Translate tab-delimited to fixed width text
Larry Bates
lbates at swamisoft.com
Tue Sep 21 14:30:52 EDT 2004
What's up with all the semi-colon (;) line terminators?
Seems like some mixture of Javascript and Python here.
I tend to do this with classes. Then I can take
advantage of using the __str__ method to build the
output string.
class outputrec:
#
# default = default value if not given (also defines length)
# fmt = format string to build output (%s, %i, %.2f, etc.)
# justify = None (left) or right
# values = list of allowable values
#
_record={'Email': {'default':100*' ',
'fmt': '%s',
'justify': None,
'values': None}},
'FirstName':{'default':20*' ',
'fmt': '%s',
'justify': None,
'values': None}},
#
# Insert remainder of fields here
#
}
_attributes=('Email','FirstName',...)
def __init__(self, **kwargs):
#-----------------------------------------------------------------------
# If any keyword arguments were passed in, update local attributes
with
# their values.
#-----------------------------------------------------------------------
for attribute in [a for a in kwargs.keys() if a in
self._attributes]:
self.__setattr__(attribute, kwargs[attribute])
return
def __setattr__(self, attribute, value):
'''
Override the __setattr__ method so that I can limit the attributes
to only those in the _record defined and so I can properly justify
the attributes according to the justification entry.
'''
#-----------------------------------------------------------------------
# Create shortcut to self.__dict__
#-----------------------------------------------------------------------
D=self.__dict__
#-----------------------------------------------------------------------
# Make sure that caller is setting a legal attribute
#-----------------------------------------------------------------------
if attribute not in self._attributes:
raise AttributeError, "accpacPRbase: Illegal attribute=%s" % \
attribute
#-----------------------------------------------------------------------
# See if the attribute has a list of allowable values and check the
# value against that list.
#-----------------------------------------------------------------------
values=self._record[attribute]['values']
if values and (value not in values):
print "attribute=%s, value=%s not in " \
"values list=%s" % (attribute, value, values))
sys.exit(2)
#-----------------------------------------------------------------------
# Check the length of string attributes to make sure they are not
too
# long
#-----------------------------------------------------------------------
if not self._record[attribute]['fmt'] == '"%s"': D[attribute]=value
else:
max_length=len(self._record[attribute]['default'])
if len(value) > max_length:
print "attribute=%s, too long, " \
"value='%s' truncated" % (attribute, value))
print "len(%s)=%i, maxlen=%i" % \
(attribute, len(attribute), max_length))
#---------------------------------------------------------------
# Chop it off at the maximum length
#---------------------------------------------------------------
value=value[:max_length]
#-------------------------------------------------------------------
# See if this value is to be justified to the right, if so
justify
# it to the length of the default value.
#-------------------------------------------------------------------
if self._record[attribute]['justify']:
D[attribute]=value.rstrip().rjust(max_length)
else:
D[attribute]=value.rstrip().ljust(max_length)
return
def __str__(self):
'''
Create and return a string representation of this record for output.
'''
#-----------------------------------------------------------------------
# Create shortcut to self.__dict__
#-----------------------------------------------------------------------
D=self.__dict__
outputlist=[]
#-----------------------------------------------------------------------
# Loop over every possible attribute and use the defined format
string
# to format it for output record. Output record is then joined and
# returned.
#-----------------------------------------------------------------------
for attribute in self._attributes:
outputlist.append(self._record[attribute]['fmt'] % D[attribute])
return ''.join(outputlist)
OREC=outputrec()
OREC.Email='user at domain.com'
OREC.FirstName='Whatever'
#
# Put more assignments here
#
# or you can do in single statement
#
OREC=outputrec(Email='user at domain.com', FirstName='Whatever')
Then you can get output record as
f.writelines(OREC)
There are LOTS of ways to accomplish this and I've tried most of
them, this method just works well for me. I do basically the same
thing for the input records. Once the record classes are debugged
the program is very easy to write. It also "hides" the record
formats from the main program which makes it easy should they
change or if you want to support multiple formats (just write
another one for the new format and conditionally instantiate it
in your main program). I have actually created a base class for
this and I define my record class with it as the baseclass. This
way much of the baseclass code (that is debugged) is reusable.
Hope this helps.
Larry Bates
"mike beck" <cmichaelbeck at hotmail.com> wrote in message
news:ead93a6.0409210948.7a80e5db at posting.google.com...
> I'm writing a script to take a tab-delimited input file where the
> format is known and transform it to a fixed-width text file. I thought
> that the script might be easier to troubleshoot or edit later (if the
> file formats change, for example) if I specified the column names for
> the input and output files, which is why I didn't just map input
> column 1 -> export column 14.
>
> Even though I've spent some quality time with the archives and the
> fine manual, I'm pretty sure that I'm doing some WRONG THINGS. Or
> perhaps some SUB-OPTIMAL THINGS. And possibly REALLY WEIRD THINGS. Or
> just maybe THINGS THAT WILL WORK FOR A LITTLE WHILE BUT WILL SURELY
> CAUSE DIRE PROBLEMS AT SOME LATER, INCREDIBLY INCONVENIENT TIME
> RESULTING IN HORRIBLE, TRAGIC CONSEQUENCES FOR EVERYONE INVOLVED.
>
> The script works, but the use of multiple dictionaries seems ... how
> should I say ... unintuitive. Is there a better way to move values
> from one dictionary to another than what I've done here? I'd
> appreciate any other tips (especially those that fit into the
> CATEGORIES listed above).
>
> And I've omitted the read from file and write to file business for the
> sake of brevity...
>
> Standing on the shoulders of giants (meaning, you all!),
> Mike
>
> ---
>
> """
> Transforms a tab delimited text file to a fixed-width text file.
> """
> import copy, string
>
> DEFFIELDSEP = '\t';
> DEFLINESEP = '\n';
> DEFHASHEADER = 1; # 1 if yes, 0 if no
>
> import_field_order = ( 'Email',
> 'First Name(s)',
> 'Last Name',
> 'Address Line 1',
> 'Address Line 2',
> 'City',
> 'State',
> 'Zip Code', );
>
> # export columns (keys) to import columns (values)
> mapped_fields = { 'EMAIL_ADDRESS_1':'Email',
> 'FIRST_NAME_1':'First Name(s)',
> 'LAST_NAME_1':'Last Name',
> 'ADDRESS_1':'Address Line 1',
> 'CITY':'City',
> 'STATE':'State',
> 'ZIP_CODE':'Zip Code', };
>
> # required length and justification of export fields
> export_fields = { 'RECORD_TYPE':[1,'left'],
> 'ACCOUNT_ID':[8,'left'],
> 'TITLE_1':[8,'left'],
> 'FIRST_NAME_1':[20,'left'],
> 'MIDDLE_NAME_1':[20,'left'],
> 'LAST_NAME_1':[20,'left'],
> 'SUFFIX_1':[8,'left'],
> 'NAME_ID_1':[2,'left'],
> 'ADDRESS_1':[256,'left'],
> 'CITY':[30,'left'],
> 'STATE':[2,'left'],
> 'ZIP_CODE':[5,'left'],
> 'ZIP_EXTENSION':[4,'left'],
> 'EMAIL_ADDRESS_1':[100,'left'],
> 'EMAIL_ADDRESS_TYPE_1':[2,'left'], };
>
> export_field_order = ( 'RECORD_TYPE',
> 'ACCOUNT_ID',
> 'TITLE_1',
> 'FIRST_NAME_1',
> 'MIDDLE_NAME_1',
> 'LAST_NAME_1',
> 'SUFFIX_1',
> 'NAME_ID_1',
> 'ADDRESS_1',
> 'CITY',
> 'STATE',
> 'ZIP_CODE',
> 'ZIP_EXTENSION',
> 'EMAIL_ADDRESS_1',
> 'EMAIL_ADDRESS_TYPE_1', );
>
> def pad ( elem, max_length, justification=None):
> """
> Pad a string to a predetermined length. Truncate the string
> if necessary. By default, left justify and pad with spaces.
> """
> padded_value = '';
> # what if elem is none?
> if (justification is 'right'):
> padded_value = elem.rjust(max_length)[:max_length]
> else:
> padded_value = elem.ljust(max_length)[:max_length]
> return padded_value
>
> def glue ( var, ref, order ):
> """
> Build a string with padded dictionary values.
> Var and ref are dictionaries, order is a tuple describing the
> field order.
> """
> finished_line = []
> # pad the items
> for k in var.keys():
> # pad (value, length, justification)
> var[k] = pad( str(var[k]), ref[k][0], ref[k][1])
> # build the line in order
> for elem in order:
> finished_line.append(dict.get(var, elem))
> # join the elements and return a string
> return ''.join([x for x in finished_line])
>
> def build ( line, field_sep = DEFFIELDSEP ):
> """
> Create a single record by transposing fields via
> multiple dictionaries.
> """
> fields = string.split(line, field_sep);
> # trim whitespace
> fields = map(string.strip, fields)
> # create dict of import column names (keys) and import
> # data (values)
> step_one = dict([(q,a) for q,a in zip(import_field_order,
> fields)])
> # create a 'throwaway' copy of the mapped fields dict to
> # do the transform
> step_two = copy.deepcopy(mapped_fields)
> # replace the column names of the import file with the
> # values from the import file
> for k,v in step_two.items():
> step_two[k] = dict.get(step_one, v)
> # create an empty dict with the export column names
> step_three = dict([(q,'') for q in export_field_order])
> dict.update(step_three, step_two)
> # special cases
> step_three['RECORD_TYPE'] = 'D'
> step_three['EMAIL_ADDRESS_TYPE_1'] = 'I'
> return glue(step_three, export_fields, export_field_order)
>
> def decapitate(infile,has_header=DEFHASHEADER,line_sep=DEFLINESEP):
> """
> Return file without header row
> """
> if (has_header):
> sans_head = string.split(infile, line_sep)[1:]
> else:
> sans_head = string.split(infile, line_sep)
> return sans_head
>
> if __name__ == '__main__':
>
> # init vars
> out_lst = []
>
> # sample data. Names have been changed ... ;)
> fin = {'data': 'Email\tFirst Name(s)\tLast Name\tAddress Line
> 1\tAddress Line 2\tCity\tState\tZip
> Code\nme at test.com\tBill\tMunroe\t416
> Main\t\tManhattan\tNY\t10471\ngodaddy at test.com\tGeorge\tScott\t111
> Blue Street 2005\t\tSan
> Francisco\tCA\t94144\nmmike at test.net\tMike\tBork\t22 Oak
> Rd\t\tAlbuquerque\tNM\t01720-5303\npsycho at test.com\tIma\tCrazy\t1111 E
> Maple Rd\t\tDenver\tCo\t80206-6139'};
>
> # get a list of lines without a header row
> lines = decapitate(fin['data']);
>
> for line in lines:
> # check for blank lines
> if (line):
> out_lst.append(build(line))
>
> print out_lst;
More information about the Python-list
mailing list