到目前为止，我的数据库有2583000个文档。它是静态的（我们不经常插入新文档）。
这是一份示例文件：
{
"_id" : ObjectId("5492d118426f72116a436a0c"),
"dst" : "string",
"n" : "All the Light We Cannot See",
"su" : "some string...",
"detail" : {
"_id" : ObjectId("982736408170649823746"),
"au" : [
"string"
],
"auu" : [
"some string..."
],
"be" : null,
"bf" : "some string...",
"bp" : 531,
"ch" : [],
"ela" : "English",
"fp" : null,
"g" : [
"Historical Fiction",
"Fiction",
"Book Club",
"History",
"World War II",
"War",
"Cultural",
"France",
"Adult",
"Adult Fiction",
"Literary Fiction",
"Cultural",
"Germany"
],
"i13" : "some string...",
"ior" : "some string...",
"isb" : "some string...",
"ism" : "some string...",
"ist" : "some string...",
"k" : "book",
"law" : [
"some string here"
],
"oec" : 17,
"oeu" : "some string",
"ot" : "All The Light We Cannot See",
"ott" : {
"English" : "All the Light We Cannot See",
"German" : "Alles Licht, das wir nicht sehen"
},
"pu" : "Published May 6th 2014 by Scribner",
"r" : 4.18,
"rc" : 27738,
"rvc" : 4220,
"stt" : [
"Paris"
],
"sy" : "sime text here...",
"tad" : null,
"tau" : null,
"updated_at" : ISODate("2014-11-03T11:36:15.436Z")
},
"sv" : "1",
"updated_at" : ISODate("2014-12-18T13:05:28.190Z"),
"created_at" : ISODate("2014-12-18T13:05:28.190Z")
}
我设定的“文字”索引：
db.items.ensureIndex(
{"n": "text", "detail.est" : "text", "detail.ot" : "text"}, {default_language: "none"}
)
请注意，并非所有索引创建时使用的字段都在所有文档中。另外，我使用default_language : none作为text_search同时通过不同的语言。
前几天我收到的文件有一半，搜索速度很快。现在，以下结果对我来说似乎很奇怪，有些事情我不太明白：
和搜索
时间：138秒
查询：db.items.find（{$ text：{$ search：“\”all \“\”\“\”light \“\”我们\“\”不能\“\”参见\ “”}}）
解释：
{
"cursor" : "TextCursor",
"n" : 2,
"nscannedObjects" : 720806,
"nscanned" : 733110,
"nscannedObjectsAllPlans" : 720806,
"nscannedAllPlans" : 733110,
"scanAndOrder" : false,
"nYields" : 13395,
"nChunkSkips" : 0,
"millis" : 161997,
"server" : "myserver.com",
"filterSet" : false,
"stats" : {
"type" : "TEXT",
"works" : 1453924,
"yields" : 13395,
"unyields" : 13395,
"invalidates" : 0,
"advanced" : 2,
"needTime" : 1453921,
"needFetch" : 0,
"isEOF" : 1,
"keysExamined" : 733110,
"fetches" : 720806,
"parsedTextQuery" : {
"terms" : [
"all",
"the",
"light",
"we",
"cannot",
"see"
],
"negatedTerms" : [],
"phrases" : [
"all",
"the",
"light",
"we",
"cannot",
"see"
],
"negatedPhrases" : []
},
"children" : []
}
}
或搜索
时间：4.4秒
查询：db.items.find（{$ text：{$ search：“我们看不到的所有灯光”}}}
解释：
{
"cursor" : "TextCursor",
"n" : 720806,
"nscannedObjects" : 720806,
"nscanned" : 733110,
"nscannedObjectsAllPlans" : 720806,
"nscannedAllPlans" : 733110,
"scanAndOrder" : false,
"nYields" : 12317,
"nChunkSkips" : 0,
"millis" : 127775,
"server" : "myserver.com",
"filterSet" : false,
"stats" : {
"type" : "TEXT",
"works" : 1453924,
"yields" : 12317,
"unyields" : 12317,
"invalidates" : 0,
"advanced" : 720806,
"needTime" : 733117,
