def testCanNamifyEmptyText(self):
self.assertEqual(_tools.namified(u""), "x")
self.assertEqual(_tools.namified(u" "), "x")
self.assertEqual(_tools.namified(u"\t"), "x")
def testCanNamifyControlCharacters(self):
self.assertEqual(_tools.namified(u"\r"), "x")
self.assertEqual(_tools.namified(u"a\rb"), "a_b")
def testCanNamifyNumber(self):
self.assertEqual(_tools.namified(u"1a"), "x1a")
self.assertEqual(_tools.namified(u"3.1415"), "x3_1415")
def testCanNamifyKeyword(self):
self.assertEqual(_tools.namified(u"if"), "if_")
def testCanNamifyControlCharacters(self):
self.assertEqual(_tools.namified(u"\r"), "x")
self.assertEqual(_tools.namified(u"a\rb"), "a_b")
def testCanNamifyText(self):
self.assertEqual(_tools.namified(u"hello"), "hello")
self.assertEqual(_tools.namified(u"hElLo"), "hElLo")
self.assertEqual(_tools.namified(u"h3LL0"), "h3LL0")
self.assertEqual(_tools.namified(u"Date of birth"), "Date_of_birth")
self.assertEqual(_tools.namified(u"a b"), "a_b")
def testCanNamifyEmptyText(self):
self.assertEqual(_tools.namified(u""), "x")
self.assertEqual(_tools.namified(u" "), "x")
self.assertEqual(_tools.namified(u"\t"), "x")
def testCanNamifyKeyword(self):
self.assertEqual(_tools.namified(u"if"), "if_")
def testCanNamifyNumber(self):
self.assertEqual(_tools.namified(u"1a"), "x1a")
self.assertEqual(_tools.namified(u"3.1415"), "x3_1415")
def testCanNamifyText(self):
self.assertEqual(_tools.namified(u"hello"), "hello")
self.assertEqual(_tools.namified(u"hElLo"), "hElLo")
self.assertEqual(_tools.namified(u"h3LL0"), "h3LL0")
self.assertEqual(_tools.namified(u"Date of birth"), "Date_of_birth")
self.assertEqual(_tools.namified(u"a b"), "a_b")
def createCidRows(readable, **keywords):
"""
Create rows for an ICD by examining the contents of ``readable``.
Optional keyword parameters are:
* ``encoding`` - the character encoding to be used in case ``readable``
contains delimited data.
* ``dataFormat`` - the data format to be assumed; default: `FORMAT_AUTO`.
* ``header`` - number of header rows to ignore for data analysis;
default: 0.
* ``stopAfter`` - number of data rows after which to stop analyzing;
0 means "analyze all data"; default: 0.
* ``fieldNames`` - Python names to refer to the fields. If this is a list of
strings, each string represents a field name. If this is a single
string, split it using comma (,) as separator to get to the names. If
this is ``None`` (the default), use the last column of the ``header``
as names. If ``header`` is ``None``, use generated field names such as
'column_a', 'column_b' and so on.
"""
assert readable is not None
dataFormat = keywords.get("dataFormat", FORMAT_AUTO)
assert dataFormat is not None
encoding = keywords.get(_ENCODING, DEFAULT_ENCODING)
assert encoding is not None
dataRowsToStopAfter = keywords.get("stopAfter", 0)
assert dataRowsToStopAfter >= 0
headerRowsToSkip = keywords.get("header", 0)
assert headerRowsToSkip >= 0
fieldNames = keywords.get("fieldNames")
if isinstance(fieldNames, basestring):
fieldNames = [name.strip() for name in fieldNames.split(",")]
elif fieldNames is not None:
assert isinstance(fieldNames, list), u"field names must be a list or string but is: %s" % type(fieldNames)
def decimalRows(hasDecimals, usesThousandsSeparator, decimalSeparator, thousandsSeparator):
result = []
if hasDecimals:
result.append(["d", data.KEY_DECIMAL_SEPARATOR, decimalSeparator])
if usesThousandsSeparator:
result.append(["d", data.KEY_THOUSANDS_SEPARATOR, thousandsSeparator])
return result
NO_COUNT = -1
_log.debug(u"find longest segment of rows with same column count")
