Skip to content

Latest commit

 

History

History
187 lines (113 loc) · 11.4 KB

CHANGELOG.md

File metadata and controls

187 lines (113 loc) · 11.4 KB
Raw

LiquidScript Change Log

Version 1.8.2

Fixes

  • Fixed comparison of strings in logical expressions. Previously we only supported comparing strings for equality with == and !=, now we support <, >, <= and >= too.
  • Fixed handling of superfluous expressions in {% else %} tags. We now silently ignore anything between else and %}, matching the behavior of Shopify/Liquid.
  • Fixed handling of extra {% else %} and {% elsif %} blocks after the first {% else %} block. We now silently ignore extraneous blocks, matching the behavior of Shopify/Liquid.

Version 1.8.1

Fixes

  • Fixed the split filter. Previously, when the split filter's argument string was equal to its input string, we returned an array with empty strings. Now we return an empty array.

Version 1.8.0

Fixes

  • Fixed comparison of simple arrays with {% if %}, {% unless %} and {% case %} tags. Previously, two simple arrays with the same elements would not be considered equal.
  • Fixed the standard map filters handling of nested input arrays. We now automatically flatten nested input arrays and coerce non-iterable inputs to a single element array, just like Ruby Liquid.
  • Fixed the {% liquid %} tag's handling of some corner cases involving nested liquid tags.

Features

  • Added the standard sum filter, which returns the sum of any numeric values in its input array (docs, source).
  • Added optional {% extends %} and {% block %} tags that add template inheritance features to Liquid templates (docs, source).
  • Added filter and tag properties to the result of Template.analyze(), containing the locations of filters and tags found during static analysis (docs).

Version 1.7.0

Features

  • Allow customization of local namespace limits by overriding RenderContext.assignScore. See #5.
  • New macro and call tags. Define parameterized Liquid snippets with the macro tag and call them using the call tag. macro and call are optional extra tags that need to be explicitly registered with a liquidscript.Environment. (docs, source)
  • New drop-in replacements for the standard output statement ({{ output }}), assign tag and echo tag that support inline conditional expressions, optionally including a logical not operator and grouping terms with parentheses. (docs, source)

Compatibility

Version 1.6.0

Fixes

  • Fixed the string representation of variable paths that contain bracketed string elements with a dot. Previously foo["x.y"].bar would be represented as foo.x.y.bar. Now we retain the quoted property name foo["x.y"].bar. It is this string representation that is exposed in the results of Template.analyze(). See #9.
  • Fixed {% cycle %} tag behavior when given a cycle group name. See Python Liquid/#43.
  • Fixed the round filter when given non-integer arguments. See Shopify/liquid#1590.
  • Allow string literals to be used as a {% for %} tag iterable. See Python Liquid/#102.

Features

  • Include comment text in a template's abstract syntax tree. We used to strip comment text out, as a space saving measure, but static analysis of comment text will be required when implementing translation message extraction.

Version 1.5.1

Fixes

  • Fixed static template analysis fails with {% break %} and {% continue %}.

Version 1.5.0

Fixes

  • Fixed case/when tag expression parsing. when expressions no longer fail when presented with a string containing a comma. Handling of comma and or separated "sub-expressions" is now consistent with the reference implementation.

Compatibility

  • for tag arguments can now be separated by commas as well as whitespace. See Shopify/liquid#1658.

Version 1.4.2

Fixes

  • The truncatewords filter now trims leading and trailing whitespace from its output.
  • The built-in for and tablerow tags now accept string arguments as well as integer literals and variables that resolve to integers.

Version 1.4.1

Fixes

  • The tablerowloop drop now exposes its row property, being the current row in the table.
  • The truncatewords filter no longer throws a FilterArgumentError if its word count argument is greater than 2147483647. If the word count argument is greater than 2147483647, the input string is returned.
  • The slice filter now clamps its arguments to between Number.MIN_SAFE_INTEGER and Number.MAX_SAFE_INTEGER. A smaller range than the reference implementation, but close enough while avoiding BigInt and custom slicing.

Version 1.4.0

Features

  • Resource limits. Optionally set limits on memory usage per template render to mitigate the impact of malicious templates. (docs)

Version 1.3.1

Fixes

  • Fixed a range expression bug where the length of a range would be incorrectly calculated if both start and stop values were the same.

Version 1.3.0

Breaking Changes

  • Objects returned from ast.Node.children must now include a token property, being the token representing the start of the AST node.

