def __init__(self, lyx, outcb=None, debugcb=None, debug_level=0):
self.app_debugcb = debugcb
self.debug_level = debug_level
if debugcb:
self.dbg = self._debugcb
else:
self.dbg = self._null_debugcb
self.stack = []
self.i = 0
if type(lyx) == list:
self.lines = lyx
elif type(lyx) == str:
self.lines = _read_lyx(lyx)
else:
assert lyx.readlines
self.lines = lyx.readlines()
self.xout = XmlStreamer(outcb or LyX2XML._outcb, 'lyx') # No DTD...
# XXX Should move the rest of this to a 'convert' method.
#
# Fix the lack of XML-ish nesting of things like \series, \emph,
# \family, \color, \shape, and \lang
self._fix_text_styling()
# Uncomment to save a copy of the .lyx with fixed styling, for
# debugging purposes:
#f = open('/tmp/f', 'w+')
#for i in range(len(self.lines)):
# f.write(self.lines[i])
# if self.lines[i][-1] != '\n':
# f.write('\n')
#f.close()
if self.debug_level >= 3:
self.dbg(3, 'Fixed lines:\n')
for i in range(len(self.lines)):
self.dbg(3, '%d %s' % (i + 1, self.lines[i]))
self._lyx2xml()
self.xout.finish()
return None
def __init__(self, lyx, outcb=None, debugcb=None, debug_level=0):
self.app_debugcb = debugcb
self.debug_level = debug_level
if debugcb:
self.dbg = self._debugcb
else:
self.dbg = self._null_debugcb
self.stack = []
self.i = 0
if type(lyx) == list:
self.lines = lyx
elif type(lyx) == str:
self.lines = _read_lyx(lyx)
else:
assert lyx.readlines
self.lines = lyx.readlines()
self.xout = XmlStreamer(outcb or LyX2XML._outcb, 'lyx') # No DTD...
# XXX Should move the rest of this to a 'convert' method.
#
# Fix the lack of XML-ish nesting of things like \series, \emph,
# \family, \color, \shape, and \lang
self._fix_text_styling()
# Uncomment to save a copy of the .lyx with fixed styling, for
# debugging purposes:
#f = open('/tmp/f', 'w+')
#for i in range(len(self.lines)):
# f.write(self.lines[i])
# if self.lines[i][-1] != '\n':
# f.write('\n')
#f.close()
if self.debug_level >= 3:
self.dbg(3, 'Fixed lines:\n')
for i in range(len(self.lines)):
self.dbg(3, '%d %s' % (i + 1, self.lines[i]))
self._lyx2xml()
self.xout.finish()
return None
class LyX2XML(object):
debug = False
def _debugcb(self, level, text):
if level <= self.debug_level:
self.app_debugcb(text)
def _null_debugcb(self, level, text): pass
def _outcb(self, text):
sys.stdout.write(text)
def __init__(self, lyx, outcb=None, debugcb=None, debug_level=0):
self.app_debugcb = debugcb
self.debug_level = debug_level
if debugcb:
self.dbg = self._debugcb
else:
self.dbg = self._null_debugcb
self.stack = []
self.i = 0
if type(lyx) == list:
self.lines = lyx
elif type(lyx) == str:
self.lines = _read_lyx(lyx)
else:
assert lyx.readlines
self.lines = lyx.readlines()
self.xout = XmlStreamer(outcb or LyX2XML._outcb, 'lyx') # No DTD...
