• 前两个字段INDEX_MAGIC_HEADER和INDEX_FORMAT_VERSION分别是magic number和索引版本号
    • 第三个字段USE_64BIT表示是否使用64位的document和word id(默认是使用).
    • 然后是写入docinfo,这个字段也就是配置中的docinfo字段(index block中)
    • 接下来将会写入schema，也就是索引的schema信息，比如当前索引的字段名，当前需要建立的属性名等等.
    2. {
    3.     // schema
    4.     fdInfo.PutDword ( tSchema.m_dFields.GetLength() );
    5.     ARRAY_FOREACH ( i, tSchema.m_dFields )
    6.         WriteSchemaColumn ( fdInfo, tSchema.m_dFields[i] );
    8.     fdInfo.PutDword ( tSchema.GetAttrsCount() );
    9.     for ( int i=0; i<tSchema.GetAttrsCount(); i++ )
    10.         WriteSchemaColumn ( fdInfo, tSchema.GetAttr(i) );
    11. }
    • 然后是写入当前索引集的最小doc id(m_uMinDocid)
    • 接下来是根据docinfo(也就是属性存储)的配置来选择是否写入行信息(当docinfo为inline的话，表示attribute value 将会存储在spd文件中).
    • 然后是写入wordlist的checkpoint.
    • 接下来是写入对应的索引配置信息
    • 写入对应的tokenizer的配置信息,
    1. void SaveTokenizerSettings ( CSphWriter & tWriter, ISphTokenizer * pTokenizer, int iEmbeddedLimit )
    2. {
    5.     const CSphTokenizerSettings & tSettings = pTokenizer->GetSettings ();
    6.     tWriter.PutByte ( tSettings.m_iType );
    7.     tWriter.PutString ( tSettings.m_sCaseFolding.cstr () );
    8.     tWriter.PutDword ( tSettings.m_iMinWordLen );
    10.     bool bEmbedSynonyms = pTokenizer->GetSynFileInfo ().m_uSize<=(SphOffset_t)iEmbeddedLimit;
    11.     tWriter.PutByte ( bEmbedSynonyms ? 1 : 0 );
    12.     if ( bEmbedSynonyms )
    13.         pTokenizer->WriteSynonyms ( tWriter );
    15.     tWriter.PutString ( tSettings.m_sSynonymsFile.cstr () );
    17.     tWriter.PutString ( tSettings.m_sIgnoreChars.cstr () );
    18.     tWriter.PutDword ( tSettings.m_iNgramLen );
    19.     tWriter.PutString ( tSettings.m_sNgramChars.cstr () );
    20.     tWriter.PutString ( tSettings.m_sBlendChars.cstr () );
    21.     tWriter.PutString ( tSettings.m_sBlendMode.cstr () );
    22. }
    • 写入dictionary的配置信息(比如stop word之类).
    • 然后是写入killlist的size（m_uKillListSize)
    • 写入m_iMinMaxIndex，这个选项也就是表示document size.
    1.     CSphFixedVector<CSphRowitem> dMinRow ( tNewSchema.GetRowSize() );
    2.     ...............
    3.     int iNewStride = DOCINFO_IDSIZE + tNewSchema.GetRowSize();
    5.     int64_t iNewMinMaxIndex = m_iDocinfo * iNewStride;
    7.     tBuildHeader.m_iMinMaxIndex = iNewMinMaxIndex;
    • 写入regex相关配置(regexp_filter)
    • 最后是写入对应的schema field长度.