old_fasta_py.txt

def read_fasta_header(f):
    '''
    read the next fasta header
    file must be at the start of the next record
    or at end of file
    return header
    header is the full header line
    do not mess about with parsing the header line at all
    '''

    line = f.readline()
    
    #end of file
    if line == '': return None

    #check the line looks like a fasta header
    assert line.startswith('>'), 'expecting fasta header: %s'%line

    #remove the leading '>' character and trailing whitespace
    header = line[1:]
    header = header.strip()

    rec = {}
    
    #header is the complete line, minus the '>'
    rec['header'] = header
    
    return rec

def next_seqline(f):
    '''
    generator function to read next sequence line
    '''
    
    while True:
        line = read_seqline(f)
        
        #end of sequence lines
        if line == None: break
        
        #yield next line
        yield line
    
def read_seqline(f):
    '''
    read lines until end of file or next fasta header
    '''
    
    while True:
        posn = f.tell()
        line = f.readline()
        
        #end of file or reached next fasta record
        if line == '' or line.startswith('>'):
            f.seek(posn)
            return None
        
        #skip blank lines
        line = line.strip()
        if line == '': continue
        
        #remove any internal spaces or tabs
        line = line.replace(' ','')
        line = line.replace('\t','')
        
        #return sequence line
        return line

def read_fasta(f,storeseq=True):
    '''
    read next fasta record
    file must be at start of record or end of file
    '''
    
    #read header into record
    rec = read_fasta_header(f)
    
    #end of file
    if rec == None: return None
        
    if storeseq:
        #read in the sequence data
        rec['seq'] = ''.join([x for x in next_seqline(f)])
        rec['len'] = len(rec['seq'])
    else:
        #determine length of sequence without storing it
        length = 0
        for x in next_seqline(f):
            length += len(x)
        
        #return record without the sequence
        rec['seq'] = None
        rec['len'] = length
        
    return rec