Home

Awesome

vark

ArduinoIDE Asynchronous HTTP client library for ESP8266, ESP32

Version 0.1.2 2/6/2021

Code tidying re std:: prefixing / namespace issues, API/functionality unchanged

Version 0.1.1

Read Release Notes 0.1.1

If you are able, please Support me on Patreon and/or subscribe to my Youtube channel (instructional videos)

Contents

What does it do?

ArmadilloHTTP sits on top of AardvarkTCP. It allows you to access remote HTTP/S sites asynchronously with huge* page sizes, using GET, POST, PUT, PATCH, DELETE,

*By "huge" we mean any packet that is larger than the total system buffers. If you compile with the "Low Memory" option on ESP8266 this will be 1072 bytes or 2920 with the "Higher Bandwith" options (as at core 2.7.4)

ESP32 and '8266 use LwIP to manage TCP communications. LwIP allows multiple buffers, for ESP8266 it's 2, so what you see in the figures above is 1072=2x536 and 2920=2x1460. The important point here is that all of the "magic numbers" 536, 1460 (the individual LwIP buffer size) and 2 (the number of buffers) are implementation-dependent and could change in the future, so "hardcoding" any of them into your own app would be a bad idea as it could cause problems in the future or prevent your code running on newer / different machines.

If you want code that runs on either ESP32 or ESP8266 with no changes, then you need something that automatically caters for those values and make is so that you can just send and receive without worrying about them.

ArmadilloHTTP solves that problem by allowing data up to aabout half of the free heap to be sent / received, no matter what any of those LwIP "magic numbers" above are. It also provides a very simple interface which seamlessly manages both unencrypted and TLS-encrypted (ESP8266 only) sessions, depending on the URL provided.

Worth a 1000 words...

This is pretty much all the code you will need. "Here's a URL, work out which protocol it needs, and give me a shout when you have the whole page ready for me". Why should it be any more complicated?

#include <ArmadilloHTTP.h>
ArmadilloHTTP aClient;

void responseHandler(ARMA_HTTP_REPLY r){ // r is a struct will all the reply fields you might need
  Serial.printf("\nRaw Data\n");
  dumphex(r.data,r.length);
}

void setup() {
  Serial.begin(115200);
  WiFi.begin("XXXXXXXX", "XXXXXXXX");
  WiFi.waitForConnectResult();
  Serial.printf("Connected %s\n",WiFi.localIP().toString().data());

  aClient.GET("http://robot.local/1328.htm",responseHandler); // <--- Yup, that's pretty much it!
}

void loop() {}

The "menagerie" roadmap

ArmadilloHTTP is one of a family of firmware packages for simple robust and rapid asynchronous IOT development on ESP8266 / ESP32

SEE Installation

roadmap

The related / dependent libraries

NameProvidesNotes
Forked AsyncTCP"Glue" to LwIP (ESP8266)Important bugfixes
Forked ESPAsyncTCP"Glue" to LwIP(ESP32)Missing features added
Forked ESPAsyncWebserverBasis of webUI in H4PluginsSeveral major bugfixes
roadmapPMB Tools'32/'8266 HAL and utility functions
roadmapAardvarkTCPSimple Large-payload Async TCPAPI-compatible with ESPAsyncTCP, seamless TLS/SSL
roadmapPangolinMQTTAsync MQTT ClientQoS 0/1/2 Fully 3.1.1 compliant. Large payloads
roadmapArmadilloHTTPAsync HTTP/S ClientSimple send/callback of large payloads
roadmapH4Scheduler/Async Timers
roadmapH4/PluginsFull Async IOT FirmwareWebserver, MQTT, OTA, NTP, HTTP etc etc

ArmadilloHTTP installation path

dillo path

API

URL defintion

The url must be specified in the following general form. The extended path and query portions are optional, as is the port. If the port is omitted it will default to 80 for URLs starting http and 443 for those starting https

http://hostname:port/path/to/resource?a=b&c=d"

or

https://hostname:port/path/to/resource?a=b&c=d"

The hostname portion my be specified as a "dotted quad" IP address e.g. "172.103.22.14" or a publicly resolvable DNS name e.g. somehost.co.uk

ESP8266 targets will happily resolve .local names. See "Known Issues" re ESP32

Valid examples

Using TLS

TLS is only currently only available on ESP8266 targets. The first step to using TLS is to edit the async_config.h file in Forked AsyncTCP and change #define ASYNC_TCP_SSL_ENABLED 0 to #define ASYNC_TCP_SSL_ENABLED 1

Note that this will significantly increase the size of the compiled app. Unless you absolutely need it, do not compile in TLS!

