core/java/android/util/JsonReader.java

/*
 * Copyright (C) 2010 The Android Open Source Project
 *
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 *
 *      http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */

package android.util;

import libcore.internal.StringPool;

import java.io.Closeable;
import java.io.EOFException;
import java.io.IOException;
import java.io.Reader;
import java.util.ArrayList;
import java.util.List;


/**
 * Reads a JSON (<a href="http://www.ietf.org/rfc/rfc4627.txt">RFC 4627</a>)
 * encoded value as a stream of tokens. This stream includes both literal
 * values (strings, numbers, booleans, and nulls) as well as the begin and
 * end delimiters of objects and arrays. The tokens are traversed in
 * depth-first order, the same order that they appear in the JSON document.
 * Within JSON objects, name/value pairs are represented by a single token.
 *
 * <h3>Parsing JSON</h3>
 * To create a recursive descent parser for your own JSON streams, first create
 * an entry point method that creates a {@code JsonReader}.
 *
 * <p>Next, create handler methods for each structure in your JSON text. You'll
 * need a method for each object type and for each array type.
 * <ul>
 *   <li>Within <strong>array handling</strong> methods, first call {@link
 *       #beginArray} to consume the array's opening bracket. Then create a
 *       while loop that accumulates values, terminating when {@link #hasNext}
 *       is false. Finally, read the array's closing bracket by calling {@link
 *       #endArray}.
 *   <li>Within <strong>object handling</strong> methods, first call {@link
 *       #beginObject} to consume the object's opening brace. Then create a
 *       while loop that assigns values to local variables based on their name.
 *       This loop should terminate when {@link #hasNext} is false. Finally,
 *       read the object's closing brace by calling {@link #endObject}.
 * </ul>
 * <p>When a nested object or array is encountered, delegate to the
 * corresponding handler method.
 *
 * <p>When an unknown name is encountered, strict parsers should fail with an
 * exception. Lenient parsers should call {@link #skipValue()} to recursively
 * skip the value's nested tokens, which may otherwise conflict.
 *
 * <p>If a value may be null, you should first check using {@link #peek()}.
 * Null literals can be consumed using either {@link #nextNull()} or {@link
 * #skipValue()}.
 *
 * <h3>Example</h3>
 * Suppose we'd like to parse a stream of messages such as the following: <pre> {@code
 * [
 *   {
 *     "id": 912345678901,
 *     "text": "How do I read JSON on Android?",
 *     "geo": null,
 *     "user": {
 *       "name": "android_newb",
 *       "followers_count": 41
 *      }
 *   },
 *   {
 *     "id": 912345678902,
 *     "text": "@android_newb just use android.util.JsonReader!",
 *     "geo": [50.454722, -104.606667],
 *     "user": {
 *       "name": "jesse",
 *       "followers_count": 2
 *     }
 *   }
 * ]}</pre>
 * This code implements the parser for the above structure: <pre>   {@code
 *
 *   public List<Message> readJsonStream(InputStream in) throws IOException {
 *     JsonReader reader = new JsonReader(new InputStreamReader(in, "UTF-8"));
 *     try {
 *       return readMessagesArray(reader);
 *     } finally {
 *       reader.close();
 *     }
 *   }
 *
 *   public List<Message> readMessagesArray(JsonReader reader) throws IOException {
 *     List<Message> messages = new ArrayList<Message>();
 *
 *     reader.beginArray();
 *     while (reader.hasNext()) {
 *       messages.add(readMessage(reader));
 *     }
 *     reader.endArray();
 *     return messages;
 *   }
 *
 *   public Message readMessage(JsonReader reader) throws IOException {
 *     long id = -1;
 *     String text = null;
 *     User user = null;
 *     List<Double> geo = null;
 *
 *     reader.beginObject();
 *     while (reader.hasNext()) {
 *       String name = reader.nextName();
 *       if (name.equals("id")) {
 *         id = reader.nextLong();
 *       } else if (name.equals("text")) {
 *         text = reader.nextString();
 *       } else if (name.equals("geo") && reader.peek() != JsonToken.NULL) {
 *         geo = readDoublesArray(reader);
 *       } else if (name.equals("user")) {
 *         user = readUser(reader);
 *       } else {
 *         reader.skipValue();
 *       }
 *     }
 *     reader.endObject();
 *     return new Message(id, text, user, geo);
 *   }
 *
 *   public List<Double> readDoublesArray(JsonReader reader) throws IOException {
 *     List<Double> doubles = new ArrayList<Double>();
 *
 *     reader.beginArray();
 *     while (reader.hasNext()) {
 *       doubles.add(reader.nextDouble());
 *     }
 *     reader.endArray();
 *     return doubles;
 *   }
 *
 *   public User readUser(JsonReader reader) throws IOException {
 *     String username = null;
 *     int followersCount = -1;
 *
 *     reader.beginObject();
 *     while (reader.hasNext()) {
 *       String name = reader.nextName();
 *       if (name.equals("name")) {
 *         username = reader.nextString();
 *       } else if (name.equals("followers_count")) {
 *         followersCount = reader.nextInt();
 *       } else {
 *         reader.skipValue();
 *       }
 *     }
 *     reader.endObject();
 *     return new User(username, followersCount);
 *   }}</pre>
 *
 * <h3>Number Handling</h3>
 * This reader permits numeric values to be read as strings and string values to
 * be read as numbers. For example, both elements of the JSON array {@code
 * [1, "1"]} may be read using either {@link #nextInt} or {@link #nextString}.
 * This behavior is intended to prevent lossy numeric conversions: double is
 * JavaScript's only numeric type and very large values like {@code
 * 9007199254740993} cannot be represented exactly on that platform. To minimize
 * precision loss, extremely large values should be written and read as strings
 * in JSON.
 *
 * <p>Each {@code JsonReader} may be used to read a single JSON stream. Instances
 * of this class are not thread safe.
 */
public final class JsonReader implements Closeable {