# XXX Should move the rest of this to a 'convert' method.
#
# Fix the lack of XML-ish nesting of things like \series, \emph,
# \family, \color, \shape, and \lang
self._fix_text_styling()
# Uncomment to save a copy of the .lyx with fixed styling, for
# debugging purposes:
#f = open('/tmp/f', 'w+')
#for i in range(len(self.lines)):
# f.write(self.lines[i])
# if self.lines[i][-1] != '\n':
# f.write('\n')
#f.close()
if self.debug_level >= 3:
self.dbg(3, 'Fixed lines:\n')
for i in range(len(self.lines)):
self.dbg(3, '%d %s' % (i + 1, self.lines[i]))
self._lyx2xml()
self.xout.finish()
return None
# Returns one past the new location of self.lines[i], after possibly
# inserting lines to close and re-open tags as implied by
# self.lines[i], to make it possible to emit proper XML.
def _close_and_reopen_styling(self, i, tag, new_style):
lines = self.lines
stack = self.stack
found = False
# XXX Could use an __in__ operator overloading for this code (or
# could use lists' in operator with a list comprehension to make
# a list with just the first element of each tuple in stack):
for k in range(len(stack) - 1, -1, -1):
if stack[k][0] == tag:
found = True
break
if not found:
# We're just opening a tag that's not already in the stack.
# Not much to do in this case.
assert new_style != mixed_tags[tag]
self.dbg(2, 'adding (\\%s, %s) to the stack (%s) at %d' % (tag, new_style, repr(stack), i))
stack.append((tag, new_style))
return i + 1
# OK, we have work to do: close all intervening tags, close this
# one, re-open all those tags and this one.
self.dbg(2, 'fixing ordering for \\%s at line %d, stack = %s' % (tag, i, repr(stack)))
m = -1
for k in range(len(stack) - 1, -1, -1):
# Close tags
m = k
# If this element in the stack is not the one we'll
# terminate at (tag) or if it is but we're opening a new
# style:
if stack[k][0] != tag or new_style != mixed_tags[tag]:
# Close the tag stack[k][0]
self.dbg(2, 'fixing ordering for \\%s by closing %s %s to re-open it' % (tag, stack[k][0], stack[k][1]))
lines.insert(i, '\\' + stack[k][0] + ' ' + mixed_tags[stack[k][0]])
i += 1 # keep i pointing at the line triggering all this
if stack[k][0] == tag:
stack[k] = (tag, new_style)
m += 1
break
else:
assert stack[k][0] == tag and new_style == mixed_tags[tag]
# This line closes a tag, so just remove the
# corresponding entry in the stack (any others before
# this one will have been closed above).
del(stack[k])
break
# Now re-open those intervening tags that we closed; we do this
# *after* the line triggering all this, since that's the one
# opening a tag.
i += 1
k = m
while k > -1 and k < len(stack):
lines.insert(i, '\\' + stack[k][0] + ' ' + stack[k][1])
k += 1
i += 1
self.dbg(5, 'foo at %d: %s' % (i, lines[i]))
return i
def _fix_text_styling(self):
lines = self.lines
self.stack = []
stack = self.stack
fixes = []
# First we'll rename the \color in the header, if any.
i = find_tokens(lines, ('\\color', '\\end_header'), 0)
if i != -1 and \
check_token(lines[i], '\\color') and \
not get_containing_layout(lines, i):
lines[i] = '\\color_in_header ' + lines[i].split()[1]
# Now let's get on with the rest
i = 0
while i < len(lines):
line = lines[i]
self.dbg(4, 'looking at line %d: %s' % (i, line))
# XXX We really should simplify all this startswith()
# nonsense.
if line.startswith('\\end_'):
self.dbg(2, 'fixing unended style tags %s' % (repr(stack)))
for k in range(len(stack) - 1, -1, -1):
lines.insert(i, '\\' + stack[k][0] + ' ' + mixed_tags[stack[k][0]])
i += 1
del(stack[0:])
i += 1
continue
if not line.startswith('\\') or line.find(' ') == -1:
self.dbg(5, '1 i++ at %d' % (i,))
i += 1
continue
# XXX And this parsing nonsense needs refactoring too
line = _chomp(line[1:])
a = line[0:line.find(' ')]
if not a in mixed_tags:
self.dbg(5, '2 i++ at %d' % (i,))
i += 1
continue
v = line[line.find(' ') + 1:]
# We're opening a new whatever it is. But we need to handle
# the possibility that we're changing the whatever it is!
# How to handle this? We could convert:
# \lang french foo \lang spanish bar \lang english foobar
# to
# <lang a="french">foo<lang a="spanish">bar</lang></lang>
# or to
# <lang a="french">foo</lang><lang a="spanish">bar</lang>
# The former might be easier: just close all tags up to the
# farthest one in the stack for the tag being closed, then
# re-open any others that were found along the way. But the
# latter is more sensible, so that's what we do.
# Invariant: we never have more than one style tag in the
# stack (XXX assert this in code).
self.dbg(4, 'seen \\%s at line %d' % (line, i))
i = self._close_and_reopen_styling(i, a, v)
self.dbg(5, '3 i++ at %d' % (i,))
def _lyxml2xml(self, start, end):
lines = self.lines
xout = self.xout
i = start
while i < end:
self.dbg(3, 'Parsing line %d: %s' % (i, lines[i]))
if lines[i].startswith('</'):
xout.end_elt(lines[i][2:-2])
i += 1
elif lines[i].startswith('<'):
(tag, attrs, start_tok, end_tok) = _parse_xml_tag(lines[i])
# There don't seem to be tags with no child nodes in .lyx,
# but let's handle them just in case.
if lines[i][-2] != '/':
e = find_end_of(lines, i, start_tok, end_tok)
self.dbg(3, 'find_end_of(%d, %s, %s) = %d' % (i, start_tok, end_tok, e))
# lyxtabular's <column> and <features> don't get closed!
# What a mess.
if e == -1:
e = None
else:
e = None
xout.start_elt(tag)
xout.attr('embedded_xml', 'true')
for a in attrs:
xout.attr(a[0], a[1])
if e:
i = self._lyxml2xml(i + 1, e)
self.dbg(3, '_lyx2xml(...) = %d, looking for %d' % (i, e))
if i + 1 == e:
i += 1
else:
self.dbg(3, '_lyx2xml() returned %d, e = %d' % (i, e))
else:
xout.end_elt(tag)
i += 1
else:
i = self._lyx2xml(i, end)
return i
def _handle_interspersed_attrs(self, start, end):
lines = self.lines
xout = self.xout
i = start
depth = 0
while i < end:
if lines[i].startswith('\\begin_') or lines[i].startswith('\\index '):
depth += 1
elif lines[i].startswith('\\end_'):
depth -= 1
elif depth == 0 and lines[i].startswith('\\'):
(a, v) = _parse_attr(lines[i])
if a and v:
xout.attr(a, v)
if a == 'language':
# Set the default language; needed for the style
# tag fix code.
mixed_tags['lang'] = v
self.dbg(2, 'Default language is %s' % (v,))
i += 1
def _lyx2xml(self, start=0, end=-1, cmd_type=None):
lines = self.lines
xout = self.xout
i = start
if end < 0:
end = len(lines)
prev_i = -1
while i < end:
self.dbg(3, 'Parsing line %d: %s' % (i, lines[i]))
assert i > prev_i
prev_i = i
if len(lines[i]) == 0 or lines[i] == ' ':
# Ignore empty lines
i += 1
cmd_type = None
elif lines[i][0] == '#':
# LyX source comment
#xout.comment(lines[i][1:])
i += 1
cmd_type = None
elif _beginner(lines[i]):
(el, start_tok, end_tok, cmd_type, rest) = _parse_begin(lines[i])
xout.start_elt(el)
if el == 'document':
# Default XML namespace
xout.attr('xmlns', 'urn:cryptonector.com:lyx-other')
# We put layouts, insets, and custome insets into
# separate namespaces
xout.attr('xmlns:layout', 'urn:cryptonector.com:lyx-layout')
xout.attr('xmlns:inset', 'urn:cryptonector.com:lyx-inset')
xout.attr('xmlns:flex', 'urn:cryptonector.com:lyx-flex')
# XXX MathML not yet implemented; we get to turn LyX
# formulas into MathML, joy! But it looks like
# LaTeX math, and there's tools for converting that
# to MathML, so, hey, it might not be too much work.
xout.attr('xmlns:math', 'http://www.w3.org/1998/Math/MathML')
# Suck in \lyxformat, which precedes \begin_document,
# and anything else that might be there (though
# nothing will be).
self._handle_interspersed_attrs(0, i)
if (el.startswith('inset:') or el.startswith('flex:')) and \
not cmd_type and (i + 1) < end and \
lines[i + 1].startswith('status '):
i += 1 # skip status open|collapsed line
status = _chomp(lines[i][lines[i].find(' ') + 1:])
else:
status = None
self.dbg(4, 'lines[%d] = %s' % (i, lines[i]))
e = find_end_of(lines, i, start_tok, end_tok)
assert e != -1
self.dbg(4, 'find_end_of(%d, %s, %s) = %d' % (i, start_tok, end_tok, e))
# XXX Here we need to find any attributes that might be
# interspersed with child nodes so we can suck them in
# first. What a PITA.
self._handle_interspersed_attrs(i + 1, e - 1)
if status:
xout.attr('status', status)
if len(rest) == 2 and el != 'inset:Formula':
xout.attr(rest[0], escape(rest[1]))
i = self._lyx2xml(i + 1, e, cmd_type)
self.dbg(4, '_lyx2xml(...) = %d, looking for %s at %d; end = %d' % (i, end_tok, e, end))
if i + 1 == e:
i += 1
else:
self.dbg(4, '_lyx2xml() returned %d, e = %d; end = %d' % (i, e, end))
assert lines[i].startswith('\\end_')
if len(rest) == 2 and el == 'inset:Formula':
xout.text(' ')
xout.text(escape(rest[1]))
xout.end_elt(el)
cmd_type = None
i += 1
elif cmd_type == 'inset':
# Parse "\begin_inset CommandInset ..." attributes
self.dbg(4, 'lines[%d] = %s' % (i, lines[i]))
while i < end and lines[i] != '' and lines[i] != ' ':
(a, v) = _parse_attr(lines[i])
if not a or not v:
break
xout.attr(a, v)
i += 1
# then suck in content
cmd_type = None
elif cmd_type == 'XML':
# Parse embedded XML contents
i = self._lyxml2xml(i, end)
cmd_type = None
elif lines[i][0] == '\\' and \
not lines[i].startswith('\\begin_') and \
not lines[i].startswith('\\end_') and \
not _key(lines[i]) in mixed_tags:
# An attribute, which we've handled above with the call to
# _handle_interspersed_attrs().
i += 1
else:
line = lines[i]
if xout.stack[-1] == 'layout':
line = _chomp(line)
key = _key(line)
if key in mixed_tags:
val = _chomp(lines[i][lines[i].find(' ') + 1:])
if val == mixed_tags[key]:
xout.end_elt(key)
else:
xout.start_elt(key)
xout.attr('type', val)
else:
xout.text(escape(line))
cmd_type = None
i += 1
return i
class LyX2XML(object):
debug = False
def _debugcb(self, level, text):
if level <= self.debug_level:
self.app_debugcb(text)
def _null_debugcb(self, level, text):
pass
def _outcb(self, text):
sys.stdout.write(text)
def __init__(self, lyx, outcb=None, debugcb=None, debug_level=0):
self.app_debugcb = debugcb
self.debug_level = debug_level
if debugcb:
self.dbg = self._debugcb
else:
self.dbg = self._null_debugcb
self.stack = []
self.i = 0
if type(lyx) == list:
self.lines = lyx
elif type(lyx) == str:
self.lines = _read_lyx(lyx)
else:
assert lyx.readlines
self.lines = lyx.readlines()
self.xout = XmlStreamer(outcb or LyX2XML._outcb, 'lyx') # No DTD...
