Files
mnote/rust/crates/mnote-web/src/routes/local_ocr.rs
T

1912 lines
68 KiB
Rust
Raw Normal View History

2026-06-01 09:29:12 +08:00
use crate::app::AppState;
use crate::context::RequestContext;
use crate::error::WebError;
use crate::routes::local_folder_source::{
decode_local_id_segment, encode_local_id_segment,
ensure_local_workspace_read_access_with_state, ensure_local_workspace_write_access_with_state,
};
use axum::extract::{Extension, Query, State};
use axum::http::{HeaderMap, StatusCode};
use axum::Json;
use serde::{Deserialize, Serialize};
use serde_json::{json, Value};
use std::collections::{BTreeMap, HashMap};
use std::fs;
use std::hash::{Hash, Hasher};
use std::io::{Cursor, Read};
use std::path::{Component, Path, PathBuf};
use std::time::{Duration, SystemTime, UNIX_EPOCH};
const OCR_INDEX_VERSION: u32 = 1;
const DEFAULT_PROVIDER: &str = "mineru";
const DEFAULT_MODEL_VERSION: &str = "vlm";
const DEFAULT_MINERU_API_BASE_URL: &str = "https://mineru.net";
#[derive(Debug, Deserialize)]
#[serde(rename_all = "camelCase")]
pub(crate) struct OcrJobRequest {
root_uri: String,
document_id: String,
#[serde(default)]
source_path: Option<String>,
source_root_relative_path: String,
#[serde(default)]
provider: Option<String>,
#[serde(default)]
force: bool,
#[serde(default)]
mock_markdown: Option<String>,
#[serde(default)]
mock_error: Option<String>,
}
#[derive(Debug, Deserialize)]
#[serde(rename_all = "camelCase")]
pub(crate) struct OcrStatusQuery {
root_uri: String,
source_root_relative_path: Option<String>,
}
#[derive(Debug, Deserialize)]
#[serde(rename_all = "camelCase")]
pub(crate) struct OcrReadQuery {
root_uri: String,
ocr_root_relative_path: String,
}
#[derive(Debug, Deserialize)]
#[serde(rename_all = "camelCase")]
pub(crate) struct OcrInsertRequest {
root_uri: String,
document_id: String,
ocr_root_relative_path: String,
#[serde(default)]
mode: Option<String>,
}
#[derive(Debug, Clone, Serialize, Deserialize)]
#[serde(rename_all = "camelCase")]
struct OcrIndex {
version: u32,
#[serde(default)]
entries: BTreeMap<String, OcrIndexEntry>,
}
#[derive(Debug, Clone, Serialize, Deserialize)]
#[serde(rename_all = "camelCase")]
pub(crate) struct OcrIndexEntry {
pub(crate) job_id: String,
pub(crate) owner_document_id: String,
pub(crate) owner_document_path: String,
pub(crate) source_root_relative_path: String,
pub(crate) ocr_root_relative_path: String,
pub(crate) provider: String,
pub(crate) model_version: String,
pub(crate) status: String,
pub(crate) source_size: u64,
pub(crate) source_mtime_ms: u128,
pub(crate) created_at_ms: u128,
pub(crate) updated_at_ms: u128,
#[serde(default)]
pub(crate) plain_text_preview: String,
#[serde(default, skip_serializing_if = "Option::is_none")]
pub(crate) error: Option<String>,
}
#[derive(Debug, Clone)]
struct SourceMetadata {
size: u64,
mtime_ms: u128,
}
#[derive(Debug, Clone)]
struct MineruClientConfig {
api_base_url: String,
token: String,
model_version: String,
poll_interval: Duration,
max_polls: usize,
}
#[derive(Debug, Clone)]
struct OcrSidecarPlan {
owner_document_path: String,
source_root_relative_path: String,
ocr_root_relative_path: String,
ocr_path: PathBuf,
provider: String,
model_version: String,
source_size: u64,
source_mtime_ms: u128,
}
#[derive(Debug, Clone)]
pub(crate) struct OcrFrontmatter {
pub provider: String,
pub owner_document: String,
pub source_path: String,
pub source_root_relative_path: String,
pub source_size: u64,
pub source_mtime_ms: u128,
pub status: String,
}
pub(crate) async fn create_job(
State(state): State<AppState>,
Extension(context): Extension<RequestContext>,
Json(body): Json<OcrJobRequest>,
) -> Result<(StatusCode, HeaderMap, Json<Value>), WebError> {
let root_uri = body.root_uri.trim();
let root = ensure_local_workspace_write_access_with_state(&state, &context, root_uri)
.map_err(|error| error.with_context(&context))?;
let provider = body
.provider
.as_deref()
.map(str::trim)
.filter(|value| !value.is_empty())
.unwrap_or(DEFAULT_PROVIDER);
let token = if provider != "mock" {
Some(mineru_token().ok_or_else(|| {
WebError::new(
StatusCode::UNAUTHORIZED,
"mineru_token_missing",
"缺少 MinerU API token",
)
.with_context(&context)
})?)
} else {
None
};
if provider != "mock" && token.is_none() {
return Err(WebError::new(
StatusCode::UNAUTHORIZED,
"mineru_token_missing",
"缺少 MinerU API token",
)
.with_context(&context));
}
let source_relative = body.source_root_relative_path.trim().replace('\\', "/");
let _source_path_hint = body
.source_path
.as_deref()
.map(str::trim)
.filter(|value| !value.is_empty())
.unwrap_or(&source_relative);
let plan = plan_ocr_sidecar_path(
&root,
&body.document_id,
&source_relative,
provider,
DEFAULT_MODEL_VERSION,
body.force,
)?;
let now = now_ms();
if let Some(error) = body.mock_error.as_deref() {
let entry = build_index_entry(&plan, "failed", now, "", Some(redact_error(error)));
upsert_ocr_index_entry(&root, entry.clone())?;
return Ok(ok_json(
&context,
json!({
"ok": true,
"job": ocr_job_payload(&root, &entry),
}),
));
}
let markdown = if provider == "mock" {
body.mock_markdown
.unwrap_or_else(|| format!("OCR mock result for {}", plan.source_root_relative_path))
} else {
match run_mineru_ocr(&root, &plan, token.unwrap_or_default()).await {
Ok(markdown) => markdown,
Err(error) => {
2026-06-01 10:07:42 +08:00
let failed_entry = build_index_entry(
&plan,
"failed",
now,
"",
Some(redact_error(error.message())),
);
2026-06-01 09:29:12 +08:00
upsert_ocr_index_entry(&root, failed_entry)?;
return Err(error.with_context(&context));
}
}
};
write_ocr_sidecar(&plan, &markdown, now)?;
let entry = build_index_entry(&plan, "done", now, &markdown, None);
upsert_ocr_index_entry(&root, entry.clone())?;
Ok(ok_json(
&context,
json!({
"ok": true,
"job": ocr_job_payload(&root, &entry),
}),
))
}
pub(crate) async fn list_jobs(
State(state): State<AppState>,
Extension(context): Extension<RequestContext>,
Query(query): Query<OcrStatusQuery>,
) -> Result<(StatusCode, HeaderMap, Json<Value>), WebError> {
let root =
ensure_local_workspace_read_access_with_state(&state, &context, query.root_uri.trim())
.map_err(|error| error.with_context(&context))?;
let index = read_ocr_index(&root)?;
let jobs = index
.entries
.values()
.map(|entry| ocr_job_payload(&root, entry))
.collect::<Vec<_>>();
Ok(ok_json(&context, json!({ "ok": true, "jobs": jobs })))
}
pub(crate) async fn status(
State(state): State<AppState>,
Extension(context): Extension<RequestContext>,
Query(query): Query<OcrStatusQuery>,
) -> Result<(StatusCode, HeaderMap, Json<Value>), WebError> {
let root =
ensure_local_workspace_read_access_with_state(&state, &context, query.root_uri.trim())
.map_err(|error| error.with_context(&context))?;
let source = query
