Navigation

$regexFindAll (aggregation)

Definition

$regexFindAll

New in version 4.2.

Provides regular expression (regex) pattern matching capability in aggregation expressions. The operator returns an array of documents that contains information on each match. If a match is not found, returns an empty array.

MongoDB uses Perl compatible regular expressions (i.e. “PCRE” ) version 8.41 with UTF-8 support.

Prior to MongoDB 4.2, aggregation pipeline can only use the query operator $regex in the $match stage. For more information on using regex in a query, see $regex.

Syntax

The $regexFindAll operator has the following syntax:

{ $regexFindAll: { input: <expression> , regex: <expression>, options: <expression> } }
Field Description
input

The string on which you wish to apply the regex pattern. Can be a string or any valid expression that resolves to a string.

regex

The regex pattern to apply. Can be any valid expression that resolves to either a string or regex pattern /<pattern>/. When using the regex /<pattern>/, you can also specify the regex options i and m (but not the s or x options):

  • "pattern"
  • /<pattern>/
  • /<pattern>/<options>

Alternatively, you can also specify the regex options with the options field. To specify the s or x options, you must use the options field.

You cannot specify options in both the regex and the options field.

options

Optional. The following <options> are available for use with regular expression.

Note

You cannot specify options in both the regex and the options field.

Option Description
i Case insensitivity to match both upper and lower cases. You can specify the option in the options field or as part of the regex field.
m

For patterns that include anchors (i.e. ^ for the start, $ for the end), match at the beginning or end of each line for strings with multiline values. Without this option, these anchors match at beginning or end of the string.

If the pattern contains no anchors or if the string value has no newline characters (e.g. \n), the m option has no effect.

x

“Extended” capability to ignore all white space characters in the pattern unless escaped or included in a character class.

Additionally, it ignores characters in-between and including an un-escaped hash/pound (#) character and the next new line, so that you may include comments in complicated patterns. This only applies to data characters; white space characters may never appear within special character sequences in a pattern.

The x option does not affect the handling of the VT character (i.e. code 11).

You can specify the option only in the options field.

s

Allows the dot character (i.e. .) to match all characters including newline characters.

You can specify the option only in the options field.

Returns

The operator returns an array:

  • If the operator does not find a match, the operator returns an empty array.

  • If the operator finds a match, the operator returns an array of documents that contains the following information for each match:

    • the matching string in the input,
    • the code point index (not byte index) of the matching string in the input, and
    • An array of the strings that corresponds to the groups captured by the matching string. Capturing groups are specified with unescaped parenthesis () in the regex pattern.
    [ { "match" : <string>, "idx" : <num>, "captures" : <array of strings> }, ... ]
    

Behavior

$regexFindAll and Collation

$regexFindAll ignores the collation specified for the collection, db.collection.aggregate(), and the index, if used.

For example, the create a sample collection with collation strength 1 (i.e. compare base character only and ignore other differences such as case and diacritics):

db.createCollection( "myColl", { collation: { locale: "fr", strength: 1 } } )

Insert the following documents:

db.myColl.insertMany([
   { _id: 1, category: "café" },
   { _id: 2, category: "cafe" },
   { _id: 3, category: "cafE" }
])

Using the collection’s collation, the following operation performs a case-insensitive and diacritic-insensitive match:

db.myColl.aggregate( [ { $match: { category: "cafe" } } ] )

The operation returns the following 3 documents:

{ "_id" : 1, "category" : "café" }
{ "_id" : 2, "category" : "cafe" }
{ "_id" : 3, "category" : "cafE" }

However, the aggregation expression $regexFind ignores collation; that is, the following regular expression pattern matching examples are case-sensitive and diacritic sensitive:

db.myColl.aggregate( [ { $addFields: { results: { $regexFindAll: { input: "$category", regex: /cafe/ }  } } } ] )
db.myColl.aggregate(
   [ { $addFields: { results: { $regexFindAll: { input: "$category", regex: /cafe/ }  } } } ],
   { collation: { locale: "fr", strength: 1 } }       // Ignored in the $regexFindAll
)

Both operations return the following:

{ "_id" : 1, "category" : "café", "results" : [ ] }
{ "_id" : 2, "category" : "cafe", "results" : [ { "match" : "cafe", "idx" : 0, "captures" : [ ] } ] }
{ "_id" : 3, "category" : "cafE", "results" : [ ] }

To perform a case-insensitive regex pattern matching, use the i Option instead. See i Option for an example.

captures Output Behavior

If your regex pattern contains capture groups and the pattern finds a match in the input, the captures array in the results corresponds to the groups captured by the matching string. Capture groups are specified with unescaped parentheses () in the regex pattern. The length of the captures array equals the number of capture groups in the pattern and the order of the array matches the order in which the capture groups appear.

Create a sample collection named contacts with the following documents:

db.contacts.insertMany([
  { "_id": 1, "fname": "Carol", "lname": "Smith", "phone": "718-555-0113" },
  { "_id": 2, "fname": "Daryl", "lname": "Doe", "phone": "212-555-8832" },
  { "_id": 3, "fname": "Polly", "lname": "Andrews", "phone": "208-555-1932" },
  { "_id": 4, "fname": "Colleen", "lname": "Duncan", "phone": "775-555-0187" },
  { "_id": 5, "fname": "Luna", "lname": "Clarke", "phone": "917-555-4414" }
])

The following pipeline applies the regex pattern /(C(ar)*)ol/ to the fname field:

db.contacts.aggregate([
  {
    $project: {
      returnObject: {
        $regexFindAll: { input: "$fname", regex: /(C(ar)*)ol/ }
      }
    }
  }
])

The regex pattern finds a match with fname values Carol and Colleen:

{ "_id" : 1, "returnObject" : [ { "match" : "Carol", "idx" : 0, "captures" : [ "Car", "ar" ] } ] }
{ "_id" : 2, "returnObject" : [ ] }
{ "_id" : 3, "returnObject" : [ ] }
{ "_id" : 4, "returnObject" : [ { "match" : "Col", "idx" : 0, "captures" : [ "C", null ] } ] }
{ "_id" : 5, "returnObject" : [ ] }

The pattern contains the capture group (C(ar)*) which contains the nested group (ar). The elements in the captures array correspond to the two capture groups. If a matching document is not captured by a group (e.g. Colleen and the group (ar)), $regexFindAll replaces the group with a null placeholder.

As shown in the previous example, the captures array contains an element for each capture group (using null for non-captures). Consider the following example which searches for phone numbers with New York City area codes by applying a logical or of capture groups to the phone field. Each group represents a New York City area code:

db.contacts.aggregate([
  {
    $project: {
      nycContacts: {
        $regexFindAll: { input: "$phone", regex: /^(718).*|^(212).*|^(917).*/ }
      }
    }
  }
])

For documents which are matched by the regex pattern, the captures array includes the matching capture group and replaces any non-capturing groups with null:

{ "_id" : 1, "nycContacts" : [ { "match" : "718-555-0113", "idx" : 0, "captures" : [ "718", null, null ] } ] }
{ "_id" : 2, "nycContacts" : [ { "match" : "212-555-8832", "idx" : 0, "captures" : [ null, "212", null ] } ] }
{ "_id" : 3, "nycContacts" : [ ] }
{ "_id" : 4, "nycContacts" : [ ] }
{ "_id" : 5, "nycContacts" : [ { "match" : "917-555-4414", "idx" : 0, "captures" : [ null, null, "917" ] } ] }

Examples

$regexFindAll and Its Options

To illustrate the behavior of the $regexFindAll operator as discussed in this example, create a sample collection products with the following documents:

db.products.insertMany([
   { _id: 1, description: "Single LINE description." },
   { _id: 2, description: "First lines\nsecond line" },
   { _id: 3, description: "Many spaces before     line" },
   { _id: 4, description: "Multiple\nline descriptions" },
   { _id: 5, description: "anchors, links and hyperlinks" },
   { _id: 6, description: "métier work vocation" }
])

