Package for working with extended CSV (XCSV) files
Project description
xcsv
xcsv is a package for reading and writing extended CSV files.
Extended CSV format
- Extended header section of parseable atttributes, introduced by '#'.
- Header row of variable name and units for each column.
- Data rows.
Example
Extended header section
- No leading/trailing whitespace.
- Each line introduced by a comment ('#') character.
- Each line contains a single header item, or a single element of a multi-line list header item.
- Key/value separator ': '.
- Multi-line values naturally continued over to the next lines following the line introducing the key.
- Continuation lines that contain the delimiter character in the value must be escaped by a leading delimiter.
- Preferably use a common vocabulary for attribute name, such as CF conventions.
- Preferably include recommended attributes from Attribute Convention for Data Discovery (ACDD).
- Preferably use units from Unified Code for Units of Measure and/or Udunits.
- Units in parentheses.
- Units are automatically parsed when they appear in parentheses at the end of a line. Hence, if you have non-units text in parentheses at the end of the line (e.g. when expanding an acronym), then ensure that the line doesn't end with a closing parenthesis to avoid the text being incorrectly parsed as units. A '.' would suffice.
- This line:
# latitude: -73.86 (degree_north)
would parse correctly as a value/unitsdict
:'latitude': {'value': '-73.86', 'units': 'degree_north'}
. - This line:
# institution: BAS (British Antarctic Survey).
would correctly avoid being parsed as a value/unitsdict
because of the '.' as the last character.
- This line:
- Certain special keys are used to further process the data, for example the
missing_value
key.
# id: 1
# title: The title
# summary: This dataset...
# The second summary paragraph.
# : The third summary paragraph. Escaped because it contains the delimiter in a URL https://dummy.domain
# authors: A B, C D
# institution: BAS (British Antarctic Survey).
# latitude: -73.86 (degree_north)
# longitude: -65.46 (degree_east)
# elevation: 1897 (m a.s.l.)
# [a]: 2012 not a complete year
Header row
- No leading/trailing whitespace.
- Preferably use a common vocabulary for variable name, such as CF conventions.
- Units in parentheses.
- Optional notes in square brackets, that reference an item in the extended header section.
time (year) [a],depth (m)
Data row
- No leading/trailing whitespace.
2012,0.575
Automated post-processing of the data
Depending on the presence of special keys in the extended header section, these will be used to automatically post-process the data. To turn off this automatic behaviour, either remove or rename these keys, or set parse_metadata=False
when reading in the data.
missing_value
: This is used to define those values in the data that are to be considered as missing values. This is typically a value that is outside the domain of the data such as-999.99
, or can be a symbolic value such asNA
. All such values appearing in the data will be masked, appearing as anNA
value to pandas (i.e.pd.isna(value)
returnsTrue
). Note that pandas itself will automatically do this for certain values regardless of this key, such as for the stringsNaN
orNA
, or the constantNone
.
A note on encodings
The default character set encoding is UTF-8, without a Byte Order Mark (BOM). If an extended CSV file has a different encoding, it can either be converted to UTF-8 (by using iconv
, for example), or the encoding can be specified when opening the file (xcsv.File(filename, encoding=encoding)
).
If the encoding of a file is UTF-8 and it begins with a BOM, then the BOM is silently skipped. This is necessary so that the extended header section is parsed correctly.
Install
The package can be installed from PyPI:
$ pip install xcsv
Using the package
The package has a general XCSV
class, that has a metadata
attribute that holds the parsed contents of the extended file header section and the parsed column headers from the data table, and a data
attribute that holds the data table (including the column headers as-is).
The metadata
attribute is a dict
, with the following general structure:
{'header': {}, 'column_headers': {}}
and the data
attribute is a pandas.DataFrame
, and so has all the features of the pandas package.
The package also has a Reader
class for reading an extended CSV file into an XCSV
object, and similarly a Writer
class for writing an XCSV
object to a file in the extended CSV format. In addition there is a File
class that provides a convenient context manager for reading and writing these files.
