Add support for single-quoted strings in mojangson parser (#53)

* Initial plan

* Add support for single-quoted strings in mojangson parser

Co-authored-by: extremeheat <[email protected]>

---------

Co-authored-by: copilot-swe-agent[bot] <[email protected]>
Co-authored-by: extremeheat <[email protected]>

Author: Copilot

grammar.ne

@@ -6,6 +6,7 @@ JVALUE -> JOBJECT {% (d) => d[0] %}
         | "'" _ JOBJECT _ "'" {% (d) => d[2] %}
         | JARRAY  {% (d) => d[0] %}
         | STRING  {% (d) => d[0] %}
+        | SINGLE_QUOTED_STRING {% (d) => d[0] %}
         | "null"  {% (d) => null %}
 
 JOBJECT -> "{" _ "}" {% (d) => { return { type: 'compound', value: {} } } %}
@@ -21,6 +22,8 @@ PAIR -> STRING _ ":" _ JVALUE {% (d) => [d[0].value, d[4]] %}
 STRING -> "\"" ( [^\\"] | "\\" ["bfnrt\/\\] | "\\u" [a-fA-F0-9] [a-fA-F0-9] [a-fA-F0-9] [a-fA-F0-9] ):* "\"" {% (d) => parseValue( JSON.parse(d.flat(3).map(b => b.replace('\n', '\\n')).join('')) ) %}
         | [^\"\'}\]:;,\s]:+ {% (d) => parseValue(d[0].join('')) %}
 
+SINGLE_QUOTED_STRING -> "'" ( [^\\'] | "\\" ["bfnrt\/\\'] | "\\u" [a-fA-F0-9] [a-fA-F0-9] [a-fA-F0-9] [a-fA-F0-9] ):* "'" {% (d) => parseSingleQuoteString(d) %}
+
 @{%
 
 // Because of unquoted strings, parsing can be ambiguous.
@@ -48,6 +51,38 @@ function parseValue (str) {
   return { value: str, type: 'string' }
 }
 
+function parseSingleQuoteString(d) {
+  // Build the string content from the parsed parts similar to double-quoted strings
+  // The structure is: ["'", [content parts], "'"]
+  // d[1] contains the content between quotes
+  const content = d[1] || []
+  let str = "'"
+  for (const part of content) {
+    if (Array.isArray(part)) {
+      str += part.flat().join('')
+    } else if (part) {
+      str += part
+    }
+  }
+  str += "'"
+  
+  // Process escape sequences to convert to actual string value
+  // Replace escaped single quotes with actual single quotes
+  // and handle other escape sequences
+  str = str.replace(/\\'/g, "\\'")  // Keep escaped single quotes for JSON parsing
+           .replace(/\\"/g, '\\"')   // Keep escaped double quotes
+  
+  // Convert to a JSON-compatible string by replacing outer single quotes with double quotes
+  str = '"' + str.slice(1, -1).replace(/"/g, '\\"').replace(/\\'/g, "'") + '"'
+  
+  try {
+    return { value: JSON.parse(str), type: 'string' }
+  } catch (e) {
+    // If JSON parsing fails, return the raw string content
+    return { value: str.slice(1, -1), type: 'string' }
+  }
+}
+
 function extractPair(kv, output) {
   if (kv[0] !== undefined) {
     output[kv[0]] = kv[1]

index.js

@@ -92,13 +92,68 @@ function parse (text) {
   try {
     const parserNE = new nearley.Parser(nearley.Grammar.fromCompiled(grammar))
     parserNE.feed(text)
-    return parserNE.results[0]
+    // When there are multiple parse results (ambiguous grammar),
+    // prefer results with more structured types (compound, list) over strings
+    const results = parserNE.results
+    if (results.length > 1) {
+      // Score each result based on how "structured" it is
+      const scored = results.map((r, i) => {
+        const score = scoreResult(r)
+        return { result: r, score }
+      })
+      // Sort by score descending (higher score = more structured)
+      scored.sort((a, b) => b.score - a.score)
+      return scored[0].result
+    }
+    return results[0]
   } catch (e) {
     e.message = `Error parsing text '${text}'`
     throw e
   }
 }
 
+function scoreResult (obj) {
+  let score = 0
+  if (!obj || typeof obj !== 'object') return score
+
+  // Prefer compound and list types over string types
+  if (obj.type === 'compound') score += 10
+  if (obj.type === 'list') score += 10
+  if (obj.type === 'string') score -= 1
+
+  // Recursively score nested structures
+  if (obj.value && typeof obj.value === 'object') {
+    if (Array.isArray(obj.value)) {
+      obj.value.forEach(item => {
+        score += scoreResult(item)
+      })
+    } else if (obj.value.value && Array.isArray(obj.value.value)) {
+      // This is a list type with value.value array
+      // The items in this array are raw values, not wrapped in { type, value }
+      // Score based on whether items are objects (compounds) vs primitives
+      obj.value.value.forEach(item => {
+        if (item && typeof item === 'object' && !Array.isArray(item)) {
+          // This is likely a compound object
+          score += 10
+          // Recursively score the object's properties
+          Object.values(item).forEach(prop => {
+            score += scoreResult(prop)
+          })
+        } else if (typeof item === 'string') {
+          score -= 1
+        }
+      })
+    } else {
+      // This is a compound with nested values
+      Object.values(obj.value).forEach(item => {
+        score += scoreResult(item)
+      })
+    }
+  }
+
+  return score
+}
+
 module.exports = {
   parse,
   simplify,