add the default xml declaration in writeGlyphToString #157
Conversation
Lib/ufoLib/glifLib.py
Outdated
| @@ -628,7 +628,10 @@ def writeGlyphToString(glyphName, glyphObject=None, drawPointsFunc=None, formatV | |||
| _writeLib(glyphObject, root, validate) | |||
| # return the text | |||
| tree = etree.ElementTree(root) | |||
| text = etree.tostring(root, encoding=unicode, pretty_print=True) | |||
There was a problem hiding this comment.
by the way, this was also doubly wrong, because if anything it should have been encoding="unicode" (as a string), not the built-in unicode function.
There was a problem hiding this comment.
Not according to the lxml doc I read, but on vacation right now so can’t dig the reference up, if it works the other way, great
There was a problem hiding this comment.
@benkiel you’re right, you can also pass the unicode function to the encoding argument:
I was misled by the fact that in the official docs I linked earlier there is no mention of that, only of “unicode” as a string.
Nevertheless, we need to restore the xml declaration, otherwise attempting to parse the xml with another parser different from xml might not work properly.
|
/cc @miguelsousa |
the writeGlyphToString will still return a unicode string, like it's always done. However, in the writeGlyph method we can call a private _writeGlyphToBytes and write UTF-8 bytes directly instead of decoding them, then re-encoding them as we write to the file.
|
@anthrotype this is breaking fontParts so something is amiss withe the xml declaration |
|
i'll fix that |
|
fixed with 4cb26a3 i'll cut a bugfix release once the bots are green. |
|
oops wrong link, 4cb26a3 |
This is a regression from ufoLib v2.1.1, see
adobe-type-tools/afdko#462 (comment)
https://travis-ci.org/adobe-type-tools/afdko/jobs/401751859#L8427
when we were using the fonttools XMLWriter the xml declaration was always included. When using lxml with encoding="unicode" there is no default xml declaration.
https://lxml.de/parsing.html#python-unicode-strings