Update the metadata testing for nonvisual content display

[mattgarrish]

This is my best effort to capture as much as we can about nonvisual reading ability, not accounting for the various edge cases we discussed in the last editor's call.

I've changed the existing variables follows:

• I added a has_nonvisual_metadata variable that checks if any accessMode or accessModeSufficient properties are set. If not, then we don't know anything about the content and this triggers the no information available message.
• I modified all_necessary_content_textual to add a check that there is only one accessMode property and it has the value textual. This fills in the gap of accessModeSufficient not being required metadata
• I changed the variable non_textual_content_images to non_textual_content and updated it to check for any accessMode that is not textual so we capture audio and video in addition to images.
• I added a check for transcripts to textual_alternatives variable since these could be used for audio or video with images of text.

With those changes, the only major modification I had to make to the instructions is to put the "if" check that has_nonvisual_metadata is not true first. This stops us from commenting on any publications for which the publisher hasn't said anything about the access modes and lets the final else capture the ambiguous cases where they have.

The check on all_necessary_content_textual will now better capture all books we know have full text alternatives.

The check on non_textual_content without textual_alternatives will capture the case that some non-textual content is not going to be accessible. (There are ways this can fail, of course, but it relies on the publisher having only provided partial metadata.)

And finally is the catch-all case where we just don't know. There's non-textual content with some kinds of text alternatives, but for whatever reason the publisher isn't claiming the text is fully readable as text.

But if you spot anything that doesn't seem right, let me know.

Note that I added the new ID ways-of-reading-nonvisual-no-metadata for the new "No information about nonvisual reading is available" string. I didn't modify the guidelines to add this.

---

Preview | Diff

[mattgarrish]

Hm, something's gone really wrong with this pull request. I think I may have undone some of my work and reverted to before I opened the new branch. I'll see if I can fix it, but I might have to open a new pull request. For now, ignore all the hundreds of changes it's flagging in the diff after the nonvisual section.

All fixed now.

[clapierre]

Not sure what this line is doing.

[mattgarrish]

It checks that there is at least one meta tag that has a property of accessMode or accessModeSufficient so we don't attempt to process a value when they haven't stated anything about the modes. The value doesn't matter.

[clapierre]

but if accessMode="visual" is present wont this variable has_nonvisual_metadata be set?

[mattgarrish]

Yes, it's not the variable that checks if the content is all available as text. This is checking if the publisher has set the nonvisual metadata we need to test for completeness.

[clapierre]

non-textual shouldn't it be <>"textual"?

[mattgarrish]

No, there's a not(...) around the value test to capture anything but textual values.

[mattgarrish]

Also, as I wrote in #633, if we accept this change the descriptive version for alternative text should not include "for images" in it. When the visual content is a video with a transcript, this description would be inaccurate.