.source_root_relative_path
.as_deref()
.map(str::trim)
.filter(|value| !value.is_empty())
.ok_or_else(|| {
WebError::bad_request_code("local_ocr_source_required", "OCR 状态查询缺少来源路径")
.with_context(&context)
})?;
let index = read_ocr_index(&root)?;
let job = index
.entries
.get(source)
.map(|entry| ocr_job_payload(&root, entry));
Ok(ok_json(&context, json!({ "ok": true, "job": job })))
}
pub(crate) async fn read(
State(state): State<AppState>,
Extension(context): Extension<RequestContext>,
Query(query): Query<OcrReadQuery>,
) -> Result<(StatusCode, HeaderMap, Json<Value>), WebError> {
let root =
ensure_local_workspace_read_access_with_state(&state, &context, query.root_uri.trim())
.map_err(|error| error.with_context(&context))?;
let relative = normalize_relative_path(&query.ocr_root_relative_path)?;
if !is_local_ocr_sidecar_relative_path(&relative) {
return Err(WebError::bad_request_code(
"local_ocr_sidecar_invalid",
"OCR 读取目标不是 OCR sidecar",
)
.with_context(&context));
}
let path = root.join(&relative);
ensure_target_under_root(&root, &path, "local_ocr_read_root_escape")?;
let markdown = fs::read_to_string(&path).map_err(|error| {
WebError::bad_request_code(
"local_ocr_read_failed",
format!("无法读取 OCR Markdown {}: {error}", path.display()),
)
.with_context(&context)
})?;
let frontmatter = parse_ocr_frontmatter(&markdown);
Ok(ok_json(
&context,
json!({
"ok": true,
"markdown": markdown,
"frontmatter": frontmatter.map(ocr_frontmatter_payload),
"ocrRootRelativePath": relative,
}),
))
}
pub(crate) async fn insert(
State(state): State<AppState>,
Extension(context): Extension<RequestContext>,
Json(body): Json<OcrInsertRequest>,
) -> Result<(StatusCode, HeaderMap, Json<Value>), WebError> {
let root =
ensure_local_workspace_write_access_with_state(&state, &context, body.root_uri.trim())
.map_err(|error| error.with_context(&context))?;
let mode = body
.mode
.as_deref()
.map(str::trim)
.filter(|value| !value.is_empty())
.unwrap_or("link");
if mode != "link" {
return Err(WebError::bad_request_code(
"local_ocr_insert_mode_unsupported",
"当前仅支持插入 OCR 链接",
)
.with_context(&context));
}
let owner_document_path = owner_document_path_from_id(&body.document_id)?;
let owner_path = root.join(&owner_document_path);
ensure_target_under_root(&root, &owner_path, "local_ocr_owner_root_escape")
.map_err(|error| error.with_context(&context))?;
if !owner_path.is_file() {
return Err(WebError::bad_request_code(
"local_ocr_owner_missing",
"OCR owner Markdown 不存在",
)
.with_context(&context));
}
let ocr_root_relative_path = normalize_relative_path(&body.ocr_root_relative_path)?;
if !is_local_ocr_sidecar_relative_path(&ocr_root_relative_path) {
return Err(WebError::bad_request_code(
"local_ocr_sidecar_invalid",
"OCR 插入目标不是 OCR sidecar",
)
.with_context(&context));
}
let ocr_path = root.join(&ocr_root_relative_path);
ensure_target_under_root(&root, &ocr_path, "local_ocr_read_root_escape")
.map_err(|error| error.with_context(&context))?;
if !ocr_path.is_file() {
return Err(
WebError::bad_request_code("local_ocr_read_failed", "OCR Markdown 不存在")
.with_context(&context),
);
}
let current = fs::read_to_string(&owner_path).map_err(|error| {
WebError::bad_request_code(
"local_ocr_owner_read_failed",
format!(
"无法读取 OCR owner Markdown {}: {error}",
owner_path.display()
),
)
.with_context(&context)
})?;
let link_target = relative_from_owner_dir(&owner_document_path, &ocr_root_relative_path);
let label = ocr_link_label(&ocr_root_relative_path);
let inserted_markdown = format!("[OCR{}]({})", label, link_target);
let mut next = current.trim_end_matches('\n').to_string();
if next.is_empty() {
next.push_str(&inserted_markdown);
next.push('\n');
} else {
next.push_str("\n\n");
next.push_str(&inserted_markdown);
next.push('\n');
}
fs::write(&owner_path, next).map_err(|error| {
WebError::bad_request_code(
"local_ocr_owner_write_failed",
format!(
"无法写入 OCR owner Markdown {}: {error}",
owner_path.display()
),
)
.with_context(&context)
})?;
Ok(ok_json(
&context,
json!({
"ok": true,
"documentId": body.document_id,
"ownerDocumentPath": owner_document_path,
"ocrRootRelativePath": ocr_root_relative_path,
"insertedMarkdown": inserted_markdown,
}),
))
}
pub(crate) fn is_local_ocr_sidecar_relative_path(relative_path: &str) -> bool {
let normalized = relative_path.trim().replace('\\', "/");
let path = Path::new(&normalized);
let Some(file_name) = path.file_name().and_then(|value| value.to_str()) else {
return false;
};
let Some(parent_name) = path
.parent()
.and_then(|value| value.file_name())
.and_then(|value| value.to_str())
else {
return false;
};
parent_name.ends_with(".ocr") && file_name.ends_with(".ocr.md")
}
pub(crate) fn parse_ocr_frontmatter(markdown: &str) -> Option<OcrFrontmatter> {
let trimmed = markdown.strip_prefix("---\n")?;
let end = trimmed.find("\n---")?;
let frontmatter = &trimmed[..end];
let mut fields = HashMap::<String, String>::new();
for line in frontmatter.lines() {
let Some((key, value)) = line.split_once(':') else {
continue;
};
fields.insert(
key.trim().to_string(),
value.trim().trim_matches('"').to_string(),
);
}
if fields.get("mnote_ocr_version").map(String::as_str) != Some("1") {
return None;
}
Some(OcrFrontmatter {
provider: fields.get("provider")?.to_string(),
owner_document: fields.get("owner_document")?.to_string(),
source_path: fields.get("source_path")?.to_string(),
source_root_relative_path: fields.get("source_root_relative_path")?.to_string(),
source_size: fields.get("source_size")?.parse().ok()?,
source_mtime_ms: fields.get("source_mtime_ms")?.parse().ok()?,
status: fields.get("status")?.to_string(),
})
}
#[cfg(test)]
fn read_ocr_index_entry(
root: &Path,
source_root_relative_path: &str,
) -> Result<Option<OcrIndexEntry>, WebError> {
let source = normalize_relative_path(source_root_relative_path)?;
Ok(read_ocr_index(root)?.entries.remove(&source))
}
pub(crate) fn ocr_index_entries(root: &Path) -> Result<Vec<OcrIndexEntry>, WebError> {
Ok(read_ocr_index(root)?.entries.into_values().collect())
}
pub(crate) fn strip_ocr_frontmatter(markdown: &str) -> &str {
let Some(trimmed) = markdown.strip_prefix("---\n") else {
return markdown;
};
let Some(end) = trimmed.find("\n---") else {
return markdown;
};
trimmed[end + 4..].trim_start_matches('\n')
}
2026-06-01 10:07:42 +08:00
async fn run_mineru_ocr(
root: &Path,
plan: &OcrSidecarPlan,
token: String,
) -> Result<String, WebError> {
2026-06-01 09:29:12 +08:00
let config = MineruClientConfig {
api_base_url: mineru_api_base_url(),
token,
model_version: plan.model_version.clone(),
poll_interval: mineru_poll_interval(),
max_polls: mineru_max_polls(),
};
let source_path = root.join(&plan.source_root_relative_path);
let bytes = fs::read(&source_path).map_err(|error| {
