eliot/dic/header.cpp

/*****************************************************************************
 * Eliot
 * Copyright (C) 2007-2009 Olivier Teulière
 * Authors: Olivier Teulière <ipkiss @@ gmail.com>
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2 of the License, or
 * (at your option) any later version.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 *****************************************************************************/

#include "config.h"

#include <cstring> // for strcpy
#include <cstdio> // for snprintf
#include <string>
#include <sstream>
#include <iostream>
#include <boost/foreach.hpp>
#include <boost/format.hpp>
#include <boost/tokenizer.hpp>

// For ntohl & Co.
#ifdef WIN32
#   include <winsock2.h>
#else
#    if HAVE_NETINET_IN_H
#       include <netinet/in.h>
#    endif
#    if HAVE_ARPA_INET_H
#       include <arpa/inet.h>
#    endif
#endif

#if ENABLE_NLS
#   include <libintl.h>
#   define _(String) gettext(String)
#else
#   define _(String) String
#endif

#include "header.h"
#include "encoding.h"
#include "dic_exception.h"

using boost::format;
using boost::wformat;


INIT_LOGGER(dic, Header);


#if defined(WORDS_BIGENDIAN)
// Nothing to do on big-endian machines
#   define ntohll(x) (x)
#   define htonll(x) (x)
#else
static inline uint64_t htonll(uint64_t host64)
{
    return (((uint64_t)htonl((host64 << 32) >> 32)) << 32) | htonl(host64 >> 32);
}

static inline uint64_t ntohll(uint64_t net64)
{
    return htonll(net64);
}
#endif


/**
 * Keyword included in dictionary headers
 * Implies little endian storage on words
 */
#define _COMPIL_KEYWORD_ "_COMPILED_DICTIONARY_"

/** Old format of the header (i.e. version 0) */
struct Dict_header_old
{
    /// Identification string
    char ident[sizeof(_COMPIL_KEYWORD_)];
    /// Version of the serialization format
    uint8_t version;
    /// Unused at the moment, reserved for future use
    char unused;
    uint32_t root;
    uint32_t nwords;
    /// Information about the compression
    //@{
    uint32_t edgesused;
    uint32_t nodesused;
    uint32_t nodessaved;
    uint32_t edgessaved;
    //@}
};

// Do not change these values, as they impact the size of the structure!
// Note: they are chosen carefully to avoid alignment issues
#define _MAX_USER_HOST_ 32
#define _MAX_DIC_NAME_SIZE_ 30
#define _MAX_LETTERS_NB_ 63
#define _MAX_LETTERS_SIZE_ 80
#define _MAX_DISPLAY_INPUT_SIZE_ 112

/** Extension of the old format (used in version 1) */
struct Dict_header_ext
{
    // Time when the dictionary was compressed
    // (number of seconds since the Epoch)
    uint64_t compressDate;
    // Build information
    char userHost[_MAX_USER_HOST_];
    // Size taken by the build information
    uint32_t userHostSize;

    // --- we have a multiple of 64 bytes here

    // Compression algorithm (1 = DAWG, 2 = GADDAG)
    uint8_t algorithm;
    // Variant used in the rules (XXX: currently unused)
    uint8_t variant;

    // --- we have a multiple of 64 bytes here

    // Dictionary official name and version (e.g.: ODS 5.0)
    char dicName[_MAX_DIC_NAME_SIZE_];
    // Size taken by the dictionary name
    uint32_t dicNameSize;

    // --- we have a multiple of 64 bytes here

    // Letters used in the dictionary
    // We should have: nbLetters <= lettersSize <= _MAX_LETTERS_SIZE_
    // and:            nbLetters <= _MAX_LETTERS_NB_
    // The letters themselves, in UTF-8
    char letters[_MAX_LETTERS_SIZE_];
    // Size taken by the letters
    uint32_t lettersSize;
    // Number of letters (XXX: in theory useless, but allows a sanity check)
    uint32_t nbLetters;

