foreach

`foreach`

Iterate through an array

Description

foreach reads an array or map from STDIN and iterates through it, running a code block for each iteration with the value of the iterated element passed to it.

By default foreach's output data type is inherited from its input data type. For example is STDIN is yaml then so will STDOUT. The only exception to this is if STDIN is json in which case STDOUT will be jsonlines (jsonl), or when additional flags are used such as --jmap.

Usage

{ code-block } reads from a variable and writes to an array / unbuffered STDOUT:

<stdin> -> foreach variable { code-block } -> <stdout>

{ code-block } reads from STDIN and writes to an array / unbuffered STDOUT:

<stdin> -> foreach { -> code-block } -> <stdout>

foreach writes to a buffered JSON map:

<stdin> -> foreach --jmap variable {
    code-block (map key)
} {
    code-block (map value)
} -> <stdout>

Examples

There are two basic ways you can write a foreach loop depending on how you want the iterated element passed to the code block.

The first option is to specify a temporary variable which can be read by the code block:

» a [1..3] -> foreach i { out $i }
1
2
3

Please note that the variable is specified without the dollar prefix, then used in the code block with the dollar prefix.

The second option is for the code block's STDIN to read the element:

» a [1..3] -> foreach { -> cat }
1
2
3

STDIN can only be read as the first command. If you cannot process the element on the first command then it is recommended you use the first option (passing a variable) instead.

Writing JSON maps

» ja [Monday..Friday] -> foreach --jmap day { out $day -> left 3 } { $day }
{
    "Fri": "Friday",
    "Mon": "Monday",
    "Thu": "Thursday",
    "Tue": "Tuesday",
    "Wed": "Wednesday"
}

Using steps to jump iterations by more than 1 (one)

You can step through an array, list or table in jumps of user definable quantities. The value passed in STDIN and $VAR will be an array of all the records within that step range. For example:

» %[1..10] -> foreach --step 3 value { out "Iteration $.i: $value" }
Iteration 1: [
    1,
    2,
    3
]
Iteration 2: [
    4,
    5,
    6
]
Iteration 3: [
    7,
    8,
    9
]
Iteration 4: [
    10
]

Flags

--jmap Write a json map to STDOUT instead of an array
--step<int> Iterates in steps. Value passed to block is an array of items in the step range. Not (yet) supported with --jmap

Detail

Meta values

Meta values are a JSON object stored as the variable $.. The meta variable will get overwritten by any other block which invokes meta values. So if you wish to persist meta values across blocks you will need to reassign $., eg

%[1..3] -> foreach {
    meta_parent = $.
    %[7..9] -> foreach {
        out "$(meta_parent.i): $.i"
    }
}

The following meta values are defined:

i: iteration number

Preserving the data type (when no flags used)

foreach will preserve the data type read from STDIN in all instances where data is being passed along the pipeline and push that data type out at the other end:

The temporary variable will be created with the same data-type as foreach's STDIN, or the data type of the array element (eg if it is a string or number)
The code block's STDIN will have the same data-type as foreach's STDIN
foreeach's STDOUT will also be the same data-type as it's STDIN (or jsonl (jsonlines) where STDIN was json because jsonl better supports streaming)

This last point means you may need to cast your data if you're writing data in a different format. For example the following is creating a YAML list however the data-type is defined as json:

» ja [1..3] -> foreach i { out "- $i" }
- 1
- 2
- 3

» ja [1..3] -> foreach i { out "- $i" } -> debug -> [[ /Data-Type/Murex ]]
json

Thus any marshalling or other data-type-aware API's would fail because they are expecting json and receiving an incompatible data format.

This can be resolved via cast:

» ja [1..3] -> foreach i { out "- $i" } -> cast yaml
- 1
- 2
- 3

» ja [1..3] -> foreach i { out "- $i" } -> cast yaml -> debug -> [[ /Data-Type/Murex ]]
yaml

The output is the same but now it's defined as yaml so any further pipelined processes will now automatically use YAML marshallers when reading that data.

