def test_build_tbl__no_seq():
builder = FormexBuilder()
table1 = Table()
table1.rows[1].insert_cell(u"text1")
table1_elem = builder.build_tbl(table1)
table2 = Table()
table2.rows[1].insert_cell(u"text2")
table2_elem = builder.build_tbl(table2)
assert table1_elem.attrib['NO.SEQ'] == u"0001"
assert table2_elem.attrib['NO.SEQ'] == u"0002"
def test_build_cell__align(cell_styles, expected):
builder = CalsBuilder()
# -- create a minimal <row> element
row_elem = etree.XML("<row/>")
# -- setup a minimal cell
p_elem = etree.XML("<p>text</p>")
cell = Cell([p_elem], x=1, y=1, styles=cell_styles)
builder.setup_table(Table([cell]))
# -- build the cell
builder.build_cell(row_elem, cell)
# -- check the '<entry>' attributes
entry_elem = row_elem[0] # type: ElementType
assert entry_elem.tag == u"entry"
# we don't want to check @colsep/@rowsep here
actual = dict(entry_elem.attrib)
actual.pop("colsep", None)
actual.pop("rowsep", None)
assert actual == expected
assert entry_elem[0] == p_elem
def test_build_cell__with_sep(cell_styles, expected):
# .. note::
#
# CALS default value for @colsep and @rowsep is "1".
builder = CalsBuilder()
# -- create a minimal <row> element
row_elem = etree.XML("<row/>")
# -- setup a minimal cell
p_elem = etree.XML("<p>text</p>")
cell_x1_y1 = Cell([p_elem], x=1, y=1, styles=cell_styles)
cell_x2_y1 = Cell([p_elem], x=2, y=1, styles=cell_styles)
cell_x1_y2 = Cell([p_elem], x=1, y=2, styles=cell_styles)
cell_x2_y2 = Cell([p_elem], x=2, y=2, styles=cell_styles)
builder.setup_table(Table([cell_x1_y1, cell_x2_y1, cell_x1_y2,
cell_x2_y2]))
# -- build the cell
builder.build_cell(row_elem, cell_x1_y1)
# -- check the '<entry>' attributes
entry_elem = row_elem[0] # type: ElementType
assert entry_elem.tag == u"entry"
assert entry_elem.attrib == expected
assert entry_elem[0] == p_elem
def test_build_title__subtitle():
table = Table()
content = [P(u"TITLE"), P(u"Subtitle 1"), P(u"Subtitle 2")]
table.rows[1].insert_cell(content, styles={"align": "center"})
builder = FormexBuilder()
tbl_elem = TBL()
builder.build_title(tbl_elem, table.rows[0])
# -- check the '<TITLE>' attributes
title_elem = tbl_elem[0] # type: etree._Element
xml_parser = etree.XMLParser(remove_blank_text=True)
# fmt: off
expected = etree.XML(u"""\
<TITLE>
<TI>
<P>TITLE</P>
</TI>
<STI>
<P>Subtitle 1</P>
<P>Subtitle 2</P>
</STI>
</TITLE>""",
parser=xml_parser)
# fmt: on
diff_list = xmldiff.main.diff_trees(title_elem, expected)
if diff_list:
print(etree.tounicode(title_elem, pretty_print=True, with_tail=False),
file=sys.stderr)
assert diff_list == []
def test_fill_missing_cells():
# fmt: off
table = Table()
table.rows[1].insert_cell("one")
table.rows[1].insert_cell("two")
table.rows[1].insert_cell("three")
table.rows[1].insert_cell("four", height=2)
table.rows[2].insert_cell("un-deux", width=2)
table.fill_missing(table.bounding_box, "???")