    // --- we have a multiple of 64 bytes here

    // Points of the letters (indexed by their code)
    // The "+ 1" is there for struct alignment
    uint8_t points[_MAX_LETTERS_NB_ + 1];
    // Frequency of the letters (indexed by their code)
    // The "+ 1" is there for struct alignment
    uint8_t frequency[_MAX_LETTERS_NB_ + 1];
    // Bitfield indicating whether letters are vowels
    uint64_t vowels;
    // Bitfield indicating whether letters are consonants
    uint64_t consonants;

    // --- we have a multiple of 64 bytes here
};

/** Extension of the extension :-) (used in version 2) */
struct Dict_header_ext_2
{
    // Additional information concerning the display strings and the
    // alternative input strings of the letters
    char displayAndInput[_MAX_DISPLAY_INPUT_SIZE_];
    // Size taken by the display/input data
    uint16_t displayAndInputSize;

    // --- we have a multiple of 64 bytes here
};


Header::Header(istream &iStream)
    : m_root(0), m_nbWords(0), m_nodesUsed(0), m_edgesUsed(0),
      m_nodesSaved(0), m_edgesSaved(0), m_type(kDAWG)
{
    // Simply delegate to the read() method
    // The code is not moved here because I find it more natural to have a
    // read() method symmetrical to the write() one
    read(iStream);
    buildCaches();
}


Header::Header(const DictHeaderInfo &iInfo)
{
    // Use the latest serialization format
    m_version = 2;

    // Sanity checks
    if (iInfo.letters.size() > _MAX_LETTERS_NB_)
    {
        format fmt(_("Header::Header: Too many different letters for "
                     "the current format; only %1% are supported"));
        fmt % _MAX_LETTERS_NB_;
        throw DicException(fmt.str());
    }
    if (iInfo.points.size() != iInfo.letters.size())
    {
        throw DicException("Header::Header: Different sizes for "
                           "iInfo.points and iInfo.letters");
    }
    if (iInfo.frequency.size() != iInfo.letters.size())
    {
        throw DicException("Header::Header: Different sizes for "
                           "iInfo.frequency and iInfo.letters");
    }
    if (iInfo.vowels.size() != iInfo.letters.size())
    {
        throw DicException("Header::Header: Different sizes for "
                           "iInfo.vowels and iInfo.letters");
    }
    if (iInfo.consonants.size() != iInfo.letters.size())
    {
        throw DicException("Header::Header: Different sizes for "
                           "iInfo.consonants and iInfo.letters");
    }

    m_compressDate = time(NULL);
    m_userHost = wfl(ELIOT_COMPILE_BY + string("@") + ELIOT_COMPILE_HOST);
    m_root = iInfo.root;
    m_nbWords = iInfo.nwords;
    m_nodesUsed = iInfo.nodesused;
    m_edgesUsed = iInfo.edgesused;
    m_nodesSaved = iInfo.nodessaved;
    m_edgesSaved = iInfo.edgessaved;
    m_type = iInfo.dawg ? kDAWG : kGADDAG;
    m_dicName = iInfo.dicName;
    m_letters = iInfo.letters;
    m_points = iInfo.points;
    m_frequency = iInfo.frequency;
    m_vowels = iInfo.vowels;
    m_consonants = iInfo.consonants;
    m_displayAndInputData = iInfo.displayInputData;

    buildCaches();
}


void Header::buildCaches()
{
    // Build the char --> code mapping
    for (unsigned int i = 0; i < m_letters.size(); ++i)
    {
        // We don't differentiate uppercase and lowercase letters
        m_mapCodeFromChar[towlower(m_letters[i])] = i + 1;
        m_mapCodeFromChar[towupper(m_letters[i])] = i + 1;
    }

    // Build the display strings cache
    m_displayCache.assign(m_letters.size() + 1, L"");
    for (unsigned int i = 0; i < m_letters.size(); ++i)
    {
        map<wchar_t, vector<wstring> >::const_iterator it =
            m_displayAndInputData.find(m_letters[i]);
        if (it == m_displayAndInputData.end())
            m_displayCache[i + 1].append(1, m_letters[i]);
        else
            m_displayCache[i + 1] = it->second[0];
    }

