(html_table)
| 71 | return "".join(html_table) |
| 72 | |
| 73 | def table_html2matrix(html_table): |
| 74 | soup = BeautifulSoup(html_table, 'html.parser') |
| 75 | table = soup.find('table') |
| 76 | rownum = len(table.find_all('tr')) |
| 77 | colnum = 0 |
| 78 | tr = table.find_all('tr')[0] |
| 79 | for td in tr.find_all('td'): |
| 80 | colnum += td.get('colspan', 1) |
| 81 | matrix = [[None for _ in range(colnum)] for _ in range(rownum)] |
| 82 | |
| 83 | rid = 0 |
| 84 | for tr in table.find_all('tr'): |
| 85 | cid = 0 |
| 86 | for td in tr.find_all('td'): |
| 87 | for c in range(cid, colnum): |
| 88 | if matrix[rid][c] is None: |
| 89 | break |
| 90 | cid = c |
| 91 | rowspan = td.get('rowspan', 1) |
| 92 | colspan = td.get('colspan', 1) |
| 93 | cell_text = td.get_text(strip=True) |
| 94 | for r in range(rid,rid+rowspan): |
| 95 | if r >= rownum: |
| 96 | raise Exception('rownum not match') |
| 97 | for c in range(cid,cid+colspan): |
| 98 | if c >= colnum: |
| 99 | raise Exception('colnum not match') |
| 100 | if matrix[r][c] is not None: |
| 101 | raise Exception('cell not match') |
| 102 | if r == rid and c == cid: |
| 103 | matrix[r][c] = cell_text |
| 104 | elif r == rid: |
| 105 | matrix[r][c] = '<l>' |
| 106 | elif c == cid: |
| 107 | matrix[r][c] = '<t>' |
| 108 | else: |
| 109 | matrix[r][c] = '<lt>' |
| 110 | cid += colspan |
| 111 | rid += 1 |
| 112 | |
| 113 | matrix_table = ['<table>'] |
| 114 | for rid in range(rownum): |
| 115 | matrix_table.append('<tr>') |
| 116 | for cid in range(colnum): |
| 117 | matrix_table.append('<td>') |
| 118 | cell_text = matrix[rid][cid] |
| 119 | matrix_table.append(cell_text) |
| 120 | matrix_table.append('</td>') |
| 121 | matrix_table.append('</tr>') |
| 122 | matrix_table.append('</table>') |
| 123 | return "".join(matrix_table) |
| 124 | |
| 125 | trans_func = { |
| 126 | "html2matrix": table_html2matrix, |
nothing calls this directly
no outgoing calls
no test coverage detected