By default, $regexFindAll performs a case-sensitive match. For example, the following aggregation performs a case-sensitive $regexFindAll on the description field. The regex pattern /line/ does not specify any grouping:

db.products.aggregate([
   { $addFields: { returnObject: { $regexFindAll: { input: "$description", regex: /line/ } } } }
])

The operation returns the following:

{
   "_id" : 1,
   "description" : "Single LINE description.",
   "returnObject" : [ ]
}
{
   "_id" : 2,
   "description" : "First lines\nsecond line",
   "returnObject" : [ { "match" : "line", "idx" : 6, "captures" : [ ]}, { "match" : "line", "idx" : 19, "captures" : [ ] } ]
}
{
   "_id" : 3,
   "description" : "Many spaces before     line",
   "returnObject" : [ { "match" : "line", "idx" : 23, "captures" : [ ] } ]
}
{
   "_id" : 4,
   "description" : "Multiple\nline descriptions",
   "returnObject" : [ { "match" : "line", "idx" : 9, "captures" : [ ] }
] }

{
   "_id" : 5,
   "description" : "anchors, links and hyperlinks",
   "returnObject" : [ ]
}
{
   "_id" : 6,
   "description" : "métier work vocation",
   "returnObject" : [ ]
}

The following regex pattern /lin(e|k)/ specifies a grouping (e|k) in the pattern:

db.products.aggregate([
   { $addFields: { returnObject: { $regexFindAll: { input: "$description", regex: /lin(e|k)/ } } } }
])

The operation returns the following:

{
   "_id" : 1,
   "description" : "Single LINE description.",
   "returnObject": [ ]
}
{
   "_id" : 2,
   "description" : "First lines\nsecond line",
   "returnObject" : [ { "match" : "line", "idx" : 6, "captures" : [ "e" ] }, { "match" : "line", "idx" : 19, "captures" : [ "e" ] } ]
}
{
   "_id" : 3,
   "description" : "Many spaces before     line",
   "returnObject" : [ { "match" : "line", "idx" : 23, "captures" : [ "e" ] } ]
}
{
   "_id" : 4,
   "description" : "Multiple\nline descriptions",
   "returnObject" : [ { "match" : "line", "idx" : 9, "captures" : [ "e" ] } ]
}
{
   "_id" : 5,
   "description" : "anchors, links and hyperlinks",
   "returnObject" : [ { "match" : "link", "idx" : 9, "captures" : [ "k" ] }, { "match" : "link", "idx" : 24, "captures" : [ "k" ] } ]
}
{
   "_id" : 6,
   "description" : "métier work vocation",
   "returnObject" : [ ]
}

In the return option, the idx field is the code point index and not the byte index. To illustrate, consider the following example that uses the regex pattern /tier/:

db.products.aggregate([
   { $addFields: { returnObject: { $regexFindAll: { input: "$description", regex: /tier/ } } } }
])

The operation returns the following where only the last record matches the pattern and the returned idx is 2 (instead of 3 if using a byte index)

{ "_id" : 1, "description" : "Single LINE description.", "returnObject" : [ ] }
{ "_id" : 2, "description" : "First lines\nsecond line", "returnObject" : [ ] }
{ "_id" : 3, "description" : "Many spaces before     line", "returnObject" : [ ] }
{ "_id" : 4, "description" : "Multiple\nline descriptions", "returnObject" : [ ] }
{ "_id" : 5, "description" : "anchors, links and hyperlinks", "returnObject" : [ ] }
{ "_id" : 6, "description" : "métier work vocation",
             "returnObject" : [ { "match" : "tier", "idx" : 2, "captures" : [ ] } ] }

i Option

Note

You cannot specify options in both the regex and the options field.

To perform case-insensitive pattern matching, include the i option as part of the regex field or in the options field:

// Specify i as part of the regex field
{ $regexFindAll: { input: "$description", regex: /line/i } }

// Specify i in the options field
{ $regexFindAll: { input: "$description", regex: /line/, options: "i" } }
{ $regexFindAll: { input: "$description", regex: "line", options: "i" } }

For example, the following aggregation performs a case-insensitive $regexFindAll on the description field. The regex pattern /line/ does not specify any grouping:

db.products.aggregate([
   { $addFields: { returnObject: { $regexFindAll: { input: "$description", regex: /line/i } } } }
])

The operation returns the following documents:

{
   "_id" : 1,
   "description" : "Single LINE description.",
   "returnObject" : [ { "match" : "LINE", "idx" : 7, "captures" : [ ] } ]
}
{
   "_id" : 2,
   "description" : "First lines\nsecond line",
   "returnObject" : [ { "match" : "line", "idx" : 6, "captures" : [ ] }, { "match" : "line", "idx" : 19, "captures" : [ ] } ]
}
{
   "_id" : 3,
   "description" : "Many spaces before     line",
   "returnObject" : [ { "match" : "line", "idx" : 23, "captures" : [ ] } ]
}
{
   "_id" : 4,
   "description" : "Multiple\nline descriptions",
   "returnObject" : [ { "match" : "line", "idx" : 9, "captures" : [ ] } ]
}
{
   "_id" : 5,
   "description" : "anchors, links and hyperlinks",
   "returnObject" : [ ]
}
{ "_id" : 6, "description" : "métier work vocation", "returnObject" : [ ] }

m Option

Note

You cannot specify options in both the regex and the options field.

To match the specified anchors (e.g. ^, $) for each line of a multiline string, include the m option as part of the regex field or in the options field:

// Specify m as part of the regex field
{ $regexFindAll: { input: "$description", regex: /line/m } }

// Specify m in the options field
{ $regexFindAll: { input: "$description", regex: /line/, options: "m" } }
{ $regexFindAll: { input: "$description", regex: "line", options: "m" } }

The following example includes both the i and the m options to match lines starting with either the letter s or S for multiline strings:

db.products.aggregate([
   { $addFields: { returnObject: { $regexFindAll: { input: "$description", regex: /^s/im } } } }
])

The operation returns the following:

{
   "_id" : 1,
   "description" : "Single LINE description.",
   "returnObject" : [ { "match" : "S", "idx" : 0, "captures" : [ ] } ]
}
{
   "_id" : 2,
   "description" : "First lines\nsecond line",
   "returnObject" : [ { "match" : "s", "idx" : 12, "captures" : [ ] } ]
}
{
   "_id" : 3,
   "description" : "Many spaces before     line",
   "returnObject" : [ ]
}
{
   "_id" : 4,
   "description" : "Multiple\nline descriptions",
   "returnObject" : [ ]
}
{
   "_id" : 5,
   "description" : "anchors, links and hyperlinks",
   "returnObject" : [ ]
}
{ "_id" : 6, "description" : "métier work vocation", "returnObject" : [ ] }

x Option

Note

You cannot specify options in both the regex and the options field.

To ignore all unescaped white space characters and comments (denoted by the un-escaped hash # character and the next new-line character) in the pattern, include the s option in the options field:

// Specify x in the options field
{ $regexFindAll: { input: "$description", regex: /line/, options: "x" } }
{ $regexFindAll: { input: "$description", regex: "line", options: "x" } }

The following example includes the x option to skip unescaped white spaces and comments:

db.products.aggregate([
   { $addFields: { returnObject: { $regexFindAll: { input: "$description", regex: /lin(e|k) # matches line or link/, options:"x" } } } }
])

The operation returns the following:

