/******************************************************************************

 *

 * Project:  Common Portability Library

 * Purpose:  Implementation of CPLKeywordParser - a class for parsing

 *           the keyword format used for files like QuickBird .RPB files.

 *           This is a slight variation on the NASAKeywordParser used for

 *           the PDS/ISIS2/ISIS3 formats.

 * Author:   Frank Warmerdam <warmerdam@pobox.com

 *

 ******************************************************************************

 * Copyright (c) 2008, Frank Warmerdam <warmerdam@pobox.com>

 * Copyright (c) 2009-2010, Even Rouault <even dot rouault at spatialys.com>

 *

 * SPDX-License-Identifier: MIT

 ****************************************************************************/

//! @cond Doxygen_Suppress

#include "cpl_port.h"

#include "cplkeywordparser.h"

#include <cctype>

#include <cstring>

#include <string>

#include "cpl_string.h"

#include "cpl_vsi.h"

/************************************************************************/

/* ==================================================================== */

/*                          CPLKeywordParser                           */

/* ==================================================================== */

/************************************************************************/

/************************************************************************/

/*                          CPLKeywordParser()                          */

/************************************************************************/

CPLKeywordParser::CPLKeywordParser() = default;

/************************************************************************/

/*                         ~CPLKeywordParser()                          */

/************************************************************************/

CPLKeywordParser::~CPLKeywordParser()

{

    CSLDestroy(papszKeywordList);

    papszKeywordList = nullptr;

}

/************************************************************************/

/*                               Ingest()                               */

/************************************************************************/

int CPLKeywordParser::Ingest(VSILFILE *fp)

{

    /* -------------------------------------------------------------------- */

    /*      Read in buffer till we find END all on its own line.            */

    /* -------------------------------------------------------------------- */

    for (; true;)

    {

        char szChunk[513] = {};

        const size_t nBytesRead = VSIFReadL(szChunk, 1, 512, fp);

        szChunk[nBytesRead] = '\0';

        osHeaderText += szChunk;

        if (nBytesRead < 512)

            break;

        const char *pszCheck = nullptr;

        if (osHeaderText.size() > 520)

            pszCheck = osHeaderText.c_str() + (osHeaderText.size() - 520);

        else

            pszCheck = szChunk;

        if (strstr(pszCheck, "\r\nEND;\r\n") != nullptr ||

            strstr(pszCheck, "\nEND;\n") != nullptr)

            break;

    }

    pszHeaderNext = osHeaderText.c_str();

    /* -------------------------------------------------------------------- */

    /*      Process name/value pairs, keeping track of a "path stack".      */

    /* -------------------------------------------------------------------- */

    return ReadGroup("", 0);

}

/************************************************************************/

/*                             ReadGroup()                              */

/************************************************************************/

bool CPLKeywordParser::ReadGroup(const char *pszPathPrefix, int nRecLevel)

{

    CPLString osName;

    CPLString osValue;

    // Arbitrary threshold to avoid stack overflow

    if (nRecLevel == 100)

        return false;

    for (; true;)

    {

        if (!ReadPair(osName, osValue))

            return false;

        if (EQUAL(osName, "BEGIN_GROUP") || EQUAL(osName, "GROUP"))

        {

            if (!ReadGroup((CPLString(pszPathPrefix) + osValue + ".").c_str(),

                           nRecLevel + 1))

                return false;

        }

        else if (STARTS_WITH_CI(osName, "END"))

        {

            return true;

        }

        else

        {

            osName = pszPathPrefix + osName;

            papszKeywordList =

                CSLSetNameValue(papszKeywordList, osName, osValue);

        }

    }

}

/************************************************************************/

/*                              ReadPair()                              */

/*                                                                      */

/*      Read a name/value pair from the input stream.  Strip off        */

/*      white space, ignore comments, split on '='.                     */

/************************************************************************/

bool CPLKeywordParser::ReadPair(CPLString &osName, CPLString &osValue)

{

    osName = "";

    osValue = "";

    if (!ReadWord(osName))

        return false;

    SkipWhite();

    if (EQUAL(osName, "END"))

        return TRUE;

    if (*pszHeaderNext != '=')

    {

        // ISIS3 does not have anything after the end group/object keyword.

        return EQUAL(osName, "End_Group") || EQUAL(osName, "End_Object");

    }

    pszHeaderNext++;

    SkipWhite();

    osValue = "";

    // Handle value lists like:     Name   = (Red, Red)

    // or list of lists like: TLCList = ( (0, 0.000000), (8299, 4.811014) );

    if (*pszHeaderNext == '(')

    {

        CPLString osWord;

        int nDepth = 0;

        const char *pszLastPos = pszHeaderNext;

        while (ReadWord(osWord) && pszLastPos != pszHeaderNext)

        {

            SkipWhite();

            pszLastPos = pszHeaderNext;

            osValue += osWord;

            const char *pszIter = osWord.c_str();

            bool bInQuote = false;

            while (*pszIter != '\0')

            {

                if (*pszIter == '"')

                    bInQuote = !bInQuote;

                else if (!bInQuote)

                {

                    if (*pszIter == '(')

                        nDepth++;

                    else if (*pszIter == ')')

                    {

                        nDepth--;

                        if (nDepth == 0)

                            break;

                    }

                }

                pszIter++;

            }

            if (*pszIter == ')' && nDepth == 0)

                break;

        }

    }

    else  // Handle more normal "single word" values.

