2
:mod:`csv` --- CSV File Reading and Writing
3
===========================================
6
:synopsis: Write and read tabular data to and from delimited files.
7
.. sectionauthor:: Skip Montanaro <skip@pobox.com>
14
The so-called CSV (Comma Separated Values) format is the most common import and
15
export format for spreadsheets and databases. There is no "CSV standard", so
16
the format is operationally defined by the many applications which read and
17
write it. The lack of a standard means that subtle differences often exist in
18
the data produced and consumed by different applications. These differences can
19
make it annoying to process CSV files from multiple sources. Still, while the
20
delimiters and quoting characters vary, the overall format is similar enough
21
that it is possible to write a single module which can efficiently manipulate
22
such data, hiding the details of reading and writing the data from the
25
The :mod:`csv` module implements classes to read and write tabular data in CSV
26
format. It allows programmers to say, "write this data in the format preferred
27
by Excel," or "read data from this file which was generated by Excel," without
28
knowing the precise details of the CSV format used by Excel. Programmers can
29
also describe the CSV formats understood by other applications or define their
30
own special-purpose CSV formats.
32
The :mod:`csv` module's :class:`reader` and :class:`writer` objects read and
33
write sequences. Programmers can also read and write data in dictionary form
34
using the :class:`DictReader` and :class:`DictWriter` classes.
38
This version of the :mod:`csv` module doesn't support Unicode input. Also,
39
there are currently some issues regarding ASCII NUL characters. Accordingly,
40
all input should be UTF-8 or printable ASCII to be safe; see the examples in
41
section :ref:`csv-examples`. These restrictions will be removed in the future.
46
:pep:`305` - CSV File API
47
The Python Enhancement Proposal which proposed this addition to Python.
55
The :mod:`csv` module defines the following functions:
58
.. function:: reader(csvfile[, dialect='excel'][, fmtparam])
60
Return a reader object which will iterate over lines in the given *csvfile*.
61
*csvfile* can be any object which supports the :term:`iterator` protocol and returns a
62
string each time its :meth:`next` method is called --- file objects and list
63
objects are both suitable. If *csvfile* is a file object, it must be opened
64
with the 'b' flag on platforms where that makes a difference. An optional
65
*dialect* parameter can be given which is used to define a set of parameters
66
specific to a particular CSV dialect. It may be an instance of a subclass of
67
the :class:`Dialect` class or one of the strings returned by the
68
:func:`list_dialects` function. The other optional *fmtparam* keyword arguments
69
can be given to override individual formatting parameters in the current
70
dialect. For full details about the dialect and formatting parameters, see
71
section :ref:`csv-fmt-params`.
73
All data read are returned as strings. No automatic data type conversion is
76
The parser is quite strict with respect to multi-line quoted fields. Previously,
77
if a line ended within a quoted field without a terminating newline character, a
78
newline would be inserted into the returned field. This behavior caused problems
79
when reading files which contained carriage return characters within fields.
80
The behavior was changed to return the field without inserting newlines. As a
81
consequence, if newlines embedded within fields are important, the input should
82
be split into lines in a manner which preserves the newline characters.
84
A short usage example::
87
>>> spamReader = csv.reader(open('eggs.csv'), delimiter=' ', quotechar='|')
88
>>> for row in spamReader:
89
... print(', '.join(row))
90
Spam, Spam, Spam, Spam, Spam, Baked Beans
91
Spam, Lovely Spam, Wonderful Spam
94
.. function:: writer(csvfile[, dialect='excel'][, fmtparam])
96
Return a writer object responsible for converting the user's data into delimited
97
strings on the given file-like object. *csvfile* can be any object with a
98
:func:`write` method. If *csvfile* is a file object, it must be opened with the
99
'b' flag on platforms where that makes a difference. An optional *dialect*
100
parameter can be given which is used to define a set of parameters specific to a
101
particular CSV dialect. It may be an instance of a subclass of the
102
:class:`Dialect` class or one of the strings returned by the
103
:func:`list_dialects` function. The other optional *fmtparam* keyword arguments
104
can be given to override individual formatting parameters in the current
105
dialect. For full details about the dialect and formatting parameters, see
106
section :ref:`csv-fmt-params`. To make it
107
as easy as possible to interface with modules which implement the DB API, the
108
value :const:`None` is written as the empty string. While this isn't a
109
reversible transformation, it makes it easier to dump SQL NULL data values to
110
CSV files without preprocessing the data returned from a ``cursor.fetch*`` call.
111
All other non-string data are stringified with :func:`str` before being written.
