htmlparser2
updated from 8.0.1 to 8.0.2 (release notes) - this fixes broken parsing in certain situations: #285;deepmerge
updated from 4.3.0 to 4.3.1 - no functional changes;All commits: 9.0.4...9.0.5
dataTable
formatter was missing some existing cells in incompletely defined tables: #282;All commits: 9.0.3...9.0.4
dom-serializer
dependency: #269.All commits: 9.0.2...9.0.3
encodeCharacters
option: #267.All commits: 9.0.1...9.0.2
BlockTextBuilder.closeTable
: #264;BlockTextBuilder.closeTable
.All commits: 9.0.0...9.0.1
All commits: 8.2.1...9.0.0
Version 9 roadmap: #240
Request for comments: #261 [RFC] Naming issue - please take a look and share opinions while you're here
Required Node version is now >=14.
Package now provides cjs
and mjs
exports.
If you use CLI then install that package instead.
The new package uses new arg parser aspargvs instead of minimist in order to deal with the vast options space of html-to-text
.
htmlparser2
updated from 6.1.0 to 8.0.1 (Release notes);he
dependency is removed. It was needed at the time it was introduced, apparently, but at this point htmlparser2
seems to do a better job itself.decodeOptions
section removed with he
dependency;fromString
method removed;BlockTextBuilder
methods are now removed.Refer to README for migration instructions.
decodeEntities
- controls whether HTML entities found in the input HTML should be decoded or left as is in the output text;encodeCharacters
- a dictionary with characters that should be replaced in the output text and corresponding escape sequences.New generic formatters blockString
, blockTag
, blockHtml
, inlineString
, inlineSurround
, inlineTag
, inlineHtml
cover some common usage scenarios such as #231.
anchor
and image
got pathRewrite
option;dataTable
formatter allows zero colSpacing
.openList
, openListItem
, closeListItem
, closeList
). Addresses #238;startNoWrap
, stopNoWrap
- allows to keep local inline content in a single line regardless of wrapping options;addLiteral
- it is like addInline
but circumvents most of the text processing logic. This should be preferred when inserting markup elements;builder.metadata
. This allows to compile the converter once and still being able to supply per-document data. Metadata object is supplied as the last optional argument to convert
function and the function returned by compile
function;closeTable
function got a required property in the options object - tableToString
function, and previously existed colSpacing
and rowSpacing
are removed (now a responsibility of the tableToString
function).tags
option support. Addresses #253.No changes in published package. Bumped dev dependencies and regenerated package-lock.json
.
Fix for the issue #249 and possibly other obscure issues when some selector options are ignored. options.selectors
array was not fully processed before.
Bump minimist
dependency, regenerate package-lock.json
.
preserveNewlines
option is used. Addresses #232;linkBrackets
option - it accepts an array of two strings (default: ['[', ']']
) or false
to remove the brackets. Addresses #236;noLinkBrackets
formatters option is now deprecated.All commits: 8.0.0...8.1.0
All commits: 7.1.1...8.0.0
Version 8 roadmap issue: #228
The main focus of this version. Addresses the most demanded user requests (#159, #179, partially #143).
It is now possible to specify formatting options or assign custom formatters not only by tag names but by almost any selectors.
See the README Selectors section for details.
Note: The new selectors
option is an array, in contrast to the tags
option introduced in version 6 (and now deprecated). Selectors have to have a well defined order and object properties is not a right tool for that.
Two new packages were created to enable this feature - parseley and selderee.
The same selectors implementation is used now to narrow down the conversion to specific HTML DOM fragments. Addresses #96. (Previous implementation had more limited selectors format.)
BREAKING CHANGE: All outermost elements matching provided selectors will be present in the output (previously it was only the first match for each selector). Addresses #215.
limits.maxBaseElements
can be used when you only need a fixed number of base elements and would like to avoid checking the rest of the source HTML document.
Base elements can be arranged in output text in the order of matched selectors (default, to keep it closer to the old implementation) or in the order of appearance in source HTML document.