    private static final String TRUE = "true";
    private static final String FALSE = "false";

    private final StringPool stringPool = new StringPool();

    /** The input JSON. */
    private final Reader in;

    /** True to accept non-spec compliant JSON */
    private boolean lenient = false;

    /**
     * Use a manual buffer to easily read and unread upcoming characters, and
     * also so we can create strings without an intermediate StringBuilder.
     * We decode literals directly out of this buffer, so it must be at least as
     * long as the longest token that can be reported as a number.
     */
    private final char[] buffer = new char[1024];
    private int pos = 0;
    private int limit = 0;

    /*
     * The offset of the first character in the buffer.
     */
    private int bufferStartLine = 1;
    private int bufferStartColumn = 1;

    private final List<JsonScope> stack = new ArrayList<JsonScope>();
    {
        push(JsonScope.EMPTY_DOCUMENT);
    }

    /**
     * The type of the next token to be returned by {@link #peek} and {@link
     * #advance}. If null, peek() will assign a value.
     */
    private JsonToken token;

    /** The text of the next name. */
    private String name;

    /*
     * For the next literal value, we may have the text value, or the position
     * and length in the buffer.
     */
    private String value;
    private int valuePos;
    private int valueLength;

    /** True if we're currently handling a skipValue() call. */
    private boolean skipping = false;

    /**
     * Creates a new instance that reads a JSON-encoded stream from {@code in}.
     */
    public JsonReader(Reader in) {
        if (in == null) {
            throw new NullPointerException("in == null");
        }
        this.in = in;
    }

    /**
     * Configure this parser to be  be liberal in what it accepts. By default,
     * this parser is strict and only accepts JSON as specified by <a
     * href="http://www.ietf.org/rfc/rfc4627.txt">RFC 4627</a>. Setting the
     * parser to lenient causes it to ignore the following syntax errors:
     *
     * <ul>
     *   <li>End of line comments starting with {@code //} or {@code #} and
     *       ending with a newline character.
     *   <li>C-style comments starting with {@code /*} and ending with
     *       {@code *}{@code /}. Such comments may not be nested.
     *   <li>Names that are unquoted or {@code 'single quoted'}.
     *   <li>Strings that are unquoted or {@code 'single quoted'}.
     *   <li>Array elements separated by {@code ;} instead of {@code ,}.
     *   <li>Unnecessary array separators. These are interpreted as if null
     *       was the omitted value.
     *   <li>Names and values separated by {@code =} or {@code =>} instead of
     *       {@code :}.
     *   <li>Name/value pairs separated by {@code ;} instead of {@code ,}.
     * </ul>
     */
    public void setLenient(boolean lenient) {
        this.lenient = lenient;
    }

