When having a loop inside a pptx table, the height of the frame was not updated if a tag in the form of : <a:ext uri="{11111111-1111-1111-1111-111111111111}">
was present in the document.
Now, those tags are ignored and the height of the table should be updated appropriately.
When having a loop inside a pptx table, the height of the frame will automatically be updated if some rows are added or removed.
In previous versions, the table would keep the previous height, meaning the added rows would not be shown, except after forcing a rerendering of the table by changing the fontsize.
Add support for "lambdas", eg if a value in the data is a function, that function will be called with the scope
and the scopeManager
.
You now can write :
const doc = new Docxtemplater(zip);
doc.render({
userGreeting: (scope) => {
return "How is it going, " + scope.user + " ? ";
},
users: [
{
name: "John",
},
{
name: "Mary",
},
],
});
With the following template :
{#users}
{userGreeting}
{/}
Add support to remove the call to setData
or resolveData
.
(The setData function and resolveData will still work, but will be dropped in Docxtemplater v4)
You can now do :
const doc = new Docxtemplater(zip, { linebreaks: true });
doc.render({
user: "John",
});
or, in the async version :
const doc = new Docxtemplater(zip, { linebreaks: true });
doc
.renderAsync({
user: new Promise(function (resolve, reject) {
resolve("John");
}),
})
.then(function () {
const zip = doc.getZip().generate({
type: "nodebuffer",
compression: "DEFLATE",
});
});
Bugfix error : Cannot read property 'type' of undefined
when having a loop inside a loop that has an empty content, such as :
{#a}
{#b}{/}{/}
Add typescript typings to InspectModule
Internal cleanup :
-
Remove internal unused function
getNearestLeftIndex
,getNearestRightIndex
,getNearestLeftIndexWithCache
,getNearestRightIndexWithCache
from doc-utils. -
Add explanation to all errors that have id
Use @xmldom/xmldom instead of xmldom
See : xmldom/xmldom#271
Bugfix to avoid loosing an image that is contained inside a paragraph loop.
The image would disappear only when the image was followed directly by the end of the loop, like this :
{#loop}
[IMAGE]
{/loop}
Bugfix to avoid error message Cannot read property 'offset' of undefined
on some rare templates.
The internal mergesort
function had a bug.
Bugfix to avoid dropping section when having a loop right before a section break
When having a loop containing a page section break, the generated document had continuous section breaks in the output.
With this version, the page section break should be rendered as page section breaks, and create as many pages as there are items in the loop.
Update InspectModule to make getTags function work well with XLSXModule
Disallow calling loadZip if using v4 constructor
Correctly handle <<tag>>
delimiters when saved in word as <<tag>>
Fixes #606
- Add resolveOffset internal property on scope managers to be able to handle the slides loop case correctly.
- Update moduleApiVersion to 3.26.0.
Internal change to allow to match tags with non-breaking space.
When entering Ctrl+Shift+Space
, a "non-breaking" space is created.
When using a non-breaking space in the tag {:subtemplate doc}
, or `{:table data}, the tag would not be replaced correctly. You need to upgrade the PRO modules as well to their latest version in order to fix the bug.
Bugfix issue with spaces that would disappear after rendering :
Correctly add xml:space="preserve" to all relevant w:t tags, ie all <w:t>
tags that contain a placeholder.
Update algorithm for loops to have more chance to show the underlying error.
For example, with the following template :
{#a}
{#b}
{#c}
{/d}
{/d}
{/c}
{/b}
{/a}
The algorithm will now show that the {/d} tags are involved in the issue.
Automatically add empty paragraph after table if the table is the last element in a document.
Change how the resolve algorithm works internally.
This fixes a bug in the slides module and image module, when used together, in async mode, where all image tags coming after a condition or loop would not be shown.
Because this induces an internal incompatible change, if you use the footnotes module, you have to upgrade the footnotes module to the version 3.2.0
Update moduleApiVersion to 3.25.0.
When having a document that has breaks to change the amount of columns, and loops surrounding the breaks, some pagebreaks could appear in the document.
Now, the section breaks should no more be transformed into page breaks.
When having a document containing a table like this :
================================
| {#users} abc |
--------------------------------
| {/users}{#cond} def {/cond} |
================================
Docxtemplater will now throw an error : id: "unbalanced_loop_tags"
,
and explanation: Unbalanced loop tags {#users}{/users}{#cond}{/cond}
,
In previous versions, this would most likely lead to a corrupt document.
The reason is that when a loop tag is spanning across multiple cells of a table, it will automatically expand to "row-mode", eg it will create 2 rows for each user in the users iterable.
However, the {#cond} loop tag is in one single cell, hence it will not expand to "row-mode".
Instead, the template should be written like this if the cond loop should be part of the users loop :
================================
| {#users} abc |
--------------------------------
| {#cond} def {/cond}{/users} |
================================
or like this if the loops are to be kept separate
===========================
| {#users} abc {/users} |
---------------------------
| {#cond} def {/cond} |
===========================
Don't override global configuration for docx/pptx when using a module, instead, a new fresh configuration is passed to a docxtemplater instance.
Fail with a clear error message when instantiating Docxtemplater with a null value, like this :
new Docxtemplater(null);
This will now throw :
The first argument of docxtemplater's constructor must be a valid zip file (jszip v2 or pizzip v3)
Fix speed regression when having big document with many rawxml tag.
This was caused by the fact that expandToOne was a recursive function.
Now that the function uses simple for loop, the rendering is way faster.
If you use the PRO word-run module, you should update the word-run-module to version 3.1.1
Bugfix when data contains double array, the scope parser would not do the right thing.
With the following template + data :
{#a}{.}{/a}
and the following data :
{ a: [[ "b", "c"]] }
This would render just the "c", but it should render the array [ "b", "c" ]
which will render as b,c
Avoid corruption when having self closing <w:sdtContent/>
in the document.
Throw an error when calling render on an invalid template.
Before this version, it was possible to do the following on an invalid template :
try {
doc.compile();
}
catch (e) {
doc.render();
}
And, this would produce, most of the times, a generated document that is corrupted.
Now, doc.render()
will also throw in this case the error "You should not call .render on a document that had compilation errors"
Update typescript bindings to be able to write doc.resolveData()
[Internal Only, for tests] : Rewrite xml-prettify to handle canonicalization of <w:t>Hello></w:t>
When there are errors both in the header and the footer, all errors are shown up instead of seeing only the errors of the first parsed file. This helps to find errors more quickly.
Huge performance improvements when using resolveData with loops with many iterations or with huge content inside the loop. On a document with loops of 8000 iterations, a total of 130000 Promises were created, now only about 8000 are created, (This is a 16x improvement in this case). This will produce noticeable improvements in particular in browser environments, where creating many promises costs more than in Node JS.
Some other improvements to call the parser()
function only when needed, and using some internal caching.
If you're using the PRO xlsx module, you should upgrade the module to version 3.3.4 or higher, or you might see some rendered values duplicated at multiple points in the generated spreadsheet.
- Throw error if calling
.resolveData(data)
before.compile()
- Add TypeScript typings
Bugfix corruption when having loops containing one section.
In that case, the generated file would be marked as corrupt by Word.
Do not mutate options when calling setOptions.
This means that if you do :
const options = { paragraphLoop: true };
doc.setOptions(options);
The options object will not be mutated at all anymore. Before this release, this could lead to fatal errors if using the options object across multiple Docxtemplater instances.
When using docxtemplater in async mode, inside loops, rejections would be ignored.
With this version, if one or more tags turn into a rejected Promise,
doc.resolveData(data)
will also reject (with a multi error containing all
suberrors)
Add support for dotx and dotm file formats
- Make expandToOne recursive, to allow to have multiple tags expand to the same level. (for example, for multiple "paragraph-placeholder" tags in the same paragraph).
- Update moduleApiVersion to 3.24.0.
- Make sure all postparse (also recursive ones) get the options such as filePath and contentType
Make docxtemplater fail with following error messages when using the v4 constructor with either setOptions or attachModule :
setOptions() should not be called manually when using the v4 constructor
attachModule() should not be called manually when using the v4 constructor
You should not write :
const doc = new Docxtemplater(zip);
doc.setOptions({
delimiters: {
start: "<",
end: ">",
},
});
doc.attachModule(new ImageModule())
You should instead write :
const doc = new Docxtemplater(zip, {
modules: [new ImageModule()],
delimiters: {
start: "<",
end: ">",
},
});
- Update moduleApiVersion to 3.23.0.
