java.lang.StackOverflowError with long lines with escapes
Opened this issue · 0 comments
GoogleCodeExporter commented
What steps will reproduce the problem?
1. find a source code with long strings with escapes (unfortunately way too
common in large codebases)
2. try to use the prettifier on them
3. watch StackOverflowError
- What is the expected output? What do you see instead?
The expected output is a prettified output. Instead, one gets an application
crash if StackOverflowError is not handled.
- What version of the product are you using? On what operating system?
Both the error and the proposed patch were observed and tested with java
1.7.0_17 under windows and linux.
- Please provide any additional information below.
The issue seem to be that java regex implementation implements the choices
using recursive calls. Thus, patterns such as (a|b)* when a abd b are sinle
characters result in a lot of stack use if the engine cannot determine a and b
are mutually exclusive (and it does not in most situations).
An example of such failure can be observed by prettifying the ltmain.sh file
from libtool, which is present in most projects, and contains large shell
multiline strings. A copy can be found at
http://svn.r-project.org/R/trunk/tools/ltmain.sh
Similar issue has been observed for java/cpp strings, and python triple quoted
strings
(for instance
https://raw.github.com/numpy/numpy/master/numpy/lib/tests/test_format.py).
The attached patch tries to avoid such issue, by rewriting the original regexes
for strings to remove most choices operators.
This patch has been tested on a very large codebase, with different languages.
Best regards,
antoine
Original issue reported on code.google.com by areil...@gmail.com on 23 Nov 2013 at 2:39
Attachments: