#ifdef HAVE_CONFIG_H
#include "config.h"
#endif

#include <algorithm>
#include <cctype>
#include <cstring>
#include <stdexcept>

#include <glib/gstdio.h>
#include <sys/stat.h>
#include <zlib.h>

#include "distance.hpp"
#include "mapfile.hpp"
#include "utils.hpp"

#include "stardict_lib.hpp"

#define TO_STR2(xstr) #xstr
#define TO_STR1(xstr) TO_STR2(xstr)

#define THROW_IF_ERROR(expr)                                                                 \
    do {                                                                                     \
        assert((expr));                                                                      \
        if (!(expr))                                                                         \
            throw std::runtime_error(#expr " not true at " __FILE__ ": " TO_STR1(__LINE__)); \
    } while (false)

// Notice: read src/tools/DICTFILE_FORMAT for the dictionary
// file's format information!

namespace
{
struct Fuzzystruct {
    char *pMatchWord;
    int iMatchWordDistance;
};

static inline bool bIsVowel(gchar inputchar)
{
    gchar ch = g_ascii_toupper(inputchar);
    return (ch == 'A' || ch == 'E' || ch == 'I' || ch == 'O' || ch == 'U');
}

static bool bIsPureEnglish(const gchar *str)
{
    // i think this should work even when it is UTF8 string :).
    for (int i = 0; str[i] != 0; i++)
        // if(str[i]<0)
        // if(str[i]<32 || str[i]>126) // tab equal 9,so this is not OK.
        //  Better use isascii() but not str[i]<0 while char is default unsigned in arm
        if (!isascii(str[i]))
            return false;
    return true;
}

static inline gint stardict_strcmp(const gchar *s1, const gchar *s2)
{
    const gint a = g_ascii_strcasecmp(s1, s2);
    if (a == 0)
        return strcmp(s1, s2);
    else
        return a;
}

static void unicode_strdown(gunichar *str)
{
    while (*str) {
        *str = g_unichar_tolower(*str);
        ++str;
    }
}
} // namespace

bool DictInfo::load_from_ifo_file(const std::string &ifofilename,
                                  bool istreedict)
{
    ifo_file_name = ifofilename;
    glib::CharStr buffer;
    if (!g_file_get_contents(ifofilename.c_str(), get_addr(buffer), nullptr, nullptr))
        return false;

    static const char TREEDICT_MAGIC_DATA[] = "StarDict's treedict ifo file";
    static const char DICT_MAGIC_DATA[] = "StarDict's dict ifo file";

    const gchar *magic_data = istreedict ? TREEDICT_MAGIC_DATA : DICT_MAGIC_DATA;
    static const unsigned char utf8_bom[] = { 0xEF, 0xBB, 0xBF, '\0' };
    if (!g_str_has_prefix(
            g_str_has_prefix(get_impl(buffer), (const gchar *)(utf8_bom)) ? get_impl(buffer) + 3 : get_impl(buffer),
            magic_data)) {
        return false;
    }

    gchar *p1 = get_impl(buffer) + strlen(magic_data) - 1;

    gchar *p2 = strstr(p1, "\nwordcount=");
    if (p2 == nullptr)
        return false;

    gchar *p3 = strchr(p2 + sizeof("\nwordcount=") - 1, '\n');

    wordcount = atol(std::string(p2 + sizeof("\nwordcount=") - 1, p3 - (p2 + sizeof("\nwordcount=") - 1)).c_str());

    if (istreedict) {
        p2 = strstr(p1, "\ntdxfilesize=");
        if (p2 == nullptr)
            return false;

        p3 = strchr(p2 + sizeof("\ntdxfilesize=") - 1, '\n');

        index_file_size = atol(std::string(p2 + sizeof("\ntdxfilesize=") - 1, p3 - (p2 + sizeof("\ntdxfilesize=") - 1)).c_str());

    } else {

        p2 = strstr(p1, "\nidxfilesize=");
        if (p2 == nullptr)
            return false;

        p3 = strchr(p2 + sizeof("\nidxfilesize=") - 1, '\n');
        index_file_size = atol(std::string(p2 + sizeof("\nidxfilesize=") - 1, p3 - (p2 + sizeof("\nidxfilesize=") - 1)).c_str());
    }

    p2 = strstr(p1, "\nbookname=");

    if (p2 == nullptr)
        return false;

    p2 = p2 + sizeof("\nbookname=") - 1;
    p3 = strchr(p2, '\n');
    bookname.assign(p2, p3 - p2);

    p2 = strstr(p1, "\nauthor=");
    if (p2) {
        p2 = p2 + sizeof("\nauthor=") - 1;
        p3 = strchr(p2, '\n');
        author.assign(p2, p3 - p2);
    }

    p2 = strstr(p1, "\nemail=");
    if (p2) {
        p2 = p2 + sizeof("\nemail=") - 1;
        p3 = strchr(p2, '\n');
        email.assign(p2, p3 - p2);
    }

    p2 = strstr(p1, "\nwebsite=");
    if (p2) {
        p2 = p2 + sizeof("\nwebsite=") - 1;
        p3 = strchr(p2, '\n');
        website.assign(p2, p3 - p2);
    }

    p2 = strstr(p1, "\ndate=");
    if (p2) {
        p2 = p2 + sizeof("\ndate=") - 1;
        p3 = strchr(p2, '\n');
        date.assign(p2, p3 - p2);
    }

    p2 = strstr(p1, "\ndescription=");
    if (p2) {
        p2 = p2 + sizeof("\ndescription=") - 1;
        p3 = strchr(p2, '\n');
        description.assign(p2, p3 - p2);
    }

    p2 = strstr(p1, "\nsametypesequence=");
    if (p2) {
        p2 += sizeof("\nsametypesequence=") - 1;
        p3 = strchr(p2, '\n');
        sametypesequence.assign(p2, p3 - p2);
    }

    p2 = strstr(p1, "\nsynwordcount=");
    syn_wordcount = 0;
    if (p2) {
        p2 += sizeof("\nsynwordcount=") - 1;
        p3 = strchr(p2, '\n');
        syn_wordcount = atol(std::string(p2, p3 - p2).c_str());
    }

    return true;
}

gchar *DictBase::GetWordData(guint32 idxitem_offset, guint32 idxitem_size)
{
    for (int i = 0; i < WORDDATA_CACHE_NUM; i++)
        if (cache[i].data && cache[i].offset == idxitem_offset)
            return cache[i].data;

    if (dictfile)
        fseek(dictfile, idxitem_offset, SEEK_SET);

    gchar *data;
    if (!sametypesequence.empty()) {
        glib::CharStr origin_data((gchar *)g_malloc(idxitem_size));

        if (dictfile) {
            const size_t nitems = fread(get_impl(origin_data), idxitem_size, 1, dictfile);
            THROW_IF_ERROR(nitems == 1);
        } else
            dictdzfile->read(get_impl(origin_data), idxitem_offset, idxitem_size);

