| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The variable which captures the body part of the tag (material
after attributes) is now a pattern. Thus the body can be
destructured, allowing the custom to effectively have
non-attribute parameters.
* README.md: Documented.
* specials (*cur-form*, *cur-env*): New special variables.
* test/simple.tl: Test case for new deftag feature.
* who.tl (with-html-output): Capture macro invocation's form
and environment, binding them to *cur-form* and *cur-env*.
This is a way of getting this information down to the custom
tag expander, where destructuring errors can occur now.
We would like them diagnosed against the right form.
The body destructuring pattern can use :env too.
(deftag): tag-body parameter renamed to body-pattern. This
is not inserted into the lambda parameter list any more;
a gensym is inserted into that position, and then that value
is destructured using body-pattern.
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The deftag macro targets the HTML markup syntax itself
rather than Lisp forms within it. deftag macros
masquerade as tag markup, but rewrite themselves into
other markup. They can produce multiple elements
using (progn ...) and can move element attribute
material into the avlue position and vice versa.
They can destructure attributes using keyword parameters,
and easily default the values of required attributes
and such.
* packages.tl (tl-who): New symbol, deftag.
* specials.tl (*tag-macro*): New special variable,
holding a hash table of deftag definitions.
* who.tl (process-tag): After a tag is parsed,
we check whether there is a macro defined for it.
If so we call its expander lambda. This is done
here because this function has the parsed pieces
of the tag. The higher level above this function
doesn't, and after this function, everything
is just a flat list of strings and other objects.
Here we recognize whether the macro put out a
progn shape, and iterate over the multiple items
it contains.
(scrub-kw-args): New function. This is a helper
function used by deftag expanders to remove, from
the attribute rest parameter, those keyword
arguments which were captured by the named parameters.
(deftag): New macro.
* test/simple.tl: New test cases 50-54 targeting deftag.
* README.md: Documented.
|
| |
|
|
|
|
|
|
|
|
|
|
| |
* who.tl (tree-to-template): Recognize string elements
and html-encode them at macro time before adding
to list. Recognize (noesc ...) syntax to defeat this.
Error if it isn't given one argument that is a constant.
* test/simple.tl: New test cases 46 to 49.
* README.md: Document new differences between
CL-WHO and TL-WHO.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
* who.tl (attr-warning-macrolet): Do not warn when
fmt and escj are used on attributes. Instead, give
them a useful definition so they do what the
programmer wants: return a string.
* test/simple.tl: Trim it
to just the tags that should warn now. Also
fix a bug that rendered this case invalid:
unquote occurred with no backquote, and this
produced an undefined variable warning on
sym, which was treated by the test case's handler
as a false positive that made it pass.
New test case 45 testing fmt and escj in
attributes.
* README.md: Added to notes on differences between
TL-WHO and CL-WHO.
|
| |
|
|
|
|
|
|
|
| |
* who.tl (with-html-output): Arrange for
tostringp to be called on argument of esc and escq.
* test/simple.tl: New test 44.
* README.tl: mention as a difference from CL-WHO.
|
| |
|
|
|
|
|
|
|
|
|
| |
* CHANGELOG: File removed. Changelogs are not useful;
they can be reconstructed from version control comments.
I will create a RELNOTES file for release notes.
* cl-who.asd: Removed; no ASDF in TXR Lisp.
* test/packages.lisp, test/tests.lisp: Files removed;
the test/simple.tl is self-contained.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The escq macro provides HTML escaping for the apostrophe
and double quote also, whereas esqj is for escaping material
to be put into a Javascript literal.
* packages.tl (tl-who): New symbols escq and escj.
* who.tl (attr-warning-macrolet): Add escq and escq to
list of WHO local macros that are not warned about if
they occur in attributes.
(with-html-output): Add escq and escj.
* test/simple.tl: Test 41 updated to include escq and
escj. New tests 42 and 43.
* README.md: Mention escq and escj.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The with-html-output ("WHO") macro defines some local macrolets.