- Add
contentType
property in options passed to parse, postparse, render and postrender. - Bugfix in apiversion check, the patch version was not taken into account at all previously
Add proofstate module to allow to remove the <w:proofState w:grammar="clean" w:spelling="clean"/>
tag during .render()
- Add support for automatically detaching modules that do not support the current filetype when using constructor v4. In previous versions, you would do the following:
let doc = new Docxtemplater();
const zip = new PizZip(buffer)
doc.loadZip(zip);
if (doc.fileType === "pptx") {
doc.attachModule(new SlidesModule());
}
Now it is possible to write the following, without needing the condition on filetype:
const zip = new PizZip(buffer)
const options = {
modules: [new SlidesModule()],
}
try {
const doc = new Docxtemplater(zip, options);
}
catch (e) {
console.log(e);
// error handler
}
- Update moduleApiVersion to 3.22.0.
- Add a constructor method that accepts zip and optionally modules and other options. This constructor will be the official constructor in docxtemplater v4 and the methods:
loadZip
,attachModule
,setOptions
andcompile
will no more be available.
You can migrate the following code:
const doc = new Docxtemplater();
doc.loadZip(zip)
doc.setOptions({
delimiters: {
start: "<",
end: ">",
},
});
doc.attachModule(new ImageModule())
doc.attachModule(new Htmlmodule())
doc.attachModule(new Pptxmodule())
try {
doc.compile();
}
catch (e) {
// error handler
}
to
const options = {
modules: [new ImageModule(), new Htmlmodule(), new Pptxmodule()],
delimiters: {
start: "<",
end: ">",
},
}
try {
const doc = new Docxtemplater(zip, options);
}
catch (e) {
// error handler
}
- This change is backward compatible, meaning that you can continue to use the constructor with no arguments for the time being.
- Add specific error (
Duplicate open tag
andDuplicate close tag
) when using{{foobar}}
in a template when the delimiters are just one single{
and}
- Avoid error
TypeError: Cannot set property 'file' of undefined
that hides the real error
- Properly decode
&gt;
into>
in templates, in previous versions, the value was decoded to>
- Pass in
{match, getValue, getValues}
functions to second argument of module.parse - Allow to have a promise at the root level in resolveData
- Update moduleApiVersion to 3.21.0
(Internal, for modules) Pass whole part instead of just part.value to explanation function
Bugfix with paragraphLoops containing pagebreaks + tables, the pagebreak got reordered
Bugfix corrupt document when using dashloop : {-a:p loop}
inside table cell which renders false
Other bugfixes related to pagebreaks and loops
Bugfix issue with paragraphLoop and pagebreaks.
In some situations, the pagebreak got removed if the start of the loop is at the top of the page (right after the w:br
element).
- Add options.lIndex and other information about the current token in the
module.parse
method. - Update moduleApiVersion to 3.20.0
Since this version, modules using the expandToOne trait will be able to throw a specific error instead of the one related to the raw-xml module, ("Raw tag not in paragraph")
Bugfix for loop module contained inside grid slides module : Do not fail with error "Unopened loop"
Bugfix for loop module and raw module : when having an "execution" error in the parser (eg on render, after the compilation of the template, for example with the angular parser when a filter is called with bad arguments), if one {#looptag}
or {@rawtag}
had an execution error, the code would immediately stop collecting other errors. Now docxtemplater will collect all errors, for simple tags, for loop tags, and for rawxml tags.
This means that if your template was like this :
{#foobar|badfilter1}test{/}
{#foobar|badfilter2}test{/}
{@rawvalue|badfilter3}
Before version 3.16.0, calling render
on this template would throw an error with just the first tag erroring.
Since version 3.16.0, a multi error is thrown with all three tags in error being in .properties.errors
.
This is fixed in sync mode (when calling doc.render
directly) and also in async mode (when calling doc.resolveData
and then doc.render
)
- Add scopePathLength to scopemanager to be able to write "$isLast".
- Update moduleApiVersion to 3.19.0
Add offset property to following errors :
"No w:p was found at the left"
"No w:p was found at the right"
Bugfix when having dashloop inside normal loop, like this :
{#foo}test {-w:p loop}loop{/}{/}
we now throw a multi error instead of just the first error encountered.
- Update moduleApiVersion to 3.18.0
- (internal) Use array of string for shouldContain to fix corruptions
- Do not add
<w:p>
in<w:sdtContent>
if it contains a<w:r>
- If you use the xlsx module, please update the xlsx module to 3.1.1 or you will get a stacktrace
Cannot set property 'parse' of undefined
- Update moduleApiVersion to 3.17.0
- Add
options.parse
function inparse(placeholder, options)
to allow "recursive" parsing - Fill empty
<w:sdtContent>
with<w:p>
to avoid corruption
- Update moduleApiVersion to 3.16.0
- Disallow to call attachModule twice on a given module
- Add options to
parse
andpostparse
such as filePath
Use {}
(empty object) instead of "false" for the fs
shim in the browser.
- Bugfix for joinUncorrupt method, which produced a corrupt document when having a paragraphLoop containing a table containing a
{-w:tr loop}
loop.
-
Update moduleApiVersion to 3.15.0
-
Add
joinUncorrupt
tooptions
inmodule.render
to be able to fix corruptions -
Fixes corruption when using
{#loops}
that contain tables with{@rawXML}
evaluating to empty.
- Add p:txBody to pptx-lexed to handle corruption on pptx
-
Update moduleApiVersion to 3.14.0
-
Add
getNearestLeftIndex
,getNearestRightIndex
,getNearestLeftIndexWithCache
,getNearestRightIndexWithCache
to doc-utils. -
Add
preparse
method to modules to be able to retrieve info about document before the parsing of the whole document by other modules. -
Add
tagShouldContain
to the filetype API to be able to make the documents valid, which would make it possible to remove thepart.emptyValue
trick.
Bugfix when using paragraphLoop section with a pagebreak : at the end of the paragraphLoop, the pagebreak was removed by mistake. This is no longer the case since this version
Bugfix speed issue when having many loop tags in the same paragraph (over 50 loop tags in one paragraph).
(internal) Update verify moduleAPI to 3.13.0 (addition of getNearestLeftWithCache
, getNearestRightWithCache
, buildNearestCache
)
Throw Error with offsets when having loop start tag inside table and loop end tag outside table
Update xmlprettify and test it
Update dependencies
Add support for .docm
format
Document the usage of PizZip instead of JSZip
Improved fix for 3.13.5, to still create a scope but that is the same as the parent scope (without deep copy)
Bugfix condition when value is truthy, the subscope was incorrect
(This is a fix on top of 3.13.3, which fixed a bug for the value true
but not for other truthy values such as integers, strings).
Fix support for Node 6 (also for Nashorn): Rebuild js after upgrade to latest babel version
Bugfix condition when value is === true, the subscope was incorrect
Auto verify moduleAPI version on attachModule
Fix undefined is not a function evaluating 'object.getOwnPropertyDescriptors', by downgrading to babel 7.4
-
Throw multi Error with offsets when having rendering error (makes it possible to use the error location module on rendering errors).
-
Avoid false positive error "The filetype for this file could not be identified" when the document uses
<Default>
in the[Content_Types].xml
file -
Add getFileType module API to add potential support for other filetypes
Add support for setting the same delimiter on the start and beginning, for example :
%user%
, by doing doc.setOptions({delimiters: {start: "%", end: "%"}});
Remove (internal) scopeManager.clone, which is no more used
Add offset property to unopened/unclosed loop error
Parse selfclosing tag correctly, such as in <w:rPr><w:noProof/></w:rPr>
- Bugfix issue with conditions in async mode conflicting sometimes
For example with the template :
{#loop}
{#cond2}
{label}
{/cond2}
{#cond3}
{label}
{/cond3}
{/loop}
and the data :
{
label: "outer",
loop: [
{
cond2: true,
label: "inner",
},
],
}
The label would render as outer
even though it should render as inner
.
- Bugfix speed issue introduced in 3.10.0 for rawXmlModule
-
Bugfix corruption when using paragraphLoop with a template containing
<w:p>
as direct childs of other<w:p>
. -
Update getLeft, getRight, getNearestLeft, getNearestRight, getLeftOrNull, getRightOrNull to take into account the nesting of tags.
-
Ensure that the
f
functor is always called when using chunkBy (like the native.map
function) -
In expandOne trait, call
getLeft
beforegetRight
- Improve detection of filetype to read information directly from
[Content_Types].xml
instead of from zip files.
- WIPED - Published by mistake.
