\ref{} is detected as end of a sentence

[atsuyaw]

When I use \ref{}, the command was detected as end of a sentence:

It is expected to detect an error in sentences such as line 199, but errors outed in line 200 and 201.

\ref{} command is used for cross-reference frequently in academic writing.
In fact, such structure of sentences were contained a lot in my file, so the output were filled with these errors.

[pddg]

@atsuyaw
Thank you for reporting the issue. I confirmed that it is a bug.
Sorry for the delay in addressing the other reported issues 😥 .
We will do our best to fix it.

[pddg]

How to reproduce

Environment

• Node.js v14.17.0
• textlint-plugin-latex2e v1.1.3
• textlint v12.0.0
• textlint-rule-ja-no-mixed-period v2.1.1

Procedure

$ mkdir textlint-latex2e-issue168 && cd $_
$ npm init -y
$ npm install --save \
    [email protected] \
    [email protected] \
    [email protected]
$ cat << EOF > sample.tex
本日は図\ref{fig:sunnyy}に示すように晴天なり。
EOF
$ cat << EOF > .textlintrc
{
    "plugins": ["latex2e"],
    "rules": {
        "ja-no-mixed-period": {
            "periodMark": "。"
        }
    }
}
EOF
$ textlint sample.tex

/Users/pudding/Factory/textlint-latex2e-issue168/sample.tex
  1:4  error  文末が"。"で終わっていません。  ja-no-mixed-period

✖ 1 problem (1 error, 0 warnings)

Cause of this issue

There seems to be an error in the way Paragraph is interpreted. There should be an inline node and a block node, but since \ref is interpreted as a block node, there are two separate Paragraphs, which should be one.

$ npx tex2tast -r sample.tex
{
  "type": "Document",
  "raw": "本日は図\\ref{fig:sunnyy}に示すように晴天なり。\n",
  "children": [
    {
      "raw": "本日は図",
      "type": "Paragraph",
      "children": [
        {
          "raw": "本日は図",
          "type": "Str",
          "value": "本日は図"
        }
      ]
    },
    {
      "raw": "\\ref{fig:sunnyy}",
      "value": "fig:sunnyy",
      "type": "Html"
    },
    {
      "raw": "に示すように晴天なり。",
      "type": "Paragraph",
      "children": [
        {
          "raw": "に示すように晴天なり。",
          "type": "Str",
          "value": "に示すように晴天なり。"
        }
      ]
    }
  ]
}

This problem can be avoided by interpreting it as an inline element, just like label and url.

[pddg]

hmm...

This problem is caused by #70 . HTML type nodes contain two types, inline and block, but there is no way to tell the difference, so they are always interpreted as block elements.

textlint-plugin-latex2e/src/paragraphize.ts

Lines 38 to 54 in 8bd0910

	for (const node of rootNode.children) {
	switch (node.type) {
	case ASTNodeTypes.CodeBlock:
	case ASTNodeTypes.Header:
	case ASTNodeTypes.Html:
	pushChild(paragraph);
	paragraph = [];
	children.push(node);
	break;
	case "parbreak":
	pushChild(paragraph);
	paragraph = [];
	break;
	default:
	paragraph.push(node);
	}
	}

@tani Do you know of any good solutions?

[pddg]

Oh, I found HtmlBlock in ASTNodeTypes definitions.

https://github.com/textlint/textlint/blob/master/packages/%40textlint/ast-node-types/src/index.ts#L26

We need to use both Html and HtmlBlock.

[pddg]

@atsuyaw
We published a new version of textlint-plugin-latex2e v1.1.4. It includes some fixes for #158 and this issue.
Please check and let me know if there are any bugs. Thank you!

[atsuyaw]

@pddg
Thank you for your corresponding!
I confirmed the problems solved.