Search to notes

Search to notes is an Anki addon to rapidly generate notes from an image web search (Anki forum).

Search to notes (S2N) is designed to rapidly generate notes from a list of terms, such as a list of anatomy names (for all of us that tend to get lists with 400 Latin names that are on the next exam).

1. Paste/load the list
2. Specify the search query template and run the query
3. Select which images to use for each term
4. Specify note details and generate notes

1. Paste/load list

Edit the search terms manually, in the Enter/paste window or load them from a text file (each line will be interpreted as a search term). The terms will be parsed and "split" on tab character and the sum as well as part available in the query template (see below). Remarks:

• Ctrl+return will close (accept) the Enter/Paste window, escape will cancel.
• When new terms are loaded (even if they are the same as the old ones) all downloaded fi.e. are removed (of course everything that has already been generated to notes is retai.e. inside Anki).

2. Specify search query template and run the query

The template will be applied to each term using the following substitutions and configurations:

• %0 - will be replaced with the complete term
• %1 - will be replaced by the first part of the term (everything up to the first tab character, i.e. if there are no tab characters it will be the same as %0)
• %2 - will be replaced by the second part of the term (everything between the first and second tab character), etc.
• Any %\<digit> that doesn't have a corresponding part in the term (i.e. %3 in the template when a term only has one tab character) is deleted.
• maxn:<digits> - limits the number of images to be downloaded (all found if omitted).

The search syntax is dependent on the search engine used, Search to notes currently ships with the following eninges (which search engine is used is set in addon configuration, Google or DuckDuckGo (API)):

• Google: Google images search using web page parsing. (instructions and instructions)
• DuckDuckGo (API): DuckDuckGo images search using the DDG API. (instructions)

The addon will try to use curl to download images as is more robust than Python requests, if the addon doesn't find curl installed on the system it will fall back to using requests. curl should come installed from Windows 10 and onwards, on macOS and some linux dists - ensure it is installed (or install it) on your platform to benefit.

Remarks

• I think all med students know of si.e. that have a lot of useful anatomy images with specific parts highlighted, I will refrain from spelling it out lest they change their site to make things like this more difficult.
• It is likely useful to set nmax - the image search and downloading is slow and the relevance gets lower rather rapidly.
• When pressing "Search" a dialog will present the terms and their respective search query to allow the user to inspect and abort if there was some mistake in the template.

3. Select which images to use for each term

Select (so that they are highlighted) the relevant images for each term (S2N remembers which images are selected for each term as you swap between them). Remarks:

• All actions can be made with the mouse (selecting term, selecting/deselecting images).
• Keyboard shortcuts for moving down and up in the list of terms (default 'S' and 'W' respectively) even when out of focus reduces the need for mouse movement to allow rapid iteration through the terms.
• Movement between the image thumbnails can be done with the arrow keys, space bar will select/deselect.
• A zoom view is available by right click/return on a thumbnail. Escape, return, right and left click closes the zoom window (it can also be left open).
• The size of the images in the thumbnail view can be set in the config.

4. Specify note generation details

Select deck, note type, and in which field the terms and images go respectively (e.g. term on front, images on back or vice versa).

• The max size of the images can be set in the config file (images that are larger are scaled down to the max retaining proportions).
• For basic style notes one note of the specified type per term will be generated to the selected deck, term and images inserted in the specified fields.
• For cloze style notes one note of the specified type will be generated and each term/image inserted as a cloze, either with the term as prompt and the image(s) inside the cloze or vice versa. With the term as prompt the clozes will be inserted with a line break between each "pair". With the image(s) as prompt the clozes will be inserted as a table (table attributes adjustable in the config, i.e. you can insert style="color: red;" or add a CSS class etc.) to be able to see which images and cloze belong together.
• Once generated a dialog will list any terms that did not get any notes/clozes (because no images were found in the search or no images were selected) to allow copying those to run another query on or create those notes some other way.

Configuration

Configuration is made through the addon configuration dialog:

• Engine: Which search engine implementation to use, shipped alternatives are Google and DuckDuckGo (API).
• Shortcut open: Shortcut to open the Search to notes dialog from the Anki main window.
• Shortcut next/previous term: Shortcut to move down/up in the list of terms in the Search to notes main window.
• Shortcut close: Shortcut to close Search to notes main window.
• Thumbnail height/width: Dimensions of the thumbnails in the Search to notes main window.
• Image height/width: Dimensions to scale images to when generating notes.
• Cloze <table>/<td> attributes: Attributes added to <table>/<td> tags when generating cloze notes, for instance to apply some sort of styling (style="border: 1px solid black; border-collapse: collapse;") or a class (class="my-own-styling-class").
• Listview light mode/Listview dark mode: Styling (notably of how the current as well as selected images are hightlighted) depending on light or dark mode.
• Internal state: Addon internal state, do not edit.

Remarks

The results returned by the DDG API are somewhat mediocre, anyone that is up for it can plug in another search engine (requirements in the source, but basically provide a search function that returns a list of matches as well as functions that return tooltip and legend strings). We will see what explodes with 2.1.50 and the move to Qt 6.

Credits

• DEEPAN for inspiration for DuckDuckGo implementation

Changelog

• 2022-01-01: Prepare code for Anki Qt5 to Qt6 transition. Added local error logging.
• 2022-05-18: Bug fixes.
• 2023-08-14: Add Google images search, code refactor, add some configurations, update README.