Merge pull request #723 from TamasPergerDWP/master

JSONML should be protected from stack overflow exceptions caused by recursion, resolving #722

Author: stleary

src/main/java/org/json/JSONML.java

@@ -27,7 +27,32 @@ private static Object parse(
         XMLTokener x,
         boolean    arrayForm,
         JSONArray  ja,
-        boolean keepStrings
+        boolean keepStrings,
+        int currentNestingDepth
+    ) throws JSONException {
+        return parse(x,arrayForm, ja,
+            keepStrings ? JSONMLParserConfiguration.KEEP_STRINGS : JSONMLParserConfiguration.ORIGINAL,
+            currentNestingDepth);
+    }
+
+    /**
+     * Parse XML values and store them in a JSONArray.
+     * @param x       The XMLTokener containing the source string.
+     * @param arrayForm true if array form, false if object form.
+     * @param ja      The JSONArray that is containing the current tag or null
+     *     if we are at the outermost level.
+     * @param config  The parser configuration:
+     *     JSONMLParserConfiguration.ORIGINAL is the default behaviour;
+     *     JSONMLParserConfiguration.KEEP_STRINGS means Don't type-convert text nodes and attribute values.
+     * @return A JSONArray if the value is the outermost tag, otherwise null.
+     * @throws JSONException if a parsing error occurs
+     */
+    private static Object parse(
+        XMLTokener x,
+        boolean    arrayForm,
+        JSONArray  ja,
+        JSONMLParserConfiguration config,
+        int currentNestingDepth
     ) throws JSONException {
         String     attribute;
         char       c;
@@ -152,7 +177,7 @@ private static Object parse(
                             if (!(token instanceof String)) {
                                 throw x.syntaxError("Missing value");
                             }
-                            newjo.accumulate(attribute, keepStrings ? ((String)token) :XML.stringToValue((String)token));
+                            newjo.accumulate(attribute, config.isKeepStrings() ? ((String)token) :XML.stringToValue((String)token));
                             token = null;
                         } else {
                             newjo.accumulate(attribute, "");
@@ -181,7 +206,12 @@ private static Object parse(
                         if (token != XML.GT) {
                             throw x.syntaxError("Misshaped tag");
                         }
-                        closeTag = (String)parse(x, arrayForm, newja, keepStrings);
+
+                        if (currentNestingDepth == config.getMaxNestingDepth()) {
+                            throw x.syntaxError("Maximum nesting depth of " + config.getMaxNestingDepth() + " reached");
+                        }
+
+                        closeTag = (String)parse(x, arrayForm, newja, config, currentNestingDepth + 1);
                         if (closeTag != null) {
                             if (!closeTag.equals(tagName)) {
                                 throw x.syntaxError("Mismatched '" + tagName +
@@ -203,7 +233,7 @@ private static Object parse(
             } else {
                 if (ja != null) {
                     ja.put(token instanceof String
-                        ? keepStrings ? XML.unescape((String)token) :XML.stringToValue((String)token)
+                        ? (config.isKeepStrings() ? XML.unescape((String)token) : XML.stringToValue((String)token))
                         : token);
                 }
             }
@@ -224,7 +254,7 @@ private static Object parse(
      * @throws JSONException Thrown on error converting to a JSONArray
      */
     public static JSONArray toJSONArray(String string) throws JSONException {
-        return (JSONArray)parse(new XMLTokener(string), true, null, false);
+        return (JSONArray)parse(new XMLTokener(string), true, null, JSONMLParserConfiguration.ORIGINAL, 0);
     }
 
 
@@ -235,8 +265,8 @@ public static JSONArray toJSONArray(String string) throws JSONException {
      * attributes, then the second element will be JSONObject containing the
      * name/value pairs. If the tag contains children, then strings and
      * JSONArrays will represent the child tags.
-     * As opposed to toJSONArray this method does not attempt to convert 
-     * any text node or attribute value to any type 
+     * As opposed to toJSONArray this method does not attempt to convert
+     * any text node or attribute value to any type
      * but just leaves it as a string.
      * Comments, prologs, DTDs, and <pre>{@code &lt;[ [ ]]>}</pre> are ignored.
      * @param string The source string.
@@ -246,7 +276,56 @@ public static JSONArray toJSONArray(String string) throws JSONException {
      * @throws JSONException Thrown on error converting to a JSONArray
      */
     public static JSONArray toJSONArray(String string, boolean keepStrings) throws JSONException {
-        return (JSONArray)parse(new XMLTokener(string), true, null, keepStrings);
+        return (JSONArray)parse(new XMLTokener(string), true, null, keepStrings, 0);
+    }
+
+
+
+    /**
+     * Convert a well-formed (but not necessarily valid) XML string into a
+     * JSONArray using the JsonML transform. Each XML tag is represented as
+     * a JSONArray in which the first element is the tag name. If the tag has
+     * attributes, then the second element will be JSONObject containing the
+     * name/value pairs. If the tag contains children, then strings and
+     * JSONArrays will represent the child tags.
+     * As opposed to toJSONArray this method does not attempt to convert
+     * any text node or attribute value to any type
+     * but just leaves it as a string.
+     * Comments, prologs, DTDs, and <pre>{@code &lt;[ [ ]]>}</pre> are ignored.
+     * @param string The source string.
+     * @param config  The parser configuration:
+     *     JSONMLParserConfiguration.ORIGINAL is the default behaviour;
+     *     JSONMLParserConfiguration.KEEP_STRINGS means values will not be coerced into boolean
+     *       or numeric values and will instead be left as strings
+     * @return A JSONArray containing the structured data from the XML string.
+     * @throws JSONException Thrown on error converting to a JSONArray
+     */
+    public static JSONArray toJSONArray(String string, JSONMLParserConfiguration config) throws JSONException {
+        return (JSONArray)parse(new XMLTokener(string), true, null, config, 0);
+    }
+
+
+    /**
+     * Convert a well-formed (but not necessarily valid) XML string into a
+     * JSONArray using the JsonML transform. Each XML tag is represented as
+     * a JSONArray in which the first element is the tag name. If the tag has
+     * attributes, then the second element will be JSONObject containing the
+     * name/value pairs. If the tag contains children, then strings and
+     * JSONArrays will represent the child content and tags.
+     * As opposed to toJSONArray this method does not attempt to convert
+     * any text node or attribute value to any type
+     * but just leaves it as a string.
+     * Comments, prologs, DTDs, and <pre>{@code &lt;[ [ ]]>}</pre> are ignored.
+     * @param x An XMLTokener.
+     * @param config  The parser configuration:
+     *     JSONMLParserConfiguration.ORIGINAL is the default behaviour;
+     *     JSONMLParserConfiguration.KEEP_STRINGS means values will not be coerced into boolean
+     *       or numeric values and will instead be left as strings
+     * @return A JSONArray containing the structured data from the XML string.
+     * @throws JSONException Thrown on error converting to a JSONArray
+     */
+    public static JSONArray toJSONArray(XMLTokener x, JSONMLParserConfiguration config) throws JSONException {
+        return (JSONArray)parse(x, true, null, config, 0);
     }
 
 
@@ -257,8 +336,8 @@ public static JSONArray toJSONArray(String string, boolean keepStrings) throws J
      * attributes, then the second element will be JSONObject containing the
      * name/value pairs. If the tag contains children, then strings and
      * JSONArrays will represent the child content and tags.
-     * As opposed to toJSONArray this method does not attempt to convert 
-     * any text node or attribute value to any type 
+     * As opposed to toJSONArray this method does not attempt to convert
+     * any text node or attribute value to any type
      * but just leaves it as a string.
      * Comments, prologs, DTDs, and <pre>{@code &lt;[ [ ]]>}</pre> are ignored.
      * @param x An XMLTokener.
@@ -268,7 +347,7 @@ public static JSONArray toJSONArray(String string, boolean keepStrings) throws J
      * @throws JSONException Thrown on error converting to a JSONArray
      */
     public static JSONArray toJSONArray(XMLTokener x, boolean keepStrings) throws JSONException {
-        return (JSONArray)parse(x, true, null, keepStrings);
+        return (JSONArray)parse(x, true, null, keepStrings, 0);
     }
 
 
@@ -285,7 +364,7 @@ public static JSONArray toJSONArray(XMLTokener x, boolean keepStrings) throws JS
      * @throws JSONException Thrown on error converting to a JSONArray
      */
     public static JSONArray toJSONArray(XMLTokener x) throws JSONException {
-        return (JSONArray)parse(x, true, null, false);
+        return (JSONArray)parse(x, true, null, false, 0);
     }
 
 
@@ -303,10 +382,10 @@ public static JSONArray toJSONArray(XMLTokener x) throws JSONException {
      * @throws JSONException Thrown on error converting to a JSONObject
      */
     public static JSONObject toJSONObject(String string) throws JSONException {
-        return (JSONObject)parse(new XMLTokener(string), false, null, false);
+        return (JSONObject)parse(new XMLTokener(string), false, null, false, 0);
     }
-    
-    
+
+
     /**
      * Convert a well-formed (but not necessarily valid) XML string into a
      * JSONObject using the JsonML transform. Each XML tag is represented as
@@ -323,10 +402,32 @@ public static JSONObject toJSONObject(String string) throws JSONException {
      * @throws JSONException Thrown on error converting to a JSONObject
      */
     public static JSONObject toJSONObject(String string, boolean keepStrings) throws JSONException {
-        return (JSONObject)parse(new XMLTokener(string), false, null, keepStrings);
+        return (JSONObject)parse(new XMLTokener(string), false, null, keepStrings, 0);
+    }
+
+
+    /**
+     * Convert a well-formed (but not necessarily valid) XML string into a
+     * JSONObject using the JsonML transform. Each XML tag is represented as
+     * a JSONObject with a "tagName" property. If the tag has attributes, then
+     * the attributes will be in the JSONObject as properties. If the tag
+     * contains children, the object will have a "childNodes" property which
+     * will be an array of strings and JsonML JSONObjects.
+
+     * Comments, prologs, DTDs, and <pre>{@code &lt;[ [ ]]>}</pre> are ignored.
+     * @param string The XML source text.
+     * @param config  The parser configuration:
+     *     JSONMLParserConfiguration.ORIGINAL is the default behaviour;
+     *     JSONMLParserConfiguration.KEEP_STRINGS means values will not be coerced into boolean
+     *       or numeric values and will instead be left as strings
+     * @return A JSONObject containing the structured data from the XML string.
+     * @throws JSONException Thrown on error converting to a JSONObject
+     */
+    public static JSONObject toJSONObject(String string, JSONMLParserConfiguration config) throws JSONException {
+        return (JSONObject)parse(new XMLTokener(string), false, null, config, 0);
     }
 
-    
+
     /**
      * Convert a well-formed (but not necessarily valid) XML string into a
      * JSONObject using the JsonML transform. Each XML tag is represented as
@@ -341,7 +442,7 @@ public static JSONObject toJSONObject(String string, boolean keepStrings) throws
      * @throws JSONException Thrown on error converting to a JSONObject
      */
     public static JSONObject toJSONObject(XMLTokener x) throws JSONException {
-           return (JSONObject)parse(x, false, null, false);
+           return (JSONObject)parse(x, false, null, false, 0);
     }
 
 
@@ -361,7 +462,29 @@ public static JSONObject toJSONObject(XMLTokener x) throws JSONException {
      * @throws JSONException Thrown on error converting to a JSONObject
      */
     public static JSONObject toJSONObject(XMLTokener x, boolean keepStrings) throws JSONException {
-           return (JSONObject)parse(x, false, null, keepStrings);
+           return (JSONObject)parse(x, false, null, keepStrings, 0);
+    }
+
+
+    /**
+     * Convert a well-formed (but not necessarily valid) XML string into a
+     * JSONObject using the JsonML transform. Each XML tag is represented as
+     * a JSONObject with a "tagName" property. If the tag has attributes, then
+     * the attributes will be in the JSONObject as properties. If the tag
+     * contains children, the object will have a "childNodes" property which
+     * will be an array of strings and JsonML JSONObjects.
+
+     * Comments, prologs, DTDs, and <pre>{@code &lt;[ [ ]]>}</pre> are ignored.
+     * @param x An XMLTokener of the XML source text.
+     * @param config  The parser configuration:
+     *     JSONMLParserConfiguration.ORIGINAL is the default behaviour;
+     *     JSONMLParserConfiguration.KEEP_STRINGS means values will not be coerced into boolean
+     *       or numeric values and will instead be left as strings
+     * @return A JSONObject containing the structured data from the XML string.
+     * @throws JSONException Thrown on error converting to a JSONObject
+     */
+    public static JSONObject toJSONObject(XMLTokener x, JSONMLParserConfiguration config) throws JSONException {
+        return (JSONObject)parse(x, false, null, config, 0);
     }
 
 
@@ -442,6 +565,7 @@ public static String toString(JSONArray ja) throws JSONException {
         return sb.toString();
     }
 
+
     /**
      * Reverse the JSONML transformation, making an XML text from a JSONObject.
      * The JSONObject must contain a "tagName" property. If it has children,

src/main/java/org/json/JSONMLParserConfiguration.java

@@ -0,0 +1,128 @@
+package org.json;
+/*
+Public Domain.
+*/
+
+/**
+ * Configuration object for the XML to JSONML parser. The configuration is immutable.
+ */
+@SuppressWarnings({""})
+public class JSONMLParserConfiguration {
+    /**
+     * Used to indicate there's no defined limit to the maximum nesting depth when parsing a XML
+     * document to JSONML.
+     */
+    public static final int UNDEFINED_MAXIMUM_NESTING_DEPTH = -1;
+
+    /**
+     * The default maximum nesting depth when parsing a XML document to JSONML.
+     */
+    public static final int DEFAULT_MAXIMUM_NESTING_DEPTH = 512;
+
+    /** Original Configuration of the XML to JSONML Parser. */
+    public static final JSONMLParserConfiguration ORIGINAL
+        = new JSONMLParserConfiguration();
+    /** Original configuration of the XML to JSONML Parser except that values are kept as strings. */
+    public static final JSONMLParserConfiguration KEEP_STRINGS
+        = new JSONMLParserConfiguration().withKeepStrings(true);
+
+    /**
+     * When parsing the XML into JSONML, specifies if values should be kept as strings (<code>true</code>), or if
+     * they should try to be guessed into JSON values (numeric, boolean, string)
+     */
+    private boolean keepStrings;
+
+    /**
+     * The maximum nesting depth when parsing a XML document to JSONML.
+     */
+    private int maxNestingDepth = DEFAULT_MAXIMUM_NESTING_DEPTH;
+
+    /**
+     * Default parser configuration. Does not keep strings (tries to implicitly convert values).
+     */
+    public JSONMLParserConfiguration() {
+        this.keepStrings = false;
+    }
+
+    /**
+     * Configure the parser string processing and use the default CDATA Tag Name as "content".
+     * @param keepStrings <code>true</code> to parse all values as string.
+     *      <code>false</code> to try and convert XML string values into a JSON value.
+     * @param maxNestingDepth <code>int</code> to limit the nesting depth
+     */
+    private JSONMLParserConfiguration(final boolean keepStrings, final int maxNestingDepth) {
+        this.keepStrings = keepStrings;
+        this.maxNestingDepth = maxNestingDepth;
+    }
+
+    /**
+     * Provides a new instance of the same configuration.
+     */
+    @Override
+    protected JSONMLParserConfiguration clone() {
+        // future modifications to this method should always ensure a "deep"
+        // clone in the case of collections. i.e. if a Map is added as a configuration
+        // item, a new map instance should be created and if possible each value in the
+        // map should be cloned as well. If the values of the map are known to also
+        // be immutable, then a shallow clone of the map is acceptable.
+        return new JSONMLParserConfiguration(
+                this.keepStrings,
+                this.maxNestingDepth
+        );
+    }
+
+    /**
+     * When parsing the XML into JSONML, specifies if values should be kept as strings (<code>true</code>), or if
+     * they should try to be guessed into JSON values (numeric, boolean, string)
+     *
+     * @return The <code>keepStrings</code> configuration value.
+     */
+    public boolean isKeepStrings() {
+        return this.keepStrings;
+    }
+
+    /**
+     * When parsing the XML into JSONML, specifies if values should be kept as strings (<code>true</code>), or if
+     * they should try to be guessed into JSON values (numeric, boolean, string)
+     *
+     * @param newVal
+     *      new value to use for the <code>keepStrings</code> configuration option.
+     *
+     * @return The existing configuration will not be modified. A new configuration is returned.
+     */
+    public JSONMLParserConfiguration withKeepStrings(final boolean newVal) {
+        JSONMLParserConfiguration newConfig = this.clone();
+        newConfig.keepStrings = newVal;
+        return newConfig;
+    }
+
+    /**
+     * The maximum nesting depth that the parser will descend before throwing an exception
+     * when parsing the XML into JSONML.
+     * @return the maximum nesting depth set for this configuration
+     */
+    public int getMaxNestingDepth() {
+        return maxNestingDepth;
+    }
+
+    /**
+     * Defines the maximum nesting depth that the parser will descend before throwing an exception
+     * when parsing the XML into JSONML. The default max nesting depth is 512, which means the parser
+     * will throw a JsonException if the maximum depth is reached.
+     * Using any negative value as a parameter is equivalent to setting no limit to the nesting depth,
+     * which means the parses will go as deep as the maximum call stack size allows.
+     * @param maxNestingDepth the maximum nesting depth allowed to the XML parser
+     * @return The existing configuration will not be modified. A new configuration is returned.
+     */
+    public JSONMLParserConfiguration withMaxNestingDepth(int maxNestingDepth) {
+        JSONMLParserConfiguration newConfig = this.clone();
+
+        if (maxNestingDepth > UNDEFINED_MAXIMUM_NESTING_DEPTH) {
+            newConfig.maxNestingDepth = maxNestingDepth;
+        } else {
+            newConfig.maxNestingDepth = UNDEFINED_MAXIMUM_NESTING_DEPTH;
+        }
+
+        return newConfig;
+    }
+}