currentSegmentColumnCount = None
longestSegmentColumnCount = NO_COUNT
longestSegmentRowCount = NO_COUNT
currentSegmentRowCount = 0
rowIndex = 0
rowIndexWhereCurrentSegmentStarted = 0
rowIndexWhereLongestSegmentStarts = None
# TODO: Cleanup code: calling both createDataFormat and createReader causes the data format to be analyzed twice.
dataFormat = createDataFormat(readable, **keywords)
readable.seek(0)
reader = createReader(readable, **keywords)
isFirstRow = True
isReadFieldNamesFromHeader = (not fieldNames and headerRowsToSkip)
for rowToAnalyze in reader:
columnCount = len(rowToAnalyze)
if isFirstRow:
currentSegmentColumnCount = columnCount
else:
isFirstRow = False
if isReadFieldNamesFromHeader and (rowIndex == headerRowsToSkip - 1):
fieldNames = rowToAnalyze
if (rowIndex >= headerRowsToSkip) and (columnCount != currentSegmentColumnCount):
_log.debug(u" segment starts in row %d after %d rows", rowIndex, currentSegmentRowCount)
if currentSegmentRowCount > longestSegmentRowCount:
rowIndexWhereLongestSegmentStarts = rowIndexWhereCurrentSegmentStarted
longestSegmentRowCount = currentSegmentRowCount
longestSegmentColumnCount = currentSegmentColumnCount
rowIndexWhereCurrentSegmentStarted = rowIndex
currentSegmentRowCount = 0
currentSegmentColumnCount = columnCount
else:
currentSegmentRowCount += 1
rowIndex += 1
# Validate field names.
if fieldNames is not None:
if isReadFieldNamesFromHeader:
location = tools.InputLocation(readable, hasCell=True)
location.advanceLine(headerRowsToSkip)
else:
location = None
if not fieldNames:
raise data.DataFormatSyntaxError(u"the field names specified must contain at least 1 name", location)
uniquefieldNames = set()
for nameIndex in range(len(fieldNames)):
fieldNameToCheck = fieldNames[nameIndex]
if isReadFieldNamesFromHeader:
fieldNameToCheck = _tools.namified(fieldNameToCheck)
fieldNameToCheck = fields.validatedFieldName(fieldNameToCheck, location)
if fieldNameToCheck in uniquefieldNames:
raise fields.FieldSyntaxError(u"field name must be unique: %s" % fieldNameToCheck, location)
fieldNames[nameIndex] = fieldNameToCheck
uniquefieldNames.add(fieldNameToCheck)
if location:
location.advanceCell()
# Handle the case that the whole file can be one large segment.
_log.debug(u"last segment started in row %d and lasted for %d rows", rowIndexWhereCurrentSegmentStarted, currentSegmentRowCount)
if currentSegmentRowCount > longestSegmentRowCount:
rowIndexWhereLongestSegmentStarts = rowIndexWhereCurrentSegmentStarted
longestSegmentRowCount = currentSegmentRowCount
longestSegmentColumnCount = currentSegmentColumnCount
if longestSegmentRowCount < 1:
raise CutplaceSniffError(u"content must contain data for format to be sniffed")
_log.debug(u"found longest segment starting in row %d lasting for %d rows having %d columns",
rowIndexWhereLongestSegmentStarts, longestSegmentRowCount, longestSegmentColumnCount)
assert rowIndexWhereLongestSegmentStarts is not None
_log.debug(u"skip %d rows until longest segment starts", rowIndexWhereLongestSegmentStarts)
readable.seek(0)
reader = createReader(readable, **keywords)
rowIndex = 0
location = tools.InputLocation(readable)
while rowIndex < rowIndexWhereLongestSegmentStarts:
reader.next()
location.advanceLine()
rowIndex += 1
_log.debug(u"analyze longest segment of rows with same column count")
columnInfos = []
for columnIndex in range(longestSegmentColumnCount):
columnInfoToAppend = _ColumnSniffInfo(columnIndex, dataFormat)
if fieldNames:
columnInfoToAppend.name = fieldNames[columnIndex]
columnInfos.append(columnInfoToAppend)
rowIndex = 0
while rowIndex < longestSegmentRowCount:
rowToAnalyze = reader.next()
if rowIndex >= headerRowsToSkip:
columnCountOfRowToAnalyze = len(rowToAnalyze)
if columnCountOfRowToAnalyze != longestSegmentColumnCount:
raise CutplaceSniffError(u"data must not change between sniffer passes, but row %d now has %d columns instead of %d" \
% (rowIndex + 1, columnCountOfRowToAnalyze, longestSegmentColumnCount), location)
for itemIndex in range(longestSegmentColumnCount):
value = rowToAnalyze[itemIndex]
columnInfos[itemIndex].process(value)
location.advanceLine()