WebError::bad_request_code(
"mineru_source_read_failed",
format!(
"无法读取 MinerU OCR 来源文件 {}: {error}",
source_path.display()
),
)
})?;
let file_name = Path::new(&plan.source_root_relative_path)
.file_name()
.and_then(|value| value.to_str())
.unwrap_or("source.pdf");
let client = reqwest::Client::new();
let upload = create_mineru_upload_task(&client, &config, file_name).await?;
upload_mineru_source(&client, &upload.upload_url, bytes).await?;
let zip_url = poll_mineru_result_zip_url(&client, &config, &upload.batch_id).await?;
let zip_bytes = download_mineru_result_zip(&client, &zip_url).await?;
extract_mineru_markdown_from_zip(&zip_bytes)
}
#[derive(Debug)]
struct MineruUploadTask {
batch_id: String,
upload_url: String,
}
async fn create_mineru_upload_task(
client: &reqwest::Client,
config: &MineruClientConfig,
file_name: &str,
) -> Result<MineruUploadTask, WebError> {
let url = format!(
"{}/api/v4/file-urls/batch",
config.api_base_url.trim_end_matches('/')
);
let payload = json!({
"files": [{ "name": file_name, "data_id": short_hash(file_name) }],
"model_version": config.model_version,
});
let response = client
.post(&url)
.bearer_auth(&config.token)
.json(&payload)
.send()
.await
.map_err(|error| {
WebError::bad_gateway_code(
"mineru_upload_task_failed",
2026-06-01 10:07:42 +08:00
format!(
"MinerU 上传任务创建失败: {}",
redact_error(&error.to_string())
),
2026-06-01 09:29:12 +08:00
)
})?;
let status = response.status();
let body = response.text().await.unwrap_or_default();
if !status.is_success() {
return Err(WebError::bad_gateway_code(
"mineru_upload_task_http_failed",
format!("MinerU 上传任务创建失败: HTTP {}", status.as_u16()),
));
}
let value = parse_mineru_json(&body, "mineru_upload_task_json_invalid")?;
2026-06-01 10:07:42 +08:00
let batch_id =
find_json_string_by_keys(&value, &["batch_id", "batchId", "id"]).ok_or_else(|| {
2026-06-01 09:29:12 +08:00
WebError::bad_gateway_code(
"mineru_upload_task_batch_missing",
"MinerU 上传任务响应缺少 batch_id",
)
})?;
let upload_url = find_upload_url(&value).ok_or_else(|| {
WebError::bad_gateway_code(
"mineru_upload_url_missing",
"MinerU 上传任务响应缺少 upload_url",
)
})?;
Ok(MineruUploadTask {
batch_id,
upload_url,
})
}
async fn upload_mineru_source(
client: &reqwest::Client,
upload_url: &str,
bytes: Vec<u8>,
) -> Result<(), WebError> {
let response = client
.put(upload_url)
.body(bytes)
.send()
.await
.map_err(|error| {
WebError::bad_gateway_code(
"mineru_source_upload_failed",
2026-06-01 10:07:42 +08:00
format!(
"MinerU 源文件上传失败: {}",
redact_error(&error.to_string())
),
2026-06-01 09:29:12 +08:00
)
})?;
if !response.status().is_success() {
return Err(WebError::bad_gateway_code(
"mineru_source_upload_http_failed",
format!("MinerU 源文件上传失败: HTTP {}", response.status().as_u16()),
));
}
Ok(())
}
async fn poll_mineru_result_zip_url(
client: &reqwest::Client,
config: &MineruClientConfig,
batch_id: &str,
) -> Result<String, WebError> {
let url = format!(
"{}/api/v4/extract-results/batch/{}",
config.api_base_url.trim_end_matches('/'),
batch_id
);
for _ in 0..config.max_polls.max(1) {
let response = client
.get(&url)
.bearer_auth(&config.token)
.send()
.await
.map_err(|error| {
WebError::bad_gateway_code(
"mineru_result_poll_failed",
format!("MinerU 结果轮询失败: {}", redact_error(&error.to_string())),
)
})?;
let status = response.status();
let body = response.text().await.unwrap_or_default();
if !status.is_success() {
return Err(WebError::bad_gateway_code(
"mineru_result_poll_http_failed",
format!("MinerU 结果轮询失败: HTTP {}", status.as_u16()),
));
}
let value = parse_mineru_json(&body, "mineru_result_poll_json_invalid")?;
if let Some(error_message) = mineru_result_error(&value) {
return Err(WebError::bad_gateway_code(
"mineru_result_failed",
format!("MinerU 识别失败: {}", redact_error(&error_message)),
));
}
if let Some(zip_url) = find_json_string_by_keys(
&value,
2026-06-01 10:07:42 +08:00
&[
"full_zip_url",
"fullZipUrl",
"zip_url",
"zipUrl",
"result_url",
"resultUrl",
],
2026-06-01 09:29:12 +08:00
) {
return Ok(zip_url);
}
tokio::time::sleep(config.poll_interval).await;
}
Err(WebError::gateway_timeout_code(
"mineru_result_poll_timeout",
"MinerU 识别结果轮询超时",
))
}
async fn download_mineru_result_zip(
client: &reqwest::Client,
zip_url: &str,
) -> Result<Vec<u8>, WebError> {
let response = client.get(zip_url).send().await.map_err(|error| {
WebError::bad_gateway_code(
"mineru_result_download_failed",
2026-06-01 10:07:42 +08:00
format!(
"MinerU 结果包下载失败: {}",
redact_error(&error.to_string())
),
2026-06-01 09:29:12 +08:00
)
})?;
if !response.status().is_success() {
return Err(WebError::bad_gateway_code(
"mineru_result_download_http_failed",
format!("MinerU 结果包下载失败: HTTP {}", response.status().as_u16()),
));
}
2026-06-01 10:07:42 +08:00
response
.bytes()
.await
.map(|bytes| bytes.to_vec())
.map_err(|error| {
WebError::bad_gateway_code(
"mineru_result_download_failed",
format!(
"MinerU 结果包读取失败: {}",
redact_error(&error.to_string())
),
)
})
2026-06-01 09:29:12 +08:00
}
fn extract_mineru_markdown_from_zip(bytes: &[u8]) -> Result<String, WebError> {
let cursor = Cursor::new(bytes);
let mut archive = zip::ZipArchive::new(cursor).map_err(|error| {
WebError::bad_gateway_code(
"mineru_result_zip_invalid",
format!("MinerU 结果包不是合法 zip: {error}"),
)
})?;
let mut candidates = Vec::<(String, String)>::new();
for index in 0..archive.len() {
let mut file = archive.by_index(index).map_err(|error| {
WebError::bad_gateway_code(
"mineru_result_zip_read_failed",
format!("MinerU 结果包读取失败: {error}"),
)
})?;
let name = file.name().replace('\\', "/");
if !name.to_ascii_lowercase().ends_with(".md") || name.contains("/.") {
continue;
}
let mut markdown = String::new();
file.read_to_string(&mut markdown).map_err(|error| {
WebError::bad_gateway_code(
"mineru_result_markdown_read_failed",
format!("MinerU Markdown 读取失败: {error}"),
)
})?;
candidates.push((name, markdown));
}
candidates
.into_iter()
.max_by_key(|(name, markdown)| {
2026-06-01 10:07:42 +08:00
let preferred =
name.ends_with("/full.md") || name == "full.md" || name.ends_with("/result.md");
2026-06-01 09:29:12 +08:00
(preferred, markdown.len())
})
.map(|(_, markdown)| markdown)
.filter(|markdown| !markdown.trim().is_empty())
.ok_or_else(|| {
WebError::bad_gateway_code(
"mineru_result_markdown_missing",
"MinerU 结果包中缺少 Markdown 文件",
)
})
}
fn parse_mineru_json(body: &str, code: &'static str) -> Result<Value, WebError> {
serde_json::from_str::<Value>(body).map_err(|error| {
WebError::bad_gateway_code(code, format!("MinerU 响应 JSON 解析失败: {error}"))
})
}
fn find_upload_url(value: &Value) -> Option<String> {
if let Some(url) = find_json_string_by_keys(value, &["upload_url", "uploadUrl"]) {
return Some(url);
}
None
}
fn find_json_string_by_keys(value: &Value, keys: &[&str]) -> Option<String> {