-
Add support for multiple elements in getRightOrNull and getLeftOrNull (to be able to specify array of elements in traits.expandToOne
-
Add postparse to expandToOne
-
Add
index
andprefix
property torender
method in the second argument (options.prefix
andoptions.render
) -
Make sure that postrender returns an array of the same length as it gets
-
Update moduleApiVersion to 3.9.0
- Update moduleApiVersion to 3.8.0
When using a paragraphLoop inside a table like this :
{#loop} Content {/loop}
When loop was falsey, the table cell caused a corruption in the resulting document.
This now fixes the issue, an empty "<w:p>/w:p" is insereted.
Performance fix in inspectModule. This performance fix concerns you only if you do doc.attachModule(inspectModule)
.
On some templates, and when reusing the same inspectModule instance, the rendering could be very slow (20 minutes in some cases). On a simple test case, the rendering time has decreased from 6 seconds to 90ms.
Render all spaces when using linebreak option
Remove .babelrc from published module
Expose Lexer
to modules (API Version 3-7-0)
Fix issue with nested loops with paragraphLoop
:
{#user}{#pets}
{name}
{/}{/}
would produce No tag "w:p" was found at the left
It now renders the same way as without paragraphLoop
-
Add possibility to change the prefix of the rawxmlmodule and of the loopmodule
-
In parsed output, also add raw containing the full tag, for example {#loop} will be parsed to : {value: "loop", raw: "#loop"}
Fix bug with {linebreaks: true} throwing a.split is not a function
Add templating of more meta data of the document, including :
Author, Title, Topic, marks, Categories, Comments, Company
Fix bug with resolveData algorithm, which raised the error : "Cannot read property value of undefined" in the render call.
Fix bug wrong styling when using linebreaks.
- Add linebreaks option for pptx and docx
Revert : Add back lIndex to parsed in addition to endLindex : Fixes issue with async rendering of multiple tags (images, qrcode, ...)
-
Add nullGetter module API
-
Update inspectModule to have :
- Unused variables (nullValues)
- filetype
- data from setData()
- templatedFiles
- list of tags
In firefox, fix : Argument 1 of DOMParser.constructor does not implement interface Principal
Fix issue of remaining xmlns=""
in some docx files
Fix issue with loopmodule eating up whitespace characters
Allow to set delimiters to {start: '<', end: '>'}
#403
Add meta context argument to custom parser with information about the tag for each types of tags
Fix infinite loop when XML is invalid (throw an explicit error instead)
Fixes open-xml-templating#398
Fix open-xml-templating#397 : Add more information to context
in the parser to be able to write {#users}{$index} - {name}
.
See https://docxtemplater.readthedocs.io/en/latest/configuration.html#custom-parser for full doc.
Add scopeManager
argument to nullGetter to know what variable is undefined.
-
Move cli out of main repository : https://github.com/open-xml-templating/docxtemplater-cli
-
Get meta attributes with raw-xml tag.
- Fix #226 when using {@rawXml} tag inside table column, the document is no longer corrupted
- Use JSZipUtils in tests
-
Add resolveData function for async data resolving
-
Fix bugs with spacing : Now space="preserve" is applied in the first and last
<w:t>
of each placeholder
- Update getAllTags to work with multiple loops by merging fields.
-
Add getAllTags to inspectModule
-
Add base-modules after
loadZip
instead of oncompile
Breaking change : The syntax of change delimiter has been changed from :
{=<% %>}
to:
{=<% %>=}
To change to <%
and %>
delimiters.
The reason is that this allows to parse the delimiters without any assumption.
For example {={{ }}}
was not possible to parse at version 3.4.0, but by adding the ending equal sign : {={{ }}=}
, the ambiguity is removed.
Add change delimiter syntax from inside template :
For example :
* {default_tags}
{=<% %>}
* <% erb_style_tags %>
<%={ }%>
* { default_tags_again }
- Add
getTags
toInspectModule
.
-
Automatically strip empty namespaces in xml files
-
Do not throw postparsed errors for tags that are unclosed/unopened
Throw error if the output contains invalid XML characters
Take into account paragraphLoop for PPTX documents
Correctly replace titles of PPTX documents
Add support for Office365 generated documents (with word/document2.xml
file)
- Fix rendering issues with
paragraphLoop
When setting paragraphLoop
, the intention is to have a special case for loops when you write :
{#users}
{name}
{/users}
Eg : both the start of the loop and the end of the loop are in a paragraph, surrounded by no other content. In that particular case, we render the content of the loop ({name}
) in this use case, in a new paragraph each time, so that there would be no additional whitespace added to the loop.
