def _get_style_borders(w_styles, style_id):
if w_styles is None or style_id is None:
return {}
w_style = value_of(w_styles,
'w:style[@w:styleId = "{0}"]'.format(style_id))
if w_style is None:
return {}
# - get parent styles (if it exist)
based_on_id = value_of(w_style, 'w:basedOn/@w:val')
parent_styles = _get_style_borders(w_styles, based_on_id)
# - get child styles
w_tbl_borders = value_of(w_style, 'w:tblPr/w:tblBorders')
table_borders = _get_table_borders(w_tbl_borders)
cell_borders = _get_cell_borders(w_tbl_borders)
child_styles = table_borders.copy()
child_styles.update(
{'x-cell-' + key: value
for key, value in cell_borders.items()})
# - *child_styles* override *parent_styles*
real_styles = parent_styles.copy()
real_styles.update({
key: value
for key, value in child_styles.items() if value is not None
})
return real_styles
def _get_border_properties(w_tbl_borders, style_xpath_mapping):
# - Get the cell properties for each direction: 'top', 'right'...
# Values are converted to HTML values, size are in 'pt'
properties = []
for style, xpath in style_xpath_mapping:
prop = {}
color = value_of(w_tbl_borders, xpath.format(attr='color'))
if color and color != "auto":
prop['color'] = "#" + color
shadow = value_of(w_tbl_borders, xpath.format(attr='shadow'))
if shadow:
prop['shadow'] = {"true": True, "false": False}[shadow]
space = value_of(w_tbl_borders, xpath.format(attr='space'))
if space:
# unit is 'pt'
prop['space'] = float(space)
sz = value_of(w_tbl_borders, xpath.format(attr='sz'))
if sz:
# convert eighths of a point to 'pt'
prop['sz'] = float(sz) / 8
val = value_of(w_tbl_borders, xpath.format(attr='val'))
if val:
val = "none" if val == "nil" else val # "nil" is "none" -- no border
prop['val'] = _BORDER_STYLE_MAPPING.get(val, 'w-' + val)
properties.append((style, prop))
return properties
def __init__(self, w_shd):
# type: (etree._Element) -> None
#: Shading Pattern
self.w_val = value_of(w_shd, "@w:val") # required
#: Shading Pattern Color
self.w_color = value_of(w_shd, "@w:color")
#: Shading Background Color
self.w_fill = value_of(w_shd, "@w:fill")
#: Shading Pattern Theme Color
self.w_themeColor = value_of(w_shd, "@w:themeColor")
#: Shading Pattern Theme Color Tint
self.w_themeTint = value_of(w_shd, "@w:themeTint")
#: Shading Pattern Theme Color Shade
self.w_themeShade = value_of(w_shd, "@w:themeShade")
#: Shading Background Theme Color
self.w_themeFill = value_of(w_shd, "@w:themeFill")
#: Shading Background Theme Color Tint
self.w_themeFillTint = value_of(w_shd, "@w:themeFillTint")
#: Shading Background Theme Color Shade
self.w_themeFillShade = value_of(w_shd, "@w:themeFillShade")
def __init__(self, w_pg_sz):
#: Page Width
self.w_w = value_of(w_pg_sz, "@w:w")
#: Page Height
self.w_h = value_of(w_pg_sz, "@w:h")
#: Page Orientation (Possible values are "landscape" and "portrait").
self.w_orient = value_of(w_pg_sz, "@w:orient")
#: Printer Paper Code
self.w_code = value_of(w_pg_sz, "@w:code")
def transform_tables(self, tree):
self._w_styles = etree.parse(
self.styles_path) if self.styles_path else None
self._w_styles = self._w_styles or value_of(tree, ".//w:styles")
for w_tbl in tree.xpath("//w:tbl", namespaces=NS):
table = self.parse_table(w_tbl)
table_elem = self.builder.generate_table_tree(table)
parent = w_tbl.getparent()
index = parent.index(w_tbl)
parent.insert(index, table_elem)
table_elem.tail = w_tbl.tail
parent.remove(w_tbl)
def parse_tr(self, w_tr):
"""
Parse a ``<w:tr>`` element.
See: `Table Row Properties <http://officeopenxml.com/WPtableRowProperties.php>`_.
:type w_tr: etree._Element
:param w_tr: Table element.
"""
# - w:tblHeader => the current row should be repeated at the top
# of each new page on which the table is displayed.
# This is a simple boolean property, so you can specify a val attribute of true or false.
#
# <w:trPr>
# <w:tblHeader/>
# </w:trPr>
#
w_tbl_header = value_of(w_tr, "w:trPr/w:tblHeader")
if w_tbl_header is not None:
w_tbl_header = value_of(w_tr,
"w:trPr/w:tblHeader/@w:val",
default=u"true")
nature = {
"true": u"header",
"false": u"body",
None: u"body"
}[w_tbl_header]
state = self._state
state.row = state.table.rows[state.row_pos]
state.row.nature = nature
# - w:trHeight => height of the row
#
# <w:trPr>
# <w:trHeight w:val="567"/>
# </w:trPr>
#
w_tr_height = value_of(w_tr, "w:trPr/w:trHeight")
if w_tr_height is not None:
h_rule = value_of(w_tr,
"w:trPr/w:tblHeader/@w:hRule",
default="auto")
# Possible values are:
# - atLeast (height should be at least the value specified),
# - exact (height should be exactly the value specified), or
# - auto (height is determined based on the height of the contents, so the value is ignored).
style = {
'atLeast': u'min-height',
'exact': u'height',
'auto': None
}[h_rule]
if style:
val = value_of(w_tr, "w:trPr/w:tblHeader/@w:val", default="0")
# Specifies the row's height, in twentieths of a point.
height = float(val) / 20 # pt
state.row.styles[style] = "{0:0.2f}pt".format(height)
# - w:ins => revision marks: A row can be marked as "inserted".
#
# <w:trPr>
# <w:ins w:id="0" w:author="Laurent Laporte" w:date="2018-11-21T18:08:00Z"/>
# </w:trPr>
#
w_ins = value_of(w_tr, "w:trPr/w:ins")
if w_ins is not None:
state.row.styles['x-ins'] = True
style_xpath_mapping = [
('x-ins-id', "w:trPr/w:ins/@w:id"),
('x-ins-author', "w:trPr/w:ins/@w:author"),
('x-ins-date', "w:trPr/w:ins/@w:date"),
]
for style, xpath in style_xpath_mapping:
value = value_of(w_tr, xpath)
if value:
state.row.styles[style] = value
def parse_tbl(self, w_tbl):
"""
Parse a ``<w:tbl>`` element.
See: `Table Properties <http://officeopenxml.com/WPtableProperties.php>`_.
:type w_tbl: etree._Element
:param w_tbl: Table element.
.. versionchanged:: 0.4.0
The section width and height are now stored in the 'x-sect-size' table style (units in 'pt').
"""
style_id = value_of(w_tbl, "w:tblPr/w:tblStyle/@w:val")
# - Table and borders are extracted from the style (if possible)
# and then from the ``w:tblPr/w:tblBorders`` properties.
style_borders = _get_style_borders(self._w_styles, style_id)
w_tbl_borders = value_of(w_tbl, 'w:tblPr/w:tblBorders')
# - Table borders (frame) and Cell borders (colsep/rowsep) use the "x-cell-" prefix
table_borders = _get_table_borders(w_tbl_borders)
real_table_borders = style_borders.copy()
real_table_borders.update({
key: value
for key, value in table_borders.items() if value is not None
})
attrs = real_table_borders.copy()
# -- Table shading
shd = Shd(value_of(w_tbl, 'w:tblPr/w:shd'))
attrs.update(shd.styles)
# -- Sections: http://officeopenxml.com/WPsection.php
# A section's properties are stored in a sectPr element.
# For all sections except the last section, the sectPr element is stored as
# a child element of the last paragraph in the section. For the last section,
# the sectPr is stored as a child element of the body element.
w_sect_pr = value_of(
w_tbl, 'following::w:p/w:pPr/w:sectPr | following::w:sectPr')
pg_sz = PgSz(value_of(w_sect_pr, 'w:pgSz'))
attrs.update(pg_sz.styles)
# - w:cols -- Specifies the set of columns for the section.
# - ``x-sect-cols``: Section column number
# Default value is "1" -- useful for @pgwide
sect_cols = value_of(w_sect_pr, 'w:cols/@w:num')
if sect_cols is None:
if w_sect_pr is None:
sect_cols = "1" # type: str
else:
sect_cols = w_sect_pr.xpath('count(w:cols/w:col)',
namespaces=NS) # type: float
sect_cols = str(
int(sect_cols)) if sect_cols else "1" # type: str
attrs['x-sect-cols'] = sect_cols
# - The HTML ``class`` attribute is not a regular style.
# We use the table ``nature``instead.
self._state.table = Table(styles=attrs, nature=style_id)
def parse_tc(self, w_tc):
"""
Parse a ``<w:tc>`` element.
See: `Table Cell Properties <http://officeopenxml.com/WPtableCellProperties.php>`_.
:type w_tc: etree._Element
:param w_tc: Table element.
.. versionchanged:: 0.5.1
XML indentation between cell paragraphs is ignored.
"""
state = self._state
# w:gridSpan => number of logical columns across which the cell spans
width = int(value_of(w_tc, "w:tcPr/w:gridSpan/@w:val", default=u"1"))
# take the colspan into account:
state.col_pos += width - 1
# w:vMerge => specifies that the cell is part of a vertically merged set of cells.
w_v_merge = value_of(w_tc, "w:tcPr/w:vMerge")
if w_v_merge is not None:
w_v_merge = value_of(w_tc,
"w:tcPr/w:vMerge/@w:val",
default=u"continue")
if w_v_merge is None:
# no merge
height = 1
elif w_v_merge == u"continue":
# the current cell continues a previously existing merge group
state.table.expand((state.col_pos, state.row.row_pos - 1),
height=1)
height = None
elif w_v_merge == u"restart":
# the current cell starts a new merge group
height = 1
else:
raise NotImplementedError(w_v_merge)
if height:
styles = {}
# -- Cell shading
shd = Shd(value_of(w_tc, 'w:tcPr/w:shd'))
styles.update(shd.styles)
# -- Vertical alignment
#
# w:vAlign => Specifies the vertical alignment for text between the top and bottom margins of the cell.
#
# Possible values are:
# - bottom - Specifies that the text should be vertically aligned to the bottom margin.
# - center - Specifies that the text should be vertically aligned to the center of the cell.
# - top - Specifies that the text should be vertically aligned to the top margin.
w_v_align = value_of(w_tc, "w:tcPr/w:vAlign")
if w_v_align is not None:
w_v_align = value_of(w_tc,
"w:tcPr/w:vAlign/@w:val",
default=u"top")
# CSS/Properties/vertical-align
# valid values: http://www.datypic.com/sc/ooxml/t-w_ST_VerticalJc.html
# fmt: off
v_align = {
"top": u"top",
"center": u"middle",
"bottom": u"bottom",
"both": u"w-both",
}[w_v_align]
# fmt: on
styles["vertical-align"] = v_align
# -- Horizontal alignment
#
# Horizontal alignment is done at paragraph level, inside the cell.
# We can calculate the cell alignment base on the paragraph properties,
# for instance ``<w:p><w:pPr><w:jc w:val="right"/>``,
# see: http://officeopenxml.com/WPalignment.php
#
# We use the most common alignment for cell alignment.
w_p_list = w_tc.xpath("w:p", namespaces=NS)
w_jc_counter = collections.Counter(
value_of(w_p, "w:pPr/w:jc/@w:val") for w_p in w_p_list)
w_jc = w_jc_counter.most_common(1)[0][0] # type: str or None
if w_jc is not None:
# CSS/Properties/text-align
# valid values: http://www.datypic.com/sc/ooxml/t-w_ST_Jc.html
align = {
"start": u"left",
"end": u"right",
"left": u"left",
"right": u"right",
"center": u"center",
"both": u"justify",
"distribute": u"justify",
# "mediumKashida": None,
# "numTab": None,
# "lowKashida": None,
# "thaiDistribute": None
}[w_jc]
styles["align"] = align
# -- Borders
w_tc_borders = value_of(w_tc, 'w:tcPr/w:tcBorders')
cell_borders = _get_table_borders(w_tc_borders)
styles.update(cell_borders)
# todo: calculate the ``@rotate`` attribute.
content = w_tc.xpath('w:p | w:tbl', namespaces=NS)
# ignore the *tail* (if the XML is indented)
for node in content:
node.tail = None
state.row.insert_cell(content,
width=width,
height=height,
styles=styles)