x/lume@v1.9.0/CHANGELOG.md
All notable changes to this project will be documented in this file.
The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.
site.copy() now accepts an array of file extensions to copy.
For example: site.copy([".jpg", ".png"]).site.copy() accepts a function as second argument to fine tuning the output file name.
For example: site.copy("assets", (file) => "/dir" + file)
and also with extensions: site.copy([".jpeg"], (file) => file.replace(".jpeg", ".jpg"))prettyUrls option allows no-html-extension value to use the same url resolution as GitHub #193.imagick plugin has the cache option enabled by default.site.loadAssets()) won't render the layout variable.
This is something that you probably always wanted.
For example after defining a default layout in a _data.yml file,
this value was used also for assets like .css or .jpg files,
and the only way to prevent this is creating another _data.yml file inside these files' directory with layout: null.
With this change, this is no longer required, and layout is applied only for pages loaded with site.loadPages().engines.ts into renderer.ts class.std, postcss_autoprefixer parcel_css, cliffy, deno_dom and esbuild dependencies.metas to add automatically <meta> tags for SEO and social networks #188.prism to use this library as code highligher #187.search.values() to return an array of all unique values of a key #191.cache to imagick plugin to cache the transformed images in the _cache folder #184.site.ignore() now supports functions #53.site.ignore() affects to static files copied with site.copy().
This allows to copy a entire directory with site.copy("statics")
but ignoring some subdirectories with site.ignore("statics/ignored-folder").afterBuild and afterUpdate events have the staticFiles property
with info about the static files that have been copied.site.files containing all static files to be copied.
It's like site.pages but for static files.parcel_css, cliffy, react, markdown_it, liquid, std, esbuild, postcss, postcss_autoprefixer and pug.printError function to print SASS errors properly #194.import_map.json and deno.json files.serve task generation in deno.json to meet the changes introduced in Deno 1.21.2
denoland/deno/issues/14459date variable of the pages is a string or number,
it's converted to Date automatically #181.std, denosass, cliffy, imagick and esbuild.afterRender event must stop the build if any listener returns false.lume upgrade, the deno.json and import_map.json files are updated accordingly #182.--version to lume upgrade, to upgrade to a specific version.bundler plugin detects automatically the import map file from deno.json.lume init command:import lume from "lume/mod.ts".import_map.json file generation:
Only lume/ import is added. lume and https://deno.land/x/lume/ are removed.std, esbuild, deno_dom, deno_graph, parcel-css, highlight.js, terser and react.date plugin accepts an array of locales (ex: ["en", "gl", "pt", "es"]),
that are loaded automatically.lume/plugins/date/locale import from the import_map.netlify_cms plugin.expires middleware for 0 values.deno_graph and std.deno.json file.imagick.netlify_cms.expires middleware to add the Expires header to responses.cache_busting middleware to include the assets version identifier in the path.redirects middleware to configure custom redirects.mergedKeys data key to configure how some keys will be merged.lume/plugins/date/locale bare import to the import_map.lume import-map command creates a deno.json file (in addition to import_map.json)
to automatically link the import map to Deno,
meaning that you no longer need to specify the import map file in the command line
(lume -- --import-map=import_map.json) because it's automatically detected.deno.json file not only includes the importMap key with the "import_map.json"
value but also a couple of tasks (build and serve) to run Lume using Deno tasks
(deno task build and deno task serve).1.20.1.inline plugin adds the id and class attributes
from the removed <img> to the inlined <svg>.lume init. You can use a deno.json file with the importMap key.--file option to lume import-map.std, esbuild, postcss, parcel_css and liquid.lume import-map doesn't update the lume version._data reloading in watching mode.console.log().deno.json file detection.https://lumeland.github.io to https://lume.land #175.std, postcss, liquid and parcel_css to the latest version.site.loadData() is optional
and use textLoader by default.parcel_css plugin.<style>.deno_graph, liquidjs and parcel_css..DS_Store files (macOS).parcel_css plugin
(the most recent version of every browser).@import.sass.languages to codeHighlight plugin #169.std/http modules to create the local server./about/ instead of /about/index.html.std, esbuild, deno_graph, markdown_it, liquidjs, parcel_css and postcss.page.data.url after change any page.dest value.search helper converts undefined and null values.
