Repository navigation
fix(i18n-tr): review fuzzy Turkish translations (part 2) - #45741
Conversation
Review the legacy fuzzy entries from "Edit %s in modal" to "Row height" and the "Set as default" theme tooltips; fix 15 adjacent confirmed ones.
|
We would appreciate it if you could provide us with more info about this issue/pr! Please do not leave the |
✅ Deploy Preview for superset-docs-preview ready!
To edit notification comments on pull requests, go to your Netlify project configuration. |
|
The description is filled in now (the first save went out empty, sorry about that). Part 1 is #45428. |
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## master #45741 +/- ##
=======================================
Coverage 84.26% 84.26%
=======================================
Files 3057 3057
Lines 201986 201993 +7
Branches 47254 47255 +1
=======================================
+ Hits 170204 170213 +9
+ Misses 28525 28523 -2
Partials 3257 3257
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
hainenber
left a comment
There was a problem hiding this comment.
Thanks again for your work and LGTM!
|
Hmm, there seems to be a uncaught regression on the babel-extraction after merging this to master. Can you help checking? |
|
Had a look. I don't think the .po file is the problem, it's the baseline. The failing runs checked out a tree from before this PR, but the job fetches master when it starts, and by then master was f93aa5b (this PR). So the 196 entries this PR de-fuzzed look like they were invalidated by the PR under test. The push run for ff0af9c and the runs for #45709 and #45640 all show Re-running won't help because the merge ref doesn't change, but a new push to those branches will. I opened #45794 to take the baseline from the parent of the checked-out commit instead of the branch tip. |
SUMMARY
Part 2 of the Turkish review that #41702 asked for, after #45428 (part 1). I used the same method and the same terminology as #45428: I looked up the call site of each
#, fuzzyentry, wrote the translation for that context and removed the flag.This PR covers the next 194 legacy fuzzy entries (the old fuzzy-matcher guesses, not the
Machine-translated via backfill_po.pydrafts), fromEdit %s in modalthroughRow height. That is every legacy fuzzy entry from where part 1 stopped to the end of R, so nothing in the range is skipped. I also did theSet as default light theme/Set as default dark themetooltips, which #45428 listed as a follow-up. They are in S, so part 3 starts atSQL Lab queriesand these two are already done.Superset serves fuzzy entries (see #45428), so Turkish users see these guesses today. The worst ones say something else:
msgidNot set(dataset of a dynamic group-by card)HaziranOR(Table V2 filter condition)saatRetryherExport All Data(Explore → Data Export Options)Tüm veriyi temizleExpand AllTümünü temizlePasswords do not match!Gösterim Paneli BulunmuyorPasteGüncelleLast login(Users list)son ayEdit User/List UsersSorguyu düzenleRefresh interval set for this sessionYenileme aralığı kaydedildiSet as default light theme/dark themeFiltrenin default değeri varCounts
Edit %s in modal…Row height, plus the 2 theme tooltips). All 196 lose their#, fuzzyflag.Alarm ekle/Rapor ekle(Edit alert,Edit report,Edit tag), add a missing colon (Password:), or replaceveri kümesi/veri setiwithveriseti(Modified 1 column in the dataset,Removed 1 column from the dataset).Email,Email link,Enable labels,Error fetching charts,Error: %s,False,Open in SQL lab.msgfmt -c. LikeAdded 1 new column to the datasetin part 1, theirmsgstr[0]uses the%sofmsgid_plural.msgstr.Last day/week/month/quarter/yearwere lowercaseson gün,son hafta, … . They are nowSon 1 gün,Son 1 hafta, … . Insuperset/utils/date_parser.pythese options are rolling ranges that end today (Last week : today), and the separateprevious calendar weekoptions are the calendar ones.Son haftaalone can also read as "the final week".Confirmed entries changed
Same rule as part 1. I changed a confirmed entry only where it is in the same control, list or menu as an entry fixed here and contradicts it: the other half of a pair left in English, a wrong meaning, or a different word for the same thing.
