For v2.0.0+ versions of Vale, there are going to be some development and release workflow changes.

Currently, the process is quite simple: development takes place on master and releases are tagged at various times.

Going forward, this is going to change. The goal will be for master to always represent the state of the latest release (in terms of both go get and the state of the live documentation) -- Essentially, it's going to be a read-only release branch. Development will take place on versioned branches (e.g., v2.0.1).

Editors + CLI tools

• [x] Sublime Text (https://github.com/ValeLint/SubVale)
• [x] Atom (https://github.com/TimKam/atomic-vale)
• [x] VS Code (https://github.com/lunaryorn/vscode-vale)
• [ ] coala CI
• [x] Vim (via ALE, thanks to @chew-z)
• [x] Emacs (https://github.com/abingham/flycheck-vale)
• [ ] JetBrains

Other

The following would be nice to have, but will probably need to access Vale through an API of some sort (see languagetool-msword10-addin, for example).

• [ ] Chrome + Firefox + Thunderbird
• [ ] MS Word
• [ ] WordPress
• [ ] Slack

There is no way to extend a base config, is there?

I mean, I'd like to have a base .vale.ini config file somewhere, and then in each project's .vale.ini do something like:

Extend = path/to/base/.vale.ini
# and here override some config

I'm trying to lint a combination of markdown and ERB templating, but I'm finding the ignore syntax isn't working.

I'm trying to exclude syntax like

<%= image_tag "#{ENV["BADGE_DOMAIN"]}/sample.svg?status=unknown" %>

and 
<% for theme in ["slack", "github", "saturn", "flickr", "solarized", "mono"] %>

The following config works for the markdown just fine, but doesn't exclude the ERB syntax.

[formats]
erb = md
txt = md

[*.erb]
TokenIgnores = (<%.*%>), \
                          (<%=.*?%>)

[*]
Vale.Spelling = error

and the same regex in BlockIgnores also doesn't work.

Thanks

• OS (e.g., macOS, Windows, Linux, etc.): Linux
• Install method (e.g., Go, Homebrew, direct download, etc.): go get
• Vale version (the result of vale -v): vale version master

I have some text in a quote block:

  | I have a very large collection

I have added the following configuration:

SkippedScopes = script, style, pre, figure, blockquote

But vale still complains about the use of very.

This is one of the most common feature requests / misunderstandings about how Vale's rules work. So, I'd like to take some time to try to get this right—whether that means changing existing functionality or simply improving the documentation on the topic.

Why things are the way they are

To provide some background: Vale is written in Go and uses its stdlib regexp package to evaluate all rules. The implementation is based on Google's RE2 library:

RE2 is a fast, safe, thread-friendly alternative to backtracking regular expression engines like those used in PCRE, Perl, and Python. It is a C++ library.

If you'd like to learn more about the decisions that went into RE2, the wiki on google/re2 goes into some detail.

A different take, but not a limitation

The academics of the issue aside, my stance has always been that I don't want this to be a limitation of Vale; all else equal, I'd rather Vale be a little slower but more capable.

I've written a lot of general-purpose regex (using many languages) and even more Vale-specific regex, and I've yet to find something that seemed impossible or unnecessarily difficult to implement with Vale in comparison to other tools. This leads me to believe that the problem is more related to documentation than implementation.

The documentation currently mentions the limitation of Go and links to a blog post that goes into detail about how Vale approaches the problem. But it's clear that this could be explained better.

Go's stdlib regex engine doesn't support lookarounds, but Vale does

Vale supports both positive and negative lookarounds. It does so by taking advantage of the YAML wrapper around its rules.

Positive lookarounds

The first variation of lookarounds are the positive lookahead and lookbehind assertions:

• lookahead (x(?=y)): "I want to match x only if it precedes y without matching y."
• lookbehind ((?<=y)x): "I want to match x only if it is preceded by y without matching y."

For Vale’s purposes, this should be straightforward to implement:

extends: existence
message: "Consider removing 'dialog'."
level: warning
ignorecase: true
tokens:
  - 'dialog (?:box)'

The basic idea here is that you don't have show the end-user every thing that the rule matched; you can use the message to gain control over what you show.

Negative lookarounds

The second variation of lookarounds are the negative lookahead and lookbehind assertions:

• lookahead (x(?!y)): "I want to match x only if it does not precede y.”
• lookbehind ((?<!y)x): "I want to match x only if it is not preceded by y."

