code to replace emoticons by "SAD" or "HAPPY" not working properly

Question

So I wanted to replace all the happy emoticons with "HAPPY" and vice versa "SAD" for sad emoticons for a text file. But the code isnt working properly. Though it detects smileys (as of now :-) ), but in the below example its not replacing the emoticon with text, its simply appending the text and that too its appending it two times for reasons I dont seem to understand.

dict_sad={":-(":"SAD", ":(":"SAD", ":-|":"SAD",  ";-(":"SAD", ";-<":"SAD", "|-{":"SAD"}
dict_happy={":-)":"HAPPY",":)":"HAPPY", ":o)":"HAPPY",":-}":"HAPPY",";-}":"HAPPY",":->":"HAPPY",";-)":"HAPPY"}

#THE INPUT TEXT#
a="guys beautifully done :-)" 

for i in a.split():
    for j in dict_happy.keys():
        if set(j).issubset(set(i)):
            print "HAPPY"
            continue
    for k in dict_sad.keys():
        if set(k).issubset(set(i)):
            print "SAD"
            continue
    if str(i)==i.decode('utf-8','replace'):
       print i

THE INPUT TEXT

a="guys beautifully done :-)"              

OUTPUT ("HAPPY" is coming two times, also the emoticon isnt getting away)

guys
-
beautifully
done
HAPPY
HAPPY
:-)

EXPECTED OUTPUT

guys
beautifully
done
HAPPY

Answer 1

You are turning each word and each emoticon to a set; this means you are looking for overlap of individual characters. You probably wanted uses exact matches at most:

for i in a.split():
    for j in dict_happy:
        if j == i:
            print "HAPPY"
            continue
    for k in dict_sad:
        if k == i:
            print "SAD"
            continue

You can iterate over dictionaries directly, no need to call .keys() there. You don't actually appear to be using the dictionary values; you could just do:

for word in a.split():
    if word in dict_happy:
        print "HAPPY"
    if word in dict_sad:
        print "SAD"

and then perhaps use sets instead of dictionaries. This then can be reduced to:

words = set(a.split())
if dict_happy.viewkeys() & words:
    print "HAPPY"
if dict_sad.viewkeys() & words:
    print "SAD"

using the dictionary view on the keys as a set. Still, it would still be better to use sets then:

sad_emoticons = {":-(", ":(", ":-|", ";-(", ";-<", "|-{"}
happy_emoticons = {":-)", ":)", ":o)", ":-}", ";-}", ":->", ";-)"}

words = set(a.split())
if sad_emoticons & words:
    print "HAPPY"
if happy_emoticons & words:
    print "SAD"

If you wanted to remove the emoticon from the text, you'll have to filter the words:

for word in a.split():
    if word in dict_happy:
        print "HAPPY"
    elif word in dict_sad:
        print "SAD"
    else:
        print word

or better still, combine the two dictionaries and use dict.get():

emoticons = {
    ":-(": "SAD", ":(": "SAD", ":-|": "SAD", 
    ";-(": "SAD", ";-<": "SAD", "|-{": "SAD",
    ":-)": "HAPPY",":)": "HAPPY", ":o)": "HAPPY",
    ":-}": "HAPPY", ";-}": "HAPPY", ":->": "HAPPY",
    ";-)": "HAPPY"
}

for word in a.split():
    print emoticons.get(word, word)

Here I pass in the current word both as the look-up key and the default; if the current word is not an emoticon, the word itself is printed, otherwise the word SAD or HAPPY is printed instead.

Answer 2

Instead of dictionaries i've used lists. Makes the code a bit simpler:

list_sad = [":(", ":-("]
list_happy = [":)", ":-)"]

a = "guys beautifully done :-)" 

for i in a.split():
    if i in list_sad:
        print ("SAD")
    elif i in list_happy:
        print ("HAPPY")
    else:
        print (i)