import re
from sqlalchemy import create_engine, MetaData, Table

__author__ = "Gavin Huttley"
__copyright__ = "Copyright 2007-2012, The Cogent Project"
__credits__ = ["Gavin Huttley"]
__license__ = "GPL"
__version__ = "1.5.3"
__maintainer__ = "Gavin Huttley"
__email__ = "Gavin.Huttley@anu.edu.au"
__status__ = "alpha"

class DisplayString(str):
    """provides a mechanism for customising the str() and repr() of objects"""
    def __new__(cls, arg, num_words=None, repr_length=None, with_quotes=False):
        new = str.__new__(cls, str(arg))
        new.num_words = num_words
        new.repr_length = repr_length or len(str(arg))
        new.with_quotes = with_quotes
        return new
    
    def __repr__(self):
        if self.num_words is not None:
            new = " ".join(self.split()[:self.num_words])
        elif self.repr_length != len(self):
            new = self[:self.repr_length]
        else:
            new = self
        if len(self) > len(new):
            new += '...'
        new = [new, "'%s'" % new][self.with_quotes]
        return new
    

class CaseInsensitiveString(str):
    """A case insensitive string class. Comparisons are case insensitive."""
    def __new__(cls, arg, h=None):
        n = str.__new__(cls, str(arg))
        n._hash=hash(''.join(list(n)).lower())
        n._lower = ''.join(list(n)).lower()
        return n
    
    def __eq__(self, other):
        return self._lower == ''.join(list(other)).lower()
    
    def __hash__(self):
        # dict hashing done via lower case
        return self._hash
    
    def __str__(self):
        return ''.join(list(self))

class LazyRecord(object):
    """a convenience class for conducting lazy evaluations of class
    properties"""
    NULL_VALUE = None
    def __init__(self):
        """blind constructor of caches"""
        self._cached = {}
        self._table_rows = {}
    
    def _get_cached_value(self, attr_name, get_attr_func):
        if attr_name not in self._cached:
            get_attr_func()
        
        return self._cached[attr_name]
    
    def _populate_cache_from_record(self, attr_column, table_name):
        """attr_column: the attribute name <-> table column key mappin
        table_name: the key in _table_rows"""
        table = self._table_rows[table_name]
        for attr, column, func in attr_column:
            if attr not in self._cached:
                self._cached[attr] = func(table[column])
    
    def _set_null_values(self, attrs, table_name = None):
        for attr in attrs:
            self._cached[attr] = self.NULL_VALUE
        
        if table_name:
            self._table_rows[table_name] = self.NULL_VALUE
    

class NoItemError(Exception):
    def __init__(self, value):
        self.value = value
    
    def __str__(self):
        return repr(self.value)
    

def convert_strand(val):
    """ensures a consistent internal representation of strand"""
    if isinstance(val, str):
        assert val in '-+', 'unknown strand "%s"' % val
        val = [-1,1][val == '+']
    elif val is not None:
        val = [-1,1][val > 0]
    else:
        val = 1
    return val

def asserted_one(items):
    """asserts items has a single value and returns it"""
    one = False
    for item in items:
        if one:
            raise ValueError('More than one: [%s]' % item.items())
        one = True
    if one:
        return item
    else:
        raise NoItemError('No items')

def what_columns(table):
    """shows what columns are in a table"""
    print [c.name for c in table.c]

def yield_selected(sqlalchemy_select, limit=100):
    """takes a SQLAlchemy select condition, yielding limit per db query
    purpose is to not get all records from the db in one go
    """
    offset = 0
    while 1:
        results = sqlalchemy_select.offset(offset).limit(limit).execute()
        count = 0
        for result in results:
            yield result
            count += 1
        offset += limit
        if count == 0 or count < limit:
            break