        guint32 data_size;
        gint sametypesequence_len = sametypesequence.length();
        // there have sametypesequence_len char being omitted.
        data_size = idxitem_size + sizeof(guint32) + sametypesequence_len;
        // if the last item's size is determined by the end up '\0',then +=sizeof(gchar);
        // if the last item's size is determined by the head guint32 type data,then +=sizeof(guint32);
        switch (sametypesequence[sametypesequence_len - 1]) {
        case 'm':
        case 't':
        case 'y':
        case 'l':
        case 'g':
        case 'x':
        case 'k':
            data_size += sizeof(gchar);
            break;
        case 'W':
        case 'P':
            data_size += sizeof(guint32);
            break;
        default:
            if (g_ascii_isupper(sametypesequence[sametypesequence_len - 1]))
                data_size += sizeof(guint32);
            else
                data_size += sizeof(gchar);
            break;
        }
        data = (gchar *)g_malloc(data_size);
        gchar *p1, *p2;
        p1 = data + sizeof(guint32);
        p2 = get_impl(origin_data);
        guint32 sec_size;
        // copy the head items.
        for (int i = 0; i < sametypesequence_len - 1; i++) {
            *p1 = sametypesequence[i];
            p1 += sizeof(gchar);
            switch (sametypesequence[i]) {
            case 'm':
            case 't':
            case 'y':
            case 'l':
            case 'g':
            case 'x':
            case 'k':
                sec_size = strlen(p2) + 1;
                memcpy(p1, p2, sec_size);
                p1 += sec_size;
                p2 += sec_size;
                break;
            case 'W':
            case 'P':
                sec_size = get_uint32(p2);
                sec_size += sizeof(guint32);
                memcpy(p1, p2, sec_size);
                p1 += sec_size;
                p2 += sec_size;
                break;
            default:
                if (g_ascii_isupper(sametypesequence[i])) {
                    sec_size = get_uint32(p2);
                    sec_size += sizeof(guint32);
                } else {
                    sec_size = strlen(p2) + 1;
                }
                memcpy(p1, p2, sec_size);
                p1 += sec_size;
                p2 += sec_size;
                break;
            }
        }
        // calculate the last item 's size.
        sec_size = idxitem_size - (p2 - get_impl(origin_data));
        *p1 = sametypesequence[sametypesequence_len - 1];
        p1 += sizeof(gchar);
        switch (sametypesequence[sametypesequence_len - 1]) {
        case 'm':
        case 't':
        case 'y':
        case 'l':
        case 'g':
        case 'x':
        case 'k':
            memcpy(p1, p2, sec_size);
            p1 += sec_size;
            *p1 = '\0'; // add the end up '\0';
            break;
        case 'W':
        case 'P':
            set_uint32(p1, sec_size);
            p1 += sizeof(guint32);
            memcpy(p1, p2, sec_size);
            break;
        default:
            if (g_ascii_isupper(sametypesequence[sametypesequence_len - 1])) {
                set_uint32(p1, sec_size);
                p1 += sizeof(guint32);
                memcpy(p1, p2, sec_size);
            } else {
                memcpy(p1, p2, sec_size);
                p1 += sec_size;
                *p1 = '\0';
            }
            break;
        }
        set_uint32(data, data_size);
    } else {
        data = (gchar *)g_malloc(idxitem_size + sizeof(guint32));
        if (dictfile) {
            const size_t nitems = fread(data + sizeof(guint32), idxitem_size, 1, dictfile);
            THROW_IF_ERROR(nitems == 1);
        } else
            dictdzfile->read(data + sizeof(guint32), idxitem_offset, idxitem_size);
        set_uint32(data, idxitem_size + sizeof(guint32));
    }
    g_free(cache[cache_cur].data);

    cache[cache_cur].data = data;
    cache[cache_cur].offset = idxitem_offset;
    cache_cur++;
    if (cache_cur == WORDDATA_CACHE_NUM)
        cache_cur = 0;
    return data;
}

bool DictBase::SearchData(std::vector<std::string> &SearchWords, guint32 idxitem_offset, guint32 idxitem_size, gchar *origin_data)
{
    int nWord = SearchWords.size();
    std::vector<bool> WordFind(nWord, false);
    int nfound = 0;

    if (dictfile)
        fseek(dictfile, idxitem_offset, SEEK_SET);
    THROW_IF_ERROR(origin_data != nullptr);
    if (dictfile) {
        const size_t nitems = fread(origin_data, idxitem_size, 1, dictfile);
        THROW_IF_ERROR(nitems == 1);
    } else
        dictdzfile->read(origin_data, idxitem_offset, idxitem_size);
    gchar *p = origin_data;
    guint32 sec_size;
    int j;
    if (!sametypesequence.empty()) {
        gint sametypesequence_len = sametypesequence.length();
        for (int i = 0; i < sametypesequence_len - 1; i++) {
            switch (sametypesequence[i]) {
            case 'm':
            case 't':
            case 'y':
            case 'l':
            case 'g':
            case 'x':
            case 'k':
                for (j = 0; j < nWord; j++)
                    if (!WordFind[j] && strstr(p, SearchWords[j].c_str())) {
                        WordFind[j] = true;
                        ++nfound;
                    }

                if (nfound == nWord)
                    return true;
                sec_size = strlen(p) + 1;
                p += sec_size;
                break;
            default:
                if (g_ascii_isupper(sametypesequence[i])) {
                    sec_size = get_uint32(p);
                    sec_size += sizeof(guint32);
                } else {
                    sec_size = strlen(p) + 1;
                }
                p += sec_size;
            }
        }
        switch (sametypesequence[sametypesequence_len - 1]) {
        case 'm':
        case 't':
        case 'y':
        case 'l':
        case 'g':
        case 'x':
        case 'k':
            sec_size = idxitem_size - (p - origin_data);
            for (j = 0; j < nWord; j++)
                if (!WordFind[j] && g_strstr_len(p, sec_size, SearchWords[j].c_str())) {
                    WordFind[j] = true;
                    ++nfound;
                }

            if (nfound == nWord)
                return true;
            break;
        }
    } else {
        while (guint32(p - origin_data) < idxitem_size) {
            switch (*p) {
            case 'm':
            case 't':
            case 'y':
            case 'l':
            case 'g':
            case 'x':
            case 'k':
                for (j = 0; j < nWord; j++)
                    if (!WordFind[j] && strstr(p, SearchWords[j].c_str())) {
                        WordFind[j] = true;
                        ++nfound;
                    }

                if (nfound == nWord)
                    return true;
                sec_size = strlen(p) + 1;
                p += sec_size;
                break;
            default:
                if (g_ascii_isupper(*p)) {
                    sec_size = get_uint32(p);
                    sec_size += sizeof(guint32);
                } else {
                    sec_size = strlen(p) + 1;
                }
                p += sec_size;
            }
        }
    }
    return false;
}

namespace
{
class OffsetIndex : public IIndexFile
{
public:
    OffsetIndex()
        : idxfile(nullptr)
    {
    }
    ~OffsetIndex()
    {
        if (idxfile)
            fclose(idxfile);
    }
    bool load(const std::string &url, gulong wc, gulong fsize, bool verbose) override;
    const gchar *get_key(glong idx) override;
    void get_data(glong idx) override { get_key(idx); }
    const gchar *get_key_and_data(glong idx) override
    {
        return get_key(idx);
    }
    bool lookup(const char *str, std::set<glong> &idxs, glong &next_idx) override;

