Skip to content

A Python codebase that uses AWS Lambda/API Gateway via the serverless framework. Generates an .svg file on S3 from given text input.

Notifications You must be signed in to change notification settings

daviseford/word-art-serverless

Repository files navigation

Word Art Serverless Endpoint

Endpoint and Parameters

Endpoint: https://gy2aa8c57c.execute-api.us-east-1.amazonaws.com/dev

Parameters:

  • text - str - Required

  • node_colors - array - An array of 2 hex codes, like ['#FFF', '#CCC']. These codes are used for the start and end node colors, respectively.

  • color - str - A hex code,, like '#FFF'. This is the color of the path

  • split - object - An object like so:

{
    "words": ["quality", "knowledge", "logic"],
    "color": "#000"
}

Requires Python 2.7

First, take care of the requirements:

  1. pip install -r requirements.txt

  2. brew install caskformula/caskformula/inkscape

  3. npm install -g serverless

  4. sls plugin install -n serverless-python-requirements

  5. Requires Docker (serverless-python-requirements uses Docker to build)

Commands

sls deploy

sls logs -f app --tail

sls deploy && sls logs -f app --tail

Application flow

  1. POST Request to the endpoint
POST /dev HTTP/1.1
Host: gy2aa8c57c.execute-api.us-east-1.amazonaws.com
Content-Type: application/json
Cache-Control: no-cache
Postman-Token: 87c7679b-0719-608d-016a-f10e6f00b276

{
	"text" : "I guess various sentences may lead to various structures. 
	          A short one. Then a quick twist and run, further than we'd anticipated. 
	          I guess. Well, I know that one thing is for certain - long, 
	          run on sentences have a place here. But! So do quick ones. Tight little turns. 
	          Another tactic would be to send out a long, long sentence, broken up by not 
	          much else other than the occasional - well, yes, that - out of nowhere, how surprising. 
	          The more sentences, the merrier. Try giving me a whole book :)",
	"node_colors":["FF101A", "#CCC"],
	"colors": ["#FF59E7"]
}
  1. The text is converted into an XML Path.

  2. The XML Path is converted into a full SVG file using svgpathtools.

  3. The contents of the file are checksummed using sha1 and named [file-checksum].svg

  4. The file is uploaded to S3. There is a check to ensure we do not upload duplicate files - hence the checksumming.

  5. The file's URL is returned as a JSON response like so: { 'statusCode': 200, 'body': {"s3_url": "https://s3.amazonaws.com/word-art-svgs/f3f5ea8d8cd19705f3d743c8897eddf0c0e3840a.svg"}

About

A Python codebase that uses AWS Lambda/API Gateway via the serverless framework. Generates an .svg file on S3 from given text input.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages