Multiple keys and values

Question

I'm a bit stuck on a project.. Basically, I get a directory to scan for files that passes certain filters.

The command files contains lines with filtering (given filters name are stored in an enum file) instructions. like this for example: suffix%txt exec%YES

if that's all, it will return all the files ending with txt (extension) and are executable.. so far no problem.

the problem starts with line like this: suffix%txt exec%YES (on the same line) in this case, it should return all the files that ends with txt OR executable..

I'm splitting the line using String.split("%") and getting it into a map with a key and value, than iterate over each key and checks what filter in the enum it is, and perform the desire checking.

I'm kinda stuck on how to identify that when i have more than 1 filter per line. I tried doing a HashMap with the 1st filter as the key, and the value is a list that contains all the other (using split(" ") for breaking up the filters.. I can't rely on the keys being on the even or odd indexes, because a filter may have another %NOT at the end of it (suffix%txt%NOT) which will return all the files that doesn't end with txt...

Any help would be appreciated! Thanks!

jacobm · Accepted Answer · 2012-03-09 03:04:02Z

2

How about this: create a Filter interface, and write a function that parses a file into a single filter based on the contents of the file. You'd have something like:

interface Filter {
  boolean passesFilter(File file);
}

class SuffixFilter implements Filter {
  SuffixFilter(String suffix) { ... }
  public boolean passesFilter(File file) {
    // return true if file has the appropriate suffix
  }
}

class ExecutableFilter implements Filter {
  ...  // filter that returns true if the file is executable
}

// now for the interesting part ...
class NegationFilter implements Filter {
  private final Filter subfilter;
  NegationFilter(Filter subfilter) {
    this.subfilter = subfilter;
  }

  public boolean passesFilter(File file) {
    return !subfilter.passesFilter(file);
  }
}

class AndFilter implements Filter {
  private final Collection<Filter> subfilters;
  AndFilter(Collection<Filter> subfilters) {
    this.subfilters = subfilters;
  }

  public boolean passesFilter(File file) {
    for (Filter subfilter : subfilters) {
      if (!subfilter.passesFilter(file)) {
        return false;
      }
    }
    return true;
  }
}

class OrFilter implements Filter {
  private final Collection<Filter> subfilters;
  OrFilter(Collection<Filter> subfilters) {
    this.subfilters = subfilters;
  }

  public boolean passesFilter(File file) {
    for (Filter subfilter : subfilters) {
      if (subfilter.passesFilter(file)) {
        return true;
      }
    }
    return false;
  }
}

With this in place, you just need to build up all the basic filters, and then adjacent elements on the same line get OrFiltered together while filters on separate lines (or-filtered or not) get AndFiltered together. Here's a sketch:

Filter readAndFilter(Iterable<String> fileLines) {
  List<Filter> subfilters = new ArrayList<Filter>();
  for (String line : fileLines) {
    subfilters.add(readOrFilter(line));
  }
  return new AndFilter(subfilters);
}

Filter readOrFilter(String fileLine) {
  List<Filter> subfilters = new ArrayList<Filter>();
  for (String oneFilter : fileLine.split(" ")) {
    Filter filter = buildOneFilter(oneFilter);
    subfilters.add(filter);
  }
  return new OrFilter(subfilters);
}

Filter buildOneFilter(String oneFilterClause) {
  // parse as you were doing before
}

You'd call readAndFilter on lines read out from your file, and what you get back is always a single Filter that you can pass files to and it'll tell you whether they pass the filter. (Note: if you wanted to, you could special-case my readAndFilter and readOrFilter methods to check if the list they're going to return is of length 1, and if so just return that one filter rather than an And or Or of one element. It doesn't affect correctness, but it might make debugging output a little easier to read.)

edited Mar 9, 2012 at 3:04

answered Mar 9, 2012 at 1:26

jacobm

14.1k1 gold badge27 silver badges29 bronze badges

Sign up to request clarification or add additional context in comments.

7 Comments

La bla bla Over a year ago

Thanks for the long answer, I think I like the way you're trying to go with, but I still have a few unclear things I'm still not quite sure on how to implement it. I got a txt files, I can create for each one of them a corresponding filter object. but from here.. if " " found (more than 2 filters per line) create a new OrFilter object. and at the end of scanning all the lines, call the AndFilter object? And thanks again, will give it a try

jacobm Over a year ago

Edited my answer to sketch out how you'd parse a file to produce a filter.

La bla bla Over a year ago

Thanks! sounds like a great idea. I'm gonna implement this.

La bla bla Over a year ago

Now that i'm thinking of it, isn't it a bit not efficient? cause for every single file in the source directory, including sub-directories, u create all the filter objects from the command file? just a thought.

jacobm Over a year ago

There's no reason I see why this strategy would make you do any more parsing of the rules definitions than any other strategy would. Also I suspect that the disk and system calls you'll be making will completely dwarf any extra costs you incur in parsing (milliseconds versus nanoseconds or microseconds).

|

Collectives™ on Stack Overflow

Multiple keys and values

1 Answer 1

7 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

7 Comments

Your Answer

Sign up or log in

Post as a guest

Related