feat(search): add wiki search

app/app.py

@@ -11,6 +11,7 @@
 from seafevents.app.config import ENABLE_METADATA_MANAGEMENT
 from seafevents.seasearch.index_task.filename_index_updater import RepoFilenameIndexUpdater
 from seafevents.repo_metadata.face_recognition_updater import RepoFaceClusterUpdater
+from seafevents.seasearch.index_task.wiki_index_updater import WikiIndexUpdater
 
 
 class App(object):
@@ -44,6 +45,7 @@ def __init__(self, config, ccnet_config, seafile_config,
                 self._slow_task_handler = SlowTaskHandler(config)
                 self._repo_face_cluster_updater = RepoFaceClusterUpdater(config)
             self._repo_filename_index_updater = RepoFilenameIndexUpdater(config)
+            self._wiki_index_updater = WikiIndexUpdater(config)
 
     def serve_forever(self):
         if self._fg_tasks_enabled:
@@ -69,3 +71,4 @@ def serve_forever(self):
                 self._slow_task_handler.start()
                 self._repo_face_cluster_updater.start()
             self._repo_filename_index_updater.start()
+            self._wiki_index_updater.start()

repo_data/__init__.py

@@ -3,6 +3,7 @@
 from sqlalchemy.sql import text
 
 from seafevents.repo_data.db import init_db_session_class
+from seafevents.seasearch.utils.constants import REPO_TYPE_WIKI
 
 logger = logging.getLogger(__name__)
 
@@ -53,6 +54,25 @@ def _get_mtime_by_repo_ids(self, repo_ids):
         finally:
             session.close()
 
+    def _get_wiki_repo_id_commit_id(self, start, count):
+        session = self.db_session()
+        try:
+            cmd = """SELECT RepoInfo.repo_id, Branch.commit_id, RepoInfo.type
+                     FROM RepoInfo
+                     INNER JOIN Branch ON RepoInfo.repo_id = Branch.repo_id
+                     WHERE Branch.name = :name
+                     AND RepoInfo.type = :repo_type
+                     limit :start, :count;"""
+            res = session.execute(text(cmd), {'name': 'master',
+                                              'repo_type': REPO_TYPE_WIKI,
+                                              'start': start,
+                                              'count': count}).fetchall()
+            return res
+        except Exception as e:
+            raise e
+        finally:
+            session.close()
+
     def _get_all_trash_repo_list(self):
         session = self.db_session()
         try:
@@ -106,7 +126,8 @@ def _get_virtual_repo_in_repos(self, repo_ids):
         if not repo_ids:
             return []
         try:
-            cmd = """SELECT repo_id from VirtualRepo WHERE repo_id IN {}""".format(tuple(repo_ids))
+            formatted_ids = ", ".join("'{}'".format(id) for id in repo_ids)
+            cmd = """SELECT repo_id from VirtualRepo WHERE repo_id IN ({})""".format(formatted_ids)
             res = session.execute(text(cmd)).fetchall()
             return res
         except Exception as e:
@@ -149,6 +170,13 @@ def get_repo_id_commit_id(self, start, count):
             logger.error(e)
             return self._get_repo_id_commit_id(start, count)
 
+    def get_wiki_repo_id_commit_id(self, start, count):
+        try:
+            return self._get_wiki_repo_id_commit_id(start, count)
+        except Exception as e:
+            logger.error(e)
+            return self._get_wiki_repo_id_commit_id(start, count)
+
     def get_repo_head_commit(self, repo_id):
         try:
             return self._get_repo_head_commit(repo_id)
@@ -163,5 +191,4 @@ def get_virtual_repo_in_repos(self, repo_ids):
             logger.error(e)
             return self._get_virtual_repo_in_repos(repo_ids)
 
-
 repo_data = RepoData()

seafevent_server/request_handler.py

@@ -207,3 +207,36 @@ def extract_file_details():
     details = add_file_details(repo_id, obj_ids, metadata_server_api)
 
     return {'details': details}, 200
+
+
+@app.route('/wiki-search', methods=['POST'])
+def search_wiki():
+    is_valid = check_auth_token(request)
+    if not is_valid:
+        return {'error_msg': 'Permission denied'}, 403
+
+    # Check seasearch is enable
+    if not index_task_manager.enabled:
+        return {'error_msg': 'Seasearch is not enabled by seafevents.conf'}
+    try:
+        data = json.loads(request.data)
+    except Exception as e:
+        logger.exception(e)
+        return {'error_msg': 'Bad request.'}, 400
+
+    query = data.get('query').strip()
+    wiki = data.get('wiki')
+
+    if not query:
+        return {'error_msg': 'query invalid.'}, 400
+    if not wiki:
+        return {'error_msg': 'wiki invalid.'}, 400
+
+    try:
+        count = int(data.get('count'))
+    except:
+        count = 20
+
+    results = index_task_manager.wiki_search(query, wiki, count)
+
+    return {'results': results}, 200

