.body && e.body.code) || "";
+                att.errorDetail = e && (e.message || e.body && e.body.message) || (typeof e === "string" ? e : "");
+              }
+            });
           });
         });
       });
@@ -1953,6 +2082,7 @@ exports.EXPIRED_ATTACHMENT_URL_ORIGIN = EXPIRED_ATTACHMENT_URL_ORIGIN;
 exports.HISTORY_TOKEN_BUDGET = HISTORY_TOKEN_BUDGET;
 exports.LINK_LABEL_MAX_DISPLAY_CHARS = LINK_LABEL_MAX_DISPLAY_CHARS;
 exports.MAX_HISTORY_MESSAGES = MAX_HISTORY_MESSAGES;
+exports.MAX_PARSED_CONTENT_CHARS = MAX_PARSED_CONTENT_CHARS;
 exports.MCP_NAME = MCP_NAME;
 exports.MIN_INPUT_TOKEN_BUDGET = MIN_INPUT_TOKEN_BUDGET;
 exports.OUTPUT_TOKEN_RESERVE = OUTPUT_TOKEN_RESERVE;
@@ -1967,6 +2097,7 @@ exports.callClaudeWithMcp = callClaudeWithMcp;
 exports.callClaudeWithPublicMcp = callClaudeWithPublicMcp;
 exports.callOpenAIWithPublicMcp = callOpenAIWithPublicMcp;
 exports.chatEngineConfig = chatEngineConfig;
+exports.clearAttachmentParsers = clearAttachmentParsers;
 exports.composeUserMessage = composeUserMessage;
 exports.configureChatEngine = configureChatEngine;
 exports.createInlineLinkRegex = createInlineLinkRegex;
@@ -1978,6 +2109,8 @@ exports.extractLastUserTextFromRequest = extractLastUserTextFromRequest;
 exports.extractOpenAIText = extractOpenAIText;
 exports.extractRemotePathFromAttachmentHref = extractRemotePathFromAttachmentHref;
 exports.filterListByClearHorizon = filterListByClearHorizon;
+exports.findAttachmentParser = findAttachmentParser;
+exports.getAttachmentParsers = getAttachmentParsers;
 exports.getChatHistory = getChatHistory;
 exports.getContextWindow = getContextWindow;
 exports.getErrorMessage = getErrorMessage;
@@ -1994,6 +2127,8 @@ exports.mapHistoryListToMessages = mapHistoryListToMessages;
 exports.normalizeAttachmentPathCandidate = normalizeAttachmentPathCandidate;
 exports.normalizeTextContent = normalizeTextContent;
 exports.notifyAgentSaveAttachment = notifyAgentSaveAttachment;
+exports.parseAttachmentContent = parseAttachmentContent;
+exports.registerAttachmentParser = registerAttachmentParser;
 exports.safeDecodeURIComponent = safeDecodeURIComponent;
 exports.sanitizeAttachmentLinksForHistory = sanitizeAttachmentLinksForHistory;
 exports.stripFileBlocksFromHistory = stripFileBlocksFromHistory;