Handle attachment parsing error (exit gracefully)

Author: dinhlongviolin1

extensions/llamacpp-extension/src/util.ts

@@ -118,14 +118,14 @@ export type EmbedBatchResult = {
   usage?: EmbedUsage
 }
 
-export function estimateTokensFromText(text: string, charsPerToken = 3): number {
+export function estimateTokensFromText(text: string, charsPerToken = 2): number {
   return Math.max(1, Math.ceil(text.length / Math.max(charsPerToken, 1)))
 }
 
 export function buildEmbedBatches(
   inputs: string[],
   ubatchSize: number,
-  charsPerToken = 3
+  charsPerToken = 2
 ): EmbedBatch[] {
   const batches: EmbedBatch[] = []
   let current: string[] = []

extensions/yarn.lock

@@ -342,73 +342,73 @@ __metadata:
 
 "@janhq/core@file:../../core/package.tgz::locator=%40janhq%2Fassistant-extension%40workspace%3Aassistant-extension":
   version: 0.1.10
-  resolution: "@janhq/core@file:../../core/package.tgz#../../core/package.tgz::hash=16006d&locator=%40janhq%2Fassistant-extension%40workspace%3Aassistant-extension"
+  resolution: "@janhq/core@file:../../core/package.tgz#../../core/package.tgz::hash=a97af4&locator=%40janhq%2Fassistant-extension%40workspace%3Aassistant-extension"
   dependencies:
     rxjs: "npm:^7.8.1"
     ulidx: "npm:^2.3.0"
   peerDependencies:
     react: 19.0.0
-  checksum: 10c0/58a966e7f4aabfe7ee6c34958ca4b90c0de1d1210be9397e8f3fa77dbb4d744c4bc9814647652abd747cdbce630a0035be5e1eb748e38c354199110a5f3dc42f
+  checksum: 10c0/350e658f44d7ace8f829a548eed13d34291f50096537abeb36f6ffdc738781e26cf4feb839ce2bc3f43c18d82742b4118c1e960969b38abaeaca1f6c428386ed
   languageName: node
   linkType: hard
 
 "@janhq/core@file:../../core/package.tgz::locator=%40janhq%2Fconversational-extension%40workspace%3Aconversational-extension":
   version: 0.1.10
-  resolution: "@janhq/core@file:../../core/package.tgz#../../core/package.tgz::hash=16006d&locator=%40janhq%2Fconversational-extension%40workspace%3Aconversational-extension"
+  resolution: "@janhq/core@file:../../core/package.tgz#../../core/package.tgz::hash=a97af4&locator=%40janhq%2Fconversational-extension%40workspace%3Aconversational-extension"
   dependencies:
     rxjs: "npm:^7.8.1"
     ulidx: "npm:^2.3.0"
   peerDependencies:
     react: 19.0.0
-  checksum: 10c0/58a966e7f4aabfe7ee6c34958ca4b90c0de1d1210be9397e8f3fa77dbb4d744c4bc9814647652abd747cdbce630a0035be5e1eb748e38c354199110a5f3dc42f
+  checksum: 10c0/350e658f44d7ace8f829a548eed13d34291f50096537abeb36f6ffdc738781e26cf4feb839ce2bc3f43c18d82742b4118c1e960969b38abaeaca1f6c428386ed
   languageName: node
   linkType: hard
 
 "@janhq/core@file:../../core/package.tgz::locator=%40janhq%2Fdownload-extension%40workspace%3Adownload-extension":
   version: 0.1.10
-  resolution: "@janhq/core@file:../../core/package.tgz#../../core/package.tgz::hash=16006d&locator=%40janhq%2Fdownload-extension%40workspace%3Adownload-extension"
+  resolution: "@janhq/core@file:../../core/package.tgz#../../core/package.tgz::hash=a97af4&locator=%40janhq%2Fdownload-extension%40workspace%3Adownload-extension"
   dependencies:
     rxjs: "npm:^7.8.1"
     ulidx: "npm:^2.3.0"
   peerDependencies:
     react: 19.0.0
-  checksum: 10c0/58a966e7f4aabfe7ee6c34958ca4b90c0de1d1210be9397e8f3fa77dbb4d744c4bc9814647652abd747cdbce630a0035be5e1eb748e38c354199110a5f3dc42f
+  checksum: 10c0/350e658f44d7ace8f829a548eed13d34291f50096537abeb36f6ffdc738781e26cf4feb839ce2bc3f43c18d82742b4118c1e960969b38abaeaca1f6c428386ed
   languageName: node
   linkType: hard
 
 "@janhq/core@file:../../core/package.tgz::locator=%40janhq%2Fllamacpp-extension%40workspace%3Allamacpp-extension":
   version: 0.1.10
-  resolution: "@janhq/core@file:../../core/package.tgz#../../core/package.tgz::hash=16006d&locator=%40janhq%2Fllamacpp-extension%40workspace%3Allamacpp-extension"
+  resolution: "@janhq/core@file:../../core/package.tgz#../../core/package.tgz::hash=a97af4&locator=%40janhq%2Fllamacpp-extension%40workspace%3Allamacpp-extension"
   dependencies:
     rxjs: "npm:^7.8.1"
     ulidx: "npm:^2.3.0"
   peerDependencies:
     react: 19.0.0
-  checksum: 10c0/58a966e7f4aabfe7ee6c34958ca4b90c0de1d1210be9397e8f3fa77dbb4d744c4bc9814647652abd747cdbce630a0035be5e1eb748e38c354199110a5f3dc42f
+  checksum: 10c0/350e658f44d7ace8f829a548eed13d34291f50096537abeb36f6ffdc738781e26cf4feb839ce2bc3f43c18d82742b4118c1e960969b38abaeaca1f6c428386ed
   languageName: node
   linkType: hard
 
 "@janhq/core@file:../../core/package.tgz::locator=%40janhq%2Frag-extension%40workspace%3Arag-extension":
   version: 0.1.10
-  resolution: "@janhq/core@file:../../core/package.tgz#../../core/package.tgz::hash=16006d&locator=%40janhq%2Frag-extension%40workspace%3Arag-extension"
+  resolution: "@janhq/core@file:../../core/package.tgz#../../core/package.tgz::hash=a97af4&locator=%40janhq%2Frag-extension%40workspace%3Arag-extension"
   dependencies:
     rxjs: "npm:^7.8.1"
     ulidx: "npm:^2.3.0"
   peerDependencies:
     react: 19.0.0
-  checksum: 10c0/58a966e7f4aabfe7ee6c34958ca4b90c0de1d1210be9397e8f3fa77dbb4d744c4bc9814647652abd747cdbce630a0035be5e1eb748e38c354199110a5f3dc42f
+  checksum: 10c0/350e658f44d7ace8f829a548eed13d34291f50096537abeb36f6ffdc738781e26cf4feb839ce2bc3f43c18d82742b4118c1e960969b38abaeaca1f6c428386ed
   languageName: node
   linkType: hard
 
 "@janhq/core@file:../../core/package.tgz::locator=%40janhq%2Fvector-db-extension%40workspace%3Avector-db-extension":
   version: 0.1.10
-  resolution: "@janhq/core@file:../../core/package.tgz#../../core/package.tgz::hash=16006d&locator=%40janhq%2Fvector-db-extension%40workspace%3Avector-db-extension"
+  resolution: "@janhq/core@file:../../core/package.tgz#../../core/package.tgz::hash=a97af4&locator=%40janhq%2Fvector-db-extension%40workspace%3Avector-db-extension"
   dependencies:
     rxjs: "npm:^7.8.1"
     ulidx: "npm:^2.3.0"
   peerDependencies:
     react: 19.0.0
-  checksum: 10c0/58a966e7f4aabfe7ee6c34958ca4b90c0de1d1210be9397e8f3fa77dbb4d744c4bc9814647652abd747cdbce630a0035be5e1eb748e38c354199110a5f3dc42f
+  checksum: 10c0/350e658f44d7ace8f829a548eed13d34291f50096537abeb36f6ffdc738781e26cf4feb839ce2bc3f43c18d82742b4118c1e960969b38abaeaca1f6c428386ed
   languageName: node
   linkType: hard
 

src-tauri/plugins/tauri-plugin-rag/src/commands.rs

@@ -1,4 +1,5 @@
-use crate::{RagError, parser};
+use crate::{parser, RagError};
+use std::panic::{catch_unwind, AssertUnwindSafe};
 
 #[tauri::command]
 pub async fn parse_document<R: tauri::Runtime>(
@@ -7,6 +8,22 @@ pub async fn parse_document<R: tauri::Runtime>(
     file_type: String,
 ) -> Result<String, RagError> {
     log::info!("Parsing document: {} (type: {})", file_path, file_type);
-    let res = parser::parse_document(&file_path, &file_type);
-    res
+    let res = catch_unwind(AssertUnwindSafe(|| parser::parse_document(&file_path, &file_type)));
+    match res {
+        Ok(result) => result,
+        Err(payload) => {
+            let reason = if let Some(s) = payload.downcast_ref::<&str>() {
+                *s
+            } else if let Some(s) = payload.downcast_ref::<String>() {
+                s.as_str()
+            } else {
+                "unknown panic"
+            };
+            log::error!("Document parsing panicked: {}", reason);
+            Err(RagError::ParseError(format!(
+                "Document parsing failed unexpectedly: {}",
+                reason
+            )))
+        }
+    }
 }