    // Create a string with all the characters possibly used
    m_inputChars.reserve(m_letters.size());
    BOOST_FOREACH(wchar_t wch, m_letters)
    {
        map<wchar_t, vector<wstring> >::const_iterator it =
            m_displayAndInputData.find(wch);
        if (it == m_displayAndInputData.end())
            m_inputChars.append(1, wch);
        else
        {
            BOOST_FOREACH(const wstring &str, it->second)
            {
                BOOST_FOREACH(wchar_t chr, str)
                {
                    if (m_inputChars.find(towupper(chr)) == string::npos)
                        m_inputChars.append(1, towupper(chr));
                }
            }
        }
    }
}


wchar_t Header::getCharFromCode(unsigned int iCode) const
{
    // Safety check
    if (iCode == 0 || iCode > m_letters.size())
    {
        format fmt(_("Header::getCharFromCode: No letter for code '%1%'"));
        fmt % iCode;
        throw DicException(fmt.str());
    }
    return m_letters[iCode - 1];
}


unsigned int Header::getCodeFromChar(wchar_t iChar) const
{
    map<wchar_t, unsigned int>::const_iterator pair =
        m_mapCodeFromChar.find(iChar);
    if (pair == m_mapCodeFromChar.end())
    {
        char s[5];
        sprintf(s, "%d", iChar);
        format fmt(_("Header::getCodeFromChar: No code for letter '%1%' (val=%2%)"));
        fmt % lfw(iChar) % s;
        throw DicException(fmt.str());
    }
    return pair->second;
}


const wdstring & Header::getDisplayStr(unsigned int iCode) const
{
    // Safety check
    if (iCode == 0 || iCode > m_letters.size())
    {
        ostringstream oss;
        oss << iCode;
        throw DicException("Header::getDisplayStr: No code for letter '" + oss.str());
    }
    return m_displayCache[iCode];
}


vector<wistring> Header::getInputStr(unsigned int iCode) const
{
    // Safety check
    if (iCode == 0 || iCode > m_letters.size())
    {
        ostringstream oss;
        oss << iCode;
        throw DicException("Header::getInputStr: No code for letter '" + oss.str());
    }
    map<wchar_t, vector<wstring> >::const_iterator it =
        m_displayAndInputData.find(m_letters[iCode - 1]);
    if (it == m_displayAndInputData.end())
    {
        vector<wistring> vect;
        vect.push_back(m_displayCache[iCode]);
        return vect;
    }
    else
        return it->second;
}


void Header::read(istream &iStream)
{
    Dict_header_old aHeader;
    iStream.read((char*)&aHeader, sizeof(Dict_header_old));
    if (iStream.gcount() != sizeof(Dict_header_old))
        throw DicException("Header::read: expected to read more bytes");

    // Check the identification string
    if (string(aHeader.ident) != _COMPIL_KEYWORD_)
        throw DicException("Header::read: incorrect header keyword; is it a dictionary file?");

    m_version = aHeader.version;

    // Version 0 corresponds to the dictionary format in the first Eliot
    // versions, supported until Eliot 1.8 (excluded).
    // The new version (version 1) was introduced in Eliot 1.6.
    if (m_version == 0)
    {
        throw DicException(_("Too old dictionary format. This format is not "
                             "supported anymore since Eliot 1.8. You can "
                             "create dictionaries in the new format with the "
                             "'compdic' tool provided with Eliot (since "
                             "version 1.6)."));
    }

    // Handle endianness
    m_root = ntohl(aHeader.root);
    m_nbWords = ntohl(aHeader.nwords);
    m_nodesUsed = ntohl(aHeader.nodesused);
    m_edgesUsed = ntohl(aHeader.edgesused);
    m_nodesSaved = ntohl(aHeader.nodessaved);
    m_edgesSaved = ntohl(aHeader.edgessaved);

    // After reading the old header, we now read the extension
    Dict_header_ext aHeaderExt;
    iStream.read((char*)&aHeaderExt, sizeof(Dict_header_ext));
    if (iStream.gcount() != sizeof(Dict_header_ext))
        throw DicException("Header::read: expected to read more bytes");

