Roosevelt

Teddy — Changelog

🔝 Scroll to top

2.0.5

  • Fixed a bug that caused markup supplied through a {variable|s} to be left out of the page entirely when it contained a <script>, <style>, <textarea>, or <pre> holding something that looks like a Teddy tag, such as a script with the string "<include src=x>" in it. The check that markup from the model opens and closes its own tags counted tags inside those elements as though they were real, so it judged the markup incomplete. Their contents are now skipped, the same as when those elements are written in the template.
  • Fixed a bug in the client-side build that caused Teddy tags inside a <script>, <style>, or <textarea> to be parsed and run, which garbled the element and could drop the markup after it. Their contents are now read as plain text, the way a browser reads them and the way the server build already did.
  • Fixed a bug in the client-side build that caused an unquoted attribute value such as class=greeting to be read as =greeting.
  • Fixed a spurious warning about a closing tag the template never opened when that closing tag appeared inside a <script>, <style>, <textarea>, or <pre>.
  • Updated dependencies.

2.0.4

  • Fixed text written inside an <include> that renders as a web component being dropped rather than passed through as the component's light DOM, so a component reading its content through a default <slot>, as <my-button>Click me</my-button> does, was given nothing to project. Only the elements written inside the include were kept, so an example mixing the two, like <my-card>before<b>bold</b>after</my-card>, lost everything but the <b>.
  • Updated dependencies.

2.0.3

  • Added support for <include> tags to render web components via a new as attribute.
  • Fixed everything inside a <template> element being dropped by the browser builds, and teddy tags written inside one never being evaluated. A template element keeps its children in a document fragment of its own rather than among its child nodes, and neither the parser nor the element queries reached into it, so a browser render lost the contents entirely while Node kept them. Node builds were unaffected.
  • Fixed custom elements being mangled by the browser builds. A tag name was read as stopping at the first character that could not begin one, so <my-card> came out as a <my> tag carrying an attribute called -card. Since a custom element's name must contain a hyphen, none of them rendered correctly in a browser. Node builds were unaffected.
  • Updated dependencies.

2.0.2

  • Added teddy.precompile and teddy.registerPrecompiled, which let a browser render from the JavaScript Teddy would otherwise have built for a template at runtime. A format option writes it as an ES module, as CommonJS, or as a plain script.
  • Fixed import teddy from 'teddy' being unable to read a template from the filesystem in the ESM build.
  • Improved performance in various places.
  • Updated dependencies.

2.0.1

  • Fixed a <loop> whose through names a {variable} in the middle of its path, rather than at its head, rendering nothing. Fixed the same in the value of a condition: An <if>, or a one line if-, whose value named a {variable} anywhere but at its head was compared against the text as written rather than against what the variable resolved to.
  • Updated dependencies.

2.0.0

  • Breaking: Removed support for non-well formed templates. A template, and any markup arriving through the model, now has to be complete on its own.
    • Related: {variable|s} will no longer repair malformed markup that arrived through the model. Teddy handed its own output back to an HTML parser on every pass, which silently closed unclosed tags and reordered crossed ones along the way, so a model value of <p>unclosed came out as <p>unclosed</p>. It now comes out as given, which is what the |s flag says it does. Well formed markup is unaffected. If you were relying on Teddy to tidy markup coming out of your model, it will no longer do that for you.
  • Breaking: Removed teddy.setMaxPasses() and the maxPasses param. Teddy no longer makes passes over a template, so there was nothing left for it to limit. A template that includes itself is now reported as that, naming the templates leading round the loop, rather than as having run out of passes.
  • Fixed a {variable} naming one of two model keys that differ only in case resolving to whichever of them came last, which depended on the order the keys happened to be in rather than on anything in the template. An exact match now wins. Lookups are otherwise unchanged and remain case insensitive: {Something} still finds something, and only a model holding both Something and something is affected.
  • Fixed a value set into the model during a render — an <include> <arg>, or a <loop> key or val — leaving behind an existing key that differed from it only in case, so that which of the two a later lookup found depended on the order the keys were in. The one being set now wins. This was most visible in the browser, where attribute names are lowercased and a model key of escapeTest would sit next to an <arg escapeTest> that had become escapetest.
  • Fixed {a{b}|s}, a variable whose name is built from another variable and which also carries a flag, rendering its value twice with a | between the copies. The name was being handed to the regular expression engine as written rather than matched as text, so the | in the flag was read as alternation.
  • Fixed a non-parsed block being emptied when the markup holding it was passed on through another variable, which is what a layout does when it is handed its page as an argument and writes it out with |s. The rendered page carries Teddy's own markers for the blocks it lifted out of parsing, and those were being mistaken for blocks of their own on the way through, so a <pre> or <noteddy> in a page rendered through a layout came out empty.
  • Fixed a whole template cache set with teddy.setCache() never being found again unless the model value it was keyed on happened to be a string. The value names an entry, and a name is a string, so a number used as one becomes its own digits; reading it back compared the two strictly and never matched. A key whose value was 0 or an empty string was also skipped, as though the model had said nothing about it, and now counts.
  • Improved performance considerably in several places.
  • Updated dependencies.

