This repository has been archived by the owner on Nov 9, 2017. It is now read-only.
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Add TM to the Angular editor. Adds a REST endpoint.
Squashed commit of the following: commit 81f4476 Author: David Mason <drdmason@gmail.com> Date: Thu Jun 11 10:32:06 2015 +1000 Remove fallback to stop-words comparison for similarity score calculation. This change is contentious, so I am removing it until an amicable resoultion to the discussion is reached. Similarity scores need more user testing to determine which algorithm will b emost useful to translators. commit 882e301 Author: David Mason <drdmason@gmail.com> Date: Thu Jun 11 10:19:01 2015 +1000 Revert "Change token regex for similarity comparison to pick up more cases." This reverts commit db71115. More test-cases are required to ensure this regex is correct. Rolling back for now so that this change does not go into the next release without thorough testing. The main problem with the new regex is that it would break on '.' in constructions such as URLs, which does not appear to be desired behaviour. commit 70cbb12 Author: David Mason <drdmason@gmail.com> Date: Tue Jun 9 13:59:05 2015 +1000 Update zanata-spa to commit 883b7b9071ef42c5e021c62905d311e985ec8596 commit db71115 Author: David Mason <drdmason@gmail.com> Date: Tue Jun 9 13:38:10 2015 +1000 Change token regex for similarity comparison to pick up more cases. The regex was not picking up punctuation that was not followed by whitespace. There are still some cases that it is not covering. I have kept this change small, and the other changes can be made when there is more time for testing. commit 276f310 Author: David Mason <drdmason@gmail.com> Date: Tue Jun 9 08:53:54 2015 +1000 Fall back on stop-words comparison for similarity when nothing else is available. A divide-by-zero condition was generating a NaN value for similarity for strings that contain only stop-words. When that condition would be met, this will instead repeat the comparison without ignoring stop-words, then finally just return 0.0 if there are still no tokens to compare. commit 35e0e53 Author: David Mason <drdmason@gmail.com> Date: Tue Jun 9 13:10:30 2015 +1000 Log more information when TM result row has an unexpected entity type. When lucene indexes are in a bad state, search can return a row that does not contain one of the expected entity types. This ensures that valid rows will still be returned in results, and an error with more information will be logged when a bad row is encountered. commit 0144585 Author: David Mason <drdmason@gmail.com> Date: Thu Jun 4 16:12:07 2015 +1000 Update zanata-spa to commit 6f20797323e6d45d05eff01cfefb0969f50e5479 commit 990e0ad Author: David Mason <drdmason@gmail.com> Date: Thu Jun 4 13:11:51 2015 +1000 Update zanata-spa to commit 194a5e237ff5edd47953b89e2e31255de67d0a1d commit c20e1b7 Author: David Mason <drdmason@gmail.com> Date: Thu May 21 20:58:11 2015 +1000 Update zanata-spa to commit 402c57d0ebff0ea957f31e4a4e508974b89d9099 commit 68b2b27 Author: David Mason <drdmason@gmail.com> Date: Thu May 21 15:05:15 2015 +1000 Add release note for suggestions endpoint. commit f643587 Author: David Mason <drdmason@gmail.com> Date: Wed May 20 08:36:50 2015 +1000 Refactor LocaleServiceImpl.getByLocaleId(String) for brevity. commit 0a4e2c5 Author: David Mason <drdmason@gmail.com> Date: Tue May 19 11:34:44 2015 +1000 Change source and translation locales for suggestions endpoint to query params. Since locales can contain non-url-friendly characters, they are better included as part of the query string. commit d98a2fa Author: David Mason <drdmason@gmail.com> Date: Tue May 19 14:59:05 2015 +1000 Remove unused media type string for suggestions. Changed to use the JSON media type for suggestions. commit 36bf6b7 Author: David Mason <drdmason@gmail.com> Date: Tue May 19 14:23:32 2015 +1000 Use Option instead of Either in SuggestionService commit 7d65aea Author: David Mason <drdmason@gmail.com> Date: Tue May 19 14:22:43 2015 +1000 Add @nullable annotation to locale lookup. commit 0306816 Author: David Mason <drdmason@gmail.com> Date: Tue May 19 11:01:23 2015 +1000 Prevent undeclared IllegalArgumentException when looking up locales. commit 68a25b8 Author: David Mason <drdmason@gmail.com> Date: Tue May 19 10:37:10 2015 +1000 Make search type lookup case-insensitive. The try block could be removed because iterating enum values will not give nulls, and equalsIgnoreCase will behave sensibly with an entered null value. commit 0b8e828 Author: David Mason <drdmason@gmail.com> Date: Tue May 19 10:24:11 2015 +1000 Keep format strings in the same scope as String.format() Format strings were being passed to a function, which seems confusing and so more error-prone. Changed to passing an already-formatted string so that the format strings can be inlined in the call to String.format() so that it is obvious that they are format strings and what the placeholders represent. commit 727c0a0 Author: David Mason <drdmason@gmail.com> Date: Sat May 9 22:09:39 2015 +1000 Add REST endpoint for suggestions for editor. commit 3b8addd Author: David Mason <drdmason@gmail.com> Date: Fri May 15 21:34:36 2015 +1000 Add method to get suggestions from TM service. I have avoided changing the existing code paths for TM search for now since it is a performance-critical component for translation copy. We should combine the code paths in a separate operation as long as it does not impact on performance. commit bee070e Author: David Mason <drdmason@gmail.com> Date: Fri May 15 22:34:51 2015 +1000 Add DTOs for suggestions and suggestion detail. These DTO classes are only configured to serialize to JSON, and I have not considered deserialization since this is query data and not expected to be uploaded to the server. commit a53c3cc Author: David Mason <drdmason@gmail.com> Date: Wed May 13 08:58:52 2015 +1000 Workaround TM result coalescing bug. Refactors the match processing code to make it clearer what type of match is being added to a result item, and makes a few small changes to minimize the incorrect behaviour: - No longer adds imported TM item id to the list that is used to look up text flows. A more complete fix is to wrap the Long id in a type so that all the ids are present and can be used sensibly. This just makes the list appropriate for the code that uses it right now, to delay spending the development time for the full fix. - Upgrades match type when a match has a higher match type, where ApprovedInternal > TranslatedInternal > Imported. Imported is treated as lowest just because the editor suppresses details display for Imported. Treating the others as higher will allow users to see text flow details for the other types. The full fix for this would be to inculde a summary of all the match types in the DTO, so the editor can display whatever makes sense.
- Loading branch information