Skip to content

Definitions (TODO)

This page is dedicated to showcasing how definitions can be easily chosen, overwritten and customized overall.

Dictionary Placement

Dictionaries from Yomitan are sorted into the following fields:

Field Description
PrimaryDefinition The highest priority monolingual or bilingual dictionary (depending on the value of opt-first-definition-type). If you want to manually select a different dictionary, see here
SecondaryDefinition All bilingual dictionaries outside of the one selected in the primary definition.
ExtraDefinition All monolingual dictionaries outside of the one selected in the primary definition.
UtilityDictionaries All dictionaries that fall outside the category of bilingual or monolingual. For example, JMnedict or JMdict Forms.

The way that the dictionaries are sorted into the appropriate fields is by assigning a category to each individual dictionary.

Verifying Categories

You can check that your dictionaries are correctly categorized with the {jpmn-test-dict-type} helper. Under the Anki Templates code, replace Card field with {jpmn-test-dict-type} and press Test.

checking dictionary categories

An example output of the above (on the word 結構) is the following: 

「旺文社国語辞典 第十一版」: monolingual
「明鏡国語辞典 第二版」: monolingual
「ハイブリッド新辞林」: monolingual
「新明解国語辞典 第五版」: monolingual
「デジタル大辞泉」: monolingual
「NHK日本語発音アクセント新辞典」: utility
「JMDict Surface Forms」: utility
「JMdict (English)」: bilingual
「JMdict (English)」: bilingual
「JMdict (English)」: bilingual
「JMdict (English)」: bilingual
「JMdict (English)」: bilingual
「新和英」: bilingual

If a dictionary is miscategorized, you will have to edit bilingual-dict-regex or utility-dict-regex at the top of the template code. Monolingual dictionaries are considered to be dictionaries that aren't either of the two above, so no handlebars code has to be changed if one were to use more monolingual dictionaries.

To see how to edit the regex, go to this section.

Ignoring a Dictionary

If you want to see the dictionary on Yomitan but not have it show on Anki, you can use the ignored-dict-regex option.

To see how to edit the option, see the section below.

Conversely, if you want to not see the dictionary on Yomitan but want it to show up on Anki, see here.

Editing the dictionary regex

To modify a regex string:

  1. Determine the exact tag your dictionary has. To see this, take a word that has a definition in the desired dictionary, and test {jpmn-test-dict-type} like above. The string inside the quotes 「」 is exactly the tag of the dictionary.

  2. Add the dictionary tag to the string, by replacing ADD_x_DICTIONARIES_HERE. For example, if your bilingual dictionary tag is Amazing Dictionary, change ADD_BILINGUAL_DICTIONARIES_HERE to Amazing Dictionary.

    If you want to add more than one dictionary, they have to be joined with the | character. For example, if you want to add the bilingual dictionaries Amazing Dictionary and Amazing-Dictionary-2, change ADD_BILINGUAL_DICTIONARIES_HERE to Amazing Dictionary|Amazing-Dictionary-2.

Primary Definition Selection

The primary definition can either be selected automatically, or manually via highlighting the dictionary name or a section of the definition.

Primary Definition Selection: Automatic

The dictionary for the primary definition is the first bilingual dictionary (that appears on Yomitan) by default.

This can be changed to the first monolingual dictionary by changing the following Yomitan template option to monolingual:

{{~! valid values: "bilingual", "monolingual" ~}}
{{~set "opt-first-definition-type" "monolingual" ~}}

TODO re-record with only automatic

Primary Definition Selection: Manual

TODO re-record with only manual

Sometimes, you may want to override the primary definition, or highlight the definition that makes sense with the context.

This is enabled by default. In case you want to disable this behavior, set opt-selection-text-enabled to false.

This manual selection behavior does the following:

  1. If nothing is selected, then the first dictionary is chosen just like normal.

  2. If a dictionary is selected, then that dictionary will replace the first definition.

    To disable this, set opt-selection-text-dictionary to false.

  3. If a section of text is selected, then that dictionary will replace the first definition. Additionally, that section of text will be highlighted (bolded).

    To disable this, set opt-selection-text-glossary to false.

    Additionally, if you do not want to use the entire dictionary, and prefer that only the selected text is shown in the first definition, then set opt-selection-text-glossary-attempt-bold to false.

When Manual Selection Fails

In an ideal world, we would have access to exactly what dictionary was selected from the handlebars. For example, a handlebars function could tell us if the n'th dictionary was highlighted. Unfortunately, we do not live in this ideal world, so we must get this information manually.

The handlebars algorithm runs thusly:

  • If the selected text exactly matches any dictionary in the exported definition, then the dictionary entry is selected as the primary definition.
  • Otherwise, we do the same search as the above, except we search through every dictionary's HTML (in Yomitan's order) for the selected text.
  • If the selected text was found in some dictionary, then that dictionary is chosen. Otherwise, we fallback to the selected text itself.

Due to us having to find the information ourselves, there are many edge-cases where this algorithm fails.

  1. Selecting formatted text

    If you select formatted parts of text then automatic bolding will fail. An incomplete list of of formatted sections is shown below:

    Line breaks

    Line breaks cannot be properly captured by the selected text.

    Furigana

    Just like line breaks, furigana cannot be properly captured by the selected text.

    List Items (common with JMdict)

    Avoid highlighting multiple items in a list if you want automatic bolding to work.

  2. Same selected text appearing in multiple dictionaries

    Manual selection may occasionally select the wrong dictionary, but this only happens if the selected text also appears in a dictionary above the selected text.

    For example, suppose you have two bilingual dictionaries. and for the word 蛸, you highlight the word "octopus" and create the card. Both bilingual dictionaries will list "octopus", so even if you highlight the word "octopus" in the second bilingual dictionary, only the first bilingual dictionary will be chosen.

    Example: 蛸

    This is usually not a problem even if the same text appears in a dictionary above, because you'll be seeing the same definition regardless. However, if you still want a specific dictionary, highlight the dictionary tag as shown above.

  3. Selected text in HTML markup

    There are very rare edge cases when the highlighted text can be found in the internal HTML markup, leading to invalid HTML and (almost certainly) the incorrect definition.

    Example: 垣根

    On the word 垣根, if you have the デジタル大辞泉 dictionary before JMdict in Yomitan and you highlight border, then デジタル大辞泉 will be incorrectly selected as the primary definition.

    This is due to a perfect storm of events:

    • Due to Yomitan's implementation details, definitions with images contain a border property in an internal style attribute.
    • デジタル大辞泉 indeed has images for the word 垣根.
    • デジタル大辞泉 is ordered before JMdict in Yomitan

    With all these three combined, デジタル大辞泉 is prioritized over JMdict, leading to an incorrect dictionary selected and invalid HTML.

Usage on Yomitan's Search Page / Clipboard Page

On Yomitan's Search Page or Clipboard Page, the word is usually highlighted within the example sentence when searching for the word. For example, the word 「人里離れた」 is highlighted in the following image:

This highlighted word interferes with the definition selector used by the handlebars, and may cause unexpected definitions to be selected as the primary definition instead.

In order to fix this, you have a few different options:

  • Simply don't use the search page or clipboard page. Instead, you can use a texthooker setup.
  • Create a new Yomitan profile that matches the desired page(s), and disable Selected matching text within your new profile in the Yomitan settings. If you must use the search page or clipboard page, this is the recommended way to deal with the issue, as it has minimal impact on the rest of your workflow.
  • Ensure you always manually select something in the definition.
  • Simply disable the opt-selection-text-glossary handlebars option completely.
Explanation (click here)

Under normal circumstances, i.e. within an embedded popup, the selected text in a sentence does not interfere with the handlebars, because the selected text is not within the popup itself. However, within the search page / clipboard page, the sentence and its selected text is found directly within the page itself.

Internally, the handlebars to get the selected text cannot distinguish between whether the text is selected within a definition or somewhere else. As mentioned before, the handlebars has no way of actually knowing which dictionary entry or dictionary is selected, and must use this algorithm to search for the correct entry. With these two combined, the selected text within the sentence is used to search the definition. This may be especially confusing to see for people who remove the first line of monolingual dictionaries, because it is usually the first line of monolingual dictionaries that is matched and bolded.

Simplifying the Definition

If you use a monolingual dictionary, there is usually a bunch of extra information in the first line. Additionally, there is usually only one dictionary entry, making the list number redundant. Both of these can be automatically hidden with the following runtime option:

"blockquotes.simplifyDefinitions.enabled": true,

Simpifying the definition per blockquote

By default, if the above option is enabled, all definitions within the Primary Definition, Secondary Definition and Extra Definitions blockquotes are simplified. One can grant finer control to only simplify definitions within these particular blockquotes:

// Remember to enable this so the options can be used!
"blockquotes.simplifyDefinitions.enabled": true,

// The following controls precisely whether the simplifying definition is enabled
// or not for each block.
"blockquotes.simplifyDefinitions.primaryDefinition.enabled": true,
"blockquotes.simplifyDefinitions.secondaryDefinition.enabled": true,
"blockquotes.simplifyDefinitions.extraDefinitions.enabled": true,

Removing the first line: Algorithms Discussion

There are plenty of ways to remove the first line within the definition. In order to better understand how these handlebars work as well as its pitfalls, pretty much all possible options are listed below.

