Table to JSON

Attempts to convert HTML tables into JSON.

Can be passed the markup for a single table as a string, a fragment of HTML or an entire page or just a URL (with an optional callback function; promises also supported).

The response is always an array. Every array entry in the response represents a table found on the page (in same the order they were found in the HTML).

As of version 2.0 tabletojson is completely written in typescript.

!!! ATTENTION !!!: Incompatible API change in version 2.0.0 since request.js got deprecated. More information here...

Conversion from version 1.+ to 2.x

• Require must be changed from const tabletojson = require('../lib/tabletojson'); to either const tabletojson = require('../lib/tabletojson').Tabletojson; or const {Tabletojson: tabletojson} = require('../lib/tabletojson');
• Replace request options by got options. More information here...

Basic usage

Install via npm

1npm install tabletojson

Remote (convertUrl)

1'use strict';
2
3const tabletojson = require('tabletojson').Tabletojson;
4
5tabletojson.convertUrl(
6    'https://en.wikipedia.org/wiki/List_of_ISO_639-1_codes',
7    function(tablesAsJson) {
8        console.log(tablesAsJson[1]);
9    }
10);

Local (convert)

Have a look in the examples.

1// example-6.js
2'use strict';
3
4const {Tabletojson: tabletojson} = require('../dist');
5const fs = require('fs');
6const path = require('path');
7
8const html = fs.readFileSync(path.resolve(__dirname, '../test/tables.html'), {encoding: 'UTF-8'});
9const converted = tabletojson.convert(html);
10
11console.log(converted);

Duplicate column headings

If there are duplicate column headings, subsequent headings are suffixed with a count:

PLACE	VALUE	PLACE	VALUE
abc	1	def	2

1[{
2  PLACE: 'abc', VALUE: '1',
3  PLACE_2: 'def', VALUE_2: '2',
4}]

Tables with rowspan

Having tables with rowspan, the content of the spawned cell must be available in the respective object.

Parent	Child	Age
Marry	Sue	15
	Steve	12
	Tom	3

1[{
2  PARENT: 'Marry', CHILD: 'Tom', AGE, '3',
3  PARENT: 'Marry', CHILD: 'Steve', AGE, '12',
4  PARENT: 'Marry', CHILD: 'Sue', AGE, '15'
5}]

Tables with complex rowspan

Having tables with complex rowspans, the content of the spawned cell must be available in the respective object.

Parent	Child	Age
Marry	Sue	15
	Steve	12
	Tom	3
Taylor
	Peter	17

1[{
2  PARENT: 'Marry', CHILD: 'Sue', AGE, '15'
3  PARENT: 'Marry', CHILD: 'Steve', AGE, '12',
4  PARENT: 'Marry', CHILD: 'Tom', AGE, '3',
5  PARENT: 'Taylor', CHILD: 'Tom', AGE, '3',
6  PARENT: 'Taylor', CHILD: 'Peter', AGE, '17'
7}]

Tables with headings in the first column

If a table contains headings in the first column you might get an unexpected result, but you can pass a second argument with options with { useFirstRowForHeadings: true } to have it treat the first column as it would any other cell.

1tabletojson.convertUrl(
2  'https://www.timeanddate.com/holidays/ireland/2017',
3  { useFirstRowForHeadings: true },
4  function(tablesAsJson) {
5    console.log(tablesAsJson);
6  }
7);

Tables with HTML

The following options are true by default, which converts all values to plain text to give you an easier more readable object to work with:

• stripHtmlFromHeadings
• stripHtmlFromCells

If your table contains HTML you want to parse (for example for links) you can set stripHtmlFromCells to false to treat it as raw text.

1tabletojson.convertUrl(
2  'https://en.wikipedia.org/wiki/List_of_ISO_639-1_codes',
3  { stripHtmlFromCells: false },
4  function(tablesAsJson) {
5    //Print out the 1st row from the 2nd table on the above webpage as JSON
6    console.log(tablesAsJson[1][0]);
7  }
8);

Note: This doesn't work with nested tables, which it will still try to parse.

You probably don't need to set stripHtmlFromHeadings to false (and setting it to false can make the results hard to parse), but if you do you can also set both at the same time by setting stripHtml to false.

Options

!!! ATTENTION !!! Since request is not actively supported we need to switch to a reliable request replacement and I decided to use got.