    /**
     * Returns true if this parser is liberal in what it accepts.
     */
    public boolean isLenient() {
        return lenient;
    }

    /**
     * Consumes the next token from the JSON stream and asserts that it is the
     * beginning of a new array.
     */
    public void beginArray() throws IOException {
        expect(JsonToken.BEGIN_ARRAY);
    }

    /**
     * Consumes the next token from the JSON stream and asserts that it is the
     * end of the current array.
     */
    public void endArray() throws IOException {
        expect(JsonToken.END_ARRAY);
    }

    /**
     * Consumes the next token from the JSON stream and asserts that it is the
     * beginning of a new object.
     */
    public void beginObject() throws IOException {
        expect(JsonToken.BEGIN_OBJECT);
    }

    /**
     * Consumes the next token from the JSON stream and asserts that it is the
     * end of the current object.
     */
    public void endObject() throws IOException {
        expect(JsonToken.END_OBJECT);
    }

    /**
     * Consumes {@code expected}.
     */
    private void expect(JsonToken expected) throws IOException {
        peek();
        if (token != expected) {
            throw new IllegalStateException("Expected " + expected + " but was " + peek());
        }
        advance();
    }

    /**
     * Returns true if the current array or object has another element.
     */
    public boolean hasNext() throws IOException {
        peek();
        return token != JsonToken.END_OBJECT && token != JsonToken.END_ARRAY;
    }

    /**
     * Returns the type of the next token without consuming it.
     */
    public JsonToken peek() throws IOException {
        if (token != null) {
          return token;
        }

        switch (peekStack()) {
            case EMPTY_DOCUMENT:
                replaceTop(JsonScope.NONEMPTY_DOCUMENT);
                JsonToken firstToken = nextValue();
                if (!lenient && token != JsonToken.BEGIN_ARRAY && token != JsonToken.BEGIN_OBJECT) {
                    throw new IOException(
                            "Expected JSON document to start with '[' or '{' but was " + token);
                }
                return firstToken;
            case EMPTY_ARRAY:
                return nextInArray(true);
            case NONEMPTY_ARRAY:
                return nextInArray(false);
            case EMPTY_OBJECT:
                return nextInObject(true);
            case DANGLING_NAME:
                return objectValue();
            case NONEMPTY_OBJECT:
                return nextInObject(false);
            case NONEMPTY_DOCUMENT:
                try {
                    JsonToken token = nextValue();
                    if (lenient) {
                        return token;
                    }
                    throw syntaxError("Expected EOF");
                } catch (EOFException e) {
                    return token = JsonToken.END_DOCUMENT; // TODO: avoid throwing here?
                }
            case CLOSED:
                throw new IllegalStateException("JsonReader is closed");
            default:
                throw new AssertionError();
        }
    }

    /**
     * Advances the cursor in the JSON stream to the next token.
     */
    private JsonToken advance() throws IOException {
        peek();

        JsonToken result = token;
        token = null;
        value = null;
        name = null;
        return result;
    }

    /**
     * Returns the next token, a {@link JsonToken#NAME property name}, and
     * consumes it.
     *
     * @throws IOException if the next token in the stream is not a property
     *     name.
     */
    public String nextName() throws IOException {
        peek();
        if (token != JsonToken.NAME) {
            throw new IllegalStateException("Expected a name but was " + peek());
        }
        String result = name;
        advance();
        return result;
    }

    /**
     * Returns the {@link JsonToken#STRING string} value of the next token,
     * consuming it. If the next token is a number, this method will return its
     * string form.
     *
     * @throws IllegalStateException if the next token is not a string or if
     *     this reader is closed.
     */
    public String nextString() throws IOException {
        peek();
        if (token != JsonToken.STRING && token != JsonToken.NUMBER) {
            throw new IllegalStateException("Expected a string but was " + peek());
        }

