def createReader(readable, **keywords):
"""
A reader fitting the contents of ``readable``. Supported formats are
delimited data (such as CSV), ODS and Excel. When iterating the reader,
it returns a Python array for each row of data.
"""
# TODO: Get rid of circular import.
import _parsers
assert readable is not None
encoding = keywords.get("encoding", DEFAULT_ENCODING)
assert encoding is not None
result = None
icdHeader = readable.read(4)
_log.debug(u"header=%r", icdHeader)
if icdHeader == _ODS_HEADER:
# Consider ICD to be ODS.
readable.seek(0)
result = _parsers.odsReader(readable)
else:
icdHeader += readable.read(4)
readable.seek(0)
if _tools.isEqualBytes(icdHeader, _EXCEL_HEADER):
# Consider ICD to be Excel.
result = _parsers.excelReader(readable)
else:
# Consider ICD to be CSV.
dialect = _parsers.DelimitedDialect()
dialect.lineDelimiter = _parsers.AUTO
dialect.itemDelimiter = _parsers.AUTO
dialect.quoteChar = "\""
dialect.escapeChar = "\""
result = _parsers.delimitedReader(readable, dialect, encoding)
return result
def createDataFormat(readable, **keywords):
"""
Data format describing the contents of ``readable``, which should be a
a raw binary input stream as returned by ``open(..., 'rb')``. Do not use
``codecs.open(...)`` because it returns Unicode strings instead of raw
strings.
Supported formats are delimited data (such as CSV), ODS and Excel.
"""
assert readable is not None
encoding = keywords.get("encoding", DEFAULT_ENCODING)
assert encoding is not None
icdHeader = readable.read(4)
_log.debug(u"header=%r", icdHeader)
if _tools.isEqualBytes(icdHeader, _ODS_HEADER):
# Consider ICD to be ODS.
dataFormatName = data.FORMAT_ODS
else:
icdHeader += readable.read(4)
assert isinstance(icdHeader, str), u"icdHeader=%r but must be a string; use open(..., 'rb') instead of codecs.open()" % icdHeader
assert isinstance(_EXCEL_HEADER, str), u"_EXCEL_HEADER=%r" % _EXCEL_HEADER
if _tools.isEqualBytes(icdHeader, _EXCEL_HEADER):
# Consider ICD to be Excel.
dataFormatName = data.FORMAT_EXCEL
else:
# Consider ICD to be CSV.
dataFormatName = data.FORMAT_DELIMITED
result = data.createDataFormat(dataFormatName)
if result.name == data.FORMAT_DELIMITED:
readable.seek(0)
options = delimitedOptions(readable, **keywords)
for key, value in options.items():
propertyName = _tools.decamelized(key)
if key == _LINE_DELIMITER:
value = _LINE_DELIMITER_TO_NAME_MAP[value]
result.set(propertyName, value)
readable.seek(0)
return result
