Skip to content

Self-evolving Context Database for AI Agents. Unify Agent Memory, Knowledge RAG and Skills.

Revert "feat(cli): download directories as zip archives (#4262)" (#4320)

Zayn Jarvis9b7bb49634e7
diff --git a/crates/ov_cli/src/client.rs b/crates/ov_cli/src/client.rs
index 3d62dba19..16d38b037 100644
--- a/crates/ov_cli/src/client.rs
+++ b/crates/ov_cli/src/client.rs
@@ -503,14 +503,8 @@ impl HttpClient {
self.post("/api/v1/system/backend/sync-retry", &body).await
}
- /// Download file bytes, or a directory as ZIP bytes
+ /// Download file as raw bytes
pub async fn get_bytes(&self, uri: &str) -> Result<Vec<u8>> {
- Ok(self.get_bytes_with_type(uri).await?.0)
- }
-
- /// Download bytes together with the response `Content-Type`, which is how
- /// the caller tells a raw file apart from a directory served as a ZIP.
- pub async fn get_bytes_with_type(&self, uri: &str) -> Result<(Vec<u8>, Option<String>)> {
let url = format!("{}/api/v1/content/download", self.base.base_url);
let params = vec![
("uri".to_string(), uri.to_string()),
@@ -538,16 +532,10 @@ impl HttpClient {
return Err(crate::base_client::api_error_from_body(&bytes, status));
}
- let content_type = response
- .headers()
- .get(reqwest::header::CONTENT_TYPE)
- .and_then(|value| value.to_str().ok())
- .map(|value| value.to_string());
-
response
.bytes()
.await
- .map(|b| (b.to_vec(), content_type))
+ .map(|b| b.to_vec())
.map_err(|e| Error::from_reqwest("Failed to read response bytes", e))
}
diff --git a/crates/ov_cli/src/commands/content.rs b/crates/ov_cli/src/commands/content.rs
index 44245049a..f78b0d534 100644
--- a/crates/ov_cli/src/commands/content.rs
+++ b/crates/ov_cli/src/commands/content.rs
@@ -3,8 +3,8 @@ use crate::error::Result;
use crate::output::OutputFormat;
use serde_json::{Value, json};
use std::collections::BTreeSet;
-use std::io::{ErrorKind, Write};
-use std::path::{Path, PathBuf};
+use std::io::Write;
+use std::path::Path;
pub async fn read(
client: &HttpClient,
@@ -88,82 +88,33 @@ pub async fn reindex(
Ok(())
}
-pub async fn get(client: &HttpClient, uri: &str, local_path: Option<&str>) -> Result<()> {
- preflight_explicit_download_target(local_path)?;
-
- // The body has to arrive before the target can be named: a directory URI
- // comes back as a ZIP, and that decides the `.zip` suffix.
- let (bytes, content_type) = client.get_bytes_with_type(uri).await?;
- let is_zip = content_type
- .as_deref()
- .is_some_and(|value| value.starts_with("application/zip"));
-
- let path = download_target(uri, local_path, is_zip);
- let display = path.display().to_string();
+pub async fn get(client: &HttpClient, uri: &str, local_path: &str) -> Result<()> {
+ // Check if target path already exists
+ let path = Path::new(local_path);
if path.exists() {
return Err(crate::error::Error::Client(format!(
- "File already exists: {display}"
+ "File already exists: {}",
+ local_path
)));
}
// Ensure parent directory exists
if let Some(parent) = path.parent() {
- if !parent.as_os_str().is_empty() {
- std::fs::create_dir_all(parent)?;
- }
+ std::fs::create_dir_all(parent)?;
}
+ // Download file
+ let bytes = client.get_bytes(uri).await?;
+
// Write to local file
- let mut file = create_download_target(&path, &display)?;
+ let mut file = create_download_target(path, local_path)?;
file.write_all(&bytes)?;
file.flush()?;
- println!("Downloaded {} bytes to {}", bytes.len(), display);
+ println!("Downloaded {} bytes to {}", bytes.len(), local_path);
Ok(())
}
-/// Resolve where the downloaded bytes go.
-///
-/// A target that is an existing directory — or omitted, meaning the current
-/// directory — receives a file named after the resource, so downloading a
-/// folder lands a real `<name>.zip` instead of an extension-less file carrying
-/// ZIP bytes. Any other path is used verbatim, which keeps
-/// `ov get <uri> ./explicit-name.zip` working as documented.
