After looking all over the Internet, I've come to this.
Let's say I have already made a text file that reads:
Hello World
Well, I want to remove the very last character (in this case d) from this text file.
So now the text file should look like this: Hello Worl
But I have no idea how to do this.
All I want, more or less, is a single backspace function for text files on my HDD.
This needs to work on Linux as that's what I'm using.
Use fileobject.seek() to seek 1 position from the end, then use file.truncate() to remove the remainder of the file:
import os
with open(filename, 'rb+') as filehandle:
filehandle.seek(-1, os.SEEK_END)
filehandle.truncate()
This works fine for single-byte encodings. If you have a multi-byte encoding (such as UTF-16 or UTF-32) you need to seek back enough bytes from the end to account for a single codepoint.
For variable-byte encodings, it depends on the codec if you can use this technique at all. For UTF-8, you need to find the first byte (from the end) where bytevalue & 0xC0 != 0x80 is true, and truncate from that point on. That ensures you don't truncate in the middle of a multi-byte UTF-8 codepoint:
with open(filename, 'rb+') as filehandle:
# move to end, then scan forward until a non-continuation byte is found
filehandle.seek(-1, os.SEEK_END)
while filehandle.read(1) & 0xC0 == 0x80:
# we just read 1 byte, which moved the file position forward,
# skip back 2 bytes to move to the byte before the current.
filehandle.seek(-2, os.SEEK_CUR)
# last read byte is our truncation point, move back to it.
filehandle.seek(-1, os.SEEK_CUR)
filehandle.truncate()
Note that UTF-8 is a superset of ASCII, so the above works for ASCII-encoded files too.
Accepted answer of Martijn is simple and kind of works, but does not account for text files with:
UTF-8 encoding containing non-English characters (which is the default encoding for text files in Python 3)
one newline character at the end of the file (which is the default in Linux editors like vim or gedit)
If the text file contains non-English characters, neither of the answers provided so far would work.
What follows is an example, that solves both problems, which also allows removing more than one character from the end of the file:
import os
def truncate_utf8_chars(filename, count, ignore_newlines=True):
"""
Truncates last `count` characters of a text file encoded in UTF-8.
:param filename: The path to the text file to read
:param count: Number of UTF-8 characters to remove from the end of the file
:param ignore_newlines: Set to true, if the newline character at the end of the file should be ignored
"""
with open(filename, 'rb+') as f:
last_char = None
size = os.fstat(f.fileno()).st_size
offset = 1
chars = 0
while offset <= size:
f.seek(-offset, os.SEEK_END)
b = ord(f.read(1))
if ignore_newlines:
if b == 0x0D or b == 0x0A:
offset += 1
continue
if b & 0b10000000 == 0 or b & 0b11000000 == 0b11000000:
# This is the first byte of a UTF8 character
chars += 1
if chars == count:
# When `count` number of characters have been found, move current position back
# with one byte (to include the byte just checked) and truncate the file
f.seek(-1, os.SEEK_CUR)
f.truncate()
return
offset += 1
How it works:
Reads only the last few bytes of a UTF-8 encoded text file in binary mode
Iterates the bytes backwards, looking for the start of a UTF-8 character
Once a character (different from a newline) is found, return that as the last character in the text file
Sample text file - bg.txt:
Здравей свят
How to use:
filename = 'bg.txt'
print('Before truncate:', open(filename).read())
truncate_utf8_chars(filename, 1)
print('After truncate:', open(filename).read())
Outputs:
Before truncate: Здравей свят
After truncate: Здравей свя
This works with both UTF-8 and ASCII encoded files.
In case you are not reading the file in binary mode, where you have only 'w' permissions, I can suggest the following.
f.seek(f.tell() - 1, os.SEEK_SET)
f.write('')
In this code above, f.seek() will only accept f.tell() b/c you do not have 'b' access. then you can set the cursor to the starting of the last element. Then you can delete the last element by an empty string.
with open(urfile, 'rb+') as f:
f.seek(0,2) # end of file
size=f.tell() # the size...
f.truncate(size-1) # truncate at that size - how ever many characters
Be sure to use binary mode on windows since Unix file line ending many return an illegal or incorrect character count.
with open('file.txt', 'w') as f:
f.seek(0, 2) # seek to end of file; f.seek(0, os.SEEK_END) is legal
f.seek(f.tell() - 2, 0) # seek to the second last char of file; f.seek(f.tell()-2, os.SEEK_SET) is legal
f.truncate()
subject to what last character of the file is, could be newline (\n) or anything else.
This may not be optimal, but if the above approaches don't work out, you could do:
with open('myfile.txt', 'r') as file:
data = file.read()[:-1]
with open('myfile.txt', 'w') as file:
file.write(data)
The code first opens the file, and then copies its content (with the exception of the last character) to the string data. Afterwards, the file is truncated to zero length (i.e. emptied), and the content of data is saved to the file, with the same name.
