I'm looking for a way to pretty-print tables like this:
=======================
| column 1 | column 2 |
=======================
| value1 | value2 |
| value3 | value4 |
=======================
I've found the asciitable library but it doesn't do the borders, etc. I don't need any complex formatting of data items, they're just strings. I do need it to auto-size columns.
Do other libraries or methods exist, or do I need to spend a few minutes writing my own?
I've read this question long time ago, and finished writing my own pretty-printer for tables: tabulate.
My use case is:
Given your example, grid is probably the most similar output format:
from tabulate import tabulate
print tabulate([["value1", "value2"], ["value3", "value4"]], ["column 1", "column 2"], tablefmt="grid")
+------------+------------+
| column 1 | column 2 |
+============+============+
| value1 | value2 |
+------------+------------+
| value3 | value4 |
+------------+------------+
Other supported formats are plain (no lines), simple (Pandoc simple tables), pipe (like tables in PHP Markdown Extra), orgtbl (like tables in Emacs' org-mode), rst (like simple tables in reStructuredText). grid and orgtbl are easily editable in Emacs.
Performance-wise, tabulate is slightly slower than asciitable, but much faster than PrettyTable and texttable.
P.S. I'm also a big fan of aligning numbers by a decimal column. So this is the default alignment for numbers if there are any (overridable).
Here's a quick and dirty little function I wrote for displaying the results from SQL queries I can only make over a SOAP API. It expects an input of a sequence of one or more namedtuples as table rows. If there's only one record, it prints it out differently.
It is handy for me and could be a starting point for you:
def pprinttable(rows):
if len(rows) > 1:
headers = rows[0]._fields
lens = []
for i in range(len(rows[0])):
lens.append(len(max([x[i] for x in rows] + [headers[i]],key=lambda x:len(str(x)))))
formats = []
hformats = []
for i in range(len(rows[0])):
if isinstance(rows[0][i], int):
formats.append("%%%dd" % lens[i])
else:
formats.append("%%-%ds" % lens[i])
hformats.append("%%-%ds" % lens[i])
pattern = " | ".join(formats)
hpattern = " | ".join(hformats)
separator = "-+-".join(['-' * n for n in lens])
print hpattern % tuple(headers)
print separator
_u = lambda t: t.decode('UTF-8', 'replace') if isinstance(t, str) else t
for line in rows:
print pattern % tuple(_u(t) for t in line)
elif len(rows) == 1:
row = rows[0]
hwidth = len(max(row._fields,key=lambda x: len(x)))
for i in range(len(row)):
print "%*s = %s" % (hwidth,row._fields[i],row[i])
Sample output:
pkid | fkn | npi -------------------------------------+--------------------------------------+---- 405fd665-0a2f-4f69-7320-be01201752ec | 8c9949b9-552e-e448-64e2-74292834c73e | 0 5b517507-2a42-ad2e-98dc-8c9ac6152afa | f972bee7-f5a4-8532-c4e5-2e82897b10f6 | 0 2f960dfc-b67a-26be-d1b3-9b105535e0a8 | ec3e1058-8840-c9f2-3b25-2488f8b3a8af | 1 c71b28a3-5299-7f4d-f27a-7ad8aeadafe0 | 72d25703-4735-310b-2e06-ff76af1e45ed | 0 3b0a5021-a52b-9ba0-1439-d5aafcf348e7 | d81bb78a-d984-e957-034d-87434acb4e97 | 1 96c36bb7-c4f4-2787-ada8-4aadc17d1123 | c171fe85-33e2-6481-0791-2922267e8777 | 1 95d0f85f-71da-bb9a-2d80-fe27f7c02fe2 | 226f964c-028d-d6de-bf6c-688d2908c5ae | 1 132aa774-42e5-3d3f-498b-50b44a89d401 | 44e31f89-d089-8afc-f4b1-ada051c01474 | 1 ff91641a-5802-be02-bece-79bca993fdbc | 33d8294a-053d-6ab4-94d4-890b47fcf70d | 1 f3196e15-5b61-e92d-e717-f00ed93fe8ae | 62fa4566-5ca2-4a36-f872-4d00f7abadcf | 1
Example
>>> from collections import namedtuple
>>> Row = namedtuple('Row',['first','second','third'])
>>> data = Row(1,2,3)
>>> data
Row(first=1, second=2, third=3)
>>> pprinttable([data])
first = 1
second = 2
third = 3
>>> pprinttable([data,data])
first | second | third
------+--------+------
1 | 2 | 3
1 | 2 | 3
For some reason when I included 'docutils' in my google searches I stumbled across texttable, which seems to be what I'm looking for.
