Please go through this question of mine:
MongoDB $group and explicit group formation with computed column
请仔细阅读我的这个问题:MongoDB $ group和带有计算列的显式组形成
But this time, I need to compare strings, not numbers.The CASE
query must have a LIKE
: CASE WHEN source LIKE '%Web%' THEN 'Web'
I then need to group by source. How to write this in Mongo? I am trying the following but not sure if $regex
is supported inside $cond
. By the way, is there a list of valid operators inside $cond
somewhere? Looks like $cond
isn't very fond of me :)
但这一次,我需要比较字符串,而不是数字.CASE查询必须有一个LIKE:CASE WHEN来源LIKE'%Web%'那么'Web'然后需要按来源分组。如何在Mongo中写这个?我正在尝试以下但不确定$ cond中是否支持$ regex。顺便问一下,$ cond里面有一个有效的运算符列表吗?看起来$ cond不是很喜欢我:)
db.Twitter.aggregate(
{ $project: {
"_id":0,
"Source": {
$cond: [
{ $regex:['$source','/.* Android.*/'] },
'Android',
{ $cond: [
{ $eq: ['$source', 'web'] }, 'Web', 'Others'
] }
]
}
} }
);
There're many other values that I need to write in there, doing a deeper nesting. This is just an example with just 'Android' and 'Web' for the sake of brevity. I have tried both with $eq
and $regex
. Using $regex
gives error of invalid operator whereas using $eq
doesn't understand the regex expression and puts everything under 'Others'. If this is possible with regex, kindly let me know how to write it for case-insensitive match.
我需要在那里写下许多其他值,进行更深层次的嵌套。为简洁起见,这只是“Android”和“Web”的一个例子。我用$ eq和$ regex试过了。使用$ regex会产生无效运算符的错误,而使用$ eq则无法理解正则表达式并将所有内容置于“其他”下。如果这可以使用正则表达式,请告诉我如何编写它以区分大小写匹配。
Thanks for any help :-)
谢谢你的帮助 :-)
2 个解决方案
#1
1
Well, it still seems to be not even scheduled to be implemented :( https://jira.mongodb.org/browse/SERVER-8892
好吧,它似乎甚至没有安排实施:( https://jira.mongodb.org/browse/SERVER-8892
I'm using 2.6 and took a peek on 3.0, but it's just not there.
我正在使用2.6并看一眼3.0,但它不存在。
There's one workaround though, if you can project your problem onto a stable substring. Then you can $substr the field and use multiple nested $cond. It's awkward, but it works.
但是,如果您可以将问题投影到稳定的子字符串上,则有一种解决方法。然后你可以$ substr字段并使用多个嵌套的$ cond。这很尴尬,但它确实有效。
#2
0
Maybe you can try it with MapReduce.
也许你可以尝试使用MapReduce。
var map = function()
{
var reg1=new RegExp("(Android)+");
var reg2=new RegExp("(web)+");
if (reg1.test(this.source)){
emit(this._id,'Android');
}
else if (reg2.test(this.source))
{
emit(this._id,'web');
}
}
var reduce = function (key,value){
var reduced = {
id:key,
source:value
}
return reduced;
}
db.Twitter.mapReduce(map,reduce,{out:'map_reduce_result'});
db.map_reduce_result.find();
You can use JavaScript regular expresions instead of MongoDB $regex.
您可以使用JavaScript常规表达式而不是MongoDB $ regex。
#1
1
Well, it still seems to be not even scheduled to be implemented :( https://jira.mongodb.org/browse/SERVER-8892
好吧,它似乎甚至没有安排实施:( https://jira.mongodb.org/browse/SERVER-8892
I'm using 2.6 and took a peek on 3.0, but it's just not there.
我正在使用2.6并看一眼3.0,但它不存在。
There's one workaround though, if you can project your problem onto a stable substring. Then you can $substr the field and use multiple nested $cond. It's awkward, but it works.
但是,如果您可以将问题投影到稳定的子字符串上,则有一种解决方法。然后你可以$ substr字段并使用多个嵌套的$ cond。这很尴尬,但它确实有效。
#2
0
Maybe you can try it with MapReduce.
也许你可以尝试使用MapReduce。
var map = function()
{
var reg1=new RegExp("(Android)+");
var reg2=new RegExp("(web)+");
if (reg1.test(this.source)){
emit(this._id,'Android');
}
else if (reg2.test(this.source))
{
emit(this._id,'web');
}
}
var reduce = function (key,value){
var reduced = {
id:key,
source:value
}
return reduced;
}
db.Twitter.mapReduce(map,reduce,{out:'map_reduce_result'});
db.map_reduce_result.find();
You can use JavaScript regular expresions instead of MongoDB $regex.
您可以使用JavaScript常规表达式而不是MongoDB $ regex。