This is basically the same as vins ms's answer, except that it doesn't use the os package, and that is used the safer 'with open' syntax. This may not be recommended if the text file is huge. (I wrote this since none of the above approaches worked out too well for me in python 3.8).
here is a dirty way (erase & recreate)...
i don't advice to use this, but, it's possible to do like this ..
x = open("file").read()
os.remove("file")
open("file").write(x[:-1])
On a Linux system or (Cygwin under Windows). You can use the standard truncate command. You can reduce or increase the size of your file with this command.
In order to reduce a file by 1G the command would be truncate -s 1G filename. In the following example I reduce a file called update.iso by 1G.
Note that this operation took less than five seconds.
chris#SR-ENG-P18 /cygdrive/c/Projects
$ stat update.iso
File: update.iso
Size: 30802968576 Blocks: 30081024 IO Block: 65536 regular file
Device: ee6ddbceh/4000177102d Inode: 19421773395035112 Links: 1
Access: (0664/-rw-rw-r--) Uid: (1052727/ chris) Gid: (1049089/Domain Users)
Access: 2020-06-12 07:39:00.572940600 -0400
Modify: 2020-06-12 07:39:00.572940600 -0400
Change: 2020-06-12 07:39:00.572940600 -0400
Birth: 2020-06-11 13:31:21.170568000 -0400
chris#SR-ENG-P18 /cygdrive/c/Projects
$ truncate -s -1G update.iso
chris#SR-ENG-P18 /cygdrive/c/Projects
$ stat update.iso
File: update.iso
Size: 29729226752 Blocks: 29032448 IO Block: 65536 regular file
Device: ee6ddbceh/4000177102d Inode: 19421773395035112 Links: 1
Access: (0664/-rw-rw-r--) Uid: (1052727/ chris) Gid: (1049089/Domain Users)
Access: 2020-06-12 07:42:38.335782800 -0400
Modify: 2020-06-12 07:42:38.335782800 -0400
Change: 2020-06-12 07:42:38.335782800 -0400
Birth: 2020-06-11 13:31:21.170568000 -0400
The stat command tells you lots of info about a file including its size.
I have written a very simple python programme, called wc.py, which mimics "bash wc" behaviour to count the number of words, lines and bytes in a file. My programme is as follow:
import sys
path = sys.argv[1]
w = 0
l = 0
b = 0
for currentLine in file:
wordsInLine = currentLine.strip().split(' ')
wordsInLine = [word for word in wordsInLine if word != '']
w += len(wordsInLine)
b += len(currentLine.encode('utf-8'))
l += 1
#output
print(str(l) + ' ' + str(w) + ' ' + str(b))
In order to execute my programme you should execute the following command:
python3 wc.py [a file to read the data from]
As the result it shows
[The number of lines in the file] [The number of words in the file] [The number of bytes in the file] [the file directory path]
The files I used to test my code is as follow:
file.txt which contains the following data:
1
2
3
4
Executing "wc file.txt" returns
4 4 8
Executing "python3 wc.py file.txt" returns 4 4 8
Download "Annual enterprise survey: 2020 financial year (provisional) – CSV" from CSV file download
Executing "wc [fileName].csv" returns
37081 500273 5881081
Executing "python3 wc.py [fileName].csv" returns
37081 500273 5844000
and a [something].pdf file
Executing "wc [something].pdf" works.
Executing "python3 code.py" throws the following errors:
Traceback (most recent call last):
File "code.py", line 10, in <module>
for currentLine in file:
File "/usr/lib/python3.8/codecs.py", line 322, in decode
(result, consumed) = self._buffer_decode(data, self.errors, final)
UnicodeDecodeError: 'utf-8' codec can't decode byte 0xbf in position 10: invalid start byte
As you can see, the output of python3 code.py [something].pdf and python3 code.py [something].csv is not the same as what wc returns. Could you help me to find the reason of this erroneous behaviour in my code?
Regarding the CSV file, if you look at the difference between your result and that of wc:
5881081 - 5844000 = 37081 which is exactly the number of lines.
That is, every line has one additional character in the original file. That character is the carriage return \r which got lost in Python because you iterate over lines and don't specify the linebreaks. If you want a byte-correct result, you have to first identify the type of linebreaks used in the file (and watch out for inconsistencies throughout the document).
I am trying to write Chinese characters to a CSV file based on their Unicode code points found in a text file in unicode.org/Public/zipped/13.0.0/Unihan.zip. For instance, one example character is U+9109.
In the example below I can get the correct output by hard coding the value (line 8), but keep getting it wrong with every permutation I've tried at generating the bytes from the code point (lines 14-16).
I'm running this in Python 3.8.3 on a Debian-based Linux distro.
Minimal working (broken) example:
1 #!/usr/bin/env python3
2
3 def main():
4
5 output = open("test.csv", "wb")
6
7 # Hardcoded values work just fine
8 output.write('\u9109'.encode("utf-8"))
9
10 # Comma separation
11 output.write(','.encode("utf-8"))
12
13 # Problem is here
14 codepoint = '9109'
15 u_str = '\\' + 'u' + codepoint
16 output.write(u_str.encode("utf-8"))
17
18 # End with newline
19 output.write('\n'.encode("utf-8"))
20
21 output.close()
22
23 if __name__ == "__main__":
24 main()
Executing and viewing results:
example $
example $./test.py
example $
example $cat test.csv
鄉,\u9109
example $
The expected output would look like this (Chinese character occurring on both sides of the comma):
example $
example $./test.py
example $cat test.csv
鄉,鄉
example $
chr is used to convert integers to code points in Python 3. Your code could use:
output.write(chr(0x9109).encode("utf-8"))
But if you specify the encoding in the open instead of using binary mode you don't have to manually encode everything. print to a file handles newlines for you as well.
with open("test.txt",'w',encoding='utf-8') as output:
for i in range(0x4e00,0x4e10):
print(f'U+{i:04X} {chr(i)}',file=output)
Output:
U+4E00 一
U+4E01 丁
U+4E02 丂
U+4E03 七
U+4E04 丄
U+4E05 丅
U+4E06 丆
U+4E07 万
U+4E08 丈
U+4E09 三
U+4E0A 上
U+4E0B 下
U+4E0C 丌
U+4E0D 不
U+4E0E 与
U+4E0F 丏
Vim offers the :profile command, which is really handy. But it is limited to Vim script -- when it comes to plugins implemented in python it isn't that helpful.
Currently I'm trying to understand what is causing a large delay on Denite. As it doesn't happen in vanilla Vim, but only on some specific conditions which I'm not sure how to reproduce, I couldn't find which setting/plugin is interfering.
So I turned to profiling, and this is what I got from :profile:
FUNCTION denite#vim#_start()
Defined: ~/.vim/bundle/denite.nvim/autoload/denite/vim.vim line 33
Called 1 time
Total time: 5.343388
Self time: 4.571928
count total (s) self (s)
1 0.000006 python3 << EOF
def _temporary_scope():
nvim = denite.rplugin.Neovim(vim)
try:
buffer_name = nvim.eval('a:context')['buffer_name']
if nvim.eval('a:context')['buffer_name'] not in denite__uis:
denite__uis[buffer_name] = denite.ui.default.Default(nvim)
denite__uis[buffer_name].start(
denite.rplugin.reform_bytes(nvim.eval('a:sources')),
denite.rplugin.reform_bytes(nvim.eval('a:context')),
)
except Exception as e:
import traceback
for line in traceback.format_exc().splitlines():
denite.util.error(nvim, line)
denite.util.error(nvim, 'Please execute :messages command.')
_temporary_scope()
if _temporary_scope in dir():
del _temporary_scope
EOF
1 0.000017 return []
(...)
FUNCTIONS SORTED ON TOTAL TIME
count total (s) self (s) function
1 5.446612 0.010563 denite#helper#call_denite()
1 5.396337 0.000189 denite#start()
1 5.396148 0.000195 <SNR>237_start()
1 5.343388 4.571928 denite#vim#_start()
(...)
I tried to use the python profiler directly by wrapping the main line:
import cProfile
cProfile.run(_temporary_scope(), '/path/to/log/file')
, but no luck -- just a bunch of errors from cProfile. Perhaps it is because the way python is started from Vim, as it is hinted here that it only works on the main thread.
I guess there should be an easier way of doing this.
The python profiler does work by enclosing the whole code,
cProfile.run("""
(...)
""", '/path/to/log/file')
, but it is not that helpful. Maybe it is all that is possible.
I am trying to create and unix file with some text in it with the below command:
ssh.sendLine("echo '"+"{1:F01ZYIBGB20AXXX0000000000}{2:O5481710NDEASES0XXXX12345678901511041511180930N}\n{4:\n:16R:GENL\n:20C::SEME//"+$TradeRef+"\n:23G:INST\n:16R:LINK\n:20C::RELA//"+$TradeRef+"\n:16S:LINK\n:16R:STAT\n:25D::MTCH//MACH\n:16S:STAT\n:16S:GENL\n:16R:SETTRAN\n:35B:ISIN DE0005933931\niShares Core DAX UCITS ETF DE\n:36B::SETT//UNIT/10,\n:97A::SAFE//8696\n:22F::SETR//TRAD\n:98A::SETT//20151118\n:98A::TRAD//20151118\n:16S:SETTRAN\n-}'"+">M548File.txt");
NOw this command is createing the file M548File.txt.
When id a cat this is what i get :
{1:F01ZYIBGB20AXXX0000000000}{2:O5481710NDEASES0XXXX12345678901511041511180930N}
{4:
:16R:GENL
:20C::SEME//11111111111111111111
:23G:INST
:16R:LINK
:20C::RELA//11111111111111111111
:16S:LINK
:16R:STAT
:25D::MTCH//MACH
:16S:STAT
:16S:GENL
:16R:SETTRAN
:35B:ISIN DE0005933931
iShares Core DAX UCITS ETF DE
:36B::SETT//UNIT/10,
:97A::SAFE//8696
:22F::SETR//TRAD
:98A::SETT//20151118
:98A::TRAD//20151118
:16S:SETTRAN
-}
However when i try to open the same file in notepad i get one extraline at the last which is basically an empty line making it a total of -- 23 lines as compared to 22 in cat.
I tried the Sed commands but it just not working
Any idea how to overcome this to get 22 lines in notepad(same as cat)?