To make it easier to follow and understand.
Both building and rebuilding now follow a four step flow:
1. Init
2. Process
3. Assemble
4. Render
And now there are only one Build method, used for both builds and rebuilds.
Updates #2297
There are currently several Params and case related issues floating around in Hugo.
This is very confusing for users and one of the most common support questions on the forum.
And while there have been done some great leg work in Viper etc., this is of limited value since this and similar doesn't work:
`Params.myCamelCasedParam`
Hugo has control over all the template method invocations, and can take care of all the lower-casing of the map lookup keys.
But that doesn't help with direct template lookups of type `Site.Params.TWITTER_CONFIG.USER_ID`.
This commit solves that by doing some carefully crafted modifications of the templates' AST -- lowercasing the params keys.
This is low-level work, but it's not like the template API wil change -- and this is important enough to defend such "bit fiddling".
Tests are added for all the template engines: Go templates, Ace and Amber.
Fixes#2615Fixes#1129Fixes#2590
Add imageConfig function which calls image.DecodeConfig and returns the height, width and color mode of the image. (#2677)
This allows for more advanced image shortcodes and templates such as those required by AMP.
layouts/shortcodes/amp-img.html
```
{{ $src := .Get "src" }}
{{ $config := imageConfig (printf "/static/%s" $src) }}
<amp-img src="{{$src}}"
height="{{$config.Height}}"
width="{{$config.Width}}"
layout="responsive">
</amp-img>
```
This commit adds a `GitInfo` object to `Page` if `EnableGitInfo` is set.
It then also sets `Lastmod` for the given `Page` to the author date provided by Git.
The Git integrations should be fairly performant, but it adds "some time" to the build, somewhat depending on the Git history size.
If you want, you can run without during development and turn it on when deploying to the live server: `hugo --enableGitInfo`.
Fixes#2102
changes:
- in hugolib/page.go, `func permalink` and `func TargetPath`
Fixed the attempt to *replace* the extension of something
that was *already* a basename.
- in source/file.go, `func NewFile`
added check for allowed languages before translating filename
Fixes#2555
All config variables starts with low-case and uses camelCase.
If there is abbreviation at the beginning of the name, the whole
abbreviation will be written in low-case.
If there is abbreviation at the end of the name, the
whole abbreviation will be written in upper-case.
For example, rssURI.
The gain, given the "real sites benchmark" below, is obvious:
```
benchmark old ns/op new ns/op delta
BenchmarkHugo-4 14497594101 13084156335 -9.75%
benchmark old allocs new allocs delta
BenchmarkHugo-4 57404335 48282002 -15.89%
benchmark old bytes new bytes delta
BenchmarkHugo-4 9933505624 9721984424 -2.13%
```
Fixes#2495
Also refactor the rendering pages test to accept more than one page source per test run, which wasn't really needed for this issue, but may be in the future.
Closes#2586Fixes#2538
Maps in Viper, Hugo's config backing store, is now properly lower-cased not just on top level, the current situation.
While this is mostly a good thing, as you don't need to know the original casing to look up a value, it will be breaking for people doing direct lookups in the ´Site.Params` map.
We will try to find a solution to this "breakage", but the recommended method to get params values is via the `.Param` methods.
This method is now implemented on `Node`, `Page` and `Site` and is case-insensitive:
* Use `.Param "someKey" ` if you want page param with fall back to site param if not found on page.
* Use `.Site.Param "someKey"` to get a site param
See #2590
This change adds a canonical alias.html template that is used for page
redirects, and passes the page as data to the template under .Page
Fixes#2533Closes#2576
Add DocumentName (path to the file being rendered) to RenderingContext
and use that information to include the path in the error print.
See #2399Closes#2567
There were some breaking changes etc. that is too late to fix for 0.17.
Let us think this through and add proper author support for Hugo 0.18.
Fixes#2464
Revert "docs: Add documentation for author profiles"
This reverts commit b6673e5309.
Revert "Add First Class Author Support"
This reverts commit cf978c0649.
This avoids having to execute these expensive operations for sites not using these values.
This commit sums up a set of wordcounting and autosummary related performance improvements.
The effect of these kind of depends on what features your site use, but a benchmark from 4 Hugo sites in the wild shows promise:
```
benchmark old ns/op new ns/op delta
BenchmarkHugo-4 21293005843 20032857342 -5.92%
benchmark old allocs new allocs delta
BenchmarkHugo-4 65290922 65186032 -0.16%
benchmark old bytes new bytes delta
BenchmarkHugo-4 9771213416 9681866464 -0.91%
```
Closes#2378
It is obviously more efficient when we do not care about the actual words.
