How to check if a line has one of the strings in a list?

Question

Possible Duplicate:
Check if multiple strings exist in another string

I am trying to find out if there is a nice and clean way to test for 3 different strings.

Basically I am looping trough a file using a for loop; then I have to check if it contains 1 of the 3 strings that I have set in a list.

So far I have found the multiple if condition check, but it does not feel like is really elegant and efficient:

for line in file
    if "string1" in line or "string2" in line or "string3" in line:
        print "found the string"

I was thinking like creating a list that contains string1, string2 and string3, and check if any of these is contained in the line, but it doesn't seems that i can just compare the list without explicitly loop trough the list, and in that case I am basically in the same conditions as in the multiple if statement that I wrote above.

Is there any smart way to check against multiple strings without writing long if statements or loop trough the elements of a list?

Do you need to search every line or just find the first one? If so, you can optimize by breaking out of the for loop. — Tyler Crompton, Dec 21 '11 at 00:18

Niklas B. · Accepted Answer · 2012-08-27T13:20:00.173

77

strings = ("string1", "string2", "string3")
for line in file:
    if any(s in line for s in strings):
        print "yay!"

edited Aug 27 '12 at 13:20

answered Dec 21 '11 at 00:11

Niklas B.

92,950
18
194
224

3

While the meaning of [`any`](http://docs.python.org/library/functions.html?highlight=any#any) is pretty logical, I've never used it so I had to look it up. Nice code @Niklas. – Brigand Dec 21 '11 at 00:13
This behaves almost exactly the same way. – Tyler Crompton Dec 21 '11 at 00:15
3

It's a function. All `any` does is check if any of the values in the sequence is true-ish. The real magic is using a generator expression for the sequence. – Karl Knechtel Dec 21 '11 at 00:50
@Niklas B.: Sorry for the late comment, but is there a way to print the string that was found? How would you do this. Thank you. – Shankar Kumar Aug 01 '13 at 01:24
@ShankarKumar: I guess something like `next((s in line for s in string), None)` would be useful. I returns the first matched string or `None` otherwise – Niklas B. Aug 01 '13 at 10:38
I meant `next((s for s in string if s in line), None)` – Niklas B. Aug 12 '13 at 09:56
@ShankarKumar I think a simple print line instead of print "yay!" will do :) – Pitto Aug 21 '13 at 08:52
@Pitto: How is that any simpler? – Niklas B. Aug 21 '13 at 19:52
@NiklasB. print line looks to me the shortest and simplest way to print the found string :) – Pitto Aug 26 '13 at 07:55
1

@Pitto: ah, I only see now that your comment was a response to Shabkar :) However, `line` is not the found string. That would be the matching element of `string` – Niklas B. Aug 26 '13 at 11:32
You're right: sorry :) – Pitto Aug 26 '13 at 13:57
How would i go about printing the matching string? `if any(s in line for s in strings): print(s)` – Lyux Jan 27 '18 at 22:22

senderle · Answer 2 · 2014-03-19T23:34:43.930

14

This still loops through the cartesian product of the two lists, but it does it one line:

>>> lines1 = ['soup', 'butter', 'venison']
>>> lines2 = ['prune', 'rye', 'turkey']
>>> search_strings = ['a', 'b', 'c']
>>> any(s in l for l in lines1 for s in search_strings)
True
>>> any(s in l for l in lines2 for s in search_strings)
False

This also have the advantage that any short-circuits, and so the looping stops as soon as a match is found. Also, this only finds the first occurrence of a string from search_strings in linesX. If you want to find multiple occurrences you could do something like this:

>>> lines3 = ['corn', 'butter', 'apples']
>>> [(s, l) for l in lines3 for s in search_strings if s in l]
[('c', 'corn'), ('b', 'butter'), ('a', 'apples')]

If you feel like coding something more complex, it seems the Aho-Corasick algorithm can test for the presence of multiple substrings in a given input string. (Thanks to Niklas B. for pointing that out.) I still think it would result in quadratic performance for your use-case since you'll still have to call it multiple times to search multiple lines. However, it would beat the above (cubic, on average) algorithm.

edited Mar 19 '14 at 23:34

answered Dec 21 '11 at 00:19

senderle

145,869
36
209
233

1

Actually there is. Check out the Aho-Corasick automaton. It can be done in linear time – Niklas B. Mar 19 '14 at 17:28
@NiklasB., thanks, that's quite interesting! Correct me if I'm wrong, but I think the result would still be quadratic, since the OP wants to test multiple lines for substring matches. But that still beats the naive `any` version (cubic, assuming average O(n) performance of `in`). – senderle Mar 19 '14 at 23:38
No it's linear. Building the automaton is linear time and feeding a string into the automaton is linear time as well – Niklas B. Mar 19 '14 at 23:39
Or to be more precise, you build the automaton from the needles once and reuse it for every line – Niklas B. Mar 19 '14 at 23:58
I thought the automaton was built from the needles... – senderle Mar 19 '14 at 23:59
yeah that's what I meant... – Niklas B. Mar 20 '14 at 00:00
Ok, but then that's `O(i * (n + m)) + O(s)`, where `i` is the number of input strings, `n` is the average length of an input string, `s` is the total number of characters in the search strings, and `m` is the average number of matches. – senderle Mar 20 '14 at 00:05
But if you just want a yes/no answer then it's O(s + i*n) = O(s + N) where N is the total input length :P Whatever, the details are not too important – Niklas B. Mar 20 '14 at 00:07

Janne Karila · Answer 3 · 2017-10-27T10:49:08.520

4

One approach is to combine the search strings into a regex pattern as in this answer.

edited Oct 27 '17 at 10:49

answered Dec 21 '11 at 08:23

Janne Karila

24,266
6
53
94

I think a regex is prettier, but doesn't regex have quite a lot of overhead? Can it really be more efficient than a simpler 'if ... or ... or ...:' ? – Sam Redway Oct 27 '17 at 09:14
@SamRedway There is some overhead, but on the other hand the search strings can be combined into one pattern that is processed "in one go". – Janne Karila Oct 27 '17 at 10:16
@SamRedway Hmm ok, a quick test makes the regex look bad in comparison – Janne Karila Oct 27 '17 at 10:36
1

So I would guess that in most use cases readability is more important than efficiency and this is a totally reasonable approach based on that metric. Would say though that the 'more efficient' may be misleading. – Sam Redway Oct 27 '17 at 10:38
@SamRedway Yeah, edited that out now. – Janne Karila Oct 27 '17 at 10:50

How to check if a line has one of the strings in a list?

3 Answers3

Linked