add the approach from mermaid-js#4910 pr

packages/parser/src/language/sankey/sankey.langium

@@ -1,26 +1,5 @@
 grammar Sankey
-
-// common stuff in ../common/common.langium
-// can't import directly because of SANKEY_LINK_NODE parse rule
-interface Common {
-  accDescr?: string;
-  accTitle?: string;
-  title?: string;
-}
-
-fragment TitleAndAccessibilities:
-  ((accDescr=ACC_DESCR | accTitle=ACC_TITLE | title=TITLE) NEWLINE+)+
-;
-
-terminal NEWLINE: /\r?\n/;
-terminal ACC_DESCR: /accDescr(?:[\t ]*:[\t ]*[^\n\r]*?(?=%%)|\s*{[^}]*})|accDescr(?:[\t ]*:[\t ]*[^\n\r]*|\s*{[^}]*})/;
-terminal ACC_TITLE: /accTitle[\t ]*:[\t ]*[^\n\r]*?(?=%%)|accTitle[\t ]*:[\t ]*[^\n\r]*/;
-terminal TITLE: /title(?:[\t ]+[^\n\r]*?|)(?=%%)|title(?:[\t ]+[^\n\r]*|)/;
-
-hidden terminal WHITESPACE: /[\t ]+/;
-hidden terminal YAML: /---[\t ]*\r?\n[\s\S]*?---[\t ]*(?!.)/;
-hidden terminal DIRECTIVE: /[\t ]*%%{[\s\S]*?}%%\s*/;
-hidden terminal SINGLE_LINE_COMMENT: /[\t ]*%%[^\n\r]*/;
+import "../common/common";
 
 // actual grammar
 interface Sankey extends Common {
@@ -38,7 +17,7 @@ entry Sankey returns Sankey:
 ;
 
 SankeyLink:
-  source=SANKEY_LINK_NODE SANKEY_LINK_COMMA target=SANKEY_LINK_NODE SANKEY_LINK_COMMA value=SANKEY_LINK_VALUE NEWLINE+
+  source=SANKEY_LINK_NODE SANKEY_LINK_COMMA target=SANKEY_LINK_NODE SANKEY_LINK_COMMA value=SANKEY_LINK_VALUE EOL
 ;
 
 terminal SANKEY_LINK_COMMA: /,/;

packages/parser/src/language/sankey/sankeyMatcher.ts

@@ -1,3 +1,8 @@
+import type { CustomPatternMatcherFunc } from 'chevrotain';
+
+import { accessibilityDescrRegex, accessibilityTitleRegex, titleRegex } from '../common/matcher.js';
+import { matchAnyRegexps } from '../../utils.js';
+
 /**
  * Matches sankey link source and target node
  */
@@ -7,3 +12,42 @@ export const sankeyLinkNodeRegex = /(?:"((?:""|[^"])+)")|([^\n\r,]+(?=%%)|[^\n\r
  * Matches sankey link value
  */
 export const sankeyLinkValueRegex = /("(?:0|[1-9]\d*)(?:\.\d+)?"|[\t ]*(?:0|[1-9]\d*)(?:\.\d+)?)/;
+
+/**
+ * Try to match the text with a SankeyNodeLink.
+ *
+ * These have to be checked first because they take precedence.
+ * Langium does not provide any way to specify precedence for grammars (or parts thereof)
+ * that are `imported` into another grammar definition file (e.g. sankey.langium).
+ * Specifically, the order of _tokens_ defined in the imported file (e.g. common.langium)
+ * may or may not come before (or after) the _tokens_ defined in the including file (e.g. sankey.langium).
+ *
+ * Thus, we have to manually handle this by first checking for matches that should take precedence
+ * (in this case title, accessibility title, and accessibility description)
+ * over matching this token.
+ *
+ * First check if the text matches a title, accessibility title, or accessibility description.
+ * If it matches one of those, return null (no match with a SankeyNodeLink).
+ *
+ * If it does not match one of those, then check to see if the text matches the
+ * SankeyNodeLink Regexp.
+ *
+ * Note that _all_ regular expressions have the sticky flag set.
+ *
+ * @param text - text to check for matches
+ * @param startingOffset - offset to start at
+ *
+ * @returns  Null if there is no match, else return the RegExpExecArray with the match.
+ */
+export const matchSankeyLinkNode: CustomPatternMatcherFunc = (text, startingOffset) => {
+  const regexpsToFail: RegExp[] = [accessibilityDescrRegex, accessibilityTitleRegex, titleRegex];
+  const targetRegexp: RegExp = sankeyLinkNodeRegex;
+
+  const matchedOtherPatterns = matchAnyRegexps(text, startingOffset, regexpsToFail);
+  if (matchedOtherPatterns) {
+    return null;
+  }
+  const stickyTargetRegexp = new RegExp(targetRegexp, targetRegexp.flags + 'y');
+  stickyTargetRegexp.lastIndex = startingOffset;
+  return stickyTargetRegexp.exec(text);
+};

packages/parser/src/utils.ts

@@ -0,0 +1,25 @@
+const STICKY_FLAG = 'y';
+
+/**
+ * Check to see if text matches any of the given RegExps. Return true as soon
+ * as there is a match. Use the sticky flag on all RegExps.
+ *
+ * @param text - the string to try to match
+ * @param startingOffset - offset to start at
+ * @param regexps - a list of RegExps to check for matches
+ * @internal
+ *
+ * @returns true if the text matches any of the RegExps (with the sticky flag set),
+ *    else returns false.
+ */
+export function matchAnyRegexps(text: string, startingOffset: number, regexps: RegExp[]): boolean {
+  const found = false;
+  for (const regexp of regexps) {
+    const currentRegexp = new RegExp(regexp, regexp.flags + STICKY_FLAG);
+    currentRegexp.lastIndex = startingOffset;
+    if (currentRegexp.exec(text) !== null) {
+      return true;
+    }
+  }
+  return found;
+}