        String result = value;
        advance();
        return result;
    }

    /**
     * Returns the {@link JsonToken#BOOLEAN boolean} value of the next token,
     * consuming it.
     *
     * @throws IllegalStateException if the next token is not a boolean or if
     *     this reader is closed.
     */
    public boolean nextBoolean() throws IOException {
        peek();
        if (token != JsonToken.BOOLEAN) {
            throw new IllegalStateException("Expected a boolean but was " + token);
        }

        boolean result = (value == TRUE);
        advance();
        return result;
    }

    /**
     * Consumes the next token from the JSON stream and asserts that it is a
     * literal null.
     *
     * @throws IllegalStateException if the next token is not null or if this
     *     reader is closed.
     */
    public void nextNull() throws IOException {
        peek();
        if (token != JsonToken.NULL) {
            throw new IllegalStateException("Expected null but was " + token);
        }

        advance();
    }

    /**
     * Returns the {@link JsonToken#NUMBER double} value of the next token,
     * consuming it. If the next token is a string, this method will attempt to
     * parse it as a double using {@link Double#parseDouble(String)}.
     *
     * @throws IllegalStateException if the next token is not a literal value.
     */
    public double nextDouble() throws IOException {
        peek();
        if (token != JsonToken.STRING && token != JsonToken.NUMBER) {
            throw new IllegalStateException("Expected a double but was " + token);
        }

        double result = Double.parseDouble(value);
        advance();
        return result;
    }

    /**
     * Returns the {@link JsonToken#NUMBER long} value of the next token,
     * consuming it. If the next token is a string, this method will attempt to
     * parse it as a long. If the next token's numeric value cannot be exactly
     * represented by a Java {@code long}, this method throws.
     *
     * @throws IllegalStateException if the next token is not a literal value.
     * @throws NumberFormatException if the next literal value cannot be parsed
     *     as a number, or exactly represented as a long.
     */
    public long nextLong() throws IOException {
        peek();
        if (token != JsonToken.STRING && token != JsonToken.NUMBER) {
            throw new IllegalStateException("Expected a long but was " + token);
        }

        long result;
        try {
            result = Long.parseLong(value);
        } catch (NumberFormatException ignored) {
            double asDouble = Double.parseDouble(value); // don't catch this NumberFormatException
            result = (long) asDouble;
            if ((double) result != asDouble) {
                throw new NumberFormatException(value);
            }
        }

        advance();
        return result;
    }

    /**
     * Returns the {@link JsonToken#NUMBER int} value of the next token,
     * consuming it. If the next token is a string, this method will attempt to
     * parse it as an int. If the next token's numeric value cannot be exactly
     * represented by a Java {@code int}, this method throws.
     *
     * @throws IllegalStateException if the next token is not a literal value.
     * @throws NumberFormatException if the next literal value cannot be parsed
     *     as a number, or exactly represented as an int.
     */
    public int nextInt() throws IOException {
        peek();
        if (token != JsonToken.STRING && token != JsonToken.NUMBER) {
            throw new IllegalStateException("Expected an int but was " + token);
        }

        int result;
        try {
            result = Integer.parseInt(value);
        } catch (NumberFormatException ignored) {
            double asDouble = Double.parseDouble(value); // don't catch this NumberFormatException
            result = (int) asDouble;
            if ((double) result != asDouble) {
                throw new NumberFormatException(value);
            }
        }

        advance();
        return result;
    }

    /**
     * Closes this JSON reader and the underlying {@link Reader}.
     */
    public void close() throws IOException {
        value = null;
        token = null;
        stack.clear();
        stack.add(JsonScope.CLOSED);
        in.close();
    }