match value {
Value::Object(map) => {
for key in keys {
if let Some(text) = map.get(*key).and_then(Value::as_str) {
let trimmed = text.trim();
if !trimmed.is_empty() {
return Some(trimmed.to_string());
}
}
}
for nested in map.values() {
if let Some(found) = find_json_string_by_keys(nested, keys) {
return Some(found);
}
}
None
}
Value::Array(items) => items
.iter()
.find_map(|item| find_json_string_by_keys(item, keys)),
_ => None,
}
}
fn mineru_result_error(value: &Value) -> Option<String> {
let status = find_json_string_by_keys(value, &["state", "status"])
.unwrap_or_default()
.to_ascii_lowercase();
if matches!(
status.as_str(),
"failed" | "fail" | "error" | "interrupted" | "canceled" | "cancelled"
) {
return find_json_string_by_keys(value, &["error", "message", "msg", "err_msg", "errMsg"])
.or_else(|| Some(status));
}
None
}
fn ok_json(context: &RequestContext, result: Value) -> (StatusCode, HeaderMap, Json<Value>) {
let mut headers = HeaderMap::new();
if let Ok(value) = axum::http::HeaderValue::from_str(&context.trace.request_id) {
headers.insert("x-request-id", value);
}
(StatusCode::OK, headers, Json(result))
}
fn plan_ocr_sidecar_path(
root: &Path,
document_id: &str,
source_root_relative_path: &str,
provider: &str,
model_version: &str,
force: bool,
) -> Result<OcrSidecarPlan, WebError> {
let owner_document_path = owner_document_path_from_id(document_id)?;
let source_root_relative_path = normalize_relative_path(source_root_relative_path)?;
let source_path = root.join(&source_root_relative_path);
ensure_target_under_root(root, &source_path, "local_ocr_source_root_escape")?;
if !source_path.is_file() {
return Err(WebError::bad_request_code(
"local_ocr_source_missing",
"OCR 来源文件不存在",
));
}
if !is_supported_ocr_source(&source_path) {
return Err(WebError::bad_request_code(
"local_ocr_source_type_unsupported",
"OCR 仅支持图片和 PDF",
));
}
let owner_path = root.join(&owner_document_path);
ensure_target_under_root(root, &owner_path, "local_ocr_owner_root_escape")?;
if !owner_path.is_file() {
return Err(WebError::bad_request_code(
"local_ocr_owner_missing",
"OCR owner Markdown 不存在",
));
}
let metadata = fs::metadata(&source_path).map_err(|error| {
WebError::bad_request_code(
"local_ocr_source_stat_failed",
format!(
"无法读取 OCR 来源文件状态 {}: {error}",
source_path.display()
),
)
})?;
let source_metadata = SourceMetadata {
size: metadata.len(),
mtime_ms: metadata
.modified()
.ok()
.and_then(|value| value.duration_since(UNIX_EPOCH).ok())
.map(|value| value.as_millis())
.unwrap_or_default(),
};
let owner_parent = Path::new(&owner_document_path)
.parent()
.map(Path::to_path_buf)
.unwrap_or_default();
let owner_stem = Path::new(&owner_document_path)
.file_stem()
.and_then(|value| value.to_str())
.unwrap_or("Page");
let ocr_dir = owner_parent.join(format!("{owner_stem}.ocr"));
let source_leaf = Path::new(&source_root_relative_path)
.file_name()
.and_then(|value| value.to_str())
.unwrap_or("source");
let base_file_name = format!("{source_leaf}.ocr.md");
let mut ocr_relative = ocr_dir.join(&base_file_name);
let default_path = root.join(&ocr_relative);
if !force && default_path.exists() {
let suffix = short_hash(&format!(
"{}:{}:{}",
source_root_relative_path, source_metadata.size, source_metadata.mtime_ms
));
ocr_relative = ocr_dir.join(format!("{source_leaf}-{suffix}.ocr.md"));
}
let ocr_root_relative_path = ocr_relative.to_string_lossy().replace('\\', "/");
let ocr_path = root.join(&ocr_root_relative_path);
ensure_target_under_root(root, &ocr_path, "local_ocr_sidecar_root_escape")?;
Ok(OcrSidecarPlan {
owner_document_path,
source_root_relative_path,
ocr_root_relative_path,
ocr_path,
provider: provider.to_string(),
model_version: model_version.to_string(),
source_size: source_metadata.size,
source_mtime_ms: source_metadata.mtime_ms,
})
}
fn write_ocr_sidecar(
plan: &OcrSidecarPlan,
markdown_body: &str,
now: u128,
) -> Result<(), WebError> {
if let Some(parent) = plan.ocr_path.parent() {
fs::create_dir_all(parent).map_err(|error| {
WebError::bad_request_code(
"local_ocr_sidecar_create_failed",
format!("无法创建 OCR sidecar 目录 {}: {error}", parent.display()),
)
})?;
}
let content = build_ocr_markdown(plan, markdown_body, "done", now);
fs::write(&plan.ocr_path, content).map_err(|error| {
WebError::bad_request_code(
"local_ocr_sidecar_write_failed",
format!("无法写入 OCR Markdown {}: {error}", plan.ocr_path.display()),
)
})
}
fn build_ocr_markdown(
plan: &OcrSidecarPlan,
markdown_body: &str,
status: &str,
timestamp_ms: u128,
) -> String {
let owner_document =
relative_from_sidecar_dir(&plan.ocr_root_relative_path, &plan.owner_document_path);
let source_path =
relative_from_owner_dir(&plan.owner_document_path, &plan.source_root_relative_path);
let timestamp = timestamp_ms.to_string();
format!(
"---\nmnote_ocr_version: 1\nprovider: {}\nmodel_version: {}\nowner_document: {}\nsource_path: {}\nsource_root_relative_path: {}\nsource_size: {}\nsource_mtime_ms: {}\nstatus: {}\ncreated_at: {}\nupdated_at: {}\n---\n\n{}\n",
plan.provider,
plan.model_version,
owner_document,
source_path,
plan.source_root_relative_path,
plan.source_size,
plan.source_mtime_ms,
status,
timestamp,
timestamp,
markdown_body.trim_end()
)
}
fn build_index_entry(
plan: &OcrSidecarPlan,
status: &str,
now: u128,
markdown: &str,
error: Option<String>,
) -> OcrIndexEntry {
OcrIndexEntry {
job_id: format!(
"ocr_{}_{}",
now,
short_hash(&plan.source_root_relative_path)
),
owner_document_id: format!(
"local-md:{}",
encode_local_id_segment(&plan.owner_document_path)
),
owner_document_path: plan.owner_document_path.clone(),
source_root_relative_path: plan.source_root_relative_path.clone(),
ocr_root_relative_path: plan.ocr_root_relative_path.clone(),
provider: plan.provider.clone(),
model_version: plan.model_version.clone(),
status: status.to_string(),
source_size: plan.source_size,
source_mtime_ms: plan.source_mtime_ms,
created_at_ms: now,
updated_at_ms: now,
plain_text_preview: markdown.chars().take(240).collect(),
error,
}
}
fn read_ocr_index(root: &Path) -> Result<OcrIndex, WebError> {
let path = ocr_index_path(root);
if !path.exists() {
return Ok(OcrIndex {
version: OCR_INDEX_VERSION,
entries: BTreeMap::new(),
});
}
let content = fs::read_to_string(&path).map_err(|error| {
WebError::bad_request_code(
"local_ocr_index_read_failed",
format!("无法读取 OCR 索引 {}: {error}", path.display()),
)
})?;
serde_json::from_str::<OcrIndex>(&content).map_err(|error| {
WebError::bad_request_code(
"local_ocr_index_invalid",
format!("OCR 索引格式非法 {}: {error}", path.display()),
)
})
}
fn write_ocr_index(root: &Path, index: &OcrIndex) -> Result<(), WebError> {
let path = ocr_index_path(root);
if let Some(parent) = path.parent() {
