# coding: utf-8
import regex as re
re.search('[0-9]', "asdfgadfg5545")
re.search(r'[0-9]', "asdfgadfg5545")
re.search(r'\d', "asdfgadfg5545")
re.search('\d', "asdfgadfg5545")
re.search(r'\d', "asdfgadfg5545")
re.findall(r'\d', "asdfgadfg5545")
re.search(r'\d', "asdfgadfg")
a=re.search(r'\d', "asdfgadfg")
a
dir(a)
type(a)
a=re.search(r'\d', "asdfgadfg")
re.search(r'\d', "asdfgadfg")
re.findall(r'\d', "asdfgadfg5545")
re.findall(r'\d', "asdfgadfg")
re.findall(r'\d', "asdfgadfg5545")
re.findall(r'[a-z]*', "asdfgadfg5545")
re.findall(r'[a-z]+', "asdfgadfg5545")
re.findall(r'[abc]', "asdfgadfg5545")
re.findall(r'abc', "asdfgadfg5545")
re.findall(r'as', "asdfgadfg5545")
re.findall(r'[as]', "asdfgadfg5545")
re.findall(r'[^as]', "asdfgadfg5545")
re.findall(r'[a-g]', "asdfgadfg5545")
re.findall(r'[as]*', "asdfgadfg5545")
re.findall(r'[as]+', "asdfgadfg5545")
re.findall(r'[sa]+', "asdfgadfg5545")
re.findall(r'[sa]d?', "asdfgadfg5545")
re.findall(r'[asdf]{2,5}', "asdfgadfg5545")
re.findall(r'[asdf]{2,3}', "asdfgadfg5545")
re.findall(r'[asdf]{,3}', "asdfgadfg5545")
re.findall(r'[asdf]{3,}', "asdfgadfg5545")
re.findall(r'[[:alpha:]]', "asdfgadfg5545")
re.findall(r'[[:alnum:]]', "asdfgadfg5545")
re.findall(r'[[:lower:]]', "aFFFFsdfgadfg5545")
re.findall(r'[[:upper:]]', "aFFFFsdfgadfg5545")
re.findall(r'[[:upper:]]', "aFFFFsdŠČŽščfgadfg5545")
import re as re2
re2.findall(r'[[:upper:]]', "aFFFFsdŠČŽščfgadfg5545")
re.findall(r'[[:upper:]]', "aFFFFsdŠČŽščfgadfg5545")
re.findall(r'[[:punct:]]', "aFFFFsdŠČŽščfgadfg,55.45")
re.findall(r'\p{Greek}', "aFFFFsdŠČŽščfgadfg,55.45")
re.findall(r'\P{Greek}', "aFFFFsdŠČŽščfgadfg,55.45")
re.findall(r'\p{Greek}', "aFFFFαsdŠČŽščfgadfg,55.45")
re.findall(r'\p{Greek}', "aFFFFαsdŠČŽščfgadfg,55.45")
re.findall(r'[[:punct:]]', "aFFFFsdŠČŽščfgadfg,55.45")
re.findall(r'[[:punct:]]', "aFFFFsdŠČŽščfgadfg,55.45")
re.split(r'[[:punct:]]', "aFFFFsdŠČŽščfgadfg,55.45")
re.split(r'[[:punct:]]', "aFFFFsd--ŠČ,Žščfgadfg,55.45")
re.split(r'[[:punct:]]', "aFFFFsd--ŠČ,Žščfgadfg,55.45")
re.split(r'[[:punct:]]', "aFFFFsd--ŠČ,Žščfgadfg,55.45")
re.split(r'[[:punct:]]', text)
f=open("genesis.txt")
text=f.read()
re.split(r'[[:punct:]]', text)
re.split(r'[ [:punct:]]', text)
text
re.findall('.', text)
re.findall('\.', text)
re.findall('[.]', text)
re.findall('[.]', text)
re.findall('\.', text)
re.findall('\. [^ ]+ ', text)
re.findall('\. ?[^ ]+ ', text)
re.findall('\.\s?[^ ]+ ', text)
re.findall('\.\s*[^ ]+\s', text)
re.findall('\.\s*[[:alpha:]]+\s', text)
re.findall('\.\s*[[:alpha:]]+[[:space:]]', text)
re.findall('\.\s*[[:alpha:]]+\s', text)
re.findall('\.\s*\w+\s', text)
re.findall('\.\s*\w+\s', text)
textnl=text.replace(". ",".\n")
textnl
re.findall('\.\s*\w+\s', textnl)
textnl
re.findall('\.\s*\w+\s', textnl)
re.findall('\.\s*(\w+)\s', textnl)
re.findall('(\.)\s*(\w+)\s', textnl)
re.findall('\.\s*(\w+)\s(\w+)', textnl)
re.findall('\.\s*(\w+)\s(\w+)', textnl)
re.findall('\.\s*(\w+)\s(\w+)', textnl)
re.findall('god', textnl)
re.findall('and', textnl)
re.findall('and', textnl, flags=re.I)
re.findall('(?i)and', textnl)
re.findall('(?i)(and|or|the|a|an)', textnl)
re.findall('', textnl)
re2.findall('(?i)(and|or|the|a|an)', textnl)
re.findall('(?i)(and|or|the|a|an)', textnl)
re.findall('(?i)(and|or|the|a|an)', textnl)
re.findall('(.)', text)
re.findall('(.)\1', text)
a="wheels and mee"
re.findall('(.)\1', a)
re.findall(r'(.)\1', a)
re.findall(r'(.){2}', a)
re.findall(r'(.{2})', a)
re.findall(r'(.)\1', a)
re.findall(r'(.)(\1)', a)
re.findall(r'(.)(\1)', text)
re.findall(r'(..(.)(\2)..)', text)
re.findall(r'(..(.)(\2)..)', text)
d="October 21, June 15 and April 12"
re.sub(r'o',r'i',d)
re.sub(r'[aeiou]',r'i',d)
re.sub(r'[ [:punct:]]',r'\n',d)
print(re.sub(r'[ [:punct:]]',r'\n',d))
re.sub(r'[ [:punct:]]',r'\n',d)
d
re.sub(r'([[:alpha:]]+) ([0-9])',r'\2 \1',d)
re.sub(r'([[:alpha:]]+) ([0-9]+)',r'\2 \1',d)
d
re.sub(r'([[:alpha:]]+) ([0-9]+)',r'\2 \1',d)
re.sub(r'([[:alpha:]]+) ([0-9]+)',r'\2 \1',d)
text
re.findall(r'....of....',text)
re.findall(r'........of........',text)
re.findall(r'[[:alpha:]]+ of [[:alpha:]]+', text)
re.findall(r'[[:alpha:]]+ of [[:alpha:]]+', text)
re.findall(r'([[:alpha:]]+) of ([[:alpha:]]+)', text)
re.sub(r'([[:alpha:]]+) of ([[:alpha:]]+)', r"\2's \1", text)
re.sub(r'([[:alpha:]]+) of (the )?([[:alpha:]]+)', r"\3's \1", text)
r=re.sub(r'([[:alpha:]]+) of (the )?([[:alpha:]]+)', r"\3's \1", text)
r
print(r)
r=re.sub(r'([[:alpha:]]+) of (the )?([[:alpha:]]+)', r"\3's \1", text)