seasearch/index_store/index_manager.py

@@ -4,7 +4,8 @@
 
 from seafevents.seasearch.utils import need_index_metadata_info
 from seafevents.db import init_db_session_class
-from seafevents.seasearch.utils.constants import ZERO_OBJ_ID, REPO_FILENAME_INDEX_PREFIX
+from seafevents.seasearch.utils.constants import ZERO_OBJ_ID, REPO_FILENAME_INDEX_PREFIX, \
+    WIKI_INDEX_PREFIX
 from seafevents.repo_metadata.metadata_server_api import MetadataServerAPI
 from seafevents.repo_metadata.utils import METADATA_TABLE
 from seafevents.utils import timestamp_to_isoformat_timestr
@@ -74,3 +75,45 @@ def delete_repo_filename_index(self, repo_id, repo_filename_index, repo_status_f
 
     def keyword_search(self, query, repos, repo_filename_index, count, suffixes=None, search_path=None, obj_type=None):
         return repo_filename_index.search_files(repos, query, 0, count, suffixes, search_path, obj_type)
+
+    def delete_wiki_index(self, wiki_id, wiki_index, wiki_status_index):
+        # first delete wiki_index
+        wiki_index_name = WIKI_INDEX_PREFIX + wiki_id
+        wiki_index.delete_index_by_index_name(wiki_index_name)
+        wiki_status_index.delete_documents_by_repo(wiki_id)
+
+    def wiki_search(self, query, wiki, wiki_index, count):
+        return wiki_index.search_wiki(wiki, query, 0, count)
+
+    def update_wiki_index(self, wiki_id, commit_id, wiki_index, wiki_status_index):
+        try:
+            new_commit_id = commit_id
+            index_name = WIKI_INDEX_PREFIX + wiki_id
+
+            wiki_index.create_index_if_missing(index_name)
+
+            wiki_status = wiki_status_index.get_repo_status_by_id(wiki_id)
+            from_commit = wiki_status.from_commit
+            to_commit = wiki_status.to_commit
+
+            if new_commit_id == from_commit:
+                return
+
+            if not from_commit:
+                commit_id = ZERO_OBJ_ID
+            else:
+                commit_id = from_commit
+
+            if wiki_status.need_recovery():
+                logger.warning('%s: wiki index inrecovery', wiki_id)
+                wiki_index.update(index_name, wiki_id, commit_id, to_commit)
+                commit_id = to_commit
+                time.sleep(1)
+            wiki_status_index.begin_update_repo(wiki_id, commit_id, new_commit_id)
+            wiki_index.update(index_name, wiki_id, commit_id, new_commit_id)
+            wiki_status_index.finish_update_repo(wiki_id, new_commit_id)
+
+            logger.info('wiki: %s, update wiki index success', wiki_id)
+
+        except Exception as e:
+            logger.exception('wiki_id: %s, update wiki index error: %s.', wiki_id, e)

seasearch/index_store/repo_file_name_index.py

@@ -1,4 +1,3 @@
-import json
 import os
 import logging
 

seasearch/index_store/repo_status_index.py

@@ -56,15 +56,16 @@ def create_index_if_missing(self):
     def check_repo_status(self, repo_id):
         return self.seasearch_api.check_document_by_id(self.index_name, repo_id).get('is_exist')
 
-    def add_repo_status(self, repo_id, commit_id, updatingto, metadata_updated_time):
-        date = {
+    def add_repo_status(self, repo_id, commit_id, updatingto,  metadata_updated_time):
+        data = {
             'repo_id': repo_id,
             'commit_id': commit_id,
             'updatingto': updatingto,
             'metadata_updated_time': metadata_updated_time,
         }
+
         doc_id = repo_id
-        self.seasearch_api.create_document_by_id(self.index_name, doc_id, date)
+        self.seasearch_api.create_document_by_id(self.index_name, doc_id, data)
 
     def begin_update_repo(self, repo_id, old_commit_id, new_commit_id, metadata_updated_time):
         self.add_repo_status(repo_id, old_commit_id, new_commit_id, metadata_updated_time)
@@ -79,6 +80,7 @@ def get_repo_status_by_id(self, repo_id):
         doc = self.seasearch_api.get_document_by_id(self.index_name, repo_id)
         if doc.get('error'):
             return RepoStatus(repo_id, None, None, None)
+
         commit_id = doc['_source']['commit_id']
         updatingto = doc['_source']['updatingto']
         metadata_updated_time = doc['_source']['metadata_updated_time']
@@ -152,12 +154,13 @@ def _repo_head_search(self, query_params):
             commit_id = hit.get('_source').get('commit_id')
             updatingto = hit.get('_source').get('updatingto')
             metadata_updated_time = hit.get('_source').get('metadata_updated_time')
-            repo_heads.append({
+            repo_head = {
                 'repo_id': repo_id,
                 'commit_id': commit_id,
                 'updatingto': updatingto,
-                'metadata_updated_time': metadata_updated_time,
-            })
+                'metadata_updated_time': metadata_updated_time
+            }
+            repo_heads.append(repo_head)
         return repo_heads, total
 
     def delete_index_by_index_name(self):