needs to implement value-class-pattern value-title support in dt-* parsing step 1

[tantek]

Per http://microformats.org/wiki/microformats2-parsing#parsing_a_dt-_property the first step is:

• parse the element for the value-class-pattern including the date and time parsing rules, if a value is found then return it.

Value-class-pattern support specifically support for value-title:

• http://microformats.org/wiki/value-class-pattern#Parsing_value_from_a_title_attribute

appears to be missing per:

• http://pin13.net/mf2/?url=https%3A%2F%2Fraw.githubusercontent.com%2Fwhatwg%2Furl%2Fmaster%2Furl.html

which returns an "updated" property of:

• "updated": ["T"]

whereas it should be:

• "updated": ["20141211"]

(or possibly a newer date)
from the title attribute title=20141211 in this markup in the source:

• <span class=dt-updated><span class=value-title title=20141211>11 December 2014</span></span>

[gRegorLove]

Reviewing the spec while working on this, I saw:

Parsing rules for value-title are the same as for value above, with the following change:

• Where a microformats property has a child element with class name of value-title, the content of the title attribute of that element must be parsed, rather than the portion of the element that would be parsed for a class name of value.

When I read further up on the page, YYYYMMDD does not appear to be a valid format for the parser (emphasis mine):

• YYYY-MM-DD
• YYYY-DDD
• then use that as the date value. For the purposes of the value class pattern, the hyphens "-" separating the year, month, day and/or ordinal day are required.
• ignore any further "value" elements that specify the date.

php-mf2 is finding the YYYYMMDD value correctly, but then it doesn't match those formats. I realize "T" isn't the correct, but I'm not sure what it should be. Blank? No 'updated' property? Or should the spec be udpated to allow YYYYMMDD?

Edited to add: the version of url.html referenced.

[tantek]

Note that the master (latest) copy of url.html is using YYYY-MM-DD so this has been fixed in the source document: https://github.com/whatwg/url/blob/master/url.html so we can ignore YYYYMMDD for now.

[gRegorLove]

Confirmed this is working properly when the date is in a correct format: http://pin13.net/mf2/?id=20170528051422204

Edit: I'm working on #115, which will resolve the "T" in the original example here.

[aaronpk]

I believe this was fixed in #131, but the original HTML in the issue is gone so I don't know the exact input it was failing on.

[gRegorLove]

@aaronpk Let's leave this open for now.

I read #60 (comment) as "authoring issue, ignore." However, I just tested the original HTML with #131 and it parses a space character for updated. I'm not sure if it should be an empty string or if updated property shouldn't appear at all.

Original HTML:
<div class="h-entry"><span class=dt-updated><span class=value-title title=20141204>4 December 2014</span></div>

Parsed using #131:

{
    "items": [
        {
            "type": [
                "h-entry"
            ],
            "properties": {
                "updated": [
                    " "
                ],
                "name": [
                    "4 December 2014"
                ]
            }
        }
    ],
    "rels": [],
    "rel-urls": []
}

Parsed with latest php-mf2 release:

{
    "items": [
        {
            "type": [
                "h-entry"
            ],
            "properties": {
                "updated": [
                    "T"
                ],
                "name": [
                    "4 December 2014"
                ]
            }
        }
    ],
    "rels": {},
    "debug": {
        "package": "https://packagist.org/packages/mf2/mf2",
        "version": "v0.3.2",
        "note": [
            "This output was generated from the php-mf2 library available at https://github.com/indieweb/php-mf2",
            "Please file any issues with the parser at https://github.com/indieweb/php-mf2/issues"
        ]
    }
}

[gRegorLove]

I did some comparisons with other parsers. For YYYYMMDD,

• Node parses it (beyond spec; doesn't seem to cause problems)
• Python returns an empty array for the property
• Ruby parses it (beyond spec, but also adds time precision: "2014-12-04 00:00:00")
• Go parses the node's textContent: "4 December 2014"

When the VCP spec says:

If by parsing the "value" element(s), at least a specific date has been found, then the "value" is overall valid...

What do we do when the "value" is not overall valid? Should a blank value be returned, or should the property not appear at all?

[gRegorLove]

Checking with current php-mf2 and I think this is resolved.

With invalid YYYYMMDD format:

<div class="h-entry"><span class=dt-updated><span class=value-title title=20141204>4 December 2014</span></div>

Parser no longer returns updated as a space or empty string (correct behavior, IMO):

{
    "items": [
        {
            "type": [
                "h-entry"
            ],
            "properties": {
                "name": [
                    "4 December 2014"
                ]
            }
        }
    ]
}

With correct YYYY-MM-DD:

<div class="h-entry"><span class=dt-updated><span class=value-title title=2014-12-04>4 December 2014</span></div>

Parser returns updated:

{
    "items": [
        {
            "type": [
                "h-entry"
            ],
            "properties": {
                "updated": [
                    "2014-12-04"
                ],
                "name": [
                    "4 December 2014"
                ]
            }
        }
    ]
}