fix: throw error if invalid link syntax is detected

Author: chrispcampbell

packages/docs-builder/src/parse.spec.ts

@@ -275,4 +275,49 @@ there
       'Unresolved reference-style link found for lang=en link=[This is a link][unknown_ref]'
     )
   })
+
+  it('should throw an error if invalid link syntax is detected', () => {
+    const links = `\
+This is a valid normal link: [page](https://climateinteractive.org)
+
+This is a valid reference-style link: [page][ref]
+
+This is an invalid normal link: [page] (https://climateinteractive.org)
+
+This is an invalid reference-style link: [page] [ref]
+`
+
+    const md = `\
+# <!-- section:section_1 -->Section 1
+
+<!-- begin-def:block_1 -->
+
+${links}
+
+<!-- end-def -->
+
+[ref]: https://climateinteractive.org
+`
+
+    function expectedMsg(lang: string) {
+      const langPart = lang === 'en' ? '' : `lang=${lang} `
+      return `\
+Detected invalid link syntax:
+[page] (https://climateinteractive.org)
+[page] [ref]
+To fix, ensure there are no spaces between link text and link url/reference, for example: [text](url) or [text][ref] (${langPart}page=page_1.md scope=section_1)`
+    }
+
+    // Verify that an error is thrown if the English content contains invalid link syntax
+    const enContext = new Context(config, 'en')
+    expect(() => parseMarkdownPageContent(enContext, 'page_1.md', md)).toThrow(expectedMsg('en'))
+
+    // Verify that an error is thrown if the translated content contains invalid link syntax
+    const deBlocks = new Map([
+      ['section_1__title', 'Section 1'],
+      ['section_1__block_1', links]
+    ])
+    const deContext = new Context(config, 'de', undefined, deBlocks)
+    expect(() => parseMarkdownPageContent(deContext, 'page_1.md', md)).toThrow(expectedMsg('de'))
+  })
 })

packages/docs-builder/src/parse.ts

@@ -517,6 +517,7 @@ function addBlockForTokens(
       sep = '\n\n'
     }
     const blockText = blockParts.join(sep)
+    checkForInvalidLinkSyntax(context, blockText)
     context.addBlock(localBlockId, blockText)
     return tokens
   } else {
@@ -525,6 +526,7 @@ function addBlockForTokens(
     const blockText = context.getTranslatedBlockText(fullBlockId)
     if (blockText) {
       // Parse the translated tokens and insert them
+      checkForInvalidLinkSyntax(context, blockText)
       // XXX: If there is more than one token (even if some are whitespace only), we
       // need to include extra newlines after the last block, otherwise marked will
       // not parse the text correctly
@@ -539,6 +541,23 @@ function addBlockForTokens(
   }
 }
 
+/**
+ * Throw an error if the given text contains invalid link syntax.
+ */
+function checkForInvalidLinkSyntax(context: Context, md: string): void {
+  const invalidLinkRegExp = /\[([^\]]+)\]\s+\(([^)]+)\)|\[([^\]]+)\]\s+\[([^\]]+)\]/g
+  const matches = md.match(invalidLinkRegExp)
+  if (matches) {
+    let msg = 'Detected invalid link syntax:\n'
+    for (const match of matches) {
+      msg += `${match}\n`
+    }
+    msg +=
+      'To fix, ensure there are no spaces between link text and link url/reference, for example: [text](url) or [text][ref]'
+    throw new Error(context.getScopedMessage(msg))
+  }
+}
+
 /**
  * Log a warning if the given text token is on a translatable page but is
  * not included in a `def` or `begin/end-def` pair.