text = str(table)
assert text == textwrap.dedent("""\
+-----------+-----------+-----------+-----------+
| one | two | three | four |
+-----------------------+-----------| |
| un-deux | ??? | |
+-----------------------+-----------+-----------+""")
def test_build_tbl__empty_cell():
builder = FormexBuilder()
table1 = Table()
table1.rows[1].insert_cell(u"")
table1_elem = builder.build_tbl(table1)
cell_elem = table1_elem.xpath('//CELL')[0]
assert len(cell_elem) == 1
assert cell_elem[0].tag == 'IE'
def test_colspan():
# fmt: off
table = Table()
table.cols[1].insert_cell("one")
table.cols[1].insert_cell("spanned", width=2)
table.cols[2].insert_cell("two")
text = str(table)
assert text == textwrap.dedent("""\
+-----------+-----------+
| one | two |
+-----------------------+
| spanned |
+-----------------------+""")
def test_rowspan():
# fmt: off
table = Table()
table.rows[1].insert_cell("one")
table.rows[1].insert_cell("spanned", height=2)
table.rows[2].insert_cell("two")
text = str(table)
assert text == textwrap.dedent("""\
+-----------+-----------+
| one | spanned |
+-----------| |
| two | |
+-----------+-----------+""")
def test_build_cell__body(kwargs, attrib):
builder = FormexBuilder()
p_elem = P(u"text")
cell_x1_y1 = Cell([p_elem], x=1, y=1, **kwargs)
table = Table([cell_x1_y1])
builder._table = table
# -- build the cell
row_elem = ROW()
row_y1 = next(iter(table.rows))
builder.build_cell(row_elem, cell_x1_y1, row_y1)
# -- check the '<CELL>' attributes
entry_elem = row_elem[0] # type: etree._Element
assert entry_elem.tag == u"CELL"
assert entry_elem.attrib == attrib
assert entry_elem[0] == p_elem
def test_parse_tc(w_tc_content, expected):
builder = BaseBuilder()
parser = OoxmlParser(builder)
# -- setup a minimal table
state = parser._state
state.table = Table()
state.row = state.table.rows[1]
state.next_col()
# -- parse a <w:tc/>
w_tc = etree.XML(w_tc_content)
parser.parse_tc(w_tc)
# -- check the styles
table = state.table
cell = table[(1, 1)]
assert expected == cell.styles
def test_build_tgroup__tgroup_sorting(tgroup_sorting, expected_tags):
# -- create a table with different natures
table = Table()
row1 = table.rows[1]
row2 = table.rows[2]
row3 = table.rows[3]
row1.nature = "header"
row2.nature = "body"
row3.nature = "footer"
row1.insert_cell("a")
row2.insert_cell("b")
row3.insert_cell("c")
# -- create a builder
table_elem = etree.Element("table")
builder = CalsBuilder(tgroup_sorting=tgroup_sorting)
builder.setup_table(table)
builder.build_tgroup(table_elem, table)
# -- check the tgroup children name and order
actual_tags = [elem.tag for elem in table_elem.xpath("tgroup/*")]
assert actual_tags == expected_tags
def test_parse_tc(w_tc_content, expected):
builder = BaseBuilder()
parser = OoxmlParser(builder)
# -- setup a minimal table
state = parser._state
state.table = Table()
state.row = state.table.rows[1]
state.next_col()
# -- parse a <w:tc/>
w_tc = etree.XML(w_tc_content)
parser.parse_tc(w_tc)
# -- check the styles
table = state.table
cell = table[(1, 1)]
# Ignore cell styles extensions (like 'x-cell-empty').
actual = {
k: v
for k, v in cell.styles.items() if not k.startswith("x-cell-")
}
assert expected == actual
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 test_build_tbl__use_cals():
# see: formex-4/samples/jo-compl-2002C_061/C_2002061EN.01000403.xml
table = Table(
styles={
"border-top": "solid",
"border-bottom": "solid",
"x-sect-orient": "landscape",
"x-sect-cols": "1",
"background-color": "blue",
"width": "180",
})
table.rows[1].nature = "header"
table.rows[1].insert_cell([P(u"Expert group")], styles={"align": "center"})
table.rows[1].insert_cell([P(u"First name and surname of the expert")],