On version 3.2.1, the paragraphLoop would change the rendering for most of the loops, for example, if you wrote :
My paragraph {#users}
{name}
{/users}
the paragraphLoop code was triggered, and if users was [], even the text "My paragraph" would be removed, which was a bug.
This release fixes that bug.
- Fix bug with tr loop inside
paragraphLoop
If doing
{#par}
======================
| {#row} | {/row} |
======================
{/par}
An unexpected error 'No "w:tr" found at the left' would be raised.
Add paragraphLoop
option, that permits to have better rendering for spaces (Fixes #272)
It is recommended to turn that option on, because the templates are more readable. However since it breaks backwards-compatibility, it is turned off by default.
Inspect postparsed in compile() method instead of render()
- Add support for self-closing tag in xmlMatcher -
- Add tag information in parsed/postparsed
- Review testing code to always use "real" docx
- Add support for setting the prefix when attaching a module.
- Bugfix when looping over selfclosing tag
Add error loop_position_invalid
when the position of the loop would produce invalid xml (for example, when putting the start of a loop in a table and the end outside the loop
Use createFolders JSZip option to avoid docx corruption
Show clear error if file is ODT, escape " as "
Template values in docProps/core.xml and docProps/app.xml
Add possibility to have RenderingErrors (if data is incorrect)
Fix RangeError: Maximum call stack size exceeded
with very big document
Handle unclosed tag <a:t /> and <w:t />
Bugfix loop over string value (fixes https://github.com/open-xml-templating/docxtemplater/issues/309\)
Add support for multi errors :
docxtemplater doesn't fail on the first error, but instead, will throw multiple errors at the time. See https://docxtemplater.readthedocs.io/en/latest/errors.html for a detailled explanation.
Add js/tests to npm package (for modules)
Reduce size of the package
Reduce size of the package
Reduce size of the package (718.9 MB to 1.0 MB), and add prepublish script to ensure the size will never exceed 1.5MB
Add expanded property when using expandTo
Bugfix : Do not decode utf8 in xmlDocuments
- When using getRenderedMap in the modules, we now pass the filepath of the file that will be generated instead of the path of the template.
- Remove cycle between traits and docutils
- Make sure fileTypeConfig is uptodate
-
Autodetection of filetype : if you use pptx, you don't have to write doc.setOptions({fileType: "pptx"}) anymore.
-
Intelligent tagging for pptx (Fixes issue #284)
- Update documentation
- Completely remove intelligentTagging (which is on for everyone)
- Performance improvements for arrays: prefer push over concat
- Add automatic module wrapping
- Add mecanism to change which files to template with modules
- The modules are now ordered in the following order : baseModules, and then attachedModules. Before version 3.0.2, the attachedModules came before baseModules. This fixes : https://github.com/open-xml-templating/docxtemplater-image-module/issues/76
This release was published by error, and should not be used at all.
-
The rendering of the templates has now multiple steps : Lexing, Parsing, and Rendering. This makes the code much more robust, they might be bugs at the beginning, but in the long run, the code is much easier to understand/debug/change.
-
The module system was completely changed, no compatibility is kept between version 2 and version 3, please make sure to update your modules if needed.
-
You can't have the same startTag and endTag (for example
$foo$
), because this would make the code more complex and the errorHandling quite impossible. -
All extended features (loop, rawxml, invertedloops), are written as v3 modules. We dogfood our own module system, and will hopefully improve it that way.
-
The constructor arguments have been removed, and you are responsible to load the JSZip.
Instead of :
var doc = new Docxtemplater(content);
You now should do :
var zip = new JSZip(content); var doc=new Docxtemplater().loadZip(zip)
-
getTags() has been removed. It is now not easily possible to get the tags. See open-xml-templating#258 for a alternate solution
The JSZip version that you use should be 2.x, the 3.x is now exclusively async, which Docxtemplater doesn't handle yet.
- Fix stacktrace not showing up (#245)
- Fixed a memory leak when using large document (10Mb document.xml) (#237)
- Add fileTypeConfig options in setOptions to define your own fileType config (#235)
- {@rawXml} has been fixed for pptx
- Add possibility to close loopTag with {/} #192
- Bug fix : Some times, docxtemplater could eat lots of memory due to the new "compilation" feature that was only experimental. As this feature is not yet used, it was completely removed from the codebase.
