Files
controller-wisetopdataserver/python/biopro/file/segment.py
T
2021-12-20 14:52:55 +08:00

266 lines
6.4 KiB
Python

"""segment a file. used to upload file to network for someone can donwload.
Main
----
segment a file.
::
python -m biopro.file.segment FILE [SIZE]
Class
-----
"""
from collections import deque
from math import ceil
from pathlib import Path
from typing import Dict, Optional
import struct
import base64
from biopro.util.json import JsonSerialize, JSON_OBJECT
from biopro.util.lock import Synchronized, synchronized
class FileSegment(Synchronized, JsonSerialize):
"""file segment information.
**json format**
::
segment_info = {
'path': str
'file_size': int
'segment_total': int
'segment_size': int
}
"""
DEFAULT_CACHE_SIZE = 16
DEFAULT_SEGMENT_SIZE = 200 * 1024 # 200KB
ALL = {}
__slots__ = '_export_path', '_segment_size', '_file_size', '_total_segment', '_cache'
def __init__(self,
export_path: Path,
segment_size = DEFAULT_SEGMENT_SIZE):
"""
:param export_path: source path
:param segment_size: size in bytes
"""
super().__init__()
if segment_size <= 0:
raise ValueError('non-positive segment_size : ' + str(segment_size))
if segment_size < 1024:
raise ValueError('non-suggest segment_size, too small : ' + str(segment_size))
self._export_path = export_path
self._segment_size = segment_size
self._file_size = 0
self._total_segment = 0
self._cache = deque([], maxlen=self.DEFAULT_CACHE_SIZE)
self.reload()
@synchronized
def reload(self):
"""reload file cache"""
if not self._export_path.exists():
raise FileNotFoundError(str(self._export_path))
self._file_size = file_size = self._export_path.stat().st_size
self._total_segment = int(ceil(file_size / self._segment_size))
self._cache.clear()
@property
def export_path(self) -> Path:
"""source path"""
return self._export_path
@property
def segment_size(self) -> int:
"""segment size in bytes."""
return self._segment_size
@property
def file_size(self) -> int:
"""original file size"""
return self._file_size
@property
def total_size(self) -> int:
"""total segment number"""
return self._total_segment
def get_segment(self, index: int) -> 'FileSegmentContent':
"""get certain segment content.
:param index:
:return:
"""
for i, content in self._cache: # type: int, str
if i == index:
return FileSegmentContent(self, index, content)
content = self._get_segment(index, 'BPRF')
self._cache.append((index, content))
return FileSegmentContent(self, index, content)
@synchronized
def _get_segment(self, index: int, _type: str) -> str:
if not (0 <= index < self._total_segment):
raise IndexError()
if _type == 'BPRF':
read = struct.calcsize('>4sIQ16s5B')
with self._export_path.open('rb') as f:
f.seek( read + index * self._segment_size )
return base64.b64encode(f.read(self._segment_size)).decode()
else:
with self._export_path.open('r') as f:
f.seek(index * self._segment_size)
return f.read(self._segment_size)
@classmethod
def get(cls,
export_path: Path,
segment_size = DEFAULT_SEGMENT_SIZE,
create=False) -> Optional['FileSegment']:
"""get/create file segment.
:param export_path:
:param segment_size:
:param create: create segment if not found
:return: file segment.
"""
print('*****1', export_path)
ret = cls.ALL.get(export_path, None)
if not create or ret is not None:
return ret
ret = FileSegment(export_path, segment_size)
cls.ALL[export_path] = ret
print('*****2', export_path)
return ret
def invalid(self):
"""invalid this segment and release the resource.
:return:
"""
self._cache.clear()
try:
del self.ALL[self._export_path]
except KeyError:
pass
@classmethod
def invalid_all(cls):
cls.ALL.clear()
def as_json(self) -> JSON_OBJECT:
return {
'path': self._export_path,
'file_size': self._file_size,
'segment_total': self._total_segment,
'segment_size': self._segment_size,
}
class FileSegmentContent(JsonSerialize):
"""File segment with file segment content.
**json format**
::
segment_info = {
'path': str
'file_size': int
'segment_total': int
'segment_size': int
'segment_index': int
'segment': str
}
"""
__slots__ = '_segment', '_index', '_content'
def __init__(self, segment: FileSegment, index: int, content: str):
self._segment = segment
self._index = index
self._content = content
def __len__(self):
return len(self._content)
def __str__(self):
return self._content
def __repr__(self):
return 'FileSegment[%s:%d/%d]' % (self.export_path, self._index, self.total_size)
@property
def export_path(self) -> Path:
"""source path"""
return self._segment.export_path
@property
def segment_size(self) -> int:
"""segment size in bytes."""
return self._segment.segment_size
@property
def file_size(self) -> int:
"""original file size"""
return self._segment.file_size
@property
def total_size(self) -> int:
"""total segment number"""
return self._segment.total_size
def as_json(self) -> JSON_OBJECT:
ret = self._segment.as_json()
ret['segment_index'] = self._index
ret['segment'] = self._content
return ret
if __name__ == '__main__':
import sys
def _main():
try:
sg_sz = int(sys.argv[2])
except (IndexError, ValueError):
sg_sz = 1024
fs = FileSegment(Path(sys.argv[1]), sg_sz)
for i in range(fs.total_size):
fc = fs.get_segment(i)
print('---', repr(fc), '---')
print(str(fc))
_main()