/** Handling of SQL arrays.
 *
 * Copyright (c) 2018, Jeroen T. Vermeulen.
 *
 * See COPYING for copyright license.  If you did not receive a file called
 * COPYING with this source code, please notify the distributor of this mistake,
 * or contact the author.
 */

// Taken from libpqxx and modified

/**
 *
 *  ArrayParser.cc
 *  An Tao
 *
 *  Copyright 2018, An Tao.  All rights reserved.
 *  https://github.com/an-tao/drogon
 *  Use of this source code is governed by a MIT license
 *  that can be found in the License file.
 *
 *  Drogon
 *
 */

#include <cassert>
#include <cstddef>
#include <cstring>
#include <utility>

#include <drogon/orm/ArrayParser.h>
#include <drogon/orm/Exception.h>

using namespace drogon::orm;

namespace
{
/// Find the end of a single-quoted SQL string in an SQL array.
/** Assumes UTF-8 or an ASCII-superset single-byte encoding.
 * This is used for array parsing, where the database may send us strings
 * stored in the array, in a quoted and escaped format.
 *
 * Returns the address of the first character after the closing quote.
 */
const char *scan_single_quoted_string(const char begin[])
{
    const char *here = begin;
    assert(*here == '\'');
    for (here++; *here; here++)
    {
        switch (*here)
        {
            case '\'':
                // Escaped quote, or closing quote.
                here++;
                // If the next character is a quote, we've got a double single
                // quote. That's how SQL escapes embedded quotes in a string.
                // Terrible idea, but it's what we have.
                if (*here != '\'')
                    return here;
                // Otherwise, we've found the end of the string.  Return the
                // address of the very next byte.
                break;
            case '\\':
                // Backslash escape.  Skip ahead by one more character.
                here++;
                if (!*here)
                    throw ArgumentError("SQL string ends in escape: " +
                                        std::string(begin));
                break;
        }
    }
    throw ArgumentError("Null byte in SQL string: " + std::string(begin));
}

/// Parse a single-quoted SQL string: un-quote it and un-escape it.
/** Assumes UTF-8 or an ASCII-superset single-byte encoding.
 */
std::string parse_single_quoted_string(const char begin[], const char end[])
{
    // There have to be at least 2 characters: the opening and closing quotes.
    assert(begin + 1 < end);
    assert(*begin == '\'');
    assert(*(end - 1) == '\'');

    std::string output;
    // Maximum output size is same as the input size, minus the opening and
    // closing quotes.  In the worst case, the real number could be half that.
    // Usually it'll be a pretty close estimate.
    output.reserve(std::size_t(end - begin - 2));
    for (const char *here = begin + 1; here < end - 1; here++)
    {
        auto c = *here;
        // Skip escapes.
        if (c == '\'' || c == '\\')
            here++;
        output.push_back(c);
    }

    return output;
}

/// Find the end of a double-quoted SQL string in an SQL array.
/** Assumes UTF-8 or an ASCII-superset single-byte encoding.
 */
const char *scan_double_quoted_string(const char begin[])
{
    const char *here = begin;
    assert(*here == '"');
    for (here++; *here; here++)
    {
        switch (*here)
        {
            case '\\':
                // Backslash escape.  Skip ahead by one more character.
                here++;
                if (!*here)
                    throw ArgumentError("SQL string ends in escape: " +
                                        std::string(begin));
                break;
            case '"':
                return here + 1;
        }
    }
    throw ArgumentError("Null byte in SQL string: " + std::string(begin));
}

/// Parse a double-quoted SQL string: un-quote it and un-escape it.
/** Assumes UTF-8 or an ASCII-superset single-byte encoding.
 */
std::string parse_double_quoted_string(const char begin[], const char end[])
{
    // There have to be at least 2 characters: the opening and closing quotes.
    assert(begin + 1 < end);
    assert(*begin == '"');
    assert(*(end - 1) == '"');

    std::string output;
    // Maximum output size is same as the input size, minus the opening and
    // closing quotes.  In the worst case, the real number could be half that.
    // Usually it'll be a pretty close estimate.
    output.reserve(std::size_t(end - begin - 2));
    for (const char *here = begin + 1; here < end - 1; here++)
    {
        auto c = *here;
        // Skip escapes.
        if (c == '\\')
            here++;
        output.push_back(c);
    }

    return output;
}

/// Find the end of an unquoted string in an SQL array.
/** Assumes UTF-8 or an ASCII-superset single-byte encoding.
 */
const char *scan_unquoted_string(const char begin[])
{
    assert(*begin != '\'');
    assert(*begin != '"');

    const char *p;
    for (p = begin; *p != ',' && *p != ';' && *p != '}'; p++)
        ;
    return p;
}

/// Parse an unquoted SQL string.
/** Here, the special unquoted value NULL means a null value, not a string
 * that happens to spell "NULL".
 */
std::string parse_unquoted_string(const char begin[], const char end[])
{
    return std::string(begin, end);
}

}  // namespace

ArrayParser::ArrayParser(const char input[]) : pos_(input)
{
}

std::pair<ArrayParser::juncture, std::string> ArrayParser::getNext()
{
    ArrayParser::juncture found;
    std::string value;
    const char *end;

    if (pos_ == nullptr)
    {
        found = juncture::done;
        end = nullptr;
    }
    else
        switch (*pos_)
        {
            case '\0':
                found = juncture::done;
                end = pos_;
                break;
            case '{':
                found = juncture::row_start;
                end = pos_ + 1;
                break;
            case '}':
                found = juncture::row_end;
                end = pos_ + 1;
                break;
            case '\'':
                found = juncture::string_value;
                end = scan_single_quoted_string(pos_);
                value = parse_single_quoted_string(pos_, end);
                break;
            case '"':
                found = juncture::string_value;
                end = scan_double_quoted_string(pos_);
                value = parse_double_quoted_string(pos_, end);
                break;
            default:
                end = scan_unquoted_string(pos_);
                value = parse_unquoted_string(pos_, end);
                if (value == "NULL")
                {
                    // In this one situation, as a special case, NULL means a
                    // null field, not a string that happens to spell "NULL".
                    value.clear();
                    found = juncture::null_value;
                }
                else
                {
                    // The normal case: we just parsed an unquoted string.  The
                    // value is what we need.
                    found = juncture::string_value;
                }
                break;
        }

    // Skip a field separator following a string (or null).
    if (end != nullptr && (*end == ',' || *end == ';'))
        end++;

    pos_ = end;
    return std::make_pair(found, value);
}