fs::create_dir_all(parent).map_err(|error| {
WebError::bad_request_code(
"local_ocr_index_create_failed",
format!("无法创建 OCR 索引目录 {}: {error}", parent.display()),
)
})?;
}
let content = serde_json::to_vec_pretty(index)
.map_err(|error| WebError::internal(format!("OCR 索引序列化失败: {error}")))?;
let tmp = path.with_extension("json.tmp");
fs::write(&tmp, [&content[..], b"\n"].concat()).map_err(|error| {
WebError::bad_request_code(
"local_ocr_index_write_failed",
format!("无法写入 OCR 临时索引 {}: {error}", tmp.display()),
)
})?;
fs::rename(&tmp, &path).map_err(|error| {
WebError::bad_request_code(
"local_ocr_index_write_failed",
format!("无法替换 OCR 索引 {}: {error}", path.display()),
)
})
}
fn upsert_ocr_index_entry(root: &Path, entry: OcrIndexEntry) -> Result<(), WebError> {
let mut index = read_ocr_index(root)?;
index.version = OCR_INDEX_VERSION;
index
.entries
.insert(entry.source_root_relative_path.clone(), entry);
write_ocr_index(root, &index)
}
fn ocr_job_payload(root: &Path, entry: &OcrIndexEntry) -> Value {
let stale = source_is_stale(root, entry);
json!({
"jobId": entry.job_id,
"ownerDocumentId": entry.owner_document_id,
"ownerDocumentPath": entry.owner_document_path,
"sourceRootRelativePath": entry.source_root_relative_path,
"ocrRootRelativePath": entry.ocr_root_relative_path,
"provider": entry.provider,
"modelVersion": entry.model_version,
"status": if stale && entry.status == "done" { "stale" } else { entry.status.as_str() },
"stageLabel": stage_label(if stale && entry.status == "done" { "stale" } else { entry.status.as_str() }),
"stale": stale,
"updatedAtMs": entry.updated_at_ms,
"finishedAtMs": if matches!(entry.status.as_str(), "done" | "failed") { Some(entry.updated_at_ms) } else { None },
"plainTextPreview": entry.plain_text_preview,
"error": entry.error,
})
}
fn source_is_stale(root: &Path, entry: &OcrIndexEntry) -> bool {
let source = root.join(&entry.source_root_relative_path);
let Ok(metadata) = fs::metadata(source) else {
return true;
};
let mtime_ms = metadata
.modified()
.ok()
.and_then(|value| value.duration_since(UNIX_EPOCH).ok())
.map(|value| value.as_millis())
.unwrap_or_default();
metadata.len() != entry.source_size || mtime_ms != entry.source_mtime_ms
}
fn stage_label(status: &str) -> &'static str {
match status {
"queued" => "排队中",
"uploading" => "上传中",
"mineru_processing" => "识别中",
"downloading" => "下载中",
"writing_sidecar" => "写入中",
"done" => "已识别",
"failed" => "识别失败",
"interrupted" => "已中断",
"stale" => "来源已变化",
_ => "未知状态",
}
}
fn ocr_frontmatter_payload(frontmatter: OcrFrontmatter) -> Value {
json!({
"provider": frontmatter.provider,
"ownerDocument": frontmatter.owner_document,
"sourcePath": frontmatter.source_path,
"sourceRootRelativePath": frontmatter.source_root_relative_path,
"sourceSize": frontmatter.source_size,
"sourceMtimeMs": frontmatter.source_mtime_ms,
"status": frontmatter.status,
})
}
fn ocr_index_path(root: &Path) -> PathBuf {
root.join(".mnote").join("ocr-index.json")
}
fn owner_document_path_from_id(document_id: &str) -> Result<String, WebError> {
let encoded = document_id
.trim()
.strip_prefix("local-md:")
.ok_or_else(|| {
WebError::bad_request_code(
"local_ocr_owner_document_invalid",
"OCR owner documentId 必须是 local-md 路径型 ID",
)
})?;
let decoded = decode_local_id_segment(encoded)?;
normalize_relative_path(&decoded)
}
fn normalize_relative_path(value: &str) -> Result<String, WebError> {
let normalized = value.trim().trim_start_matches('/').replace('\\', "/");
if normalized.is_empty() {
return Err(WebError::bad_request_code(
"local_ocr_path_required",
"OCR 路径不能为空",
));
}
let path = Path::new(&normalized);
if path.is_absolute()
|| path.components().any(|component| {
matches!(
component,
Component::ParentDir | Component::RootDir | Component::Prefix(_)
)
})
{
return Err(WebError::bad_request_code(
"local_ocr_path_escape",
"OCR 相对路径不能越过 root",
));
}
Ok(normalized)
}
fn ensure_target_under_root(
root: &Path,
target: &Path,
code: &'static str,
) -> Result<(), WebError> {
let parent = target.parent().unwrap_or(root);
let canonical_parent = parent
.canonicalize()
.unwrap_or_else(|_| parent.to_path_buf());
let canonical_root = root.canonicalize().unwrap_or_else(|_| root.to_path_buf());
if !canonical_parent.starts_with(&canonical_root) {
return Err(WebError::bad_request_code(code, "OCR 路径不能越过 root"));
}
Ok(())
}
fn is_supported_ocr_source(path: &Path) -> bool {
let ext = path
.extension()
.and_then(|value| value.to_str())
.unwrap_or_default()
.to_ascii_lowercase();
matches!(
ext.as_str(),
"pdf" | "png" | "jpg" | "jpeg" | "webp" | "bmp" | "tif" | "tiff"
)
}
fn relative_from_owner_dir(owner_document_path: &str, source_root_relative_path: &str) -> String {
let owner_parent = Path::new(owner_document_path)
.parent()
.unwrap_or(Path::new(""));
Path::new(source_root_relative_path)
.strip_prefix(owner_parent)
.map(|value| format!("./{}", value.to_string_lossy().replace('\\', "/")))
.unwrap_or_else(|_| source_root_relative_path.to_string())
}
fn relative_from_sidecar_dir(ocr_root_relative_path: &str, owner_document_path: &str) -> String {
let sidecar_parent = Path::new(ocr_root_relative_path)
.parent()
.unwrap_or_else(|| Path::new(""));
let owner = Path::new(owner_document_path);
let sidecar_components = sidecar_parent
.components()
.map(|component| component.as_os_str().to_string_lossy().to_string())
.collect::<Vec<_>>();
let owner_components = owner
.components()
.map(|component| component.as_os_str().to_string_lossy().to_string())
.collect::<Vec<_>>();
let common = sidecar_components
.iter()
.zip(owner_components.iter())
.take_while(|(left, right)| left == right)
.count();
let mut parts = Vec::<String>::new();
for _ in common..sidecar_components.len() {
parts.push("..".to_string());
}
parts.extend(owner_components.into_iter().skip(common));
if parts.is_empty() {
".".to_string()
} else {
parts.join("/")
}
}
fn short_hash(value: &str) -> String {
let mut hasher = std::collections::hash_map::DefaultHasher::new();
value.hash(&mut hasher);
format!("{:x}", hasher.finish()).chars().take(6).collect()
}
fn now_ms() -> u128 {
SystemTime::now()
.duration_since(UNIX_EPOCH)
.map(|value| value.as_millis())
.unwrap_or_default()
}
fn redact_error(error: &str) -> String {
let text = error.trim().replace('\n', " ");
let lower = text.to_ascii_lowercase();
if lower.contains("token") || lower.contains("signature") || lower.contains("x-oss-") {
return "provider_error_redacted".to_string();
}
text.chars().take(160).collect()
}
fn ocr_link_label(ocr_root_relative_path: &str) -> String {
Path::new(ocr_root_relative_path)
.file_name()
.and_then(|value| value.to_str())
.unwrap_or("OCR")
.trim_end_matches(".ocr.md")
.replace(['[', ']'], "")
}
fn mineru_token() -> Option<String> {