Note also that the version of TLS that ships with ESPAsyncTCP is very weak and there are many sites that will refuse to connect as they require stronger ciphers or client certificates etc.

tls

"Pre-flighting" and phases

In the vast majority of cases, you will just call the GET function with a URL and optionally a fingerprint (if using TLS on ESp8266). Some sites however require that you "pre-flight" them, meaning that they will refuse a GET , POST etc request until you have "announced your intentions" first. This is usually done by sending HTTP OPTIONS request, laying out what you intend to retrieve etc.

A second consideration is that both ESP32 and especially ESP8266 are very limited in terms of processing power and memory, so while they may be acting like a browser in some ways, they can only perform a tiny proportion of the functions of a typical browser. This means that asking them to retrieve a large webpage is likely to fail. Currently anything over about 9 or 10k will "break the bank", so knowing in advance how big the reply will be can save a lot of wasted time, effort and potential crashes.

Many apps do this by sending HTTP HEAD request, when the server will respond with all the headerer information to allow the app to calculate the predicted size, but not actually send any data. Then if the expected size will "break the bank", the actual GET , POST etc request will not be sent, and a "too big" error will be sent back to your code.

Thus there are three "phases" of a worst-case scenario where a "pre-flighting" site can send large responses:

  1. Pre-Flight (ARMA_PHASE_PREFLIGHT): send OPTIONS and confirm intended method available
  2. Measure (ARMA_PHASE_MEASURE): send HEAD and confirm expected size is within limits
  3. Execute (ARMA_PHASE_EXECUTE): send actual HTTP method (GET , POST etc) to do the deed

All of the HTTP method calls take the starting phase value as the final parameter. In 99% of cases you will just want to dive straight in with ARMA_PHASE_EXECUTE. If however you choose ARMA_PHASE_PREFLIGHT or ARMA_PHASE_MEEASURE then ArmadilloHTTP will automatically step through the following phases, analysing the responses and only continuing when the next phase is likely to succeed, or returning an error message when it cannot.

Your response handler function will be called exactly once, after the ARMA_PHASE_EXECUTE phase is complete. (see next section)

The optimum strategy depends entirely on the type of sites you will be visiting and the nature of the responses they provide. There is no "one-size-fits-all" solution - the best way for you will be found by experimentation (and probably a crash or two) before you get your on code "tuned in".

Respsonse handling

All of the information from the response is packed into a struct and a copy of that is passed to your response handler function.

The ARMA_HTTP_REPLY struct

struct ArmadilloHTTPresponse {
    uint32_t        httpResponseCode; // the HTTP reply: 200, 404 etc
    VARK_NVP_MAP    responseHeaders; // std:map<std::string,std::string> name / value pairs e.g. { ... {"Content-Type","text/html"} ... } etc
    ARMA_METHODS    allowedMethods; // std::unordered+map<std::string> allowed methods e.g. {"GET","POST"}
    const uint8_t*  data; // The raw binary data of the reply N.B. THIS IS NOT NECESSARILY A "STRING"...DO NOT ASSUME THAT IT IS!!!
    size_t          length; // the total length of data

    std::string     asJsonstring(); // ONLY IF REPLY IS VALID JSON: will return a std:string that can be passed to e.g. ArduinoJson lib
    VARK_NVP_MAP    asSimpleJson(); // ONLY IF REPLY IS "SIMPLE JSON" see docs for what we mean by "simple json"
    std::string     asStdstring(); // returns a std::string of <length> characters. There is NO GUARANTEE that they are valid "visible" characters

Handling the data

You need to know in advance the type and general content of the reply you will receive before trying to use it. The functions in the response structure will almost certainly fail (and cause a crash) if you force them to operate on a response that does not correctly match the actual data received.

For this reason it is generally wise to examine the message contents (if any) only when the httpResponseCode is in the range 200-299.