    /**
     * Skips the next value recursively. If it is an object or array, all nested
     * elements are skipped. This method is intended for use when the JSON token
     * stream contains unrecognized or unhandled values.
     */
    public void skipValue() throws IOException {
        skipping = true;
        try {
            if (!hasNext() || peek() == JsonToken.END_DOCUMENT) {
                throw new IllegalStateException("No element left to skip");
            }
            int count = 0;
            do {
                JsonToken token = advance();
                if (token == JsonToken.BEGIN_ARRAY || token == JsonToken.BEGIN_OBJECT) {
                    count++;
                } else if (token == JsonToken.END_ARRAY || token == JsonToken.END_OBJECT) {
                    count--;
                }
            } while (count != 0);
        } finally {
            skipping = false;
        }
    }

    private JsonScope peekStack() {
        return stack.get(stack.size() - 1);
    }

    private JsonScope pop() {
        return stack.remove(stack.size() - 1);
    }

    private void push(JsonScope newTop) {
        stack.add(newTop);
    }

    /**
     * Replace the value on the top of the stack with the given value.
     */
    private void replaceTop(JsonScope newTop) {
        stack.set(stack.size() - 1, newTop);
    }

    private JsonToken nextInArray(boolean firstElement) throws IOException {
        if (firstElement) {
            replaceTop(JsonScope.NONEMPTY_ARRAY);
        } else {
            /* Look for a comma before each element after the first element. */
            switch (nextNonWhitespace()) {
                case ']':
                    pop();
                    return token = JsonToken.END_ARRAY;
                case ';':
                    checkLenient(); // fall-through
                case ',':
                    break;
                default:
                    throw syntaxError("Unterminated array");
            }
        }

        switch (nextNonWhitespace()) {
            case ']':
                if (firstElement) {
                    pop();
                    return token = JsonToken.END_ARRAY;
                }
                // fall-through to handle ",]"
            case ';':
            case ',':
                /* In lenient mode, a 0-length literal means 'null' */
                checkLenient();
                pos--;
                value = "null";
                return token = JsonToken.NULL;
            default:
                pos--;
                return nextValue();
        }
    }

    private JsonToken nextInObject(boolean firstElement) throws IOException {
        /*
         * Read delimiters. Either a comma/semicolon separating this and the
         * previous name-value pair, or a close brace to denote the end of the
         * object.
         */
        if (firstElement) {
            /* Peek to see if this is the empty object. */
            switch (nextNonWhitespace()) {
                case '}':
                    pop();
                    return token = JsonToken.END_OBJECT;
                default:
                    pos--;
            }
        } else {
            switch (nextNonWhitespace()) {
                case '}':
                    pop();
                    return token = JsonToken.END_OBJECT;
                case ';':
                case ',':
                    break;
                default:
                    throw syntaxError("Unterminated object");
            }
        }

        /* Read the name. */
        int quote = nextNonWhitespace();
        switch (quote) {
            case '\'':
                checkLenient(); // fall-through
            case '"':
                name = nextString((char) quote);
                break;
            default:
                checkLenient();
                pos--;
                name = nextLiteral(false);
                if (name.isEmpty()) {
                    throw syntaxError("Expected name");
                }
        }

        replaceTop(JsonScope.DANGLING_NAME);
        return token = JsonToken.NAME;
    }

    private JsonToken objectValue() throws IOException {
        /*
         * Read the name/value separator. Usually a colon ':'. In lenient mode
         * we also accept an equals sign '=', or an arrow "=>".
         */
        switch (nextNonWhitespace()) {
            case ':':
                break;
            case '=':
                checkLenient();
                if ((pos < limit || fillBuffer(1)) && buffer[pos] == '>') {
                    pos++;
                }
                break;
            default:
                throw syntaxError("Expected ':'");
        }

        replaceTop(JsonScope.NONEMPTY_OBJECT);
        return nextValue();
    }

    private JsonToken nextValue() throws IOException {
        int c = nextNonWhitespace();
        switch (c) {
            case '{':
                push(JsonScope.EMPTY_OBJECT);
                return token = JsonToken.BEGIN_OBJECT;

            case '[':
                push(JsonScope.EMPTY_ARRAY);
                return token = JsonToken.BEGIN_ARRAY;

            case '\'':
                checkLenient(); // fall-through
            case '"':
                value = nextString((char) c);
                return token = JsonToken.STRING;

            default:
                pos--;
                return readLiteral();
        }
    }

