1
00:00:02,170 --> 00:00:09,670
Speaking of multikey indexes, there is a special kind of multikey index, a text index.

2
00:00:09,790 --> 00:00:16,870
Let's say we have this text stored in a field in our document, could be the product description of a

3
00:00:16,870 --> 00:00:18,220
product.

4
00:00:18,220 --> 00:00:24,670
Now if you want to search that text, we saw before that we can use the regex operator and that

5
00:00:24,670 --> 00:00:27,460
is however not a really great way of searching text,

6
00:00:27,480 --> 00:00:35,780
it offers a very low performance. Better is to use a text index and a text index is a special kind of

7
00:00:35,780 --> 00:00:42,590
index supported by mongodb which will essentially turn this text into an array of single words and

8
00:00:42,590 --> 00:00:43,940
it will store it as such,

9
00:00:43,940 --> 00:00:50,150
so it stores it essentially as if you had an array of these single words, one extra thing it does for

10
00:00:50,150 --> 00:00:58,970
you is it removes all the stop words and it stems all words, so that you have an array of keywords essentially

11
00:00:59,210 --> 00:01:05,450
and things like is or the or a are not stored there because that is typically something you don't search

12
00:01:05,450 --> 00:01:07,690
for because it's all over the place,

13
00:01:07,700 --> 00:01:11,060
the keywords are one matter for text searches typically.

14
00:01:11,420 --> 00:01:14,020
So this is the text index,

15
00:01:14,030 --> 00:01:16,560
now let's have a look at such a text index

16
00:01:16,580 --> 00:01:24,360
and for that I'll, whoops not index, insert many products into a newly created products collection here.

17
00:01:24,370 --> 00:01:27,280
So the first product will have a name or

18
00:01:27,280 --> 00:01:37,880
let's say a title to mix it up, title, a book and it will have a description, this is an awesome

19
00:01:37,910 --> 00:01:41,030
book about a young artist.

20
00:01:42,050 --> 00:01:53,350
The second document I add here should also describe product will have a title of red t-shirt and there,

21
00:01:53,890 --> 00:02:06,070
we'll have a, whoops we'll have a description of this t-shirt is red and it's pretty awesome.

22
00:02:10,410 --> 00:02:14,090
So after finding that missing quotation mark, it now worked fine for me

23
00:02:14,310 --> 00:02:20,380
and now if I have a look at my products and I pretty print this, here are my two products.

24
00:02:20,790 --> 00:02:22,700
Now let's use a text index,

25
00:02:22,710 --> 00:02:28,260
let's use it on the description field let's say. So first of all, I'll put our products again so that we

26
00:02:28,260 --> 00:02:29,190
can see them

27
00:02:29,400 --> 00:02:34,640
and now we can create a new text index, again with create index as we create all indexes

28
00:02:35,010 --> 00:02:38,110
and I say I want to create it on description and now important,

29
00:02:38,400 --> 00:02:40,100
don't add 1 or -1,

30
00:02:40,110 --> 00:02:47,460
you could do this but then it would simply index this as a single field index and you could then search

31
00:02:47,460 --> 00:02:54,010
for exactly this text to utilize the index but not for individual keywords. You need the text index

32
00:02:54,150 --> 00:02:56,410
so that mongodb splits this up,

33
00:02:56,460 --> 00:02:58,880
so using this will not work so

34
00:02:58,980 --> 00:03:07,580
let's quickly drop that index like this and let's recreate the index but now not with a 1 but with the

35
00:03:07,580 --> 00:03:14,270
special text keyword, text in quotation marks. This will create a text index which is a special kind of

36
00:03:14,270 --> 00:03:21,170
index where mongodb will go ahead and as I mentioned, remove all the stop words and store all the

37
00:03:21,170 --> 00:03:25,610
keywords in an array essentially,

38
00:03:25,730 --> 00:03:27,000
so let's have a look at this.

39
00:03:27,020 --> 00:03:29,230
This is the data we have in there in general

40
00:03:29,540 --> 00:03:33,690
and now let's use products and find and

41
00:03:33,830 --> 00:03:34,580
here

42
00:03:35,000 --> 00:03:38,260
let's now use the special $text key

43
00:03:39,370 --> 00:03:45,990
and search and for that, you pass a document as a value for $text and there, you need

44
00:03:46,000 --> 00:03:46,960
$search.

45
00:03:46,960 --> 00:03:52,000
Now you might be wondering why do I not need to specify the field in which I want to search,

46
00:03:52,000 --> 00:03:54,120
why don't I have to add description,

47
00:03:54,310 --> 00:03:57,290
instead we just add hey I want to search for some text.

48
00:03:57,340 --> 00:04:04,300
The reason for that is that you may only have one text index per collection because text indexes are

49
00:04:04,300 --> 00:04:06,610
pretty expensive as you can imagine,

50
00:04:06,610 --> 00:04:11,410
if you have a lot of long text that has to be split up, you don't want to do this like 10 times per

51
00:04:11,410 --> 00:04:16,080
collection and therefore, you only have one text index where this could look into.

52
00:04:16,240 --> 00:04:21,730
You can actually merge multiple fields into one text index as I will show you in a second and you will

53
00:04:21,730 --> 00:04:24,020
then look through all of them automatically

54
00:04:24,250 --> 00:04:25,990
but you can only do it like this,

55
00:04:25,990 --> 00:04:31,050
you can't say hey I want to search for text and description like this, this won't work.

56
00:04:31,090 --> 00:04:34,640
So now for search, we simply enter the words we want to look for

57
00:04:34,870 --> 00:04:38,480
like awesome and the casing is not important here by the way,

58
00:04:38,590 --> 00:04:41,380
everything is stored as lowercase.

59
00:04:41,380 --> 00:04:49,360
If I hit enter and I pretty print this, you'll see I find both products because in both products, we have

60
00:04:49,360 --> 00:04:51,150
the term awesome.

61
00:04:51,650 --> 00:04:53,090
Now let's repeat this

62
00:04:53,140 --> 00:04:56,540
but let's now just search for the term, book.

63
00:04:56,680 --> 00:05:00,960
So here I'll search for book

64
00:05:01,130 --> 00:05:06,780
and now you see I only get well this text or this document where we have book in the text,

65
00:05:06,800 --> 00:05:11,860
now what if I search for red book, I have red in the document.

66
00:05:12,170 --> 00:05:17,810
Well then I find both again because this is now actually not treated as one connected phrase where

67
00:05:17,810 --> 00:05:23,150
it would look for a red book but it simply looks for documents that have a red text or some red

68
00:05:23,180 --> 00:05:30,120
in a text and for documents that have book in a text. Of course you can also search for a specific phrase

69
00:05:30,120 --> 00:05:36,920
though, you can search for phrases by wrapping that phrase in double quotes and since we are in

70
00:05:36,920 --> 00:05:41,500
double quotes already, we have to escape them with a backslash double quote.

71
00:05:41,690 --> 00:05:46,580
We have this at the beginning and at the end of the phrase and now we don't find anything because

72
00:05:46,580 --> 00:05:51,380
we have no red book phrase anywhere in our text, for example

73
00:05:51,380 --> 00:05:53,210
awesome book would work though,

74
00:05:53,210 --> 00:05:58,250
so if I look for the awesome book phrase, we would find this document because we have awesome book right

75
00:05:58,250 --> 00:05:59,080
there.

76
00:05:59,090 --> 00:06:02,870
Now this is really powerful and much faster than regular expressions,

77
00:06:03,050 --> 00:06:07,450
so this is definitely the way to go if you need to look for keywords in text.