-fn download_target(uri: &str, local_path: Option<&str>, is_zip: bool) -> PathBuf {
- let base = Path::new(local_path.unwrap_or("."));
- if local_path.is_some() && !base.is_dir() {
- return base.to_path_buf();
- }
-
- let mut name = uri
- .trim_end_matches('/')
- .rsplit('/')
- .find(|segment| !segment.is_empty() && !segment.ends_with(':'))
- .unwrap_or("download")
- .to_string();
- if is_zip && !name.ends_with(".zip") {
- name.push_str(".zip");
- }
- base.join(name)
-}
-
-fn preflight_explicit_download_target(local_path: Option<&str>) -> Result<()> {
- let Some(local_path) = local_path else {
- return Ok(());
- };
- let path = Path::new(local_path);
- if path.is_dir() {
- return Ok(());
- }
- match path.symlink_metadata() {
- Ok(_) => Err(crate::error::Error::Client(format!(
- "File already exists: {local_path}"
- ))),
- Err(error) if error.kind() == ErrorKind::NotFound => Ok(()),
- Err(error) => Err(error.into()),
- }
-}
-
fn create_download_target(path: &Path, local_path: &str) -> Result<std::fs::File> {
std::fs::OpenOptions::new()
.write(true)
@@ -285,83 +236,6 @@ mod tests {
output
}
- #[test]
- fn download_target_names_archives_inside_a_directory() {
- use super::download_target;
- use std::path::PathBuf;
-
- let dir = tempfile::tempdir().expect("tempdir");
- let dir_path = dir.path().to_str().expect("utf-8 tempdir");
-
- // Omitted target -> current directory, named after the resource.
- assert_eq!(
- download_target("viking://resources/myfolder", None, true),
- PathBuf::from("./myfolder.zip")
- );
- assert_eq!(
- download_target("viking://resources/logo.png", None, false),
- PathBuf::from("./logo.png")
- );
- // Existing directory -> the archive lands inside it, not over it.
- assert_eq!(
- download_target("viking://resources/myfolder", Some(dir_path), true),
- dir.path().join("myfolder.zip")
- );
- // An explicit non-directory path stays verbatim.
- assert_eq!(
- download_target("viking://resources/myfolder", Some("./explicit.zip"), true),
- PathBuf::from("./explicit.zip")
- );
- // Already-suffixed names are not doubled up.
- assert_eq!(
- download_target("viking://resources/bundle.zip", None, true),
- PathBuf::from("./bundle.zip")
- );
- // Scope-only URIs still produce a usable name.
- assert_eq!(
- download_target("viking://", None, true),
- PathBuf::from("./download.zip")
- );
- }
-
- #[test]
- fn preflight_rejects_explicit_existing_file_target() {
- let dir = tempfile::tempdir().expect("tempdir should be created");
- let path = dir.path().join("result.bin");
- std::fs::write(&path, b"existing").expect("fixture should be written");
-
- let error = super::preflight_explicit_download_target(Some(&path.to_string_lossy()))
- .expect_err("explicit existing file targets must fail before download");
-
- assert!(matches!(error, Error::Client(_)));
- }
-
- #[test]
- fn preflight_allows_directory_or_missing_explicit_targets() {
- let dir = tempfile::tempdir().expect("tempdir should be created");
- let missing_path = dir.path().join("new-result.bin");
-
- super::preflight_explicit_download_target(None).expect("omitted target is named later");
- super::preflight_explicit_download_target(Some(&missing_path.to_string_lossy()))
- .expect("missing explicit file target is usable");
- super::preflight_explicit_download_target(Some(&dir.path().to_string_lossy()))
- .expect("directory target is named after the response type");
- }
-
- #[cfg(unix)]
- #[test]
- fn preflight_rejects_broken_symlink_target() {
- let dir = tempfile::tempdir().expect("tempdir should be created");
- let link_path = dir.path().join("broken-link");
- std::os::unix::fs::symlink(dir.path().join("missing-target"), &link_path)
- .expect("symlink fixture should be created");
-
- let error = super::preflight_explicit_download_target(Some(&link_path.to_string_lossy()))
- .expect_err("existing symlink targets must fail before download");
-
- assert!(matches!(error, Error::Client(_)));
- }
-
#[test]
fn table_output_renders_profiled_scalar_content() {
let result = json!({
diff --git a/crates/ov_cli/src/error.rs b/crates/ov_cli/src/error.rs
index 65950a42a..d449d60a3 100644
--- a/crates/ov_cli/src/error.rs
+++ b/crates/ov_cli/src/error.rs
@@ -123,7 +123,6 @@ fn code_from_http_status(status: Option<u16>) -> &'static str {
Some(404) => "NOT_FOUND",
Some(409) => "CONFLICT",
Some(412) => "FAILED_PRECONDITION",
- Some(413) => "PAYLOAD_TOO_LARGE",
Some(429) => "RESOURCE_EXHAUSTED",
Some(502 | 503) => "UNAVAILABLE",
Some(504) => "DEADLINE_EXCEEDED",
diff --git a/crates/ov_cli/src/handlers.rs b/crates/ov_cli/src/handlers.rs
index c61703cf3..d5ac097f3 100644
--- a/crates/ov_cli/src/handlers.rs
+++ b/crates/ov_cli/src/handlers.rs
@@ -1377,9 +1377,9 @@ pub async fn handle_reindex(
.await
}
-pub async fn handle_get(uri: String, local_path: Option<String>, ctx: CliContext) -> Result<()> {
+pub async fn handle_get(uri: String, local_path: String, ctx: CliContext) -> Result<()> {
let client = ctx.get_client();
- commands::content::get(&client, &uri, local_path.as_deref()).await
+ commands::content::get(&client, &uri, &local_path).await
}
pub async fn handle_find(
diff --git a/crates/ov_cli/src/main.rs b/crates/ov_cli/src/main.rs
index a5895f2c7..7aa6ff826 100644
--- a/crates/ov_cli/src/main.rs
+++ b/crates/ov_cli/src/main.rs
@@ -678,15 +678,14 @@ enum Commands {
#[arg(long, default_value = "false")]
recursive: bool,
},
- /// [Data] Download a file, or archive a directory as ZIP, to a local path
+ /// [Data] Download file to local path (supports binaries/images)
Get {
/// Viking URI
#[arg(value_name = "uri")]
uri: String,
- /// Destination directory, or an exact file path. A directory (the
- /// default: the current one) receives <name>.zip for a directory URI.