    /**
     * Returns true once {@code limit - pos >= minimum}. If the data is
     * exhausted before that many characters are available, this returns
     * false.
     */
    private boolean fillBuffer(int minimum) throws IOException {
        // Before clobbering the old characters, update where buffer starts
        for (int i = 0; i < pos; i++) {
            if (buffer[i] == '\n') {
                bufferStartLine++;
                bufferStartColumn = 1;
            } else {
                bufferStartColumn++;
            }
        }

        if (limit != pos) {
            limit -= pos;
            System.arraycopy(buffer, pos, buffer, 0, limit);
        } else {
            limit = 0;
        }

        pos = 0;
        int total;
        while ((total = in.read(buffer, limit, buffer.length - limit)) != -1) {
            limit += total;

            // if this is the first read, consume an optional byte order mark (BOM) if it exists
                if (bufferStartLine == 1 && bufferStartColumn == 1
                        && limit > 0 && buffer[0] == '\ufeff') {
                pos++;
                bufferStartColumn--;
            }

            if (limit >= minimum) {
                return true;
            }
        }
        return false;
    }

    private int getLineNumber() {
        int result = bufferStartLine;
        for (int i = 0; i < pos; i++) {
            if (buffer[i] == '\n') {
                result++;
            }
        }
        return result;
    }

    private int getColumnNumber() {
        int result = bufferStartColumn;
        for (int i = 0; i < pos; i++) {
            if (buffer[i] == '\n') {
                result = 1;
            } else {
                result++;
            }
        }
        return result;
    }

    private int nextNonWhitespace() throws IOException {
        while (pos < limit || fillBuffer(1)) {
            int c = buffer[pos++];
            switch (c) {
                case '\t':
                case ' ':
                case '\n':
                case '\r':
                    continue;

                case '/':
                    if (pos == limit && !fillBuffer(1)) {
                        return c;
                    }

                    checkLenient();
                    char peek = buffer[pos];
                    switch (peek) {
                        case '*':
                            // skip a /* c-style comment */
                            pos++;
                            if (!skipTo("*/")) {
                                throw syntaxError("Unterminated comment");
                            }
                            pos += 2;
                            continue;

                        case '/':
                            // skip a // end-of-line comment
                            pos++;
                            skipToEndOfLine();
                            continue;

                        default:
                            return c;
                    }

                case '#':
                    /*
                     * Skip a # hash end-of-line comment. The JSON RFC doesn't
                     * specify this behaviour, but it's required to parse
                     * existing documents. See http://b/2571423.
                     */
                    checkLenient();
                    skipToEndOfLine();
                    continue;

                default:
                    return c;
            }
        }

        throw new EOFException("End of input");
    }

    private void checkLenient() throws IOException {
        if (!lenient) {
            throw syntaxError("Use JsonReader.setLenient(true) to accept malformed JSON");
        }
    }

    /**
     * Advances the position until after the next newline character. If the line
     * is terminated by "\r\n", the '\n' must be consumed as whitespace by the
     * caller.
     */
    private void skipToEndOfLine() throws IOException {
        while (pos < limit || fillBuffer(1)) {
            char c = buffer[pos++];
            if (c == '\r' || c == '\n') {
                break;
            }
        }
    }

    private boolean skipTo(String toFind) throws IOException {
        outer:
        for (; pos + toFind.length() <= limit || fillBuffer(toFind.length()); pos++) {
            for (int c = 0; c < toFind.length(); c++) {
                if (buffer[pos + c] != toFind.charAt(c)) {
                    continue outer;
                }
            }
            return true;
        }
        return false;
    }

    /**
     * Returns the string up to but not including {@code quote}, unescaping any
     * character escape sequences encountered along the way. The opening quote
     * should have already been read. This consumes the closing quote, but does
     * not include it in the returned string.
     *
     * @param quote either ' or ".
     * @throws NumberFormatException if any unicode escape sequences are
     *     malformed.
     */
    private String nextString(char quote) throws IOException {
        StringBuilder builder = null;
        do {
            /* the index of the first character not yet appended to the builder. */
            int start = pos;
            while (pos < limit) {
                int c = buffer[pos++];

                if (c == quote) {
                    if (skipping) {
                        return "skipped!";
                    } else if (builder == null) {
                        return stringPool.get(buffer, start, pos - start - 1);
                    } else {
                        builder.append(buffer, start, pos - start - 1);
                        return builder.toString();
                    }

                } else if (c == '\\') {
                    if (builder == null) {
                        builder = new StringBuilder();
                    }
                    builder.append(buffer, start, pos - start - 1);
                    builder.append(readEscapeCharacter());
                    start = pos;
                }
            }

            if (builder == null) {
                builder = new StringBuilder();
            }
            builder.append(buffer, start, pos - start);
        } while (fillBuffer(1));

        throw syntaxError("Unterminated string");
    }

    /**
     * Reads the value up to but not including any delimiter characters. This
     * does not consume the delimiter character.
     *
     * @param assignOffsetsOnly true for this method to only set the valuePos
     *     and valueLength fields and return a null result. This only works if
     *     the literal is short; a string is returned otherwise.
     */
    private String nextLiteral(boolean assignOffsetsOnly) throws IOException {
        StringBuilder builder = null;
        valuePos = -1;
        valueLength = 0;
        int i = 0;

        findNonLiteralCharacter:
        while (true) {
            for (; pos + i < limit; i++) {
                switch (buffer[pos + i]) {
                case '/':
                case '\\':
                case ';':
                case '#':
                case '=':
                    checkLenient(); // fall-through
                case '{':
                case '}':
                case '[':
                case ']':
                case ':':
                case ',':
                case ' ':
                case '\t':
                case '\f':
                case '\r':
                case '\n':
                    break findNonLiteralCharacter;
                }
            }

            /*
             * Attempt to load the entire literal into the buffer at once. If
             * we run out of input, add a non-literal character at the end so
             * that decoding doesn't need to do bounds checks.
             */
            if (i < buffer.length) {
                if (fillBuffer(i + 1)) {
                    continue;
                } else {
                    buffer[limit] = '\0';
                    break;
                }
            }

            // use a StringBuilder when the value is too long. It must be an unquoted string.
            if (builder == null) {
                builder = new StringBuilder();
            }
            builder.append(buffer, pos, i);
            valueLength += i;
            pos += i;
            i = 0;
            if (!fillBuffer(1)) {
                break;
            }
        }

        String result;
        if (assignOffsetsOnly && builder == null) {
            valuePos = pos;
            result = null;
        } else if (skipping) {
            result = "skipped!";
        } else if (builder == null) {
            result = stringPool.get(buffer, pos, i);
        } else {
            builder.append(buffer, pos, i);
            result = builder.toString();
        }
        valueLength += i;
        pos += i;
        return result;
    }

    @Override public String toString() {
        return getClass().getSimpleName() + " near " + getSnippet();
    }

    /**
     * Unescapes the character identified by the character or characters that
     * immediately follow a backslash. The backslash '\' should have already
     * been read. This supports both unicode escapes "u000A" and two-character
     * escapes "\n".
     *
     * @throws NumberFormatException if any unicode escape sequences are
     *     malformed.
     */
    private char readEscapeCharacter() throws IOException {
        if (pos == limit && !fillBuffer(1)) {
            throw syntaxError("Unterminated escape sequence");
        }

        char escaped = buffer[pos++];
        switch (escaped) {
            case 'u':
                if (pos + 4 > limit && !fillBuffer(4)) {
                    throw syntaxError("Unterminated escape sequence");
                }
                String hex = stringPool.get(buffer, pos, 4);
                pos += 4;
                return (char) Integer.parseInt(hex, 16);

            case 't':
                return '\t';

            case 'b':
                return '\b';

            case 'n':
                return '\n';

            case 'r':
                return '\r';

            case 'f':
                return '\f';

            case '\'':
            case '"':
            case '\\':
            default:
                return escaped;
        }
    }