Features

  • Template static analysis. Report template variable usage using Template.analyze() and Template.analyzeSync(). (docs)

Version 1.2.0-beta.1

Features

  • New general purpose argument list parser. Parse Liquid expressions containing any number of named or keyword arguments, with a choice of key/value separator. See src/expressions/arguments/parse.ts.
  • New inline comment tag {% # .. %}. See Shopify Liquid PR #1498.

Fixes

  • Fixed a bug with the assign tag where it would incorrectly throw a LiquidSyntaxError when parts of its expression were split over multiple lines.
  • Fixed a bug with the where filter. It is now consistent with the reference implementation when given nil or Undefined as its second argument.
  • Fixed a bug with the sort filter. It now raises a LiquidTypeError at render time if the items to be sorted are incompatible. The reference implementation raises an equivalent Liquid::ArgumentError in this situation.

Docs

  • New example Jekyll-style include tag.

Version 1.1.0-beta.1

  • New "if not" tag. A drop-in replacement for the standard if tag that supports logical not and grouping with parentheses.
  • Moved some module-level constants to static class variables for easier subclassing of tags.
  • Allow ObjectChain to contain push, pop and size properties. This is potentially a breaking change for early adopters writing custom tags using context.scope.push and context.scope.pop. The symbols chainPush and chainPop would be needed instead.
  • Enforce maxContextDepth when extending a render context scope as well as copying a RenderContext.
  • New FalsyStrictUndefined type. FalsyStrictUndefined is similar to StrictUndefined, but can be tested for truthiness and compared to other values in an if/unless expression without throwing an error.

Version 1.0.0-beta.1

  • Breaking Change: Both Environment.fromString() and Template.fromString() have been changed to accept a template source string (as before), render context globals as the second argument, and a TemplateContext object as an optional third argument.
  • Breaking Change: The children method of the Node interface is now optional. If implemented it should return an array of ChildNodes, whereas before it would have been an array of Nodes.
  • Breaking Change: TemplateParser.parseBlock() now accepts an optional token argument. If given, this token will be assigned to the resulting block. The current token in the stream will be used otherwise. All built-in block tags now pass their initial tag to parseBlock() instead of the tag for the first node in the block.

Version 0.5.3-alpha.1

  • Fixed a bug where the loaderContext object passed to RenderContext.getTemplate and RenderContext.getTemplateSync would not get passed on to template loaders.
  • Fixed a bug where the abstract Loader class was not passing matter, upToDate and upToDateSync on to Environment.fromString.
  • Renamed TemplateSource.uptoDate to TemplateSource.upToDate and TemplateSource.uptoDateSync to TemplateSource.upToDateSync.

Version 0.5.2-alpha.1

  • Fixed a bug with NodeFileSystemLoader and CachingNodeFileSystemLoader where template authors could escape the template search path.

Version 0.5.1-alpha.1

  • Declare DOM types for browser specific loaders.

Version 0.5.0-alpha.2

  • Don't remove browser specific loaders in Node builds.

Version 0.5.0-alpha.1

  • New RenderContext.extend() and RenderContext.extendSync() helpers for extending the current scope for the duration of a callback function.
  • New Environment.addTag() method for adding custom tags to an environment.
  • New makeTokenizer() functions for reusing built-in expression tokenizers.
  • New extra with tag. Extra tags are not not part of "standard" Liquid and are not registered automatically.

Version 0.4.0-alpha.1

  • Parse unix timestamps with the date filter.
  • Renamed expression.Filter to ExpressionFilter. This avoids conflicts with The filter implementation type.

Version 0.3.0-alpha.1

  • Fixed line numbers in error messages involving tag expressions containing newlines.
  • Improved LiquidSyntaxError messages.
    • Error messages now include any offending operators rather than the name of the token that represents the operator.
    • The template name and line number of the error is included. Previously this only happened for render-time errors.
  • More efficient parsing of liquid tag expressions. The TemplateParser class now includes a parseLiquid() method that is similar to parseBlock(), but does not needlessly check an empty set for the name of an end tag.
  • Added toLiquidSync to the drop protocol.

Version 0.2.0-alpha.1

  • New FetchLoader. A template loader for web browsers that uses the Fetch API to get templates.
  • New XMLHttpRequestLoader. A template loader for web browsers that uses XMLHttpRequest to get templates.
  • Move src/builtin/loaders/choice.ts to src/builtin/loaders/choice_loader.ts
  • Move src/builtin/loaders/file_system.ts to src/builtin/loaders/file_system_loader.ts
  • Change browser tests to use Mocha's "tdd" interface.