Previously, only true and false values were converted.1.5.0 that makes that some pages doesn't have the url value.esbuild.parcel_css.Watcher in the core, with support for events.lume import-map to create/update import maps files #164.1.18.1.core/utils.ts.deno.json and import_map.json by default.Event interface to SiteEvent.searchByExtension function in core/utils.ts because it's not needed anymore.--only for lume init.page.document must return undefined if the page is not a HTML page #163.liquidjs, deno_graph and std.liquid, std, deno_dom and deno_graph.This version of Lume has a big internal code refactor but maintaining the public API,
so your sites should work without changes (unless you depend on internal undocumented features).
The only important change is the removal of metrics feature but probably you are not using it.
_components
that you can use in your template engines.
It's compatible with any engine (nunjucks, js/ts modules, jsx, liquid, pug, etc).site.loadComponents() to set up more components loaders.
For example: site.loadComponents([".jsx"], moduleLoader, jsxEngine).date plugin to change the helpers' name #150.pug plugin registers the pug filter.njk, pug, liquid...)
can access to shared data (like search, paginate etc)site.process("*", processFunction).layout values
and when include templates using the template engine syntax (ex: {% include "./template.njk" %}).site.includes() to define different directories for specific extensions.
For example: site.includes([".css"], "_styles");extensions option of some plugins allows to define different extensions
for different purposes. For example:
{ pages: [".tmpl.js"], data: [".js"], components: [".comp.js"]}.search.page() to return the first page found instead an array with all pages.1.x there are many interface (like: Site)
independent of the implementation (like: LumeSite implements Site).
This was not easy to maintain and it makes hard to explore the Lume code in VSCode
because everything is linked to the interfaces instead of the code.
So I decided to remove these generic interfaces.2022-12-05_pagename.md)
use UTC timezone for consistency with the dates defined in the front matter.metrics feature was removed. It was not very useful and I think maybe it can be implemented in better ways in the future.std, deno_dom, deno_graph, liquid, pug, liquid and postcss to the latest version.1.x you have to load .windi.css before .css to prevent conflict. This was fixed in 2.x..git directory are no longer detected by the watcher.Renderer exposes the extraData property.Engine interface has a new method renderSync.url plugin to change the helpers' name #149.sourceMap of postcss plugin accepts the same options as postcss library #147.
Note: Previously, sourceMap: true created a .map file.
Now it inlines the source map in the css file (because it's the default behavior of postcss).
Set the value to sourceMap: { inline: false } to keep the old behavior.jsx elements have the toString() function to render to html automatically.std/hash dependency and use Web Crypto API.std, postcss, postcss-nesting, deno_dom and deno_graph.on_demand to implement page rendering on demand.beforeRenderOnDemand.watcher to configure the watcher of the live reload
(the debounce interval and a list of ignored paths).lume init has the option --only to initialize only
the config file (lume init --only=config)
or VSCode (lume init --only=vscode).--. For example: lume --serve -- --import-map="import_map.json".
Note: This supersedes flags feature that is no longer available (use env variables for that).scopedUpdates to define independent scopes and avoid to rebuild the entire site on update.options argument to addEventListener:once To remove the listener after the first execution.signal To provide an AbortSignal to remove the listener at any time.1.16.1.core.ts.Renderer to render the site pages. This allows to create your own renderer to replace the default one.Emitter to emit the site pages and static files (save them in _site folder). This allows to create your own emitter to replace the default one.beforeBuild, beforeSave, beforeUpdate returns false
the build/update process is stopped.deno.land/std/io/util.ts dependency because it's not needed anymore.flags options. Use env variables to pass arbitrary data.