1.2.1

  • Fixed <include> tags inside a <loop> so that the included partial's <if>, <unless>, one line conditionals, and nested <loop> tags can read the loop's key and val. Previously the partial's contents were parsed against the model as it stood outside the loop, so a partial could print the loop's val through a {variable} but any tag that branched on it behaved as though it were not there. Includes are now expanded by the loop, once per iteration, against that iteration's model, which is how partials behave in other templating engines.
  • Updated dependencies.

1.2.0

  • Breaking: Disabled template caching for templates read from the filesystem by default. Teddy used to keep every template it read and never read it again, so editing a template had no effect until the process restarted or teddy.clearTemplates() was called. Caching is now off by default, which is how most other templating engines (e.g. ejs or pug) treat their own caching, so a template that changes on disk is picked up on the next render.
    • If you use Teddy as an Express view engine you do not need to do anything: Express supplies its own view cache setting, which it enables in production and disables in development, and Teddy now uses it. That means production keeps caching and development stops needing a restart.
    • Templates registered with teddy.setTemplate are unaffected. Those are always kept, since registering one by hand is the only way to supply a template in the browser.
    • To get the old behavior back, call teddy.setCacheTemplates(true) when instantiating Teddy, or pass cache: true in the model on each render.
  • Added teddy.setCacheTemplates(boolean) to control the above, and support for a cache option in the model, which takes precedence over both that setter and Express's view cache.
  • Altered teddy.clearTemplates() to now clear the templates that were read from the filesystem, not just the ones registered by hand.
  • Fixed <include> resolving its partial through the template registry rather than through the loader, which meant an included template could not be re-read after it changed.
  • Fixed one line conditionals so that a sequence of them on the same element is evaluated in full, rather than all but one being discarded. <p if-a true='class="x"' if-b true='data-y="z"'> now applies both outcomes.
  • Improved performance considerably in several places.
  • Made various improvements to the automated test suite.
  • Updated dependencies.

1.1.4

  • Added slicker live demo docs.
  • Added some performance improvements.
  • Fixed a bug when escaping escape tags or escape comments.
  • Updated dependencies.

1.1.3

  • Added <!--# escape comments -->.
  • Updated dependencies.

1.1.2

  • Added teddy.clearTemplates() method to clear the template cache.
  • Updated dependencies.

1.1.1

  • Added support for variables-with-dashes-in-their-names.
  • Updated dependencies.

1.1.0

  • Added <escape> tag.
  • Excluded <pre> tags from Teddy parsing. Add a parse attribute to prevent this behavior.
  • Fixed a bug that could cause infinite loops in variable parsing.
  • Updated dependencies.

1.0.1

  • Fixed a bug that caused client-side Teddy to try to fetch resources from the server inappropriately while parsing the template.
  • Updated dependencies.

1.0.0

  • Added support for a new <!--! server-side comments --> syntax.
  • Changed the behavior of <include> tags to display an error when attempting to include a template that does not exist.
  • Updated various dependencies.

0.6.26

  • Fixed a bug that could cause ${templateLiteral} strings to lose their $ character if they were meant to print as strings and not be parsed as Teddy variables.
  • Updated various dependencies.

0.6.25

  • Fixed bugs associated with selected-value and checked-value attributes.