I too wrote my own solution to this. I tried to keep it simple.
https://github.com/Robpol86/terminaltables
from terminaltables import AsciiTable
table_data = [
['Heading1', 'Heading2'],
['row1 column1', 'row1 column2'],
['row2 column1', 'row2 column2']
]
table = AsciiTable(table_data)
print table.table
+--------------+--------------+
| Heading1 | Heading2 |
+--------------+--------------+
| row1 column1 | row1 column2 |
| row2 column1 | row2 column2 |
+--------------+--------------+
table.inner_heading_row_border = False
print table.table
+--------------+--------------+
| Heading1 | Heading2 |
| row1 column1 | row1 column2 |
| row2 column1 | row2 column2 |
+--------------+--------------+
table.inner_row_border = True
table.justify_columns[1] = 'right'
table.table_data[1][1] += '\nnewline'
print table.table
+--------------+--------------+
| Heading1 | Heading2 |
+--------------+--------------+
| row1 column1 | row1 column2 |
| | newline |
+--------------+--------------+
| row2 column1 | row2 column2 |
+--------------+--------------+
I just released termtables for this purpose. For example, this
import termtables as tt
tt.print(
[[1, 2, 3], [613.23236243236, 613.23236243236, 613.23236243236]],
header=["a", "bb", "ccc"],
style=tt.styles.ascii_thin_double,
padding=(0, 1),
alignment="lcr"
)
gets you
+-----------------+-----------------+-----------------+
| a | bb | ccc |
+=================+=================+=================+
| 1 | 2 | 3 |
+-----------------+-----------------+-----------------+
| 613.23236243236 | 613.23236243236 | 613.23236243236 |
+-----------------+-----------------+-----------------+
By default, the table is rendered with Unicode box-drawing characters,
┌─────────────────┬─────────────────┬─────────────────┐
│ a │ bb │ ccc │
╞═════════════════╪═════════════════╪═════════════════╡
│ 1 │ 2 │ 3 │
├─────────────────┼─────────────────┼─────────────────┤
│ 613.23236243236 │ 613.23236243236 │ 613.23236243236 │
└─────────────────┴─────────────────┴─────────────────┘
termtables are very configurable; check out the tests for more examples.