For example, use domain=example.com lume instead of lume -- example.com.deno_dom, cliffy, deno_graph, liquid and svgo dependencies.any instead of unknown for the Helper type for more flexibility.postcss-nesting, liquid, std, deno_graph and cliffy
to their latest version.url option for paginate helper.liquid to use the Liquid template engine.search and paginate plugins. Now you can configure the helpers names.lume/plugins.ts that exports all available plugins.search plugin to paginate and search plugins for coherence
(both are enabled by default).--quiet is more quiet on ci.ts #139.postcss, deno_graph and std.dest folder if it's inside src and doesn't start with _ or ..lume --serve --experimental or lume --watch --experimental).cli/watch.ts and cli/server.ts from Site.
These modules no longer require a Site instance to work, only the necessary options.svgo, std, cliffy, postcss and terser.base_path plugin may not be executed if the location option is changed later.resolve_urls that transform links with source files to final urls.
For example: /posts/post-1.md -> /post/post-1/.
This allows to write markdown files that are navigable in GitHub.modify_urls that allows to edit the urls of HTML pages.
It's used internally by other plugins like relative_urls, base_path or resolve_urls..jsx pages can export a JSX element as default (previously only a function was accepted).https://deno.land/x/lume/ in addition to lume/.
This allows to use the _config.js file with or without import maps.keepDefaultPlugins to postcss and markdown.
Set to true append your plugins instead override the defaults.json and module plugins.lume init can configure VS Code.server.page404 option works also with pretty urls (ex: /404/)lume init and convert it to an interactive command._data files and folders in the same directory is now merged.
Previously it was overriden, causing inconsistencies.301 response for folders without trailing slash.
For example: /about-us -> /about-us/.resolve_links plugin.entries argument of bundler plugin.
(sorry, it was a bad idea)_data files no more are converted to plain object.inline plugin.// are not normalized because they are absolute urls
For example: //domain.comstd, deno_dom.data of the pages is now enumerable.
This makes pages compatible with some filters of Nunjucks.bundler plugin:entries array items dont't need to start with /.options.bundler is undefined.includes option for bundler plugin.
They are detected and included automatically now!bundler plugin with the following additions:entries to only emit some files.includes to download and include external dependencies.Deno.EmitOptions.bundle to emit a single file with all dependencies.std, terser and svgo.react to include typings._data to save arbitrary values in the pages internally.
Used by (pre)processors.1.13.1 (from 1.12.2).std/http/server and std/ws.bundler plugin:.ts, .tsx and .jsx extensions with .js.
This fixes the import errors.std, cliffy and deno_dom.No changes.
1.12.2..js files.
Use the .ts extension to import Lume modules.install.js, ci.js and cli.js (with the .js extension)
no longer exists.
Use the .ts extension (install.ts, ci.ts and cli.ts).--plugins option of lume init.search.pages("url!=/").std, cliffy, postcss, highlight.js and markdown-it.Source.getOrCreateDirectory() a private method._data files and folders.
This opens the door to implement
DPR
in Lume.PluginOption argument typing of the lume() function.yaml loader should return an empty object instead of undefined
on empty files.std and cliffy.Engine is converted to a TypeScript interface.Site, Page, Directory, Source, Metrics, etc.core/ directory.types.ts to core.ts.highlight.js and eta.date accepts now as a value to format the current time.Metrics class,
so it’s decoupled from the CLI.site.options.cwd.--quiet mode is now even quieter._config.ts file automatically if _config.js does not exist.site.includes() method since it’s not consistent with custom includes.lume init command
provided completions for the plugins enabled by default.includes configuration for plugins is not correctly resolved
if the root is different to Deno.cwd().bundler on Windows.postcss with multiple includes.import lume from "lume"; in your config file
(instead of import lume from "lume/mod.ts";).options configuration key to the eta plugin to configure Eta..ts extension.