0.6.24

  • Added to force hide or {varName} to force display variable. This is a per-var override for:
    • teddy.setEmptyVarBehavior('hide'): Will make it possible for variables which don't resolve to display as empty strings instead of displaying the variable.
      • Default: 'display'.
  • Added support for sourcing most Teddy attribute values dynamically from {variables}.
  • Added boolean logic to one-line if statements.
  • Added selected-value and checked-value attributes for setting what option to select or what checkbox / radio to check in a more concise way than using one-line ifs.
  • Fixed a bug that prevented boolean logic from working if there were multiple and or multiple or attributes on an element.
  • Updated various dependencies.

0.6.23

  • Added new <inline> tag for adding inline CSS or JS using Teddy variables. Usage is optional, but it can help you avoid syntax error warnings in your code editor.
  • Fixed a bug preventing <include> args from parsing correctly in certain circumstances.
  • Updated various dependencies.

0.6.22

  • Made variable name lookups case insensitive in server-side Teddy to match client-side Teddy so as to match HTML grammar rules.
  • Added some performance improvements to loops and includes.
  • Fixed a bug preventing <noparse> and <noteddy> tags from working in markup passed to include tags.
  • Updated various dependencies.

0.6.21

  • Fixed missing exports so you can require/import teddy less verbosely in your projects.
  • Updated docs to clarify what the different builds of Teddy are meant to be used for.
  • Removed minified cjs build since there is no use case for it.
  • Updated various dependencies.

0.6.20

  • Fixed double-encoding HTML entity bug in client-side mode.
  • Updated various dependencies.

0.6.19

  • Made client-side Teddy parser a bit more tolerant of bad markup.

0.6.18

  • Fixed a bug which caused client-side Teddy to rename yet more form elements accidentally.

0.6.17

  • Fixed a bug which caused client-side Teddy to handle array length lookups differently from cheerio-driven Teddy.
  • Fixed a bug which caused client-side Teddy to crash if extra spaces were in markup attribute lists.
  • Fixed a bug which caused client-side Teddy to return the wrong data with 2 or more layers of object lookups if the object keys were camelCase.
  • Fixed a bug which caused client-side Teddy to rename some form elements accidentally.
  • Updated various dependencies.

0.6.16

  • Fixed a bug which caused client-side Teddy to fail in some situations like putting a <loop> in a <select> element.
  • Deprecated a test that tests for passing numeric arguments to include tags, since this violates HTML grammar and never should've worked to begin with. It may still work with cheerio-driven Teddy because cheerio's parser is more forgiving than a standards-compliant one unless and until cheerio deprecates support for that itself. Client-side Teddy will not support it, so for consistency the test has been removed.
  • Updated various dependencies.

0.6.15

  • Fixed a bug which caused the cheerio-driven modules to not work client-side if you choose to use them there.
  • Updated various dependencies.

0.6.14

  • Finished work on cheerioPolyfill.js which makes it possible for Teddy to execute in client-side contexts without using cheerio, allowing for a very small bundle size for client-side Teddy (17kb minified).
  • Fixed client-side tests to test Teddy in the browser properly.
  • Refactored tests to improve maintainability.
  • Updated various dependencies.

0.6.13

  • Removed xregexp and html-entities dependencies. Also Replaced cheerio with cheerio/slim. These changes have reduced bundle size significantly.
  • Began work on cheerioPolyfill.js which will allow Teddy to execute in client-side contexts without using cheerio. This will result in even smaller bundle sizes (currently around 16kb minified), however the work is unfinished.
  • Updated various dependencies.

0.6.12

  • Downgraded cheerio to prevent webpack errors when using Teddy on frontend.
  • Updated various dependencies.

0.6.11

  • Added new setting teddy.setEmptyVarBehavior('hide') that will make it possible for variables which don't resolve to display as empty strings instead of displaying the variable.

0.6.10

  • Added support for template literal ${templateLiteral} variables.
  • Fixed bug that caused Teddy to crash if you attempted to loop through a Set.
  • Replaced internal HTML entities scanner with the html-entities npm package.
  • Updated various dependencies.

0.6.9

  • Fixed issue with rendering variables with empty strings piped through variables with flags.
  • Updated various dependencies.

0.6.8

  • Fixed issue with rendering special characters correctly when piped through a teddy noparse flagged variable.
  • Updated various dependencies.

0.6.7

  • Fixed issue causing server-side comments to not be stripped from templates parsed as strings instead of as files.
  • Updated various dependencies.

