Skip to content

Latest commit

 

History

History
222 lines (166 loc) · 8.13 KB

README.md

File metadata and controls

222 lines (166 loc) · 8.13 KB

tmpl

tmpl is a lightweight, shell script friendly document renderer. Something like envsubst just different.

tmpl is inspired by recent extensions of the shell toolset like jq, which allows to work with complex, structured (JSON) data from the shell. This tool provides a solution for a special case in that particular niche: automated generation of complex documents out of structured data using smart templates.

Examples

# hello world
$ tmpl -d 'env:' -t 'envsubst:hello ${USER}, how are you?'

# print out all env vars as JSON
$ tmpl -d 'env:' -t 'template:{{ json .data }}'

# use source from remote location to render local template and store the result
# in a local file
$ tmpl --data-location https://config.acme.tld/node/$(uname -n).json \
    --template-location /etc/apache/templates/homepage.tmpl \
    --output /var/www/home/index.html

# get data from output of execution, then pull template from remote # location
# and pipe output to script that sends alarm
$ tmpl --data-location "shell://elasticdump ... args ..." \
    --decoder json \
    --template-location "https://wiki.acme.tld/templates/report.tmpl?auth=$(tokengen.sh)" | \
    send-alarm.sh ...

There are many possible applications or integrations. Here are a few:

  • Reporting, eg generate HTML reports or summaries based on "native JSON sources" like ElasticSearch or MongoDB
  • Build pipeline, eg
    • transform generic YAML/JSON into domain specific configuration files (think YAML -> NGINX config)
    • render templates from environment variables
  • Monitoring / alerting, eg generate system alert emails with complex information rendered in a human readable way
  • Transactional messaging, eg create rich documents from raw JSON (eg JSON -> DocX)
  • you can think of something ..

Data Locations

A data location contains input data (structures). Some locations indicate a data format, and thereby a Decoder. support guessing of the decoder, e.g. an URL like http://acme.tld/foo.json implies JSON format.

Supported locations are:

  • env: or env:SOME_PREFIX_ converting all (prefix) matching env vars into a flat data map; for example env:FOO_ converts FOO_PARAM=x and FOO_Other=yy into {"PARAM":"x", "Other": "yy"}; uses JSON internally
  • http:// or https://: arbitrary, GETable HTTP(S) URLs; decoder guessed from file ending like .json or .yaml ending of file in URL path (https://acme.tld/my/file.json)
  • file:///path/to/file or /path/to/file: arbitrary local files; decoder guessed from file ending like .json or .yaml
  • shell://: arbitrary, atomic shell command lines, eg shell://date +%F, which would execute echo '{"foo":"bar"}' or anything that would return JSON/YAML on STDOUT. Don't use pipes or somesuch..
  • -: STDIN, requires decoder specification

Data Decoder

Supported decoders are JSON and YAML. Per default, tmpl tries to guess the format from the URL. The decoder can be set explicitly with --decoder <json|yaml> (or -d <json|yaml>).

Template Locations

tmpl supports multiple template render engines:

The same as with data sources: specify the renderer explicitly with --renderer | -r <name> or let tmpl try to guess it from the template location:

  • Template URLs with .envsubst file name endings default to envsubst engine. Examples:
    • http://acme.tld/templates/vhost.envsubst?t=123456
    • file:///etc/apache/vhost.envsubst
  • Template URLs with .tmpl or .template file names default to template engine. Examples:
    • http://acme.tld/templates/vhost.tmpl?foo=bar
    • file:///etc/apache/vhost.template
  • Template URLs with .pongo2 or .pongo files name endings default to pongo2 engine. Examples:
    • http://acme.tld/templates/vhost.pongo2?t=123456
    • file:///etc/apache/vhost.pongo

For template and pongo2 see the examples below, assume the following data structure & content:

{
  "name": "www.acme.tld",
  "aliases": ["acme.tld", "blog.acme.tld"],
  "directory": "/var/www/homepage",
  "directories": [
    {"path": "/foo", "users": ["bar"]},
    {"path": "/lorem"}
  ]
}

and the following expected result (+/- a few empty lines, see optimized templates here):

<VirtualHost www.acme.tld:80>
    ServerName www.acme.tld
    ServerAlias acme.tld blog.acme.tld
    DocumentRoot "/var/www/homepage"

    <Directory "/var/www/homepage/foo">
        Require user foo bar
    </Directory>
    <Directory "/var/www/homepage/lorem">
        Require valid-user
    </Directory>
</VirtualHost>

template engine

<VirtualHost {{.data.name}}:80>
    ServerName {{.data.name}}
    ServerAlias{{range .data.aliases}} {{.}}{{end}}
    DocumentRoot "{{.data.directory}}"
    {{range $idx, $directory := .data.directories}}
    <Directory "{{$.data.directory}}{{$directory.path}}">
        {{if $directory.users}}
        Require user{{range $directory.users}} {{.}}{{end}}
        {{else}}
        Require valid-user
        {{end}}
    </Directory>
    {{end}}
</VirtualHost>

Find more examples for including additional templates, working with macro like blocks, ..

pongo2 engine

Example template:

<VirtualHost {{ data.name }}:80>
    ServerName {{ data.name }}
    ServerAlias {{ data.aliases | join:" " }}
    DocumentRoot "{{ data.directory }}"
    {% for directory in data.directories %}
    <Directory "{{ data.directory }}{{ directory.path }}">
        {% if directory.users %}
        Require user {{ directory.users | join:" " }}
        {% else %}
        Require valid-user
        {% endif %}
    </Directory>
    {% endfor %}
</VirtualHost>

Find more examples for including additional templates, macros, functions, .. here and here

envsubst engine

This engine only supports flat data structures and is intended to use with the env: data location.

Assuming the following env vars are set:

USER=myself
HOME=/home/myself
APP_NAME=the-app
APP_DOMAIN=the-domain.tld

Using all env variables

With the following template

Hello ${USER}, here is your home: ${HOME}. Your application is named ${APP_NAME}.

executed with tmpl -d 'env:' -t 'file:///path/to/template' would render:

Hello myself, here is your home: /home/myself. Your application is named the-app.

Using prefixed env variables

With the following template

App name ${NAME} has domain ${DOMAIN}

executed with tmpl -d 'env:APP_' -t 'file:///path/to/template' would render:

App name the-app has domain the-domain.tld

Use as library

tmpl is (mostly) structured in the Standard Go Project Layout and follows ('ish) the Standard Package Layout, as defined by Ben Johnson.

  • Note: Parts of the library in pkg/ is using singletons, intended to be used in a program.
  • Note: Check out the example folder, the BuildTmpl facade function in tmpl.go and the integration tests in imports_test.go to get an understanding on how to use.

To use guessers and builders, you can import the whole bundle:

package mypackage

import (
	"fmt"
	"..."
	_ "github.com/ukautz/tmpl/pkg/imports"
)

// --- %< ---

renderer, err := tmpl.GuessRenderer("http://some/url.tmpl")  // or "file:///etc/file.pongo" or ..
source, err := tmpl.GuessSource("http://some/url") // or "file:///path" or ..
decoder, err := tmpl.GuessDecoder("http://some/url.json") // or "/srv/path/file.yaml" or ..