    /**
     * Reads a null, boolean, numeric or unquoted string literal value.
     */
    private JsonToken readLiteral() throws IOException {
        value = nextLiteral(true);
        if (valueLength == 0) {
            throw syntaxError("Expected literal value");
        }
        token = decodeLiteral();
        if (token == JsonToken.STRING) {
          checkLenient();
        }
        return token;
    }

    /**
     * Assigns {@code nextToken} based on the value of {@code nextValue}.
     */
    private JsonToken decodeLiteral() throws IOException {
        if (valuePos == -1) {
            // it was too long to fit in the buffer so it can only be a string
            return JsonToken.STRING;
        } else if (valueLength == 4
                && ('n' == buffer[valuePos    ] || 'N' == buffer[valuePos    ])
                && ('u' == buffer[valuePos + 1] || 'U' == buffer[valuePos + 1])
                && ('l' == buffer[valuePos + 2] || 'L' == buffer[valuePos + 2])
                && ('l' == buffer[valuePos + 3] || 'L' == buffer[valuePos + 3])) {
            value = "null";
            return JsonToken.NULL;
        } else if (valueLength == 4
                && ('t' == buffer[valuePos    ] || 'T' == buffer[valuePos    ])
                && ('r' == buffer[valuePos + 1] || 'R' == buffer[valuePos + 1])
                && ('u' == buffer[valuePos + 2] || 'U' == buffer[valuePos + 2])
                && ('e' == buffer[valuePos + 3] || 'E' == buffer[valuePos + 3])) {
            value = TRUE;
            return JsonToken.BOOLEAN;
        } else if (valueLength == 5
                && ('f' == buffer[valuePos    ] || 'F' == buffer[valuePos    ])
                && ('a' == buffer[valuePos + 1] || 'A' == buffer[valuePos + 1])
                && ('l' == buffer[valuePos + 2] || 'L' == buffer[valuePos + 2])
                && ('s' == buffer[valuePos + 3] || 'S' == buffer[valuePos + 3])
                && ('e' == buffer[valuePos + 4] || 'E' == buffer[valuePos + 4])) {
            value = FALSE;
            return JsonToken.BOOLEAN;
        } else {
            value = stringPool.get(buffer, valuePos, valueLength);
            return decodeNumber(buffer, valuePos, valueLength);
        }
    }

    /**
     * Determine whether the characters is a JSON number. Numbers are of the
     * form -12.34e+56. Fractional and exponential parts are optional. Leading
     * zeroes are not allowed in the value or exponential part, but are allowed
     * in the fraction.
     */
    private JsonToken decodeNumber(char[] chars, int offset, int length) {
        int i = offset;
        int c = chars[i];

        if (c == '-') {
            c = chars[++i];
        }

        if (c == '0') {
            c = chars[++i];
        } else if (c >= '1' && c <= '9') {
            c = chars[++i];
            while (c >= '0' && c <= '9') {
                c = chars[++i];
            }
        } else {
            return JsonToken.STRING;
        }

        if (c == '.') {
            c = chars[++i];
            while (c >= '0' && c <= '9') {
                c = chars[++i];
            }
        }

        if (c == 'e' || c == 'E') {
            c = chars[++i];
            if (c == '+' || c == '-') {
                c = chars[++i];
            }
            if (c >= '0' && c <= '9') {
                c = chars[++i];
                while (c >= '0' && c <= '9') {
                    c = chars[++i];
                }
            } else {
                return JsonToken.STRING;
            }
        }

        if (i == offset + length) {
            return JsonToken.NUMBER;
        } else {
            return JsonToken.STRING;
        }
    }

    /**
     * Throws a new IO exception with the given message and a context snippet
     * with this reader's content.
     */
    private IOException syntaxError(String message) throws IOException {
        throw new MalformedJsonException(message
                + " at line " + getLineNumber() + " column " + getColumnNumber());
    }

    private CharSequence getSnippet() {
        StringBuilder snippet = new StringBuilder();
        int beforePos = Math.min(pos, 20);
        snippet.append(buffer, pos - beforePos, beforePos);
        int afterPos = Math.min(limit - pos, 20);
        snippet.append(buffer, pos, afterPos);
        return snippet;
    }
}