# XXX Should move the rest of this to a 'convert' method.
#
# Fix the lack of XML-ish nesting of things like \series, \emph,
# \family, \color, \shape, and \lang
self._fix_text_styling()
# Uncomment to save a copy of the .lyx with fixed styling, for
# debugging purposes:
#f = open('/tmp/f', 'w+')
#for i in range(len(self.lines)):
# f.write(self.lines[i])
# if self.lines[i][-1] != '\n':
# f.write('\n')
#f.close()
if self.debug_level >= 3:
self.dbg(3, 'Fixed lines:\n')
for i in range(len(self.lines)):
self.dbg(3, '%d %s' % (i + 1, self.lines[i]))
self._lyx2xml()
self.xout.finish()
return None
# Returns one past the new location of self.lines[i], after possibly
# inserting lines to close and re-open tags as implied by
# self.lines[i], to make it possible to emit proper XML.
def _close_and_reopen_styling(self, i, tag, new_style):
lines = self.lines
stack = self.stack
found = False
# XXX Could use an __in__ operator overloading for this code (or
# could use lists' in operator with a list comprehension to make
# a list with just the first element of each tuple in stack):
for k in range(len(stack) - 1, -1, -1):
if stack[k][0] == tag:
found = True
break
if not found:
# We're just opening a tag that's not already in the stack.
# Not much to do in this case.
assert new_style != mixed_tags[tag]
self.dbg(
2, 'adding (\\%s, %s) to the stack (%s) at %d' %
(tag, new_style, repr(stack), i))
stack.append((tag, new_style))
return i + 1
# OK, we have work to do: close all intervening tags, close this
# one, re-open all those tags and this one.
self.dbg(
2, 'fixing ordering for \\%s at line %d, stack = %s' %
(tag, i, repr(stack)))
m = -1
for k in range(len(stack) - 1, -1, -1):
# Close tags
m = k
# If this element in the stack is not the one we'll
# terminate at (tag) or if it is but we're opening a new
# style:
if stack[k][0] != tag or new_style != mixed_tags[tag]:
# Close the tag stack[k][0]
self.dbg(
2,
'fixing ordering for \\%s by closing %s %s to re-open it' %
(tag, stack[k][0], stack[k][1]))
lines.insert(
i, '\\' + stack[k][0] + ' ' + mixed_tags[stack[k][0]])
i += 1 # keep i pointing at the line triggering all this
if stack[k][0] == tag:
stack[k] = (tag, new_style)
m += 1
break
else:
assert stack[k][0] == tag and new_style == mixed_tags[tag]
# This line closes a tag, so just remove the
# corresponding entry in the stack (any others before
# this one will have been closed above).
del (stack[k])
break
# Now re-open those intervening tags that we closed; we do this
# *after* the line triggering all this, since that's the one
# opening a tag.
i += 1
k = m
while k > -1 and k < len(stack):
lines.insert(i, '\\' + stack[k][0] + ' ' + stack[k][1])
k += 1
i += 1
self.dbg(5, 'foo at %d: %s' % (i, lines[i]))
return i
def _fix_text_styling(self):
lines = self.lines
self.stack = []
stack = self.stack
fixes = []
# First we'll rename the \color in the header, if any.
i = find_tokens(lines, ('\\color', '\\end_header'), 0)
if i != -1 and \
check_token(lines[i], '\\color') and \
not get_containing_layout(lines, i):
lines[i] = '\\color_in_header ' + lines[i].split()[1]
# Now let's get on with the rest
i = 0
while i < len(lines):
line = lines[i]
self.dbg(4, 'looking at line %d: %s' % (i, line))