```
BenchmarkTotalWords-4 100000 18795 ns/op 0 B/op 0 allocs/op
BenchmarkTotalWordsOld-4 30000 46751 ns/op 6400 B/op 1 allocs/op
```
For people using autogenerated summaries, this is one of the hot spots in the memory department.
We don't need to split al the content into words to do proper summary truncation.
This is obviously more effective:
```
BenchmarkTestTruncateWordsToWholeSentence-4 300000 4720 ns/op 0 B/op 0 allocs/op
BenchmarkTestTruncateWordsToWholeSentenceOld-4 100000 17699 ns/op 3072 B/op 3 allocs/op
```
Hugo 0.16 announced support for symbolic links for the root folders, /content, /static etc., but this got broken pretty fast.
The main problem this commit tries to solve is the matching of file change events to "what changed".
An example:
ContentDir: /mysites/site/content where /mysites/site/content is a symlink to /mycontent
/mycontent:
/mypost1.md
/post/mypost2.md
* A change to mypost1.md (on OS X) will trigger a file change event with name "/mycontent/mypost1.md"
* A change to mypost2.md gives event with name "/mysites/site/content/mypost2.md"
The first change will not trigger a correct update of Hugo before this commit. This commit fixes this by doing a two-step check:
1. Check if "/mysites/site/content/mypost2.md" is within /mysites/site/content
2. Check if "/mysites/site/content/mypost2.md" is within the real path that /mysites/site/content points to
Fixes#2265Closes#2273
Atomic operations with 64 bit values must be aligned for 64-bit on x86-32.
According to the spec:
"The first word in a global variable or in an allocated struct or slice can be relied upon to be 64-bit aligned."
The above wasn't enough for the `paginationPageCount` on `SiteInfo`, maybe due to how `SiteInfo` is embedded.
This commit adds a 4 byte padding before the `uint64` that creates the correct alignment.
Fixes#2415
This is the nth attempt to fix an issue by changing the placeholder token pattern, but
now we actually have tests for all the historic trouble cases.
Fixes#2223
So the Permalink gets correct when listing translations.
I have also checked the other relevant places to make sure we do not overwrite node values we need later.
Pointers can be tricky, but lesson learned is: A copy is cheap.
Updates #2309
In a multi-language setup, before this commit the Node's Translations() method
would return some "dummy nodes" that would point to the correct page (Permalink),
but would not be the same as the node it points to -- it would not have the translated
title etc.
The node creation is, however, so mingled with rendering, whihc is too early to have any global state,
so the nodes has to be split in a prepare and a render phase. This commits does that with as small
a change as possible. This implementation is a temp solution until we fix#2297.
Updates #2309
The current "rendering language" is needed outside of Site. This commit moves the Language type to the helpers package, and then used to get correct correct language configuration in the markdownify template func.
This commit also adds two new template funcs: relLangURL and absLangURL.
See #2309
This is needed to verify #2309, but it closes a big hole in Hugo's automated tests.
The loading of the config is now moved to `hugolib` and the same default settings are now used
in production and tests.
As Viper now uses Afero as its filesystem, we now can write fairly complete integration tests with ease.
See #2309
And in the same go adjusted some minor parts of the language API:
Add LanguagePrefix alias to Node and rename the Multilingual config section to Languages.
See #2309
This is needed to make shortcode users happy with the new multilanguage support,
but it will also solve many other related posts about "stuff not available in the shortcode".
We will have to revisit this re the handler chain at some point, but that will be easier
now as the integration test story has improved so much.
As part of this commit, the site-building tests in page_test.go is refreshed, they now
tests for all the rendering engines (when available), and all of them now uses the
same code-path as used in production.
Fixes#1229Fixes#2323
Fixes ##1076
Work In Progress!
This commit makes a rework of the build and rebuild process to better suit a multi-site setup.
This also includes a complete overhaul of the site tests. Previous these were a messy mix that
were testing just small parts of the build chain, some of it testing code-paths not even used in
"real life". Now all tests that depends on a built site follows the same and real production code path.
See #2309Closes#2211Closes#477Closes#1744
This commit also consolidates URLs on Node vs Page, so now .Permalink should be interoperable.
Note that this implementations should be fairly short-livded, waiting for #2297, but the API should be stable.
Setting the language to use when loading the language bundles just doesn't work.
The template system is unfortanetely a global, and the last languate processed won ...
And a Hugo global variable which contains the site under build.
This is really needed to get some level of control of the "multiple languages" in play.
There are still work related to this scattered around, but that will come.
With this commit, the multilingual feature is starting to work.
Implements:
* support to render:
* content/post/whatever.en.md to /en/2015/12/22/whatever/index.html
* content/post/whatever.fr.md to /fr/2015/12/22/whatever/index.html
* gets enabled when `Multilingual:` is specified in config.
* support having language switchers in templates, that know
where the translated page is (with .Page.Translations)
(when you're on /en/about/, you can have a "Francais" link pointing to
/fr/a-propos/)
* all translations are in the `.Page.Translations` map, including the current one.
* easily tweak themes to support Multilingual mode
* renders in a single swift, no need for two config files.
Adds a couple of variables useful for multilingual sites
Adds documentation (content/multilingual.md)
Added language prefixing for all URL generation/permalinking see in the
code base.
Implements i18n. Leverages the great github.com/nicksnyder/go-i18n lib.. thanks Nick.
* Adds "i18n" and "T" template functions..
We have to figure out another way. There are perfectly valid reasons not having a layout for a page (I have have some broken sites as result of this).
See #1313
This reverts commit b15934008f.
This issue was introduced as a fix to shortcode not working in RST.
One could argue that Blackfriday and friends should handle `#` in titles, but that will be a discussion
for another day.
The new placeholder pattern should be RST safe and work with titles.
And now with a test so this doesn't break again.
Fixes#2192Fixes#2209Closes#2210
Useful if using or sharing files with users that use editors that
append a unicode byte order marker header (like Windows notepad).
This will still assume files are UTF-8 encoded.
Closes#2075
If Page.Markup was not set by the user, it will now be set after
guessing from the file extension. This means, Page.Markup will be set in
any case. It can be used by a theme to differentiate between markup
types.
Fixes#1950
Meny people, including me, have a custom robots.txt in static.
Also remove that option from the command line; it doesn't feel
important enough.
Fixes ##2049
This also includes a refactor of the hugofs package and its usage.
The motivation for that is:
The Afero filesystems are brilliant. Hugo's way of adding a dozen of global variables for the different filesystems was a mistake. In readFile (and also in some other places in Hugo today) we need a way to restrict the access inside the working dir. We could use ioutil.ReadFile and implement the path checking, checking the base path and the dots ("..") etc. But it is obviously better to use an Afero BasePathFs combined witha ReadOnlyFs. We could create a use-once-filesystem and handle the initialization ourselves, but since this is also useful to others and the initialization depends on some other global state (which would mean to create a new file system on every invocation), we might as well do it properly and encapsulate the predefined set of filesystems. This change also leads the way, if needed, to encapsulate the file systems in a struct, making it possible to have several file system sets in action at once (parallel multilanguage site building? With Moore's law and all...)
Fixes#1551
It is a fairly costly operation:
```
benchmark old ns/op new ns/op delta
BenchmarkScratchGet-4 109 31.6 -71.01%
benchmark old allocs new allocs delta
BenchmarkScratchGet-4 0 0 +0.00%
benchmark old bytes new bytes delta
BenchmarkScratchGet-4 0 0 +0.00%
´´´
This uses the Emoji map from https://github.com/kyokomi/emoji -- but with a custom replacement implementation.
The built-in are fine for most use cases, but in Hugo we do care about pure speed.
The benchmarks below are skewed in Hugo's direction as the source and result is a byte slice,
Kyokomi's implementation works best with strings.
Curious: The easy-to-use `strings.Replacer` is also plenty fast.
```
BenchmarkEmojiKyokomiFprint-4 20000 86038 ns/op 33960 B/op 117 allocs/op
BenchmarkEmojiKyokomiSprint-4 20000 83252 ns/op 38232 B/op 122 allocs/op
BenchmarkEmojiStringsReplacer-4 100000 21092 ns/op 17248 B/op 25 allocs/op
BenchmarkHugoEmoji-4 500000 5728 ns/op 624 B/op 13 allocs/op
```
Fixes#1891
We can of course skip reading the entire byte slice again and again.
This was a slip in the original implementation; functionally the same,
but is slightly faster, esp. for larger data sets with many shortcodes:
```
benchmark old ns/op new ns/op delta
BenchmarkReplaceShortcodeTokens-4 15505 14753 -4.85%
benchmark old allocs new allocs delta
BenchmarkReplaceShortcodeTokens-4 1 1 +0.00%
benchmark old bytes new bytes delta
BenchmarkReplaceShortcodeTokens-4 3072 3072 +0.00%
```
