class Hyperlink(XmlModel):
XML_TAG = 'hyperlink'
hyperlink_id = XmlAttribute(name='id')
anchor = XmlAttribute(name='anchor')
children = XmlCollection(Run, )
@memoized
def get_target_uri(self):
if not self.container:
return None
if not self.container.package_part:
return None
package_part = self.container.package_part
try:
relationship = package_part.get_relationship(
relationship_id=self.hyperlink_id, )
except KeyError:
return None
if self.anchor:
return '{0}#{1}'.format(relationship.target_uri, self.anchor)
else:
return relationship.target_uri
@property
def target_uri(self):
return self.get_target_uri()
@target_uri.setter
def target_uri(self, target_uri):
self.get_target_uri.memo.set_cache(target_uri, self)
class Style(XmlModel):
XML_TAG = 'style'
style_type = XmlAttribute(name='type', default='paragraph')
style_id = XmlAttribute(name='styleId', default='')
name = XmlChild(attrname='val', default='')
run_properties = XmlChild(type=RunProperties)
parent_style = XmlChild(name='basedOn', attrname='val')
def is_a_heading(self):
if not self.name:
return False
return self.name.lower().startswith('heading')
class SimpleField(XmlModel):
XML_TAG = 'fldSimple'
instr = XmlAttribute()
children = XmlCollection(
Run,
Hyperlink,
SmartTagRun,
InsertedRun,
DeletedRun,
SdtRun,
)
def _parse_instr_into_field_type_and_arg_string(self):
return re.match('^\s*([^\s]+)\s*(.*)$', self.instr)
def _parse_instr_arg_string_to_args(self, arg_string):
return re.findall(r'\s*(?:"([^"]+)"|([^\s]+))+', arg_string)
def parse_instr(self):
m = self._parse_instr_into_field_type_and_arg_string()
if not m:
return
field_type = m.group(1)
raw_field_args = m.group(2)
if not raw_field_args:
return field_type, None
m = self._parse_instr_arg_string_to_args(raw_field_args)
if not m:
return field_type, None
field_args = [args[0] if args[0] else args[1] for args in m]
return field_type, field_args
class NumberingInstance(XmlModel):
XML_TAG = 'num'
num_id = XmlAttribute(name='numId')
abstract_num_id = XmlChild(name='abstractNumId', attrname='val')
level_overrides = XmlCollection(LevelOverride)
class AbstractNum(XmlModel):
XML_TAG = 'abstractNum'
abstract_num_id = XmlAttribute(name='abstractNumId')
name = XmlChild(attrname='val')
levels = XmlCollection(Level)
def __init__(self, **kwargs):
super(AbstractNum, self).__init__(**kwargs)
self._levels = {}
for level in self.levels:
self._levels[level.level_id] = level
def get_level(self, level_id):
return self._levels.get(level_id)
def get_indentation_between_levels(self):
"""
Depending on the word version we may get a different default indentation between
levels. For this we will only check first 2 levels as the other follow the same step.
"""
try:
lvl0_ind = self.levels[0].paragraph_properties.to_int(
'indentation_left', default=0)
lvl1_ind = self.levels[1].paragraph_properties.to_int(
'indentation_left', default=0)
ind_step = lvl1_ind - lvl0_ind
except IndexError:
ind_step = 720 # default one
return ind_step
class Break(XmlModel):
XML_TAG = 'br'
break_type = XmlAttribute(name='type')
def is_page_break(self):
return self.break_type == 'page'
class RFonts(XmlModel):
XML_TAG = 'rFonts'
hint = XmlAttribute(name='hint')
ascii = XmlAttribute(name='ascii')
h_ansi = XmlAttribute(name='hAnsi')
east_asia = XmlAttribute(name='eastAsia')
cs = XmlAttribute(name='cs')
ascii_theme = XmlAttribute(name='asciiTheme')
h_ansi_theme = XmlAttribute(name='hAnsiTheme')
east_asia_theme = XmlAttribute(name='eastAsiaTheme')
cs_theme = XmlAttribute(name='cstheme')
def is_symbol(self):
return self.h_ansi == 'Symbol'
class Endnote(XmlModel):
XML_TAG = 'endnote'
endnote_id = XmlAttribute(name='id')
children = XmlCollection(
Paragraph,
Table,
InsertedRun,
DeletedRun,
)
class Shape(XmlModel):
XML_TAG = 'shape'
style = XmlAttribute()
children = XmlCollection(ImageData, 'vml.Textbox')
# TODO perhaps we could have a prepare_style, or clean_style convention?
def get_style(self):
if self.style:
return dict(
item.split(':', 1) for item in self.style.split(';') if item)
return {}
class ImageData(XmlModel):
XML_TAG = 'imagedata'
# TODO We need namespaced attributes, because of conflicts like this. This
# attribute is in the relationship namespace, and there's another attribute
# named "id" which is in the default (in this case VML) namespace.
# See https://msdn.microsoft.com/en-us/library/documentformat.openxml.vml.imagedata%28v=office.14%29.aspx # noqa
relationship_id = XmlAttribute(name='id')
def get_picture_extents(self):
style = self.parent.get_style()
width = style.get('width', 0)
height = style.get('height', 0)
# TODO if width/height are missing units, "px" is implied
return width, height
class EndnoteReference(XmlModel):
XML_TAG = 'endnoteReference'
endnote_id = XmlAttribute(name='id')
@property
def endnote(self):
if not self.endnote_id:
return
part = self.container.endnotes_part
if not part:
return
endnotes = part.endnotes
endnote = endnotes.get_endnote_by_id(endnote_id=self.endnote_id)
return endnote
class Level(XmlModel):
XML_TAG = 'lvl'
level_id = XmlAttribute(name='ilvl')
start = XmlChild(attrname='val')
num_format = XmlChild(name='numFmt', attrname='val')
restart = XmlChild(name='lvlRestart', attrname='val')
paragraph_style = XmlChild(name='pStyle', attrname='val')
run_properties = XmlChild(type=RunProperties)
paragraph_properties = XmlChild(type=ParagraphProperties)
def is_bullet_format(self):
return self.num_format == 'bullet'
def format_is_none(self):
if not self.num_format:
return True
return self.num_format.lower() == 'none'
class AbstractNum(XmlModel):
XML_TAG = 'abstractNum'
abstract_num_id = XmlAttribute(name='abstractNumId')
name = XmlChild(attrname='val')
levels = XmlCollection(Level)
def __init__(self, **kwargs):
super(AbstractNum, self).__init__(**kwargs)
self._levels = {}
for level in self.levels:
self._levels[level.level_id] = level
def get_level(self, level_id):
return self._levels.get(level_id)
class FieldChar(XmlModel):
XML_TAG = 'fldChar'
_char_type = XmlAttribute(name='fldCharType')
@property
def char_type(self):
if not self._char_type:
return
return self._char_type.lower()
def is_type_begin(self):
return self.char_type == 'begin'
def is_type_separate(self):
return self.char_type == 'separate'
def is_type_end(self):
return self.char_type == 'end'
class Document(XmlModel):
XML_TAG = 'document'
conformance = XmlAttribute(name='conformance')
body = XmlChild(type=Body)
class OrangeModel(XmlModel):
XML_TAG = 'orange'
type = XmlAttribute(default='Organic')
class LevelOverride(XmlModel):
XML_TAG = 'lvlOverride'
level_id = XmlAttribute(name='ilvl')
start_override = XmlChild(name='startOverride', attrname='val')
level = XmlChild(type=Level)
class Blip(XmlModel):
XML_TAG = 'blip'
embedded_picture_id = XmlAttribute(name='embed')
linked_picture_id = XmlAttribute(name='link')
class Transform2D(XmlModel):
XML_TAG = 'xfrm'
extents = XmlChild(type=Extents)
rotate = XmlAttribute(name='rot', default=None)
class DocPr(XmlModel):
XML_TAG = 'docPr'
title = XmlAttribute(name='title')
descr = XmlAttribute(name='descr')
class Extents(XmlModel):
XML_TAG = 'ext'
length = XmlAttribute(name='cx')
width = XmlAttribute(name='cy')
class Bookmark(XmlModel):
XML_TAG = 'bookmarkStart'
name = XmlAttribute(name='name')
class AppleModel(XmlModel):
XML_TAG = 'apple'
type = XmlAttribute(default='Honey Crisp')