Tips when writing JSON inside for loops

One of the drawbacks (or maybe advantages, depending on your perspective) of JSON is that parsers generally expect a complete file for processing in that the JSON specification requires closing tags for every opening tag. This means it's not always suitable for streaming. For example

» ja [1..3] -> foreach i { out ({ "$i": $i }) }
{ "1": 1 }
{ "2": 2 }
{ "3": 3 }

What does this even mean and how can you build a JSON file up sequentially?

One answer if to write the output in a streaming file format and convert back to JSON

» ja [1..3] -> foreach i { out (- "$i": $i) }
- "1": 1
- "2": 2
- "3": 3

» ja [1..3] -> foreach i { out (- "$i": $i) } -> cast yaml -> format json
[
    {
        "1": 1
    },
    {
        "2": 2
    },
    {
        "3": 3
    }
]

What if I'm returning an object rather than writing one?

The problem with building JSON structures from existing structures is that you can quickly end up with invalid JSON due to the specifications strict use of commas.

For example in the code below, each item block is it's own object and there are no [ ... ] encapsulating them to denote it is an array of objects, nor are the objects terminated by a comma.

» config -> [ shell ] -> formap k v { $v -> alter /Foo Bar }
{
    "Data-Type": "bool",
    "Default": true,
    "Description": "Display the interactive shell's hint text helper. Please note, even when this is disabled, it will still appear when used for regexp searches and other readline-specific functions",
    "Dynamic": false,
    "Foo": "Bar",
    "Global": true,
    "Value": true
}
{
    "Data-Type": "block",
    "Default": "{ progress $PID }",
    "Description": "Murex function to execute when an `exec` process is stopped",
    "Dynamic": false,
    "Foo": "Bar",
    "Global": true,
    "Value": "{ progress $PID }"
}
{
    "Data-Type": "bool",
    "Default": true,
    "Description": "ANSI escape sequences in Murex builtins to highlight syntax errors, history completions, {SGR} variables, etc",
    "Dynamic": false,
    "Foo": "Bar",
    "Global": true,
    "Value": true
}
...

Luckily JSON also has it's own streaming format: JSON lines (jsonl). We can cast this output as jsonl then format it back into valid JSON:

» config -> [ shell ] -> formap k v { $v -> alter /Foo Bar } -> cast jsonl -> format json
[
    {
        "Data-Type": "bool",
        "Default": true,
        "Description": "Write shell history (interactive shell) to disk",
        "Dynamic": false,
        "Foo": "Bar",
        "Global": true,
        "Value": true
    },
    {
        "Data-Type": "int",
        "Default": 4,
        "Description": "Maximum number of lines with auto-completion suggestions to display",
        "Dynamic": false,
        "Foo": "Bar",
        "Global": true,
        "Value": "6"
    },
    {
        "Data-Type": "bool",
        "Default": true,
        "Description": "Display some status information about the stop process when ctrl+z is pressed (conceptually similar to ctrl+t / SIGINFO on some BSDs)",
        "Dynamic": false,
        "Foo": "Bar",
        "Global": true,
        "Value": true
    },
...

`foreach` will automatically cast it's output as `jsonl` if it's STDIN type is `json`

» ja [Tom,Dick,Sally] -> foreach name { out Hello $name }
Hello Tom
Hello Dick
Hello Sally

» ja [Tom,Dick,Sally] -> foreach name { out Hello $name } -> debug -> [[ /Data-Type/Murex ]]
jsonl

» ja [Tom,Dick,Sally] -> foreach name { out Hello $name } -> format json
[
    "Hello Tom",
    "Hello Dick",
    "Hello Sally"
]

foreach

# foreach

# Description

# Usage

# Examples

# Writing JSON maps

# Using steps to jump iterations by more than 1 (one)

# Flags

# Detail

# Meta values

# Preserving the data type (when no flags used)

# Tips when writing JSON inside for loops

# foreach will automatically cast it's output as jsonl if it's STDIN type is json

# See Also