diff options
| author | Bryan Newbold <bnewbold@archive.org> | 2020-11-06 18:32:35 -0800 |
|---|---|---|
| committer | Bryan Newbold <bnewbold@archive.org> | 2020-11-06 18:32:35 -0800 |
| commit | 175019c96fced3e21d0f60ea1a4a37da6b8872ac (patch) | |
| tree | f42fbbe9c8ac06ae9eb06373ab9eec96d2b3a177 /python/sandcrawler/xml.py | |
| parent | b0b66c20c6ffb9d8acc626068964d7dfd5d3bcdc (diff) | |
| parent | 47ca1a273912c8836630b0930b71a4e66fd2c85b (diff) | |
| download | sandcrawler-175019c96fced3e21d0f60ea1a4a37da6b8872ac.tar.gz sandcrawler-175019c96fced3e21d0f60ea1a4a37da6b8872ac.zip | |
Merge branch 'bnewbold-html-ingest'
Diffstat (limited to 'python/sandcrawler/xml.py')
| -rw-r--r-- | python/sandcrawler/xml.py | 7 |
1 files changed, 7 insertions, 0 deletions
diff --git a/python/sandcrawler/xml.py b/python/sandcrawler/xml.py new file mode 100644 index 0000000..7a0086d --- /dev/null +++ b/python/sandcrawler/xml.py @@ -0,0 +1,7 @@ + +import xml.etree.ElementTree as ET + + +def xml_reserialize(raw: bytes) -> str: + root = ET.fromstring(raw) + return '<?xml version="1.0" encoding="UTF-8"?>\n' + ET.tostring(root, encoding="unicode") |