For Vale’s purposes, these cases can be emulated using substitution checks:

extends: substitution
message: "Use '%s' instead of '%s'"
level: error
ignorecase: false
swap:
  '(?:[^\s]*) ?VS Code': Microsoft VS Code

This rule will flag any instance of "VS Code" that isn't preceded by "Microsoft"; it's functionally equivalent to (?<!Micrsoft )VS Code. The key takeaway is that you can write patterns that include the accepted case since Vale will only raise alerts for those that don't match it.

So, instead of looking for (?<!Micrsoft ) (anything that isn't 'Microsoft'), you can look for anything and Vale will compare it to the user-provided accepted case.

Feedback

I know I haven't written rules that cover every possible use case, so I'd be quite interested in seeing examples of rules you either can't implement or find hard to implement without using raw lookarounds.

@ravlen: I know this specifically came up in your research session. Do you have any thoughts on the existing support or hard-to-implement examples you could share?

After upgrading from 1.7.1 to 2.1.0, Vale is splitting the contents of URLs apart on word boundaries and testing the individual words in spell-check. I've been reading over the spelling style description to see if there were any scope parameters we've been missing, and I don't think there are.

The branch is docs-aqualls-vale-spelling, and the Spelling.yml file contains:

extends: spelling
message: 'Spelling check: "%s"?'
level: warning
ignore:
  - gitlab/spelling-exceptions.txt

(Note that in the branch it's warning but I'm modifying it locally to error so the problems are easier to see.)

Here's the command I run, and the results:

$ vale --no-wrap --minAlertLevel error doc/administration/troubleshooting/gitlab_rails_cheat_sheet.md

 doc/administration/troubleshooting/gitlab_rails_cheat_sheet.md
 288:82  error  Spelling check: "dev"?     gitlab.Spelling
 675:29  error  Spelling check: "gitlab"?  gitlab.Spelling
 675:70  error  Spelling check: "ee"?      gitlab.Spelling
 693:38  error  Spelling check: "gitlab"?  gitlab.Spelling
 736:5   error  Spelling check: "ee"?      gitlab.Spelling

When I look at gitlab_rails_cheat_sheet.md:

# Line 675
Features listed in <https://gitlab.com/gitlab-org/gitlab/blob/master/ee/app/models/license.rb>.

# Line 693
From [Zendesk ticket #91083](https://gitlab.zendesk.com/agent/tickets/91083) (internal)

At first I thought it was the <…> format of the links, but I've tried converting them over to the more standard Markdown syntax of [link word](url) and I'm still getting the same issue.

Ideas?

• OS (e.g., macOS, Windows, Linux, etc.) - MacOS
• Install method: Homebrew
• Vale version: 2.1.0

We use Hugo and have translated pages living in the same repository as our standard language content. (article.md, article.fr.md, and so on.)

Most of our Vale rules are specific to the English content (article.md), but some operate on all files.

I’ve been using a configuration file like this:

[*.{de,es,fr,it,jp,ru}.md]
BasedOnStyles = 
OurStyle.general_rule = YES

[*.md]
BasedOnStyles = OurStyle

I did it this way to avoid turning off all the rules that don’t apply to the translations. Is this the best way to accomplish this, or is there a better way?

(Copied from WTD Slack for posterity and visibility.)

Hello,

I'm trying to run vale from the command line on Windows 10 and pass text via stdin.

Running vale against a file vale --config="D:\tools\tmp\vale\.vale.ini" sample.md gets me a correct warning:

sample.md
 1:1  warning  Use the Oxford comma in a list  base.OxfordComa
               of three or more items...

But when I try this: vale --config="D:\tools\tmp\vale\.vale.ini" --ext='.md' 'A,B and C' I get no warning whatsoever.

• OS: Windows 10
• Install method: direct download
• Vale version: 2.2.2

I apologize if it's something obvious that escapes me...

Thank you, Alex

I can't get BlockIgnores to work. I want to ignore text in code blocks. I verified the regex on https://regex101.com/r/prulD0/2

My .mdx file.

---
title: "Doc title"
description: "Title description."
tags: ["tag1", "tag2"]
published: true
featured: false
---

import customComponent from "components/customComponent";
import customComponent2 from "components/customComponent2";

Customers interact with products using different input methods: keyboard, mouse, touch, voice, and more. So use generic verbs that work with any input method. Don't use input-specific verbs, such as click or swipe. Instead, use the verbs listed here.