113
A short usage example::
116
>>> spamWriter = csv.writer(open('eggs.csv', 'w'), delimiter=' ',
117
... quotechar='|', quoting=QUOTE_MINIMAL)
118
>>> spamWriter.writerow(['Spam'] * 5 + ['Baked Beans'])
119
>>> spamWriter.writerow(['Spam', 'Lovely Spam', 'Wonderful Spam'])
122
.. function:: register_dialect(name[, dialect][, fmtparam])
124
Associate *dialect* with *name*. *name* must be a string. The
125
dialect can be specified either by passing a sub-class of :class:`Dialect`, or
126
by *fmtparam* keyword arguments, or both, with keyword arguments overriding
127
parameters of the dialect. For full details about the dialect and formatting
128
parameters, see section :ref:`csv-fmt-params`.
131
.. function:: unregister_dialect(name)
133
Delete the dialect associated with *name* from the dialect registry. An
134
:exc:`Error` is raised if *name* is not a registered dialect name.
137
.. function:: get_dialect(name)
139
Return the dialect associated with *name*. An :exc:`Error` is raised if
140
*name* is not a registered dialect name. This function returns an immutable
143
.. function:: list_dialects()
145
Return the names of all registered dialects.
148
.. function:: field_size_limit([new_limit])
150
Returns the current maximum field size allowed by the parser. If *new_limit* is
151
given, this becomes the new limit.
154
The :mod:`csv` module defines the following classes:
156
.. class:: DictReader(csvfile[, fieldnames=None[, restkey=None[, restval=None[, dialect='excel'[, *args, **kwds]]]]])
158
Create an object which operates like a regular reader but maps the information
159
read into a dict whose keys are given by the optional *fieldnames* parameter.
160
If the *fieldnames* parameter is omitted, the values in the first row of the
161
*csvfile* will be used as the fieldnames. If the row read has fewer fields than
162
the fieldnames sequence, the value of *restval* will be used as the default
163
value. If the row read has more fields than the fieldnames sequence, the
164
remaining data is added as a sequence keyed by the value of *restkey*. If the
165
row read has fewer fields than the fieldnames sequence, the remaining keys take
166
the value of the optional *restval* parameter. Any other optional or keyword
167
arguments are passed to the underlying :class:`reader` instance.
170
.. class:: DictWriter(csvfile, fieldnames[, restval=''[, extrasaction='raise'[, dialect='excel'[, *args, **kwds]]]])
172
Create an object which operates like a regular writer but maps dictionaries onto
173
output rows. The *fieldnames* parameter identifies the order in which values in
174
the dictionary passed to the :meth:`writerow` method are written to the
175
*csvfile*. The optional *restval* parameter specifies the value to be written
176
if the dictionary is missing a key in *fieldnames*. If the dictionary passed to
177
the :meth:`writerow` method contains a key not found in *fieldnames*, the
178
optional *extrasaction* parameter indicates what action to take. If it is set
179
to ``'raise'`` a :exc:`ValueError` is raised. If it is set to ``'ignore'``,
180
extra values in the dictionary are ignored. Any other optional or keyword
181
arguments are passed to the underlying :class:`writer` instance.
183
Note that unlike the :class:`DictReader` class, the *fieldnames* parameter of
184
the :class:`DictWriter` is not optional. Since Python's :class:`dict` objects
185
are not ordered, there is not enough information available to deduce the order
186
in which the row should be written to the *csvfile*.
191
The :class:`Dialect` class is a container class relied on primarily for its
192
attributes, which are used to define the parameters for a specific
193
:class:`reader` or :class:`writer` instance.
198
The :class:`excel` class defines the usual properties of an Excel-generated CSV
199
file. It is registered with the dialect name ``'excel'``.
202
.. class:: excel_tab()
204
The :class:`excel_tab` class defines the usual properties of an Excel-generated
205
TAB-delimited file. It is registered with the dialect name ``'excel-tab'``.
210
The :class:`Sniffer` class is used to deduce the format of a CSV file.
212
The :class:`Sniffer` class provides two methods:
214
.. method:: sniff(sample[, delimiters=None])
216
Analyze the given *sample* and return a :class:`Dialect` subclass
217
reflecting the parameters found. If the optional *delimiters* parameter
218
is given, it is interpreted as a string containing possible valid
219
delimiter characters.
222
.. method:: has_header(sample)
224
Analyze the sample text (presumed to be in CSV format) and return
225
:const:`True` if the first row appears to be a series of column headers.
