Simian is a software that detects duplicates in files such as C, C++, Java etc. programming languages and even plain text files.
Here is a demonstrate of c/p detection on Libre Office bug tdf#39593: copy/paste code detector/fixing.
To get a space separated list of .cxx files, run:
find <directory> -name "*.cxx" | grep -v unxlng | tr '\n' ' ' > files
All .cxx file names have been written to
files.txt. The Simian call to get code pieces with at least 30 equivalent is for example:
java -jar simian-2.3.31.jar -threshold=30 -language=c++ `cat files`
Output would be:
Found 34 duplicate lines with fingerprint fda93f5a150ea8d915b59a498eac4979 in the following files: Between lines 639 and 726 in /.../core/vcl/source/bitmap/bitmap.cxx Between lines 766 and 853 in /.../core/vcl/source/bitmap/bitmap.cxx Found 45 duplicate lines with fingerprint fda93f5a150ea8d915b59a498eac4979 in the following files: Between lines 1468 and 1538 in /.../core/sw/qa/extras/uiwriter/uiwriter2.cxx Between lines 1381 and 1451 in /.../core/sw/qa/extras/uiwriter/uiwriter2.cxx Found 59 duplicate lines with fingerprint fda93f5a150ea8d915b59a498eac4979 in the following files: Between lines 1748 and 1820 in /.../core/sc/qa/unit/pivottable_filters_test.cxx Between lines 2278 and 2350 in /.../core/sc/qa/unit/pivottable_filters_test.cxx Found 75 duplicate lines with fingerprint 026938a138dede7a72201d666a827760 in the following files: Between lines 932 and 1049 in /.../core/sax/source/tools/converter.cxx Between lines 1060 and 1177 in /.../core/sax/source/tools/converter.cxx ...
We have selected /…/core/sax/source/tools/converter.cxx file to reduce c/p code. It’s clearly visible that only two lines are different on two functions. Between 932 - 1049 and 1060 - 1177 lines could shrink within one function.
convertDuration(double& rfTime, std::u16string_view rString) convertDuration(double& rfTime, std::string_view rString)
Since two functions are include the same data type of
rfTime variables, there is no need to create a
template. However, data type of
rString variables are different. Consequently, we have to create a template only for rString.
Finally, deleting duplicate code pieces will make the functions look even better and readable.
The patch can be seen at https://gerrit.libreoffice.org/c/core/+/111008. This was my first medium-size contribution to Libre Office.