If you want a table with column and row spans, then try my library dashtable
from dashtable import data2rst
table = [
["Header 1", "Header 2", "Header3", "Header 4"],
["row 1", "column 2", "column 3", "column 4"],
["row 2", "Cells span columns.", "", ""],
["row 3", "Cells\nspan rows.", "- Cells\n- contain\n- blocks", ""],
["row 4", "", "", ""]
]
# [Row, Column] pairs of merged cells
span0 = ([2, 1], [2, 2], [2, 3])
span1 = ([3, 1], [4, 1])
span2 = ([3, 3], [3, 2], [4, 2], [4, 3])
my_spans = [span0, span1, span2]
print(data2rst(table, spans=my_spans, use_headers=True))
Which outputs:
+----------+------------+----------+----------+
| Header 1 | Header 2 | Header3 | Header 4 |
+==========+============+==========+==========+
| row 1 | column 2 | column 3 | column 4 |
+----------+------------+----------+----------+
| row 2 | Cells span columns. |
+----------+----------------------------------+
| row 3 | Cells | - Cells |
+----------+ span rows. | - contain |
| row 4 | | - blocks |
+----------+------------+---------------------+
Version using w3m designed to handle the types MattH's version accepts:
import subprocess
import tempfile
import html
def pprinttable(rows):
esc = lambda x: html.escape(str(x))
sour = "<table border=1>"
if len(rows) == 1:
for i in range(len(rows[0]._fields)):
sour += "<tr><th>%s<td>%s" % (esc(rows[0]._fields[i]), esc(rows[0][i]))
else:
sour += "<tr>" + "".join(["<th>%s" % esc(x) for x in rows[0]._fields])
sour += "".join(["<tr>%s" % "".join(["<td>%s" % esc(y) for y in x]) for x in rows])
with tempfile.NamedTemporaryFile(suffix=".html") as f:
f.write(sour.encode("utf-8"))
f.flush()
print(
subprocess
.Popen(["w3m","-dump",f.name], stdout=subprocess.PIPE)
.communicate()[0].decode("utf-8").strip()
)
from collections import namedtuple
Row = namedtuple('Row',['first','second','third'])
data1 = Row(1,2,3)
data2 = Row(4,5,6)
pprinttable([data1])
pprinttable([data1,data2])
results in:
┌───────┬─┐
│ first │1│
├───────┼─┤
│second │2│
├───────┼─┤
│ third │3│
└───────┴─┘
┌─────┬───────┬─────┐
│first│second │third│
├─────┼───────┼─────┤
│1 │2 │3 │
├─────┼───────┼─────┤
│4 │5 │6 │
└─────┴───────┴─────┘
You can try BeautifulTable. It does what you want to do. Here's an example from it's documentation
>>> from beautifultable import BeautifulTable
>>> table = BeautifulTable()
>>> table.columns.header = ["name", "rank", "gender"]
>>> table.rows.append(["Jacob", 1, "boy"])
>>> table.rows.append(["Isabella", 1, "girl"])
>>> table.rows.append(["Ethan", 2, "boy"])
>>> table.rows.append(["Sophia", 2, "girl"])
>>> table.rows.append(["Michael", 3, "boy"])
>>> print(table)
+----------+------+--------+
| name | rank | gender |
+----------+------+--------+
| Jacob | 1 | boy |
+----------+------+--------+
| Isabella | 1 | girl |
+----------+------+--------+
| Ethan | 2 | boy |
+----------+------+--------+
| Sophia | 2 | girl |
+----------+------+--------+
| Michael | 3 | boy |
+----------+------+--------+
from sys import stderr, stdout
def create_table(table: dict, full_row: bool = False) -> None:
min_len = len(min((v for v in table.values()), key=lambda q: len(q)))
max_len = len(max((v for v in table.values()), key=lambda q: len(q)))
if min_len < max_len:
stderr.write("Table is out of shape, please make sure all columns have the same length.")