You should edit your config file to import .ts files instead of .js.svg to svgo.std, terser and markdown-it.-w/--watch option to watch changes without starting a web server.
#109._includes directory configurable: #115.includes option for site and the eta, nunjucks and pug pluginssite.includes() method, similar to site.src().options configuration value to the pug plugin
to set options to the pug compiler.lume completions to generate shell completions for Lume.--no-check option to lume upgrade, install.js and ci.js
to reduce the execution time.build command. Use lume instead of lume build.lume -V or lume --version
instead of lume -v (in lowercase).--verbose option with --quiet.import_map.json file because it’s not used.
It was kept only for backward compatibility of lume upgrade
from old versions.--verbose option. Use --quiet for the same behavior as --verbose=1.inline plugin.slugify_urls plugin produced empty path segments
(for example, x/@/y now becomes x/-/y and not x//y as before).lume --metrics to show the metrics in the CLI
or lume --metrics=filename.json to save the data in a file.
You can also configure it in the config file.--verbose option to configure the level of details logged:0: only important things1: normal details (the default option)2: high details (for debugging purposes)src.path value.relative_urls plugin should ignore data URLs. #107.search.pages() without arguments.slugify_urls plugin.page.search().
This allows to insert spaces in the values.
For example: page.search("'tag with spaces' title='Title with spaces'").slugify_urls plugin.attributes is disabled by default.slugifyUrls option. Use the slugify_urls plugin for that.url variable.terser dependency with a Deno version.std and postcss.afterRender and beforeSave.site.addHelper() to register different types of helpers.addFilter() method with more generic addHelper()
that allows to register not only filters but also custom tags
and other features supported by some engines, like Nunjucks.image. Moved to another repository.Page and Directory classes always have the src object.postcss and eta.limit to search.pages().--location.std.postcss to support Deno 1.11.0.deno run <script> accepts several scripts at the same time.deno run <script>.The CLI options are applied before lume() returns the site instance.
This allows to access these options in _config.js.
For example, after running lume --dev,
you can include conditions in the _config.js file like this:
const site = lume();
if (site.options.dev) {
// Development stuff
}Previously,
these overrides were applied after _config.js exports the site instance.
Source cache.search.pages() with no arguments
returns pages with other extensions than .html.nunjucks (a new fork) and highlight.js.ci.js and the command lume init on Windows.base_path to automatically search and fix all URLs in the HTML
by adding the path prefix of the location option.
This reduces the need to use the url filter everywhere.search.pages() helper allows multiple values to sort by.url variable of the pages must start with /, ./ or ../.url filter shouldn’t add the path prefix if it’s already added.search.pages() must return only HTML pages.
(Previously, it also returned assets.)import_map.json to not break upgrades from older versions.postcss and terser filters. #97.--plugins argument to lume init, so you can load and use plugins.
Example: lume init --plugins=postcss,terser,pug.document to Page that returns the parsed HTML.code_highlight to automatically highlight all code
inside pre code. Previously, it was part of the Markdown plugin.
Now it’s decoupled, so it can be used by any template engine.extensions and attribute options to the inline plugin.--dev (or -d) for lume upgrade
to upgrade to the latest development version..html files by default.loadPages and loadAssets is now optional,
and the text loader is used by default.templateEngine in a layout.code_highlight plugin.ws.js..markdown extension.
Use .md or configure the Markdown plugin to enable it.site.engine(). Use the third argument of site.loadPages().
For example: site.loadPages([".html"], textLoader, nunjucksEngine).dom.
It’s no longer needed because pages can easily return the parsed HTML.
For example: site.process([".html"], (page) => modify(page.document)).std, postcss, deno_dom, date_fns
and highlight.js (to 11.x)./bin/bash or PowerShell.exe executable.
