RediSearch Movie App Demo

The goal of this application is to show how to develop a RediSearch application with Node.js.

This application uses RediSearch that is based on Redis.

This application exposes various endpoint that are directly consumable in a front end.

How it works?

How the data is stored:

As a Redis developer, one of the first things to look when building your application is to define the structure of the key and data (data design/data modeling).

A common way of defining the keys in Redis is to use specific patterns in them. For example in this application where the database will probably deal with various business objects: movies, actors, theaters, users, ... we can use the following pattern:

• business_object:key

For example:

• movie:001 for the movie with the id 001
• user:001 the user with the id 001

and for the movies information you should use a Redis Hash.

A Redis Hash allows the application to structure all the movie attributes in individual fields; also RediSearch will index the fields based on the index definition.

Movies

The file /redisearch-docker/dataset/import_movies.redis is a script that creates 922 Hashes.

The movie hashes contain the following fields.

• movie:id : The unique ID of the movie, internal to this database (used as the key of the hash)
• title : The title of the movie.
• plot : A summary of the movie.
• genre : The genre of the movie, for now a movie will only have a single genre.
• release_year : The year the movie was released as a numerical value.
• rating : A numeric value representing the public's rating for this movie.
• votes : Number of votes.
• poster : Link to the movie poster.
• imdb_id : id of the movie in the IMDB database.

Sample Data: movie:521

Field	Value
title	Spider-Man
plot	When bitten by a genetically modified spider a nerdy shy and awkward high school student gains spider-like abilities that he eventually must use to fight evil as a superhero after tragedy befalls his family.
genre	Action
release_year	2002
rating	7.3
votes	662219
poster	https://m.media-amazon.com/images/M/MV5BZDEyN2NhMjgtMjdhNi00MmNlLWE5YTgtZGE4MzNjMTRlMGEwXkEyXkFqcGdeQXVyNDUyOTg3Njg@._V1_SX300.jpg
imdb_id	tt0145487

Actors

The file /redisearch-docker/dataset/import_actors.redis is a script that creates 1319 Hashes.

The movie hashes contain the following fields.

• actor:id : The unique ID of the actor
• first_name : The first name of the actor.
• last_name : The last name of the actor.
• date_of_birth : The birth year of the actor

Sample Data: actor:521

Field	Value
first_name	Renee
last_name	Olstead
date_of_birth	1989

Users

The file /redisearch-docker/dataset/import_users.redis is a script that creates 5996 Hashes.

The user hashes contain the following fields.

• user:id : The unique ID of the user.
• first_name : The first name of the user.
• last_name : The last name of the user.
• email : The email of the user.
• gender : The gender of the user (female/male).
• country : The country name of the user.
• country_code : The country code of the user.
• city : The city of the user.
• longitude : The longitude of the user.
• latitude : The latitude of the user.
• last_login : The last login time for the user, as EPOC time.
• ip_address : The IP address of the user.

Sample Data: user:3233

Field	Value
first_name	Rosetta
last_name	Olyff
email	rolyff6g@163.com
gender	female
country	China
country_code	CN
city	Huangdao
longitude	120.04619
latitude	35.872664
last_login	1570386621
ip_address	218.47.90.79

How the data is accessed:

One of the goals of RediSearch is to provide rich querying capabilities such as:

• simple and complex conditions
• sorting
• pagination
• counting

Conditions

The best way to start to work with RediSearch query capabilities is to look at the various conditions options.

Find all the movies that contain the word 'heat' or related to 'heat'

> FT.SEARCH "idx:movie" "heat" RETURN 2 title plot

1) (integer) 4
2) "movie:1141"
3) 1) "title"
   2) "Heat"
   3) "plot"
   4) "A group of professional bank robbers start to feel the heat from police when they unknowingly leave a clue at their latest heist."
4) "movie:818"
5) 1) "title"
   2) "California Heat"
   3) "plot"
   4) "A lifeguard bets he can be true to just one woman."
6) "movie:736"
7) 1) "title"
   2) "Chicago Justice"
   3) "plot"
   4) "The State's Attorney's dedicated team of prosecutors and investigators navigates heated city politics and controversy head-on,while fearlessly pursuing justice."
8) "movie:1109"
9) 1) "title"
   2) "Love & Hip Hop: Miami"
   3) "plot"
   4) "'Love and Hip Hop Miami' turns up the heat and doesn't hold back in making the 305 the place to be. Multi-platinum selling hip-hop legend Trick Daddy is back in the studio collaborating ..."