227
An example for :class:`Sniffer` use::
229
csvfile = open("example.csv")
230
dialect = csv.Sniffer().sniff(csvfile.read(1024))
232
reader = csv.reader(csvfile, dialect)
233
# ... process CSV file contents here ...
236
The :mod:`csv` module defines the following constants:
240
Instructs :class:`writer` objects to quote all fields.
243
.. data:: QUOTE_MINIMAL
245
Instructs :class:`writer` objects to only quote those fields which contain
246
special characters such as *delimiter*, *quotechar* or any of the characters in
250
.. data:: QUOTE_NONNUMERIC
252
Instructs :class:`writer` objects to quote all non-numeric fields.
254
Instructs the reader to convert all non-quoted fields to type *float*.
259
Instructs :class:`writer` objects to never quote fields. When the current
260
*delimiter* occurs in output data it is preceded by the current *escapechar*
261
character. If *escapechar* is not set, the writer will raise :exc:`Error` if
262
any characters that require escaping are encountered.
264
Instructs :class:`reader` to perform no special processing of quote characters.
266
The :mod:`csv` module defines the following exception:
271
Raised by any of the functions when an error is detected.
276
Dialects and Formatting Parameters
277
----------------------------------
279
To make it easier to specify the format of input and output records, specific
280
formatting parameters are grouped together into dialects. A dialect is a
281
subclass of the :class:`Dialect` class having a set of specific methods and a
282
single :meth:`validate` method. When creating :class:`reader` or
283
:class:`writer` objects, the programmer can specify a string or a subclass of
284
the :class:`Dialect` class as the dialect parameter. In addition to, or instead
285
of, the *dialect* parameter, the programmer can also specify individual
286
formatting parameters, which have the same names as the attributes defined below
287
for the :class:`Dialect` class.
289
Dialects support the following attributes:
292
.. attribute:: Dialect.delimiter
294
A one-character string used to separate fields. It defaults to ``','``.
297
.. attribute:: Dialect.doublequote
299
Controls how instances of *quotechar* appearing inside a field should be
300
themselves be quoted. When :const:`True`, the character is doubled. When
301
:const:`False`, the *escapechar* is used as a prefix to the *quotechar*. It
302
defaults to :const:`True`.
304
On output, if *doublequote* is :const:`False` and no *escapechar* is set,
305
:exc:`Error` is raised if a *quotechar* is found in a field.
308
.. attribute:: Dialect.escapechar
310
A one-character string used by the writer to escape the *delimiter* if *quoting*
311
is set to :const:`QUOTE_NONE` and the *quotechar* if *doublequote* is
312
:const:`False`. On reading, the *escapechar* removes any special meaning from
313
the following character. It defaults to :const:`None`, which disables escaping.
316
.. attribute:: Dialect.lineterminator
318
The string used to terminate lines produced by the :class:`writer`. It defaults
323
The :class:`reader` is hard-coded to recognise either ``'\r'`` or ``'\n'`` as
324
end-of-line, and ignores *lineterminator*. This behavior may change in the
328
.. attribute:: Dialect.quotechar
330
A one-character string used to quote fields containing special characters, such
331
as the *delimiter* or *quotechar*, or which contain new-line characters. It
335
.. attribute:: Dialect.quoting
337
Controls when quotes should be generated by the writer and recognised by the
338
reader. It can take on any of the :const:`QUOTE_\*` constants (see section
339
:ref:`csv-contents`) and defaults to :const:`QUOTE_MINIMAL`.
342
.. attribute:: Dialect.skipinitialspace
344
When :const:`True`, whitespace immediately following the *delimiter* is ignored.
345
The default is :const:`False`.
351
Reader objects (:class:`DictReader` instances and objects returned by the
352
:func:`reader` function) have the following public methods:
355
.. method:: csvreader.next()
357
Return the next row of the reader's iterable object as a list, parsed according
358
to the current dialect.
360
Reader objects have the following public attributes:
363
.. attribute:: csvreader.dialect
365
A read-only description of the dialect in use by the parser.
368
.. attribute:: csvreader.line_num
370
The number of lines read from the source iterator. This is not the same as the
371
number of records returned, as records can span multiple lines.
375
DictReader objects have the following public attribute:
378
.. attribute:: csvreader.fieldnames
380
If not passed as a parameter when creating the object, this attribute is
381
initialized upon first access or when the first record is read from the
389
:class:`Writer` objects (:class:`DictWriter` instances and objects returned by
390
the :func:`writer` function) have the following public methods. A *row* must be
391
a sequence of strings or numbers for :class:`Writer` objects and a dictionary
392
mapping fieldnames to strings or numbers (by passing them through :func:`str`
393
first) for :class:`DictWriter` objects. Note that complex numbers are written
394
out surrounded by parens. This may cause some problems for other programs which
395
read CSV files (assuming they support complex numbers at all).
