'''Support for importing files with list of tags.'''

import os
import re
import traceback

def nicesplit(string):
    '''Split string at commas or spaces, do not mind multiple delimiters.'''
    space_sep = string.replace(',', ' ')
    all_parts = space_sep.split(' ')
    
    # Return non-empty parts
    return [p.strip() for p in all_parts if p.strip()]

NAME_REGEXP = re.compile('^[a-zA-Z0-9_]+(/[a-zA-Z0-9_]+)*$')
def validname(string):
    '''Check if a string is a valid tag name.'''
    return bool(NAME_REGEXP.match(string))

class TagCollection:
    '''A collection of tags and associated files.'''
    
    def __init__(self):
        # Mapping of tags to set of paths
        self.tags = {}
        
        # List of import messages
        self.msgs = []
    
    def import_tagfile(self, infile, relpath):
        '''Import a file with tags. Relpath is the path to the directory.'''
        for line in infile:
            if ':' in line:
                # Individual files listed after :
                parts = line.split(':', 1)
                tags = nicesplit(parts[0])
                files = nicesplit(parts[1])
                
                # Add directory to the file names
                targets = [os.path.join(relpath, f) for f in files]
            else:
                # Tags listed for the whole folder
                tags = nicesplit(line)
                targets = [relpath]
            
            # Strip any .. or other evil things from the path
            targets = [os.path.normpath(p.replace('..', '')) for p in targets]
            
            for tag in tags:
                if not self.tags.has_key(tag):
                    if not validname(tag):
                        self.msgs.append(infile.name + ': Invalid tag ' + tag)
                        continue
                    
                    self.tags[tag] = set(targets)
                else:
                    already_exists = self.tags[tag].intersection(targets)
                    if already_exists:
                        self.msgs.append(infile.name +
                            ': Following targets already exist for tag ' +
                            tag + ': ' + ', '.join(already_exists)
                        )
                    
                    self.tags[tag].update(targets)

    def import_recursively(self, basepath, tagfilename):
        '''Import all tag files recursively from basepath.'''
        
        def callback(arg, dirname, fnames):
            '''os.walk callback function to read tagfiles.'''
            if tagfilename in fnames:
                assert dirname.startswith(basepath)
                relpath = dirname[len(basepath):].lstrip('/')
                
                tagfilepath = os.path.join(dirname, tagfilename)
                try:
                    tagfile = open(tagfilepath, 'rU')
                    self.import_tagfile(tagfile, relpath)
                except:
                    self.msgs.append('Importing tagfile ' + tagfilepath +
                        ' failed with exception:\n' + traceback.format_exc())
        
        os.path.walk(basepath, callback, None)