- Performance : The code has been made a little bit faster.
- Speed Improvements : docxtemplater had a regression causing it to be slow for loops. The regression would cause the code to run in O(n²) instead of O(n) where n is the length of the loops (with {#users}{name}{/users}. The bug is now fixed, and docxtemplater gained a lot of speed for users of lengthy loops.
- Breaking : To choose between docx or pptx, you now have to pass doc.setOptions({fileType:'docx'}) where the fileTypes are one of 'pptx', 'docx' (default is 'docx')
- Using es6 instead of coffeescript to code (code is still compiled to es5, to be usable with node v0.{10,12} or in the browser)
- Add finalize step in render function to throw an error if a tag is unclosed
- Module API has been updated, notably the tagXml property doesn't exist anymore, you should use the properties in
fileTypeConfig
- You should check if your modules have updated for the new module API, for instance, you should use version 1.0 of the docxtemplater image module
For example :
var xmlTemplater = this.manager.getInstance("xmlTemplater");
var tagXml = xmlTemplater.fileTypeConfig.tagsXmlArray[0];
- It is now possible to use tags inside of docx equations
- This release adds error management, all errors that are thrown by docxtemplater are now typed and contain additional properties.
- fix : When using loops with images inside, the modulemanager was not updated and would continue to return 'image' for the type of the tag even if the type changed
- feature : it is now possible to set the nullgetter for simple tags and for raw xml tags.
- bugfix for the modulemanager : it was not in sync in some cases
- They now is a default value for rawtags {@rawXml}, which is '' (this will delete the paragraph)
- bugfix (There was still '{' and '}' hardcoded)
- It is now possible to output the delimiters in the output (for example output "Mark } Z" with the template {name}
- scopeManager now doesn't return the string 'undefined' when the parser returns null. That behaviour is moved to the xmlTemplater class
- docxtemplater is now much faster to process loops #131
- docutils should not be used any more except for
DocUtils.defaults
(eg docutils is deprecated) - Module maintainers: please don't rely on docUtils anymore, except for docUtils.defaults
- Some templates would output corrupt templates, this should not happen anymore (if it still does, please open an issue)
Upgrade guide :
- If you use modules (image-module, chart-module, or others) its best to update those because they shouldn't use DocUtils anymore
- Add ScopeManager.loopOverValue
- {@rawXml} works in pptx
- Created new docxtemplater.com website
- Using mocha instead of jasmine-node
- New FAQ section in docs
- Corrupt documents when
<w:t>}{/body}</w:t>
- New events for moduleManager :
xmlRendering
andxmlRendered
- pptx generation was repaired 2be10b69d47e8c4ba0f541e4d201b29ef6281505
- header generation works with any amount of headers/footers
- slide generation works with any amount of slides
- doc.setOptions({delimiters:{start:”[“,end:”]”}}) now works e2d06dedd88860d2dac3d598b590bf81e2d113a6
- allowing same tag as start and end delimiter (eg @user@ instead of {user}) 32b9a7645f659ae835fd695e4d8ea99cc6bbec94
- made it possible to use objects to loop over (eg {#user} {name} {/user}) 97411cb3537be08f48ff707ac34d6aac8b008c50
- docxtemplater doesn’t depend on fs anymore, for transparency and security reasons.
loadFromFile
has been removed. You now just have to pass the content to Docxtemplater’s constructor.setTags
has been renamed tosetData
applyTags
has been renamed torender
- the constructor has changed: eg
new Docxtemplater(content,options)
will now callJSzip.load(content,options)
. - To pass options (such as the parser), you will have to call
setOptions
- The
output
function has been removed. You should now callgetZip().generate(options)
where the options are documented here: https://stuk.github.io/jszip/documentation/api_jszip/generate.html - the qrcode module has been removed, and will be developped in an other package that will be attached to docxtemplater
The QrCode Option had a security issue. If you don’t upgrade according to this, the functionality should continue to work but the leak will still be there. If you are running in the browser, the vulnerability will not affect you (no access to the filesystem). If the users can’t change the qrCodes or their value, you’re safe too.
If you set qrCode:true, you are affected. The Command Line is not affected as of v0.7.0 (but was on 0.6.3 and less). However the command line as of v0.7.0 is not more able to load images over the filesystem.
You should set qrCode to a function now, according to https://docxtemplater.readthedocs.io/en/latest/configuration.html#image-replacing.