Difficulty utilizing Python struct.Struct.unpack_from with different format strings

Question

First time poster, long-time lurker. Have searched high and low for an answer to this but it's got to that stage...!

I am having some trouble implementing the answer given by John Machin to this past question:

How to efficiently parse fixed width files?

At a very high level I am using this code to split up fixed format text files and import them into a PostgreSQL database. I have successfully used this code to implement the solution for one text file, however I am now trying to expand my program to work with different text files with different fixed formats, and am continuously running into the same error:

struct.error: unpack_from requires a buffer of at least [x] bytes

Of course, I get a different value for x depending on the format string I am feeding to the function - my problem is that it continues to work for one and only one format, and not any others. The only thing I am changing is the variable used to calculate the format string, and the variable names in the script which relate to the format.

So for example this works fine:

cnv_text = lambda s: str(s.strip())
cnv_int = lambda s: int(s) if s.isspace() is False else s.strip()
cnv_date_ymd = lambda s: datetime.datetime.strptime(s, '%Y%m%d') if s.isspace() is False else s.strip() # YYYY-MM-DD

unpack_len = 0
unpack_fmt = ""
splitData = []

conn = psycopg2.connect("[connection info]")
cur = conn.cursor()

Table1specs = [
    ('A', 6, 14, cnv_text),
    ('B', 20, 255, cnv_text),
    ('C', 275, 1, cnv_text),
    ('D', 276, 1, cnv_text),
    ('E', 277, 1, cnv_text),
    ('F', 278, 1, cnv_text),
    ('G', 279, 1, cnv_text),
    ('H', 280, 1, cnv_text),
    ('I', 281, 8, cnv_date_ymd),
    ('J', 289, 8, cnv_date_ymd),
    ('K', 297, 8, cnv_date_ymd),
    ('L', 305, 8, cnv_date_ymd),
    ('M', 313, 8, cnv_date_ymd),
    ('N', 321, 1, cnv_text),
    ('O', 335, 2, cnv_text),
    ('P', 337, 2, cnv_int),
    ('Q', 339, 5, cnv_int),
    ('R', 344, 255, cnv_text),
    ('S', 599, 1, cnv_int),
    ('T', 600, 1, cnv_int),
    ('U', 601, 5, cnv_int),
    ('V', 606, 10, cnv_text)
    ]

#for each column in the spec variable...
for column in Table1specs:
    start = column[1] - 1
    end = start + column[2]
    if start > unpack_len:
    unpack_fmt += str(start - unpack_len) + "x"
    unpack_fmt += str(end - start) + "s"
unpack_len = end
field_indices = range(len(Table1specs))
print unpack_len, unpack_fmt
#set unpacker
unpacker = struct.Struct(unpack_fmt).unpack_from

class Record(object):
    pass

filename = "Table1Data.txt"

f = open(filename, 'r')
for line in f:
    raw_fields = unpacker(line)
    r = Record()
    for x in field_indices:
        setattr(r, Table1specs[x][0], Table1specs[x][3](raw_fields[x]))
    splitData.append(r.__dict__)

All the data is appended to splitData, which I then cycle through in a loop and work into SQL statements for input into the database via psycopg2. When I change the specs to something else (and the other variables also to reflect this), then I receive the error. It is thrown from the 'raw_fields = unpacker(line)' line.

I have exhausted all resources and am at a loose end... any thoughts or ideas welcomed.

(Could it be to do with the text file I am importing from?)

Best regards.

Answer 1

Have since solved this issue: Problem was being caused by the text files I was parsing. The lines were not long enough so I have written a function that writes spaces to the end of each line to make them the correct length:

def checkLineLength(checkFile, minLength):
    print ('Checking length of lines in file '+ checkFile+', where minimum line length is '+str(minLength))
    counter = 0
    fixedFile = 'fixed'+checkFile
    src = open(checkFile, 'r')
    dest = open(fixedFile, 'w')
    lines = src.readlines()
    for line in lines:
        if len(line) < minLength:
            x = (line.rstrip('\r\n') + (" "*(minLength-(len(line)-1))+'\r\n'))
            dest.write(x)
            counter += 1
        else:
            dest.write(line)
    if counter > 0:
        os.remove(checkFile)
        os.rename(fixedFile, checkFile)
        print (str(counter) + " lines fixed in "+ checkFile)

else:
    print('Line length in '+checkFile+' is fine.' )
    os.remove(fixedFile)