styles={"align": "center"})
table.rows[2].insert_cell([P(u"Control of infectious diseases")])
table.rows[2].insert_cell([P(u"Michael Angelo BORG")])
builder = FormexBuilder(use_cals=True, cals_ns=None)
table_elem = builder.build_tbl(table)
xml_parser = etree.XMLParser(remove_blank_text=True)
# fmt: off
expected = etree.XML(u"""\
<TBL NO.SEQ="0001" COLS="2" PAGE.SIZE="SINGLE.LANDSCAPE">
<CORPUS frame="topbot" colsep="0" rowsep="0" orient="land" pgwide="1" bgcolor="blue" width="180.00mm">
<colspec colname="c1" colnum="1"/>
<colspec colname="c2" colnum="2"/>
<ROW TYPE="HEADER">
<CELL COL="1" align="center">
<P>Expert group</P>
</CELL>
<CELL COL="2" align="center">
<P>First name and surname of the expert</P>
</CELL>
</ROW>
<ROW>
<CELL COL="1">
<P>Control of infectious diseases</P>
</CELL>
<CELL COL="2">
<P>Michael Angelo BORG</P>
</CELL>
</ROW>
</CORPUS>
</TBL>""",
parser=xml_parser)
# fmt: on
for elem in table_elem.xpath("//*"):
elem.text = elem.text or None
for elem in expected.xpath("//*"):
elem.text = elem.text or None
diff_list = xmldiff.main.diff_trees(table_elem, expected)
if diff_list:
print(etree.tounicode(table_elem, pretty_print=True, with_tail=False),
file=sys.stderr)
assert diff_list == []
def setup_table(self, styles=None, nature=None):
table = Table(styles=styles, nature=nature)
self._state.table = table
return self._state
def test_build_tbl__orient(orient, size, expected):
builder = FormexBuilder()
table = Table(styles={'x-sect-orient': orient, 'x-sect-size': size})
table.rows[1].insert_cell(u"text")
table_elem = builder.build_tbl(table)
assert table_elem.attrib == expected
def test_build_tbl__with_title():
# see: formex-4/samples/jo-compl-2002C_280/C_2002280EN.01000101.xml
table = Table()
table.rows[1].insert_cell([P(u"1 euro =")],
width=3,
styles={"align": "center"})
table.rows[2].nature = "header"
table.rows[2].insert_cell([P()], styles={"x-cell-empty": "true"})
table.rows[2].insert_cell([P(u"Currency")])
table.rows[2].insert_cell([P(u"Exchange rate")])
table.rows[3].insert_cell([P(u"USD")])
table.rows[3].insert_cell([P(u"US dollar")])
table.rows[3].insert_cell([P(u"1,0029")])
table.rows[4].insert_cell([P(u"JPY")])
table.rows[4].insert_cell([P(u"Japanese yen")])
table.rows[4].insert_cell([P(u"121,05")])
builder = FormexBuilder(detect_titles=True)
table_elem = builder.build_tbl(table)
xml_parser = etree.XMLParser(remove_blank_text=True)
# fmt: off
expected = etree.XML(u"""\
<TBL COLS="3" NO.SEQ="0001">
<TITLE>
<TI>
<P>1 euro =</P>
</TI>
<STI/>
</TITLE>
<CORPUS>
<ROW TYPE="HEADER">
<CELL COL="1">
<IE/>
</CELL>
<CELL COL="2">
<P>Currency</P>
</CELL>
<CELL COL="3">
<P>Exchange rate</P>
</CELL>
</ROW>
<ROW>
<CELL COL="1">
<P>USD</P>
</CELL>
<CELL COL="2">
<P>US dollar</P>
</CELL>
<CELL COL="3">
<P>1,0029</P>
</CELL>
</ROW>
<ROW>
<CELL COL="1">
<P>JPY</P>
</CELL>
<CELL COL="2">
<P>Japanese yen</P>
</CELL>
<CELL COL="3">
<P>121,05</P>
</CELL>
</ROW>
</CORPUS>
</TBL>""",
parser=xml_parser)
# fmt: on
for elem in table_elem.xpath("//*"):
elem.text = elem.text or None
for elem in expected.xpath("//*"):
elem.text = elem.text or None
diff_list = xmldiff.main.diff_trees(table_elem, expected)
if diff_list:
print(etree.tounicode(table_elem, pretty_print=True, with_tail=False),
file=sys.stderr)
assert diff_list == []
def test_build_tbl():
# see: formex-4/samples/jo-compl-2002C_061/C_2002061EN.01000403.xml
table = Table()
table.rows[1].nature = "header"
table.rows[1].insert_cell([P(u"Expert group")])
table.rows[1].insert_cell([P(u"First name and surname of the expert")])
table.rows[2].insert_cell([P(u"Control of infectious diseases")])
table.rows[2].insert_cell([P(u"Michael Angelo BORG")])
