Commits

Author Commit Message Labels Comments Date
Thomas Waldmann
search: add name/content key terms as suggestions to search result view
Branches
pytest2
Thomas Waldmann
theme search field: do not display the submit button to save some space - hide it via css
Branches
pytest2
Thomas Waldmann
rename template search_results.html to search.html
Branches
pytest2
Thomas Waldmann
add a separate +search view, remove incomplete search paging support, cleanup .../+search -> get a search form to fill out .../+search?q=foo -> search for "foo" use "search" view url as action for search forms rename "search_in_all" to "history" (clearer, shorter for URL in GET requests) remove most search related code from "show" view (except feeding the empty search form to the template) fix css so the small theme searchform is right-aligned again
Branches
pytest2
Thomas Waldmann
search results template: cosmetic changes
Branches
pytest2
Thomas Waldmann
whooshify sitemap and urls_names
Branches
pytest2
Thomas Waldmann
whooshify itemsize
Branches
pytest2
Thomas Waldmann
whooshify similar_names
Branches
pytest2
Thomas Waldmann
backrefs: remove some unused code
Branches
pytest2
Thomas Waldmann
whooshify backrefs
Branches
pytest2
Thomas Waldmann
indexing: fix remove_item to only use uuid (not name), rearrange remove_rev code so it is more similar
Branches
pytest2
Thomas Waldmann
move indexing related code to MoinMoin.storage.backends.indexing remove Item.before_revision_commit hook - wrong place, such stuff must be done in indexing middleware or it does not happen when non-UI commits happen, e.g. when items are unserialized from xml. I noticed this because ITEMLINKS and ITEMTRANSCLUSIONS where missing for data coming from xml. note: this is likely still not the final place. stuff like current user / ip are not relevant / not available…
Branches
pytest2
Thomas Waldmann
whooshify _compute_item_sets (used for wanteds/orphans view) remove the tests for it (they can't work any more the way they were designed) added some simple highlevel test that makes sure the wanteds/orphans view code is executed and does not crash.
Branches
pytest2
Thomas Waldmann
refactor wanteds/orphans - do same stuff much simpler wanted and orphaned item names can be computed from same item sets: existing items, linked items, transcluded items, no-revisions items items without revisions are invalid for this computation, so get rid of them: valid = existing - norevs it doesn't matter for this computation whether some item is linked or transcluded, so we compute a "referred" set: referred = linked | transcluded now it is …
Branches
pytest2
Thomas Waldmann
refactored some search code, make it more flexible, reduce code duplication moved stuff dealing with whoosh / indexes from frontend.views to indexing mixin. added a .searcher method to indexing mixin that returns a searcher either for all or latest revs index. used that method also for .search and .search_page to reduce duplication. make the .query_parser method more intelligent, so it automatically chooses a MultifieldParser if there are multiple default_fields. fix ca…
Branches
pytest2
Thomas Waldmann
MoinMoin.storage.terms + tests removed (unused), please use whoosh queries
Branches
pytest2
Thomas Waldmann
user: do not use MoinMoin.storage.terms, which is deprecated
Branches
pytest2
Thomas Waldmann
get rid of storage level search_items method (uses deprecated code) user module still needed that functionality (we have no whoosh indexing for user items yet as they use item metadata not revision metadata/data). fixed some comments/docstrings that referred to it
Branches
pytest2
Thomas Waldmann
user: simplify some code, do less, only get_items_by_filter is using search_items
Branches
pytest2
Thomas Waldmann
remove unused Item.search_items/list_items code (which called deprecated backend code)
Branches
pytest2
Thomas Waldmann
whooshify include converter (use wildcard instead of regex term for now) whoosh has upcoming regex term support, but it is not released yet, so we use a wildcard term for now and get rid of the leading ^ that Include macro needs for getting into regex mode. that ^ was doubled to ^^ due to escaping done in the Include macro code. the data_unescape property of XPointer Entry looks like to be intended to reverse that, so we fixed one place in the code to use .data_unesca…
Branches
pytest2
Thomas Waldmann
add backend.documents(), simplify tags code (by using .search and .documents)
Branches
pytest2
Thomas Waldmann
remove count_items (unused, uses deprecated code)
Branches
pytest2
Thomas Waldmann
save memory by not including content in history list. Fixes #42
Branches
pytest2
Thomas Waldmann
better error message for conversion exceptions, including item name and revno, fixes #52
Branches
pytest2
Thomas Waldmann
whooshify get_index()
Branches
pytest2
Thomas Waldmann
whooshify get_templates, adapt MimeType analyzer so it yields also a full and normalized contenttype
Branches
pytest2
Thomas Waldmann
add query_parser building method to storage, use index-based search in some scripts moin maint_reduce_revisions and maint_set_meta now have a --query option that takes a whoosh query whoosh can not do regex yet, but you can use wildcards. also, you can search in other fields than in name_exact (which is the default). remove MoinMoin.storage.serialization.TermMatch - it is not used and we need to solve this differently. deprecate MoinMoin.storage.terms, use whoosh indexed searc…
Branches
pytest2
Thomas Waldmann
killed backend.history(), replace it by .search() and appropriate query/reverse/sortedby/limit params killed the .history() tests, added a few tests for .search(). Note: whoosh has own tests, so we maybe do not need that many tests there. global history view got a major speedup when the user uses a bookmark (because we then can add a DateRange term to massively reduce the search results in big wikis). atom feed limits to 100 results now. some minor refactorings.
Branches
pytest2
Thomas Waldmann
merged pytest2 repo
Branches
pytest2
  1. Prev
  2. Next