+ /// Local path (must not exist yet)
#[arg(value_name = "local-path")]
- local_path: Option<String>,
+ local_path: String,
},
/// [Data] Run semantic retrieval
Find {
diff --git a/docs/en/api/01-overview.md b/docs/en/api/01-overview.md
index 523af51e4..fb9e31373 100644
--- a/docs/en/api/01-overview.md
+++ b/docs/en/api/01-overview.md
@@ -321,7 +321,6 @@ JSON output - error:
| `UNAUTHENTICATED` | 401 | Missing or invalid API key |
| `PERMISSION_DENIED` | 403 | Insufficient permissions |
| `RESOURCE_EXHAUSTED` | 429 | Rate limit exceeded |
-| `PAYLOAD_TOO_LARGE` | 413 | Response body exceeds a hard size limit (not retryable) |
| `FAILED_PRECONDITION` | 412 | Precondition failed |
| `CONFLICT` | 409 | Operation conflicts with an in-progress task or existing state |
| `DEADLINE_EXCEEDED` | 504 | Operation timed out |
@@ -376,7 +375,7 @@ This catalog follows the routes actually mounted by the server. Each group headi
| GET | `/api/v1/content/read` | Read full content (L2) |
| GET | `/api/v1/content/abstract` | Read an abstract (L0) |
| GET | `/api/v1/content/overview` | Read an overview (L1) |
-| GET | `/api/v1/content/download` | Download original file bytes, or a directory as a ZIP archive |
+| GET | `/api/v1/content/download` | Download original file bytes |
| POST | `/api/v1/content/write` | Write content and refresh semantic indexes |
| POST | `/api/v1/content/batch-write` | Apply preconditioned multi-file writes |
| POST | `/api/v1/content/set_tags` | Set retrieval tags |
diff --git a/docs/en/api/12-content.md b/docs/en/api/12-content.md
index 25f540c1a..89f21e121 100644
--- a/docs/en/api/12-content.md
+++ b/docs/en/api/12-content.md
@@ -433,11 +433,11 @@ The TypeScript and Go SDKs and the CLI do not currently expose batch write direc
### download()
-Download a file as raw bytes, or a directory as a ZIP archive. File responses use `application/octet-stream`; directory responses use `application/zip` and preserve the downloaded directory as the archive root. Both return the filename through `Content-Disposition`. Directory downloads are limited to 10 MiB of source content, 10 MiB final ZIP size, and 10,000 tree entries; oversized archives return `PAYLOAD_TOO_LARGE` (HTTP `413`). That failure is a property of the directory, not a rate limit, so it is permanent for that URI — do not retry it.
+Download a file as raw bytes. This is intended for images, PDFs, and other non-text content. The response uses `application/octet-stream` and returns the filename through `Content-Disposition`.
| Parameter | Type | Required | Description |
|-----------|------|----------|-------------|
-| `uri` | string | Yes | File or directory URI to download |
+| `uri` | string | Yes | File URI to download |
**HTTP API**
@@ -450,10 +450,6 @@ curl --get http://localhost:1933/api/v1/content/download \
-H "X-API-Key: your-key" \
--data-urlencode "uri=viking://resources/images/logo.png" \
--output logo.png
-
-# A directory is returned as a ZIP archive.
-# Writes ./project.zip, named after the directory.
-ov get viking://resources/project
```
**Response**
@@ -468,7 +464,7 @@ Content-Disposition: attachment; filename*=UTF-8''logo.png
<binary body>
```
-The CLI exposes the same endpoint through `ov get <uri> [local-path]`. When the local path is an existing directory — or omitted, which means the current directory — the download is written inside it under the resource's own name, with a `.zip` suffix for a directory URI. Any other local path is used verbatim, and must not already exist. The CLI never extracts the archive.
+The public SDKs and CLI do not currently expose a dedicated raw-byte download method, so this section shows only the HTTP tab.
---
diff --git a/docs/en/context-compilation/03-knowledge-graph.md b/docs/en/context-compilation/03-knowledge-graph.md
index 61dd8d629..ff3d56651 100644
--- a/docs/en/context-compilation/03-knowledge-graph.md
+++ b/docs/en/context-compilation/03-knowledge-graph.md
@@ -63,14 +63,24 @@ ov read viking://resources/journal-kg/entities/sun-wukong.md
Unlike the LLM Wiki script, `knowledge_graph.py` reads from a **local directory** (it needs both `entities/` and `relations.jsonl` on disk). So download the output first, then generate the HTML.
-Download the whole artifact tree as a ZIP archive, then extract it:
+Download the whole artifact tree. `ov get` downloads one file at a time; combine it with `ov ls -r -s` to pull every path:
```bash
-ov get viking://resources/journal-kg
-unzip -o ./journal-kg.zip
+SRC="viking://resources/journal-kg"
+DST="./journal-kg"
+mkdir -p "$DST"
+ov ls -r -s "$SRC" | while read -r uri; do
+ # only download files (entities/*.md and relations.jsonl), skip directories
+ case "$uri" in
+ */entities|"$SRC") continue ;;
+ esac
+ rel="${uri#$SRC/}"
+ mkdir -p "$DST/$(dirname "$rel")"
+ ov get "$uri" "$DST/$rel"
+done
```
-> `ov get` writes `./journal-kg.zip`, named after the directory. It refuses to overwrite, so remove the old archive (`rm -f ./journal-kg.zip`) before re-downloading. `unzip -o` overwrites the previously extracted `./journal-kg/` instead of prompting.
+> `ov get` requires the local target path to not exist yet, so clear the old directory (`rm -rf ./journal-kg`) before re-downloading.
Confirm the local layout is correct:
diff --git a/docs/zh/api/01-overview.md b/docs/zh/api/01-overview.md
index 85fb49c23..278039a87 100644
--- a/docs/zh/api/01-overview.md
+++ b/docs/zh/api/01-overview.md
@@ -316,7 +316,6 @@ JSON 输出 - 错误:
| `UNAUTHENTICATED` | 401 | 缺少或无效的 API Key |
| `PERMISSION_DENIED` | 403 | 权限不足 |
| `RESOURCE_EXHAUSTED` | 429 | 超出速率限制 |
-| `PAYLOAD_TOO_LARGE` | 413 | 响应体超出硬性大小限制(不可重试) |
| `FAILED_PRECONDITION` | 412 | 前置条件不满足 |
| `CONFLICT` | 409 | 操作与正在进行的任务或已有状态冲突 |
| `DEADLINE_EXCEEDED` | 504 | 操作超时 |
@@ -371,7 +370,7 @@ JSON 输出 - 错误:
| GET | `/api/v1/content/read` | 读取完整内容(L2) |
| GET | `/api/v1/content/abstract` | 读取摘要(L0) |
| GET | `/api/v1/content/overview` | 读取概览(L1) |
-| GET | `/api/v1/content/download` | 下载原始文件字节,或将目录打包为 ZIP 归档 |
+| GET | `/api/v1/content/download` | 下载原始文件字节 |
| POST | `/api/v1/content/write` | 写入内容并刷新语义索引 |
| POST | `/api/v1/content/batch-write` | 执行带前置条件的多文件写入 |
| POST | `/api/v1/content/set_tags` | 设置检索标签 |
diff --git a/docs/zh/api/12-content.md b/docs/zh/api/12-content.md
index 7bac00334..21c052ff6 100644
--- a/docs/zh/api/12-content.md
+++ b/docs/zh/api/12-content.md
@@ -433,11 +433,11 @@ TypeScript、Go SDK 和 CLI 当前不直接暴露 batch write。
### download()
-以原始字节流下载文件,或将目录打包成 ZIP 下载。文件响应使用 `application/octet-stream`;目录响应使用 `application/zip`,并以所下载目录作为压缩包根目录。两者都会通过 `Content-Disposition` 返回文件名。目录下载的源文件总大小、最终 ZIP 大小均限制为 10 MiB,且最多包含 10,000 个目录树条目;超限返回 `PAYLOAD_TOO_LARGE`(HTTP `413`)。该失败取决于目录本身而非速率限制,对同一 URI 是永久性的,不要重试。
+以原始字节流下载文件,适用于图片、PDF 和其他非文本内容。响应使用 `application/octet-stream`,并通过 `Content-Disposition` 返回文件名。
| 参数 | 类型 | 必填 | 说明 |
|------|------|------|------|
-| `uri` | string | 是 | 要下载的文件或目录 URI |
+| `uri` | string | 是 | 要下载的文件 URI |
**HTTP API**
@@ -450,10 +450,6 @@ curl --get http://localhost:1933/api/v1/content/download \
-H "X-API-Key: your-key" \
--data-urlencode "uri=viking://resources/images/logo.png" \
--output logo.png
-
-# 目录会以 ZIP 压缩包返回
-# 写出 ./project.zip,以目录名命名
-ov get viking://resources/project
```
**响应**
@@ -468,7 +464,7 @@ Content-Disposition: attachment; filename*=UTF-8''logo.png
<binary body>
```
-CLI 通过 `ov get <uri> [local-path]` 调用同一接口。当本地路径是已存在的目录、或省略(即当前目录)时,文件会以资源自身的名字写入该目录下,目录 URI 会带上 `.zip` 后缀。其他本地路径按原样使用,且必须尚不存在。CLI 不会自动解压。
+公共 SDK 和 CLI 当前没有独立的原始字节下载方法,因此本节只展示 HTTP Tab。
---
diff --git a/docs/zh/context-compilation/03-knowledge-graph.md b/docs/zh/context-compilation/03-knowledge-graph.md
index d3c67d666..d2531ca12 100644
--- a/docs/zh/context-compilation/03-knowledge-graph.md
+++ b/docs/zh/context-compilation/03-knowledge-graph.md
@@ -63,14 +63,24 @@ ov read viking://resources/journal-kg/entities/孙悟空.md
与 LLM Wiki 的脚本不同,`knowledge_graph.py` 读取的是**本地目录**(需要 `entities/` 和 `relations.jsonl` 都在本地)。所以先把产物拉到本地,再生成 HTML。
-先把整棵工件树打包成 ZIP 下载,再解压:
+先把整棵工件树下载下来。`ov get` 一次下载一个文件,配合 `ov ls -r -s` 列出全部路径即可批量拉取:
```bash
-ov get viking://resources/journal-kg
-unzip -o ./journal-kg.zip
+SRC="viking://resources/journal-kg"
+DST="./journal-kg"
+mkdir -p "$DST"
+ov ls -r -s "$SRC" | while read -r uri; do
+ # 只下载文件(entities/*.md 和 relations.jsonl),跳过目录
+ case "$uri" in
+ */entities|"$SRC") continue ;;
+ esac
+ rel="${uri#$SRC/}"
+ mkdir -p "$DST/$(dirname "$rel")"
+ ov get "$uri" "$DST/$rel"
+done
```
-> `ov get` 会写出 `./journal-kg.zip`,以目录名命名。它不会覆盖已有文件,所以重新下载前先删除旧压缩包(`rm -f ./journal-kg.zip`)。`unzip -o` 会直接覆盖上次解压出的 `./journal-kg/`,不会停下来等确认。
+> `ov get` 要求本地目标路径尚不存在,所以重新下载前先清掉旧目录(`rm -rf ./journal-kg`)。
确认本地目录结构正确:
diff --git a/openviking/server/app.py b/openviking/server/app.py
index d4112ee68..33355abcb 100644
--- a/openviking/server/app.py
+++ b/openviking/server/app.py
@@ -185,7 +185,6 @@ _FRAMEWORK_HTTP_STATUS_TO_ERROR_CODE = {
403: "PERMISSION_DENIED",
404: "NOT_FOUND",
409: "CONFLICT",
- 413: "PAYLOAD_TOO_LARGE",
422: "INVALID_ARGUMENT",
429: "RESOURCE_EXHAUSTED",
502: "UNAVAILABLE",
diff --git a/openviking/server/models.py b/openviking/server/models.py
index 5e182e1c7..67c08cf53 100644
--- a/openviking/server/models.py
+++ b/openviking/server/models.py
@@ -36,7 +36,6 @@ ERROR_CODE_TO_HTTP_STATUS = {
"PERMISSION_DENIED": 403,
"UNAUTHENTICATED": 401,
"RESOURCE_EXHAUSTED": 429,
- "PAYLOAD_TOO_LARGE": 413,
"FAILED_PRECONDITION": 412,
"ABORTED": 409,
"DEADLINE_EXCEEDED": 504,
diff --git a/openviking/server/routers/content.py b/openviking/server/routers/content.py
index 9a8b48044..4912deeb1 100644
--- a/openviking/server/routers/content.py
+++ b/openviking/server/routers/content.py
@@ -2,10 +2,6 @@
# SPDX-License-Identifier: AGPL-3.0
"""Content endpoints for OpenViking HTTP Server."""
-import asyncio
-import io
-import zipfile
-from pathlib import PurePosixPath
from typing import Literal
from urllib.parse import quote
@@ -32,117 +28,11 @@ from openviking.server.identity import RequestContext, Role
from openviking.server.models import Response
from openviking.server.telemetry import run_operation
from openviking.telemetry import TelemetryRequest
-from openviking_cli.exceptions import (
- InvalidArgumentError,
- NotFoundError,
- PayloadTooLargeError,
- PermissionDeniedError,
-)
+from openviking_cli.exceptions import InvalidArgumentError, NotFoundError, PermissionDeniedError
from openviking_cli.utils import get_logger
logger = get_logger(__name__)
-_DIRECTORY_ARCHIVE_MAX_BYTES = 10 * 1024 * 1024
-_DIRECTORY_ARCHIVE_MAX_ENTRIES = 10_000
-
-
-def _archive_size_limit_error(uri: str) -> PayloadTooLargeError:
- # 413, not 429: the limit is a property of the directory, so retrying the
- # same URI can only fail again. A 429 would send clients into backoff-retry.
- return PayloadTooLargeError(
- f"Directory archive exceeds the {_DIRECTORY_ARCHIVE_MAX_BYTES}-byte download limit: {uri}"
- )
-
-
-def _archive_entry_limit_error(uri: str) -> PayloadTooLargeError:
- return PayloadTooLargeError(
- f"Directory archive exceeds the {_DIRECTORY_ARCHIVE_MAX_ENTRIES}-entry download limit: {uri}"
- )
-
-
-def _safe_archive_path(root_name: str, rel_path: str = "") -> str:
- """Build a portable ZIP member path from server-produced tree entries."""
- relative = PurePosixPath(rel_path)
- if relative.is_absolute():
- raise InvalidArgumentError(f"Unsafe path in directory download: {rel_path}")
- parts = [root_name, *relative.parts]
- if any(part in {"", ".", ".."} or "\\" in part for part in parts):
- raise InvalidArgumentError(f"Unsafe path in directory download: {rel_path or root_name}")
- return "/".join(parts)
-
-
-async def _build_directory_archive(
- service, uri: str, stat: dict, ctx: RequestContext
-) -> tuple[bytes, str]:
- """Pack one visible Viking directory tree into an in-memory ZIP archive.