    // Handle endianness in the extension
    aHeaderExt.compressDate = ntohll(aHeaderExt.compressDate);
    aHeaderExt.userHostSize = ntohl(aHeaderExt.userHostSize);
    aHeaderExt.dicNameSize = ntohl(aHeaderExt.dicNameSize);
    aHeaderExt.lettersSize = ntohl(aHeaderExt.lettersSize);
    aHeaderExt.nbLetters = ntohl(aHeaderExt.nbLetters);
    aHeaderExt.vowels = ntohll(aHeaderExt.vowels);
    aHeaderExt.consonants = ntohll(aHeaderExt.consonants);

    m_compressDate = aHeaderExt.compressDate;

    if (aHeaderExt.algorithm == kDAWG)
        m_type = kDAWG;
    else if (aHeaderExt.algorithm == kGADDAG)
        m_type = kGADDAG;
    else
        throw DicException("Header::read: unrecognized algorithm type");

    m_userHost = readFromUTF8(string(aHeaderExt.userHost,
                                     aHeaderExt.userHostSize),
                              "user and host information");

    // Convert the dictionary letters from UTF-8 to wchar_t*
    m_dicName = readFromUTF8(string(aHeaderExt.dicName,
                                    aHeaderExt.dicNameSize),
                             "dictionary name");

    // Convert the dictionary letters from UTF-8 to wchar_t*
    m_letters = readFromUTF8(string(aHeaderExt.letters,
                                    aHeaderExt.lettersSize),
                             "dictionary letters");
    // Safety check: correct number of letters?
    if (m_letters.size() != aHeaderExt.nbLetters)
    {
        throw DicException("Header::read: inconsistent header");
    }

    // Letters points and frequency
    for (unsigned int i = 0; i < m_letters.size(); ++i)
    {
        m_points.push_back(aHeaderExt.points[i]);
        m_frequency.push_back(aHeaderExt.frequency[i]);
    }

    // Vowels and consonants
    for (unsigned int i = 0; i < m_letters.size(); ++i)
    {
        m_vowels.push_back(aHeaderExt.vowels & (1 << i));
        m_consonants.push_back(aHeaderExt.consonants & (1 << i));
    }

    // Read the additional display/input data
    if (m_version >= 2)
    {
        // Read the extension of the extension...
        Dict_header_ext_2 aHeaderExt2;
        iStream.read((char*)&aHeaderExt2, sizeof(Dict_header_ext_2));
        if (iStream.gcount() != sizeof(Dict_header_ext_2))
            throw DicException("Header::read: expected to read more bytes (ext2)");

        // Handle endianness
        aHeaderExt2.displayAndInputSize = ntohs(aHeaderExt2.displayAndInputSize);

        // Convert the dictionary letters from UTF-8 to wchar_t*
        wstring serialized = readFromUTF8(string(aHeaderExt2.displayAndInput,
                                                 aHeaderExt2.displayAndInputSize),
                                          "display and input data");
        // Parse this string and structure the data
        readDisplayAndInput(serialized);
    }
}


void Header::write(ostream &oStream) const
{
    Dict_header_old aHeader;
    strcpy(aHeader.ident, _COMPIL_KEYWORD_);
    aHeader.version = m_version;
    aHeader.unused = 0;
    aHeader.root = htonl(m_root);
    aHeader.nwords = htonl(m_nbWords);
    aHeader.nodesused = htonl(m_nodesUsed);
    aHeader.edgesused = htonl(m_edgesUsed);
    aHeader.nodessaved = htonl(m_nodesSaved);
    aHeader.edgessaved = htonl(m_edgesSaved);

    oStream.write((char*)&aHeader, sizeof(Dict_header_old));
    if (!oStream.good())
        throw DicException("Header::write: error when writing to file");

    Dict_header_ext aHeaderExt;
    aHeaderExt.compressDate = m_compressDate;
    aHeaderExt.userHostSize =
        writeInUTF8(m_userHost, aHeaderExt.userHost,
                    _MAX_USER_HOST_, "user and host information");
    aHeaderExt.algorithm = m_type;

    // Convert the dictionary name to UTF-8
    aHeaderExt.dicNameSize =
        writeInUTF8(m_dicName, aHeaderExt.dicName,
                    _MAX_DIC_NAME_SIZE_, "dictionary name");

