I'm trying to use regexMatcher from String Manipulation in KNIME but it doesn't work. I'm writing the following: regexMatcher($Document$,"/\w") when I want to extract all sentences that have /s or /p or w/p or /200. However even though I have such cases in my table nothing is retrieved. I will appreciate your help.
Asked
Active
Viewed 1,816 times
0
-
Try `regexMatcher($Document$,"/\\w") = "TRUE"` – Wiktor Stribiżew Oct 04 '16 at 16:53
-
I have updated my question. I have tried to use your suggestion but it didn't work – Regina Oct 04 '16 at 17:00
-
Try `"^\\w*/\\w+$"` – Wiktor Stribiżew Oct 04 '16 at 17:10
-
It doesn't work too. Looks like this functionality not working at all. Can it be the case? – Regina Oct 04 '16 at 18:09
-
If you can find any documentation or source code, we could check. BTW, from what I know, the regex flavor is the same as Java. – Wiktor Stribiżew Oct 04 '16 at 18:10
-
The point is that it works but I get as a result that everything is false and nothing matches the regular expression. Maybe you are familiar with other way of doing the same. I just have to filter these type of rows. – Regina Oct 04 '16 at 18:12
1 Answers
1
I got the following:
|Document |isOK |other|strict|
|--------------|-----|-----|------|
|Some /p with q|True |False|False |
|/200 |True |True |False |
|/p |True |True |True |
|/s |True |True |True |
|w/p |True |False|False |
|no slash |False|False|False |
For the expressions:
- isOK:
regexMatcher($Document$, ".*?/\\w.*")(I guess this is what you are after.) - other:
regexMatcher($Document$, "/\\w.*") - strict:
regexMatcher($Document$, "/\\w")
(Document contains no content after the last visible character.)
The problem you might run into is the escaping for the string manipulator node and the semantics of regexMatcher.
The String literal within there is just a Java String, so you have to escape the \ (and some other characters), so it becomes \\.
The semantics of regexMatcher is to match the whole String, so you have to add .*? (non-greedy match anything) before the value you are looking for and .* (greedy match anything) after the expression you are looking for.
(Obviously if I misunderstood your question, the semantics is probably already is what you want.)
BTW: in case you want to filter, you should check the Rule-based Row Filter node as it offers an option to directly filter by regex. It uses a different escaping rule (for the isOK option):
$Document$ MATCHES ".*?/\w.*" => TRUE(escaping is not allowed within quotes)$Document$ MATCHES /.*?\/\\w.*/ => TRUE(escaping is allowed within slashes (and/,\are need to be escaped, but"is not required))
Gábor Bakos
- 8,982
- 52
- 35
- 52