mesejo / trex Goto Github PK

Add missing documentation detailing how to integrate trrex with different libraries. The documentation should use the
pydata-sphinx-theme

See this stackoverflow question

Add experiments comparing the efficiency of trrex vs union-regex vs flashtext

The names left and right for parameters are not very informative, change the names to prefix and suffix

Hello Developer,

Below is the query sent and it is generating extra ")".

code:
import trrex as tx
tx.make(['IBS(.|)TECH','IBS(.|)SOL','IBS(.|)TEHK'], prefix=r"\b(", suffix=r")")

output='\b(IBS(.(?:|)SOL|*|)TE(?:HK|CH)))'

Please let me know if any further information is required on the same.

Thanks,
Rangam

There could be a parameter to replace the whitespace character for usage in multi-token keywords.

First, great package. Love using it! Makes my life much easier, and the speed is phenomenal!

I find that it works with a list of words, such as ['Love', 'Hello', ' Book',...]

Does it work with a list of regex patterns?

For example

regex_patterns = [
    r'(?!^\d+)(?=.*)(\b\d+$\b)'               #Remove any numbers that end string i.e "SHELL 545436"
   ,r'^\bSQ\b'                                #Remove the "SQ" if it starts the string ie. "SQ NORDSTROM"
   ,r'(?!\b(ST|^\w{1,2})\b$)\b\w{1,2}\b$'     #Remove any words that are one or two chars at end of string i.e. "Burger King CA" -- except if they = "ST"

]

tx_maker = tx.make(entity_final_clean_regex_patterns,  prefix=r"", suffix=r"") 


#clean column
df['clean'] = df.STRING_COLUMN.str.replace(tx_maker, "", regex=True)

Sort in order of decreasing length to match the largest output first

What is happening?
Currently we escape every character of the input words even if they do not need escaping.

What should happen?
Remove this behavior because it prevents the end user from adding patterns and also it makes the processing slow.
Document how to escape regex characters.

The compile function offers a vanilla wrapper for using with the re Python module, is better to remove this function and then add it later if needed.