diff options
| -rw-r--r-- | mammoth/documents.py | 7 | ||||
| -rw-r--r-- | mammoth/docx/body_xml.py | 92 | ||||
| -rw-r--r-- | tests/docx/body_xml_tests.py | 95 |
3 files changed, 183 insertions, 11 deletions
diff --git a/mammoth/documents.py b/mammoth/documents.py index 7f17f3e..36e3458 100644 --- a/mammoth/documents.py +++ b/mammoth/documents.py @@ -49,6 +49,7 @@ class TableRow(HasChildren): @cobble.data class TableCell(HasChildren): colspan = cobble.field() + rowspan = cobble.field() @cobble.data class LineBreak(Element): @@ -111,10 +112,12 @@ bookmark = Bookmark table = Table table_row = TableRow -def table_cell(children, colspan=None): +def table_cell(children, colspan=None, rowspan=None): if colspan is None: colspan = 1 - return TableCell(children=children, colspan=colspan) + if rowspan is None: + rowspan = 1 + return TableCell(children=children, colspan=colspan, rowspan=rowspan) line_break = LineBreak diff --git a/mammoth/docx/body_xml.py b/mammoth/docx/body_xml.py index a44169a..8aacc1c 100644 --- a/mammoth/docx/body_xml.py +++ b/mammoth/docx/body_xml.py @@ -155,6 +155,7 @@ def _create_reader(numbering, content_types, relationships, styles, docx_file, f def table(element): return _read_xml_elements(element.children) \ + .flat_map(calculate_row_spans) \ .map(documents.table) @@ -164,8 +165,8 @@ def _create_reader(numbering, content_types, relationships, styles, docx_file, f def table_cell(element): - gridspan = element \ - .find_child_or_null("w:tcPr") \ + properties = element.find_child_or_null("w:tcPr") + gridspan = properties \ .find_child_or_null("w:gridSpan") \ .attributes.get("w:val") @@ -175,11 +176,61 @@ def _create_reader(numbering, content_types, relationships, styles, docx_file, f colspan = int(gridspan) return _read_xml_elements(element.children) \ - .map(lambda children: documents.table_cell( - children=children, - colspan=colspan + .map(lambda children: _add_attrs( + documents.table_cell( + children=children, + colspan=colspan + ), + _vmerge=read_vmerge(properties), )) + def read_vmerge(properties): + vmerge_element = properties.find_child("w:vMerge") + if vmerge_element is None: + return False + else: + val = vmerge_element.attributes.get("w:val") + return val == "continue" or not val + + + def calculate_row_spans(rows): + unexpected_non_rows = any( + not isinstance(row, documents.TableRow) + for row in rows + ) + if unexpected_non_rows: + return _elements_result_with_messages(rows, [results.warning( + "unexpected non-row element in table, cell merging may be incorrect" + )]) + + unexpected_non_cells = any( + not isinstance(cell, documents.TableCell) + for row in rows + for cell in row.children + ) + if unexpected_non_cells: + return _elements_result_with_messages(rows, [results.warning( + "unexpected non-cell element in table row, cell merging may be incorrect" + )]) + + columns = {} + for row in rows: + cell_index = 0 + for cell in row.children: + if cell._vmerge and cell_index in columns: + columns[cell_index].rowspan += 1 + else: + columns[cell_index] = cell + cell._vmerge = False + cell_index += cell.colspan + + for row in rows: + row.children = lists.filter(lambda cell: not cell._vmerge, row.children) + for cell in row.children: + del cell._vmerge + + return _success(rows) + def read_child_elements(element): return _read_xml_elements(element.children) @@ -362,22 +413,38 @@ class _ReadResult(object): self.messages = messages def map(self, func): + elements = func(self.elements) + if not isinstance(elements, list): + elements = [elements] return _ReadResult( - [func(self.elements)], + elements, self.extra, self.messages) + def flat_map(self, func): + result = func(self.elements) + return _ReadResult( + result.elements, + self.extra + result.extra, + self.messages + result.messages) + + def to_extra(self): return _ReadResult([], _concat(self.extra, self.elements), self.messages) def append_extra(self): return _ReadResult(_concat(self.elements, self.extra), [], self.messages) -def _success(element): - return _ReadResult([element], [], []) +def _success(elements): + if not isinstance(elements, list): + elements = [elements] + return _ReadResult(elements, [], []) def _element_result_with_messages(element, messages): - return _ReadResult([element], [], messages) + return _elements_result_with_messages([element], messages) + +def _elements_result_with_messages(elements, messages): + return _ReadResult(elements, [], messages) _empty_result = _ReadResult([], [], []) @@ -390,3 +457,10 @@ def _concat(*values): for element in value: result.append(element) return result + + +def _add_attrs(obj, **kwargs): + for key, value in kwargs.items(): + setattr(obj, key, value) + + return obj diff --git a/tests/docx/body_xml_tests.py b/tests/docx/body_xml_tests.py index 8fb35cf..2773398 100644 --- a/tests/docx/body_xml_tests.py +++ b/tests/docx/body_xml_tests.py @@ -247,6 +247,88 @@ class ReadXmlElementTests(object): @istest + def vmerge_is_read_as_rowspan_for_table_cell(self): + element = xml_element("w:tbl", {}, [ + w_tr(w_tc()), + w_tr(w_tc(properties=[w_vmerge("restart")])), + w_tr(w_tc(properties=[w_vmerge("continue")])), + w_tr(w_tc(properties=[w_vmerge("continue")])), + w_tr(w_tc()) + ]) + result = _read_and_get_document_xml_element(element) + expected_result = documents.table([ + documents.table_row([documents.table_cell([])]), + documents.table_row([documents.table_cell([], rowspan=3)]), + documents.table_row([]), + documents.table_row([]), + documents.table_row([documents.table_cell([])]), + ]) + assert_equal(expected_result, result) + + + @istest + def vmerge_without_val_is_treated_as_continue(self): + element = xml_element("w:tbl", {}, [ + w_tr(w_tc(properties=[w_vmerge("restart")])), + w_tr(w_tc(properties=[w_vmerge(None)])), + ]) + result = _read_and_get_document_xml_element(element) + expected_result = documents.table([ + documents.table_row([documents.table_cell([], rowspan=2)]), + documents.table_row([]), + ]) + assert_equal(expected_result, result) + + + @istest + def vmerge_accounts_for_cells_spanning_columns(self): + element = xml_element("w:tbl", {}, [ + w_tr(w_tc(), w_tc(), w_tc(properties=[w_vmerge("restart")])), + w_tr(w_tc(properties=[w_gridspan("2")]), w_tc(properties=[w_vmerge("continue")])), + w_tr(w_tc(), w_tc(), w_tc(properties=[w_vmerge("continue")])), + w_tr(w_tc(), w_tc(), w_tc()), + ]) + result = _read_and_get_document_xml_element(element) + expected_result = documents.table([ + documents.table_row([documents.table_cell([]), documents.table_cell([]), documents.table_cell([], rowspan=3)]), + documents.table_row([documents.table_cell([], colspan=2)]), + documents.table_row([documents.table_cell([]), documents.table_cell([])]), + documents.table_row([documents.table_cell([]), documents.table_cell([]), documents.table_cell([])]), + ]) + assert_equal(expected_result, result) + + + @istest + def no_vertical_cell_merging_if_merged_cells_do_not_line_up(self): + element = xml_element("w:tbl", {}, [ + w_tr(w_tc(properties=[w_gridspan("2")]), w_tc(properties=[w_vmerge("restart")])), + w_tr(w_tc(), w_tc(properties=[w_vmerge("continue")])), + ]) + result = _read_and_get_document_xml_element(element) + expected_result = documents.table([ + documents.table_row([documents.table_cell([], colspan=2), documents.table_cell([])]), + documents.table_row([documents.table_cell([]), documents.table_cell([])]), + ]) + assert_equal(expected_result, result) + + + @istest + def warning_if_non_row_in_table(self): + element = xml_element("w:tbl", {}, [xml_element("w:p")]) + result = _read_document_xml_element(element) + expected_warning = results.warning("unexpected non-row element in table, cell merging may be incorrect") + assert_equal([expected_warning], result.messages) + + + @istest + def warning_if_non_cell_in_table_row(self): + element = xml_element("w:tbl", {}, [w_tr(xml_element("w:p"))]) + result = _read_document_xml_element(element) + expected_warning = results.warning("unexpected non-cell element in table row, cell merging may be incorrect") + assert_equal([expected_warning], result.messages) + + + @istest def children_of_w_ins_are_converted_normally(self): element = xml_element("w:p", {}, [ xml_element("w:ins", {}, [ @@ -662,3 +744,16 @@ def _linked_blip(relationship_id): def _blip(attributes): return xml_element("a:blip", attributes) + + +def w_tr(*children): + return xml_element("w:tr", {}, children) + +def w_tc(properties=None, *children): + return xml_element("w:tc", {}, [xml_element("w:tcPr", {}, properties)] + list(children)) + +def w_gridspan(val): + return xml_element("w:gridSpan", {"w:val": val}) + +def w_vmerge(val): + return xml_element("w:vMerge", {"w:val": val}) |