From 786d4faf7474d62838d3129efa1f62f9548bffc8 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Piotr=20Ma=C5=9Blanka?= <piotr.maslanka@henrietta.com.pl>
Date: Fri, 11 Dec 2020 19:09:27 +0100
Subject: [PATCH] added an unit test
---
docs/varlen.rst | 13 ++++++++--
tempsdb/varlen.pxd | 7 ++++++
tempsdb/varlen.pyx | 59 ++++++++++++++++++++++++++++++++++++--------
tests/test_varlen.py | 17 +++++++++++++
4 files changed, 84 insertions(+), 12 deletions(-)
create mode 100644 tests/test_varlen.py
diff --git a/docs/varlen.rst b/docs/varlen.rst
index 04579d7..4275eaa 100644
--- a/docs/varlen.rst
+++ b/docs/varlen.rst
@@ -20,12 +20,17 @@ of 5 normal time series created to accomodate it, with length of:
* 255
* 255
-Each entry is also prefixed by it's length. The size of that field is described by an
-extra parameter called `size_struct`.
+Note that an entry is written to enough series so that it fits. For example, a 8 byte piece of data
+would be written to only to the first series.
+
+Each entry is also prefixed by it's length, so the actual size of the first
+series is larger by that. The size of that field is described by an
+extra parameter called `size_struct`. It represents an unsigned number.
Note that the only valid sizes of `size_struct` are:
* 1 for maximum length of 255
* 2 for maximum length of 65535
+* 3 for maximum length of 16777215
* 4 for maximum length of 4294967295
Accessing them
@@ -39,6 +44,10 @@ Use methods :meth:`tempsdb.database.Database.create_varlen_series` and
:members:
+.. autoclass:: tempsdb.varlen.VarlenIterator
+ :members:
+
+
.. autoclass:: tempsdb.varlen.VarlenEntry
:members:
diff --git a/tempsdb/varlen.pxd b/tempsdb/varlen.pxd
index 7d5baa0..a5b5b34 100644
--- a/tempsdb/varlen.pxd
+++ b/tempsdb/varlen.pxd
@@ -29,6 +29,13 @@ cdef class VarlenSeries:
elif self.size_field == 4:
return 0xFFFFFFFF
+cdef class VarlenIterator:
+ cdef:
+ object parent
+ unsigned long long start
+ unsigned long long stop
+
+
cdef class VarlenEntry:
cdef:
list chunks
diff --git a/tempsdb/varlen.pyx b/tempsdb/varlen.pyx
index e0ee320..71263bf 100644
--- a/tempsdb/varlen.pyx
+++ b/tempsdb/varlen.pyx
@@ -72,7 +72,7 @@ cdef class VarlenEntry:
:param start: position to start at
:param stop: position to stop at
:return: a slice of this entry
- :raises ValueError: stop was smaller than start
+ :raises ValueError: stop was smaller than start or indices were invalid
"""
if stop < start:
raise ValueError('stop smaller than start')
@@ -96,20 +96,26 @@ cdef class VarlenEntry:
pointer += next_chunk_len
segment += 1
-
cdef:
int write_pointer = 0
int chunk_len = self.parent.get_length_for(segment)
int len_to_read = self.parent.get_length_for(segment) - start_reading_at
Chunk chunk = self.chunks[segment]
- while write_pointer < length:
- if chunk_len > start_reading_at + length:
- ... # We end this right here
-
- b[write_pointer:write_pointer+len_to_read] = chunk.get_slice_of_piece_at(self.item_no[segment],
- start_reading_at,
- 0)
+ bytes temp_data
+ while write_pointer < length and len(self.chunks) < segment:
+ if chunk_len-start_reading_at >= + (length - write_pointer):
+ # We have all the data that we require
+ b[write_pointer:length] = chunk.get_slice_of_piece_at(self.item_no[segment],
+ 0, length-write_pointer)
+ return bytes(b)
+
+ temp_data = chunk.get_slice_of_piece_at(self.item_no[segment], 0, chunk_len)
+ b[write_pointer:write_pointer+chunk_len] = temp_data
+ write_pointer += chunk_len
+ segment += 1
+ start_reading_at = 0
+ raise ValueError('invalid indices')
cpdef bytes to_bytes(self):
"""
@@ -136,6 +142,31 @@ cdef class VarlenEntry:
def __len__(self) -> int:
return self.length()
+STRUCT_L = struct.Struct('<L')
+class ThreeByteStruct:
+ __slots__ = ()
+ def pack(self, v: int) -> bytes:
+ return STRUCT_L.pack(v)[0:3]
+
+ def unpack(self, v: bytes) -> tp.Tuple[int]:
+ return STRUCT_L.unpack(v+b'\x00')
+
+
+cdef class VarlenIterator:
+ """
+ A result of a varlen series query
+ """
+ def __init__(self, parent: VarlenSeries, start: int, stop: int):
+ self.parent = parent
+ self.start = start
+ self.stop = stop
+
+ def __next__(self) -> tp.Tuple[int, VarlenEntry]:
+ ...
+
+ def __iter__(self):
+ return self
+
cdef class VarlenSeries:
"""
@@ -170,6 +201,8 @@ cdef class VarlenSeries:
self.size_struct = struct.Struct('<B')
elif self.size_field == 2:
self.size_struct = struct.Struct('<H')
+ elif self.size_field == 3:
+ self.size_struct = ThreeByteStruct()
elif self.size_field == 4:
self.size_struct = struct.Struct('<L')
else:
@@ -286,6 +319,12 @@ cpdef VarlenSeries create_varlen_series(str path, str name, int size_struct, lis
"""
Create a variable length series
+ :param path: path where the directory will be placed
+ :param name: name of the series
+ :param size_struct: size of the length indicator. Must be one of 1, 2, 3 or 4.
+ :param length_profile: series' length profile
+ :param max_entries_per_chunk: maximum entries per a chunk file
+ :return: newly created VarlenSeries
:raises AlreadyExists: directory exists at given path
:raises ValueError: invalid length profile or max_entries_per_chunk or size_struct
"""
@@ -293,7 +332,7 @@ cpdef VarlenSeries create_varlen_series(str path, str name, int size_struct, lis
raise AlreadyExists('directory present at paht')
if not length_profile or not max_entries_per_chunk:
raise ValueError('invalid parameter')
- if size_struct not in (1, 2, 4):
+ if not (1 <= size_struct <= 4):
raise ValueError('invalid size_struct')
os.mkdir(path)
diff --git a/tests/test_varlen.py b/tests/test_varlen.py
new file mode 100644
index 0000000..9c64dd2
--- /dev/null
+++ b/tests/test_varlen.py
@@ -0,0 +1,17 @@
+import os
+import unittest
+
+from tempsdb.varlen import create_varlen_series
+
+
+class TestVarlen(unittest.TestCase):
+ def test_varlen(self):
+ varlen = create_varlen_series('test_dir', 'test_dir', 2, [10, 20, 10], 20)
+ try:
+ varlen.append(0, b'test skarabeusza')
+ self.assertEqual(len(os.listdir('test_dir')), 2)
+
+ varlen.append(10, b'test skarabeuszatest skarabeusza')
+ self.assertEqual(len(os.listdir('test_dir')), 3)
+ finally:
+ varlen.close()
--
GitLab