    {

        // Special case to handle non-conformant IMD files generated by

        // previous GDAL version where we omit to surround values that have

        // spaces with double quotes.

        // So we use a heuristics to handle things like:

        //       key = value with spaces without single or double quotes at

        //       beginning of value;[\r]\n

        const char *pszNextLF = strchr(pszHeaderNext, '\n');

        if (pszNextLF)

        {

            std::string osTxt(pszHeaderNext, pszNextLF - pszHeaderNext);

            const auto nCRPos = osTxt.find('\r');

            const auto nSemiColonPos = osTxt.find(';');

            const auto nQuotePos = osTxt.find('\'');

            const auto nDoubleQuotePos = osTxt.find('"');

            const auto nLTPos = osTxt.find('<');

            if (nSemiColonPos != std::string::npos &&

                (nCRPos == std::string::npos || (nCRPos + 1 == osTxt.size())) &&

                ((nCRPos != std::string::npos &&

                  (nSemiColonPos + 1 == nCRPos)) ||

                 (nCRPos == std::string::npos &&

                  (nSemiColonPos + 1 == osTxt.size()))) &&

                (nQuotePos == std::string::npos || nQuotePos != 0) &&

                (nDoubleQuotePos == std::string::npos ||

                 nDoubleQuotePos != 0) &&

                (nLTPos == std::string::npos ||

                 osTxt.find('>') == std::string::npos))

            {

                pszHeaderNext = pszNextLF;

                osTxt.resize(nSemiColonPos);

                osValue = osTxt;

                while (!osValue.empty() && osValue.back() == ' ')

                    osValue.pop_back();

                return true;

            }

        }

        if (!ReadWord(osValue))

            return false;

    }

    SkipWhite();

    // No units keyword?

    if (*pszHeaderNext != '<')

        return true;

    // Append units keyword.  For lines that like like this:

    //  MAP_RESOLUTION               = 4.0 <PIXEL/DEGREE>

    CPLString osWord;

    osValue += " ";

    while (ReadWord(osWord))

    {

        SkipWhite();

        osValue += osWord;

        if (osWord.back() == '>')

            break;

    }

    return true;

}

/************************************************************************/

/*                              ReadWord()                              */

/************************************************************************/

bool CPLKeywordParser::ReadWord(CPLString &osWord)

{

    osWord = "";

    SkipWhite();

    if (*pszHeaderNext == '\0' || *pszHeaderNext == '=')

        return false;

    while (*pszHeaderNext != '\0' && *pszHeaderNext != '=' &&

           *pszHeaderNext != ';' &&

           !isspace(static_cast<unsigned char>(*pszHeaderNext)))

    {

        if (*pszHeaderNext == '"')

        {

            osWord += *(pszHeaderNext++);

            while (*pszHeaderNext != '"')

            {

                if (*pszHeaderNext == '\0')

                    return false;

                osWord += *(pszHeaderNext++);

            }

            osWord += *(pszHeaderNext++);

        }

        else if (*pszHeaderNext == '\'')

        {

            osWord += *(pszHeaderNext++);

            while (*pszHeaderNext != '\'')

            {

                if (*pszHeaderNext == '\0')

                    return false;

                osWord += *(pszHeaderNext++);

            }

            osWord += *(pszHeaderNext++);

        }

        else

        {

            osWord += *pszHeaderNext;

            pszHeaderNext++;

        }

    }

    if (*pszHeaderNext == ';')

        pszHeaderNext++;

    return true;

}

/************************************************************************/

/*                             SkipWhite()                              */

/************************************************************************/

void CPLKeywordParser::SkipWhite()

{

    for (; true;)

    {

        // Skip white space (newline, space, tab, etc )

        if (isspace(static_cast<unsigned char>(*pszHeaderNext)))

        {

            pszHeaderNext++;

            continue;

        }

        // Skip C style comments

        if (*pszHeaderNext == '/' && pszHeaderNext[1] == '*')

        {

            pszHeaderNext += 2;

            while (*pszHeaderNext != '\0' &&

                   (*pszHeaderNext != '*' || pszHeaderNext[1] != '/'))

            {

                pszHeaderNext++;

            }

            if (*pszHeaderNext == '\0')

                break;

            pszHeaderNext += 2;

            continue;

        }

        // Skip # style comments

        if (*pszHeaderNext == '#')

        {

            pszHeaderNext += 1;

            // consume till end of line.

            while (*pszHeaderNext != '\0' && *pszHeaderNext != 10 &&

                   *pszHeaderNext != 13)

            {

                pszHeaderNext++;

            }

            continue;

        }

        // not white space, return.

        return;

    }

}

/************************************************************************/

/*                             GetKeyword()                             */

/************************************************************************/

const char *CPLKeywordParser::GetKeyword(const char *pszPath,

                                         const char *pszDefault)

{

    const char *pszResult = CSLFetchNameValue(papszKeywordList, pszPath);

    if (pszResult == nullptr)

        return pszDefault;

    return pszResult;

}

//! @endcond