4
\$\begingroup\$

I learnt to code a long, long time ago.

Concerns

Existing approach is messy and will be hard to maintain. System errors are not trapped and the error messages could be better implemented. Final hashcode calculation occurs after the data file is written.

I have been through a few python tutorials some time ago but it is hard without a real-world problem. I'd like to use this as an OOP learning opportunity. This code actually does the job and is well on the way to finalising a file validation feature yet to be implemented.

Everything is up for review here.

Situation

Data files are created by a road model and this program rapidly creates scenarios for option testing. It involves overwriting two 'barrier' fields and and one of two overtaking lane fields in the road (ROD) file. I don't understand or need to care about the many other fields but they must not be changed as some involve complex calculations. A hashcode is being introduced to prevent inadvertent ROD file changes but they must remain human readable. Two other files form part of the data set but the changes are trivial - renaming both files and replacing the first line of one with the filename.

So, a database file properties are a filename, with the filename and description on the first two lines then eight other lines before the 109 character records start. The first six characters in each record are the distance (starting at some arbitrary value near 0.0) incremented by 0.1 in each record. Whole numbers have no decimal or trailing zero. There are two other files in the dataset with different extensions.

The program provides a service of creating a new option from an existing road file set. (New option could be updating existing dataset.) The file and command line are checked and the road file data structure analysed. The hashcode register provides a validation and updating service.

#!/usr/bin/python
# With existing road file create new overtaking lane option file(s).
#
# Copy file, insert overtaking lane and barrier line codes for specified
# chainage, replace first line with new filename, second line with description
# and update MD5 HashCode (future).
#
# eg python OptionEditor.py Demo.ROD 1.5 2.2 P Option.ROD Destination_Description
#
# By Neale Irons version 25/02/2018 (CC BY-SA 4.0)


import sys  
import os
import shutil
import tempfile
import hashlib

def main():  

    def copyFile(src, dest):
    # From https://www.pythoncentral.io/how-to-copy-a-file-in-python-with-shutil/
        try:
            shutil.copy(src, dest)
        # eg. src and dest are the same file
        except shutil.Error as e:
            print('Error: %s' % e)
        # eg. source or destination doesn't exist
        except IOError as e:
            print('Error: %s' % e.strerror)
        return

    def HashCode_register(HASHFILE, src, action):
#        with open (hashfile) as hf:
#        hash, fn = (open(HASHFILE).read().split('\n').split(' *')
        hashcode = hashlib.md5(open(src, 'rb').read()).hexdigest()
        print hashcode+ " *" + src                                  # Debug ****
        lines = open(HASHFILE).read().split('\n')
        for item in lines:
            if action == 'validate':
                if hashcode in item: # Maybe uppercase compare
                    return True
            elif action == "update":
                if src in item: # Maybe uppercase compare
                    print lines                                      # Debug - no-op
#                    if item_number_hashcode != hashcode:            # FIX THIS ***
#                        item_hashcode(update_number) = hashcode     # need write access here FIX THIS ***
#                else:
#                    write_append_HASFILE(hashcode + " *" + src)     # need write access here FIX THIS ***
#                    return True
        return False


    def replace_textfile_line(text_filename, new_text, description):
    # Replace first line with new_text.
    #
    # Adapted from ReplFHdr.py
    # By Neale Irons version 14/02/2018 (CC BY-SA 4.0)
    # import sys, os, tempfile
        with open(text_filename) as src:
            with tempfile.NamedTemporaryFile('w',
              dir=os.path.dirname(text_filename), delete=False) as dst:
                line = src.readline()               # Read first line
                dst.write(new_text + '\n')          # Replace first line
                if description:                     # If description not empty
                    line = src.readline()           # Read second line
                    dst.write(description + '\n')   # Replace second line
                shutil.copyfileobj(src, dst)        # Copy rest of file
        os.unlink(text_filename)                    # remove old version
        os.rename(dst.name, text_filename)          # rename new version
        return()


    #Constants
    ODO_STEP_KM = 0.1
    RECORD_SIZE = 109
    BARRIER_COLUMN1 = 11
    BARRIER_COLUMN2 = 16
    BARRIER_VALUE = '+' # Debug - use '+' instead of '-'
    OTLANE_COLUMN1 = 22
    OTLANE_COLUMN2 = 27
    OTLANE_VALUE = 'Y'  # Debug - use 'Y' instead of 'T'
    OTLANE_DIRECTION_CODES = "PC"
    HASHFILE = "Hashcode.MD5"
    True = not False