  1. Remove the first line completely, using Handlebars.

    Advantanges:

    • The biggest advantage this has is the exported definition is as clean and minimalistic it can possibly be.

    Disadvantages:

    • This can lead to invalid HTML, because Handlebars does not expose a HTML parser. Additionally, removing the first line completely will remove info from the export.

  2. Use JavaScript within the note template to remove the first line.

    Advantages:

    • No special handlebars are particularly required in order for this method to work, i.e. default handlebars can usually do the trick.
    • All info is present / not lost.
    • Anki's card reviewer is based off a web browser, so the JavaScript has access to a html parser. In other words, html should remain valid.
    • This approach should theoretically work on previously exported cards.

    Disadvantages:

    • The JavaScript algorithm must assume the definition is of a particular HTML structure, i.e. all <ol> elements contain a list of definitions. This is problematic if the user edits the field, as they can change it to pretty much whatever they want.
    • Javascript must be ran, which will slow down card loads.
    • This is not immediately portable between note types, because the JS must be copied/pasted for each note type.
    • If the user decides that they want to include the first line, it will likely be non-trivial to do so. Editing the raw HTML may be required to simply specify whether the first line should be removed or not.

  3. Wrap the first line with some HTML, using Handlebars.

    Advantages:

    • This does not require JS to be run at all; Only special CSS has to be used. Note that internally, some extra JS is actually run in order to further filter between dictionaries (blockquotes.simplifyDefinitions.dictsOverride.hideFirstLineMode). However, this JS is faster than the above option, because the JS only adds CSS classes to particular elements, and does not do any string or HTML parsing.
    • All info is present / not lost.
    • This algorithm is relatively simple to implement.
    • Including the first line is slightly more intuitive compared to the 2nd option: Delete the <li> node and create a fresh node.

    Disadvantages:

    • Just like the first option, this can lead to invalid HTML.
    • This is technically not immediately portable, because the CSS must be copied/pasted for each note type.
    • This solution will not work on older cards exported with different handlebars, i.e. handlebars that do not wrap the first line.

    This is currently the chosen method for this note, primarily due to speed, no info loss and simplicity. The second option may be implemented in the future as a fallback, or for special case handling for specific dictionaries.

When HTML can break

New in version 0.12.0.0 (latest version: 0.12.0.0-prerelease-19)

As mentioned in the algorithms discussion, the method used by this note can lead to invalid HTML. This is because the resulting HTML is parsed with regex in order to match that first line. In computer science, it is recommended that you should not parse HTML with regex, because it is mathematically impossible to fully describe and parse HTML with regex. Unfortunately, Yomitan's handlebars does not expose any HTML parser, so we are left with using regex to parse our HTML. Due to this, an unexpected dictionary format may cause the resulting export to be invalid HTML.

If you do not plan on using this feature, you should set opt-wrap-first-line-spans to false to remove the possibility of invalid HTML.

Hide the first line for select dictionaries

New in version 0.12.0.0 (latest version: 0.12.0.0-prerelease-19)

The following two options are useful if you want to ignore certain problematic dictionaries, that do not play well with the regex above.

{{~set "opt-first-line-regex-mode" "except"~}}
{{~#set "opt-first-line-dicts-regex"~}} ^(JMdict.*|Nico/Pixiv)$ {{~/set~}}

opt-first-line-dicts-regex specifies the list of dictionaries that should be ignored or kept. Whether the dictionaries are ignored or kept is defined in opt-first-line-regex-mode (except means that the specified dictionaries are ignored, and only means that only the specified dictionaries can have their first lines removed.)

Exporting only one dictionary entry

A "dictionary entry" in this context is the single section of text corresponding to a number indicated by Yomitan, to the very far left.

In the following example, 旺文社国語辞典 第十一版 and 明鏡国語辞典 第二版 each have one single entry, corresponding to 1 and 2 respectively. JMdict (English) has two entries, corresponding to 3 and 4.

As you may have noticed, it is almost never the case that monolingual dictionaries has multiple dictionary entries. Instead, most monolingual dictionaries store the definition as one gigantic entry.

By default, all dictionary entries are exported in the primary definition. However, if opt-primary-def-one-dict-entry-only is set to true, then only the first (or manually selected) dictionary entry will be imported into the primary definition.

Legacy JMdict Support

By default, JMdict is exported in a list format, which allows the note's CSS to re-compact the list. This is required over plaintext, because JMdict Extra cannot compact export a compact definition normally. This is a known issue with Yomitan's default handlebars.

If you prefer using a legacy version of JMdict and prefer a full line of plaintext instead of a CSS list, you can set opt-jmdict-list-format to false.

Last update: May 17, 2024