Content on the first screen (also called above the fold) is the most likely to be read. Many readers won’t scroll further without a compelling reason. So as always, keep it short, and lead with what’s most important to the customer. Keep in mind that what appears above the fold varies by device and screen resolution.

Numerous studies have shown that readers look at specific areas of a page first. In left-to-right languages, people read in an F shape, giving the most attention to the upper-left corner of a page. Put your most important information there. Or try placing an offer or a Buy or Download button there. If the navigation is on the left, put your most important information in the upper-left corner of the text pane.

1. Run the `install` script to configure XYZ, 

    ```bash
    unzip -d foldername filename.zip

    ./foldername/subfolder/install.sh
    ```
2. Run second command
	```bash
		command -i more | grep something
	```

Vale config

# Example Vale config file (`.vale.ini` or `_vale.ini`)

# Core settings
StylesPath = vale

# The minimum alert level to display (suggestion, warning, or error).
#
# CI builds will only fail on error-level alerts.
MinAlertLevel = warning

# Global settings (applied to every syntax)
[*]
# List of styles to load
BasedOnStyles = Microsoft
# Style.Rule = {YES, NO} to enable or disable a specific rule
# vale.Editorializing = YES
# You can also change the level associated with a rule
# vale.Hedging = error
BlockIgnores = (?s) *(\x60\x60\x60[a-z]*\n[\s\S]*?\x60\x60\x60)

# Syntax-specific settings
# These overwrite any conflicting global settings
# [*.{mdx}]

• OS Ubuntu 18.04 on Windows
• Install method Go
• Vale version master

I'm currently evaluating using Vale to lint the GitHub documentation at https://github.com/github/docs but I'm running into a problem with TokenIgnores seemingly not ignoring content it should be. I have a minimal repro case at https://github.com/lee-dohm/vale-test. Is there something I'm doing wrong?

Steps to reproduce

1. Clone the repo at https://github.com/lee-dohm/vale-test
2. Execute the command: vale test.md

Expected: No errors
Actual: 1 error

Notice that the first Liquid tag containing the text is flagged as an error, but the second identical Liquid tag on the same line is not.

Extra info

• I've created a test case at regex101 that shows that the pattern that I'm using for TokenIgnores is ignoring what I expect, no more, no less
• Using vale ls-config shows that there doesn't seem to be a parsing error in the configuration file

Configuration

• OS: macOS 11.2.1
• Installation method: Homebrew
• Vale version: vale version 2.9.1

• OS Ubuntu Linux
• Install method (first snap, then overwritten with direct download)
• vale version 2.15.4

Maybe I'm missing something trying to follow Quick Start

My .vale.ini:

StylesPath = styles

MinAlertLevel = suggestion

Packages = Google, write-good

[*]
BasedOnStyles = Vale, Google, write-good

Now I run vale sync, and it gives an error:

E201 Invalid value provided [{project_path}/.vale.ini:1:14]:

   1* StylesPath = styles
   2  
   3  MinAlertLevel = suggestion

The path '{project_path}/styles' does not exist.

So I mkdir styles and I get

E100 [loadStyles] Runtime error

style 'Google' does not exist on StylesPath

I'm confused, looks like sync both tries to load styles for me and expects them to exist already.

• OS: MacOS Silicon running a VSCode docker dev container with Linux (Linux 4ceca93ca1ad 5.15.49-linuxkit #1 SMP PREEMPT Tue Sep 13 07:51:32 UTC 2022 aarch64 GNU/Linux)
• Install method: direct download
• Version: vale version 2.21.3

Hi,

I'm having a hard time understanding how spell check works with a custom dictionary. Can you please point me in the right direction of what I'm doing wrong here?

This is what I've tried doing so far:

• Read through all other issues here to see if there's something obvious that I'm not seeing. No luck though.
• I've downloaded the two en_UK-files from here and placed them in the fixtures folder.

My general/Spelling.yml looks like so:

# https://extensions.libreoffice.org/en/extensions/show/english-dictionaries
# Download and unzip to retrieve the needed files
extends: spelling
message: "'%s' is misspelled"
# Relative to config file
dicpath: ./fixtures/
append: false
level: error
dictionaries:
  - en_GB

This is my .vale.ini:

StylesPath = styles
MinAlertLevel = suggestion

[*]
BasedOnStyles = my, general, Vale

Running this command and I would have expected at least some errors due to the bastardisation of the English language:

(.venv) vscode ➜ /workspaces/python-vale (main ✗) $ vale --config /workspaces/python-vale/.vale.ini "This is quite obvisly a teached afair."
✔ 0 errors, 0 warnings and 0 suggestions in stdin.