msgidEnter the user's usernameKullanıcının adı girinKullanıcının kullanıcı adını girinKullanıcının adını girin)EqualsEşitlikler(equalities)EşittirNot Equal,Greater Than,In Rangefixed hereLess Than/Less Than or Equal/Greater Than or EqualDaha az/Daha az veya eşit/ EnglishKüçüktür/Küçük veya Eşittir/Büyük veya EşittirBüyüktürhere andİçerir(Contains) in part 1Label ascendingEtiket yükseldi(the label went up)Etikete göre artanLabel descendingScrollKaydırmalıListPin LeftSola SabitlePin RightPixelsPikselMetersList RolesListe RollerRolleri ListeleList Users/List GroupsError reading Columnar fileCümleyi oku(read the sentence)Sütunlu dosya okunurken hata oluştuError reading CSV file/Error reading Excel fileExport successful: %s/Processing export for %s/Processing export...İhracat başarılı: %s/%s için ihracat/İşleme ihracat ...Dışa aktarma başarılı: %s/%s dışa aktarılıyor/Dışa aktarılıyor...Export failed/Export cancelledfixed hereExport to pivoted .CSV.CSVPivotlanmış CSV’ye AktarExport to Pivoted Excel. Same text as the dashboard'sExport to Pivoted .CSVherePixelsis also an option of the Point Cluster Map "Point Radius Unit" select, whereMilesandKilometersare still in English.Terminology decisions
Everything from #45428 still applies (dashboard →
gösterim paneli, dataset →veriseti, column →sütun, alert →alarm, type →tür, light theme →açık, extension →uzantı, formal siz imperative). New choices in this PR, with counts of confirmed entries on master whosemsgidcontains the English term:dışa aktar/dışa aktarmaDışa aktar2, short… Aktar4 (CSV’ye Aktar), of 23ihracat(12) is more common, but it means the export of goods (foreign trade). TheExportbutton itself saysDışa aktar.ikonsimge(2). The GeoJSON descriptions next to these labels already sayikon.göstergeefsane(12) means a myth or a tale, and 5 entries keepLegendin English. OnlyLegend Typeuses the term here. The otherefsanestrings need their own PR.Ad/Soyadİsim/Soyisim.İsimis already the catalog's word for a generic name (Name is required→İsim gereklidir).Eşittir,Büyüktür,İçermez,İle Biter,ArasındaContains→İçerirfrom part 1Out of scope (noticed, for follow-ups)
SQL Lab queriesonward (part 3). Some of them are next to strings fixed here:True→SAL(the abbreviation of Tuesday) next toFalse,Sort by data→Sıralain the "Sort legend" select,Unpin→çalışıyor("working") next toPin Left/Pin Right,Starts Withnext toEnds With, andSelect All→Veritabanını sil("delete database").Machine-translated via backfill_po.pydrafts in this alphabetical range. They are a different review workflow.Inside top(English),Inside bottom→alt altta,Inside top left→Üstte sol,Inside bottom left→altta solun solunda,Inside bottom right→alt sağda. They are in the same select asInside,Inside left,Inside rightandInside top rightfixed here.Topis also stillTopin the label position and legend orientation selects.Autosize all columns→Tüm sütunlar("all columns"),Autosize Column→Autosize Köşe.Icon size unit→Icon boyut ünitesi,Label size unit→Etiket boyutu ünitesi.Use current extent→Mevcut ölçüde kullanın.Native filter values has no values→Yerli filtre değerlerinin değerleri yoktur(yerlimeans domestic).ihracat, for exampleExport to PDFandExport to original .CSV, and the 12efsanestrings for "legend".Featured color palettes→Özel renk paletleri, the text fix(i18n-tr): review fuzzy Turkish translations (part 1) #45428 givesCustom color palettes.Error executing query.+ theOpen in SQL lablink +to check for details.. The confirmed Turkish of the last one,Detaylar için kontrol etmek., lost its leading space, so it runs into the link text. A single string with a placeholder would fix this in code and allow Turkish word order.BEFORE/AFTER SCREENSHOTS OR ANIMATED GIF
This PR only changes the catalog. The full before/after list is below. To see the strings, run with the
trlocale and open:Last login,Login count,Is active?, and the empty states (No users yet, …)Registration date,Referrer,No Logs yet, theSet as default light/dark themetooltips, theLocaltag,Import themesPassword:,Please enter your username,Passwords do not match!Last day/week/month/quarter/yearNot Equal,Greater Than,In Range,Ends With,OR), column menu (Pin Column,Expand All,Excel Export,Paste), pagination (Page,Page Size:)Export All Data,Export Current View,Export to Pivoted .CSV / Excel; dashboard → ⋯ → Download:Export YAML,Export as ExampleGeneral information,File settings,Header row,Index column,Preview uploaded fileMeters/Pixels), Pie (Rose Type), Radar (Inside …), Waterfall (Increase color/label), legendTypeandSort legendQuery State),Retry fetching results, table preview (Indexes (%s),Refresh table schema)TESTING INSTRUCTIONS
msgfmtonly checks the format of non-fuzzy entries. To confirm that no de-fuzzed entry hides a format defect, clear all fuzzy flags and recompile:This reports 1 fatal error both before and after this change: the plural
%(suggestion)s instead of …entry, which stays fuzzy on purpose (see #45428).Other checks (gettext 1.0, Babel 2.18.0):
scripts/translations/check_translation_regression.py --counton master's catalogs, then--compareon this branch:No translation regressions.andtr: translated 3852 -> 4048 (+196), fuzzy 989 -> 793 (-196). Every other language shows+0. I ran it on the committed catalogs, withoutbabel_update.sh.#, fuzzyflag lines andmsgstrlines. Reading the catalog with Babel'sread_poand writing it back withwrite_po(width 76, aspybabel updatewrites it) gives the same small unrelated difference on master and on this branch, so the edited entries are already in Babel's format.%s/%d/%(name)s/{name}and HTML tags betweenmsgid(ormsgid_plural) andmsgstrfor all 211 edited entries. They all match.ADDITIONAL INFORMATION
All 211 changed entries (before → after)
msgid