    // Convert the dictionary letters to UTF-8
    aHeaderExt.lettersSize =
        writeInUTF8(m_letters, aHeaderExt.letters,
                    _MAX_LETTERS_SIZE_, "dictionary letters");
    aHeaderExt.nbLetters = (uint32_t)m_letters.size();

    // Letters points and frequency
    for (unsigned int i = 0; i < m_letters.size(); ++i)
    {
        aHeaderExt.points[i] = m_points[i];
        aHeaderExt.frequency[i] = m_frequency[i];
    }

    // Vowels and consonants
    aHeaderExt.vowels = 0;
    aHeaderExt.consonants = 0;
    for (unsigned int i = 0; i < m_letters.size(); ++i)
    {
        if (m_vowels[i])
            aHeaderExt.vowels |= 1 << i;
        if (m_consonants[i])
            aHeaderExt.consonants |= 1 << i;
    }

    // Handle endianness in the extension
    aHeaderExt.userHostSize = htonl(aHeaderExt.userHostSize);
    aHeaderExt.compressDate = htonll(aHeaderExt.compressDate);
    aHeaderExt.dicNameSize = htonl(aHeaderExt.dicNameSize);
    aHeaderExt.lettersSize = htonl(aHeaderExt.lettersSize);
    aHeaderExt.nbLetters = htonl(aHeaderExt.nbLetters);
    aHeaderExt.vowels = htonll(aHeaderExt.vowels);
    aHeaderExt.consonants = htonll(aHeaderExt.consonants);

    // Write the extension
    oStream.write((char*)&aHeaderExt, sizeof(Dict_header_ext));
    if (!oStream.good())
        throw DicException("Header::write: error when writing to file");

    // Write the second extension
    Dict_header_ext_2 aHeaderExt2;
    const wstring &serialized = writeDisplayAndInput();

    // Convert the serialized data to UTF-8
    aHeaderExt2.displayAndInputSize =
        writeInUTF8(serialized, aHeaderExt2.displayAndInput,
                    _MAX_DISPLAY_INPUT_SIZE_, "display and input data");

    // Handle endianness
    aHeaderExt2.displayAndInputSize = htons(aHeaderExt2.displayAndInputSize);

    // Write the extension
    oStream.write((char*)&aHeaderExt2, sizeof(Dict_header_ext_2));
    if (!oStream.good())
        throw DicException("Header::write: error when writing to file (ext2)");
}


void Header::readDisplayAndInput(const wstring &serialized)
{
    // The format is the following:
    // "X|DISPX|INP1|INP2|INP3 Y|DISPY Z|DISPZ|INP4|INP5"
    // where X, Y and Z are internal chars (i.e. chars of m_letters),
    // DISPX, DISPY and DISPZ are the corresponding display strings,
    // and the INP* are input strings (in addition to the display string,
    // which is always considered as an input string)

    // Use a more friendly type name
    typedef boost::tokenizer<boost::char_separator<wchar_t>,
            std::wstring::const_iterator,
            std::wstring> Tokenizer;

    // Split the string on double spaces
    static const boost::char_separator<wchar_t> sep1(L" ");
    static const boost::char_separator<wchar_t> sep2(L"|");
    Tokenizer tok(serialized, sep1);
    Tokenizer::iterator it;
    for (it = tok.begin(); it != tok.end(); ++it)
    {
        // Split the token on single space
        Tokenizer tok2(*it, sep2);
        vector<wstring> pieces(tok2.begin(), tok2.end());
        // Some sanity checks...
        if (pieces.size() < 2)
            throw DicException("Header::readDisplayAndInput: no display "
                               "string. Corrupted dictionary?");
        // The first piece must be a single char, present in m_letters
        if (pieces[0].size() != 1 ||
            m_letters.find(pieces[0][0]) == wstring::npos)
        {
            throw DicException("Header::readDisplayAndInput: invalid internal"
                               " letter. Corrupted dictionary?");
        }
        wchar_t chr = pieces[0][0];
        if (m_displayAndInputData.find(chr) != m_displayAndInputData.end())
        {
            throw DicException("Header::readDisplayAndInput: found 2 display"
                               " data sections for the same letter. Corrupted"
                               " dictionary?");
        }
        // OK, save the data
        pieces.erase(pieces.begin());
        m_displayAndInputData[chr] = pieces;
    }
}