Examples
Simple read and print
Read in a file and print the contents to stdout
. This shows how the contents of the extended CSV file are stored in the XCSV
object. Note how multi-line values, such as summary
here, are stored in a list. Given the following script called, say, simple_read.py
:
import argparse
import xcsv
parser = argparse.ArgumentParser()
parser.add_argument('filename', help='filename.csv')
args = parser.parse_args()
with xcsv.File(args.filename) as f:
content = f.read()
print(content.metadata)
print(content.data)
Running it would produce:
$ python3 simple_read.py example.csv
{'header': {'id': '1', 'title': 'The title', 'summary': ['This dataset...', 'The second summary paragraph.', 'The third summary paragraph. Escaped because it contains the delimiter in a URL https://dummy.domain'], 'authors': 'A B, C D', 'institution': 'BAS (British Antarctic Survey).', 'latitude': {'value': '-73.86', 'units': 'degree_north'}, 'longitude': {'value': '-65.46', 'units': 'degree_east'}, 'elevation': {'value': '1897', 'units': 'm a.s.l.'}, '[a]': '2012 not a complete year'}, 'column_headers': {'time (year) [a]': {'name': 'time', 'units': 'year', 'notes': 'a'}, 'depth (m)': {'name': 'depth', 'units': 'm', 'notes': None}}}
time (year) [a] depth (m)
0 2012 0.575
1 2011 1.125
2 2010 2.225
Simple read and print with missing values
If the above example header section included the following:
# missing_value: -999.99
and the data section looked like:
time (year) [a],depth (m)
2012,0.575
2011,1.125
2010,2.225
2009,-999
2008,999
2007,-999.99
2006,999.99
2005,NA
2004,NaN
Running it would produce:
$ python3 simple_read.py missing_example.csv
{'header': {'id': '1', 'title': 'The title', 'summary': ['This dataset...', 'The second summary paragraph.', 'The third summary paragraph. Escaped because it contains the delimiter in a URL https://dummy.domain'], 'authors': 'A B, C D', 'institution': 'BAS (British Antarctic Survey).', 'latitude': {'value': '-73.86', 'units': 'degree_north'}, 'longitude': {'value': '-65.46', 'units': 'degree_east'}, 'elevation': {'value': '1897', 'units': 'm a.s.l.'}, 'missing_value': '-999.99', '[a]': '2012 not a complete year'}, 'column_headers': {'time (year) [a]': {'name': 'time', 'units': 'year', 'notes': 'a'}, 'depth (m)': {'name': 'depth', 'units': 'm', 'notes': None}}}
time (year) [a] depth (m)
0 2012 0.575
1 2011 1.125
2 2010 2.225
3 2009 -999.000
4 2008 999.000
5 2007 NaN
6 2006 999.990
7 2005 NaN
8 2004 NaN
Note that the -999.99
value has been automatically masked as a missing value (shown as NaN
in the printed pandas DataFrame
), as well as the NA
and NaN
strings in the original data, which pandas automatically masks itself, irrespective of the missing_value
header item.
Simple read and plot
Read a file and plot the data:
import argparse
import matplotlib.pyplot as plt
import xcsv
parser = argparse.ArgumentParser()
parser.add_argument('filename', help='filename.csv')
args = parser.parse_args()
with xcsv.File(args.filename) as f:
content = f.read()
content.data.plot(x='depth (m)', y='time (year) [a]')
plt.show()
Simple read and write
Read a file in, manipulate the data in some way, and write this modified XCSV
object out to a new file:
import argparse
import xcsv
parser = argparse.ArgumentParser()
parser.add_argument('in_filename', help='in_filename.csv')
parser.add_argument('out_filename', help='out_filename.csv')
args = parser.parse_args()
with xcsv.File(args.in_filename) as f:
content = f.read()
# Manipulate the data...
with xcsv.File(args.out_filename, mode='w') as f:
f.write(xcsv=content)
Project details
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
File details
Details for the file xcsv-0.5.0.tar.gz
.
File metadata
- Download URL: xcsv-0.5.0.tar.gz
- Upload date:
- Size: 17.2 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: poetry/1.2.2 CPython/3.8.10 Linux/5.15.0-91-generic
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 | 9ee06c6fafda1dcd97bc0a2e7e011bf56abcd5b7dcba6e202d8e0ac9e6f6e32d |
|
MD5 | a54dfaff8cb0948501ec68363999831d |
|
BLAKE2b-256 | 5e249313f7d446cbb471ff4bc0ba4580452387a303b0368ad9f21cf784a017d1 |
File details
Details for the file xcsv-0.5.0-py3-none-any.whl
.
File metadata
- Download URL: xcsv-0.5.0-py3-none-any.whl
- Upload date:
- Size: 14.6 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via: poetry/1.2.2 CPython/3.8.10 Linux/5.15.0-91-generic
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 | 9c5afbb1f12f9e006aba9c310b8dac2fbf3712f0eba333b04745976462410b6d |
|
MD5 | 8ee2aad3dd6729923dac1c2aefb1ca8d |
|
BLAKE2b-256 | b5508d4ecfbd2c351875b714836a43dc0602ef352a4bd1fb2dbc0e9a74d580f0 |