{
   "_id" : 1,
   "description" : "Single LINE description.",
   "returnObject" : [ ]
}
{
   "_id" : 2,
   "description" : "First lines\nsecond line",
   "returnObject" : [ { "match" : "line", "idx" : 6, "captures" : [ "e" ] }, { "match" : "line", "idx" : 19, "captures" : [ "e" ] } ]
}
{
   "_id" : 3,
   "description" : "Many spaces before     line",
   "returnObject" : [ { "match" : "line", "idx" : 23, "captures" : [ "e" ] } ]
}
{
   "_id" : 4,
   "description" : "Multiple\nline descriptions",
   "returnObject" : [ { "match" : "line", "idx" : 9, "captures" : [ "e" ] } ]
}
{
   "_id" : 5,
   "description" : "anchors, links and hyperlinks",
   "returnObject" : [ { "match" : "link", "idx" : 9, "captures" : [ "k" ] }, { "match" : "link", "idx" : 24, "captures" : [ "k" ] } ]
}
{ "_id" : 6, "description" : "métier work vocation", "returnObject" : [ ] }

s Option

Note

You cannot specify options in both the regex and the options field.

To allow the dot character (i.e. .) in the pattern to match all characters including the new line character, include the s option in the options field:

// Specify s in the options field
{ $regexFindAll: { input: "$description", regex: /m.*line/, options: "s" } }
{ $regexFindAll: { input: "$description", regex: "m.*line", options: "s" } }

The following example includes the s option to allow the dot character (i.e. .) to match all characters including new line as well as the i option to perform a case-insensitive match:

db.products.aggregate([
   { $addFields: { returnObject: { $regexFindAll: { input: "$description", regex:/m.*line/, options: "si"  } } } }
])

The operation returns the following:

{
   "_id" : 1,
   "description" : "Single LINE description.",
   "returnObject" : [ ]
}
{
   "_id" : 2,
   "description" : "First lines\nsecond line",
   "returnObject" : [ ]
}
{
   "_id" : 3,
   "description" : "Many spaces before     line",
   "returnObject" : [ { "match" : "Many spaces before line", "idx" : 0, "captures" : [ ] } ]
}
{
   "_id" : 4,
   "description" : "Multiple\nline descriptions",
   "returnObject" : [ { "match" : "Multiple\nline", "idx" : 0, "captures" : [ ] } ]
}
{
   "_id" : 5,
   "description" : "anchors, links and hyperlinks",
   "returnObject" : [ ]
}
{ "_id" : 6, "description" : "métier work vocation", "returnObject" : [ ] }

Use $regexFindAll to Parse Email from String

Create a sample collection feedback with the following documents:

db.feedback.insertMany([
   { "_id" : 1, comment: "Hi, I'm just reading about MongoDB -- [email protected]"  },
   { "_id" : 2, comment: "I wanted to concatenate a string" },
   { "_id" : 3, comment: "How do I convert a date to string? Contact me at either [email protected] or [email protected]" },
   { "_id" : 4, comment: "It's just me. I'm testing.  [email protected]" }
])

The following aggregation uses the $regexFindAll to extract all emails from the comment field (case insensitive).

db.feedback.aggregate( [
    { $addFields: {
       "email": { $regexFindAll: { input: "$comment", regex: /[a-z0-9_.+-]+@[a-z0-9_.+-]+\.[a-z0-9_.+-]+/i } }
    } },
    { $set: { email: "$email.match"} }
] )
First Stage

The stage uses the $addFields stage to add a new field email to the document. The new field is an array that contains the result of performing the $regexFindAll on the comment field:

{ "_id" : 1, "comment" : "Hi, I'm just reading about MongoDB -- [email protected]", "email" : [ { "match" : "[email protected]", "idx" : 38, "captures" : [ ] } ] }
{ "_id" : 2, "comment" : "I wanted to concatenate a string", "email" : [ ] }
{ "_id" : 3, "comment" : "How do I convert a date to string? Contact me at either [email protected] or [email protected]", "email" : [ { "match" : "[email protected]", "idx" : 56, "captures" : [ ] }, { "match" : "[email protected]", "idx" : 75, "captures" : [ ] } ] }
{ "_id" : 4, "comment" : "It's just me. I'm testing.  [email protected]", "email" : [ { "match" : "[email protected]", "idx" : 28, "captures" : [ ] } ] }
Second Stage

The stage use the $set stage to reset the email array elements to the "email.match" value(s). If the current value of email is null, the new value of email is set to null.