This adds support for more features, like pipes, etc.pug.nunjucks,
when the src is in a subdirectory.install.js for easy installation.install from the CLI.pretty and slugify for the url page variable
to override the corresponding prettyUrls and slugifyUrls site options
in particular pages. #95.import_map.json to the installation process. This allows to import Lume
in the _config.js file with import lume from "lume/mod.js";.lume install to install Lume easily using the import map.ci.js to execute the CLI in a CI environment
(without installing it or defining the import map).--import-map for the lume init command
to enable or disable the import map in the _config.js file.
By default, it’s enabled.1.10.0.prettyUrls to not apply to the url page variable
if it’s a string. To generate a custom pretty URL: #95.{ path: /about-me })/about-me/)/about-me/index.html).url values to not assume that the page is HTML.
This means the .html extension won’t be added by default. #95.lume init to generate a _config.js file using the import map.
Use lume init --import-map=false to use the old URLs.ext from the url page variable,
because path now includes it. #95.lume update. It’s not needed thanks to import maps.std dependency.url page variable supports an object with path and ext properties
to fully customize the output filename. #83.relative_urls to convert all URLs to relative. #85.slugifyUrls setting: #94.lowercase: to enable/disable lowercase conversion (true by default)alphanumeric: to convert all characters to ASCII (true by default)separator: to customize the separator (- by default)replace: an object to replace some special characters.includes to customize the path (or paths)
used to resolve the @imported files in the postcss plugin.
Use includes: false to disable it.autoprefixer is included by default in the postcss plugin.paginate helper to relative: ./page-${page}.200,000*7 becomes 200-000-7 and not 2000007!2 / 3% becomes 2/3 and not -2-/-3-Who is?.txt becomes who-is.txt and not who-is-.txt.permalink variable in pages. Use url instead.njk filter didn’t work with async filters. #93.page.html.md). #83.inline plugin.no and not No)../ or ../,
so URLs like .foo, ..foo or .../foo are interpreted correctly._data/*/* files weren’t updated correctly on --serve.postcss and postcss_import dependencies.lume() function
to configure the default plugins. #45.search helper: #69.~= with *=.<, <=, > and >= operators.OR operator with pipes,
like tag1|tag2 or title=value1|value2.date("ATOM") filter uses Z instead of +00:00. #64.hljs class from the blocks of unhighlighted code. #71.2021-04-24-18-00_post.md.-o/--open to open the browser automatically
when running lume --serve. #62..html.js and .html.ts introduced in 0.17.0.
They can be confusing because not all templates must generate HTML pages,
so .tmpl.js/.tmpl.ts are more agnostic and fit all cases.postcss version to fix source maps.postcss_import version to fix the @charset at-rule.lume --serve. #55..html.js and .html.ts are processed as pages
(in order to replace to .tmpl.js and .tmpl.ts
that will be removed in the future).search. #57.permalink page variable. Use url instead. #46.permalink variable in the paginate helper that uses sprintf.
Now it accepts only the url option that must be a function.slugifyUrls to disable the slugifier introduced in 0.16.0. #44.only rebuild the site if it's needed
due to regression issues.permalink to generate it dynamically.permalink value can have a relative path
(must start with ./ or ../) that will be resolved to the directory name.url filter allows URLs starting with ~ to reference to source files
that will be automatically resolved to the final URL.lume --serve,
some ignored files weren’t correctly filtered.lume --serve to only rebuild the site if it’s needed.image to resize images automatically.renderOrder to control the rendering order of the pages.-s for --serve.--serve).date plugin accepts strings or integer arguments
(they will be converted to a Date with the new Date(value) constructor).cli.js
to set a site directly.extends were cached indefinitely. #39.--root CLI argument.search.pages("header.categories=my-category").search.pages("header.categories=my-category", "my.custom.order.field").search.data()
to return the data assigned to any page or directory.lume --run.search.tags().
Replaced with a filter like the one in search.pages().search.directory() to return a directory instance.