BREAKING CHANGE: previous implementation was treating id selectors in the same way as class selectors (could match <foo id="a b">
with foo#a
selector). New implementation is closer to the spec and doesn't expect multiple ids on an element. You can achieve the old behavior with foo[id~=a]
selector in case you rely on it for some poorly formatted documents (note that it has different specificity though).
Since options preprocessing is getting more involved with selectors compilation, it seemed reasonable to break the single htmlToText()
function into compilation and convertation steps. It might provide some performance benefits in client code.
compile(options)
returns a function of a single argument (html string);htmlToText(html, options)
is now an alias to convert(html, options)
function and works as before.baseElement
;returnDomByDefault
;tables
;tags
.Refer to README for migration instructions.
No previously deprecated stuff is removed in this version. Significant cleanup is planned for version 9 instead.
Bump minimist
dependency and dev dependencies, regenerate package-lock.json
.
Regenerate package-lock.json
.
htmlparser2
updated from 6.0.0 to 6.1.0 (Release notes);Required Node version is now >=10.23.2.
lodash
dependency is removed;htmlparser2
updated from 4.1.0 to 6.0.0 (Release notes, also domhandler). There is a slim chance you can run into some differences in case you're relying on it heavily in your custom formatters;BlockTextBuilder methods now accept option objects for optional arguments. This improves client code readability and allows to introduce extra options with ease. It will see some use in future updates.
Positional arguments introduced in version 6.0.0 are now deprecated. Formatters written for the version 6.0.0 should keep working for now but the compatibility layer is rather inconvenient and will be removed with the next major version.
See the commit f50f10f. Changes in lib/formatter.js
file are illustrative for how to migrate to the new API.
All commits: 6.0.0...7.0.0
Version 7 roadmap issue: #222
This is a major update. No code left untouched. While the goal was to keep as much compatibility as possible, some client-facing changes were unavoidable.
Since the library has the only exported function, it is now self-titled.
Formatting code was rewritten almost entirely to make it aware of block-level tags and to handle HTML whitespace properly. One of popular requests was to support divs, and it is here now, after a lot of effort.
Options are reorganized to make room for some extra format options while making everything more structured. Now tag-specific options live within that tag configuration.
For the majority of changed options there is a compatibility layer that will remain until next major release. But you are encouraged to explore new options since they provide a bit more flexibility.
Because formatters are integral part of the formatting code (as the name suggests), it wasn't possible to provide a compatibility layer.
Please refer to the Readme to see how things are wired now, in case you were using them for anything other than dealing with the lack of block-level tags support.
Cells can make use of extra space with colspan and rowspan attributes. Max column width is defined separately from global wordwrap limit.
Multiple options to cut content in large HTML documents.
By default, any input longer than 16 million characters will be truncated.
Required Node version is now >=8.10.0.
Dependency versions are bumped.
https://github.com/html-to-text/node-html-to-text is the new home.
GitHub should handle all redirects from the old url, so it shouldn't break anything, even if you have a local fork pointing at the old origin. But it is still a good idea to update the url.
Version 6 roadmap issue: #200
preserveNewLines
whitespace issue fixed #162The function fromFile
is removed. It was the main reason html-to-text
could not be used in the browser #164.
You can get the fromFile
functionality back by using the following code
```js const fs = require('fs'); const { fromString } = require('html-to-text');
// Callback version const fromFile = (file, options, callback) => { if (!callback) { callback = options; options = {}; } fs.readFile(file, 'utf8', (err, str) => { if (err) return callback(err); callback(null, fromString(str, options)); }); };
// Promise version const fromFile = (file, option) => fs.promises.readFile(file, 'utf8').then(html => fromString(html, options));
// Sync version const fromFileSync = (file, options) => fromString(fs.readFileSync(file, 'utf8'), options); ```
Node versions < 6 are no longer supported.
unorderedListItemPrefix
added.format
option #128ol
type values 1
, a
and A
noLinksBrackets
options added #119htmlparser
to htmlparser2
#113uppercaseHeadings
added. #86baseElement
option. #83longWordSplit
option. #83decodeOptions
added.Breaking Changes: