Fixed-Width Gallery#
Fixed-width tables are those where each column has the same width for every row in the table. This is commonly used to make tables easy to read for humans or Fortran codes. It also reduces issues with quoting and special characters, for example:
Col1 Col2 Col3 Col4
---- --------- ---- ----
1.2 "hello" 1 a
2.4 's worlds 2 2
There are a number of common variations in the formatting of fixed-width tables
which astropy.io.ascii can read and write. The most significant
difference is whether there is no header line (FixedWidthNoHeader), one
header line (FixedWidth), or two header lines
(FixedWidthTwoLine). Next, there are variations in
the delimiter character, like whether the delimiter appears on either end
(“bookends”), or if there is padding around the delimiter.
Details are available in the class API documentation, but the easiest way to understand all of the options and their interactions is by example.
Reading#
Fixed Width#
Nice, typical, fixed-format table:
>>> from astropy.io import ascii
>>> table = """
... # comment (with blank line above)
... | Col1 | Col2 |
... | 1.2 | "hello" |
... | 2.4 |'s worlds|
... """
>>> ascii.read(table, format='fixed_width')
<Table length=2>
Col1 Col2
float64 str9
------- ---------
1.2 "hello"
2.4 's worlds
Typical fixed-format table with col names provided:
>>> table = """
... # comment (with blank line above)
... | Col1 | Col2 |
... | 1.2 | "hello" |
... | 2.4 |'s worlds|
... """
>>> ascii.read(table, format='fixed_width', names=['name1', 'name2'])
<Table length=2>
name1 name2
float64 str9
------- ---------
1.2 "hello"
2.4 's worlds
Weird input table with data values chopped by col extent:
>>> table = """
... Col1 | Col2 |
... 1.2 "hello"
... 2.4 sdf's worlds
... """
>>> ascii.read(table, format='fixed_width')
<Table length=2>
Col1 Col2
float64 str7
------- -------
1.2 "hel
2.4 df's wo
Table with double delimiters:
>>> table = """
... || Name || Phone || TCP||
... | John | 555-1234 |192.168.1.10X|
... | Mary | 555-2134 |192.168.1.12X|
... | Bob | 555-4527 | 192.168.1.9X|
... """
>>> ascii.read(table, format='fixed_width')
<Table length=3>
Name Phone TCP
str4 str8 str12
---- -------- ------------
John 555-1234 192.168.1.10
Mary 555-2134 192.168.1.12
Bob 555-4527 192.168.1.9
Table with space delimiter:
>>> table = """
... Name --Phone- ----TCP-----
... John 555-1234 192.168.1.10
... Mary 555-2134 192.168.1.12
... Bob 555-4527 192.168.1.9
... """
>>> ascii.read(table, format='fixed_width', delimiter=' ')
<Table length=3>
Name --Phone- ----TCP-----
str4 str8 str12
---- -------- ------------
John 555-1234 192.168.1.10
Mary 555-2134 192.168.1.12
Bob 555-4527 192.168.1.9
Table with no header row and auto-column naming:
Use header_start and data_start keywords to indicate no header line.
>>> table = """
... | John | 555-1234 |192.168.1.10|
... | Mary | 555-2134 |192.168.1.12|
... | Bob | 555-4527 | 192.168.1.9|
... """
>>> ascii.read(table, format='fixed_width',
... header_start=None, data_start=0)
<Table length=3>
col1 col2 col3
str4 str8 str12
---- -------- ------------
John 555-1234 192.168.1.10
Mary 555-2134 192.168.1.12
Bob 555-4527 192.168.1.9
Table with no header row and with col names provided:
Second and third rows also have hanging spaces after final “|”. Use header_start and data_start keywords to indicate no header line.
>>> table = ["| John | 555-1234 |192.168.1.10|",
... "| Mary | 555-2134 |192.168.1.12| ",
... "| Bob | 555-4527 | 192.168.1.9| "]
>>> ascii.read(table, format='fixed_width',
... header_start=None, data_start=0,
... names=('Name', 'Phone', 'TCP'))
<Table length=3>
Name Phone TCP
str4 str8 str12
---- -------- ------------
John 555-1234 192.168.1.10
Mary 555-2134 192.168.1.12
Bob 555-4527 192.168.1.9
Fixed Width No Header#
Table with no header row and auto-column naming. Use the ``fixed_width_no_header`` format for convenience:
>>> table = """
... | John | 555-1234 |192.168.1.10|
... | Mary | 555-2134 |192.168.1.12|
... | Bob | 555-4527 | 192.168.1.9|
... """
>>> ascii.read(table, format='fixed_width_no_header')
<Table length=3>
col1 col2 col3
str4 str8 str12
---- -------- ------------
John 555-1234 192.168.1.10
Mary 555-2134 192.168.1.12
Bob 555-4527 192.168.1.9
Table with no delimiter with column start and end values specified:
This uses the col_starts and col_ends keywords. Note that the col_ends values are inclusive so a position range of zero to five will select the first six characters.
>>> table = """
... # 5 9 17 18 28 <== Column start / end indexes
... # | | || | <== Column separation positions
... John 555- 1234 192.168.1.10
... Mary 555- 2134 192.168.1.12
... Bob 555- 4527 192.168.1.9
... """
>>> ascii.read(table, format='fixed_width_no_header',
... names=('Name', 'Phone', 'TCP'),
... col_starts=(0, 9, 18),
... col_ends=(5, 17, 28),
... )
<Table length=3>
Name Phone TCP
str4 str9 str10
---- --------- ----------
John 555- 1234 192.168.1.
Mary 555- 2134 192.168.1.
Bob 555- 4527 192.168.1
Table with no delimiter with only column start or end values specified:
If only the col_starts keyword is given, it is assumed that each column ends where the next column starts, and the final column ends at the same position as the longest line of data.
Conversely, if only the col_ends keyword is given, it is assumed that the first column starts at position zero and that each successive column starts immediately after the previous one.
The two examples below read the same table and produce the same result.
>>> table = """
... #1 9 19 <== Column start indexes
... #| | | <== Column start positions
... #<------><--------><-------------> <== Inferred column positions
... John 555- 1234 192.168.1.10
... Mary 555- 2134 192.168.1.123
... Bob 555- 4527 192.168.1.9
... Bill 555-9875 192.255.255.255
... """
>>> ascii.read(table,
... format='fixed_width_no_header',
... names=('Name', 'Phone', 'TCP'),
... col_starts=(1, 9, 19),
... )
<Table length=4>
Name Phone TCP
str4 str9 str15
---- --------- ---------------
John 555- 1234 192.168.1.10
Mary 555- 2134 192.168.1.123
Bob 555- 4527 192.168.1.9
Bill 555-9875 192.255.255.255
>>> ascii.read(table,
... format='fixed_width_no_header',
... names=('Name', 'Phone', 'TCP'),
... col_ends=(8, 18, 32),
... )
<Table length=4>
Name Phone TCP
str4 str9 str14
---- --------- --------------
John 555- 1234 192.168.1.10
Mary 555- 2134 192.168.1.123
Bob 555- 4527 192.168.1.9
Bill 555-9875 192.255.255.25
Fixed Width Two Line#
Typical fixed-format table with two header lines with some cruft:
>>> table = """
... Col1 Col2
... ---- ---------
... 1.2xx"hello"
... 2.4 's worlds
... """
>>> ascii.read(table, format='fixed_width_two_line')
<Table length=2>
Col1 Col2
float64 str9
------- ---------
1.2 "hello"
2.4 's worlds
reStructuredText table:
>>> table = """
... ======= ===========
... Col1 Col2
... ======= ===========
... 1.2 "hello"
... 2.4 's worlds
... ======= ===========
... """
>>> ascii.read(table, format='fixed_width_two_line',
... header_start=1, position_line=2, data_end=-1)
<Table length=2>
Col1 Col2
float64 str9
------- ---------
1.2 "hello"
2.4 's worlds
Text table designed for humans and test having position line before the header line:
>>> table = """
... +------+----------+
... | Col1 | Col2 |
... +------|----------+
... | 1.2 | "hello" |
... | 2.4 | 's worlds|
... +------+----------+
... """
>>> ascii.read(table, format='fixed_width_two_line', delimiter='+',
... header_start=1, position_line=0, data_start=3, data_end=-1)
<Table length=2>
Col1 Col2
float64 str9
------- ---------
1.2 "hello"
2.4 's worlds
Writing#
Fixed Width#
Define input values ``dat`` for all write examples:
>>> table = """
... | Col1 | Col2 | Col3 | Col4 |
... | 1.2 | "hello" | 1 | a |
... | 2.4 | 's worlds | 2 | 2 |
... """
>>> dat = ascii.read(table, format='fixed_width')
Write a table as a normal fixed-width table:
>>> ascii.write(dat, format='fixed_width')
| Col1 | Col2 | Col3 | Col4 |
| 1.2 | "hello" | 1 | a |
| 2.4 | 's worlds | 2 | 2 |
Write a table as a fixed-width table with no padding:
>>> ascii.write(dat, format='fixed_width', delimiter_pad=None)
|Col1| Col2|Col3|Col4|
| 1.2| "hello"| 1| a|
| 2.4|'s worlds| 2| 2|
Write a table as a fixed-width table with no bookend:
>>> ascii.write(dat, format='fixed_width', bookend=False)
Col1 | Col2 | Col3 | Col4
1.2 | "hello" | 1 | a
2.4 | 's worlds | 2 | 2
Write a table as a fixed-width table with no delimiter:
>>> ascii.write(dat, format='fixed_width', bookend=False, delimiter=None)
Col1 Col2 Col3 Col4
1.2 "hello" 1 a
2.4 's worlds 2 2
Write a table as a fixed-width table with no delimiter and formatting:
>>> ascii.write(dat, format='fixed_width',
... formats={'Col1': '%-8.3f', 'Col2': '%-15s'})
| Col1 | Col2 | Col3 | Col4 |
| 1.200 | "hello" | 1 | a |
| 2.400 | 's worlds | 2 | 2 |
Fixed Width No Header#
Write a table as a normal fixed-width table:
>>> ascii.write(dat, format='fixed_width_no_header')
| 1.2 | "hello" | 1 | a |
| 2.4 | 's worlds | 2 | 2 |
Write a table as a fixed-width table with no padding:
>>> ascii.write(dat, format='fixed_width_no_header', delimiter_pad=None)
|1.2| "hello"|1|a|
|2.4|'s worlds|2|2|
Write a table as a fixed-width table with no bookend:
>>> ascii.write(dat, format='fixed_width_no_header', bookend=False)
1.2 | "hello" | 1 | a
2.4 | 's worlds | 2 | 2
Write a table as a fixed-width table with no delimiter:
>>> ascii.write(dat, format='fixed_width_no_header', bookend=False,
... delimiter=None)
1.2 "hello" 1 a
2.4 's worlds 2 2
Fixed Width Two Line#
Write a table as a normal fixed-width table:
>>> ascii.write(dat, format='fixed_width_two_line')
Col1 Col2 Col3 Col4
---- --------- ---- ----
1.2 "hello" 1 a
2.4 's worlds 2 2
Write a table as a fixed width table with space padding and ‘=’ position_char:
>>> ascii.write(dat, format='fixed_width_two_line',
... delimiter_pad=' ', position_char='=')
Col1 Col2 Col3 Col4
==== ========= ==== ====
1.2 "hello" 1 a
2.4 's worlds 2 2
Write a table as a fixed-width table with no bookend:
>>> ascii.write(dat, format='fixed_width_two_line', bookend=True, delimiter='|')
|Col1| Col2|Col3|Col4|
|----|---------|----|----|
| 1.2| "hello"| 1| a|
| 2.4|'s worlds| 2| 2|
Custom Header Rows#
The fixed_width, fixed_width_two_line, and rst formats normally include a
single row with the column names in the header. However, for these formats you can
customize the column attributes which appear as header rows. The available
column attributes are name, dtype, format, description and
unit. This is done by listing the desired the header rows using the
header_rows keyword argument.
- ::
>>> from astropy.table.table_helpers import simple_table >>> dat = simple_table(size=3, cols=4) >>> dat["a"].info.unit = "m" >>> dat["d"].info.unit = "m/s" >>> dat["b"].info.format = ".2f" >>> dat["c"].info.description = "C column" >>> ascii.write( ... dat, ... format="fixed_width", ... header_rows=["name", "unit", "format", "description"], ... ) | a | b | c | d | | m | | | m / s | | | .2f | | | | | | C column | | | 1 | 1.00 | c | 4 | | 2 | 2.00 | d | 5 | | 3 | 3.00 | e | 6 |
In this example the 1st row is the name, the 2nd row is the unit, and
so forth. You must supply the name value in the header_rows list in
order to get an output with the column name included.
A table with non-standard header rows can be read back in the same way, using
the same list of header_rows:
>>> txt = """\
... | int32 | float32 | <U4 | uint8 |
... | a | b | c | d |
... | m | | | m / s |
... | | .2f | | |
... | | | C column | |
... | 1 | 1.00 | c | 4 |
... | 2 | 2.00 | d | 5 |
... | 3 | 3.00 | e | 6 |
... """
>>> dat = ascii.read(
... txt,
... format="fixed_width",
... header_rows=["dtype", "name", "unit", "format", "description"],
... )
>>> dat.info
<Table length=3>
name dtype unit format description
---- ------- ----- ------ -----------
a int32 m
b float32 .2f
c str4 C column
d uint8 m / s
The same idea can be used with the fixed_width_two_line format:
>>> txt = """\
... a b c d
... int64 float64 <U1 int64
... m m / s
... ----- ------- -------- -----
... 1 1.00 c 4
... 2 2.00 d 5
... 3 3.00 e 6
... """
>>> dat = ascii.read(
... txt,
... format="fixed_width_two_line",
... header_rows=["name", "dtype", "unit"],
... )
>>> dat
<Table length=3>
a b c d
m m / s
int64 float64 str1 int64
----- ------- ---- -----
1 1.0 c 4
2 2.0 d 5
3 3.0 e 6
Note that the two_line in the fixed_width_two_line format name refers to
the default situation where the header consists two lines, a row of column names
and a row of separator lines. This is a bit of a misnomer when using
header_rows.