private:
    static const gint ENTR_PER_PAGE = 32;
    static const char *CACHE_MAGIC;

    std::vector<guint32> wordoffset;
    FILE *idxfile;
    gulong wordcount;

    gchar wordentry_buf[256 + sizeof(guint32) * 2]; // The length of "word_str" should be less than 256. See src/tools/DICTFILE_FORMAT.
    struct index_entry {
        glong idx;
        std::string keystr;
        void assign(glong i, const std::string &str)
        {
            idx = i;
            keystr.assign(str);
        }
    };
    index_entry first, last, middle, real_last;

    struct page_entry {
        gchar *keystr;
        guint32 off, size;
    };
    std::vector<gchar> page_data;
    struct page_t {
        glong idx = -1;
        page_entry entries[ENTR_PER_PAGE];

        page_t() {}
        void fill(gchar *data, gint nent, glong idx_);
    } page;
    gulong load_page(glong page_idx);
    const gchar *read_first_on_page_key(glong page_idx);
    const gchar *get_first_on_page_key(glong page_idx);
    bool load_cache(const std::string &url);
    bool save_cache(const std::string &url, bool verbose);
    static std::list<std::string> get_cache_variant(const std::string &url);
};

const char *OffsetIndex::CACHE_MAGIC = "StarDict's Cache, Version: 0.2";
#define CACHE_MAGIC_BYTES 0x51a4d1c1

class WordListIndex : public IIndexFile
{
public:
    WordListIndex()
        : idxdatabuf(nullptr)
    {
    }
    ~WordListIndex() { g_free(idxdatabuf); }
    bool load(const std::string &url, gulong wc, gulong fsize, bool verbose) override;
    const gchar *get_key(glong idx) override { return wordlist[idx]; }
    void get_data(glong idx) override;
    const gchar *get_key_and_data(glong idx) override
    {
        get_data(idx);
        return get_key(idx);
    }
    bool lookup(const char *str, std::set<glong> &idxs, glong &next_idx) override;

private:
    gchar *idxdatabuf;
    std::vector<gchar *> wordlist;
};

void OffsetIndex::page_t::fill(gchar *data, gint nent, glong idx_)
{
    idx = idx_;
    gchar *p = data;
    glong len;
    for (gint i = 0; i < nent; ++i) {
        entries[i].keystr = p;
        len = strlen(p);
        p += len + 1;
        entries[i].off = g_ntohl(get_uint32(p));
        p += sizeof(guint32);
        entries[i].size = g_ntohl(get_uint32(p));
        p += sizeof(guint32);
    }
}

inline const gchar *OffsetIndex::read_first_on_page_key(glong page_idx)
{
    fseek(idxfile, wordoffset[page_idx], SEEK_SET);
    guint32 page_size = wordoffset[page_idx + 1] - wordoffset[page_idx];
    const size_t nitems = fread(wordentry_buf,
                                std::min(sizeof(wordentry_buf), static_cast<size_t>(page_size)),
                                1, idxfile);
    THROW_IF_ERROR(nitems == 1);
    // TODO: check returned values, deal with word entry that strlen>255.
    return wordentry_buf;
}

inline const gchar *OffsetIndex::get_first_on_page_key(glong page_idx)
{
    if (page_idx < middle.idx) {
        if (page_idx == first.idx)
            return first.keystr.c_str();
        return read_first_on_page_key(page_idx);
    } else if (page_idx > middle.idx) {
        if (page_idx == last.idx)
            return last.keystr.c_str();
        return read_first_on_page_key(page_idx);
    } else
        return middle.keystr.c_str();
}

bool OffsetIndex::load_cache(const std::string &url)
{
    const std::list<std::string> vars = get_cache_variant(url);

    for (const std::string &item : vars) {
        struct ::stat idxstat, cachestat;
        if (g_stat(url.c_str(), &idxstat) != 0 || g_stat(item.c_str(), &cachestat) != 0)
            continue;
        if (cachestat.st_mtime < idxstat.st_mtime)
            continue;
        MapFile mf;
        if (!mf.open(item.c_str(), cachestat.st_size))
            continue;
        if (strncmp(mf.begin(), CACHE_MAGIC, strlen(CACHE_MAGIC)) != 0)
            continue;
        guint32 tmp;
        memcpy(&tmp, mf.begin() + strlen(CACHE_MAGIC), sizeof(tmp));
        if (tmp != CACHE_MAGIC_BYTES)
            continue;
        memcpy(&wordoffset[0], mf.begin() + strlen(CACHE_MAGIC) + sizeof(guint32), wordoffset.size() * sizeof(wordoffset[0]));
        return true;
    }

    return false;
}

std::list<std::string> OffsetIndex::get_cache_variant(const std::string &url)
{
    std::list<std::string> res = { url + ".oft" };
    if (!g_file_test(g_get_user_cache_dir(), G_FILE_TEST_EXISTS) && g_mkdir(g_get_user_cache_dir(), 0700) == -1)
        return res;

    const std::string cache_dir = std::string(g_get_user_cache_dir()) + G_DIR_SEPARATOR_S + "sdcv";

    if (!g_file_test(cache_dir.c_str(), G_FILE_TEST_EXISTS)) {
        if (g_mkdir(cache_dir.c_str(), 0700) == -1)
            return res;
    } else if (!g_file_test(cache_dir.c_str(), G_FILE_TEST_IS_DIR))
        return res;

    gchar *base = g_path_get_basename(url.c_str());
    res.push_back(cache_dir + G_DIR_SEPARATOR_S + base + ".oft");
    g_free(base);
    return res;
}

bool OffsetIndex::save_cache(const std::string &url, bool verbose)
{
    const std::list<std::string> vars = get_cache_variant(url);
    for (const std::string &item : vars) {
        FILE *out = fopen(item.c_str(), "wb");
        guint32 magic = CACHE_MAGIC_BYTES;
        if (!out)
            continue;
        if (fwrite(CACHE_MAGIC, 1, strlen(CACHE_MAGIC), out) != strlen(CACHE_MAGIC))
            continue;
        if (fwrite(&magic, 1, sizeof(magic), out) != sizeof(magic))
            continue;
        if (fwrite(&wordoffset[0], sizeof(wordoffset[0]), wordoffset.size(), out) != wordoffset.size())
            continue;
        fclose(out);
        if (verbose) {
            printf("save to cache %s\n", url.c_str());
        }
        return true;
    }
    return false;
}

bool OffsetIndex::load(const std::string &url, gulong wc, gulong fsize, bool verbose)
{
    wordcount = wc;
    gulong npages = (wc - 1) / ENTR_PER_PAGE + 2;
    wordoffset.resize(npages);
    if (!load_cache(url)) { // map file will close after finish of block
        MapFile map_file;
        if (!map_file.open(url.c_str(), fsize))
            return false;
        const gchar *idxdatabuffer = map_file.begin();

        const gchar *p1 = idxdatabuffer;
        gulong index_size;
        guint32 j = 0;
        for (guint32 i = 0; i < wc; i++) {
            index_size = strlen(p1) + 1 + 2 * sizeof(guint32);
            if (i % ENTR_PER_PAGE == 0) {
                wordoffset[j] = p1 - idxdatabuffer;
                ++j;
            }
            p1 += index_size;
        }
        wordoffset[j] = p1 - idxdatabuffer;
        if (!save_cache(url, verbose))
            fprintf(stderr, "cache update failed\n");
    }

    if (!(idxfile = fopen(url.c_str(), "rb"))) {
        wordoffset.resize(0);
        return false;
    }

    first.assign(0, read_first_on_page_key(0));
    last.assign(wordoffset.size() - 2, read_first_on_page_key(wordoffset.size() - 2));
    middle.assign((wordoffset.size() - 2) / 2, read_first_on_page_key((wordoffset.size() - 2) / 2));
    real_last.assign(wc - 1, get_key(wc - 1));

    return true;
}

inline gulong OffsetIndex::load_page(glong page_idx)
{
    gulong nentr = ENTR_PER_PAGE;
    if (page_idx == glong(wordoffset.size() - 2))
        if ((nentr = (wordcount % ENTR_PER_PAGE)) == 0)
            nentr = ENTR_PER_PAGE;

    if (page_idx != page.idx) {
        page_data.resize(wordoffset[page_idx + 1] - wordoffset[page_idx]);
        fseek(idxfile, wordoffset[page_idx], SEEK_SET);
        const size_t nitems = fread(&page_data[0], 1, page_data.size(), idxfile);
        THROW_IF_ERROR(nitems == page_data.size());

        page.fill(&page_data[0], nentr, page_idx);
    }

    return nentr;
}

const gchar *OffsetIndex::get_key(glong idx)
{
    load_page(idx / ENTR_PER_PAGE);
    glong idx_in_page = idx % ENTR_PER_PAGE;
    wordentry_offset = page.entries[idx_in_page].off;
    wordentry_size = page.entries[idx_in_page].size;

    return page.entries[idx_in_page].keystr;
}

bool OffsetIndex::lookup(const char *str, std::set<glong> &idxs, glong &next_idx)
{
    bool bFound = false;

    if (stardict_strcmp(str, first.keystr.c_str()) < 0) {
        next_idx = 0;
        return false;
    } else if (stardict_strcmp(str, real_last.keystr.c_str()) > 0) {
        next_idx = INVALID_INDEX;
        return false;
    }

    // Search for the first page where the word is likely to be located.
    glong iFrom = 0, iTo = wordoffset.size() - 2;
    glong iPage = 0, iThisIndex = 0;
    while (iFrom <= iTo) {
        iThisIndex = (iFrom + iTo) / 2;
        glong cmpint = stardict_strcmp(str, get_first_on_page_key(iThisIndex));
        if (cmpint > 0)
            iFrom = iThisIndex + 1;
        else if (cmpint < 0)
            iTo = iThisIndex - 1;
        else {
            bFound = true;
            break;
        }
    }

    if (bFound) {
        // We can use this found index (even though it might not be the first)
        // because we will search backwards later and catch any entries on
        // previous pages.
        iPage = iThisIndex;
        iThisIndex = 0; // first item in the page
    } else {
        iPage = iTo; // prev
        // Not found at the start of a page, so search within the page that
        // should contain it. Binary search here is slightly overkill (we're
        // searching at most ENTR_PER_PAGE = 32 elements) but this way next_idx
        // is treated the same as other Lookup methods.
        gulong netr = load_page(iPage);
        iFrom = 0;
        iTo = netr - 1;
        while (iFrom <= iTo) {
            iThisIndex = (iFrom + iTo) / 2;
            glong cmpint = stardict_strcmp(str, page.entries[iThisIndex].keystr);
            if (cmpint > 0)
                iFrom = iThisIndex + 1;
            else if (cmpint < 0)
                iTo = iThisIndex - 1;
            else {
                bFound = true;
                break;
            }
        }
    }

    if (!bFound)
        next_idx = iPage * ENTR_PER_PAGE + iFrom; // next
    else {
        // Convert the found in-page index to the dict index.
        iThisIndex = iPage * ENTR_PER_PAGE + iThisIndex;
        // In order to return all idxs that match the search string, walk
        // linearly behind and ahead of the found index.
        glong iHeadIndex = iThisIndex - 1; // do not include iThisIndex
        while (iHeadIndex >= 0 && stardict_strcmp(str, get_key(iHeadIndex)) == 0)
            idxs.insert(iHeadIndex--);
        do // no need to double-check iThisIndex -- we know it's a match already
            idxs.insert(iThisIndex++);
        while (iThisIndex <= real_last.idx && stardict_strcmp(str, get_key(iThisIndex)) == 0);
    }
    return bFound;
}

bool WordListIndex::load(const std::string &url, gulong wc, gulong fsize, bool)
{
    gzFile in = gzopen(url.c_str(), "rb");
    if (in == nullptr)
        return false;

    idxdatabuf = (gchar *)g_malloc(fsize);

    const int len = gzread(in, idxdatabuf, fsize);
    gzclose(in);
    if (len < 0)
        return false;

    if (gulong(len) != fsize)
        return false;

    wordlist.resize(wc + 1);
    gchar *p1 = idxdatabuf;
    guint32 i;
    for (i = 0; i < wc; i++) {
        wordlist[i] = p1;
        p1 += strlen(p1) + 1 + 2 * sizeof(guint32);
    }
    wordlist[wc] = p1;

    return true;
}

void WordListIndex::get_data(glong idx)
{
    gchar *p1 = wordlist[idx] + strlen(wordlist[idx]) + sizeof(gchar);
    wordentry_offset = g_ntohl(get_uint32(p1));
    p1 += sizeof(guint32);
    wordentry_size = g_ntohl(get_uint32(p1));
}

bool WordListIndex::lookup(const char *str, std::set<glong> &idxs, glong &next_idx)
{
    bool bFound = false;
    glong iLast = wordlist.size() - 2;

    if (stardict_strcmp(str, get_key(0)) < 0) {
        next_idx = 0;
    } else if (stardict_strcmp(str, get_key(iLast)) > 0) {
        next_idx = INVALID_INDEX;
    } else {
        glong iThisIndex = 0;
        glong iFrom = 0, iTo = iLast;
        gint cmpint;
        while (iFrom <= iTo) {
            iThisIndex = (iFrom + iTo) / 2;
            cmpint = stardict_strcmp(str, get_key(iThisIndex));
            if (cmpint > 0)
                iFrom = iThisIndex + 1;
            else if (cmpint < 0)
                iTo = iThisIndex - 1;
            else {
                bFound = true;
                break;
            }
        }
        if (!bFound)
            next_idx = iFrom; // next
        else {
            // In order to return all idxs that match the search string, walk
            // linearly behind and ahead of the found index.
            glong iHeadIndex = iThisIndex - 1; // do not include iThisIndex
            while (iHeadIndex >= 0 && stardict_strcmp(str, get_key(iHeadIndex)) == 0)
                idxs.insert(iHeadIndex--);
            do // no need to double-check iThisIndex -- we know it's a match already
                idxs.insert(iThisIndex++);
            while (iThisIndex <= iLast && stardict_strcmp(str, get_key(iThisIndex)) == 0);
        }
    }
    return bFound;
}
} // namespace