src-tauri/plugins/tauri-plugin-rag/src/parser.rs

@@ -1,20 +1,37 @@
 use crate::RagError;
+use std::borrow::Cow;
 use std::fs;
-use std::io::{Read, Cursor};
-use zip::read::ZipArchive;
-use quick_xml::events::Event;
-use quick_xml::Reader;
+use std::io::{Cursor, Read};
+use std::panic::{catch_unwind, AssertUnwindSafe};
+use calamine::{open_workbook_auto, DataType, Reader as _};
+use chardetng::EncodingDetector;
 use csv as csv_crate;
-use calamine::{Reader as _, open_workbook_auto, DataType};
 use html2text;
-use chardetng::EncodingDetector;
 use infer;
-use std::borrow::Cow;
+use quick_xml::events::Event;
+use quick_xml::Reader;
+use zip::read::ZipArchive;
 
 pub fn parse_pdf(file_path: &str) -> Result<String, RagError> {
     let bytes = fs::read(file_path)?;
-    let text = pdf_extract::extract_text_from_mem(&bytes)
-        .map_err(|e| RagError::ParseError(format!("PDF parse error: {}", e)))?;
+    // pdf-extract can panic on some malformed PDFs; guard to avoid crashing the app
+    let text = match catch_unwind(AssertUnwindSafe(|| pdf_extract::extract_text_from_mem(&bytes))) {
+        Ok(Ok(t)) => t,
+        Ok(Err(e)) => return Err(RagError::ParseError(format!("PDF parse error: {}", e))),
+        Err(payload) => {
+            let reason = if let Some(s) = payload.downcast_ref::<&str>() {
+                *s
+            } else if let Some(s) = payload.downcast_ref::<String>() {
+                s.as_str()
+            } else {
+                "unknown parser panic"
+            };
+            return Err(RagError::ParseError(format!(
+                "PDF parsing failed unexpectedly: {}",
+                reason
+            )));
+        }
+    };
 
     // Validate that the PDF has extractable text (not image-based/scanned)
     // Count meaningful characters (excluding whitespace)

web-app/src/lib/attachmentProcessing.ts

@@ -29,6 +29,49 @@ export type AttachmentProcessingResult = {
   hasEmbeddedDocuments: boolean
 }
 
+const formatAttachmentError = (err: unknown): string => {
+  if (!err) return 'Unknown error'
+  if (err instanceof Error) return err.message || err.toString()
+  if (typeof err === 'string') return err
+  if (Array.isArray(err)) {
+    const parts = err.map((e) => formatAttachmentError(e)).filter(Boolean)
+    return parts.length ? Array.from(new Set(parts)).join('; ') : 'Unknown error'
+  }
+  if (typeof err === 'object') {
+    const obj = err as Record<string, unknown>
+    const candidates = [
+      obj.message,
+      obj.reason,
+      obj.detail,
+    ]
+    for (const val of candidates) {
+      if (typeof val === 'string' && val.trim().length > 0) {
+        return val
+      }
+    }
+    const nestedSources = [obj.error, obj.cause]
+    for (const nested of nestedSources) {
+      if (nested && typeof nested === 'object') {
+        const nestedMsg = formatAttachmentError(nested)
+        if (nestedMsg && nestedMsg !== 'Unknown error') {
+          return nestedMsg
+        }
+      } else if (typeof nested === 'string' && nested.trim().length > 0) {
+        return nested
+      }
+    }
+    if (typeof obj.code === 'string' && obj.code.trim().length > 0) {
+      return obj.code
+    }
+    try {
+      return JSON.stringify(obj)
+    } catch {
+      return String(err)
+    }
+  }
+  return String(err)
+}
+
 export const processAttachmentsForSend = async (
   options: AttachmentProcessingOptions
 ): Promise<AttachmentProcessingResult> => {
@@ -83,9 +126,9 @@ export const processAttachmentsForSend = async (
         if (updateAttachmentProcessing) {
           updateAttachmentProcessing(img.name, 'error')
         }
-        const desc = err instanceof Error ? err.message : String(err)
+        const desc = formatAttachmentError(err)
         toast.error('Failed to ingest image attachment', { description: desc })
-        throw err
+        throw err instanceof Error ? err : new Error(desc)
       }
     }
   }
@@ -204,9 +247,9 @@ export const processAttachmentsForSend = async (
         if (updateAttachmentProcessing) {
           updateAttachmentProcessing(doc.name, 'error')
         }
-        const desc = err instanceof Error ? err.message : String(err)
+        const desc = formatAttachmentError(err)
         toast.error('Failed to index attachments', { description: desc })
-        throw err
+        throw err instanceof Error ? err : new Error(desc)
       }
     }
   }