Skip to content

mmap Module Complexity¶

The mmap module provides memory-mapped file support for efficient random access to file contents.

Complexity Reference¶

Operation	Time	Space	Notes
`mmap.mmap(fileno, length, ...)`	Varies	Varies	Depends on OS and mapping size
`mmap.read(n)`	O(n)	O(n)	n = bytes read
`mmap.write(data)`	O(n)	O(1)	n = bytes written
`mmap.seek(pos[, whence])`	O(1)	O(1)	Pointer movement
`mmap.find(sub[, start[, end]])`	O(n)	O(1)	n = search range
`mmap.flush()`	Varies	O(1)	Depends on OS and dirty pages
`mmap.close()`	O(1)	O(1)	Unmap and close

Platform-dependent behavior

mmap performance depends on the operating system, filesystem, and page cache behavior. The costs shown here describe typical asymptotic behavior, not wall-clock guarantees.

Common Operations¶

Creating a Mapping¶

import mmap

with open('data.bin', 'rb') as f:
    mm = mmap.mmap(f.fileno(), 0, access=mmap.ACCESS_READ)
    print(mm.size())  # total size
    mm.close()

Reading and Writing¶

import mmap

with open('data.bin', 'r+b') as f:
    mm = mmap.mmap(f.fileno(), 0)

    # Read 16 bytes
    chunk = mm.read(16)

    # Write bytes in place
    mm.seek(0)
    mm.write(b'HELLO')

    mm.flush()
    mm.close()

Searching¶

import mmap

with open('data.bin', 'rb') as f:
    mm = mmap.mmap(f.fileno(), 0, access=mmap.ACCESS_READ)

    idx = mm.find(b'needle')  # O(n) search
    if idx != -1:
        print('Found at', idx)

    mm.close()

Performance Notes¶

Mapping large files avoids copying data into Python space for random access.
Reads/writes still move data between kernel and user space when accessed.
For sequential reads of large files, buffered I/O can be comparable or faster.