def test_tok_doctype():
e = [(doctype, '<!doctype>')]
t = list(tok(io.StringIO('<!doctype>')))
assert_equal(t, e)
def test_tok_selfclosing():
e = [(selfclosing, '<foo/>')]
t = list(tok(io.StringIO('<foo/>')))
assert_equal(t, e)
def test_tok_error_comment():
e = [(error, '<!--foo>')]
t = list(tok(io.StringIO('<!--foo>')))
assert_equal(t, e)
def test_tok_error_instruction():
e = [(error, '<?instruction>')]
t = list(tok(io.StringIO('<?instruction>')))
assert_equal(t, e)
def test_tok_opening():
e = [(opening, '<foo>')]
t = list(tok(io.StringIO('<foo>')))
assert_equal(t, e)
def test_tok_instruction():
e = [(instruction, '<?foo?>')]
t = list(tok(io.StringIO('<?foo?>')))
assert_equal(t, e)
def test_empty_text():
t = list(tok(io.StringIO('')))
e = []
assert_equal(t, e)
def tree(xmlfile):
"""Parses XML file and prints a `tree` representation of it."""
from sax.parser.core import pp, xml
print(pp(xml(lt.tok(xmlfile))))
def test_text_opening_text_closing():
bs = 'eww<foo>bar</foo>'
i = list(tok(io.StringIO(bs)))
e = [(text, 'eww'), (opening, '<foo>'), (text, 'bar'), (closing, '</foo>')]
assert_equal(i, e)
def test_instruction_text_instruction():
bs = '<?xml version="1.0" encoding="UTF-8"?>text<?instruction?>'
i = list(tok(io.StringIO(bs)))
e = [(instruction, '<?xml version="1.0" encoding="UTF-8"?>'),
(text, 'text'), (instruction, '<?instruction?>')]
assert_equal(i, e)
def test_instruction_too_short():
bs = '<?xml version="1.0" encoding="UTF-8"?'
i = list(tok(io.StringIO(bs)))
e = [(error, '<?xml version="1.0" encoding="UTF-8"?')]
assert_equal(i, e)
def test_selfclosing():
bs = '<foo/>'
i = list(tok(io.StringIO(bs)))
e = [(selfclosing, '<foo/>')]
assert_equal(i, e)
def test_opening_attrs():
bs = '<foo a="a" b="b">'
i = list(tok(io.StringIO(bs)))
e = [(opening, '<foo a="a" b="b">')]
assert_equal(i, e)
def test_text_opening():
bs = 'text<foo>'
t = list(tok(io.StringIO(bs)))
e = [(text, 'text'), (opening, '<foo>')]
assert_equal(t, e)
def test_tok_text():
e = [(text, 'foo')]
t = list(tok(io.StringIO('foo')))
assert_equal(t, e)
def test_opening_closing():
bs = '<foo></foo>'
i = list(tok(io.StringIO(bs)))
e = [(opening, '<foo>'), (closing, '</foo>')]
assert_equal(i, e)
import os
import click
import sax.tokenizer.gen as gt
import sax.tokenizer.loop as lt
src = './samples/dbus-systemd1.xml'
tsg = list(gt.tok(open(src)))
tsl = list(lt.tok(open(src)))
def k(n=32):
for k, t in tsl[:n]:
print(k, t)
diffs = [(g, l) for g, l in zip(tsg, tsl) if g != l]
'''
Intersting, parsers have different streams. Mostly out of
sync because of the selfclosing `tokenistic sugar` trick,
causing more tokens emitted by gen.tok:
<sctag/> -> (opening, sctag), (closing, sctag)
instead of
<sctag/> -> (selfclosing, sctag)
but may resolve the same trees when parsed.