    # Validate command line
    if len(sys.argv) < 6 or len(sys.argv) > 7:
        format_string = ('  Usage: python {} Source.ROD Start_km End_km P|C Destination.ROD ["Destination_Description"]')
        print(format_string.format(sys.argv[0]))
        print('    Use existing road files to create new overtaking lane option and optionally insert second line description.)')
        sys.exit()

    src = sys.argv[1]
    otl_start_km = float(sys.argv[2])
    otl_end_km = float(sys.argv[3])
    otl_direction = sys.argv[4]
    dest = sys.argv[5]
    if len(sys.argv) < 6:
        description = sys.argv[6]
    else:
        description = ''

    if otl_start_km > otl_end_km:
        print("Invalid start/end overtaking lane chainage - start must be less than end. Exiting...")
        sys.exit() 

    if otl_direction not in OTLANE_DIRECTION_CODES:
        print("Direction {} is invalid, must be P or C. Exiting..."
          .format(sys.argv[4]))
        sys.exit() 


    # Validate file
    if not os.path.isfile(src):
        print("File path {} does not exist. Exiting..."
          .format(src))
        sys.exit()

    if src <> dest:
        # Validate support files
        src_name, src_ext = os.path.splitext(src)
        if not os.path.isfile(src_name+'.MLT'):
            print("Requires support file {}. Exiting..."
              .format(src_name+'.MLT'))
            sys.exit()
        if not os.path.isfile(src_name+'.OBS'):
            print("Requires support file {}. Exiting..."
              .format(src_name+'.OBS'))
            sys.exit()

    # Validate HashCode
    if not HashCode_register(HASHFILE, src, "validate"):
        print("Invalid file - failed HashCode. Exiting...")

    # Read header
    with open(src) as fp:
        # *** if fails raise "Invalid file format - first 10 lines must be file header
        for x in range(1, 10):
            line = fp.readline()
        header_length=fp.tell() # Calculate header length
        fp.seek(0,2)
        file_length = fp.tell()
    fp.close()

    if (file_length - header_length) % RECORD_SIZE != 0:
        print("Invalid file format - must be 10 lines and {} character records. Exiting..."
          .format((file_length - header_length) % RECORD_SIZE))
        sys.exit()

    nrecs = (file_length - header_length) // RECORD_SIZE
    if nrecs < 1:
        print("Invalid file format - file must contain at least one record. Exiting...")
        sys.exit()

    # dest_tmp = NamedTemporaryFile(delete=False)
    dest_tmp = src + ".tmp"
#    print("Temp file: {}".format(dest_tmp))   # Debug
    copyFile(src, dest_tmp)


    # Open file for random read/write
    with open(dest_tmp, 'r+b') as fp:
        fp.seek(header_length,0)
        odo_start_km = float((fp.read(6)))        
        fp.seek(RECORD_SIZE * -1, 2)
        odo_end_km = float((fp.read(6)))

        if otl_start_km <= odo_start_km:
            print("Invalid start overtaking lane chainage - Overtaking lane can't start before first road chainage. Exiting...")
            sys.exit()

        if otl_end_km >= odo_end_km:
            print("Invalid end overtaking lane chainage - Overtaking lane can't ends after last road chainage. Exiting...")
            sys.exit()

        if int(round((odo_end_km - odo_start_km)/ODO_STEP_KM) + 1) != nrecs:
            print("Invalid file format - incorrect number of lines between start and end distances. Exiting...")
            sys.exit()

        while otl_start_km < otl_end_km:
            fpos=int(round(header_length + ((otl_start_km - odo_start_km) / ODO_STEP_KM * RECORD_SIZE) - 1))

            fp.seek(BARRIER_COLUMN1 + fpos)
            fp.write(BARRIER_VALUE)
            fp.seek(BARRIER_COLUMN2 + fpos)
            fp.write(BARRIER_VALUE)

            if otl_direction == "P":
                fp.seek(OTLANE_COLUMN1 + fpos)
            elif otl_direction == "C":
                fp.seek(OTLANE_COLUMN2 + fpos)
            fp.write(OTLANE_VALUE)
            otl_start_km += ODO_STEP_KM

        fp.close()

    if dest <> src :
        # Create new supporting files
        dest_name, dest_ext = os.path.splitext(dest)
        copyFile(src_name+'.MLT', dest_name+'.MLT')
        replace_textfile_line(src_name+'.MLT', src_name+'.MLT', '')
        copyFile(src_name+'.OBS', dest_name+'.OBS')        

    replace_textfile_line(dest_tmp, dest, description)
    HashCode_register(HASHFILE, src, "update")                          # Not working yet ****    

    from shutil import move
    move(dest_tmp, dest)

if __name__ == '__main__':
    main()

