From 6a286d422d8d513d86a5219fe8db85bfd9b92591 Mon Sep 17 00:00:00 2001
From: Tianyang <tianyanggong17@gmail.com>
Date: Sat, 1 Aug 2026 02:28:58 +0800
Subject: [PATCH] feat: add DocumentService coordinating MinIO writes and
 search indexing

---
 backend/app/document_service.py        |  87 +++++++++++++++
 backend/tests/test_document_service.py | 146 +++++++++++++++++++++++++
 2 files changed, 233 insertions(+)
 create mode 100644 backend/app/document_service.py
 create mode 100644 backend/tests/test_document_service.py

diff --git a/backend/app/document_service.py b/backend/app/document_service.py
new file mode 100644
index 0000000..5c730b7
--- /dev/null
+++ b/backend/app/document_service.py
@@ -0,0 +1,87 @@
+import logging
+from datetime import datetime, timezone
+
+from app.text_extract import html_to_text, markdown_to_text
+
+logger = logging.getLogger(__name__)
+
+
+class DocumentNotFoundError(Exception):
+    pass
+
+
+def _guess_content_type(key: str) -> str:
+    if key.endswith(".md"):
+        return "text/markdown"
+    if key.endswith(".html") or key.endswith(".htm"):
+        return "text/html"
+    return "application/octet-stream"
+
+
+class DocumentService:
+    def __init__(self, minio_client, search_client):
+        self._minio_client = minio_client
+        self._search_client = search_client
+
+    def list_documents(self, path: str) -> dict:
+        prefix = path if not path or path.endswith("/") else f"{path}/"
+        entries = self._minio_client.list_objects(prefix)
+        folders = []
+        files = []
+        for entry in entries:
+            key = entry["key"]
+            if entry.get("is_dir"):
+                name = key[len(prefix):].rstrip("/")
+                if name:
+                    folders.append(name)
+            else:
+                name = key[len(prefix):]
+                files.append({
+                    "key": key,
+                    "name": name,
+                    "size": entry["size"],
+                    "modifiedAt": entry["modifiedAt"],
+                })
+        return {"folders": folders, "files": files}
+
+    def get_document(self, key: str) -> dict:
+        try:
+            data = self._minio_client.get_object(key)
+        except FileNotFoundError as exc:
+            raise DocumentNotFoundError(key) from exc
+        return {
+            "key": key,
+            "content": data.decode("utf-8"),
+            "contentType": _guess_content_type(key),
+        }
+
+    def _extract_text(self, content: bytes, content_type: str) -> str:
+        text = content.decode("utf-8", errors="ignore")
+        if content_type == "text/html":
+            return html_to_text(text)
+        if content_type == "text/markdown":
+            return markdown_to_text(text)
+        return text
+
+    def save_document(self, key: str, content: bytes, content_type: str) -> dict:
+        self._minio_client.put_object(key, content, content_type)
+        try:
+            text = self._extract_text(content, content_type)
+            title = key.rsplit("/", 1)[-1]
+            path = key.rsplit("/", 1)[0] if "/" in key else ""
+            self._search_client.index_document(key, path, title, text)
+        except Exception:
+            logger.exception("Failed to update search index for %s", key)
+        return {"key": key, "updatedAt": datetime.now(timezone.utc).isoformat()}
+
+    def delete_document(self, key: str) -> None:
+        self._minio_client.delete_object(key)
+        try:
+            self._search_client.delete_document(key)
+        except Exception:
+            logger.exception("Failed to delete search index entry for %s", key)
+
+    def upload_document(self, key: str, data: bytes, content_type: str) -> dict:
+        self.save_document(key, data, content_type)
+        name = key.rsplit("/", 1)[-1]
+        return {"key": key, "name": name, "size": len(data)}
diff --git a/backend/tests/test_document_service.py b/backend/tests/test_document_service.py
new file mode 100644
index 0000000..b62007d
--- /dev/null
+++ b/backend/tests/test_document_service.py
@@ -0,0 +1,146 @@
+import pytest
+
+from app.document_service import DocumentService, DocumentNotFoundError
+
+
+class FakeMinioClient:
+    def __init__(self):
+        self.entries = []
+        self.stored = {}
+        self.deleted_keys = []
+        self.put_calls = []
+
+    def list_objects(self, prefix):
+        return [e for e in self.entries if e["key"].startswith(prefix)]
+
+    def get_object(self, key):
+        if key not in self.stored:
+            raise FileNotFoundError(key)
+        return self.stored[key]
+
+    def put_object(self, key, data, content_type):
+        self.stored[key] = data
+        self.put_calls.append((key, content_type))
+
+    def delete_object(self, key):
+        if key not in self.stored:
+            raise FileNotFoundError(key)
+        del self.stored[key]
+        self.deleted_keys.append(key)
+
+
+class FakeSearchClient:
+    def __init__(self, fail_on_index=False, fail_on_delete=False):
+        self.indexed = []
+        self.deleted = []
+        self.fail_on_index = fail_on_index
+        self.fail_on_delete = fail_on_delete
+
+    def index_document(self, key, path, title, content):
+        if self.fail_on_index:
+            raise RuntimeError("meilisearch unavailable")
+        self.indexed.append({"key": key, "path": path, "title": title, "content": content})
+
+    def delete_document(self, key):
+        if self.fail_on_delete:
+            raise RuntimeError("meilisearch unavailable")
+        self.deleted.append(key)
+
+
+def test_list_documents_at_root_returns_folders_and_files():
+    minio_client = FakeMinioClient()
+    minio_client.entries = [
+        {"key": "产品文档/", "is_dir": True, "size": 0, "modifiedAt": None},
+        {"key": "团队规范/", "is_dir": True, "size": 0, "modifiedAt": None},
+        {"key": "README.md", "is_dir": False, "size": 128, "modifiedAt": "2026-08-01T09:00:00Z"},
+    ]
+    service = DocumentService(minio_client, FakeSearchClient())
+
+    result = service.list_documents("")
+
+    assert result["folders"] == ["产品文档", "团队规范"]
+    assert result["files"] == [
+        {"key": "README.md", "name": "README.md", "size": 128, "modifiedAt": "2026-08-01T09:00:00Z"},
+    ]
+
+
+def test_get_document_returns_content_and_content_type():
+    minio_client = FakeMinioClient()
+    minio_client.stored["产品文档/架构设计.md"] = "# 架构设计\n\nDocHub 使用 MinIO 存储文档。".encode("utf-8")
+    service = DocumentService(minio_client, FakeSearchClient())
+
+    result = service.get_document("产品文档/架构设计.md")
+
+    assert result["key"] == "产品文档/架构设计.md"
+    assert result["contentType"] == "text/markdown"
+    assert "架构设计" in result["content"]
+
+
+def test_get_document_raises_not_found_for_missing_key():
+    service = DocumentService(FakeMinioClient(), FakeSearchClient())
+
+    with pytest.raises(DocumentNotFoundError):
+        service.get_document("不存在/文档.md")
+
+
+def test_save_document_writes_to_minio_then_indexes_content():
+    minio_client = FakeMinioClient()
+    search_client = FakeSearchClient()
+    service = DocumentService(minio_client, search_client)
+    content = "# 架构设计\n\nDocHub 使用 **MinIO** 存储文档，使用 Meilisearch 做全文检索。".encode("utf-8")
+
+    result = service.save_document("产品文档/架构设计.md", content, "text/markdown")
+
+    assert minio_client.stored["产品文档/架构设计.md"] == content
+    assert search_client.indexed[0]["key"] == "产品文档/架构设计.md"
+    assert search_client.indexed[0]["path"] == "产品文档"
+    assert search_client.indexed[0]["title"] == "架构设计.md"
+    assert "MinIO" in search_client.indexed[0]["content"]
+    assert "#" not in search_client.indexed[0]["content"]
+    assert "key" in result and "updatedAt" in result
+
+
+def test_save_document_does_not_fail_when_index_update_fails():
+    minio_client = FakeMinioClient()
+    search_client = FakeSearchClient(fail_on_index=True)
+    service = DocumentService(minio_client, search_client)
+    content = b"# doc"
+
+    result = service.save_document("产品文档/架构设计.md", content, "text/markdown")
+
+    assert minio_client.stored["产品文档/架构设计.md"] == content
+    assert result["key"] == "产品文档/架构设计.md"
+
+
+def test_delete_document_removes_from_minio_and_search_index():
+    minio_client = FakeMinioClient()
+    minio_client.stored["产品文档/旧文档.md"] = b"old content"
+    search_client = FakeSearchClient()
+    service = DocumentService(minio_client, search_client)
+
+    service.delete_document("产品文档/旧文档.md")
+
+    assert "产品文档/旧文档.md" not in minio_client.stored
+    assert search_client.deleted == ["产品文档/旧文档.md"]
+
+
+def test_delete_document_does_not_fail_when_index_delete_fails():
+    minio_client = FakeMinioClient()
+    minio_client.stored["产品文档/旧文档.md"] = b"old content"
+    search_client = FakeSearchClient(fail_on_delete=True)
+    service = DocumentService(minio_client, search_client)
+
+    service.delete_document("产品文档/旧文档.md")
+
+    assert "产品文档/旧文档.md" not in minio_client.stored
+
+
+def test_upload_document_writes_file_and_returns_metadata():
+    minio_client = FakeMinioClient()
+    service = DocumentService(minio_client, FakeSearchClient())
+    data = b"# uploaded doc"
+
+    result = service.upload_document("产品文档/新上传.md", data, "text/markdown")
+
+    assert result == {"key": "产品文档/新上传.md", "name": "新上传.md", "size": len(data)}
+    assert minio_client.stored["产品文档/新上传.md"] == data