I saw somewhere here that you had used the Hunspell command directly so I tried that as well. This throws some errors as expected.

(.venv) vscode ➜ /workspaces/python-vale (main ✗) $ echo "This is quite obvisly a teached afair." | hunspell -d fixtures/en_GB
Hunspell 1.7.0
*
*
*
& obvisly 1 14: obviously
*
& teached 10 24: reached, teaches, teacher, leached, peached, beached, cacheted, cheated, teach, attached
& afair 5 32: afar, fair, affair, a fair, AFAIK

• OS : Arch Linux
• Install method : Installed as a plugin in neovim IDE LunarVim
• Vale version : 2.21.0

What's happening

1. In french dictionary, the flag NOSUGGEST is followed by -- (two hyphens), which makes vale exit with an error By replacing the -- by ! in the .aff file, it works but the NOSUGGEST flag in the .dic file is not taking care of. Though, my guess is that it can be fixed by replacing all -- by ! in the .dic file.

2. The spell checking for french dictionary is not complete. Using my Arch Linux hunspell dictionary, it simply does not work. As an exemple, conjugated verbs and adjectives are labelled as not correct. In this example peut is a conjugated form.

Source files https://github.com/pcapiod/dotfiles/tree/main/dics

In the files labelled warnings, I changed the NOSUGGEST to ! so it can work. In the files labelled not_working, they are original file from my Arch Linux and give the NOSUGGEST error.

If you need more information, let me know ! Thanks :)

Installed via Homebrew on macOS.

Given the following input:

<html>
<body>
  <p>This paragraph has has a duplication.</p>
  <p>This paragraph has <b>has</b> a duplication.</p>
  <p>
    This paragraph has has a
    duplication on multiple lines.
  </p>

  <p>This paragraph has <b>has</b> an <font>obsolete</font> tag</p>
  <p>
    This paragraph has <b>has</b> a
    duplication on multiple lines.
  </p>
  <p>
    This paragraph has
    has a duplication
    across lines.
  </p>
</body>
</html>

I get the following output:

 sample.html
 3:21  error  'has' is repeated!  Vale.Repetition 
 4:1   error  'has' is repeated!  Vale.Repetition 
 6:20  error  'has' is repeated!  Vale.Repetition 

✖ 3 errors, 0 warnings and 0 suggestions in 1 file.

Despite the example containing six instances of the same error, only three are actually being reported. Additionally, the second report (<p>… has <b>has</b> …</p>) is described as covering the entire line, rather than more useful description of the error location.

Hi,

I encountered some unexpected behaviour, which is shown only in a very special situation. I defined a substitution rule, which searches for a simple term:

extends: substitution
message: Use '%s' instead of '%s'
level: error
swap:
  'term': Term

Then I defined some text, which must be exactly as follow:

• It must contain the term
• then a URL containing the same term (can be in the same line as the first term)
• at least one line break
• the exact same URL again

The substitution is unexpectedly reported in the URL, not in the first word "term".

term                            <- i'd expect the substitution to be applied here, ....
https://abc.com/term

https://abc.com/term            <- ..., but the substitution is applied here. the word "term" in the URL is reported to be bad

The CLI reports following:

 some-path/error.md
 4:17  error  Use 'Term' instead of 'term'  Rules.test 

The rule works as expected, when you add following change to above example:

• remove the line break between the two URLs
• make the URLs different (for example remove an 'a' from any of the URL)
• remove one of both URLs
• remove the term from at least one of the both URLs

Any of above changes lead to following outout (which is correct in my opinion)

 some-path/error.md
 1:1  error  Use 'Term' instead of 'term'  Rules.test 

The issue feels kind of similar to https://github.com/errata-ai/vale/issues/455, but I am not able to tell that for sure.

Setup

• macOS Monterey, 12.3.1
• I installed vale through the nix package manager
• vale version 2.20.0

.vale.ini

StylesPath = styles

[*.md]
BasedOnStyles = Vale, Rules

Folder Structure

styles/Rules/test.yml (contains substitution shown above)
some-path/error.md
.vale.ini

vale Command

I execute vale ./some-path/*

It would be great if one could run vale on files and provide either a single style or a subset of styles to apply.

This would help in testing new styles and allows creating reports for a specific violation.

AFAIK this option is currently not available. I guess one could work around it by creating a specific configuration for running a specific style but that is quite cumbersome in comparison to providing styles as command line options.