std::env::var("MNOTE_MINERU_API_TOKEN")
.ok()
.or_else(|| std::env::var("MINERU_API_TOKEN").ok())
.map(|value| value.trim().to_string())
.filter(|value| !value.is_empty())
}
fn mineru_api_base_url() -> String {
std::env::var("MNOTE_MINERU_API_BASE_URL")
.ok()
.or_else(|| std::env::var("MINERU_API_BASE_URL").ok())
.map(|value| value.trim().trim_end_matches('/').to_string())
.filter(|value| !value.is_empty())
.unwrap_or_else(|| DEFAULT_MINERU_API_BASE_URL.to_string())
}
fn mineru_poll_interval() -> Duration {
std::env::var("MNOTE_MINERU_POLL_INTERVAL_MS")
.ok()
.and_then(|value| value.trim().parse::<u64>().ok())
.filter(|value| *value > 0)
.map(Duration::from_millis)
.unwrap_or_else(|| Duration::from_secs(2))
}
fn mineru_max_polls() -> usize {
std::env::var("MNOTE_MINERU_MAX_POLLS")
.ok()
.and_then(|value| value.trim().parse::<usize>().ok())
.filter(|value| *value > 0)
.unwrap_or(90)
}
#[cfg(test)]
mod tests {
use super::*;
use crate::app::{build_app, AppConfig, AppState};
use axum::body::{to_bytes, Body};
use axum::http::Request;
use std::io::Write as _;
use std::sync::atomic::{AtomicUsize, Ordering};
use std::sync::Arc;
use tower::util::ServiceExt;
fn temp_root(name: &str) -> PathBuf {
let root = std::env::temp_dir().join(format!("{name}-{}", std::process::id()));
let _ = fs::remove_dir_all(&root);
fs::create_dir_all(root.join("docs").join("Page.assets")).expect("create assets");
fs::create_dir_all(root.join(".mnote")).expect("create metadata");
root
}
fn app() -> axum::Router {
build_app(AppState::new(AppConfig {
service_name: "mnote-web".into(),
service_version: "0.1.0".into(),
bind_addr: "127.0.0.1:0".into(),
public_bind_addr: "127.0.0.1:3000".into(),
legacy_next_base_url: Some("http://127.0.0.1:3100".into()),
enable_legacy_next_compat: true,
enable_debug_shell_routes: false,
enable_editor_actor: true,
hermes_base_path: "/api/hermes".into(),
compat_next_base_path: "/api/compat/next".into(),
convex_url: None,
convex_admin_key: None,
allow_dev_fixtures: true,
query_fixtures_json: None,
mutation_fixtures_json: None,
dev_user_id: "dev-user".into(),
dev_user_name: "开发用户".into(),
dev_user_email: "dev@mnote.local".into(),
}))
}
fn write_workspace_manifest(root: &Path) {
fs::write(
root.join(".mnote").join("workspace.json"),
r#"{"workspaceId":"local-ws-ocr","ownerId":"user_test","createdAt":"2026-06-01T00:00:00Z","capabilities":["local_files"]}"#,
)
.expect("manifest");
}
async fn post_ocr_job(root: &Path, body: Value) -> (StatusCode, Value) {
let response = app()
.oneshot(
Request::builder()
.method("POST")
.uri("/api/local-folder/ocr/jobs")
.header("content-type", "application/json")
.header("x-mnote-actor-id", "user_test")
.header("x-mnote-actor-type", "user")
.body(Body::from(body.to_string()))
.expect("request"),
)
.await
.expect("response");
let status = response.status();
let body = to_bytes(response.into_body(), usize::MAX)
.await
.expect("response body");
let payload = serde_json::from_slice(&body).expect("response json");
let _ = fs::remove_dir_all(root);
(status, payload)
}
fn query_escape(value: &str) -> String {
value
.bytes()
.map(|byte| match byte {
b'A'..=b'Z' | b'a'..=b'z' | b'0'..=b'9' | b'-' | b'_' | b'.' | b'~' => {
(byte as char).to_string()
}
_ => format!("%{byte:02X}"),
})
.collect()
}
fn build_test_mineru_zip(markdown: &str) -> Vec<u8> {
let mut bytes = Cursor::new(Vec::<u8>::new());
{
let mut writer = zip::ZipWriter::new(&mut bytes);
writer
.start_file("full.md", zip::write::SimpleFileOptions::default())
.expect("zip start file");
2026-06-01 10:07:42 +08:00
writer.write_all(markdown.as_bytes()).expect("zip markdown");
2026-06-01 09:29:12 +08:00
writer.finish().expect("zip finish");
}
bytes.into_inner()
}
#[test]
fn local_ocr_sidecar_path_and_frontmatter_contract() {
let root = temp_root("mnote-local-ocr-sidecar");
fs::write(root.join("docs").join("Page.md"), "# Page\n").expect("page");
fs::write(
root.join("docs").join("Page.assets").join("photo.png"),
b"png",
)
.expect("photo");
let plan = plan_ocr_sidecar_path(
&root,
"local-md:docs~2FPage.md",
"docs/Page.assets/photo.png",
"mineru",
"vlm",
false,
)
.expect("plan");
assert_eq!(
plan.ocr_root_relative_path,
"docs/Page.ocr/photo.png.ocr.md"
);
let markdown = build_ocr_markdown(&plan, "识别文本", "done", 1780000000000);
assert!(markdown.contains("mnote_ocr_version: 1"));
assert!(markdown.contains("provider: mineru"));
assert!(markdown.contains("model_version: vlm"));
assert!(markdown.contains("owner_document: ../Page.md"));
assert!(markdown.contains("source_path: ./Page.assets/photo.png"));
assert!(markdown.contains("source_root_relative_path: docs/Page.assets/photo.png"));
assert!(markdown.contains("status: done"));
let parsed = parse_ocr_frontmatter(&markdown).expect("frontmatter");
assert_eq!(
parsed.source_root_relative_path,
"docs/Page.assets/photo.png"
);
let _ = fs::remove_dir_all(root);
}
#[test]
fn local_ocr_index_marks_stale_and_redacts_error() {
let root = temp_root("mnote-local-ocr-index");
fs::write(root.join("docs").join("Page.md"), "# Page\n").expect("page");
let source = root.join("docs").join("Page.assets").join("scan.pdf");
fs::write(&source, b"pdf").expect("pdf");
let plan = plan_ocr_sidecar_path(
&root,
"local-md:docs~2FPage.md",
"docs/Page.assets/scan.pdf",
"mock",
"vlm",
false,
)
.expect("plan");
let entry = build_index_entry(
&plan,
"failed",
1780000000000,
"",
Some(redact_error("token=secret upload_url=https://example.test")),
);
upsert_ocr_index_entry(&root, entry.clone()).expect("write index");
let stored = read_ocr_index_entry(&root, "docs/Page.assets/scan.pdf")
.expect("read index")
.expect("entry");
assert_eq!(stored.error.as_deref(), Some("provider_error_redacted"));
fs::write(&source, b"pdf changed").expect("change source");
assert!(source_is_stale(&root, &stored));
let _ = fs::remove_dir_all(root);
}
#[tokio::test]
async fn local_ocr_jobs_route_writes_mock_sidecar_and_reads_status() {
let root = temp_root("mnote-local-ocr-route");
write_workspace_manifest(&root);
fs::write(root.join("docs").join("Page.md"), "# Page\n").expect("page");
fs::write(
root.join("docs").join("Page.assets").join("photo.png"),
b"png",
)
.expect("photo");
let root_uri = format!("file://{}", root.display());
let response = app()
.oneshot(
Request::builder()
.method("POST")
.uri("/api/local-folder/ocr/jobs")
.header("content-type", "application/json")
.header("x-mnote-actor-id", "user_test")
.header("x-mnote-actor-type", "user")
.body(Body::from(
json!({
"rootUri": root_uri,
"documentId": "local-md:docs~2FPage.md",
"sourceRootRelativePath": "docs/Page.assets/photo.png",
"provider": "mock",
"mockMarkdown": "Route OCR Token"
})
.to_string(),
))
.expect("request"),
)
.await
.expect("response");
