Importing Items from a File

Choose a file of Japanese words or kanji. eevy looks up each row in its dictionary, and you add what you confirm to a collection. It links entries already in eevy and creates none. Words eevy doesn't have can go to your glossary's Needs review list, for you to make yourself.

Importing a file

Choose your file. eevy works out the file type itself.
If the file holds several sheets, tables or decks, check the ones you want under Parts.
Check the preview and the column mapping.
Set your options, then press Find matches.
Review the matches and add them to a collection.

The file opens on your device. Only its text is sent when you find matches.
An unfinished import stays listed on the page. Resume or cancel it.

Files with several parts

Excel sheets, Word tables and Anki decks are parts. Check the ones you want.
Drag a part to reorder. A higher part comes first and keeps an item two parts share.
Every part uses the first part's setup until you change it. Set up or uncheck a part marked layout differs.
Send All parts into one collection, or Each part into its own collection.

File types

Open a file type to see how it is read.

CSV

Files: .csv, .tsv, .txt

One item per row. Name the first row with these headers: id, codepoint, writtenForm, kana, type, meaning, title, cardType, content, markdown, before. Keep only the columns you have. One of id, writtenForm or kana is enough.

id	eevy's own ID for the entry. Only if you have it.
codepoint	A character by its codepoint, like 04e00 for 一.
writtenForm	The Japanese, as you would write it.
kana	Its reading, in kana or romaji.
type	Word or Character (or 1 or 2). Leave empty to let eevy decide.
meaning	Your meaning. Helps pick the right entry and meaning.
title	A knowledge card's title. Card rows only.
cardType	Explanation, Pattern, Cultural Note, Tip, Nuance. Card rows only.
content	A card's text as eevy content. Card rows only.
markdown	A card's text in Markdown. Card rows only.
before	The item a card goes in front of. Card rows only.

A single column of Japanese with no header works too. Turn off First row contains headers and map it to Written form / Japanese text.

Comma, tab and semicolon all work. UTF-8, Shift-JIS, EUC-JP, ISO-2022-JP and UTF-16 all read. If the Japanese looks garbled, pick another encoding under Read as.

Wrap a cell in double quotes when it holds a comma, a line break or a quote.

Excel workbooks

Files: .xlsx, .xlsm

Check the sheets under Parts, then map their columns as for a CSV. A hidden sheet is tagged hidden and starts unchecked.

Cell values are read. A formula gives the value Excel saved. Dates read as YYYY-MM-DD.

Number formats, pictures, comments and phonetic guides are not read. Put readings in a column of their own.

Format an ID column as Text before typing, so 0123 keeps its zero.

Save .xls, .xlsb, .ods and Numbers files as .xlsx first. Workbooks up to 20.0 MiB.

Word tables

Files: .docx, .docm

Check the tables under Parts. The one with the most rows starts checked. A table is named by its caption, or the heading above it. Map the columns as for a CSV. Headers such as 単語, 読み and 意味 map themselves.

Phonetic Guide ruby in a cell gives the reading.

A merged cell's value sits in its first cell. The rest read empty.

A document with no table goes to Extract Words from Text. Save a .doc as .docx first. Documents up to 100.0 MiB.

eevy JSON

Files: .json

Nothing to map. Every field is fixed.

Only files exported from eevy work here. For other data, save it as CSV and import that.

{
  "format": "eevy-content",
  "schemaVersion": 1,
  "items": [
    { "type": "Word", "writtenForm": "食べる", "kana": "たべる", "meaning": "to eat" },
    { "type": "Character", "writtenForm": "食" }
  ]
}

An item can carry type, id, writtenForm, kana and meaning, nothing else.
type is "Word", "Character", "Kanji" or "Kana" (or 1 or 2).
A word needs id, writtenForm or kana. A character needs id or writtenForm.
Leave out a field you have no value for. An empty or null field is an error, except a blank meaning.
An id decides the match, whatever the text says. An id eevy can't find matches on the item's text instead.
Each item says what it is, so there is no Match as choice. A word with a selected meaning keeps that meaning. Auto select meaning shows only when some word still has a meaning to match.
A word's id is its eevy ID. A kanji's or kana's id is the character itself: 食 is 食. In a CSV, a codepoint column names a character too: 098df.
A meaning over 500 characters is cut.

