| 1 |
|
|---|
| 2 | ##
|
|---|
| 3 | ## Stream Base Class
|
|---|
| 4 | ##
|
|---|
| 5 | class Stream(object):
|
|---|
| 6 |
|
|---|
| 7 | def __init__(self, data):
|
|---|
| 8 |
|
|---|
| 9 | self.data = data
|
|---|
| 10 |
|
|---|
| 11 | def __del__(self):
|
|---|
| 12 |
|
|---|
| 13 | self.data.close()
|
|---|
| 14 |
|
|---|
| 15 | def close(self):
|
|---|
| 16 |
|
|---|
| 17 | self.data.close()
|
|---|
| 18 |
|
|---|
| 19 | def records(self):
|
|---|
| 20 |
|
|---|
| 21 | records = []
|
|---|
| 22 |
|
|---|
| 23 | for record in self:
|
|---|
| 24 | records.append(record)
|
|---|
| 25 |
|
|---|
| 26 | return records
|
|---|
| 27 |
|
|---|
| 28 | ##
|
|---|
| 29 | ## ParsedStream
|
|---|
| 30 | ##
|
|---|
| 31 | class ParsedStream(object):
|
|---|
| 32 |
|
|---|
| 33 | def __init__(self, stream, parse=None, parse_args=[], parse_kwargs={}):
|
|---|
| 34 |
|
|---|
| 35 | self.stream = stream
|
|---|
| 36 | self.parse = parse
|
|---|
| 37 | self.parse_args = parse_args
|
|---|
| 38 | self.parse_kwargs = parse_kwargs
|
|---|
| 39 |
|
|---|
| 40 | def __iter__(self):
|
|---|
| 41 |
|
|---|
| 42 | return self
|
|---|
| 43 |
|
|---|
| 44 | def next(self):
|
|---|
| 45 |
|
|---|
| 46 | try:
|
|---|
| 47 | record = self.stream.next()
|
|---|
| 48 | args = self.parse_args
|
|---|
| 49 | kwargs = self.parse_kwargs
|
|---|
| 50 | return self.parse(record, *args, **kwargs)
|
|---|
| 51 | except StopIteration:
|
|---|
| 52 | self.stream.close()
|
|---|
| 53 | raise StopIteration
|
|---|
| 54 |
|
|---|
| 55 |
|
|---|
| 56 | ##
|
|---|
| 57 | ## Stream of records separated by newlines
|
|---|
| 58 | ##
|
|---|
| 59 | class NewLineStream(Stream):
|
|---|
| 60 |
|
|---|
| 61 | def __init__(self, data, header=0):
|
|---|
| 62 |
|
|---|
| 63 | Stream.__init__(self, data)
|
|---|
| 64 |
|
|---|
| 65 | for i in range(header):
|
|---|
| 66 | line = self.data.readline()
|
|---|
| 67 |
|
|---|
| 68 | def __iter__(self):
|
|---|
| 69 |
|
|---|
| 70 | return self
|
|---|
| 71 |
|
|---|
| 72 | def next(self):
|
|---|
| 73 |
|
|---|
| 74 | line = self.data.readline()
|
|---|
| 75 |
|
|---|
| 76 | if line:
|
|---|
| 77 | return line.rstrip('\r\n')
|
|---|
| 78 | else:
|
|---|
| 79 | raise StopIteration
|
|---|
| 80 |
|
|---|
| 81 |
|
|---|
| 82 | ##
|
|---|
| 83 | ## Stream of a single record
|
|---|
| 84 | ##
|
|---|
| 85 | class SingleRecordStream(Stream):
|
|---|
| 86 |
|
|---|
| 87 | def __init__(self, data, header=0):
|
|---|
| 88 |
|
|---|
| 89 | Stream.__init__(self, data)
|
|---|
| 90 |
|
|---|
| 91 | for i in range(header):
|
|---|
| 92 | line = self.data.readline()
|
|---|
| 93 |
|
|---|
| 94 | def __iter__(self):
|
|---|
| 95 |
|
|---|
| 96 | return self
|
|---|
| 97 |
|
|---|
| 98 | def next(self):
|
|---|
| 99 |
|
|---|
| 100 | record = []
|
|---|
| 101 | lines = self.data.readlines()
|
|---|
| 102 |
|
|---|
| 103 | if lines:
|
|---|
| 104 | for line in lines:
|
|---|
| 105 | record.append(line.rstrip('\r\n'))
|
|---|
| 106 | return record
|
|---|
| 107 | else:
|
|---|
| 108 | raise StopIteration
|
|---|
| 109 |
|
|---|
| 110 |
|
|---|
| 111 | ##
|
|---|
| 112 | ## Stream of records separated by blank lines
|
|---|
| 113 | ##
|
|---|
| 114 | class BlankLineStream(Stream):
|
|---|
| 115 |
|
|---|
| 116 | def __init__(self, data, header=0):
|
|---|
| 117 |
|
|---|
| 118 | Stream.__init__(self, data)
|
|---|
| 119 |
|
|---|
| 120 | for i in range(header):
|
|---|
| 121 | line = self.data.readline()
|
|---|
| 122 |
|
|---|
| 123 | def __iter__(self):
|
|---|
| 124 |
|
|---|
| 125 | return self
|
|---|
| 126 |
|
|---|
| 127 | def next(self):
|
|---|
| 128 |
|
|---|
| 129 | record = []
|
|---|
| 130 | line = self.data.readline()
|
|---|
| 131 |
|
|---|
| 132 | if not line:
|
|---|
| 133 | raise StopIteration
|
|---|
| 134 |
|
|---|
| 135 | while line not in ['\n', '\r\n']:
|
|---|
| 136 | record.append(line.rstrip('\r\n'))
|
|---|
| 137 | line = self.data.readline()
|
|---|
| 138 | if not line:
|
|---|
| 139 | return record
|
|---|
| 140 |
|
|---|
| 141 | return record
|
|---|
| 142 |
|
|---|
| 143 |
|
|---|
| 144 | ##
|
|---|
| 145 | ## Stream of records separated by a single line separator
|
|---|
| 146 | ##
|
|---|
| 147 | class LineSeparatorStream(Stream):
|
|---|
| 148 |
|
|---|
| 149 | def __init__(self, data, separator='\n', header=0):
|
|---|
| 150 |
|
|---|
| 151 | Stream.__init__(self, data)
|
|---|
| 152 |
|
|---|
| 153 | if separator[-1] != '\n':
|
|---|
| 154 | separator = '%s\n' % (separator)
|
|---|
| 155 |
|
|---|
| 156 | self.separator = separator
|
|---|
| 157 |
|
|---|
| 158 | for i in range(header):
|
|---|
| 159 | line = self.data.readline()
|
|---|
| 160 |
|
|---|
| 161 | def __iter__(self):
|
|---|
| 162 |
|
|---|
| 163 | return self
|
|---|
| 164 |
|
|---|
| 165 | def next(self):
|
|---|
| 166 |
|
|---|
| 167 | record = []
|
|---|
| 168 | line = self.data.readline()
|
|---|
| 169 |
|
|---|
| 170 | if not line:
|
|---|
| 171 | raise StopIteration
|
|---|
| 172 |
|
|---|
| 173 | while line != self.separator:
|
|---|
| 174 | record.append(line.rstrip('\r\n'))
|
|---|
| 175 | line = self.data.readline()
|
|---|
| 176 | if not line:
|
|---|
| 177 | return record
|
|---|
| 178 |
|
|---|
| 179 | return record
|
|---|
| 180 |
|
|---|