Use search.data().file to sort the results of search.pages(). Use url instead.inline plugin when the site is located in a subdirectory.Page class:page.data.tags
(they were previously contained by a Set in page.tags).page.fullData property. page.data contains the merged data.--) to commands:lume --upgrade to lume upgradelume --update to lume updatelume --init to lume initlume --run=<script> to lume run <script>.lume run --help.cwd:
instead of lume path/to/site, use lume --root=path/to/site.--help. #32.1.7.0.lume --run doesn’t return the proper exit code until
a bug in Deno is resolved.lume --update to update the Lume version
used by any _config.js file to the same installed globally in CLI.search helper to return the original array instead a clone,
so it can be modified outside.njk registered by the nunjucks plugin.url filter to a plugin that also creates
an additional filter htmlUrl to search and fix URLs in HTML code.date to format date and time values using the library
date-fns.search.pages() accepts any key
(in addition to date and file).attributes and classname to the new attributes plugin
that’s enabled by default.url filter with null values.search.pages() by implementing a cache system.css to postcss and renamed some of its options:map to sourceMappostcssPlugins to plugins.beforeUpdate and afterUpdate got a property files
with the names of all changed files.postcssPlugins: an array of the PostCSS plugins #26.map: set true to generate source maps.paginate helper didn’t always return the latest page.false if the script fails.--location option being unknown.TypeError on reloading the server..ts, .js) and fixed some issues.data filter is disabled by default..html extension to files with a subextension.
For example scripts.js.njk should be saved as scripts.js
instead of scripts.js.html. #13.lume --upgrade.<pre> elements in Markdown.search.previousPage() and search.nextPage().<dl>) in Markdown.404 error page.
Now it displays a list of files and subdirectories.templateEngine to configure the template engine
used for every page. #11/index.html.url filter in Markdown.url filter to handle hashes and queries
(such as #hashid and ?query=value)css plugin uses only the postcss-import and postcss-nesting plugins
because the others fail on Deno.lume --upgrade error.--upgrade command.lume.script()
along with CLI commands.--src and --dest to override the corresponding options.site.flags that saves all arguments after double dash,
like the ones in lume --serve -- flag1 flag2.dev.jspm.io with jspm.dev for dependencies.denopkg.com with cdn.jsdelivr.net for dependencies. #8.highlight.js to 10.x.lume --init.&& and &. #7.attr filter.server to configure the local server,
having properties port and page404.lume -v.lume -v.--upgrade.--version shortcut -V to -v.denjucks installation. #6.@import CSS of the css() plugin, using _includes as fallback.script() to execute scripts like a task runner.404.html in the built-in server to handle the 404 responses.node_modules directory by default.cwd is not a directory.attr filter by default.ignore() to ignore files and directories.lume --version.loadAssets() to register asset loaders._config.js file.load() to loadPages() and removed the asset parameter.data() to loadData().helper() to data().beforeUpdate and afterUpdate (instead of beforeBuild and afterBuild).import URL in the _config.js file
generated with --init.try…catch to prevent death on error.prettyUrls option. #1.beforeBuild and afterBuild.paginate().site.process().prettyUrls, which is true by default.site.beforeRender() and site.afterRender().
Use site.process() instead, which is an equivalent to afterRender.url filter with non-string values.search.pages() to sort pages alphabetically.--help and aliases -h and -V for CLI.eta to support the Eta template engine.helper() to register global helpers
that can be used in templates.path and recursive in search.pages().url filter with relative paths.version variable.pathPrefix and url with location.url filter bugs.search.folder().--dev to build in development mode.url filter.beforeRender transformers are executed only once.svg to optimize SVG files.dom to manipulate HTML using the DOM API.classname to manipulate CSS classes.attributes to manipulate HTML attributes.explorer to search.server.js to avoid reading problems.lume --version.lume --init to create a _config.js file.postcss and stylecow plugins in the new css plugin.The first version.