Text files & Regular Expressions

Accessing files with NLTK
Regular Expressions
Accessing additional files
• Python has tools for accessing files from
the local directories and also for
obtaining files from the web.
Python module for web access
• urllib2
– Note – this is for Python 2.x, not Python 3
• Python 3 splits the urllib2 materials over several modules
– import urllib2
– urllib2.urlopen(url [,data][, timeout])
• Establish a link with the server identified in the url and send
either a GET or POST request to retrieve the page.
• The optional data field provides data to send to the server as
part of the request. If the data field is present, the HTTP
request used is POST instead of GET
– Use to fetch content that is behind a form, perhaps a login page
– If used, the data must be encoded properly for including in an
HTTP request. See
• timeout defines time in seconds to be used for blocking
operations such as the connection attempt. If it is not
provided, the system wide default value is used.
URL fetch and use
• urlopen returns a file-like object with
– Same as for files: read(), readline(), readlines(),
fileno(), close()
– New for this class:
• info() – returns meta information about the
document at the URL
• getcode() – returns the HTTP status code sent with
the response (ex: 200, 404)
• geturl() – returns the URL of the page, which may
be different from the URL requested if the server
redirected the request
Reminder, file access
• file.close()
– File no longer available
Where file is the internal name
of the file object
• file.fileno()
– returns the file descriptor, not usually needed.
• file.read([size])
– read at most size bytes. If size not specified, read to
end of file.
• file.readline([size])
– read one line. If size provided, read that many bytes.
Empty string returned if EOF encountered immediately
• file.readlines([sizehint])
– return a list of lines. If sizehint present, return
approximately that number of lines, possibly rounding
to fill a buffer.
URL info
• info() provides the header information that
http returns when the HEAD request is
• ex:
>>> print mypage.info()
Date: Mon, 12 Sep 2011 14:23:44 GMT
Server: Apache/1.3.27 (Unix)
Last-Modified: Tue, 02 Sep 2008 21:12:03 GMT
ETag: "2f0d4-215f-48bdac23"
Accept-Ranges: bytes
Content-Length: 8543
Connection: close
Content-Type: text/html
URL status and code
>>> print mypage.getcode()
>>> print mypage.geturl()
Messy HTML
• HTML is not always perfect.
– Browsers may be forgiving.
– Human and computerized html generators make
• Tools for dealing with imperfect html include
Beautiful Soup.
– Beautiful Soup parses anything you give it, and
does the tree traversal stuff for you. You can tell it
"Find all the links", or "Find all the links of class
externalLink", or "Find all the links whose urls
match "foo.com", or "Find the table heading that's
got bold text, then give me that text."
Exceptions: How to Deal with Error
number = 0
while not 1 <= number <= 10:
number= int(raw_input('Enter number from 1 to 10: '))
if not 1 <= number <= 10:
print 'Your number must be from 1 to 10:'
except ValueError:
print 'That is not a valid integer.'
Here: recognize an error condition and deal with it
book slide
If the named error occurs, the “except” clause is
executed and the loop is terminated.
Checking for failed url fetch
import urllib2
url = raw_input("Enter the URL of the page to fetch: ")
result = page.getcode()
if result == 200:
for line in page:
print line
print page.info()
print page.getcode()
print "Page contains ",linecount," lines."
print "\nBad URL: ", url
The except clause is triggered by any error in the try
The NLP pipeline
Character encoding
• ASCII, Unicode
– American Standard Code for Information
• Everything stored in the computer must be
expressed as a bit pattern.
– For numbers, easy – convert to binary
• For integers, direct conversion
• For real numbers, floating point
– somewhat arbitrary choice of how to represent where the
decimal point is, how much precision for the whole number
part, how much for the exponent.
– For non-numeric characters, some arbitrary choice
of what bit pattern to assign to each character
Coding considerations
• If the numeric interpretation of the bit string
assigned to one character is less than that for
another character, the first will sort to an earlier
• Thus, assign the codes in the sort order desired.
Clearly, A before B
A before or after a?
8 before or after A?
* before or after A, 8?
• Once the choices are made and the code is
constructed, sort order is determined. Any need to
change will have to be dealt with in individual
Representing the bit patterns
• All the encodings can be represented as
numeric values. Example ASCII code for
“K” – two bytes: 0100 1011
– Decimal 75
• familiar, but not really convenient for
representing bits.
– Hexadecimal 4B
• one character for each four bits.
– Octal 113 (_01 001 011)
• one character for each three bits, from the right
The ASCII code
Limitations of ASCII
• Original ASCII used only 7 of the
available 8 bits
– last bit kept for parity checking
• Limited to the number of characters that
can be represented.
• Extended – use the 8th bit
– There are several variations
– See http://www.ascii-code.com/
Extended ASCII Hex 80 to FF
• ASCII is just one encoding example
• ASCII, even extended, does not have
enough space for all needed encodings.
• Different schemes in use present
potential conflict – different codes for
the same symbol, different symbols with
the same code if you deal with more
than one scheme.
• Enter unicode. See unicode.org
From unicode.org
Unicode provides a unique number for every character, no
matter what the platform, no matter what the program, no
matter what the language. The Unicode Standard has been
adopted by such industry leaders as Apple, HP, IBM,
JustSystems, Microsoft, Oracle, SAP, Sun, Sybase, Unisys and
many others. Unicode is required by modern standards such
as XML, Java, ECMAScript (JavaScript), LDAP, CORBA 3.0,
WML, etc., and is the official way to implement ISO/IEC
10646. It is supported in many operating systems, all modern
browsers, and many other products. The emergence of the
Unicode Standard, and the availability of tools supporting it,
are among the most significant recent global software
technology trends.
• There are three encoding forms:
– 8, 16, 32 bits
– UTF-8 includes the ASCII codes
– UTF-16 all commonly used symbols, other
symbols available in pairs of 16-bit units
– UTF-32 when size is not an issue. All
symbols in 32 bit string of bits
Using unicode
Regular Expressions
• Processing text often involves selecting
for specific characteristics
• Regular expressions
– powerful tool for describing the
characteristics of interest
• Access in python: import
– Raw string notation: precede a string with r
– r’\n’means backslash then n, not new line
Regular Expression special characters
• ‘^’ (Caret) Matches the start of the string
• ‘$” matches the end of the string, or just before newline at the end of
a string
• ‘*’ match 0 or more repetitions of the preceding re. 0*1 matches any
number of 0s followed by 1: 1, 01, 001, 0001, etc.
• ‘+’ matches 1 or more repetition. 0+1 matches 01, 001, 0001, etc.,
but not 1
• ‘?’ matches 0 or 1 repetitions. 0?1 matches 1 and 01 only
• {m,n} matches between m and n repetitions. If no n specified,
matches only exactly m repetitions. 0{2,4}1 matches 001, 0001, 00001
• {m,n}? match as few as possible of these. 0{2,4}1 will match 001 if it
is available, or 0001 if no 001 is available, or 00001 if no shorter string
is available.
• \ escape special character, so you can search for * or ? etc
• [ ] used to indicate a set of characters. [abc] will match a or b or c
– range: [0-9A-Za-z] will match any digit or letter, upper or lower case
– Special characters lose meaning in set: [\*] matches \ or *
– ^ = negate the set [^0-9] will match anything except a digit
• | means “or” A|B means the character A or the character B. Options
are tested left to right and the search quits when a match is found.
This gives priority to the symbol listed first.
Python re
import nltk
import re
wordlist = [w for w in nltk.corpus.words.words('en') if w.islower()]
print [w for w in wordlist if re.search('ed$', w)]
matches all words in the list that end in ed
Wildcard . matches any single character
Crossword match example:
[w for w in wordlist if re.search('^..j..t..$', w)]
Crossword match example:
['abjectly', 'adjuster', 'dejected', 'dejectly', 'injector', 'majestic',
'objectee', 'objector', 'rejecter', 'rejector', 'unjilted', 'unjolted', 'unjustly’]
Spot check
• Your Turn: The caret symbol ^ matches the start
of a string, just like the $ matches the end. What
results do we get with the above example if we
leave out both of these, and search for «..j..t..»?
– Think about it first. What do you expect?
– Then run it.
Crossword match example:
['abjectedness', 'abjection', 'abjective', 'abjectly', 'abjectness', 'adjection',
'adjectional', 'adjectival', 'adjectivally', 'adjective', 'adjectively', 'adjectivism',
'adjectivitis', 'adjustable', 'adjustably', 'adjustage', 'adjustation', 'adjuster',
'adjustive', 'adjustment', 'antejentacular', 'antiprojectivity', 'bijouterie',
'coadjustment', 'cojusticiar', 'conjective', 'conjecturable', 'conjecturably',
'conjectural', 'conjecturalist', 'conjecturality', 'conjecturally', 'conjecture',
'conjecturer', 'coprojector', 'counterobjection', 'dejected', 'dejectedly',
'dejectedness', 'dejectile', 'dejection', …
? as optional character
• ? indicates 0 or 1 occurrences
– ^e-?mail$
– matches either email or e-mail
– ^[Ee]-?mail$
• allows either upper or lower case E
• Note that [^Ee] matches anything that is not E,e
– the negation is inside the [ ]
Texting example
[w for w in wordlist if re.search('^[ghi][mno][jlk][def]$', w)
• First letter from ghi, second from mno,
then jlk, then def
['gold', 'golf', 'hold', 'hole']
• Take away the ^ and $
'tinkerlike', 'tinkerly', 'tinkershire', 'tinkershue', 'tinkerwise', 'tinlet', 'titleholder',
'toolholder', 'toolholding', 'touchhole', 'trainless', 'traphole', 'trinkerman',
'trinket', 'trinketer', 'trinketry', 'trinkety', 'triole', 'trioleate', 'triolefin', 'trioleic’, …
Python use of re
• re.search(pattern, string[,flags])
– scan through string looking for pattern. Return
None if not found.
• re.match(pattern, string)
– if zero or more characters at the beginning of
string match the re pattern, return a
correstponding MatchObject instance. Return
None if string does not match the pattern.
• re.split(pattern,string)
– Split string by occurrences of pattern.
from: http://docs.python.org/library/re.html some options not included
\w = word class: equivalent to [a-zA-Z0-9_]
\W = complement of \w – all characters other than letters and digits
>>> re.split('\W+', 'Words, words, words.')
['Words', 'words', 'words', '']
>>> re.split('(\W+)', 'Words, words, words.')
['Words', ', ', 'words', ', ', 'words', '.', '']
>>> re.split('\W+', 'Words, words, words.', 1)
['Words', 'words, words.']
>>> re.split('[a-f]+', '0a3B9', flags=re.IGNORECASE)
['0', '3', '9']
• re.findall(pattern, string[,flags])
– return all non-overlapping matches of
pattern in string, as a list of strings. String
scanned left-to-right. Matches returned in
order found.
Applications of re
• Extract word pieces
> word = 'supercalifragilisticexpialidocious'
>>> re.findall(r'[aeiou]', word)
['u', 'e', 'a', 'i', 'a', 'i', 'i', 'i', 'e', 'i', 'a', 'i', 'o', 'i', 'o', 'u']
>>> len(re.findall(r'[aeiou]', word))
• another
>>> wsj = sorted(set(nltk.corpus.treebank.words()))
>>> fd = nltk.FreqDist(vs for word in wsj
for vs in re.findall(r'[aeiou]{2,}', word))
>>> fd.items()
vu50390:ch3 lcassel$ python re2.py
[('io', 549), ('ea', 476), ('ie', 331), ('ou', 329), ('ai', 261), ('ia', 253), ('ee', 217), ('oo',
174), ('ua', 109), ('au', 106), ('ue', 105), ('ui', 95), ('ei', 86), ('oi', 65), ('oa', 59),
('eo', 39), ('iou', 27), ('eu', 18), ('oe', 15), ('iu', 14), ('ae', 11), ('eau', 10), ('uo', 8),
('ao', 6), ('oui', 6), ('eou', 5), ('uou', 5), ('uee', 4), ('aa', 3), ('ieu', 3), ('uie', 3), ('eei',
2), ('aia', 1), ('aii', 1), ('aiia', 1), ('eea', 1), ('iai', 1), ('iao', 1), ('ioa', 1), ('oei', 1),
('ooi', 1), ('ueui', 1), ('uu', 1)]
Spot check
Your Turn: In the W3C Date Time Format, dates are
represented like this: 2009-12-31. Replace the ? in the
following Python code with a regular expression, in order to
convert the string '2009-12-31' to a list of integers [2009, 12,
[int(n) for n in re.findall(?, '2009-12-31')]
Processing some text
Noting redundancy in English and eliminating internal word vowels:
>>> regexp = r'^[AEIOUaeiou]+|[AEIOUaeiou]+$|[^AEIOUaeiou]'
>>> def compress(word):
... pieces = re.findall(regexp, word)
... return ''.join(pieces)
>>> english_udhr = nltk.corpus.udhr.words('English-Latin1')
>>> print nltk.tokenwrap(compress(w) for w in english_udhr[:75])
Unvrsl Dclrtn of Hmn Rghts Prmble Whrs rcgntn of the inhrnt dgnty and
of the eql and inlnble rghts of all mmbrs of the hmn fmly is the fndtn
of frdm , jstce and pce in the wrld , Whrs dsrgrd and cntmpt fr hmn
rghts hve rsltd in brbrs acts whch hve outrgd the cnscnce of mnknd ,
and the advnt of a wrld in whch hmn bngs shll enjy frdm of spch and
Tabulating combinations
>>> rotokas_words = nltk.corpus.toolbox.words('rotokas.dic')
>>> cvs = [cv for w in rotokas_words for cv in
re.findall\(r'[ptksvr][aeiou]', w)]
>>> cfd = nltk.ConditionalFreqDist(cvs)
>>> cfd.tabulate()
a e i o u
k 418 148 94 420 173
p 83 31 105 34 51
r 187 63 84 89 79
s 0 0 100 2 1
t 47 8 0 148 37
v 93 27 105 48 49
Inspecting the words behind the
>>> cv_word_pairs = [(cv, w) for w in rotokas_words
for cv in re.findall(r'[ptksvr][aeiou]', w)]
>>> cv_index = nltk.Index(cv_word_pairs)
>>> cv_index['su']
>>> cv_index['po']
['kaapo', 'kaapopato', 'kaipori', 'kaiporipie', 'kaiporivira', 'kapo', 'kapoa', 'kapokao',
'kapokapo', 'kapokapo', 'kapokapoa', 'kapokapoa', 'kapokapora', 'kapokapora',
'kapokaporo', 'kapokaporo', 'kapokari', 'kapokarito', 'kapokoa', 'kapoo', 'kapooto',
'kapoovira', 'kapopaa', 'kaporo', 'kaporo', 'kaporopa', 'kaporoto', 'kapoto',
'karokaropo', 'karopo', 'kepo', 'kepoi', 'keposi', 'kepoto']
• Simple approach:
>>> def stem(word):
... for suffix in ['ing', 'ly', 'ed', 'ious', 'ies',
'ive', 'es', 's', 'ment']:
if word.endswith(suffix):
return word[:-len(suffix)]
... return word
Building a stemmer
• Build a disjunction of all suffixes
re.findall(r'^.*(ing|ly|ed|ious|ies|ive|es|s|ment)$', 'processing')
• Take a look. What do we have here?
– r – raw string. Interpret everything just as what you
– ^ from the beginning
– . match anything
– * repeat the match anything 0 or more times
– (ing|ly|ed|ious|ies|ive|es|s|ment) – look for one of
– $ at the end of the string
– ‘processing’ -- the string
– result = ['ing']
To get the whole word
• Need to add ?:
>>> re.findall(r'^.*(?:ing|ly|ed|ious|ies|ive|es|s|ment)$', 'processing')
Split the word into stem and suffix
• Some subtleties involved
>>> re.findall(r'^(.*)(ing|ly|ed|ious|ies|ive|es|s|ment)$', 'processing')
[('process', 'ing')]
Looks ok, but
>>> re.findall(r'^(.*)(ing|ly|ed|ious|ies|ive|es|s|ment)$', 'processes')
[('processe', 's')]
The * is a greedy operator. It takes as much as it can get.
>>> re.findall(r'^(.*?)(ing|ly|ed|ious|ies|ive|es|s|ment)$', 'processes')
[('process', 'es')]
*? is non greedy version.
>>> re.findall(r'^(.*?)(ing|ly|ed|ious|ies|ive|es|s|ment)?$', 'language')
[('language', '')]
? makes the suffix list optional, matches when none
A stemming function
>>> def stem(word):
... regexp = r'^(.*?)(ing|ly|ed|ious|ies|ive|es|s|ment)?$'
... stem, suffix = re.findall(regexp, word)[0]
... return stem
>>> raw = """DENNIS: Listen, strange women lying in ponds distributing swords
... is no basis for a system of government. Supreme executive power derives
... a mandate from the masses, not from some farcical aquatic ceremony."""
>>> tokens = nltk.word_tokenize(raw)
>>> [stem(t) for t in tokens]
['DENNIS', ':', 'Listen', ',', 'strange', 'women', 'ly', 'in', 'pond',
'distribut', 'sword', 'i', 'no', 'basi', 'for', 'a', 'system', 'of', 'govern',
'.', 'Supreme', 'execut', 'power', 'deriv', 'from', 'a', 'mandate', 'from',
'the', 'mass', ',', 'not', 'from', 'some', 'farcical', 'aquatic', 'ceremony', '.']
Note some strange “words” returned as the stem: basi from basis and
deriv and execut etc.
The Porter Stemmer
• Official home:
• The python version
• http://tartarus.org/martin/PorterStemm
>>> from nltk.corpus import gutenberg, nps_chat
>>> moby = nltk.Text(gutenberg.words('melville-moby_dick.txt'))
>>> moby.findall(r"<a> (<.*>) <man>")
monied; nervous; dangerous; white; white; white; pious; queer; good;
mature; white; Cape; great; wise; wise; butterless; white; fiendish;
pale; furious; better; certain; complete; dismasted; younger; brave;
brave; brave; brave
>>> chat = nltk.Text(nps_chat.words())
>>> chat.findall(r"<.*> <.*> <bro>")
you rule bro; telling you bro; u twizted bro
>>> chat.findall(r"<l.*>{3,}")
lol lol lol; lmao lol lol; lol lol lol; la la la la la; la la la; la
la la; lovely lol lol love; lol lol lol.; la la la; la la la
import nltk, re
sent = "Colorless green ideas sleep furiously"
Co{l}or{l}ess green ideas s{l}eep furious{l}y
Colorless {gree}n ideas sleep furiously
Word patterns
>>> from nltk.corpus import brown
>>> hobbies_learned = nltk.Text(brown.words(categories=['hobbies', 'learned']))
>>> hobbies_learned.findall(r"<\w*> <and> <other> <\w*s>")
speed and other activities; water and other liquids; tomb and other
landmarks; Statues and other monuments; pearls and other jewels;
charts and other items; roads and other features; figures and other
objects; military and other areas; demands and other factors;
abstracts and other compilations; iron and other metals
Spot Check
• How would you find all instances of the
pattern as x as y
• example: as easy as pie
• Can you handle this: as pretty as a
More on Stemming
>>> porter = nltk.PorterStemmer()
>>> lancaster = nltk.LancasterStemmer()
>>> [porter.stem(t) for t in tokens]
['DENNI', ':', 'Listen', ',', 'strang', 'women', 'lie', 'in', 'pond',
'distribut', 'sword', 'is', 'no', 'basi', 'for', 'a', 'system', 'of', 'govern',
'.', 'Suprem', 'execut', 'power', 'deriv', 'from', 'a', 'mandat', 'from',
'the', 'mass', ',', 'not', 'from', 'some', 'farcic', 'aquat', 'ceremoni', '.']
>>> [lancaster.stem(t) for t in tokens]
['den', ':', 'list', ',', 'strange', 'wom', 'lying', 'in', 'pond', 'distribut',
'sword', 'is', 'no', 'bas', 'for', 'a', 'system', 'of', 'govern', '.', 'suprem',
'execut', 'pow', 'der', 'from', 'a', 'mand', 'from', 'the', 'mass', ',', 'not',
'from', 'som', 'farc', 'aqu', 'ceremony', '.']