Data record snippet (not much happening in this sample):

   0.8    -1   -1    F    F    9    9
   0.9    -1   -1    F    F   10  201
     1    -1   -1    F    F   11  202
   1.1    -1   -1    F    F   12   12
   1.2     1   -1    F    F   13   13
   1.3     1   -1    F    F   14   14
   1.4     1   -1    F    F   15   15
   1.5    -1    1    F    F   16   16
   1.6    -1    1    F    F   17   17
   1.7    -1    1    F    F   18   18
   1.8    -1   -1    F    F   19   19
   1.9    -1   -1    F    F   20   20
     2    -1   -1    F    F   21   21
   2.1    -1   -1    F    F   22   22
   2.2    -1   -1    F    F   23   23
   2.3    -1   -1    F    F   24   24

Hashcode.MD5:

8d443f2e93a3f0b67f442e4f1d5a4d6d *md5.exe
eb574b236133e60c989c6f472f07827b *md5sum.exe
c1f62b08d050f2a30790c5b2f3d29f5c *Demo.ROD
9f036e40755daaa5a5fc141d3cbfbb99 *nGetPid.exe
\$\endgroup\$
  • \$\begingroup\$ Is this code working? FIX THIS and Not working yet make it seem that this code isn't ready for review yet. \$\endgroup\$ – Gerrit0 Feb 25 '18 at 22:17
  • \$\begingroup\$ Yes - I am using this code and guarantee it works. New functionality to extend Hashcode_Register for update action is under development and not implemented yet. Those are the comments you are referring to. Making these changes prompted me to seek a code review. \$\endgroup\$ – nealei Feb 26 '18 at 1:37
5
\$\begingroup\$
#!/usr/bin/python

Consider using #! /usr/bin/env python instead (or python3). Then you can manage import dependencies with venv or, better, conda env update. Conda will adjust PATH to point at the appropriate python interpreter which has the necessary pypi libraries that inevitably get added to a project.

# By Neale Irons version 25/02/2018 (CC BY-SA 4.0)

You probably wanted to use the word 'Copyright', if you want to assert IP rights and modify them using Creative Commons.

def main():

The main function is entirely too long. Limit yourself to one screenful per function. At a minimum, promote any helper functions that might possibly be useful for other functions to call or re-use.

def copyFile(src, dest):

Please follow PEP-8 naming conventions, and call it copy_file.

    try:
        shutil.copy(src, dest)

Joey Payne was just illustrating exceptions that one might want to catch. Here, it's not clear that you really want to catch them, given that you don't actually "handle" the error, you merely swallow it after emitting some noise on stdout. If copy_file was merely an alias for copy, that would be fine. But here, it appears to me that copy_file is actually worse, it manages to do a disservice for the caller, by swallowing exceptions. The usual post-condition would be "this method guarantees dest file exists upon return", but as written the post-condition is empty set, the method makes no correctness guarantees at all. That makes things harder for maintenance programmers reading any calling code a few months down the road.

def HashCode_register(HASHFILE, src, action):

Ok, now you're just going crazy with case. Use lowercase with underscores, please. Yes, I know you're passing in a manifest constant in the caller code. But the function definition should use lower case for the formal parameter.

#        with open (hashfile) as hf:

I like with! Why did you choose to take on the responsibility of issuing a close(), and then failed to close hf?

    print hashcode+ " *" + src

You appear to be using python2. I urge you to use python3 instead. It makes some things simpler, for example getting I/O encode / decode right. Run flake8 *.py on your sources, and follow its advice regarding whitespace around the + operator.

    lines = open(HASHFILE).read().split('\n')

This is equivalent to the .readlines() function. But it would be more natural (and less memory intensive) to phrase it like this:

    with open (hashfile) as hf:
        for line in hf:
            line = line.rstrip()
            ...