398
.. method:: csvwriter.writerow(row)
400
Write the *row* parameter to the writer's file object, formatted according to
404
.. method:: csvwriter.writerows(rows)
406
Write all the *rows* parameters (a list of *row* objects as described above) to
407
the writer's file object, formatted according to the current dialect.
409
Writer objects have the following public attribute:
412
.. attribute:: csvwriter.dialect
414
A read-only description of the dialect in use by the writer.
422
The simplest example of reading a CSV file::
425
reader = csv.reader(open("some.csv", "rb"))
429
Reading a file with an alternate format::
432
reader = csv.reader(open("passwd", "rb"), delimiter=':', quoting=csv.QUOTE_NONE)
436
The corresponding simplest possible writing example is::
439
writer = csv.writer(open("some.csv", "wb"))
440
writer.writerows(someiterable)
442
Registering a new dialect::
446
csv.register_dialect('unixpwd', delimiter=':', quoting=csv.QUOTE_NONE)
448
reader = csv.reader(open("passwd", "rb"), 'unixpwd')
450
A slightly more advanced use of the reader --- catching and reporting errors::
453
filename = "some.csv"
454
reader = csv.reader(open(filename, "rb"))
458
except csv.Error as e:
459
sys.exit('file %s, line %d: %s' % (filename, reader.line_num, e))
461
And while the module doesn't directly support parsing strings, it can easily be
465
for row in csv.reader(['one,two,three']):
468
The :mod:`csv` module doesn't directly support reading and writing Unicode, but
469
it is 8-bit-clean save for some problems with ASCII NUL characters. So you can
470
write functions or classes that handle the encoding and decoding for you as long
471
as you avoid encodings like UTF-16 that use NULs. UTF-8 is recommended.
473
:func:`unicode_csv_reader` below is a :term:`generator` that wraps :class:`csv.reader`
474
to handle Unicode CSV data (a list of Unicode strings). :func:`utf_8_encoder`
475
is a :term:`generator` that encodes the Unicode strings as UTF-8, one string (or row) at
476
a time. The encoded strings are parsed by the CSV reader, and
477
:func:`unicode_csv_reader` decodes the UTF-8-encoded cells back into Unicode::
481
def unicode_csv_reader(unicode_csv_data, dialect=csv.excel, **kwargs):
482
# csv.py doesn't do Unicode; encode temporarily as UTF-8:
483
csv_reader = csv.reader(utf_8_encoder(unicode_csv_data),
484
dialect=dialect, **kwargs)
485
for row in csv_reader:
486
# decode UTF-8 back to Unicode, cell by cell:
487
yield [unicode(cell, 'utf-8') for cell in row]
489
def utf_8_encoder(unicode_csv_data):
490
for line in unicode_csv_data:
491
yield line.encode('utf-8')
493
For all other encodings the following :class:`UnicodeReader` and
494
:class:`UnicodeWriter` classes can be used. They take an additional *encoding*
495
parameter in their constructor and make sure that the data passes the real
496
reader or writer encoded as UTF-8::
498
import csv, codecs, io
502
Iterator that reads an encoded stream and reencodes the input to UTF-8
504
def __init__(self, f, encoding):
505
self.reader = codecs.getreader(encoding)(f)
511
return next(self.reader).encode("utf-8")
515
A CSV reader which will iterate over lines in the CSV file "f",
516
which is encoded in the given encoding.
519
def __init__(self, f, dialect=csv.excel, encoding="utf-8", **kwds):
520
f = UTF8Recoder(f, encoding)
521
self.reader = csv.reader(f, dialect=dialect, **kwds)
524
row = next(self.reader)
525
return [unicode(s, "utf-8") for s in row]
532
A CSV writer which will write rows to CSV file "f",
533
which is encoded in the given encoding.
536
def __init__(self, f, dialect=csv.excel, encoding="utf-8", **kwds):
537
# Redirect output to a queue
538
self.queue = io.StringIO()
539
self.writer = csv.writer(self.queue, dialect=dialect, **kwds)
541
self.encoder = codecs.getincrementalencoder(encoding)()
543
def writerow(self, row):
544
self.writer.writerow([s.encode("utf-8") for s in row])
545
# Fetch UTF-8 output from the queue ...
546
data = self.queue.getvalue()
547
data = data.decode("utf-8")
548
# ... and reencode it into the target encoding
549
data = self.encoder.encode(data)
550
# write to the target stream
551
self.stream.write(data)
553
self.queue.truncate(0)
555
def writerows(self, rows):