This is an incompatible change in version 2++ so keep this in mind and follow Sindre's Migration Guide.

For special features like using a proxy you should follow this instructions: Proxies

got (only convertUrl)

We are using got to fetch remoter HTML pages. So if you need to get data from a remote server you can call tabletojson.convertUrl and pass any got-options (proxy, headers,...) by adding a got object to the options passed to convertUrl. for more information on how to configure request please have a look at Proxies

1tabletojson.convertUrl('https://www.timeanddate.com/holidays/ireland/2017', {
2    useFirstRowForHeadings: true,
3    got: {
4        agent: tunnel.httpOverHttp({
5            proxy: {
6                host: 'proxy:8080'
7            }
8        })
9    }
10});

stripHtmlFromHeadings

Strip any HTML from heading cells. Default is true.

1KEY | <b>VALUE</b>
2----|-------------
3abc |            1
4dev |            2

1// Example output with stripHtmlFromHeadings:true
2[
3    {
4        KEY: 'abc', VALUE: '1'
5    },
6    {
7        KEY: 'dev', VALUE: '2'
8    }
9]
10// Example output with stripHtmlFromHeadings:false
11[
12    {
13        KEY: 'abc', '<b>VALUE</b>': '1'
14    },
15    {
16        KEY: 'dev', '<b>VALUE</b>': '2'
17    }
18]

stripHtmlFromCells

Strip any HTML from tableBody cells. Default is true.

1KEY |    VALUE
2----|---------
3abc | <i>1</i>
4dev | <i>2</i>

1// Example output with stripHtmlFromHeadings:true
2[
3    {
4        KEY: 'abc', VALUE: '1'
5    },
6    {
7        KEY: 'dev', VALUE: '2'
8    }
9]
10// Example output with stripHtmlFromHeadings:false
11[
12    {
13        KEY: 'abc', 'VALUE': '<i>1</i>'
14    },
15    {
16        KEY: 'dev', 'VALUE': '<i>2</i>'
17    }
18]

forceIndexAsNumber

Instead of using column text (that sometime re-order the data), force an index as a number (string number).

1// Some JSON (Other rows)
2{
3  "0": "",
4  "1": "Ａ会",
5  "2": "Ｂ会",
6  "3": "Ｃ会",
7  "4": "Something",
8  "5": "Else",
9  "6": ""
10}
11// Some JSON (Other rows)

countDuplicateHeadings

Default is true. If set to false, duplicate headings will not get a trailing number. The value of the field will be the last value found in the table row:

PLACE	VALUE	PLACE	VALUE
abc	1	def	2
ghi	3	jkl	4

1// Example output with countDuplicateHeadings:false
2[
3    {
4        PLACE: 'def', VALUE: '2'
5    },
6    {
7        PLACE: 'jkl', VALUE: '4'
8    }
9]

ignoreColumns

Array of indexes to be ignored, starting with 0. Default is 'null/undefined'.

NAME	PLACE	WEIGHT	SEX	AGE
Mel	1	58	W	23
Tom	2	78	M	54
Bill	3	92	M	31

1// Example output with ignoreColumns: [2, 3]
2[
3    {
4        NAME: 'Mel', PLACE: '1', AGE: '23'
5    },
6    {
7        NAME: 'Tom', PLACE: '2', AGE: '54'
8    },
9    {
10        NAME: 'Bill', PLACE: '3', AGE: '31'
11    }
12]

onlyColumns

Array of indexes that are taken, starting with 0. Default is 'null/undefined'. If given, this option overrides ignoreColumns.

NAME	PLACE	WEIGHT	SEX	AGE
Mel	1	58	W	23
Tom	2	78	M	54
Bill	3	92	M	31

1// Example output with onlyColumns: [0, 4]
2[
3    {
4        NAME: 'Mel', AGE: '23'
5    },
6    {
7        NAME: 'Tom', AGE: '54'
8    },
9    {
10        NAME: 'Bill', AGE: '31'
11    }
12]

ignoreHiddenRows

Indicates if hidden rows (display:none) are ignored. Default is true:

NAME	PLACE	WEIGHT	SEX	AGE
Mel	1	58	W	23
Tom	2	78	M	54
Bill	3	92	M	31

• Cat | 4 | 4 | W | 2*

1// Example output with ignoreHiddenRows:true
2[
3    {
4        NAME: 'Mel', PLACE: '1', WEIGHT: '58', SEX: 'W', AGE: '23'
5    },
6    {
7        NAME: 'Tom', PLACE: '2', WEIGHT: '78', SEX: 'M', AGE: '54'
8    },
9    {
10        NAME: 'Bill', PLACE: '3', WEIGHT: '92', SEX: 'M', AGE: '31'
11    }
12]
13// Example output with ignoreHiddenRows:false
14[
15    {
16        NAME: 'Mel', PLACE: '1', WEIGHT: '58', SEX: 'W', AGE: '23'
17    },
18    {
19        NAME: 'Tom', PLACE: '2', WEIGHT: '78', SEX: 'M', AGE: '54'
20    },
21    {
22        NAME: 'Bill', PLACE: '3', WEIGHT: '92', SEX: 'M', AGE: '31'
23    }
24    },
25    {
26        NAME: 'Cat', PLACE: '4', WEIGHT: '4', SEX: 'W', AGE: '2'
27    }
28]

headings

Array of Strings to be used as headings. Default is null/undefined.

If more headings are given than columns exist the overcounting ones will be ignored. If less headings are given than existing values the overcounting values are ignored.

NAME	PLACE	WEIGHT	SEX	AGE
Mel	1	58	W	23
Tom	2	78	M	54
Bill	3	92	M	31

• Cat | 4 | 4 | W | 2*

1// Example output with headings: ['A','B','C','D','E']
2[
3    {
4        A: 'Mel', B: '1', C: '58', D: 'W', E: '23'
5    },
6    {
7        A: 'Tom', B: '2', C: '78', D: 'M', E: '54'
8    },
9    {
10        A: 'Bill', B: '3', C: '92', D: 'M', E: '31'
11    }
12]
13// Example output with headings: ['A','B','C']
14[
15    {
16        A: 'Mel', B: '1', C: '58'
17    },
18    {
19        A: 'Tom', B: '2', C: '78'
20    },
21    {
22        A: 'Bill', B: '3', C: '92'
23    }
24]
25// Example output with headings: ['A','B','C','D','E','F','G','H']
26[
27    {
28        A: 'Mel', B: '1', C: '58', D: 'W', E: '23'
29    },
30    {
31        A: 'Tom', B: '2', C: '78', D: 'M', E: '54'
32    },
33    {
34        A: 'Bill', B: '3', C: '92', D: 'M', E: '31'
35    }
36]
37// Example output with headings: ['A','B','C'] && ignoreColumns: [2, 3]
38[
39    {
40        A: 'Mel', B: 'W', C: '23'
41    },
42    {
43        A: 'Tom', B: 'M', C: '54'
44    },
45    {
46        A: 'Bill', B: 'M', C: '31'
47    }
48]
49

limitrows

Number of rows to which the resulting object should be limited to. Default is null/undefined.

Huge Table (see test/tables.html)

Roleplayer Number	Name	Text to say
0	Raife Parkinson	re dolor in hendrerit in vulputate ve
1	Hazel Schultz	usto duo dolores et ea rebum. Ste
2	Montana Delgado	psum dolor sit amet. Lorem ipsum dolor
3	Dianne Mcbride	sit ame olor sit amet. Lorem ipsum
4	Xena Lynch	us est Lorem ipsum dol
5	Najma Holding	akimata sanctus est Lorem ipsum dolor sit
6	Kiki House	ame nvidunt ut
...
197	Montana Delgado	lores et ea rebum. Stet clita kasd gu a
198	Myrtle Conley	rebum. Stet clita kasd gubergren, no sea
199	Hanna Ellis	kimata sanctus est Lorem ipsum dolor si

Example output with limitrows: 5

1[ { 'Roleplayer Number': '0',
2        Name: 'Raife Parkinson',
3        'Text to say': 're dolor in hendrerit in vulputate ve' },
4      { 'Roleplayer Number': '1',
5        Name: 'Hazel Schultz',
6        'Text to say': 'usto duo dolores et ea rebum. Ste' },
7      { 'Roleplayer Number': '2',
8        Name: 'Montana Delgado',
9        'Text to say': 'psum dolor sit amet. Lorem ipsum dolor sit ame' },
10      { 'Roleplayer Number': '3',
11        Name: 'Dianne Mcbride',
12        'Text to say': 'olor sit amet. Lorem ipsum' },
13      { 'Roleplayer Number': '4',
14        Name: 'Xena Lynch',
15        'Text to say': 'us est Lorem ipsum dol' } ]

containsClasses

Array of classes to find a specific table using this class. Default is null/ undefined.

Known issues and limitations

This module only supports parsing basic tables with a simple horizontal set of <th></th> headings and corresponding <td></td> cells.

It can give useless or weird results on tables that have complex structures (such as nested tables) or multiple headers (such as on both X and Y axis).

You'll need to handle things like work out which tables to parse and (in most cases) clean up the data. You might want to combine it it with modules like json2csv or CsvToMarkdownTable.

You might want to use it with a module like 'cheerio' if you want to parse specific tables identified by id or class (i.e. select them with cheerio and pass the HTML of them as a string).

Example usage

1// Convert an HTML blob into an array of all the tables on the page
2var tabletojson = require('tabletojson').Tabletojson;
3var tablesAsJson = tabletojson.convert(html);
4var firstTableAsJson = tablesAsJson[0];
5var secondTableAsJson = tablesAsJson[1];
6...

1// Fetch a URL and parse all it's tables into JSON, using a callback
2var tabletojson = require('tabletojson').Tabletojson;
3var url = 'https://en.wikipedia.org/wiki/List_of_ISO_639-1_codes';
4tabletojson.convertUrl(url, function(tablesAsJson) {
5  var listofSovereignStates = tablesAsJson[0];
6});

1// Fetch a URL and parse all it's tables into JSON, using promises
2var tabletojson = require('tabletojson').Tabletojson;
3var url = 'http://en.wikipedia.org/wiki/List_of_countries_by_credit_rating';
4tabletojson.convertUrl(url)
5.then(function(tablesAsJson) {
6  var standardAndPoorRatings = tablesAsJson[1];
7  var fitchRatings = tablesAsJson[2];
8});

1// Fetch a table from Wikipedia and combine with json2csv to convert to CSV
2var tabletojson = require('tabletojson').Tabletojson;
3var json2csv = require('json2csv');
4var url = 'http://en.wikipedia.org/wiki/List_of_countries_by_credit_rating';
5tabletojson.convertUrl(url)
6.then(function(tablesAsJson) {
7  var standardAndPoorCreditRatings = tablesAsJson[1];
8  json2csv({ data: standardAndPoorCreditRatings,
9             fields: [ 'Country', 'Outlook']
10           }, function(err, csv) {
11            console.log(csv);
12            /* Example output
13              "Country","Outlook"
14              "Abu Dhabi, UAE","Stable"
15              "Albania","Stable"
16              "Andorra","Negative"
17              "Angola","Stable"
18              "Argentina","Negative"
19              "Aruba","Stable"
20              "Australia","Stable"
21              "Austria","Negative"
22              "Azerbaijan","Positive"
23              ...
24            */
25          });
26});

Issues

Right now the table needs to be "well formatted" to be convertable. Tables in tables with not be processed.

1<thead>
2    <tr>
3        <th>Header</th>
4    <tr>
5</thead>

Contributing

Improvements, fixes and suggestions for better written modules that other people have created are welcome, as are bug reports against specific tables it is unable to handle.

You can find basic tests in the test folder. I implemented the most straight forward way in using the library. Nonetheless there are some edge cases that need to be tested and I would like to ask for support here. Feel free to fork and create PRs here. Every bit of help is appreciated.

To get also an insight you can use Iain's examples located in the example folder included with this project that shows usage and would be a good start.

If you submit a pull request, please add an example for your use case, so I can understand what you want it to do (as I want to get around to writing tests for this and want to understand the sort of use cases people have).

Thanks

June 2018 - Very special thanks to the originator of the library, Iain Collins (@iaincollins). Without his investigation in website grasping and mastering cheerio this lib would have not been where it is right now. Also I would personally like to say "Thank you" for your trust in passing me the ownership. Marius (@maugenst)

Additional thanks to

• @roryok
• Max Thyen (@maxthyen)
• Thor Jacobsen (@twjacobsen)
• Michael Keller (@mhkeller)
• Jesús Leganés-Combarro (@piranna)
• João Otávio Ferreira Barbosa (@joaobarbosa)

for improvements and bug fixes.