To loop you write:

    for item in lines:

This is a little odd. The usual idiom would be:

    for line in lines:

In general, if you have some plural iterable xs, iterate with for x in xs.

        if action == 'validate': ...
        elif action == "update":

There are only two values, which suggests you might choose to pass in a boolean. If three values are needed, consider using enum: https://docs.python.org/3/library/enum.html

Ok, I read the function, and I still don't know what it does. Please write a docstring explaining what promise it makes about its boolean return value. The 'update' case, given the current non-comment code, appears to be a constant False function.

In replace_textfile_line, delete the return() line.

True = not False

That's not funny. Do not redefine builtin values with your own computed values. And don't even think about assigning 3.2 to math.pi as the Indiana legislature once did.

if len(sys.argv) < 6 or len(sys.argv) > 7:

Your command line appears to be sufficiently complex to warrant the use of a library: https://docs.python.org/3/library/argparse.html Plus, then you get --help for free!

    format_string = ('  Usage: python {} Source.ROD Start_km End_km P|C Destination.ROD ["Destination_Description"]')

The identifier you were looking for was: usage = ...

if not os.path.isfile(src):

This is correct. But, consider letting open do the validation for you. If the file can't be read, we'll find out soon enough. Same remarks for your subsequent file validations.

if src <> dest:

This is correct but odd. Please use the modern != operator.

if not HashCode_register(HASHFILE, src, "validate"):
    print("Invalid file - failed HashCode. Exiting...")

I don't believe you. I don't see an immediate exit. Consider turning that function into one that raises a fatal exception upon finding a file that doesn't validate.

    ...
    header_length=fp.tell()

This logic is convoluted. Please write a helper function for it. If reading the entire file is acceptable, then this might suffice:

    header_length = len(''.join(fp.readlines()[:10]))
    file_length = os.path.getsize(src)

The with handler is a good style - keep using it.

fp.close()

fp is already closed at this point, thanks to the with.

if (file_length - header_length) % RECORD_SIZE != 0:

This is OK as far as it goes, but I would prefer to see a more helpful diagnostic, e.g.

lines = fp.readlines()
for line in lines[:10]:
    if len(line) != RECORD_SIZE:  # then show some helpful diagnostic

Oh, wait, I see the whole thing must conform. So use for line in lines:. Also, nrec might come from len(lines).

    fp.seek(RECORD_SIZE * -1, 2)

You're making me nervous here. Using lines[i] and lines[j] would be more robust, in case short or long lines crept into the input file. You seem to believe that every 109th character will be a newline, but you don't validate that.

        fp.seek(BARRIER_COLUMN1 + fpos)
        fp.write(BARRIER_VALUE)

Maybe you have a giant file and it's import to patch it up in-place. But your code would be more testable / readable / maintainable if it read lines from a source file and wrote modified lines to a dest file, perhaps with a file rename upon success. You used that pattern very nicely in a previous code fragment. I'm concerned about the possibility of silently corrupting a file you haven't validated. The current code might work great, but after months go by, along with personnel, someone will be faced with a maintenance task that is harder than it needs to be. After all, sometimes file formats change, and fixed record size of 109 might not last until the end of time.

if dest <> src :

Please use !=, and pay attention to flake8 when it tells you to delete the blank before the colon.

from shutil import move

Please keep your imports at top of file. The identifier is generic enough that it might be clearer to phrase it as import shutil and then shutil.move(...)

Keep at it! Getting people to read your code will have a good effect on the clarity of code you write in future.

\$\endgroup\$
  • \$\begingroup\$ Python 3 is not an option and I don't know usual python so your comments, especially flake8 advice, are excellent guidance. Exactly which copy do you recommend? Why wouldn't you read hashcode file into memory to update (replace) hashcode? Is there a nicer way to handle those verification errors? (Say msg strings and all exit in same function.) I suppose system errors should generate 'sys.exit(1)'. The datafile and hashcode files are objects - main question: Is there a better OOP way? \$\endgroup\$ – nealei Mar 1 '18 at 21:29
  • \$\begingroup\$ @nealei Python 3 should be an option. Especially for new code. \$\endgroup\$ – Mathias Ettinger May 7 '18 at 8:29
  • \$\begingroup\$ No - restricted to 2.7 on corporate system. \$\endgroup\$ – nealei May 7 '18 at 8:56

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.