-
- The archive is capped at `_DIRECTORY_ARCHIVE_MAX_BYTES`, so it is built in
- memory rather than in a temp file: a temp file served through FileResponse
- outlives the request whenever the response never reaches its cleanup
- callback (a malformed or unsatisfiable Range header returns early, and
- cancellation skips it entirely).
- """
- root_name = str(stat.get("name") or uri.rstrip("/").rsplit("/", 1)[-1] or "download")
- root_path = _safe_archive_path(root_name)
-
- entries = await service.fs.tree(
- uri,
- ctx=ctx,
- output="original",
- show_all_hidden=True,
- node_limit=_DIRECTORY_ARCHIVE_MAX_ENTRIES + 1,
- level_limit=None,
- )
- if len(entries) > _DIRECTORY_ARCHIVE_MAX_ENTRIES:
- raise _archive_entry_limit_error(uri)
-
- # Reject oversized trees before loading file contents into memory. The
- # ZIP itself is checked below as well because headers can push a highly
- # fragmented archive over the limit even when its files do not.
- declared_total = 0
- for entry in entries:
- if entry.get("isDir", False):
- continue
- size = entry.get("size")
- if isinstance(size, bool) or not isinstance(size, int) or size < 0:
- file_stat = await service.fs.stat(str(entry["uri"]), ctx=ctx)
- size = file_stat.get("size")
- if isinstance(size, bool) or not isinstance(size, int) or size < 0:
- raise InvalidArgumentError(
- f"Cannot determine file size for directory download: {entry['uri']}"
- )
- declared_total += size
- if declared_total > _DIRECTORY_ARCHIVE_MAX_BYTES:
- raise _archive_size_limit_error(uri)
-
- buffer = io.BytesIO()
- actual_total = 0
- with zipfile.ZipFile(
- buffer,
- mode="w",
- compression=zipfile.ZIP_DEFLATED,
- allowZip64=True,
- ) as archive:
- archive.writestr(f"{root_path}/", b"")
- for entry in entries:
- member_path = _safe_archive_path(root_name, str(entry["rel_path"]))
- if entry.get("isDir", False):
- archive.writestr(f"{member_path.rstrip('/')}/", b"")
- else:
- content = await service.fs.read_file_bytes(str(entry["uri"]), ctx=ctx)
- actual_total += len(content)
- if actual_total > _DIRECTORY_ARCHIVE_MAX_BYTES:
- raise _archive_size_limit_error(uri)
- await asyncio.to_thread(archive.writestr, member_path, content)
- # Bound the buffer while it grows. Per-entry ZIP headers are not
- # counted by `actual_total`, so a tree of many empty directories or
- # empty files would otherwise only be rejected once fully built.
- if buffer.tell() > _DIRECTORY_ARCHIVE_MAX_BYTES:
- raise _archive_size_limit_error(uri)
- if buffer.tell() > _DIRECTORY_ARCHIVE_MAX_BYTES:
- raise _archive_size_limit_error(uri)
- return buffer.getvalue(), f"{root_name}.zip"
-
class WriteContentRequest(BaseModel):
"""Request to write, append, or create text content to a file."""
@@ -301,27 +191,10 @@ async def download(
uri: str = Query(..., description="Viking URI"),
_ctx: RequestContext = Depends(get_request_context),
):
- """Download a file as raw bytes or a directory as a ZIP archive."""
+ """Download file as raw bytes (for images, binaries, etc.)."""
service = get_service()
uri = validate_request_viking_uri(resolve_path_variables(uri), _ctx)
try:
- stat = await service.fs.stat(uri, ctx=_ctx)
- if stat.get("isDir", False):
- archive, filename = await _build_directory_archive(service, uri, stat, _ctx)
- # Buffering the whole archive is only safe because of the 10 MiB
- # cap in _build_directory_archive; directories over it fail with
- # 413 rather than downloading. To raise that ceiling, stream from a
- # temp file with starlette.responses.FileResponse instead — but
- # note that a BackgroundTask cleanup is NOT sufficient on its own:
- # starlette skips `background` on the Range-header error paths and
- # on cancellation, so the temp archive leaks. Whatever replaces
- # this must own its own cleanup (e.g. unlink the file as soon as it
- # is opened, and stream from the surviving descriptor).