table.rows[3].insert_cell([P(u"Information society")], height=3)
table.rows[3].insert_cell([P(u"Tony HEY")])
table.rows[4].insert_cell([P(u"José L. ENCARNAÇÃO")])
table.rows[5].insert_cell([P(u"Berit SVENDSEN")])
table.rows[6].insert_cell([P(u"Controlled thermonuclear fusion")])
table.rows[6].insert_cell([P(u"Pekka PIRILÄ")])
builder = FormexBuilder()
table_elem = builder.build_tbl(table)
xml_parser = etree.XMLParser(remove_blank_text=True)
# fmt: off
expected = etree.XML(u"""\
<TBL COLS="2" NO.SEQ="0001">
<CORPUS>
<ROW TYPE="HEADER">
<CELL COL="1">
<P>Expert group</P>
</CELL>
<CELL COL="2">
<P>First name and surname of the expert</P>
</CELL>
</ROW>
<ROW>
<CELL COL="1">
<P>Control of infectious diseases</P>
</CELL>
<CELL COL="2">
<P>Michael Angelo BORG</P>
</CELL>
</ROW>
<ROW>
<CELL COL="1" ROWSPAN="3">
<P>Information society</P>
</CELL>
<CELL COL="2">
<P>Tony HEY</P>
</CELL>
</ROW>
<ROW>
<CELL COL="2">
<P>José L. ENCARNAÇÃO</P>
</CELL>
</ROW>
<ROW>
<CELL COL="2">
<P>Berit SVENDSEN</P>
</CELL>
</ROW>
<ROW>
<CELL COL="1">
<P>Controlled thermonuclear fusion</P>
</CELL>
<CELL COL="2">
<P>Pekka PIRILÄ</P>
</CELL>
</ROW>
</CORPUS>
</TBL>""",
parser=xml_parser)
# fmt: on
for elem in table_elem.xpath("//*"):
elem.text = elem.text or None
for elem in expected.xpath("//*"):
elem.text = elem.text or None
diff_list = xmldiff.main.diff_trees(table_elem, expected)
if diff_list:
print(etree.tounicode(table_elem, pretty_print=True, with_tail=False),
file=sys.stderr)
assert diff_list == []
def test_build_table():
# see: formex-4/samples/jo-compl-2002C_061/C_2002061EN.01000403.xml
table = Table(
styles={
"border-top": "solid",
"border-bottom": "solid",
"x-sect-orient": "landscape",
"x-sect-cols": "1",
"background-color": "yellow",
"width": "247",
})
table.rows[1].nature = "header"
table.rows[1].insert_cell(u"Expert group")
table.rows[1].insert_cell(u"First name and surname of the expert")
table.rows[2].insert_cell(u"Control of infectious diseases")
table.rows[2].insert_cell(u"Michael Angelo BORG")
table.rows[3].insert_cell(u"Information society", height=3)
table.rows[3].insert_cell(u"Tony HEY")
table.rows[4].insert_cell(u"José L. ENCARNAÇÃO")
table.rows[5].insert_cell(u"Berit SVENDSEN")
table.rows[6].insert_cell(u"Controlled thermonuclear fusion")
table.rows[6].insert_cell(u"Pekka PIRILÄ")
builder = CalsBuilder()
table_elem = builder.build_table(table)
xml_parser = etree.XMLParser(remove_blank_text=True)
# fmt: off
expected = etree.XML(u"""\
<table frame="topbot" colsep="0" rowsep="0" orient="land" pgwide="1" bgcolor="yellow" width="247.00mm">
<tgroup cols="2">
<colspec colnum="1" colname="c1"/>
<colspec colnum="2" colname="c2"/>
<thead>
<row>
<entry>Expert group</entry>
<entry>First name and surname of the expert</entry>
</row>
</thead>
<tbody>
<row>
<entry>Control of infectious diseases</entry>
<entry>Michael Angelo BORG</entry>
</row>
<row>
<entry morerows="2">Information society</entry>
<entry>Tony HEY</entry>
</row>
<row>
<entry>José L. ENCARNAÇÃO</entry>
</row>
<row>
<entry>Berit SVENDSEN</entry>
</row>
<row>
<entry>Controlled thermonuclear fusion</entry>
<entry>Pekka PIRILÄ</entry>
</row>
</tbody>
</tgroup>
</table>""",
parser=xml_parser)
# fmt: on
for elem in table_elem.xpath("//*"):
elem.text = elem.text or None
for elem in expected.xpath("//*"):
elem.text = elem.text or None
diff_list = xmldiff.main.diff_trees(table_elem, expected)
if diff_list:
print(etree.tounicode(table_elem, pretty_print=True, with_tail=False),
file=sys.stderr)
assert diff_list == []
def test_setup_table():
builder = CalsBuilder()
table = Table()
result = builder.setup_table(table)
assert result == table