apoc.load.html

This is the APOC Extended documentation.

APOC Extended is not supported by Neo4j. For the officially supported APOC Core, go to the APOC Core page.

Procedure Apoc Extended

apoc.load.html('url',{name: jquery, name2: jquery}, config) YIELD value - Load Html page and return the result as a Map

Signature

apoc.load.html(url :: STRING?, query = {} :: MAP?, config = {} :: MAP?) :: (value :: MAP?)

Input parameters

Name Type Default

url

STRING?

null

query

MAP?

{}

config

MAP?

{}

Config parameters

The procedure support the following config parameters:

Table 1. Config parameters
name type default description

browser

Enum [NONE, CHROME, FIREFOX]

NONE

If it is set to "CHROME" or "FIREFOX", is used Selenium Web Driver to read the dynamically generated js. In case it is "NONE" (default), it is not possible to read dynamic contents. Note that to use the Chrome or Firefox driver, you need to have them installed on your machine and you have to download additional jars into the plugin folder. See below

wait

long

0

If greater than 0, it waits until it finds at least one element for each of those entered in the query parameter (up to a maximum of defined seconds, otherwise it continues execution). Useful to handle elements which can be rendered after the page is loaded (i.e. slow asynchronous calls).

charset

String

"UTF-8"

the character set of the page being scraped, if http-equiv meta-tag is not set.

headless

boolean

true

Valid with browser not equal to NONE, allow to run browser in headless mode, that is without actually opening the browser UI (recommended).

acceptInsecureCerts

boolean

true

If true, allow to read html from insecure certificates

baseUri

String

""

Base URI used to resolve relative paths

failSilently

Enum [FALSE, WITH_LOG, WITH_LIST]

FALSE

If the parse fails with one or more elements, using FALSE it throws a RuntimeException, using WITH_LOG a log.warn is created for each incorrect item and using WITH_LIST an errorList key is added to the result with the failed tags.

htmlString

boolean

true

to use a string instead of an url as 1st parameter

Output parameters

Name Type

value

MAP?

Usage Examples

We can extract the metadata and h2 heading from the Wikipedia home page, by running the following query:

CALL apoc.load.html("https://en.wikipedia.org/",{metadata:"meta", h2:"h2"});
Table 2. Results
Output 
{
   "metadata":[
      {
         "tagName":"meta",
         "attributes":{
            "charset":"UTF-8"
         }
      },
      {
         "tagName":"meta",
         "attributes":{
            "name":"ResourceLoaderDynamicStyles"
         }
      },
      {
         "tagName":"meta",
         "attributes":{
            "name":"generator",
            "content":"MediaWiki 1.36.0-wmf.16"
         }
      },
      {
         "tagName":"meta",
         "attributes":{
            "name":"referrer",
            "content":"origin"
         }
      },
      {
         "tagName":"meta",
         "attributes":{
            "name":"referrer",
            "content":"origin-when-crossorigin"
         }
      },
      {
         "tagName":"meta",
         "attributes":{
            "name":"referrer",
            "content":"origin-when-cross-origin"
         }
      },
      {
         "tagName":"meta",
         "attributes":{
            "property":"og:image",
            "content":"https://upload.wikimedia.org/wikipedia/commons/1/1c/Orion_pulse_unit_%28transparent%29.png"
         }
      }
   ],
   "h2":[
      {
         "attributes":{
            "class":"mp-h2",
            "id":"mp-tfa-h2"
         },
         "text":"From today's featured article",
         "tagName":"h2"
      },
      {
         "attributes":{
            "class":"mp-h2",
            "id":"mp-dyk-h2"
         },
         "text":"Did you know ...",
         "tagName":"h2"
      },
      {
         "attributes":{
            "class":"mp-h2",
            "id":"mp-itn-h2"
         },
         "text":"In the news",
         "tagName":"h2"
      },
      {
         "attributes":{
            "class":"mp-h2",
            "id":"mp-otd-h2"
         },
         "text":"On this day",
         "tagName":"h2"
      },
      {
         "attributes":{
            "class":"mp-h2",
            "id":"mp-tfl-h2"
         },
         "text":"From today's featured list",
         "tagName":"h2"
      },
      {
         "attributes":{
            "class":"mp-h2",
            "id":"mp-tfp-h2"
         },
         "text":"Today's featured picture",
         "tagName":"h2"
      },
      {
         "attributes":{
            "class":"mp-h2",
            "id":"mp-other"
         },
         "text":"Other areas of Wikipedia",
         "tagName":"h2"
      },
      {
         "attributes":{
            "class":"mp-h2",
            "id":"mp-sister"
         },
         "text":"Wikipedia's sister projects",
         "tagName":"h2"
      },
      {
         "attributes":{
            "class":"mp-h2",
            "id":"mp-lang"
         },
         "text":"Wikipedia languages",
         "tagName":"h2"
      },
      {
         "tagName":"h2",
         "text":"Navigation menu"
      }
   ]
}

Let’s suppose we have a test.html file like this:

<!DOCTYPE html>
<html class="client-nojs" lang="en" dir="ltr">
  <h6 i d="error">test</h6>
  <h6 id="correct">test</h6>
</html>

We can handle the parse error caused by i d through failSilently configuration. So, we can execute:

CALL apoc.load.html("test.html",{h6:"h6"});
Table 3. Results
Failed to invoke procedure apoc.load.html: Caused by: java.lang.RuntimeException: Error during parsing element: <h6 i d="error">test</h6>

or with failSilently WITH_LIST:

CALL apoc.load.html("test.html",{h6:"h6"}, {failSilently: 'WITH_LIST'});
Table 4. Results
Output 
{
  "errorList": [
    "<h6 i d="error">test</h6>"
  ],
  "h6": [
    {
      "attributes": {
        "id": "correct"
      },
      "text": "test",
      "tagName": "h6"
    }
  ]
}

or with failSilently WITH_LOG (note that will be created a log.warn("Error during parsing element: <h6 i d="error">test</h6>") ):

CALL apoc.load.html("test.html",{h6:"h6"}, {failSilently: 'WITH_LOG'});
Table 5. Results
Output 
{
  "h6": [
    {
      "attributes": {
        "id": "correct"
      },
      "text": "test",
      "tagName": "h6"
    }
  ]
}

Load from runtime generated file

If we have a test.html file with a jQuery script like:

<!DOCTYPE html>
<html>
    <head>
      <script src="https://code.jquery.com/jquery-1.9.1.min.js"></script>
      <script type="text/javascript">
        $(() => {
            var newP = document.createElement("strong");
            var textNode = document.createTextNode("This is a new text node");
            newP.appendChild(textNode);
            document.getElementById("appendStuff").appendChild(newP);
        });
      </script>
    </head>
    <body>
        <div id="appendStuff"></div>
    </body>
</html>

we can read the generated js through the browser config. Note that to use the browser config (except with "NONE" value), you have to install additional dependencies which can be downloaded from this link.

For example, with the above file we can execute:

CALL apoc.load.html("test.html",{strong: "strong"}, {browser: "FIREFOX"});
Table 6. Results
Output 
{
  "strong": [
    {
      "tagName": "strong",
      "text": "This is a new text node"
    }
  ]
}

If we have to parse a tag from a slow async call, we can use wait config to waiting for 10 second (in this example):

CALL apoc.load.html("test.html",{asyncTag: "#asyncTag"}, {browser: "FIREFOX", wait: 10});

We can also pass an HTML string into the 1st parameter by putting as a config parameter htmlString: true, for example:

CALL apoc.load.html("<!DOCTYPE html> <html> <body> <p class='firstClass'>My first paragraph.</p> </body> </html>",{body:"body"}, {htmlString: true})
YIELD value
RETURN value["body"] as body
Table 7. Results
body 
[{
  "attributes": {},
  "text": "My first paragraph.",
  "tagName": "body"
}]

Css / jQuery selectors

The jsoup class org.jsoup.nodes.Element provides a set of functions that can be used. Anyway, we can emulate all of them using the appropriate css/jQuery selectors in these ways (except for the last one, we can substitute the with a tag name to search into it instead of everywhere. Furthermore, by removing the selector will be returned the same result):

jsoup function css/jQuery selector description

getElementById(id)

#id

Find an element by ID, including or under this element.

getElementsByTag(tag)

tag

Finds elements, including and recursively under this element, with the specified tag name.

getElementsByClass(className)

.className

Find elements that have this class, including or under this element.

getElementsByAttribute(key)

[key]

Find elements that have a named attribute set.

getElementsByAttributeStarting(keyPrefix)

*[^keyPrefix]

Find elements that have an attribute name starting with the supplied prefix. Use data

to find elements that have HTML5 datasets.

getElementsByAttributeValue(key,value)

*[key=value]

Find elements that have an attribute with the specific value.

getElementsByAttributeValueContaining(key,match)

[key=match]

Find elements that have attributes whose value contains the match string.

getElementsByAttributeValueEnding(key,valueSuffix)

*[class$="test"]

Find elements that have attributes that end with the value suffix.

getElementsByAttributeValueMatching(key,regex)

*[id~=content]

Find elements that have attributes whose values match the supplied regular expression.

getElementsByAttributeValueNot(key,value)

*:not([key="value"])

Find elements that either do not have this attribute, or have it with a different value.

getElementsByAttributeValueStarting(key,valuePrefix)

*[key^=valuePrefix]

Find elements that have attributes that start with the value prefix.

getElementsByIndexEquals(index)

*:nth-child(index)

Find elements whose sibling index is equal to the supplied index.

getElementsByIndexGreaterThan(index)

*:gt(index)

Find elements whose sibling index is greater than the supplied index.

getElementsByIndexLessThan(index)

*:lt(index)

Find elements whose sibling index is less than the supplied index.

getElementsContainingOwnText(searchText)

*:containsOwn(searchText)

Find elements that directly contain the specified string.

getElementsContainingText(searchText)

*:contains('searchText')

Find elements that contain the specified string.

getElementsMatchingOwnText(regex)

*:matches(regex)

Find elements whose text matches the supplied regular expression.

getElementsMatchingText(pattern)

*:matchesOwn(pattern)

Find elements whose text matches the supplied regular expression.

getAllElements()

*

For example, we can execute:

CALL apoc.load.html($url, {nameKey: '#idName'})
Table 8. Results
Output 
{
  "h6": [
    {
      "attributes": {
        "id": "idName"
      },
      "text": "test",
      "tagName": "h6"
    }
  ]
}

Html plain text representation

If, instead of a map of json list results, you want to get a map of plain text representations, you can use the apoc.load.htmlPlainText procedure, which use the same syntax, logic and config parameters as apoc.load.html.