rowIndex += 1
# Make sure that decimal fields either use comma or point.
decimalCommaColumnCount = 0
decimalPointColumnCount = 0
usesThousandSeparator = False
for columnInfo in columnInfos:
if columnInfo.isNumber and not columnInfo.isInteger:
if columnInfo.isDecimalComma:
_log.debug(u"field is decimal with comma as separator: %s", columnInfo.name)
decimalCommaColumnCount += 1
if columnInfo.usesThousandsSeparator:
_log.debug(u" decimal field uses point as thousands separator: %s", columnInfo.name)
usesThousandSeparator = True
elif columnInfo.isDecimalPoint:
_log.debug(u"field is decimal with point as separator: %s", columnInfo.name)
decimalPointColumnCount += 1
if columnInfo.usesThousandsSeparator:
_log.debug(u" decimal field uses comma as thousands separator: %s", columnInfo.name)
usesThousandSeparator = True
if (decimalCommaColumnCount > 0) and (decimalPointColumnCount > 0):
_log.warning(u"columns use different decimal separators: %d use comma, %d use point",
decimalCommaColumnCount, decimalPointColumnCount)
hasToChangeDecimalComma = (decimalCommaColumnCount < decimalPointColumnCount)
if hasToChangeDecimalComma:
decimalCommaColumnCount = 0
else:
decimalPointColumnCount = 0
for columnInfo in columnInfos:
if (columnInfo.isDecimalComma and hasToChangeDecimalComma) or columnInfo.isDecimalPoint and not hasToChangeDecimalComma:
_log.warn(u" change '%s' to text field", columnInfo.name)
columnInfo.changeToTextField()
assert (decimalCommaColumnCount == 0) or (decimalPointColumnCount == 0)
for columnIndex in range(longestSegmentColumnCount):
_log.debug(u" %s" % columnInfos[columnIndex].asFieldFormat())
# Build rows for CID: data data format.
icdRows = []
icdRows.append(["", "Interface: <Name>"])
icdRows.append([])
for dataFormatRow in dataFormat.asIcdRows():
dataFormatCsvRow = ['d']
dataFormatCsvRow.extend(dataFormatRow)
icdRows.append(dataFormatCsvRow)
_log.info(u"number of decimal fields: %d with comma, %d with point", decimalCommaColumnCount, decimalPointColumnCount)
icdRows.extend(decimalRows(decimalCommaColumnCount > 0, usesThousandSeparator, ",", "."))
icdRows.extend(decimalRows(decimalPointColumnCount > 0, usesThousandSeparator, ".", ","))
icdRows.append([])
# Build rows for CID: data field formats.
icdRows.append(["", "Field", "Example", "Empty?", "Length", "Type", "Rule"])
for columnInfo in columnInfos:
fieldFormat = columnInfo.asFieldFormat()
fieldRow = ["f"]
fieldRow.extend(fieldFormat.asIcdRow())
icdRows.append(fieldRow)
_log.debug(u" %s", fieldRow)
return icdRows