assert_eq!(response.status(), StatusCode::OK);
assert!(root
.join("docs")
.join("Page.ocr")
.join("photo.png.ocr.md")
.is_file());
let escaped_root = query_escape(&root_uri);
let status_response = app()
.oneshot(
Request::builder()
.uri(format!(
"/api/local-folder/ocr/status?rootUri={escaped_root}&sourceRootRelativePath=docs%2FPage.assets%2Fphoto.png"
))
.header("x-mnote-actor-id", "user_test")
.header("x-mnote-actor-type", "user")
.body(Body::empty())
.expect("request"),
)
.await
.expect("status response");
assert_eq!(status_response.status(), StatusCode::OK);
let status_body = to_bytes(status_response.into_body(), usize::MAX)
.await
.expect("status body");
let status_payload: Value = serde_json::from_slice(&status_body).expect("status json");
assert_eq!(status_payload["job"]["status"].as_str(), Some("done"));
let read_response = app()
.oneshot(
Request::builder()
.uri(format!(
"/api/local-folder/ocr/read?rootUri={escaped_root}&ocrRootRelativePath=docs%2FPage.ocr%2Fphoto.png.ocr.md"
))
.header("x-mnote-actor-id", "user_test")
.header("x-mnote-actor-type", "user")
.body(Body::empty())
.expect("request"),
)
.await
.expect("read response");
assert_eq!(read_response.status(), StatusCode::OK);
let read_body = to_bytes(read_response.into_body(), usize::MAX)
.await
.expect("read body");
let read_payload: Value = serde_json::from_slice(&read_body).expect("read json");
assert!(read_payload["markdown"]
.as_str()
.is_some_and(|markdown| markdown.contains("Route OCR Token")));
let _ = fs::remove_dir_all(root);
}
#[tokio::test]
async fn local_ocr_jobs_route_rejects_missing_mineru_token() {
let old_mnote_token = std::env::var("MNOTE_MINERU_API_TOKEN").ok();
let old_mineru_token = std::env::var("MINERU_API_TOKEN").ok();
std::env::remove_var("MNOTE_MINERU_API_TOKEN");
std::env::remove_var("MINERU_API_TOKEN");
let root = temp_root("mnote-local-ocr-token");
write_workspace_manifest(&root);
let root_uri = format!("file://{}", root.display());
let (status, payload) = post_ocr_job(
&root,
json!({
"rootUri": root_uri,
"documentId": "local-md:docs~2FPage.md",
"sourceRootRelativePath": "docs/Page.assets/photo.png",
"provider": "mineru"
}),
)
.await;
if let Some(value) = old_mnote_token {
std::env::set_var("MNOTE_MINERU_API_TOKEN", value);
}
if let Some(value) = old_mineru_token {
std::env::set_var("MINERU_API_TOKEN", value);
}
assert_eq!(status, StatusCode::UNAUTHORIZED);
assert_eq!(payload["code"].as_str(), Some("mineru_token_missing"));
}
#[tokio::test]
async fn local_ocr_jobs_route_runs_mineru_runtime_against_http_mock() {
let listener = tokio::net::TcpListener::bind("127.0.0.1:0")
.await
.expect("mock mineru bind");
let base_url = format!("http://{}", listener.local_addr().expect("mock addr"));
let upload_count = Arc::new(AtomicUsize::new(0));
let poll_count = Arc::new(AtomicUsize::new(0));
let zip_bytes = Arc::new(build_test_mineru_zip("# MinerU Result\n\n识别文本"));
let mock_mineru = axum::Router::new()
.route(
"/api/v4/file-urls/batch",
axum::routing::post({
let base_url = base_url.clone();
|| async move {
Json(json!({
"batch_id": "batch_1",
"file_urls": [{ "upload_url": format!("{base_url}/upload/source") }]
}))
}
}),
)
.route(
"/upload/source",
axum::routing::put({
let upload_count = upload_count.clone();
move |body: axum::body::Bytes| {
let upload_count = upload_count.clone();
async move {
assert!(!body.is_empty());
upload_count.fetch_add(1, Ordering::SeqCst);
StatusCode::OK
}
}
}),
)
.route(
"/api/v4/extract-results/batch/batch_1",
axum::routing::get({
let base_url = base_url.clone();
let poll_count = poll_count.clone();
move || {
let poll_count = poll_count.clone();
async move {
poll_count.fetch_add(1, Ordering::SeqCst);
Json(json!({
"status": "done",
"full_zip_url": format!("{base_url}/result.zip")
}))
}
}
}),
)
.route(
"/result.zip",
axum::routing::get({
let zip_bytes = zip_bytes.clone();
move || {
let zip_bytes = zip_bytes.clone();
async move {
(
[(axum::http::header::CONTENT_TYPE, "application/zip")],
(*zip_bytes).clone(),
)
}
}
}),
);
let mock_handle = tokio::spawn(async move {
2026-06-01 10:07:42 +08:00
axum::serve(listener, mock_mineru)
.await
.expect("mock mineru server");
2026-06-01 09:29:12 +08:00
});
let old_token = std::env::var("MNOTE_MINERU_API_TOKEN").ok();
let old_base = std::env::var("MNOTE_MINERU_API_BASE_URL").ok();
let old_interval = std::env::var("MNOTE_MINERU_POLL_INTERVAL_MS").ok();
let old_max_polls = std::env::var("MNOTE_MINERU_MAX_POLLS").ok();
std::env::set_var("MNOTE_MINERU_API_TOKEN", "test-mineru-token");
std::env::set_var("MNOTE_MINERU_API_BASE_URL", &base_url);
std::env::set_var("MNOTE_MINERU_POLL_INTERVAL_MS", "1");
std::env::set_var("MNOTE_MINERU_MAX_POLLS", "2");
let root = temp_root("mnote-local-ocr-mineru-runtime");
write_workspace_manifest(&root);
fs::write(root.join("docs").join("Page.md"), "# Page\n").expect("page");
fs::write(
root.join("docs").join("Page.assets").join("photo.png"),
b"png",
)
.expect("photo");
let root_uri = format!("file://{}", root.display());
let response = app()
.oneshot(
Request::builder()
.method("POST")
.uri("/api/local-folder/ocr/jobs")
.header("content-type", "application/json")
.header("x-mnote-actor-id", "user_test")
.header("x-mnote-actor-type", "user")
.body(Body::from(
json!({
"rootUri": root_uri,
"documentId": "local-md:docs~2FPage.md",
"sourceRootRelativePath": "docs/Page.assets/photo.png",
"provider": "mineru"
})
.to_string(),
))
.expect("request"),
)
.await
.expect("response");
if let Some(value) = old_token {
std::env::set_var("MNOTE_MINERU_API_TOKEN", value);
} else {
std::env::remove_var("MNOTE_MINERU_API_TOKEN");
}
if let Some(value) = old_base {
std::env::set_var("MNOTE_MINERU_API_BASE_URL", value);
} else {
std::env::remove_var("MNOTE_MINERU_API_BASE_URL");
}
if let Some(value) = old_interval {
std::env::set_var("MNOTE_MINERU_POLL_INTERVAL_MS", value);
} else {
std::env::remove_var("MNOTE_MINERU_POLL_INTERVAL_MS");
}
if let Some(value) = old_max_polls {
std::env::set_var("MNOTE_MINERU_MAX_POLLS", value);
} else {
std::env::remove_var("MNOTE_MINERU_MAX_POLLS");
}
assert_eq!(response.status(), StatusCode::OK);
let body = to_bytes(response.into_body(), usize::MAX)
.await
.expect("response body");
let payload: Value = serde_json::from_slice(&body).expect("response json");
assert_eq!(payload["job"]["status"].as_str(), Some("done"));
assert_eq!(payload["job"]["provider"].as_str(), Some("mineru"));
assert_eq!(upload_count.load(Ordering::SeqCst), 1);
assert_eq!(poll_count.load(Ordering::SeqCst), 1);
2026-06-01 10:07:42 +08:00
let sidecar =
fs::read_to_string(root.join("docs").join("Page.ocr").join("photo.png.ocr.md"))
.expect("sidecar");
2026-06-01 09:29:12 +08:00
assert!(sidecar.contains("provider: mineru"));