The first line contains the number of documents (4) that match the query condition, then the list of movies.

This query is a "fieldless" condition, this means that the query engine has:

• searched in all the TEXT fields of the index(title and plot)
• for the word heat and related words, this is why the movie:736 is returned since it has the word heated in the plot (stemming)
• returned the result sorted by score, remember that the title has a weight of 1.0, and the plot a weight of 0.5. So when the word or related words are found in the title the score is larger.

---

Find all the movies with a title that contains the word 'heat' or related to 'heat'

In this case you have to set the criteria to a the field title using the @title notation.

> FT.SEARCH "idx:movie" "@title:heat" RETURN 2 title plot
1) (integer) 2
2) "movie:1141"
3) 1) "title"
   2) "Heat"
   3) "plot"
   4) "A group of professional bank robbers start to feel the heat from police when they unknowingly leave a clue at their latest heist."
4) "movie:818"
5) 1) "title"
   2) "California Heat"
   3) "plot"
   4) "A lifeguard bets he can be true to just one woman."

So only 2 movies are returned.

---

Find all the movies where the title contains 'heat' and does NOT contains 'california'

For this you add parentheses around the field condition and add the - sign to 'california'.

> FT.SEARCH "idx:movie" "@title:(heat -california)" RETURN 2 title plot
1) (integer) 1
2) "movie:1141"
3) 1) "title"
   2) "Heat"
   3) "plot"
   4) "A group of professional bank robbers start to feel the heat from police when they unknowingly leave a clue at their latest heist."

Only one movie is returned.

If you do not put the ( .. ) the -california condition will be applied to all the text fields.

You can do test this with the following queries:

> FT.SEARCH "idx:movie" "@title:(heat -woman)" RETURN 2 title plot

> FT.SEARCH "idx:movie" "@title:heat -woman" RETURN 2 title plot

As you can see the first query only searches for woman in the title and returns two movies "Heat" and "California Heat", where the second query eliminates "California Heat" from the list since the plot contains the word woman.

---

Find all the 'Drama' movies that have 'heat' in the title

As you have seen earlier the movie index contains:

• the title and plot as TEXT
• the genre as TAG.

You saw earlier how to place a condition on a TEXT field.

The TAG is a little bit different as the index engine does not do any stemming.

To set a condition on this field you must use the @field:{value} notation, the {...} indicates that it is a TAG condition

>  FT.SEARCH "idx:movie" "@title:(heat) @genre:{Drama} " RETURN 3 title plot genre
1) (integer) 1
2) "movie:1141"
3) 1) "title"
   2) "Heat"
   3) "plot"
   4) "A group of professional bank robbers start to feel the heat from police when they unknowingly leave a clue at their latest heist."
   5) "genre"
   6) "Drama"

As you can see this query applies conditions to two different fields with an exact match on the TAG.

TAG is the structure to use when you want to do exact matches on strings/words.

Find all the 'Drama' or 'Comedy' movies that have 'heat' in the title

This is similar to the previous query, you can pass a list of values with the | to represent the OR.

> FT.SEARCH "idx:movie" "@title:(heat)  @genre:{Drama|Comedy} " RETURN 3 title plot genre

1) (integer) 2
2) "movie:1141"
3) 1) "title"
   2) "Heat"
   3) "plot"
   4) "A group of professional bank robbers start to feel the heat from police when they unknowingly leave a clue at their latest heist."
   5) "genre"
   6) "Drama"
4) "movie:818"
5) 1) "title"
   2) "California Heat"
   3) "plot"
   4) "A lifeguard bets he can be true to just one woman."
   5) "genre"
   6) "Comedy"

You can also put the '|' between all the conditions to search for example all movies that have "heat" in the title, or that are Comedy or that are Drama. The query will look like:

FT.SEARCH "idx:movie" "@title:(heat) | @genre:{Drama|Comedy} " RETURN 3 title plot genre

---

Find all 'Mystery' OR 'Thriller' movies, released in 2014 OR 2018

In this query, the new item is the query on a numeric field (release_year).

Like before, for the condition you have to use the @field: notation, but for a numeric field you have to put the interval of the condition.

In this query it will be two conditions with an OR (|).

> FT.SEARCH "idx:movie" "@genre:{Mystery|Thriller} (@release_year:[2018 2018] | @release_year:[2014 2014] )"   RETURN 3 title release_year genre

1) (integer) 3
2) "movie:461"
3) 1) "title"
   2) "The Boat ()"
   3) "release_year"
   4) "2018"
   5) "genre"
   6) "Mystery"
4) "movie:65"
5) 1) "title"
   2) "The Loft"
   3) "release_year"
   4) "2014"
   5) "genre"
   6) "Mystery"
6) "movie:989"
7) 1) "title"
   2) "Los Angeles Overnight"
   3) "release_year"
   4) "2018"
   5) "genre"
   6) "Thriller"

---

Summary

• Fieldless queries apply to all TEXT fields and use the words and their base form (stemming)
• To apply a condition to a specific field you must use the @field: notation
• Multiple conditions are "intersection" (AND condition), to do a "union" (OR condition), you have to use the "|" character.

---

Sort

A very common use case when querying data is to sort the data on a specific field, and paginate over the result.

Query all the `Action` movies, sorted by release year from most recent to the oldest

> FT.SEARCH "idx:movie" "@genre:{Action}"  SORTBY release_year DESC RETURN 2 title release_year
 1) (integer) 186
 2) "movie:360"
 3) 1) "release_year"
    2) "2019"
    3) "title"
    4) "Spider-Man: Far from Home"
 ...
20) "movie:278"
21) 1) "release_year"
    2) "2016"
    3) "title"
    4) "Mechanic: Resurrection"

The first line contains the number of documents (186) that match the query condition.

The FT.SEARCH command, by default, returns the first ten documents. You will see in the next query how to paginate.

You can only use one SORTBY clause in an FT.SEARCH query, if you want to sort on multiple fields, for example sorting movies by genre ascending and release_year descending, you have to use an FT.AGGREGATE, this is covered in the next section.

Note: The field used in the SORTBY should be part of the index schema and defined as SORTABLE.

---

Paginate

Query all the `Action` movies, sorted by release year from the oldest to the most recent one, returning the record by batch of 100 movies

> FT.SEARCH "idx:movie" "@genre:{Action}" LIMIT 0 100  SORTBY release_year ASC RETURN 2 title release_year
  1) (integer) 186
  2) "movie:892"
  3) 1) "release_year"
     2) "1966"
     3) "title"
     4) "Texas,Adios"
...
200) "movie:12"
201) 1) "release_year"
     2) "2014"
     3) "title"
     4) "Fury"

The result is very similar to the previous query:

• 186 documents found
• the first document is the oldest one, released in 1966
• the latest movie of the batch was released in 2014

To paginate to the next batch you need to change the limit as follows:

> FT.SEARCH "idx:movie" "@genre:{Action}" LIMIT 100 200  SORTBY release_year ASC RETURN 2 title release_year

---

---

Count

Count the number of 'Action' movies

Based on the sample queries that you have seen earlier, if you specify LIMIT 0 0 it will give you the number of documents based on the query condition.

> FT.SEARCH "idx:movie" "@genre:{Action}" LIMIT 0 0

1) (integer) 186

---

Count the number of 'Action' movies released in 2017

Based on the sample queries that you have seen earlier, if you specify LIMIT 0 0 it will give you the number of documents based on the query condition.

> FT.SEARCH "idx:movie" "@genre:{Action}" FILTER release_year 2017 2017 LIMIT 0 0

1) (integer) 5

You can also use the following syntax:

> FT.SEARCH "idx:movie" "@genre:{Action} @release_year:[2017 2017]" LIMIT 0 0

1) (integer) 5

---

---

How to run it locally?

Running the application in Docker

The application and all the services, including RediSearch, are available as a Docker Compose application.

To run the application:

> docker-compose up --force-recreate --build

This Docker Compose will start:

1. RediSearch instance on port 6380, and import all movies, actors and create indexes
2. The Node.js REST Service available on port 8086
3. The frontend on port 8084

Once started you can access the application and its services using the following URLs:

• http://localhost:8084
• http://localhost:8086/api/1.0/movies/search?q=star&offset=0&limit=10

Stop and Delete Everything

Run the following command to delete the containers & images:

> docker-compose down -v --rmi local --remove-orphans