fix(tool): read_file falls back to chat-upload attachment by basename

This commit is contained in:
matevip 2026-04-26 20:24:00 +08:00
parent 30e7e67bb6
commit 76956cf990

View File

@ -38,6 +38,9 @@ public class ReadFileTool {
private static final int DEFAULT_MAX_LINES = 1000;
private static final int MAX_OUTPUT_BYTES = 30 * 1024; // 30KB
/** Chat attachment upload root, mirrored from ChatController.uploadRoot. */
private static final Path CHAT_UPLOAD_ROOT = Paths.get("data", "chat-uploads");
/**
* 二进制文档扩展名集合 - 这些文件不应使用 read_file 读取
*/
@ -64,12 +67,30 @@ public class ReadFileTool {
try {
path = vip.mate.tool.guard.WorkspacePathGuard.validatePath(filePath);
} catch (IllegalArgumentException e) {
return errorResult(filePath, e.getMessage());
// Sandbox rejected the literal path. The LLM may have hallucinated
// a Linux-style path (e.g. /app/Dockerfile) for a chat-upload that
// actually lives under data/chat-uploads/{conversationId}/. Retry
// by basename before surfacing the boundary error.
Path attachment = resolveChatUploadAttachment(filePath);
if (attachment == null) {
return errorResult(filePath, e.getMessage());
}
path = attachment;
}
// 文件存在性和类型校验
if (!Files.exists(path)) {
return errorResult(filePath, i18n.msg("tool.read_file.error.not_found", path));
// The user-uploaded chat attachment is rendered to the LLM as
// "[附件] foo.txt" without its stored path, so LLMs often pass
// just the basename or a guessed absolute path. Fall back to
// looking up the basename inside the current conversation's
// chat-upload directory before reporting not-found.
Path attachment = resolveChatUploadAttachment(filePath);
if (attachment == null) {
return errorResult(filePath, i18n.msg("tool.read_file.error.not_found", path));
}
log.info("[ReadFile] Resolved chat-upload attachment fallback: {} -> {}", filePath, attachment);
path = attachment;
}
if (Files.isDirectory(path)) {
return errorResult(filePath, i18n.msg("tool.read_file.error.is_directory", path));
@ -193,6 +214,60 @@ public class ReadFileTool {
}
}
/**
* Try to resolve {@code rawPath} against the current conversation's chat-upload
* directory. Uploaded files land at {@code data/chat-uploads/{conversationId}/{timestamp}_{safeFilename}}
* (see ChatController#upload), but the LLM only sees the original filename in
* the rendered message ("[附件] foo.txt"). When the LLM passes a guessed path
* that doesn't exist, this fallback rescues the call by matching basenames.
*
* @return absolute path of the matched attachment, or null if no match
*/
private Path resolveChatUploadAttachment(String rawPath) {
if (rawPath == null || rawPath.isBlank()) {
return null;
}
String conversationId = ToolExecutionContext.conversationId();
if (conversationId == null || conversationId.isBlank()) {
return null;
}
Path uploadDir = CHAT_UPLOAD_ROOT.resolve(conversationId).toAbsolutePath().normalize();
if (!Files.isDirectory(uploadDir)) {
return null;
}
String basename;
try {
Path requested = Paths.get(rawPath).getFileName();
basename = requested != null ? requested.toString() : null;
} catch (Exception e) {
return null;
}
if (basename == null || basename.isBlank()) {
return null;
}
// 1) direct match (LLM passed the stored filename)
Path direct = uploadDir.resolve(basename);
if (Files.isRegularFile(direct)) {
return direct;
}
// 2) timestamp-prefixed match: stored as "{millis}_{safeFilename}"
String safeBasename = basename.replaceAll("[^a-zA-Z0-9._-]", "_");
String suffix = "_" + safeBasename;
try (var stream = Files.list(uploadDir)) {
return stream
.filter(Files::isRegularFile)
.filter(p -> p.getFileName().toString().endsWith(suffix))
.findFirst()
.orElse(null);
} catch (IOException e) {
log.warn("[ReadFile] Failed to scan chat-upload dir {}: {}", uploadDir, e.getMessage());
return null;
}
}
private String errorResult(String filePath, String message) {
JSONObject result = new JSONObject();
result.set("filePath", filePath);