(1) The converter was tested in detail for the same eaf-files which were used for the problematic dictionary. The described problem is not reproduced any more.
(2) Implemented bidirectional_links mutation to fix one-way links in the problematic dictionary. Found 326 not correct links have been fixed. But the main problem is that linked lexical entries were manually deleted before and so, big amount of links became failed. About 76 thousand links were broken this way.
*** bidirectional_links mutation
can be called with curl request using Administrator token:
curl 'http://localhost:6543/graphql' \
-H 'Content-Type: application/json' \
-H 'Cookie: auth_tkt=_tkn_!userid_type:int' \
--data-raw '{ \
"operationName": "bidirectionalLinks", \
"variables":{"dictionaryIdList":[[4755,333300]]}, \
"query": \
"mutation bidirectionalLinks($dictionaryIdList: [LingvodocID]!) { \
bidirectional_links(dictionary_id_list: $dictionaryIdList, debug_flag: true) { \
triumph }}"}'
#! Remove any newline after --data-raw to call this from Ubuntu.
#! Change _tkn_ to actual token for administrator.
#! Be careful, used token may be not actual after database restart.
collects all links over each dictionary from dictionaryIdList
iterates over the found links and looking for back one. So, if link from entry_id (3) to linked_id (4) exists, the back link (4) to (3) should be in that list as well
missing back links are created using this mutation
just not deleted lexical entries and perspectives are processed
The dictionary http://lingvodoc.ispras.ru/dictionary/4755/333300/perspective/4755/333302/view contains too few links between lexical entries and paradigmatic forms. Seems like large part of links is not built in convertation from .eaf file
(1)
or some of them are not bidirectional(2)
.(1)
The converter was tested in detail for the same eaf-files which were used for the problematic dictionary. The described problem is not reproduced any more.(2)
Implemented bidirectional_links mutation to fix one-way links in the problematic dictionary. Found 326 not correct links have been fixed. But the main problem is that linked lexical entries were manually deleted before and so, big amount of links became failed. About 76 thousand links were broken this way.*** bidirectional_links mutation
dictionaryIdList
entry_id (3)
tolinked_id (4)
exists, the back link(4) to (3)
should be in that list as well