DataTable¶

DataTable.py

INTRODUCTION

This class is useful for representing a table of data arranged by named columns, where each row in the table can be thought of as a record:

name   phoneNumber
------ -----------
Chuck  893-3498
Bill   893-0439
John   893-5901

This data often comes from delimited text files which typically have well-defined columns or fields with several rows each of which can be thought of as a record.

Using a DataTable can be as easy as using lists and dictionaries:

table = DataTable('users.csv')
for row in table:
    print(row['name'], row['phoneNumber'])

Or even:

table = DataTable('users.csv')
for row in table:
    print('{name} {phoneNumber}'.format(**row))

The above print statement relies on the fact that rows can be treated like dictionaries, using the column headings as keys.

You can also treat a row like an array:

table = DataTable('something.tabbed', delimiter='   ')
for row in table:
    for item in row:
        print(item, end=' ')
    print()

COLUMNS

Column headings can have a type specification like so:

name, age:int, zip:int

Possible types include string, int, float and datetime.

String is assumed if no type is specified but you can set that assumption when you create the table:

table = DataTable(headings, defaultType='float')

Using types like int and float will cause DataTable to actually convert the string values (perhaps read from a file) to these types so that you can use them in natural operations. For example:

if row['age'] > 120:
    self.flagData(row, 'age looks high')

As you can see, each row can be accessed as a dictionary with keys according the column headings. Names are case sensitive.

ADDING ROWS

Like Python lists, data tables have an append() method. You can append TableRecords, or you pass a dictionary, list or object, in which case a TableRecord is created based on given values. See the method docs below for more details.

FILES

By default, the files that DataTable reads from are expected to be comma-separated value files.

Limited comments are supported: A comment is any line whose very first character is a #. This allows you to easily comment out lines in your data files without having to remove them.

Whitespace around field values is stripped.

You can control all this behavior through the arguments found in the initializer and the various readFoo() methods:

...delimiter=',', allowComments=True, stripWhite=True

For example:

table = DataTable('foo.tabbed', delimiter=' ',
    allowComments=False, stripWhite=False)

You should access these parameters by their name since additional ones could appear in the future, thereby changing the order.

If you are creating these text files, we recommend the comma-separated-value format, or CSV. This format is better defined than the tab delimited format, and can easily be edited and manipulated by popular spreadsheets and databases.

MICROSOFT EXCEL

On Microsoft Windows systems with Excel and the PyWin32 package (https://github.com/mhammond/pywin32), DataTable will use Excel (via COM) to read “.xls” files:

from MiscUtils import DataTable
assert DataTable.canReadExcel()
table = DataTable.DataTable('foo.xls')

With consistency to its CSV processing, DataTable will ignore any row whose first cell is ‘#’ and strip surrounding whitespace around strings.

TABLES FROM SCRATCH

Here’s an example that constructs a table from scratch:

table = DataTable(['name', 'age:int'])
table.append(['John', 80])
table.append({'name': 'John', 'age': 80})
print(table)

QUERIES

A simple query mechanism is supported for equality of fields:

matches = table.recordsEqualTo({'uid': 5})
if matches:
    for match in matches:
        print(match)
else:
    print('No matches.')

COMMON USES

Programs can keep configuration and other data in simple comma- separated text files and use DataTable to access them. For example, a web site could read its sidebar links from such a file, thereby allowing people who don’t know Python (or even HTML) to edit these links without having to understand other implementation parts of the site.

Servers can use DataTable to read and write log files.

FROM THE COMMAND LINE

The only purpose in invoking DataTable from the command line is to see if it will read a file:

> python DataTable.py foo.csv

The data table is printed to stdout.

CACHING

DataTable uses “pickle caching” so that it can read .csv files faster on subsequent loads. You can disable this across the board with:

from MiscUtils.DataTable import DataTable
DataTable._usePickleCache = False

Or per instance by passing “usePickleCache=False” to the constructor.

See the docstring of PickleCache.py for more information.

MORE DOCS

Some of the methods in this module have worthwhile doc strings to look at. See below.

TO DO

Allow callback parameter or setting for parsing CSV records.

Perhaps TableRecord should inherit list and dict and override methods as appropriate?

_types and _blankValues aren’t really packaged, advertised or documented for customization by the user of this module.

DataTable:

Parameterize the TextColumn class.

Parameterize the TableRecord class.

More list-like methods such as insert()

writeFileNamed() is flawed: it doesn’t write the table column type

Should it inherit from list?

Add error checking that a column name is not a number (which could cause problems).

