🥼 Improve suffix citation parsing

.changeset/spotty-peaches-invent.md

@@ -0,0 +1,5 @@
+---
+'markdown-it-myst': patch
+---
+
+Improve the suffix label parsing for citations

packages/markdown-it-myst/src/citations.ts

@@ -25,7 +25,8 @@ export interface Citation {
 export const citationsPlugin: PluginWithOptions = (md) => {
   const regexes = {
     citation: /^([^^-]|[^^].+?)?(-)?@([\w][\w:.#$%&\-+?<>~/]*)(.+)?$/,
-    inText: /^@((?:[\w|{][\w:.#$%&\-+?<>~/]*[\w|}])|\w)(\s*)(\[)?/,
+    // Only allow a short [suffix] for in text citations (e.g. 50 characters)
+    inText: /^@((?:[\w|{][\w:.#$%&\-+?<>~/]*[\w|}])|\w)(\s*)(\[([^\]]{1,50})\])?/,
     allowedBefore: /^[^a-zA-Z.0-9]$/,
   };
 
@@ -50,24 +51,14 @@ export const citationsPlugin: PluginWithOptions = (md) => {
           (token as any).col = [state.pos];
         }
         if (match[3]) {
-          // suffix is there
-          const suffixStart = state.pos + match[0].length;
-          const suffixEnd = state.md.helpers.parseLinkLabel(state, suffixStart);
-          const charAfter = state.src.codePointAt(suffixEnd + 1);
-          if (suffixEnd > 0 && charAfter != 0x28 && charAfter != 0x5b /* ( or [ */) {
-            const suffix = state.src.slice(suffixStart, suffixEnd);
-            citation.suffix = state.md.parseInline(suffix, state.env);
-            state.pos += match[0].length + suffixEnd - suffixStart + 1;
-            if (token) {
-              token.content = match[0] + suffix + ']';
-              (token as any).col.push(state.pos);
-            }
-          } else {
-            state.pos += match[0].length - match[2].length - match[3].length;
-            if (token) {
-              token.content = match[0];
-              (token as any).col.push(state.pos);
-            }
+          // The in-text citation is followed by [suffix]
+          // Another way to do this is to use `state.md.helpers.parseLinkLabel(state, suffixStart);`
+          const suffix = match[4];
+          citation.suffix = state.md.parseInline(suffix, state.env);
+          state.pos += match[0].length;
+          if (token) {
+            token.content = match[0];
+            (token as any).col.push(state.pos);
           }
         } else {
           state.pos += match[0].length - match[2].length;

packages/markdown-it-myst/tests/citations.yml

@@ -202,3 +202,28 @@ cases:
           - type: text
             content: ', the authors...'
       - type: paragraph_close
+  - title: Nested labels
+    md: 'These include experimentally produced alkaline magmas from @iacovino2016 [`alkaline.xlsx`], basaltic melt inclusions from Kilauea [@tucker2019] and Gakkel Ridge [@bennett2019 `basalts.xlsx`], basaltic melt inclusions from Cerro Negro volcano, Nicaragua [@roggensack2001 `cerro_negro.xlsx`], and rhyolite melt inclusions from the Taupo Volcanic Center, New Zealand [@myers2019] and a topaz rhyolite from the Rio Grande Rift @mercer2015 [`rhyolites.xlsx`].'
+    tokens:
+      - type: paragraph_open
+      - type: inline
+        children:
+          - type: text
+            content: 'These include experimentally produced alkaline magmas from '
+          - type: cite
+            content: '@iacovino2016 [`alkaline.xlsx`]'
+            meta:
+              label: iacovino2016
+              kind: narrative
+              suffix:
+                - content: '`alkaline.xlsx`'
+          - type: text
+            content: ', basaltic melt inclusions from Kilauea '
+          - type: cite
+            content: '@tucker2019'
+            meta:
+              label: tucker2019
+              kind: parenthetical
+          - type: text
+            content: ' and Gakkel Ridge '
+      - type: paragraph_close