>>> wnl = nltk.WordNetLemmatizer()
Only keeps stems if in dictionary
>>> [wnl.lemmatize(t) for t in tokens]
['DENNIS', ':', 'Listen', ',', 'strange', 'woman', 'lying', 'in', 'pond',
'distributing', 'sword', 'is', 'no', 'basis', 'for', 'a', 'system', 'of',
'government', '.', 'Supreme', 'executive', 'power', 'derives', 'from', 'a',
'mandate', 'from', 'the', 'mass', ',', 'not', 'from', 'some', 'farcical',
'aquatic', 'ceremony', '.']
• We have done split, but it was not very
• Built in re abbreviation for any kind of
white space: \s
>>> re.split(r'\s+', raw)
['Dennis:', 'Listen,', 'strange', 'women', 'lying', 'in', 'ponds',
'distributing', 'swords', 'is', 'no', 'basis', 'for', 'a', 'system', 'of',
'government.', 'Supreme', 'executive', 'power', 'derives',
'from', 'a', 'mandate', 'from', 'the', 'masses,', 'not', 'from',
'some', 'farcical', 'aquatic', 'ceremony.']
• Split on anything other than a word
character (A-Za-z0-9)
>>> re.split(r'\W+', raw)
['', 'When', 'I', 'M', 'a', 'Duchess', 'she', 'said', 'to', 'herself', 'not', 'in',
'a', 'very', 'hopeful', 'tone', 'though', 'I', 'won', 't', 'have', 'any', 'pepper',
'in', 'my', 'kitchen', 'AT', 'ALL', 'Soup', 'does', 'very', 'well', 'without',
'Maybe', 'it', 's', 'always', 'pepper', 'that', 'makes', 'people', 'hot', 'tempered',
Note: I’M became I M
re.findall(r'\w+', raw) Splits on the words, instead of the separators
will first try to match any sequence of word characters. If no match is found, it
will try to match any non-whitespace character (\S is the complement of \s)
followed by further word characters. This means that punctuation is grouped
with any following letters (e.g. 's) but that sequences of two or more punctuation
characters are separated.
Getting there
>>> re.findall(r'\w+|\S\w*', raw)
["'When", 'I', "'M", 'a', 'Duchess', ',', "'", 'she', 'said', 'to', 'herself', ',',
'(not', 'in', 'a', 'very', 'hopeful', 'tone', 'though', ')', ',', "'I", 'won', "'t",
'have', 'any', 'pepper', 'in', 'my', 'kitchen', 'AT', 'ALL', '.', 'Soup', 'does',
'very', 'well', 'without', '-', '-Maybe', 'it', "'s", 'always', 'pepper', 'that',
'makes', 'people', 'hot', '-tempered', ',', "'", '.', '.', '.']
Now get internal marks – ‘M and ‘t
Regular expression symbols
• Summary
Symbol Function
\b Word boundary (zero width)
\d Any decimal digit (equivalent to [0-9])
\D Any non-digit character (equivalent to [^0-9])
\s Any whitespace character (equivalent to [ \t\n\r\f\v]
\S Any non-whitespace character (equivalent to [^ \t\n\r\f\v])
\w Any alphanumeric character (equivalent to [a-zA-Z0-9_])
\W Any non-alphanumeric character (equivalent to [^a-zA-Z0-9_])
\t The tab character
\n The newline character
Tokenizer in Python
>>> text = 'That U.S.A. poster-print costs $12.40...'
>>> pattern = r'''(?x) # set flag to allow verbose regexps
... ([A-Z]\.)+
# abbreviations, e.g. U.S.A.
... | \w+(-\w+)*
# words with optional internal hyphens
... | \$?\d+(\.\d+)?%? # currency and percentages, e.g. $12.40, 82%
... | \.\.\.
# ellipsis
... | [][.,;"'?():-_`] # these are separate tokens
... '''
>>> nltk.regexp_tokenize(text, pattern)
['That', 'U.S.A.', 'poster-print', 'costs', '$12.40', '...']
Spot Check
☼ Describe the class of strings matched by the
following regular expressions.
Test your answers using nltk.re_show().
Exercise for next week
◑ Read in some text from a corpus, tokenize it, and
print the list of all wh-word types that occur. (whwords in English are used in questions, relative
clauses and exclamations: who, which, what, and so
on.) Print them in order. Are any words duplicated in
this list, because of the presence of case distinctions
or punctuation?

similar documents