# XXX We really should simplify all this startswith()
# nonsense.
if line.startswith('\\end_'):
self.dbg(2, 'fixing unended style tags %s' % (repr(stack)))
for k in range(len(stack) - 1, -1, -1):
lines.insert(
i, '\\' + stack[k][0] + ' ' + mixed_tags[stack[k][0]])
i += 1
del (stack[0:])
i += 1
continue
if not line.startswith('\\') or line.find(' ') == -1:
self.dbg(5, '1 i++ at %d' % (i, ))
i += 1
continue
# XXX And this parsing nonsense needs refactoring too
line = _chomp(line[1:])
a = line[0:line.find(' ')]
if not a in mixed_tags:
self.dbg(5, '2 i++ at %d' % (i, ))
i += 1
continue
v = line[line.find(' ') + 1:]
# We're opening a new whatever it is. But we need to handle
# the possibility that we're changing the whatever it is!
# How to handle this? We could convert:
# \lang french foo \lang spanish bar \lang english foobar
# to
# <lang a="french">foo<lang a="spanish">bar</lang></lang>
# or to
# <lang a="french">foo</lang><lang a="spanish">bar</lang>
# The former might be easier: just close all tags up to the
# farthest one in the stack for the tag being closed, then
# re-open any others that were found along the way. But the
# latter is more sensible, so that's what we do.
# Invariant: we never have more than one style tag in the
# stack (XXX assert this in code).
self.dbg(4, 'seen \\%s at line %d' % (line, i))
i = self._close_and_reopen_styling(i, a, v)
self.dbg(5, '3 i++ at %d' % (i, ))
def _lyxml2xml(self, start, end):
lines = self.lines
xout = self.xout
i = start
while i < end:
self.dbg(3, 'Parsing line %d: %s' % (i, lines[i]))
if lines[i].startswith('</'):
xout.end_elt(lines[i][2:-2])
i += 1
elif lines[i].startswith('<'):
(tag, attrs, start_tok, end_tok) = _parse_xml_tag(lines[i])
# There don't seem to be tags with no child nodes in .lyx,
# but let's handle them just in case.
if lines[i][-2] != '/':
e = find_end_of(lines, i, start_tok, end_tok)
self.dbg(
3, 'find_end_of(%d, %s, %s) = %d' %
(i, start_tok, end_tok, e))
# lyxtabular's <column> and <features> don't get closed!
# What a mess.
if e == -1:
e = None
else:
e = None
xout.start_elt(tag)
xout.attr('embedded_xml', 'true')
for a in attrs:
xout.attr(a[0], a[1])
if e:
i = self._lyxml2xml(i + 1, e)
self.dbg(3, '_lyx2xml(...) = %d, looking for %d' % (i, e))
if i + 1 == e:
i += 1
else:
self.dbg(3, '_lyx2xml() returned %d, e = %d' % (i, e))
else:
xout.end_elt(tag)
i += 1
else:
i = self._lyx2xml(i, end)
return i
def _handle_interspersed_attrs(self, start, end):
lines = self.lines
xout = self.xout
i = start
depth = 0
while i < end:
if lines[i].startswith('\\begin_') or lines[i].startswith(
'\\index '):
depth += 1
elif lines[i].startswith('\\end_'):
depth -= 1
elif depth == 0 and lines[i].startswith('\\'):
(a, v) = _parse_attr(lines[i])
if a and v:
xout.attr(a, v)
if a == 'language':
# Set the default language; needed for the style
# tag fix code.
mixed_tags['lang'] = v
self.dbg(2, 'Default language is %s' % (v, ))
i += 1
def _lyx2xml(self, start=0, end=-1, cmd_type=None):
lines = self.lines
xout = self.xout
i = start
if end < 0:
end = len(lines)
prev_i = -1
while i < end:
self.dbg(3, 'Parsing line %d: %s' % (i, lines[i]))
assert i > prev_i
prev_i = i
if len(lines[i]) == 0 or lines[i] == ' ':
# Ignore empty lines
i += 1
cmd_type = None
elif lines[i][0] == '#':
# LyX source comment
#xout.comment(lines[i][1:])
i += 1
cmd_type = None
elif _beginner(lines[i]):
(el, start_tok, end_tok, cmd_type,
rest) = _parse_begin(lines[i])
xout.start_elt(el)
if el == 'document':
# Default XML namespace
xout.attr('xmlns', 'urn:cryptonector.com:lyx-other')
# We put layouts, insets, and custome insets into
# separate namespaces
xout.attr('xmlns:layout',
'urn:cryptonector.com:lyx-layout')
xout.attr('xmlns:inset', 'urn:cryptonector.com:lyx-inset')
xout.attr('xmlns:flex', 'urn:cryptonector.com:lyx-flex')
# XXX MathML not yet implemented; we get to turn LyX
# formulas into MathML, joy! But it looks like
# LaTeX math, and there's tools for converting that
# to MathML, so, hey, it might not be too much work.
xout.attr('xmlns:math',
'http://www.w3.org/1998/Math/MathML')
# Suck in \lyxformat, which precedes \begin_document,
# and anything else that might be there (though
# nothing will be).
self._handle_interspersed_attrs(0, i)
if (el.startswith('inset:') or el.startswith('flex:')) and \
not cmd_type and (i + 1) < end and \
lines[i + 1].startswith('status '):
i += 1 # skip status open|collapsed line
status = _chomp(lines[i][lines[i].find(' ') + 1:])
else:
status = None
self.dbg(4, 'lines[%d] = %s' % (i, lines[i]))
e = find_end_of(lines, i, start_tok, end_tok)
assert e != -1
self.dbg(
4, 'find_end_of(%d, %s, %s) = %d' %
(i, start_tok, end_tok, e))
# XXX Here we need to find any attributes that might be
# interspersed with child nodes so we can suck them in
# first. What a PITA.
self._handle_interspersed_attrs(i + 1, e - 1)
if status:
xout.attr('status', status)
if len(rest) == 2 and el != 'inset:Formula':
xout.attr(rest[0], escape(rest[1]))
i = self._lyx2xml(i + 1, e, cmd_type)
self.dbg(
4, '_lyx2xml(...) = %d, looking for %s at %d; end = %d' %
(i, end_tok, e, end))
if i + 1 == e:
i += 1
else:
self.dbg(
4, '_lyx2xml() returned %d, e = %d; end = %d' %
(i, e, end))
assert lines[i].startswith('\\end_')
if len(rest) == 2 and el == 'inset:Formula':
xout.text(' ')
xout.text(escape(rest[1]))
xout.end_elt(el)
cmd_type = None
i += 1
elif cmd_type == 'inset':
# Parse "\begin_inset CommandInset ..." attributes
self.dbg(4, 'lines[%d] = %s' % (i, lines[i]))
while i < end and lines[i] != '' and lines[i] != ' ':
(a, v) = _parse_attr(lines[i])
if not a or not v:
break
xout.attr(a, v)
i += 1
# then suck in content
cmd_type = None
elif cmd_type == 'XML':
# Parse embedded XML contents
i = self._lyxml2xml(i, end)
cmd_type = None
elif lines[i][0] == '\\' and \
not lines[i].startswith('\\begin_') and \
not lines[i].startswith('\\end_') and \
not _key(lines[i]) in mixed_tags:
# An attribute, which we've handled above with the call to
# _handle_interspersed_attrs().
i += 1
else:
line = lines[i]
if xout.stack[-1] == 'layout':
line = _chomp(line)
key = _key(line)
if key in mixed_tags:
val = _chomp(lines[i][lines[i].find(' ') + 1:])
if val == mixed_tags[key]:
xout.end_elt(key)
else:
xout.start_elt(key)
xout.attr('type', val)
else:
xout.text(escape(line))
cmd_type = None
i += 1
return i