bool SynFile::load(const std::string &url, gulong wc)
{
    struct stat stat_buf;
    if (!stat(url.c_str(), &stat_buf)) {

        if (!synfile.open(url.c_str(), stat_buf.st_size))
            return false;

        synlist.resize(wc + 1);
        gchar *p1 = synfile.begin();

        for (unsigned long i = 0; i < wc; i++) {
            // each entry in a syn-file is:
            // - 0-terminated string
            // 4-byte index into .dict file in network byte order

            synlist[i] = p1;
            p1 += strlen(p1) + 1 + 4;
        }
        synlist[wc] = p1;

        return true;
    } else {
        return false;
    }
}

bool SynFile::lookup(const char *str, std::set<glong> &idxs, glong &next_idx)
{
    bool bFound = false;
    glong iLast = synlist.size() - 2;
    if (iLast < 0)
        return false;

    if (stardict_strcmp(str, get_key(0)) < 0) {
        next_idx = 0;
    } else if (stardict_strcmp(str, get_key(iLast)) > 0) {
        next_idx = INVALID_INDEX;
    } else {
        glong iThisIndex = 0;
        glong iFrom = 0, iTo = iLast;
        gint cmpint;
        while (iFrom <= iTo) {
            iThisIndex = (iFrom + iTo) / 2;
            cmpint = stardict_strcmp(str, get_key(iThisIndex));
            if (cmpint > 0)
                iFrom = iThisIndex + 1;
            else if (cmpint < 0)
                iTo = iThisIndex - 1;
            else {
                bFound = true;
                break;
            }
        }
        if (!bFound)
            next_idx = iFrom; // next
        else {
            // In order to return all idxs that match the search string, walk
            // linearly behind and ahead of the found index.
            glong iHeadIndex = iThisIndex - 1; // do not include iThisIndex
            while (iHeadIndex >= 0 && stardict_strcmp(str, get_key(iHeadIndex)) == 0) {
                const gchar *key = get_key(iHeadIndex--);
                idxs.insert(g_ntohl(get_uint32(key + strlen(key) + 1)));
            }
            do {
                // no need to double-check iThisIndex -- we know it's a match already
                const gchar *key = get_key(iThisIndex++);
                idxs.insert(g_ntohl(get_uint32(key + strlen(key) + 1)));
            } while (iThisIndex <= iLast && stardict_strcmp(str, get_key(iThisIndex)) == 0);
        }
    }
    return bFound;
}

bool Dict::Lookup(const char *str, std::set<glong> &idxs, glong &next_idx)
{
    bool found = false;
    found |= syn_file->lookup(str, idxs, next_idx);
    found |= idx_file->lookup(str, idxs, next_idx);
    return found;
}

bool Dict::load(const std::string &ifofilename, bool verbose)
{
    gulong idxfilesize;
    if (!load_ifofile(ifofilename, idxfilesize))
        return false;

    std::string fullfilename(ifofilename);
    fullfilename.replace(fullfilename.length() - sizeof("ifo") + 1, sizeof("ifo") - 1, "dict.dz");

    if (g_file_test(fullfilename.c_str(), G_FILE_TEST_EXISTS)) {
        dictdzfile.reset(new DictData);
        if (!dictdzfile->open(fullfilename, 0)) {
            // g_print("open file %s failed!\n",fullfilename);
            return false;
        }
    } else {
        fullfilename.erase(fullfilename.length() - sizeof(".dz") + 1, sizeof(".dz") - 1);
        dictfile = fopen(fullfilename.c_str(), "rb");
        if (!dictfile) {
            // g_print("open file %s failed!\n",fullfilename);
            return false;
        }
    }

    fullfilename = ifofilename;
    fullfilename.replace(fullfilename.length() - sizeof("ifo") + 1, sizeof("ifo") - 1, "idx.gz");

    if (g_file_test(fullfilename.c_str(), G_FILE_TEST_EXISTS)) {
        idx_file.reset(new WordListIndex);
    } else {
        fullfilename.erase(fullfilename.length() - sizeof(".gz") + 1, sizeof(".gz") - 1);
        idx_file.reset(new OffsetIndex);
    }

    if (!idx_file->load(fullfilename, wordcount, idxfilesize, verbose))
        return false;

    fullfilename = ifofilename;
    fullfilename.replace(fullfilename.length() - sizeof("ifo") + 1, sizeof("ifo") - 1, "syn");
    syn_file.reset(new SynFile);
    syn_file->load(fullfilename, syn_wordcount);

    // g_print("bookname: %s , wordcount %lu\n", bookname.c_str(), narticles());
    return true;
}

bool Dict::load_ifofile(const std::string &ifofilename, gulong &idxfilesize)
{
    DictInfo dict_info;
    if (!dict_info.load_from_ifo_file(ifofilename, false))
        return false;
    if (dict_info.wordcount == 0)
        return false;

    ifo_file_name = dict_info.ifo_file_name;
    wordcount = dict_info.wordcount;
    syn_wordcount = dict_info.syn_wordcount;
    bookname = dict_info.bookname;

    idxfilesize = dict_info.index_file_size;

    sametypesequence = dict_info.sametypesequence;

    return true;
}

bool Dict::LookupWithRule(GPatternSpec *pspec, glong *aIndex, int iBuffLen)
{
    int iIndexCount = 0;

    for (guint32 i = 0; i < narticles() && iIndexCount < (iBuffLen - 1); i++)
        if (g_pattern_spec_match_string(pspec, get_key(i)))
            aIndex[iIndexCount++] = i;

    aIndex[iIndexCount] = -1; // -1 is the end.

    return iIndexCount > 0;
}

Libs::~Libs()
{
    for (Dict *p : oLib)
        delete p;
}

void Libs::load_dict(const std::string &url)
{
    Dict *lib = new Dict;
    if (lib->load(url, verbose_))
        oLib.push_back(lib);
    else
        delete lib;
}

void Libs::load(const std::list<std::string> &dicts_dirs,
                const std::list<std::string> &order_list,
                const std::list<std::string> &disable_list)
{
    for_each_file(dicts_dirs, ".ifo", order_list, disable_list,
                  [this](const std::string &url, bool disable) -> void {
                      if (!disable)
                          load_dict(url);
                  });
}

bool Libs::LookupSimilarWord(const gchar *sWord, std::set<glong> &iWordIndices, int iLib)
{
    bool bFound = false;
    gchar *casestr;

    if (!bFound) {
        // to lower case.
        casestr = g_utf8_strdown(sWord, -1);
        if (strcmp(casestr, sWord)) {
            if (oLib[iLib]->Lookup(casestr, iWordIndices))
                bFound = true;
        }
        g_free(casestr);
        // to upper case.
        if (!bFound) {
            casestr = g_utf8_strup(sWord, -1);
            if (strcmp(casestr, sWord)) {
                if (oLib[iLib]->Lookup(casestr, iWordIndices))
                    bFound = true;
            }
            g_free(casestr);
        }
        // Upper the first character and lower others.
        if (!bFound) {
            gchar *nextchar = g_utf8_next_char(sWord);
            gchar *firstchar = g_utf8_strup(sWord, nextchar - sWord);
            nextchar = g_utf8_strdown(nextchar, -1);
            casestr = g_strdup_printf("%s%s", firstchar, nextchar);
            g_free(firstchar);
            g_free(nextchar);
            if (strcmp(casestr, sWord)) {
                if (oLib[iLib]->Lookup(casestr, iWordIndices))
                    bFound = true;
            }
            g_free(casestr);
        }
    }

    if (bIsPureEnglish(sWord)) {
        // If not Found , try other status of sWord.
        int iWordLen = strlen(sWord);
        bool isupcase;

        gchar *sNewWord = (gchar *)g_malloc(iWordLen + 1);