wstring Header::writeDisplayAndInput() const
{
    wstring serialized;
    bool first = true;
    map<wchar_t, vector<wstring> >::const_iterator it;
    for (it = m_displayAndInputData.begin();
         it != m_displayAndInputData.end(); ++it)
    {
        if (first)
            first = false;
        else
            serialized += L" ";
        serialized.append(1, it->first);
        BOOST_FOREACH(const wstring &str, it->second)
        {
            // Make sure the string is uppercase
            wstring upStr = str;
            std::transform(upStr.begin(), upStr.end(), upStr.begin(), towupper);
            serialized += L"|" + upStr;
        }
    }
    return serialized;
}


void Header::print() const
{
#define fmt(x) boost::format(x)
    cout << fmt(_("Dictionary name: %1%")) % lfw(m_dicName) << endl;
    char buf[150];
    strftime(buf, sizeof(buf), "%c", gmtime(&m_compressDate));
    cout << fmt(_("Compressed on: %1%")) % buf << endl;
    cout << fmt(_("Compressed using a binary compiled by: %1%")) % lfw(m_userHost) << endl;
    cout << fmt(_("Dictionary type: %1%")) % (m_type == kDAWG ? "DAWG" : "GADDAG") << endl;
    cout << fmt(_("Letters: %1%")) % lfw(m_letters) << endl;
    cout << fmt(_("Number of letters: %1%")) % m_letters.size() << endl;
    cout << fmt(_("Number of words: %1%")) % m_nbWords << endl;
    long unsigned int size = sizeof(Dict_header_old) +
        sizeof(Dict_header_ext) + sizeof(Dict_header_ext_2);
    cout << fmt(_("Header size: %1% bytes")) % size << endl;
    cout << fmt(_("Root: %1% (edge)")) % m_root << endl;
    cout << fmt(_("Nodes: %1% used + %2% saved")) % m_nodesUsed % m_nodesSaved << endl;
    cout << fmt(_("Edges: %1% used + %2% saved")) % m_edgesUsed % m_edgesSaved << endl;
#undef fmt
    cout << "====================================================================" << endl;
    cout << format("%1% | %2% | %3% | %4% | %5% | %6% | %7%")
        % _("Letter") % _("Points") % _("Frequency") % _("Vowel")
        % _("Consonant") % _("Display") % _("Alt. input") << endl;
    cout << "-------+--------+-----------+-------+-----------+-------+------" << endl;
#define sz(x) strlen(_(x))
    for (unsigned int i = 0; i < m_letters.size(); ++i)
    {
        format fmter("%1% | %2% | %3% | %4% | %5% | %6% | %7%");
        fmter % centerAndConvert(wstring(1, m_letters[i]), sz("Letter"));
        fmter % centerAndConvert(str(wformat(L"%1%") % m_points[i]), sz("Points"));
        fmter % centerAndConvert(str(wformat(L"%1%") % m_frequency[i]), sz("Frequency"));
        fmter % centerAndConvert(str(wformat(L"%1%") % m_vowels[i]), sz("Vowel"));
        fmter % centerAndConvert(str(wformat(L"%1%") % m_consonants[i]), sz("Consonant"));
        map<wchar_t, vector<wstring> >::const_iterator it =
            m_displayAndInputData.find(m_letters[i]);
        if (it != m_displayAndInputData.end())
        {
            const vector<wstring> &inputs = it->second;
            fmter % centerAndConvert(str(wformat(L"%1%") % inputs[0]), sz("Display"));
            bool first = true;
            string s;
            for (uint8_t j = 1; j < inputs.size(); ++j)
            {
                if (first)
                    first = false;
                else
                    s += " ";
                s += lfw(inputs[j]);
            }
            fmter % s;
        }
        else
        {
            fmter % string(sz("Display"), ' ') % string(sz("Alt. input"), ' ');
        }
        cout << fmter.str() << endl;
    }
#undef sz
    cout << "====================================================================" << endl;
}