"needFetch" : 0,
"isEOF" : 1,
"keysExamined" : 733110,
"fetches" : 720806,
"parsedTextQuery" : {
"terms" : [
"all",
"the",
"light",
"we",
"cannot",
"see"
],
"negatedTerms" : [],
"phrases" : [],
"negatedPhrases" : []
},
"children" : []
}
}
完全搜索
时间：0.06秒
查询：db.items.find（{“n”：“我们看不见的所有光”}）
解释：
{
"cursor" : "BtreeCursor n_1",
"isMultiKey" : false,
"n" : 2,
"nscannedObjects" : 2,
"nscanned" : 2583249,
"nscannedObjectsAllPlans" : 2,
"nscannedAllPlans" : 2583249,
"scanAndOrder" : false,
"indexOnly" : false,
"nYields" : 20183,
"nChunkSkips" : 0,
"millis" : 9056,
"indexBounds" : {
"n" : [
[
"",
{}
],
[
/All the Light We Cannot See/i,
/All the Light We Cannot See/i
]
]
},
"server" : "myserver.com",
"filterSet" : false,
"stats" : {
"type" : "FETCH",
"works" : 2583250,
"yields" : 20183,
"unyields" : 20183,
"invalidates" : 0,
"advanced" : 2,
"needTime" : 2583247,
"needFetch" : 0,
"isEOF" : 1,
"alreadyHasObj" : 0,
"forcedFetches" : 0,
"matchTested" : 0,
"children" : [
{
"type" : "IXSCAN",
"works" : 2583249,
"yields" : 20183,
"unyields" : 20183,
"invalidates" : 0,
"advanced" : 2,
"needTime" : 2583247,
"needFetch" : 0,
"isEOF" : 1,
"keyPattern" : "{ n: 1.0 }",
"isMultiKey" : 0,
"boundsVerbose" : "field #0['n']: [\"\", {}), [/All the Light We Cannot See/i, /All the Light We Cannot See/i]",
"yieldMovedCursor" : 0,
"dupsTested" : 0,
"dupsDropped" : 0,
"seenInvalidated" : 0,
"matchTested" : 2,
"keysExamined" : 2583249,
"children" : []
}
]
}
}
正则表达式搜索
时间：9.64秒
查询：db.items.find（{“n”：/我们看不到的所有光/ i}）
解释：
{
"cursor" : "BtreeCursor n_1",
"isMultiKey" : false,
"n" : 2,
"nscannedObjects" : 2,
"nscanned" : 2583249,
"nscannedObjectsAllPlans" : 2,
"nscannedAllPlans" : 2583249,
"scanAndOrder" : false,
"indexOnly" : false,
"nYields" : 20183,
"nChunkSkips" : 0,
"millis" : 9056,
"indexBounds" : {
"n" : [
[
"",
{}
],
[
/All the Light We Cannot See/i,
/All the Light We Cannot See/i
]
]
},
"server" : "myserver.com0",
"filterSet" : false,
"stats" : {
"type" : "FETCH",
"works" : 2583250,
"yields" : 20183,
"unyields" : 20183,
"invalidates" : 0,
"advanced" : 2,
"needTime" : 2583247,
"needFetch" : 0,
"isEOF" : 1,
"alreadyHasObj" : 0,
"forcedFetches" : 0,
"matchTested" : 0,
"children" : [
{
"type" : "IXSCAN",
"works" : 2583249,
"yields" : 20183,
"unyields" : 20183,
"invalidates" : 0,
"advanced" : 2,
"needTime" : 2583247,
"needFetch" : 0,
"isEOF" : 1,
"keyPattern" : "{ n: 1.0 }",
"isMultiKey" : 0,
"boundsVerbose" : "field #0['n']: [\"\", {}), [/All the Light We Cannot See/i, /All the Light We Cannot See/i]",
"yieldMovedCursor" : 0,
"dupsTested" : 0,
"dupsDropped" : 0,
"seenInvalidated" : 0,
"matchTested" : 2,
"keysExamined" : 2583249,
"children" : []
}
]
}
}
请注意，确切的搜索结果是最快的。我不明白的是，如果搜索中涉及的字段被编入索引，那么AND搜索需要花费更多时间。我的猜测是MongoDB文本索引没有针对AND进行优化，在最简单的表达式（OR搜索）中，它比常规单字段索引慢得多。
这里有两个问题：