        // cut one char "s" or "d"
        if (!bFound && iWordLen > 1) {
            isupcase = sWord[iWordLen - 1] == 'S' || !strncmp(&sWord[iWordLen - 2], "ED", 2);
            if (isupcase || sWord[iWordLen - 1] == 's' || !strncmp(&sWord[iWordLen - 2], "ed", 2)) {
                strcpy(sNewWord, sWord);
                sNewWord[iWordLen - 1] = '\0'; // cut "s" or "d"
                if (oLib[iLib]->Lookup(sNewWord, iWordIndices))
                    bFound = true;
                else if (isupcase || g_ascii_isupper(sWord[0])) {
                    casestr = g_ascii_strdown(sNewWord, -1);
                    if (strcmp(casestr, sNewWord)) {
                        if (oLib[iLib]->Lookup(casestr, iWordIndices))
                            bFound = true;
                    }
                    g_free(casestr);
                }
            }
        }

        // cut "ly"
        if (!bFound && iWordLen > 2) {
            isupcase = !strncmp(&sWord[iWordLen - 2], "LY", 2);
            if (isupcase || (!strncmp(&sWord[iWordLen - 2], "ly", 2))) {
                strcpy(sNewWord, sWord);
                sNewWord[iWordLen - 2] = '\0'; // cut "ly"
                if (iWordLen > 5 && sNewWord[iWordLen - 3] == sNewWord[iWordLen - 4]
                    && !bIsVowel(sNewWord[iWordLen - 4]) && bIsVowel(sNewWord[iWordLen - 5])) { // doubled

                    sNewWord[iWordLen - 3] = '\0';
                    if (oLib[iLib]->Lookup(sNewWord, iWordIndices))
                        bFound = true;
                    else {
                        if (isupcase || g_ascii_isupper(sWord[0])) {
                            casestr = g_ascii_strdown(sNewWord, -1);
                            if (strcmp(casestr, sNewWord)) {
                                if (oLib[iLib]->Lookup(casestr, iWordIndices))
                                    bFound = true;
                            }
                            g_free(casestr);
                        }
                        if (!bFound)
                            sNewWord[iWordLen - 3] = sNewWord[iWordLen - 4]; // restore
                    }
                }
                if (!bFound) {
                    if (oLib[iLib]->Lookup(sNewWord, iWordIndices))
                        bFound = true;
                    else if (isupcase || g_ascii_isupper(sWord[0])) {
                        casestr = g_ascii_strdown(sNewWord, -1);
                        if (strcmp(casestr, sNewWord)) {
                            if (oLib[iLib]->Lookup(casestr, iWordIndices))
                                bFound = true;
                        }
                        g_free(casestr);
                    }
                }
            }
        }

        // cut "ing"
        if (!bFound && iWordLen > 3) {
            isupcase = !strncmp(&sWord[iWordLen - 3], "ING", 3);
            if (isupcase || !strncmp(&sWord[iWordLen - 3], "ing", 3)) {
                strcpy(sNewWord, sWord);
                sNewWord[iWordLen - 3] = '\0';
                if (iWordLen > 6 && (sNewWord[iWordLen - 4] == sNewWord[iWordLen - 5])
                    && !bIsVowel(sNewWord[iWordLen - 5]) && bIsVowel(sNewWord[iWordLen - 6])) { // doubled
                    sNewWord[iWordLen - 4] = '\0';
                    if (oLib[iLib]->Lookup(sNewWord, iWordIndices))
                        bFound = true;
                    else {
                        if (isupcase || g_ascii_isupper(sWord[0])) {
                            casestr = g_ascii_strdown(sNewWord, -1);
                            if (strcmp(casestr, sNewWord)) {
                                if (oLib[iLib]->Lookup(casestr, iWordIndices))
                                    bFound = true;
                            }
                            g_free(casestr);
                        }
                        if (!bFound)
                            sNewWord[iWordLen - 4] = sNewWord[iWordLen - 5]; // restore
                    }
                }
                if (!bFound) {
                    if (oLib[iLib]->Lookup(sNewWord, iWordIndices))
                        bFound = true;
                    else if (isupcase || g_ascii_isupper(sWord[0])) {
                        casestr = g_ascii_strdown(sNewWord, -1);
                        if (strcmp(casestr, sNewWord)) {
                            if (oLib[iLib]->Lookup(casestr, iWordIndices))
                                bFound = true;
                        }
                        g_free(casestr);
                    }
                }
                if (!bFound) {
                    if (isupcase)
                        strcat(sNewWord, "E"); // add a char "E"
                    else
                        strcat(sNewWord, "e"); // add a char "e"
                    if (oLib[iLib]->Lookup(sNewWord, iWordIndices))
                        bFound = true;
                    else if (isupcase || g_ascii_isupper(sWord[0])) {
                        casestr = g_ascii_strdown(sNewWord, -1);
                        if (strcmp(casestr, sNewWord)) {
                            if (oLib[iLib]->Lookup(casestr, iWordIndices))
                                bFound = true;
                        }
                        g_free(casestr);
                    }
                }
            }
        }

        // cut two char "es"
        if (!bFound && iWordLen > 3) {
            isupcase = (!strncmp(&sWord[iWordLen - 2], "ES", 2) && (sWord[iWordLen - 3] == 'S' || sWord[iWordLen - 3] == 'X' || sWord[iWordLen - 3] == 'O' || (iWordLen > 4 && sWord[iWordLen - 3] == 'H' && (sWord[iWordLen - 4] == 'C' || sWord[iWordLen - 4] == 'S'))));
            if (isupcase || (!strncmp(&sWord[iWordLen - 2], "es", 2) && (sWord[iWordLen - 3] == 's' || sWord[iWordLen - 3] == 'x' || sWord[iWordLen - 3] == 'o' || (iWordLen > 4 && sWord[iWordLen - 3] == 'h' && (sWord[iWordLen - 4] == 'c' || sWord[iWordLen - 4] == 's'))))) {
                strcpy(sNewWord, sWord);
                sNewWord[iWordLen - 2] = '\0';
                if (oLib[iLib]->Lookup(sNewWord, iWordIndices))
                    bFound = true;
                else if (isupcase || g_ascii_isupper(sWord[0])) {
                    casestr = g_ascii_strdown(sNewWord, -1);
                    if (strcmp(casestr, sNewWord)) {
                        if (oLib[iLib]->Lookup(casestr, iWordIndices))
                            bFound = true;
                    }
                    g_free(casestr);
                }
            }
        }

