Skip to content

Commit

Permalink
reduce redundancy in host/IPv6 parsing
Browse files Browse the repository at this point in the history
  • Loading branch information
@rubys
rubys committed Nov 29, 2014
1 parent 055d995 commit 3967bb2
Show file tree
Hide file tree
Showing 3 changed files with 56 additions and 437 deletions.
258 changes: 34 additions & 224 deletions url.bs
View file
Expand Up @@ -341,9 +341,8 @@ Representation of IPv4 and IPv6 Addresses</a> for some history.

<h3 id=host-parsing>Host parsing</h3>

<p>The <dfn id=concept-host-parser title='host parser'>host parser</dfn> takes a string
<var>input</var> and optionally a <var>Unicode flag</var>, and then runs
these steps:
<p>The <dfn id=concept-host-parser title='host parser'>host parser</dfn> takes
a string <var>input</var> and then runs these steps:

<ol>
<li><p>If <var>input</var> is the empty string, return failure.
Expand All @@ -353,20 +352,6 @@ these steps:

Could potentially move this check to the URL parser if deemed problematic. -->

<li>
<p>If <var>input</var> starts with "<code>[</code>", run these
substeps:

<ol>
<li><p>If <var>input</var> does not end with
"<code>]</code>", <a>parse exception</a>, return failure.

<li><p>Return the result of
<a title='IPv6 parser'>IPv6 parsing</a> <var>input</var>
with its leading "<code>[</code>" and trailing
"<code>]</code>" removed.
</ol>

<li><p>Let <var>domain</var> be the result of
<a>utf-8 decode without BOM</a> on the
<a title="percent decode">percent decoding</a> of
Expand Down Expand Up @@ -397,186 +382,9 @@ these steps:
"<code>]</code>",<!-- 5D -->
return failure.

<li><p>Return <var>asciiDomain</var> if the <var>Unicode flag</var> is unset,
and the result of running <a>domain to Unicode</a>
on <var>asciiDomain</var> otherwise.
</ol>

<p>The <dfn id=concept-ipv6-parser title='IPv6 parser'>IPv6 parser</dfn> takes a string
<var>input</var> and then runs these steps:

<ol>
<li><p>Let <var>address</var> be a new
<a title='IPv6'>IPv6 address</a> with its
<a title='IPv6 piece'>16-bit pieces</a> initialized to 0.

<li><p>Let <var>piece pointer</var> be a pointer into
<var>address</var>'s
<a title='IPv6 piece'>16-bit pieces</a>, initially zero
(pointing to the first <a title='IPv6 piece'>16-bit piece</a>),
and let <var>piece</var> be the
<a title='IPv6 piece'>16-bit piece</a> it points to.

<li><p>Let <var>compress pointer</var> be another pointer into
<var>address</var>'s <a title='IPv6 piece'>16-bit pieces</a>, initially
null and pointing to nothing.

<li><p>Let <var>pointer</var> be a pointer into
<var>input</var>, initially zero (pointing to the first code point).

<li>
<p>If <a>c</a> is "<code>:</code>", run these substeps:

<ol>
<li><p>If <a>remaining</a> does not start with
"<code>:</code>", <a>parse exception</a>, return failure.

<li><p>Increase <var>pointer</var> by two.

<li><p>Increase <var>piece pointer</var> by one and then set
<var>compress pointer</var> to <var>piece pointer</var>.
</ol>

<li>
<p><dfn id=concept-ipv6-parser-main title='IPv6 parser Main'>Main</dfn>:
While <a>c</a> is not the <a>EOF code point</a>, run these
substeps:

<ol>
<li><p>If <var>piece pointer</var> is eight,
<a>parse exception</a>, return failure.

<li>
<p>If <a>c</a> is "<code>:</code>", run these inner
substeps:

<ol>
<li><p>If <var>compress pointer</var> is not null,
<a>parse exception</a>, return failure.

<li>Increase <var>pointer</var> and <var>piece pointer</var> by one, set
<var>compress pointer</var> to <var>piece pointer</var>,
and then jump to <a title='IPv6 parser Main'>Main</a>.
</ol>

<li><p>Let <var>value</var> and <var>length</var> be 0.

<li><p>While <var>length</var> is less than 4 and
<a>c</a> is an
<a title="ASCII hex digits">ASCII hex digit</a>, set
<var>value</var> to
<var>value</var> &times; 0x10 + <a>c</a> interpreted as hexadecimal number,
and increase <var>pointer</var> and <var>length</var> by one.

<li>
<p>Based on <a>c</a>:

<dl class=switch>
<dt>"<code>.</code>"
<dd>
<p>If <var>length</var> is 0, <a>parse exception</a>,
return failure.
<p>Decrease <var>pointer</var> by <var>length</var>.
<p>Jump to <a title='IPv6 parser IPv4'>IPv4</a>.