- return FastAPIResponse(
- content=archive,
- media_type="application/zip",
- headers={"Content-Disposition": f"attachment; filename*=UTF-8''{quote(filename)}"},
- )
content = await service.fs.read_file_bytes(uri, ctx=_ctx)
except AGFSNotFoundError:
raise NotFoundError(uri, "file")
@@ -331,7 +204,14 @@ async def download(
raise mapped from e
raise
- filename = stat.get("name", "download")
+ # Try to get filename from stat
+ filename = "download"
+ try:
+ stat = await service.fs.stat(uri, ctx=_ctx)
+ if stat and "name" in stat:
+ filename = stat["name"]
+ except Exception:
+ pass
filename = quote(filename)
return FastAPIResponse(
content=content,
diff --git a/openviking/utils/exceptions.py b/openviking/utils/exceptions.py
index 9376cbd90..d3b8444b2 100644
--- a/openviking/utils/exceptions.py
+++ b/openviking/utils/exceptions.py
@@ -10,7 +10,6 @@ HTTP_STATUS_TO_ERROR_CODE = {
404: "NOT_FOUND",
408: "DEADLINE_EXCEEDED",
409: "CONFLICT",
- 413: "PAYLOAD_TOO_LARGE",
422: "INVALID_ARGUMENT",
429: "RESOURCE_EXHAUSTED",
500: "UNAVAILABLE",
diff --git a/openviking_cli/client/_http_compat.py b/openviking_cli/client/_http_compat.py
index b8e960bfc..d9fe260b5 100644
--- a/openviking_cli/client/_http_compat.py
+++ b/openviking_cli/client/_http_compat.py
@@ -25,7 +25,6 @@ from openviking_cli.exceptions import (
NotFoundError,
NotInitializedError,
OpenVikingError,
- PayloadTooLargeError,
PermissionDeniedError,
ProcessingError,
ResourceExhaustedError,
@@ -50,7 +49,6 @@ ERROR_CODE_TO_EXCEPTION = {
"UNAUTHENTICATED": UnauthenticatedError,
"PERMISSION_DENIED": PermissionDeniedError,
"RESOURCE_EXHAUSTED": ResourceExhaustedError,
- "PAYLOAD_TOO_LARGE": PayloadTooLargeError,
"UNAVAILABLE": UnavailableError,
"INTERNAL": InternalError,
"DEADLINE_EXCEEDED": DeadlineExceededError,
@@ -92,7 +90,6 @@ def _raise_legacy_exception(error: Dict[str, Any]) -> None:
if exc_class in (
InvalidArgumentError,
FailedPreconditionError,
- PayloadTooLargeError,
ResourceExhaustedError,
AbortedError,
UnimplementedError,
diff --git a/openviking_cli/exceptions.py b/openviking_cli/exceptions.py
index 87db9fb9f..9305a2c6d 100644
--- a/openviking_cli/exceptions.py
+++ b/openviking_cli/exceptions.py
@@ -136,17 +136,6 @@ class ResourceExhaustedError(OpenVikingError):
super().__init__(message, code="RESOURCE_EXHAUSTED", details=details)
-class PayloadTooLargeError(OpenVikingError):
- """Response or request body exceeds a hard size limit.
-
- Distinct from ResourceExhaustedError: this failure is permanent for the
- given input, so clients must not retry it the way they back off on 429.
- """
-
- def __init__(self, message: str = "Payload too large", details: Optional[dict] = None):
- super().__init__(message, code="PAYLOAD_TOO_LARGE", details=details)
-
-
class InternalError(OpenVikingError):
"""Internal server error."""
diff --git a/tests/server/test_api_fs_content_endpoint_suite.py b/tests/server/test_api_fs_content_endpoint_suite.py
index 42edabc3e..f7c6137bd 100644
--- a/tests/server/test_api_fs_content_endpoint_suite.py
+++ b/tests/server/test_api_fs_content_endpoint_suite.py
@@ -1,16 +1,11 @@
# Copyright (c) 2026 Beijing Volcano Engine Technology Co., Ltd.
# SPDX-License-Identifier: AGPL-3.0
-import io
-import zipfile
from types import SimpleNamespace
import httpx
-import pytest
from openviking.pyagfs.exceptions import AGFSHTTPError
-from openviking.server.models import ERROR_CODE_TO_HTTP_STATUS
-from openviking.server.routers import content as content_router
def _assert_error(
@@ -169,197 +164,6 @@ async def test_download_returns_attachment_response(client_with_resource):
assert "attachment;" in response.headers["content-disposition"]
-async def test_download_directory_returns_zip_archive(client_with_resource):
- client, uri = client_with_resource
- response = await client.get("/api/v1/content/download", params={"uri": uri})
-
- assert response.status_code == 200
- assert response.headers["content-type"] == "application/zip"
- assert ".zip" in response.headers["content-disposition"]
-
- with zipfile.ZipFile(io.BytesIO(response.content)) as archive:
- names = archive.namelist()
- assert names
- root = names[0].rstrip("/")
- assert names[0].endswith("/")
- assert any(name.startswith(f"{root}/") and not name.endswith("/") for name in names)
- assert all(not name.startswith("/") and ".." not in name.split("/") for name in names)
-
-
-async def test_download_directory_over_limit_returns_payload_too_large(
- client_with_resource, monkeypatch
-):
- client, uri = client_with_resource
- monkeypatch.setattr(content_router, "_DIRECTORY_ARCHIVE_MAX_BYTES", 1)
-
- response = await client.get("/api/v1/content/download", params={"uri": uri})
-
- _assert_error(response, status_code=413, error_code="PAYLOAD_TOO_LARGE")
-
-
-async def test_build_directory_archive_preserves_tree_and_empty_directories():
- class FakeFS:
- async def tree(self, *args, **kwargs):
- return [
- {"uri": "viking://resources/project/empty", "rel_path": "empty", "isDir": True},
- {
- "uri": "viking://resources/project/docs/readme.md",
- "rel_path": "docs/readme.md",
- "isDir": False,
- "size": 5,
- },
- ]
-
- async def read_file_bytes(self, uri, ctx):
- assert uri.endswith("docs/readme.md")
- return b"hello"
-
- service = SimpleNamespace(fs=FakeFS())
- payload, filename = await content_router._build_directory_archive(
- service,
- "viking://resources/project",
- {"name": "project", "isDir": True},
- SimpleNamespace(),
- )
- assert filename == "project.zip"
- with zipfile.ZipFile(io.BytesIO(payload)) as archive:
- assert archive.namelist() == [
- "project/",
- "project/empty/",
- "project/docs/readme.md",
- ]
- assert archive.read("project/docs/readme.md") == b"hello"
-
-
-async def test_build_directory_archive_rejects_declared_size_over_limit():
- class FakeFS:
- async def tree(self, *args, **kwargs):
- return [
- {
- "uri": "viking://resources/project/large.bin",
- "rel_path": "large.bin",
- "isDir": False,
- "size": content_router._DIRECTORY_ARCHIVE_MAX_BYTES + 1,
- }
- ]
-
- async def read_file_bytes(self, *args, **kwargs):
- raise AssertionError("oversized file must not be read")
-
- service = SimpleNamespace(fs=FakeFS())
- with pytest.raises(Exception, match="exceeds the .* download limit"):
- await content_router._build_directory_archive(
- service,
- "viking://resources/project",
- {"name": "project", "isDir": True},
- SimpleNamespace(),
- )
-
-
-async def test_build_directory_archive_rejects_too_many_entries_before_reading():
- observed = {}
-
- class FakeFS:
- async def tree(self, *args, **kwargs):
- observed["node_limit"] = kwargs.get("node_limit")
- return [
- {
- "uri": f"viking://resources/project/d{index}",
- "rel_path": f"d{index}",
- "isDir": True,
- "size": 0,
- }
- for index in range(content_router._DIRECTORY_ARCHIVE_MAX_ENTRIES + 1)
- ]
-
- async def read_file_bytes(self, *args, **kwargs):
- raise AssertionError("entry cap must fail before reading file contents")
-
- service = SimpleNamespace(fs=FakeFS())
- with pytest.raises(Exception, match="entry download limit"):
- await content_router._build_directory_archive(
- service,
- "viking://resources/project",
- {"name": "project", "isDir": True},
- SimpleNamespace(),
- )
-
- assert observed["node_limit"] == content_router._DIRECTORY_ARCHIVE_MAX_ENTRIES + 1
-
-
-async def test_build_directory_archive_rejects_zip_over_limit(monkeypatch):
- class FakeFS:
- async def tree(self, *args, **kwargs):
- return [
- {
- "uri": "viking://resources/project/empty",
- "rel_path": "empty",
- "isDir": True,
- "size": 0,
- }
- ]
-
- monkeypatch.setattr(content_router, "_DIRECTORY_ARCHIVE_MAX_BYTES", 1)
- service = SimpleNamespace(fs=FakeFS())
- with pytest.raises(Exception, match="exceeds the 1-byte download limit"):
- await content_router._build_directory_archive(
- service,
- "viking://resources/project",
- {"name": "project", "isDir": True},
- SimpleNamespace(),
- )
-
-
-async def test_build_directory_archive_stops_writing_once_over_limit(monkeypatch):
- """Header-only entries must not grow the archive buffer past the limit."""
- limit = 4096
- peak = {"bytes": 0}
-
- class FakeFS:
- async def tree(self, *args, **kwargs):
- return [
- {
- "uri": f"viking://resources/project/d{index}",
- "rel_path": f"d{index}",
- "isDir": True,
- }
- for index in range(20000)
- ]
-
- class TrackedBytesIO(io.BytesIO):
- def write(self, data):
- count = super().write(data)
- peak["bytes"] = max(peak["bytes"], self.tell())
- return count
-
- monkeypatch.setattr(content_router, "_DIRECTORY_ARCHIVE_MAX_BYTES", limit)
- monkeypatch.setattr(content_router, "io", SimpleNamespace(BytesIO=TrackedBytesIO))
-
- with pytest.raises(Exception, match="download limit"):
- await content_router._build_directory_archive(
- SimpleNamespace(fs=FakeFS()),
- "viking://resources/project",
- {"name": "project", "isDir": True},
- SimpleNamespace(),
- )
-
- # Without the in-loop guard this reaches ~1.9 MB for 20k entries.
- assert peak["bytes"] < limit * 4
-
-
-def test_archive_size_limit_error_is_not_retryable():
- """Oversize is a property of the directory, so it must be 413, not 429."""
- error = content_router._archive_size_limit_error("viking://resources/big")
- assert error.code == "PAYLOAD_TOO_LARGE"
- assert ERROR_CODE_TO_HTTP_STATUS[error.code] == 413
-
-
[email protected]("path", ["../escape", "/absolute", r"..\\escape"])
-def test_safe_archive_path_rejects_unsafe_members(path):
- with pytest.raises(Exception, match="Unsafe path"):
- content_router._safe_archive_path("project", path)
-
-
async def test_download_missing_uri_returns_not_found(app, service, monkeypatch):
async def fake_read_file_bytes(*args, **kwargs):
raise FileNotFoundError("download target missing")