[bug] json output is off by one

[cameronbrill]

⏯ Playground Link

Playground link with relevant code

💻 Code

Code:

ObjectName.increment_counter(
  foo="bar",
  tags=[
    f"tag_name:'{tag_value.an_id}'"
  ],
)

Pattern:

rule:
  kind: keyword_argument
  regex: (?i)(?:\{id\}|\{.+_id\}|\{.+\.id\})
  inside:
    pattern: ObjectName.increment_counter($$$)
    stopBy: end

🙁 Actual behavior

The start and end lines for all json output are off by one. I am using a VS Code task with a problem matcher to display squiggly lines when ast-grep finds an error, but the squiggly lines currently show up on the wrong line because of this.

• ✅ ast-grep scan shows the correct start line and column (3:3)
• ✅ ast-grep scan --format github shows the correct start and end lines (line=3,endLine=5)
• ❌ ast-grep scan --json=stream lines off by one ({start: {line: 2}, end: {line:4}})
• ❌ ast-grep scan --json=pretty lines off by one ({start: {line: 2}, end: {line:4}})
• ❌ ast-grep scan --json=compact lines off by one ({start: {line: 2}, end: {line:4}})

🙂 Expected behavior

I expect the start and end lines in the json output to be the same as those from other output formats.

• ✅ ast-grep scan --json=stream should show ({start: {line: 3}, end: {line:5}})
• ✅ ast-grep scan --json=pretty should show ({start: {line: 3}, end: {line:5}})
• ✅ ast-grep scan --json=compact should show ({start: {line: 3}, end: {line:5}})

[HerringtonDarkholme]

JSON's output line and column is zero-based because most of programming language is zero based.

[HerringtonDarkholme]

Also, if you need to display errors. https://marketplace.visualstudio.com/items?itemName=ast-grep.ast-grep-vscode the official LSP plugin is recommended.

[cameronbrill]

okay that's fair, we were trying this because the vs code extension doesn't work on my teams codebase right now. but, that's unrelated to this repo, so I'll just open an issue on the vs code extension repo