<dt>"<code>:</code>"
<dd>
<p>Increase <var>pointer</var> by one.
<p>If <a>c</a> is the <a>EOF code point</a>,
<a>parse exception</a>, return failure.

<dt>Anything but the <a>EOF code point</a>
<dd><p><a>Parse exception</a>, return failure.
</dl>

<li><p>Set <var>piece</var> to <var>value</var>.

<li><p>Increase <var>piece pointer</var> by one.
</ol>

<li><p>If <a>c</a> is the <a>EOF code point</a>, jump to
<a title='IPv6 parser Finale'>Finale</a>.

<li><p><dfn id=concept-ipv6-parser-ipv4 title='IPv6 parser IPv4'>IPv4</dfn>:
If <var>piece pointer</var> is greater than six,
<a>parse exception</a>, return failure.

<li><p>Let <var>dots seen</var> be 0.

<li>
<p>While <a>c</a> is not the <a>EOF code point</a>, run
these substeps:

<ol>
<li><p>Let <var>value</var> be null.

<li><p>If <a>c</a> is not an <a title="ASCII digits">ASCII digit</a>,
<a>parse exception</a>, return failure. <!-- prevent the empty string -->

<li>
<p>While <a>c</a> is an
<a title="ASCII digits">ASCII digit</a>, run these subsubsteps:

<ol>
<li><p>Let <var>number</var> be <a>c</a> interpreted as decimal number.

<li>
<p>If <var>value</var> is null, set <var>value</var> to <var>number</var>.

<p>Otherwise, if <var>value</var> is 0, <a>parse exception</a>, return failure.

<p>Otherwise, set <var>value</var> to <var>value</var> &times; 10 + <var>number</var>.

<li><p>Increase <var>pointer</var> by one.

<li><p>If <var>value</var> is greater than 255, <a>parse exception</a>,
return failure.
</ol>

<li><p>If <var>dots seen</var> is less than 3 and
<a>c</a> is not a "<code>.</code>",
<a>parse exception</a>, return failure.

<li><p>Set <var>piece</var> to
<var>piece</var> &times; 0x100 + <var>value</var>.

<li><p>If <var>dots seen</var> is 1 or 3, increase
<var>piece pointer</var> by one.

<li><p>Increase <var>pointer</var> by one.

<li><p>If <var>dots seen</var> is 3 and <a>c</a> is not
the <a>EOF code point</a>,
<a>parse exception</a>, return failure.

<li><p>Increase <var>dots seen</var> by one.
</ol>

<li>
<p><dfn id=concept-ipv6-parser-finale title='IPv6 parser Finale'>Finale</dfn>:
If <var>compress pointer</var> is not null, run these substeps:

<ol>
<li><p>Let <var>swaps</var> be
<var>piece pointer</var> &minus; <var>compress pointer</var>.

<li><p>Set <var>piece pointer</var> to seven.

<li><p>While <var>piece pointer</var> is not zero and <var>swaps</var> is
greater than zero, swap <var>piece</var> with the
<a title='IPv6 piece'>piece</a> at pointer
<var>compress pointer</var> + <var>swaps</var> &minus; 1, and then
decrease both <var>piece pointer</var> and <var>swaps</var> by one.
</ol>

<li><p>Otherwise, if <var>compress pointer</var> is null and
<var>piece pointer</var> is not eight, <a>parse exception</a>,
return failure.

<li><p>Return <var>address</var>.
<li><p>Return <var>asciiDomain</var>.
</ol>


<h3 id=host-serializing>Host serializing</h3>

<p>The <dfn id=concept-host-serializer title='host serializer'>host serializer</dfn> takes null or a
Expand Down Expand Up @@ -1386,17 +1194,15 @@ Return the <a title=cleanse>cleansed</a> result using the
<pre class=railroad>
Choice:
Sequence:
T: [
N: ipv6addr
T: ]
Sequence:
N: ipv4addr
ZeroOrMore:
T: any except [:/\?#]
</pre>

If the input contains an <code class=grammar-rule><a href=#ipv6addr>ipv6addr</a></code>, return "<code>[</code>" plus
the result returned by <code class=grammar-rule><a href=#ipv6addr>ipv6addr</a></code> plus "<code>]</code>".
If the input contains an <code class=grammar-rule><a href=#ipv6addr>ipv6addr</a></code>,
the result returned by <code class=grammar-rule><a href=#ipv6addr>ipv6addr</a></code>.

If the input contains an <code class=grammar-rule><a href=#ipv4addr>ipv4addr</a></code>, return
the result returned by <code class=grammar-rule><a href=#ipv4addr>ipv4addr</a></code>.
Expand Down Expand Up @@ -1433,20 +1239,23 @@ may change the way domain names and trailing dots are handled.