Anki decks

Files: .apkg, .colpkg

In Anki, choose File > Export and pick Anki Deck Package (.apkg). Untick Include media to keep the file small. If eevy can't read the file, tick Support older Anki versions and export again.

Under Parts, check your decks. A deck brings its subdecks. Split subdecks into parts makes each subdeck a part of its own.
For each note type, set Match as and map the fields that hold the Japanese, the reading and the meaning. Front and Back never map themselves.
Furigana such as 食[た]べる gives the reading.
Pictures, sound, review history, scheduling and tags are not imported.
With New collection, its name starts as the deck's name.
Deck files up to 500.0 MiB, and 64.0 MiB of cards without media.

Mining-tool exports

Files: .csv, .tsv, .txt, .zip

Exports from Migaku, Yomitan note types such as Lapis, Kaishi and Core decks map their word, reading and meaning columns themselves. Check the mapping, then find matches.

Two columns for one field, such as Word and Lemma? Neither is used until you pick one. Lemma and MainDefinition are usually right.
Formatting and sound tags are removed. Furigana such as 食[た]べる gives the reading.
Sentence columns, known-word status and review history are not imported. Words marked Ignored in Migaku are left out.
A mapped column holds cells up to 16 KiB. Map the short definition, not a long glossary.
Language Reactor: choose the zip it downloads, or the file inside. Saved phrases are left out, and the reading comes from matching.
An Anki Notes in Plain Text export has no headers. Map each column by position.

Zips

Files: .zip

A zip holding one readable file reads that file. If it holds several, pick one under Choose a file from this zip. For another file, choose the zip again.

A zip inside a zip is not opened. Unzip it first.

Zips up to 1 GiB. An Anki deck inside one is read up to 128.0 MiB.

Books, subtitles and manga go to Extract Words from Text.

Knowledge cards

A CSV or eevy JSON file can hold knowledge cards, alone or beside words.

How to write cards in a file

A card row has a type of Card, a title, and its text as markdown or as eevy content. A CSV of only cards needs no Type column.
cardType is Explanation, Pattern, Cultural Note, Tip, Nuance. Left blank, it's an Explanation.
A card goes in front of the next word in the file, or in front of the item named in before. A card with neither goes at the very end.
Furigana works as in the card editor: 四時[よじ]. The reading sits over the kanji right before the bracket.
Color a word with a tag pair: {primary}word{/primary}. The colors are primary, secondary, accent, info, success, warning, error.
Lists stay one level: a list can't go inside another list.
A card with the same title and type as one of your cards offers Update existing card (the default) or Make a new card on its row, whatever collection you pick. An update changes the card everywhere it's used and adds it to your collection if it's not there.
Up to 500 cards in one import. A title holds up to 200 characters, and a card's text up to 5,000.

Options

Column mapping: Headers eevy knows map themselves. Change any mapping, or set a column to Ignore. If two columns claim one field, pick one.
Match as: Auto matches each row to a word or a character. The other choices match only that kind. Choices: Auto, Words, Characters. Default: Auto. An Anki deck sets it for each note type, so one note type can be words and another kanji. Not shown when your Type column fills every row.
Maximum mastery: Leaves out words you know above this level. Choices: Any level, Proficient, Familiar, Learning, Novice, New only. Default: Any level. Left-out rows go to Skipped on Review. See Mastery Dashboard.
Use this column to pick matches: Shown under the Meaning column. On by default. eevy uses your meaning to pick the entry: かける with "to bet" finds 賭ける. If two meanings fit almost equally, the row waits in Needs your choice.
Auto select meaning: On by default. Also links the meaning that matched, so study uses it.
Item types: Shown under the Type column: Words, Characters, Cards, all on by default. Rows of a type you turn off are left out. Rows with no type are always kept. Keep at least one on.

Find matches compares meanings 500 rows at a time. If some fail, choose Retry or Continue without.

On an import you reopen, changing an option turns Continue to Review into Find matches. It reads the file again and replaces your Review choices.