Those are in scope of the entire thing, but they can only be
meaningfully used in the imperative Lisp forms that are placed
into the HTML tag body. If they are used in an attribute
expression, it's probably a programming error; an output
side effect doesn't belong there. We implement a diagnostic
strategy which produces a warning in this situation.
* who.tl (attr-warning-macrolet): New function, which
wraps macrolet code around the expression passed in.
These macrolets intercept the local macros and generate
a warning. Then they decline to expand, deferring to the
real macros.
(convert-attributes): Wrap the warning macrolets around
the run-time evaluation of the val expression. We don't
do anything for the compile-time evaluation of a constant
attribute expression, because the macro calls we want
to intercept do not appear in a constant expression.
* test/simple.tl: New test 41 validating that warnings
are generated for all the local macros when they are called
from an attr expression.
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
* packages.tl (tl-who): Add noesc-fmt symbol.
* who.tl (with-html-output): Add escaping to fmt,
which is disabled under *cl-who-compat*.
Add noesc-fmt macro that looks like former definition
of fmt.
* test/simple.tl: New test cases 38-40.
* README.md: Mention that TL-WHO provides noesc-fmt.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
CL-WHO tries to handle the case when attribute values are
constant NIL and T values, but it bungles it; it applies
the correct behavior only when the constants are literally
these symbols, not when they are constant expressions
which evaluate to these values.
Secondly, CL-WHO neglects to HTML-escape attribute values.
We fix this behavior and introduce a noesc operator to
selectively revert it, as well as a *cl-who-compat* special
to revert the behavior more pervasively, for the daredevils.
* packages.tl (tl-who): New symbols *cl-who-compat* and noesc.
* specials.tl (*cl-who-compat*): New special variable.
* who.tl (convert-attributes): When treating a constant
expression, evaluate it first, then check for nil or t.
Escape interpolated text with html-encode.
Check the original expression for (noesc ...) pattern,
or the presence of *cl-who-compat*. In these cases, don't
generate the escape call.
* test/simple.tl: New tests 28 to 37 providing some coverage
to all these changes.
* README.md: Document.
|
| |
|
|
|
|
|
| |
* simple.tl: Translate and enable all tests.
* who.tl (process-tag): Fix double nreverse bug
causing attr list to be truncated to one tag.
|
| | |
|
| |
|
|
| |
* test/simple.tl: Port tests 7 through 8 and move (exit).
|
| |
|
|
|
|
|
|
|
|
|
| |
* who.tl (convert-attributes): In the run-time case,
we must generate code to print the calculated
string, not just to calculate it.
(convert-tag-to-string-list): Test the original tag
keyword against the *html-empty-tags*, not the
transformed string.
* test/simple.tl: Port and enable tests 1 to 7.
|
| |
|
|
|
|
|
| |
* test/simple: renamed to simple.tl. Just a self-contained
file we can run. The first test has been rewritten to TL.
After that we (exit). The first test is failing; it is
not rendering any attributes.
|
| |
|
|
|
| |
* Fix keyword bloating in DEFPACKAGE :EXPORT clauses
* Fix keyword bloating in DEFPACKAGE :SHADOW clause
|
| |
|
|
|
|
|
|
| |
Tags in this list don't have their body indented even if
*INDENT* is true.
<pre> in particular is indentation sensitive, and it's irritating
when rendering changes as :indent is flipped.
|
| |
|
|
|
|
|
| |
Previously the tests implicitly assumed that EVAL does macroexpansion
at runtime -- which is legal, but not required.
Add explicit EVAL calls to tests requiring that.
|
| |
|
|
| |
A bit ugly, but...
|
| |
|
|
|
| |
Just spent 20 minutes trying to figure out why a test that
wasn't failing was...
|
| | |
|
|
|
git-svn-id: svn://bknr.net/svn/trunk/thirdparty/cl-who@4336 4281704c-cde7-0310-8518-8e2dc76b1ff0
|