<pre class=railroad>
Sequence:
Optional:
Sequence:
ZeroOrMore:
Sequence:
N: h16
T: :
T: :
ZeroOrMore:
Sequence:
T: [
Sequence:
Optional:
Sequence:
ZeroOrMore:
Sequence:
N: h16
T: :
T: :
ZeroOrMore:
Sequence:
N: h16
T: :
Choice:
N: h16
T: :
Choice:
N: h16
N: ls32
N: ls32
T: ]
</pre>

Let <code>pre</code>, <code>post</code>, and <code>last</code> be the <code class=grammar-rule><a href=#h16>h16</a></code> values before the double colon if
Expand All @@ -1472,8 +1281,8 @@ Sequence:
the sum of the lengths of the <code>pre</code> and <code>post</code> array is seven.
* Append <code>last</code> to <code>pre</code>.

Return the <a href=#concept-ipv6-serializer>ipv6
serialized</a> value of <code>pre</code> as a string.
Return '[' plus the <a href=#concept-ipv6-serializer>ipv6
serialized</a> value of <code>pre</code> as a string, plus ']'.

<p class=XXX>The resolution of
<a href="https://www.w3.org/Bugs/Public/show_bug.cgi?id=27234">bug <code>27234</code></a>
Expand Down Expand Up @@ -2507,10 +2316,13 @@ static method, when invoked, must run these steps:

<ol>
<li><p>Let <var>asciiDomain</var> be the result of
<a title='host parser'>host parsing</a> <var>domain</var>.
invoking <code class=grammar-rule><a href=#host>host</a></code> with
<var>domain</var> as input.

<li><p>If <var>asciiDomain</var> is an <a title='IPv6'>IPv6 address</a>
or failure, return the empty string.
<li><p>If <var>asciiDomain</var> matches
<code class=grammar-rule><a href=#ipv6addr>ipv6addr</a></code> or
<code class=grammar-rule><a href=#ipv4addr>ipv4addr</a></code> or
failure, return the empty string.

<li><p>Return <var>asciiDomain</var>.
</ol>
Expand All @@ -2520,14 +2332,12 @@ static method, when invoked, must run these steps:
static method, when invoked, must run these steps:

<ol>
<li><p>Let <var>unicodeDomain</var> be the result of
<a title='host parser'>host parsing</a> <var>domain</var> with the
<var>Unicode flag</var> set.

<li><p>If <var>unicodeDomain</var> is an
<a title='IPv6'>IPv6 address</a> or failure, return the empty string.
<li><p>Let <var>asciiDomain</var> be the result of invoking
<a link-for=URL method title=domainToASCII()>domainToASCII</a> with
<var>domain</var> as input.

<li><p>Return <var>unicodeDomain</var>.
<li><p>Return the result of running <a>domain to Unicode</a>
on <var>asciiDomain</var>.
</ol>

<p class=XXX>Add domainToUI() which follows the UA conventions for when to use the Unicode
Expand Down
16 changes: 8 additions & 8 deletions url.pegjs
View file
Expand Up @@ -536,8 +536,8 @@ Password
}

/*
If the input contains an @IPv6Addr, return "[" plus
the result returned by @IPv6Addr plus "]".
If the input contains an @IPv6Addr,
the result returned by @IPv6Addr.
If the input contains an @IPv4Addr, return
the result returned by @IPv4Addr.
Expand Down Expand Up @@ -571,10 +571,10 @@ Password
may change the way domain names and trailing dots are handled.
*/
Host
= '[' addr:IPv6Addr ']'
= addr:IPv6Addr
&{ return lookahead(/^([\\\/?#:]|$)/) }
{
return '[' + addr + ']'
return addr
}

/ addr:IPv4Addr
Expand Down Expand Up @@ -674,15 +674,15 @@ Host
the sum of the lengths of the $pre and $post array is seven.
* Append $last to $pre.
Return the <a href=#concept-ipv6-serializer>ipv6
serialized</a> value of $pre as a string.
Return '[' plus the <a href=#concept-ipv6-serializer>ipv6
serialized</a> value of $pre as a string, plus ']'.
<p class=XXX>The resolution of
<a href="https://www.w3.org/Bugs/Public/show_bug.cgi?id=27234">bug 27234</a>
may add support for link-local addresses.
*/
IPv6Addr
= addr:(((H16 ':')* ':')? (H16 ':')* (H16 / LS32))
= '[' addr:(((H16 ':')* ':')? (H16 ':')* (H16 / LS32)) ']'
{
var pre = [];
var post = [];
Expand Down Expand Up @@ -716,7 +716,7 @@ IPv6Addr
ipv4 = addr[2]
};
return Url.canonicalizeIpv6(pre, post, ipv4)
return '[' + Url.canonicalizeIpv6(pre, post, ipv4) + ']'
}

/*
Expand Down

0 comments on commit 3967bb2

Please sign in to comment.