How cells are looked up

A tilde is dropped: ～人 finds 人, お～ finds お. Suffix and prefix entries come first.
Forms split by ; or :, such as 会う;逢う, are each looked up. The first is the one shown.
If the reading fits none of the spelling's entries, eevy looks up the spelling alone. The row then waits in Needs your choice.
Full-width and ordinary letters and digits count as the same: Tシャツ finds Ｔシャツ. A reading in romaji, such as taberu, is read as たべる.

Review

Review sorts your rows into groups. Only Needs your choice starts open.

Needs your choice:: More than one entry fits, or the row needs fixing. Choose one, or leave it out.
Matched by meaning:: eevy picked these from the meanings in your file. Nothing here needs checking.
Cards to make:: Knowledge cards eevy makes from your file. Preview one, or leave it out.
Your choices:: Rows you saved, or a bulk button answered.
Matched:: The spelling found exactly one entry.
Not found in eevy:: Nothing in eevy's dictionary matched. Use Search, press Add to glossary, or let the row be skipped.
Add to glossary:: Words you'll make yourself. Don't add sends one back where it came from.
Skipped:: Rows left out by you or by Maximum mastery, each with its reason. Restore brings one back.

A row shows the meaning from your file, when it has one.
With several parts, pick one above the groups to see only its rows.
Back to Upload takes you back to the Upload page to fix the mapping.

Choosing an entry

Pick a candidate, or one of its meanings. Whole word links the whole entry.
A linked meaning is the one study uses for that word.
Press Save to keep your choice. Clear drops it. Nothing is kept until you save.
A saved row moves to Your choices.
Search finds any entry in eevy when the right one isn't listed. Save it like any choice.
Exclude leaves a row out.

Add to glossary

On a Not found in eevy or Needs your choice row, press Add to glossary to make the word yourself. The row moves to the Add to glossary group.
Don't add sends a row back to the group it came from.
The word starts as your file wrote it, with a mark like 〜 kept, and you can edit it.
Add all to glossary adds every Not found row at once. Clear all sends every added row back.
After the import, those words wait in Needs review. You review each one before it is saved.

Answering many rows at once

These buttons sit above Needs your choice. They only answer rows nobody has answered yet.

Choose top match for all:: Takes each row's top match. A word is taken whole, never one meaning.
Treat all as Words, Treat all as Characters:: Takes each row's one entry of that kind. A row with none is left alone.
Skip unresolved rows:: Leaves out every row still waiting. The button shows how many.

Adding to a collection

Under Destination, choose Existing collection or New collection.
With several parts, choose All parts into one collection or Each part into its own collection, then Existing or New for each part.
Skip items already in an earlier part drops repeats. Off by default.
Under Order in collection, choose the order items are added in.
If some rows are skipped or not found, tick the box that says so. Rows going to your glossary don't count.
Press the import button at the foot of the page.

A full collection shows Full and cannot be chosen. Others show how much room they have.
If a collection has no room for the import, nothing is added and your choices stay. With several parts, pick another collection for the whole part. With one part, start with a smaller import.

File order (default):: Keeps the order of your file.
Frequency:: Common items first. Items without frequency data go last, in gojūon order.
Gojūon:: Sorted by reading.

Import complete

The page shows ten totals: Items attached, Already in collection, Meanings linked, Duplicates in file, Source records, Matched, Matched by meaning, Your choices, Not found in eevy, Skipped. Skipped counts every row the collection didn't get: left out, not found, invalid or sent to your glossary.

Below them, each collection the import filled shows what it got. Open collection opens it. Edit collection opens it in the Collection Editor. Back to import returns to the import start page.

When words went to your glossary, Review words opens Needs review.

Limits and privacy

Text files up to 16.0 MiB. The text sent from any file up to 8.0 MiB.
Up to 10,000 rows or notes in one import, across all its parts.
Up to 10,000 items in one collection. An import that would go past that is stopped, and nothing is added.
Up to 50 parts in one import.
Up to 5 imports in progress. Finish or cancel one to start another.
Your file is private. Adding it to a collection deletes the saved file; the items stay.
Importing needs an active Studio subscription.

Extracting Words from Text Collection Editor Collections Guide Word List Items Character Pages