        // cut "ed"
        if (!bFound && iWordLen > 3) {
            isupcase = !strncmp(&sWord[iWordLen - 2], "ED", 2);
            if (isupcase || !strncmp(&sWord[iWordLen - 2], "ed", 2)) {
                strcpy(sNewWord, sWord);
                sNewWord[iWordLen - 2] = '\0';
                if (iWordLen > 5 && (sNewWord[iWordLen - 3] == sNewWord[iWordLen - 4])
                    && !bIsVowel(sNewWord[iWordLen - 4]) && bIsVowel(sNewWord[iWordLen - 5])) { // doubled
                    sNewWord[iWordLen - 3] = '\0';
                    if (oLib[iLib]->Lookup(sNewWord, iWordIndices))
                        bFound = true;
                    else {
                        if (isupcase || g_ascii_isupper(sWord[0])) {
                            casestr = g_ascii_strdown(sNewWord, -1);
                            if (strcmp(casestr, sNewWord)) {
                                if (oLib[iLib]->Lookup(casestr, iWordIndices))
                                    bFound = true;
                            }
                            g_free(casestr);
                        }
                        if (!bFound)
                            sNewWord[iWordLen - 3] = sNewWord[iWordLen - 4]; // restore
                    }
                }
                if (!bFound) {
                    if (oLib[iLib]->Lookup(sNewWord, iWordIndices))
                        bFound = true;
                    else if (isupcase || g_ascii_isupper(sWord[0])) {
                        casestr = g_ascii_strdown(sNewWord, -1);
                        if (strcmp(casestr, sNewWord)) {
                            if (oLib[iLib]->Lookup(casestr, iWordIndices))
                                bFound = true;
                        }
                        g_free(casestr);
                    }
                }
            }
        }

        // cut "ied" , add "y".
        if (!bFound && iWordLen > 3) {
            isupcase = !strncmp(&sWord[iWordLen - 3], "IED", 3);
            if (isupcase || (!strncmp(&sWord[iWordLen - 3], "ied", 3))) {
                strcpy(sNewWord, sWord);
                sNewWord[iWordLen - 3] = '\0';
                if (isupcase)
                    strcat(sNewWord, "Y"); // add a char "Y"
                else
                    strcat(sNewWord, "y"); // add a char "y"
                if (oLib[iLib]->Lookup(sNewWord, iWordIndices))
                    bFound = true;
                else if (isupcase || g_ascii_isupper(sWord[0])) {
                    casestr = g_ascii_strdown(sNewWord, -1);
                    if (strcmp(casestr, sNewWord)) {
                        if (oLib[iLib]->Lookup(casestr, iWordIndices))
                            bFound = true;
                    }
                    g_free(casestr);
                }
            }
        }

        // cut "ies" , add "y".
        if (!bFound && iWordLen > 3) {
            isupcase = !strncmp(&sWord[iWordLen - 3], "IES", 3);
            if (isupcase || (!strncmp(&sWord[iWordLen - 3], "ies", 3))) {
                strcpy(sNewWord, sWord);
                sNewWord[iWordLen - 3] = '\0';
                if (isupcase)
                    strcat(sNewWord, "Y"); // add a char "Y"
                else
                    strcat(sNewWord, "y"); // add a char "y"
                if (oLib[iLib]->Lookup(sNewWord, iWordIndices))
                    bFound = true;
                else if (isupcase || g_ascii_isupper(sWord[0])) {
                    casestr = g_ascii_strdown(sNewWord, -1);
                    if (strcmp(casestr, sNewWord)) {
                        if (oLib[iLib]->Lookup(casestr, iWordIndices))
                            bFound = true;
                    }
                    g_free(casestr);
                }
            }
        }

        // cut "er".
        if (!bFound && iWordLen > 2) {
            isupcase = !strncmp(&sWord[iWordLen - 2], "ER", 2);
            if (isupcase || (!strncmp(&sWord[iWordLen - 2], "er", 2))) {
                strcpy(sNewWord, sWord);
                sNewWord[iWordLen - 2] = '\0';
                if (oLib[iLib]->Lookup(sNewWord, iWordIndices))
                    bFound = true;
                else if (isupcase || g_ascii_isupper(sWord[0])) {
                    casestr = g_ascii_strdown(sNewWord, -1);
                    if (strcmp(casestr, sNewWord)) {
                        if (oLib[iLib]->Lookup(casestr, iWordIndices))
                            bFound = true;
                    }
                    g_free(casestr);
                }
            }
        }

        // cut "est".
        if (!bFound && iWordLen > 3) {
            isupcase = !strncmp(&sWord[iWordLen - 3], "EST", 3);
            if (isupcase || (!strncmp(&sWord[iWordLen - 3], "est", 3))) {
                strcpy(sNewWord, sWord);
                sNewWord[iWordLen - 3] = '\0';
                if (oLib[iLib]->Lookup(sNewWord, iWordIndices))
                    bFound = true;
                else if (isupcase || g_ascii_isupper(sWord[0])) {
                    casestr = g_ascii_strdown(sNewWord, -1);
                    if (strcmp(casestr, sNewWord)) {
                        if (oLib[iLib]->Lookup(casestr, iWordIndices))
                            bFound = true;
                    }
                    g_free(casestr);
                }
            }
        }

        g_free(sNewWord);
    }
#if 0
    else {
        //don't change iWordIndex here.
        //when LookupSimilarWord all failed too, we want to use the old LookupWord index to list words.
        //iWordIndex = INVALID_INDEX;
    }
#endif
    return bFound;
}

bool Libs::SimpleLookupWord(const gchar *sWord, std::set<glong> &iWordIndices, int iLib)
{
    bool bFound = oLib[iLib]->Lookup(sWord, iWordIndices);
    if (!bFound && fuzzy_)
        bFound = LookupSimilarWord(sWord, iWordIndices, iLib);
    return bFound;
}

bool Libs::LookupWithFuzzy(const gchar *sWord, gchar *reslist[], gint reslist_size)
{
    if (sWord[0] == '\0')
        return false;

    Fuzzystruct oFuzzystruct[reslist_size];

    for (int i = 0; i < reslist_size; i++) {
        oFuzzystruct[i].pMatchWord = nullptr;
        oFuzzystruct[i].iMatchWordDistance = iMaxFuzzyDistance;
    }
    int iMaxDistance = iMaxFuzzyDistance;
    int iDistance;
    bool Found = false;
    EditDistance oEditDistance;

    glong iCheckWordLen;
    const char *sCheck;
    gunichar *ucs4_str1, *ucs4_str2;
    glong ucs4_str2_len;

    ucs4_str2 = g_utf8_to_ucs4_fast(sWord, -1, &ucs4_str2_len);
    unicode_strdown(ucs4_str2);

    for (size_t iLib = 0; iLib < oLib.size(); ++iLib) {
        if (progress_func)
            progress_func();

        // if (stardict_strcmp(sWord, poGetWord(0,iLib))>=0 && stardict_strcmp(sWord, poGetWord(narticles(iLib)-1,iLib))<=0) {
        // there are Chinese dicts and English dicts...