assert!(sidecar.contains("识别文本"));
mock_handle.abort();
let _ = fs::remove_dir_all(root);
}
#[test]
fn local_ocr_mineru_runtime_config_uses_safe_defaults_and_env_overrides() {
let old_base = std::env::var("MNOTE_MINERU_API_BASE_URL").ok();
let old_legacy_base = std::env::var("MINERU_API_BASE_URL").ok();
let old_interval = std::env::var("MNOTE_MINERU_POLL_INTERVAL_MS").ok();
let old_max_polls = std::env::var("MNOTE_MINERU_MAX_POLLS").ok();
std::env::remove_var("MNOTE_MINERU_API_BASE_URL");
std::env::remove_var("MINERU_API_BASE_URL");
std::env::remove_var("MNOTE_MINERU_POLL_INTERVAL_MS");
std::env::remove_var("MNOTE_MINERU_MAX_POLLS");
assert_eq!(mineru_api_base_url(), DEFAULT_MINERU_API_BASE_URL);
assert_eq!(mineru_poll_interval(), Duration::from_secs(2));
assert_eq!(mineru_max_polls(), 90);
std::env::set_var("MNOTE_MINERU_API_BASE_URL", "https://mineru.example.test/");
std::env::set_var("MNOTE_MINERU_POLL_INTERVAL_MS", "25");
std::env::set_var("MNOTE_MINERU_MAX_POLLS", "3");
assert_eq!(mineru_api_base_url(), "https://mineru.example.test");
assert_eq!(mineru_poll_interval(), Duration::from_millis(25));
assert_eq!(mineru_max_polls(), 3);
if let Some(value) = old_base {
std::env::set_var("MNOTE_MINERU_API_BASE_URL", value);
} else {
std::env::remove_var("MNOTE_MINERU_API_BASE_URL");
}
if let Some(value) = old_legacy_base {
std::env::set_var("MINERU_API_BASE_URL", value);
} else {
std::env::remove_var("MINERU_API_BASE_URL");
}
if let Some(value) = old_interval {
std::env::set_var("MNOTE_MINERU_POLL_INTERVAL_MS", value);
} else {
std::env::remove_var("MNOTE_MINERU_POLL_INTERVAL_MS");
}
if let Some(value) = old_max_polls {
std::env::set_var("MNOTE_MINERU_MAX_POLLS", value);
} else {
std::env::remove_var("MNOTE_MINERU_MAX_POLLS");
}
}
#[tokio::test]
async fn local_ocr_jobs_route_rejects_root_escape_source() {
let root = temp_root("mnote-local-ocr-escape");
write_workspace_manifest(&root);
fs::write(root.join("docs").join("Page.md"), "# Page\n").expect("page");
let root_uri = format!("file://{}", root.display());
let (status, payload) = post_ocr_job(
&root,
json!({
"rootUri": root_uri,
"documentId": "local-md:docs~2FPage.md",
"sourceRootRelativePath": "../outside.png",
"provider": "mock",
"mockMarkdown": "should not write"
}),
)
.await;
assert_eq!(status, StatusCode::BAD_REQUEST);
assert_eq!(payload["code"].as_str(), Some("local_ocr_path_escape"));
}
#[tokio::test]
async fn local_ocr_jobs_route_rejects_non_image_pdf_source() {
let root = temp_root("mnote-local-ocr-unsupported");
write_workspace_manifest(&root);
fs::write(root.join("docs").join("Page.md"), "# Page\n").expect("page");
fs::write(
root.join("docs").join("Page.assets").join("notes.txt"),
b"text",
)
.expect("text");
let root_uri = format!("file://{}", root.display());
let (status, payload) = post_ocr_job(
&root,
json!({
"rootUri": root_uri,
"documentId": "local-md:docs~2FPage.md",
"sourceRootRelativePath": "docs/Page.assets/notes.txt",
"provider": "mock",
"mockMarkdown": "should not write"
}),
)
.await;
assert_eq!(status, StatusCode::BAD_REQUEST);
assert_eq!(
payload["code"].as_str(),
Some("local_ocr_source_type_unsupported")
);
}
#[tokio::test]
async fn local_ocr_jobs_route_writes_failed_mock_entry_with_redacted_error() {
let root = temp_root("mnote-local-ocr-failed");
write_workspace_manifest(&root);
fs::write(root.join("docs").join("Page.md"), "# Page\n").expect("page");
fs::write(
root.join("docs").join("Page.assets").join("photo.png"),
b"png",
)
.expect("photo");
let root_uri = format!("file://{}", root.display());
let response = app()
.oneshot(
Request::builder()
.method("POST")
.uri("/api/local-folder/ocr/jobs")
.header("content-type", "application/json")
.header("x-mnote-actor-id", "user_test")
.header("x-mnote-actor-type", "user")
.body(Body::from(
json!({
"rootUri": root_uri,
"documentId": "local-md:docs~2FPage.md",
"sourceRootRelativePath": "docs/Page.assets/photo.png",
"provider": "mock",
"mockError": "token=secret https://signed.example.test/full.zip?X-Oss-Signature=abc"
})
.to_string(),
))
.expect("request"),
)
.await
.expect("response");
assert_eq!(response.status(), StatusCode::OK);
let body = to_bytes(response.into_body(), usize::MAX)
.await
.expect("response body");
let payload: Value = serde_json::from_slice(&body).expect("response json");
assert_eq!(payload["job"]["status"].as_str(), Some("failed"));
assert_eq!(
payload["job"]["error"].as_str(),
Some("provider_error_redacted")
);
let stored = read_ocr_index_entry(&root, "docs/Page.assets/photo.png")
.expect("read index")
.expect("entry");
assert_eq!(stored.status, "failed");
assert_eq!(stored.error.as_deref(), Some("provider_error_redacted"));
assert!(!root
.join("docs")
.join("Page.ocr")
.join("photo.png.ocr.md")
.exists());
let _ = fs::remove_dir_all(root);
}
#[tokio::test]
async fn local_ocr_insert_route_appends_explicit_ocr_link() {
let root = temp_root("mnote-local-ocr-insert");
write_workspace_manifest(&root);
fs::write(root.join("docs").join("Page.md"), "# Page\n\n正文\n").expect("page");
fs::write(
root.join("docs").join("Page.assets").join("photo.png"),
b"png",
)
.expect("photo");
let root_uri = format!("file://{}", root.display());
let create_response = app()
.oneshot(
Request::builder()
.method("POST")
.uri("/api/local-folder/ocr/jobs")
.header("content-type", "application/json")
.header("x-mnote-actor-id", "user_test")
.header("x-mnote-actor-type", "user")
.body(Body::from(
json!({
"rootUri": root_uri,
"documentId": "local-md:docs~2FPage.md",
"sourceRootRelativePath": "docs/Page.assets/photo.png",
"provider": "mock",
"mockMarkdown": "识别文本"
})
.to_string(),
))
.expect("request"),
)
.await
.expect("create response");
assert_eq!(create_response.status(), StatusCode::OK);
let insert_response = app()
.oneshot(
Request::builder()
.method("POST")
.uri("/api/local-folder/ocr/insert")
.header("content-type", "application/json")
.header("x-mnote-actor-id", "user_test")
.header("x-mnote-actor-type", "user")
.body(Body::from(
json!({
"rootUri": root_uri,
"documentId": "local-md:docs~2FPage.md",
"ocrRootRelativePath": "docs/Page.ocr/photo.png.ocr.md",
"mode": "link"
})
.to_string(),
))
.expect("request"),
)
.await
.expect("insert response");
assert_eq!(insert_response.status(), StatusCode::OK);
let body = to_bytes(insert_response.into_body(), usize::MAX)
.await
.expect("insert body");
let payload: Value = serde_json::from_slice(&body).expect("insert json");
assert_eq!(
payload["insertedMarkdown"].as_str(),
Some("[OCRphoto.png](./Page.ocr/photo.png.ocr.md)")
);
let owner = fs::read_to_string(root.join("docs").join("Page.md")).expect("owner");
assert_eq!(
owner,
"# Page\n\n正文\n\n[OCRphoto.png](./Page.ocr/photo.png.ocr.md)\n"
);
let _ = fs::remove_dir_all(root);
}
}