0.6.6

  • Fixed issue causing the template caching feature to not work properly in Express apps.
  • Updated various dependencies.

0.6.5

  • Fixed scenario where one-line if with only a false condition could crash if it's fed by arguments from an include element.
  • Fixed crash associated with embedding JSON strings in script tags.
  • Fixed bug causing "{}" to be inappropriately stripped from templates.
  • Reduced unminified bundle size from 2.4mb to 1.5mb and added a minified bundle that reduces it further to 449kb.
  • Improved logging and refactored the code a bit.
  • Updated various dependencies.

0.6.4

  • Fixed crash that could occur in an unusual edge case.
  • Various dependencies bumped.

0.6.3

  • Restored template-level caching feature with an improved API.
  • Restored compile method but now with an API similar to other templating systems. It will now take a template string and return a function which when given model data will render HTML from the template and model data.
  • Now hosting the Teddy playground on https://rooseveltframework.github.io/teddy/playground.html. Thanks to jsDelivr for CDN hosting.
  • Various dependencies bumped.

0.6.2

  • Significant performance improvements.
  • Added a new <cache> element that can be used to improve template parsing performance further.
  • Various dependencies bumped.

0.6.1

  • Fixed memory leak related to the no parse block feature.
  • Various dependencies bumped.

0.6.0

  • Breaking: <include> elements will no longer automatically disable escaping content of <arg> variables. Use {var|s} to disable escaping yourself.
  • Breaking: One-line if statements like if-something='' will now evaluate as true if something is present in the model.
  • Breaking: You can no longer apply more than one one-line if statement to the same element.
  • Breaking: Boolean logic will no longer allow you to query the same variable twice in the same if statement or use the same boolean logic operator twice in the same if statement.
  • Breaking: You can no longer use Teddy tags in elements that interpret child text as plain text, e.g. <style>, <script>, <textarea>, etc.
  • Breaking: Removed the noparse and noteddy attribute feature from <include> tags.
  • Breaking: Removed Teddy's internal minifier since third party HTML minifiers are better.
  • Breaking: Removed caching feature. A better-designed one may arrive in the future.
  • Breaking: Server-side comments will now always be removed, even if they are contained in a no-parse block. To preserve them for display, you must manually escape them with HTML entities.
  • Breaking: teddy.compile was removed as a public method. To set a new template, you must now call teddy.setTemplate to create it. Accordingly the teddy.compileAtEveryRender method has been removed as well.
  • Breaking: The |p flag will now also apply the same logic as |s as well. If you have any {variables} written like {variable|s|p}, you can now optionally write it as {variable|p} instead to get identical functionality.
  • Breaking: The in attribute as a synonym for through in <loop> has been removed.
  • Significant stability improvements; lots of bugs fixed.
  • Some performance improvements, some performance downgrades (notably looping through large datasets is slightly slower now).
  • Simplified API.
  • maxPasses default changed from 25000 to 1000.
  • Various dependencies bumped.

0.5.10

0.5.9

  • Fixed an issue where model content could get printed without being escaped even when the |s flag is not present.
  • Various dependencies bumped.

0.5.8

  • Fixed bug in client-side support where file extensions would be inappropriately appended to template names in some cases.
  • Dropped Node 15 support. Added Node 16 support.
  • Various dependencies bumped.

0.5.7

0.5.6

0.5.5

0.5.4

0.5.3

  • Fixed some Windows-exclusive bugs related to how newlines are parsed in Windows vs. other operating systems.
  • Various dependencies bumped.
  • CI improvements.

0.5.2

  • Undo accidental removal of package-lock.json.

0.5.1

  • Removed postinstall script since it created a problematic prompt to install webpack-cli during npm installs.

0.5.0

  • Variables with spaces in them will now be parsed.
  • Fixed issue where one line if statements couldn't use variables as part of the condition.
  • Fixed issue where recursive variable resolution could cause an infinite loop.
  • Significant performance improvements.
  • Total rewrite into a much cleaner codebase:
    • Less reliance on regex and more reliance on character counting in base algorithm.
    • Code now split into separate files for development but bundled into a single JS file with Webpack during deployment.
  • Various dependencies bumped.
  • CI improvements.

0.4.28 and below

Here be dragons…