stderr.flush()
return
additional_spacing = 1
heading_separator = '| '
horizontal_split = '| '
rc_separator = ''
key_list = list(table.keys())
rc_len_values = []
for key in key_list:
rc_len = len(max((v for v in table[key]), key=lambda q: len(str(q))))
rc_len_values += ([rc_len, [key]] for n in range(len(table[key])))
heading_line = (key + (" " * (rc_len + (additional_spacing + 1)))) + heading_separator
stdout.write(heading_line)
rc_separator += ("-" * (len(key) + (rc_len + (additional_spacing + 1)))) + '+-'
if key is key_list[-1]:
stdout.flush()
stdout.write('\n' + rc_separator + '\n')
value_list = [v for vl in table.values() for v in vl]
aligned_data_offset = max_len
row_count = len(key_list)
next_idx = 0
newline_indicator = 0
iterations = 0
for n in range(len(value_list)):
key = rc_len_values[next_idx][1][0]
rc_len = rc_len_values[next_idx][0]
line = ('{:{}} ' + " " * len(key)).format(value_list[next_idx], str(rc_len + additional_spacing)) + horizontal_split
if next_idx >= (len(value_list) - aligned_data_offset):
next_idx = iterations + 1
iterations += 1
else:
next_idx += aligned_data_offset
if newline_indicator >= row_count:
if full_row:
stdout.flush()
stdout.write('\n' + rc_separator + '\n')
else:
stdout.flush()
stdout.write('\n')
newline_indicator = 0
stdout.write(line)
newline_indicator += 1
stdout.write('\n' + rc_separator + '\n')
stdout.flush()
Example:
table = {
"uid": ["0", "1", "2", "3"],
"name": ["Jon", "Doe", "Lemma", "Hemma"]
}
create_table(table)
Output:
uid | name |
------+------------+-
0 | Jon |
1 | Doe |
2 | Lemma |
3 | Hemma |
------+------------+-
I know it the question is a bit old but here's my attempt at this:
https://gist.github.com/lonetwin/4721748
It is a bit more readable IMHO (although it doesn't differentiate between single / multiple rows like @MattH's solutions does, nor does it use NamedTuples).
Here's my solution:
def make_table(columns, data):
"""Create an ASCII table and return it as a string.
Pass a list of strings to use as columns in the table and a list of
dicts. The strings in 'columns' will be used as the keys to the dicts in
'data.'
Not all column values have to be present in each data dict.
>>> print(make_table(["a", "b"], [{"a": "1", "b": "test"}]))
| a | b |
|----------|
| 1 | test |
"""
# Calculate how wide each cell needs to be
cell_widths = {}
for c in columns:
values = [str(d.get(c, "")) for d in data]
cell_widths[c] = len(max(values + [c]))
# Used for formatting rows of data
row_template = "|" + " {} |" * len(columns)
# CONSTRUCT THE TABLE
# The top row with the column titles
justified_column_heads = [c.ljust(cell_widths[c]) for c in columns]
header = row_template.format(*justified_column_heads)
# The second row contains separators
sep = "|" + "-" * (len(header) - 2) + "|"
# Rows of data
rows = []
for d in data:
fields = [str(d.get(c, "")).ljust(cell_widths[c]) for c in columns]
row = row_template.format(*fields)
rows.append(row)
return "\n".join([header, sep] + rows)
I use this small utility function.
def get_pretty_table(iterable, header):
max_len = [len(x) for x in header]
for row in iterable:
row = [row] if type(row) not in (list, tuple) else row
for index, col in enumerate(row):
if max_len[index] < len(str(col)):
max_len[index] = len(str(col))
output = '-' * (sum(max_len) + 1) + '\n'
output += '|' + ''.join([h + ' ' * (l - len(h)) + '|' for h, l in zip(header, max_len)]) + '\n'
output += '-' * (sum(max_len) + 1) + '\n'
for row in iterable:
row = [row] if type(row) not in (list, tuple) else row
output += '|' + ''.join([str(c) + ' ' * (l - len(str(c))) + '|' for c, l in zip(row, max_len)]) + '\n'
output += '-' * (sum(max_len) + 1) + '\n'
return output
print get_pretty_table([[1, 2], [3, 4]], ['header 1', 'header 2'])
output
-----------------
|header 1|header 2|
-----------------
|1 |2 |
|3 |4 |
-----------------
This can be done with only builtin modules fairly compactly using list and string comprehensions. Accepts a list of dictionaries all of the same format...
def tableit(dictlist):
lengths = [ max(map(lambda x:len(x.get(k)), dictlist) + [len(k)]) for k in dictlist[0].keys() ]
lenstr = " | ".join("{:<%s}" % m for m in lengths)
lenstr += "\n"
outmsg = lenstr.format(*dictlist[0].keys())
outmsg += "-" * (sum(lengths) + 3*len(lengths))
outmsg += "\n"
outmsg += "".join(
lenstr.format(*v) for v in [ item.values() for item in dictlist ]
)
return outmsg