During the development, you should use tools like Fiddler, Postman, Wireshark etc so that you know the format of the data you will get back from any given URL you will retrieve. A quicker route is to use ArmadilloHTTP's own builtin function: dumphex, passing it the data pointer and length.

In short, do not call any of the response functions unless you know 100% that the response data is of the appropriate type!

Cheap and cheerful "simple json" handling

If a JSON response does not have a complex structure, or contain e.g. arrays and it amounts to a simple collection of name / value pairs, then it can be dealt with in a few lines of code without needing the heavy overhead of a "proper" Json library such as the excellent ArduinoJson.

This is especially easy to arrange in cases where you may be the website's author, but you can "get lucky" with some sites that have APIs that return only small amounts of data. If you can obey the limits below, it enables some very efficient code.

Example:

{"name":"phil","firmware":"H4","numeric","666"}

If you can stick to that regime, then the PMB Tools functions:

Will come in very handy and keep you app binary very small.

Example response handler function

void responseHandler(ARMA_HTTP_REPLY r){ // r is a struct with all the necessary fields in it: see above
  Serial.printf("HTTP Response Code: %d\n",r.httpResponseCode);
  Serial.printf("Response Headers:\n");
  for(auto const h:r.responseHeaders) Serial.printf("%s=%s\n",h.first.data(),h.second.data());
  Serial.printf("\nRaw Data\n");
  dumphex(r.data,r.length);
  Serial.printf("\nAs a std::string - BE CAREFUL, IT MAY NOT BE A STRING!!!\n%s\n",r.asStdstring().data()); // Data may NOT be a string -> crash!!!
  //Serial.printf("\nAs simple JSON - BE CAREFUL, complex will break it...AND it may not be JSON!!!\n");
  //for(auto const& j:r.asSimpleJson()) Serial.printf("%s=%s\n",j.first.data(),j.second.data());
  ...or
  //
  // std::string complex=r.asJsonString();
  // pass complex to a "proper" Json handling function
}

Functions

ArmadilloHTTP(); // default constructor

void addRequestHeader(const std::string& hdr,const std::string& value); // adds a request header, needs to be called before any of the HTTP methods :)


void DELETE(const std::string& url,ARMA_FN_HTTP rx,const uint8_t* fingerprint=nullptr,uint32_t phase=ARMA_PHASE_EXECUTE);
void GET(const std::string& url,ARMA_FN_HTTP rx,const uint8_t* fingerprint=nullptr,uint32_t phase=ARMA_PHASE_EXECUTE);
void PATCH(const std::string& url,const VARK_NVP_MAP& fields,ARMA_FN_HTTP rx;{ _prepare(phase,"PATCH",url,rx,fields); }
void POST(const std::string& url,const VARK_NVP_MAP& fields,ARMA_FN_HTTP rx,const uint8_t* fingerprint=nullptr,uint32_t phase=ARMA_PHASE_EXECUTE);
void PUT(const std::string& url,const VARK_NVP_MAP& fields,ARMA_FN_HTTP rx,const uint8_t* fingerprint=nullptr,uint32_t phase=ARMA_PHASE_EXECUTE);

Example Sketch - basic introduction

Example Sketch using requestHeaders to set API key

Known issues / limitations

Installation

Please see H4 Installer

Issues

If you want a quick resolution, please follow these rules:

  1. As with all H4 and H4Plugins libraries, please make sure you have read all the relevant documentation relating to the issue and watched any videos on the Youtube channel (instructional videos). Please also subscribe to the channel for notifications of news and updates.

  2. If you still think there is a problem, then join the Facebook H4 Support / Discussion group and report the issue briefly there. This is because I visit the group every day, whereas I do not have time to visit 11 github repos every day. Furthermore, it alerts other users to potential problems and allows an initial assessment.

  3. If there is a genuine issue then you will be referred to Raising H4/H4Plugins issues after which you are advised to create a full github issue report.

  4. Failing to make an initial report in the Facebook H4 Support / Discussion group and simply starting with a github issue, or failing to include all of the information required in Raising H4/H4Plugins issues is likely to result in a long delay before it gets picked up.

(c) 2021 Phil Bowles h4plugins@gmail.com