{ "_id" : 1, "comment" : "Hi, I'm just reading about MongoDB -- [email protected]", "email" : [ "[email protected]" ] }
{ "_id" : 2, "comment" : "I wanted to concatenate a string", "email" : [ ] }
{ "_id" : 3, "comment" : "How do I convert a date to string? Contact me at either [email protected] or [email protected]", "email" : [ "[email protected]", "[email protected]" ] }
{ "_id" : 4, "comment" : "It's just me. I'm testing.  [email protected]", "email" : [ "[email protected]" ] }

Use Captured Groupings to Parse User Name

Create a sample collection feedback with the following documents:

db.feedback.insertMany([
   { "_id" : 1, comment: "Hi, I'm just reading about MongoDB -- [email protected]"  },
   { "_id" : 2, comment: "I wanted to concatenate a string" },
   { "_id" : 3, comment: "How do I convert a date to string? Contact me at either [email protected] or [email protected]" },
   { "_id" : 4, comment: "It's just me. I'm testing.  [email protected]" }
])

To reply to the feedback, assume you want to parse the local-part of the email address to use as the name in the greetings. Using the captured field returned in the $regexFindAll results, you can parse out the local part of each email address:

db.feedback.aggregate( [
    { $addFields: {
       "names": { $regexFindAll: { input: "$comment", regex: /([a-z0-9_.+-]+)@[a-z0-9_.+-]+\.[a-z0-9_.+-]+/i } },
    } },
    { $set: { names: { $reduce: { input:  "$names.captures", initialValue: [ ], in: { $concatArrays: [ "$$value", "$$this" ] } } } } }
] )
First Stage

The stage uses the $addFields stage to add a new field names to the document. The new field contains the result of performing the $regexFindAll on the comment field:

{
   "_id" : 1,
   "comment" : "Hi, I'm just reading about MongoDB -- [email protected]",
   "names" : [ { "match" : "[email protected]", "idx" : 38, "captures" : [ "aunt.arc.tica" ] } ]
}

{ "_id" : 2, "comment" : "I wanted to concatenate a string", "names" : [ ] }
{
   "_id" : 3,
   "comment" : "How do I convert a date to string? Contact me at either [email protected] or [email protected]",
   "names" : [
      { "match" : "[email protected]", "idx" : 56, "captures" : [ "cam" ] },
      { "match" : "[email protected]", "idx" : 75, "captures" : [ "c.dia" ] }
    ]
}
{
   "_id" : 4,
   "comment" : "It's just me. I'm testing.  [email protected]",
   "names" : [ { "match" : "[email protected]", "idx" : 28, "captures" : [ "fred" ] } ]
}
Second Stage

The stage use the $set stage with the $reduce operator to reset names to an array that contains the "$names.captures" elements.

{
   "_id" : 1,
   "comment" : "Hi, I'm just reading about MongoDB -- [email protected]",
   "names" : [ "aunt.arc.tica" ]
}
{ "_id" : 2, "comment" : "I wanted to concatenate a string", "names" : [ ] }
{
   "_id" : 3,
   "comment" : "How do I convert a date to string? Contact me at either [email protected] or [email protected]",
   "names" : [ "cam", "c.dia" ]
}
{
   "_id" : 4,
   "comment" : "It's just me. I'm testing.  [email protected]",
   "names" : [ "fred" ]
}

See also

For more information on the behavior of the captures array and additional examples, see captures Output Behavior.