seafile_ocr

seahub/ai/apis.py

@@ -17,7 +17,7 @@
 from seahub.utils.repo import is_valid_repo_id_format, is_repo_admin
 from seahub.ai.utils import search, get_file_download_token, get_search_repos, \
     RELATED_REPOS_PREFIX, RELATED_REPOS_CACHE_TIMEOUT, SEARCH_REPOS_LIMIT, \
-    format_repos
+    format_repos, ocr
 from seahub.utils import is_org_context, normalize_cache_key, HAS_FILE_SEASEARCH
 from seahub.views import check_folder_permission
 
@@ -121,3 +121,48 @@ def post(self, request):
                     f['fullpath'] = f['fullpath'].split(origin_path)[-1]
 
         return Response(resp_json, resp.status_code)
+
+
+class OCR(APIView):
+    authentication_classes = (TokenAuthentication, SessionAuthentication)
+    permission_classes = (IsAuthenticated, )
+    throttle_classes = (UserRateThrottle, )
+
+    def post(self, request):
+        repo_id = request.data.get('repo_id')
+        path = request.data.get('path')
+
+        if not repo_id:
+            return api_error(status.HTTP_400_BAD_REQUEST, 'repo_id invalid')
+        if not path:
+            return api_error(status.HTTP_400_BAD_REQUEST, 'path invalid')
+
+        repo = seafile_api.get_repo(repo_id)
+        if not repo:
+            error_msg = 'Library %s not found.' % repo_id
+            return api_error(status.HTTP_404_NOT_FOUND, error_msg)
+
+        file_obj = seafile_api.get_dirent_by_path(repo_id, path)
+        if not file_obj:
+            error_msg = 'File %s not found.' % path
+            return api_error(status.HTTP_404_NOT_FOUND, error_msg)
+
+        # permission check
+        if not check_folder_permission(request, repo_id, path):
+            error_msg = 'Permission denied.'
+            return api_error(status.HTTP_403_FORBIDDEN, error_msg)
+
+        params = {
+            'repo_id': repo_id,
+            'path': path,
+            'obj_id': file_obj.obj_id,
+        }
+
+        try:
+            ocr_result = ocr(params)
+        except Exception as e:
+            logger.error(e)
+            error_msg = 'Internal Server Error'
+            return api_error(status.HTTP_500_INTERNAL_SERVER_ERROR, error_msg)
+
+        return Response({'ocr_result': ocr_result}, status.HTTP_200_OK)

seahub/ai/utils.py

@@ -4,7 +4,7 @@
 import time
 from urllib.parse import urljoin
 
-from seahub.settings import SECRET_KEY, SEAFEVENTS_SERVER_URL
+from seahub.settings import SECRET_KEY, SEAFEVENTS_SERVER_URL, SEAFILE_AI_SECRET_KEY, SEAFILE_AI_SERVER_URL
 from seahub.utils import get_user_repos
 
 from seaserv import seafile_api
@@ -18,6 +18,12 @@
 RELATED_REPOS_CACHE_TIMEOUT = 2 * 60 * 60
 
 
+def gen_headers():
+    payload = {'exp': int(time.time()) + 300, }
+    token = jwt.encode(payload, SEAFILE_AI_SECRET_KEY, algorithm='HS256')
+    return {"Authorization": "Token %s" % token}
+
+
 def search(params):
     payload = {'exp': int(time.time()) + 300, }
     token = jwt.encode(payload, SECRET_KEY, algorithm='HS256')
@@ -65,3 +71,10 @@ def format_repos(repos):
             continue
         repos_map[real_repo_id] = (real_repo_id, origin_path, repo_name)
     return searched_repos, repos_map
+
+
+def ocr(params):
+    headers = gen_headers()
+    url = urljoin(SEAFILE_AI_SERVER_URL, '/api/v1/ocr/')
+    resp = requests.post(url, json=params, headers=headers, timeout=30)
+    return resp

seahub/settings.py

@@ -777,6 +777,13 @@ def genpassword():
 ENABLE_GLOBAL_ADDRESSBOOK = True
 ENABLE_ADDRESSBOOK_OPT_IN = False
 
+#####################
+# Seafile AI #
+#####################
+SEAFILE_AI_SERVER_URL = ''
+SEAFILE_AI_SECRET_KEY = ''
+ENABLE_SEAFILE_AI = False
+
 ####################
 # Guest Invite     #
 ####################

seahub/urls.py

@@ -203,7 +203,7 @@
 from seahub.seadoc.views import sdoc_revision, sdoc_revisions, sdoc_to_docx
 from seahub.ocm.settings import OCM_ENDPOINT
 
-from seahub.ai.apis import Search
+from seahub.ai.apis import Search, OCR
 from seahub.wiki2.views import wiki_view
 from seahub.api2.endpoints.wiki2 import Wikis2View, Wiki2View, Wiki2ConfigView, Wiki2PagesView, Wiki2PageView, \
     Wiki2DuplicatePageView, WikiPageTrashView
@@ -1044,3 +1044,8 @@
         re_path(r'^api/v2.1/repos/(?P<repo_id>[-0-9a-f]{36})/metadata/move-views/$', MetadataViewsMoveView.as_view(), name='api-v2.1-metadata-views-move'),
 
     ]
+
+if settings.ENABLE_SEAFILE_AI:
+    urlpatterns += [
+        re_path(r'^api/v2.1/ai/ocr/$', OCR.as_view(), name='api-v2.1-ai-ocr'),
+    ]