Fix positional info on tokens, nodes, also for tabs

Related-to: mdx-js/eslint-mdx#435.

package.json

@@ -26,12 +26,16 @@
   ],
   "devDependencies": {
     "@types/acorn": "^4.0.0",
+    "@types/estree": "^1.0.0",
     "@types/node": "^20.0.0",
     "acorn": "^8.0.0",
     "acorn-jsx": "^5.0.0",
     "c8": "^7.0.0",
     "micromark": "^3.0.0",
     "micromark-build": "^1.0.0",
+    "micromark-util-character": "^1.0.0",
+    "micromark-util-symbol": "^1.0.0",
+    "micromark-util-types": "^1.0.0",
     "prettier": "^2.0.0",
     "remark-cli": "^11.0.0",
     "remark-preset-wooorm": "^9.0.0",

packages/micromark-extension-mdx-expression/dev/lib/syntax.js

@@ -27,12 +27,12 @@
  *   `micromark-extension-mdx-jsx` to prohobit empty attribute values).
  */
 
-import {ok as assert} from 'uvu/assert'
 import {factoryMdxExpression} from 'micromark-factory-mdx-expression'
 import {factorySpace} from 'micromark-factory-space'
 import {markdownLineEnding, markdownSpace} from 'micromark-util-character'
 import {codes} from 'micromark-util-symbol/codes.js'
 import {types} from 'micromark-util-symbol/types.js'
+import {ok as assert} from 'uvu/assert'
 
 /**
  * Create an extension for `micromark` to enable MDX expression syntax.
@@ -111,6 +111,7 @@ export function mdxExpression(options) {
      */
     function start(code) {
       // To do: in `markdown-rs`, constructs need to parse the indent themselves.
+      // This should also be introduced in `micromark-js`.
       assert(code === codes.leftCurlyBrace, 'expected `{`')
       return before(code)
     }

packages/micromark-factory-mdx-expression/dev/index.js

@@ -10,30 +10,29 @@
 
 /**
  * @typedef MdxSignalOk
+ *   Good result.
  * @property {'ok'} type
+ *   Type.
  * @property {Program | undefined} estree
+ *   Value.
  *
  * @typedef MdxSignalNok
+ *   Bad result.
  * @property {'nok'} type
+ *   Type.
  * @property {VFileMessage} message
- *
- * @typedef MdxSignalEof
- *   Currently not given back.
- * @property {'eof'} type
- * @property {VFileMessage} message
+ *   Value.
  *
  * @typedef {MdxSignalOk | MdxSignalNok} MdxSignal
  */
 
-import {ok as assert} from 'uvu/assert'
-import {markdownLineEnding, markdownSpace} from 'micromark-util-character'
+import {markdownLineEnding} from 'micromark-util-character'
+import {eventsToAcorn} from 'micromark-util-events-to-acorn'
 import {codes} from 'micromark-util-symbol/codes.js'
 import {types} from 'micromark-util-symbol/types.js'
-import {constants} from 'micromark-util-symbol/constants.js'
-import {factorySpace} from 'micromark-factory-space'
 import {positionFromEstree} from 'unist-util-position-from-estree'
+import {ok as assert} from 'uvu/assert'
 import {VFileMessage} from 'vfile-message'
-import {eventsToAcorn} from 'micromark-util-events-to-acorn'
 
 /**
  * @this {TokenizeContext}
@@ -232,34 +231,21 @@ export function factoryMdxExpression(
       )
     }
 
-    if (markdownSpace(code)) {
-      // Idea: investigate if we’d need to use more complex stripping.
-      // Take this example:
-      //
-      // ```markdown
-      // >  aaa <b c={`
-      // >      d
-      // >  `} /> eee
-      // ```
-      //
-      // Currently, the “paragraph” starts at `> | aaa`, so for the next line
-      // here we split it into `>␠|␠␠␠␠|␠d` (prefix, this indent here,
-      // expression data).
-      // The intention above is likely for the split to be as `>␠␠|␠␠␠␠|d`,
-      // which is impossible, but we can mimick it with `>␠|␠␠␠␠␠|d`.
-      //
-      // To improve the situation, we could take `tokenizer.line_start` at
-      // the start of the expression and move past whitespace.
-      // For future lines, we’d move at most to
-      // `line_start_shifted.column + 4`.
-      return factorySpace(
-        effects,
-        before,
-        types.linePrefix,
-        constants.tabSize
-      )(code)
-    }
-
+    // Idea: investigate if we’d need to use more complex stripping.
+    // Take this example:
+    //
+    // ```markdown
+    // >  aaa <b c={`
+    // >      d
+    // >  `} /> eee
+    // ```
+    //
+    // The block quote takes one space from each line, the paragraph doesn’t.
+    // The intent above is *perhaps* for the split to be as `>␠␠|␠␠␠␠|d`,
+    // Currently, we *don’t* do anything at all, it’s `>␠|␠␠␠␠␠|d` instead.
+    //
+    // Note: we used to have some handling here, and `markdown-rs` still does,
+    // which should be removed.
     return before(code)
   }
 }

packages/micromark-factory-mdx-expression/package.json

@@ -39,7 +39,7 @@
     "default": "./index.js"
   },
   "dependencies": {
-    "micromark-factory-space": "^1.0.0",
+    "@types/estree": "^1.0.0",
     "micromark-util-character": "^1.0.0",
     "micromark-util-events-to-acorn": "^1.0.0",
     "micromark-util-symbol": "^1.0.0",