Word Count HotCRP

A live word counter for HotCRP submissions. It keeps a word count in the status bar that is measured the same way HotCRP measures text, so the number you see while writing is the number the submission form checks against its word limits — for abstracts, reviews and comments alike.

The default counting rule is a direct port of HotCRP's count_words() (lib/base.php): a word is a run of non-whitespace text whose first character is not one of - . , ; : < > ! ? * _ ~ # |. Markdown markers such as #, -, >and---are skipped, while symbols such as©, emoji, parentheses and URLs count. A link label` is one word, and a run of Chinese or Japanese text with no spaces is one word.

A local fork of Microsoft's ms-vscode.wordcount sample extension. Like the original, it puts a live word count in the status bar. It adds:

1. HotCRP's measurement, by default. The hotcrp counting rule above, so you can write to a word limit without pasting text into the submission form to check it.
2. Selection word count, live. Select text and the status bar shows how many words are in the selection out of the whole document, updating as you drag, extend with Shift, or type. Multi-cursor selections are added up.
3. A character counter you can turn on and off. Click the status bar item, or run Word Count HotCRP: Toggle Character Count from the Command Palette. The choice is saved in the wordcount.showCharacterCount setting.
4. A choice of counting rules. wordcount.countMode also offers Markdown-aware counting, the rules online word counters use, and plain whitespace splitting, for when you are writing for something other than HotCRP.

What the status bar shows

Hover over the item for detail: the number of selected ranges, and, when the character counter is on, the character count without whitespace.

Settings

If wordcount.showCharacterCount is set in the workspace's .vscode/settings.json, the toggle changes it there. Otherwise it changes your user settings.

How things are counted

Words follow the rules of wordcount.countMode:

| Mode | A word is | ## Hi © [docs](https://x.io) - ok | | --- | --- | --- | | hotcrp (default) | A run of non-whitespace text whose first character is not one of - . , ; : < > ! ? * _ ~ # |— HotCRP'scount_words()rule. Leading markers are skipped, so--wordcounts but---does not;©, emoji, ()and whole URLs count. | 4 | |markdown| A run of non-whitespace text with at least one letter or digit, so Markdown markers such as#, -, >and---are not counted. | 3 | |online| Asmarkdown, but symbols such as ©, ™and emoji count as words, and a URL is a word of its own, so a linklabelcounts as two words. Use it to match online word counters that you paste the raw text into. | 5 | |whitespace| Any run of non-whitespace characters,#and-included. The originalms-vscode.wordcountandwc -w` count this way. | 6 |

Chinese and Japanese text has no spaces between words. In markdown and online modes each Han, Hiragana or Katakana character counts as one word, as in Microsoft Word. HotCRP does not split them: in hotcrp mode, as in whitespace mode, a run of them with no spaces counts as one word.

Characters are what a reader sees as one character: an emoji with a skin tone or an accented letter counts once. A line break counts once, whether the file uses LF or CRLF line endings. The counting mode does not change the character count.

Build, test, install

npm install
npm test                      # compile, then run the unit and extension tests with node --test
npm run package               # builds word-count-HotCRP-<version>.vsix
code --install-extension word-count-HotCRP-0.4.0.vsix

The original ms-vscode.wordcount also adds a word count to the status bar. Disable or uninstall it so the two items don't sit side by side.

License

MIT; see the LICENSE file. Based on ms-vscode.wordcount 0.1.0, © Microsoft Corporation.