Reading Excel sheets with xlrd, not only with win32com.

class MiscUtils.DataTable.DataTable(filenameOrHeadings=None, delimiter=',', allowComments=True, stripWhite=True, encoding=None, defaultType=None, usePickleCache=None)¶

Bases: object

Representation of a data table.

See the doc string for this module.

__init__(filenameOrHeadings=None, delimiter=',', allowComments=True, stripWhite=True, encoding=None, defaultType=None, usePickleCache=None)¶

append(obj)¶

Append an object to the table.

If obj is not a TableRecord, then one is created, passing the object to initialize the TableRecord. Therefore, obj can be a TableRecord, list, dictionary or object. See TableRecord for details.

static canReadExcel()¶

commit()¶

createNameToIndexMap()¶

Create speed-up index.

Invoked by self to create the nameToIndexMap after the table’s headings have been read/initialized.

dictKeyedBy(key)¶

Return a dictionary containing the contents of the table.

The content is indexed by the particular key. This is useful for tables that have a column which represents a unique key (such as a name, serial number, etc.).

filename()¶

hasHeading(name)¶

heading(index)¶

headings()¶

nameToIndexMap()¶

Speed-up index.

Table rows keep a reference to this map in order to speed up index-by-names (as in row[‘name’]).

numHeadings()¶

readExcel(worksheet=1, row=1, column=1)¶

readFile(file, delimiter=',', allowComments=True, stripWhite=True)¶

readFileNamed(filename, delimiter=',', allowComments=True, stripWhite=True, encoding=None, worksheet=1, row=1, column=1)¶

readLines(lines, delimiter=',', allowComments=True, stripWhite=True)¶

readString(string, delimiter=',', allowComments=True, stripWhite=True)¶

recordsEqualTo(record)¶

save()¶

setHeadings(headings)¶

Set table headings.

Headings can be a list of strings (like [‘name’, ‘age:int’]) or a list of TableColumns or None.

writeFile(file)¶

Write the table out as a file.

This doesn’t write the column types (like int) back out.

It’s notable that a blank numeric value gets read as zero and written out that way. Also, values None are written as blanks.

writeFileNamed(filename, encoding='utf-8')¶

exception MiscUtils.DataTable.DataTableError¶

Bases: Exception

Data table error.

__init__(*args, **kwargs)¶

args¶

with_traceback()¶: Exception.with_traceback(tb) – set self.__traceback__ to tb and return self.

class MiscUtils.DataTable.TableColumn(spec)¶

Bases: object

Representation of a table column.

A table column represents a column of the table including name and type. It does not contain the actual values of the column. These are stored individually in the rows.

__init__(spec)¶

Initialize the table column.

The spec parameter is a string such as ‘name’ or ‘name:type’.

name()¶

setType(type_)¶

Set the type (by a string containing the name) of the heading.

Usually invoked by DataTable to set the default type for columns whose types were not specified.

type()¶

valueForRawValue(value)¶

Set correct type for raw value.

The rawValue is typically a string or value already of the appropriate type. TableRecord invokes this method to ensure that values (especially strings that come from files) are the correct types (e.g., ints are ints and floats are floats).

class MiscUtils.DataTable.TableRecord(table, values=None, headings=None)¶

Bases: object

Representation of a table record.

__init__(table, values=None, headings=None)¶

Initialize table record.

Dispatches control to one of the other init methods based on the type of values. Values can be one of three things:

A TableRecord

A list

A dictionary

Any object responding to hasValueForKey() and valueForKey().

asDict()¶: Return a dictionary whose key-values match the table record.

asList()¶

Return a sequence whose values are the same as the record’s.

The order of the sequence is the one defined by the table.

get(key, default=None)¶

has_key(key)¶

initFromDict(values)¶

initFromObject(obj)¶

Initialize from object.

The object is expected to response to hasValueForKey(name) and valueForKey(name) for each of the headings in the table. It’s alright if the object returns False for hasValueForKey(). In that case, a “blank” value is assumed (such as zero or an empty string). If hasValueForKey() returns True, then valueForKey() must return a value.

initFromSequence(values)¶

items()¶

iteritems()¶

iterkeys()¶

itervalues()¶

keys()¶

valueForAttr(attr, default=<class 'MiscUtils.NoDefault'>)¶

valueForKey(key, default=<class 'MiscUtils.NoDefault'>)¶

values()¶

MiscUtils.DataTable.canReadExcel()¶

MiscUtils.DataTable.main(args=None)¶

DataTable¶

Webware for Python 3

Navigation

Related Topics