Hi,
This is a tool I’m using on my own files to save me time. Basically or most of
the tracks were imported with different version iTunes over the years. There
are two problems:
1. File System characters are replaced (you can’t have ‘/‘ or ‘:’ in a file
name).
2. Smart Quotes were added at some point, these need to replaced.
3. Other character based of name being of a non-english origin.
If find others I’ll add them.
I’m using MusicBrainz to do a fuzzy match and get the correct name.
it’s not perfect, but works for 99% of files which is good enough for me!
Cheers
Dave
> On 8 Jun 2022, at 18:23, Avi Gross via Python-list <python-list@python.org>
> wrote:
>
> Dave,
>
> Your goal is to compare titles and there can be endless replacements needed
> if you allow the text to contain anything but ASCII.
>
> Have you considered stripping out things instead? I mean remove lots of stuff
> that is not ASCII in the first place and perhaps also remove lots of extra
> punctuation likesingle quotes or question marks or redundant white space and
> compare the sort of skeletons of the two?
>
> And even if that fails, could you have a measure of how different they are
> and tolerate if they were say off by one letter albeit "My desert" matching
> "My Dessert" might not be a valid match with one being a song about an arid
> environment and the other about food you don't need!
>
> Your seemingly simple need can expand into a fairly complex project. There
> may be many ideas on how to deal with it but not anything perfect enough to
> catch all cases as even a trained human may have to make decisions at times
> and not match what other humans do. We have examples like the TV show
> "NUMB3RS" that used a perfectly valid digit 3 to stand for an "E" but yet is
> often written when I look it up as NUMBERS. You have obvious cases where
> titles of songs may contain composite symbols like "œ" which will not compare
> to one where it is written out as "oe" so the idea of comparing is quite
> complex and the best you might do is heuristic.
>
> UNICODE has many symbols that are almost the same or even look the same or
> maybe in one font versus another. There are libraries of functions that allow
> some kinds of comparisons or conversions that you could look into but the
> gain for you may not be worth it. Nothing stops a person from naming a song
> any way they want and I speak many languages and often see a song re-titled
> in the local language and using the local alphabet mixed often with another.
>
> Your original question is perhaps now many questions, depending on what you
> choose. You started by wanting to know how to compare and it is moving on to
> how to delete parts or make substitutions or use regular expressions and it
> can get worse. You can, for example, take a string and identify the words
> within it and create a regular expression that inserts sequences between the
> words that match any zero or one or more non-word characters such as spaces,
> tabs, punctuation or non-ASCII, so that song titles with the same words in a
> sequence match no matter what is between them. The possibilities are endless
> but consider some of the techniques that are used by some programs that parse
> text and suggest alternate spellings or even programs like Google Translate
> that can take a sentence and then suggest you may mean a slightly altered
> sentence with one word changed to fit better.
>
> You need to decide what you want to deal with and what will be mis-classified
> by your program. Some of us have suggested folding the case of the words but
> that means asong about a dark skinned person in Poland called "Black Polish"
> would match a song about keeping your shoes dark with "black polish" so I
> keep repeating it is very hard or frankly impossible, to catch every case I
> can imagine and the many I can't!
>
> But the emphasis here is not your overall problem. It is about whether and
> how the computer language called python, and perhaps some add-on modules, can
> be used to solve each smaller need such as recognizing a pattern or replacing
> text. It can do quite a bit but only when the specification of the problem is
> exact.
>
>
>
>
> -----Original Message-----
> From: Dave <d...@looktowindward.com>
> To: python-list@python.org
> Sent: Wed, Jun 8, 2022 5:09 am
> Subject: Re: How to replace characters in a string?
>
> Hi,
>
> Thanks for this!
>
> So, is there a copy function/method that returns a MutableString like in
> objective-C? I’ve solved this problems before in a number of languages like
> Objective-C and AppleScript.
>
> Basically there is a set of common characters that need “normalizing” and I
> have a method that replaces them in a string, so:
>
> myString = [myString normalizeCharacters];
>
> Would return a new string with all the “common” replacements applied.
>
> Since the following gives an error :
>
> myString = 'Hello'
> myNewstring = myString.replace(myString,'e','a’)
>
> TypeError: 'str' object cannot be interpreted as an integer
>
> I can’t see of a way to do this in Python?
>
> All the Best
> Dave
>
>
>> On 8 Jun 2022, at 10:14, Chris Angelico <ros...@gmail.com> wrote:
>>
>> On Wed, 8 Jun 2022 at 18:12, Dave <d...@looktowindward.com> wrote:
>>
>>> I tried the but it doesn’t seem to work?
>>> myCompareFile1 = ascii(myTitleName)
>>> myCompareFile1.replace("\u2019", "'")
>>
>> Strings in Python are immutable. When you call ascii(), you get back a
>> new string, but it's one that has actual backslashes and such in it.
>> (You probably don't need this step, other than for debugging; check
>> the string by printing out the ASCII version of it, but stick to the
>> original for actual processing.) The same is true of the replace()
>> method; it doesn't change the string, it returns a new string.
>>
>>>>> word = "spam"
>>>>> print(word.replace("sp", "h"))
>> ham
>>>>> print(word)
>> spam
>>
>> ChrisA
>> --
>> https://mail.python.org/mailman/listinfo/python-list
>
> --
> https://mail.python.org/mailman/listinfo/python-list
> --
> https://mail.python.org/mailman/listinfo/python-list
--
https://mail.python.org/mailman/listinfo/python-list