        const int iwords = narticles(iLib);
        for (int index = 0; index < iwords; index++) {
            sCheck = poGetWord(index, iLib);
            // tolower and skip too long or too short words
            iCheckWordLen = g_utf8_strlen(sCheck, -1);
            if (iCheckWordLen - ucs4_str2_len >= iMaxDistance || ucs4_str2_len - iCheckWordLen >= iMaxDistance)
                continue;
            ucs4_str1 = g_utf8_to_ucs4_fast(sCheck, -1, nullptr);
            if (iCheckWordLen > ucs4_str2_len)
                ucs4_str1[ucs4_str2_len] = 0;
            unicode_strdown(ucs4_str1);

            iDistance = oEditDistance.CalEditDistance(ucs4_str1, ucs4_str2, iMaxDistance);
            g_free(ucs4_str1);
            if (iDistance < iMaxDistance && iDistance < ucs4_str2_len) {
                // when ucs4_str2_len=1,2 we need less fuzzy.
                Found = true;
                bool bAlreadyInList = false;
                int iMaxDistanceAt = 0;
                for (int j = 0; j < reslist_size; j++) {
                    if (oFuzzystruct[j].pMatchWord && strcmp(oFuzzystruct[j].pMatchWord, sCheck) == 0) { // already in list
                        bAlreadyInList = true;
                        break;
                    }
                    // find the position,it will certainly be found (include the first time) as iMaxDistance is set by last time.
                    if (oFuzzystruct[j].iMatchWordDistance == iMaxDistance) {
                        iMaxDistanceAt = j;
                    }
                }
                if (!bAlreadyInList) {
                    if (oFuzzystruct[iMaxDistanceAt].pMatchWord)
                        g_free(oFuzzystruct[iMaxDistanceAt].pMatchWord);
                    oFuzzystruct[iMaxDistanceAt].pMatchWord = g_strdup(sCheck);
                    oFuzzystruct[iMaxDistanceAt].iMatchWordDistance = iDistance;
                    // calc new iMaxDistance
                    iMaxDistance = iDistance;
                    for (int j = 0; j < reslist_size; j++) {
                        if (oFuzzystruct[j].iMatchWordDistance > iMaxDistance)
                            iMaxDistance = oFuzzystruct[j].iMatchWordDistance;
                    } // calc new iMaxDistance
                } // add to list
            } // find one
        } // each word

    } // each lib
    g_free(ucs4_str2);

    if (Found) // sort with distance
        std::sort(oFuzzystruct, oFuzzystruct + reslist_size, [](const Fuzzystruct &lh, const Fuzzystruct &rh) -> bool {
            if (lh.iMatchWordDistance != rh.iMatchWordDistance)
                return lh.iMatchWordDistance < rh.iMatchWordDistance;

            if (lh.pMatchWord && rh.pMatchWord)
                return stardict_strcmp(lh.pMatchWord, rh.pMatchWord) < 0;

            return false;
        });

    for (gint i = 0; i < reslist_size; ++i)
        reslist[i] = oFuzzystruct[i].pMatchWord;

    return Found;
}

gint Libs::LookupWithRule(const gchar *word, gchar **ppMatchWord)
{
    glong aiIndex[MAX_MATCH_ITEM_PER_LIB + 1];
    gint iMatchCount = 0;
    GPatternSpec *pspec = g_pattern_spec_new(word);

    for (std::vector<Dict *>::size_type iLib = 0; iLib < oLib.size(); iLib++) {
        // if(oLibs.LookdupWordsWithRule(pspec,aiIndex,MAX_MATCH_ITEM_PER_LIB+1-iMatchCount,iLib))
        //  -iMatchCount,so save time,but may got less result and the word may repeat.

        if (oLib[iLib]->LookupWithRule(pspec, aiIndex, MAX_MATCH_ITEM_PER_LIB + 1)) {
            if (progress_func)
                progress_func();
            for (int i = 0; aiIndex[i] != -1; i++) {
                const gchar *sMatchWord = poGetWord(aiIndex[i], iLib);
                bool bAlreadyInList = false;
                for (int j = 0; j < iMatchCount; j++) {
                    if (strcmp(ppMatchWord[j], sMatchWord) == 0) { // already in list
                        bAlreadyInList = true;
                        break;
                    }
                }
                if (!bAlreadyInList)
                    ppMatchWord[iMatchCount++] = g_strdup(sMatchWord);
            }
        }
    }
    g_pattern_spec_free(pspec);

    if (iMatchCount) // sort it.
        std::sort(ppMatchWord, ppMatchWord + iMatchCount, [](const char *lh, const char *rh) -> bool {
            return stardict_strcmp(lh, rh) < 0;
        });

    return iMatchCount;
}

bool Libs::LookupData(const gchar *sWord, std::vector<gchar *> *reslist)
{
    std::vector<std::string> SearchWords;
    std::string SearchWord;
    const char *p = sWord;
    while (*p) {
        if (*p == '\\') {
            p++;
            switch (*p) {
            case ' ':
                SearchWord += ' ';
                break;
            case '\\':
                SearchWord += '\\';
                break;
            case 't':
                SearchWord += '\t';
                break;
            case 'n':
                SearchWord += '\n';
                break;
            default:
                SearchWord += *p;
            }
        } else if (*p == ' ') {
            if (!SearchWord.empty()) {
                SearchWords.push_back(SearchWord);
                SearchWord.clear();
            }
        } else {
            SearchWord += *p;
        }
        p++;
    }
    if (!SearchWord.empty()) {
        SearchWords.push_back(SearchWord);
        SearchWord.clear();
    }
    if (SearchWords.empty())
        return false;

    guint32 max_size = 0;
    gchar *origin_data = nullptr;
    for (std::vector<Dict *>::size_type i = 0; i < oLib.size(); ++i) {
        if (!oLib[i]->containSearchData())
            continue;
        if (progress_func)
            progress_func();
        const gulong iwords = narticles(i);
        const gchar *key;
        guint32 offset, size;
        for (gulong j = 0; j < iwords; ++j) {
            oLib[i]->get_key_and_data(j, &key, &offset, &size);
            if (size > max_size) {
                origin_data = (gchar *)g_realloc(origin_data, size);
                max_size = size;
            }
            if (oLib[i]->SearchData(SearchWords, offset, size, origin_data))
                reslist[i].push_back(g_strdup(key));
        }
    }
    g_free(origin_data);

    std::vector<Dict *>::size_type i;
    for (i = 0; i < oLib.size(); ++i)
        if (!reslist[i].empty())
            break;

    return i != oLib.size();
}

/**************************************************/
query_t analyze_query(const char *s, std::string &res)
{
    if (!s || !*s) {
        res = "";
        return qtSIMPLE;
    }
    if (*s == '/') {
        res = s + 1;
        return qtFUZZY;
    }

    if (*s == '|') {
        res = s + 1;
        return qtDATA;
    }

    bool regexp = false;
    const char *p = s;
    res = "";
    for (; *p; res += *p, ++p) {
        if (*p == '\\') {
            ++p;
            if (!*p)
                break;
            continue;
        }
        if (*p == '*' || *p == '?')
            regexp = true;
    }
    if (regexp)
        return qtREGEXP;

    return qtSIMPLE;
}