For 100[percnt] assurance you need a SCIENTIFIC way to validate and trust a beautifier program. The method described in this section will enable the beautifier program to be accepted as "trust-worthy" and reliable.
In order to verify that beautifier programs like bcpp , indent or cb is not damaging or changing the input source-code after formatting, you can use one of the following technique -
bash$ man diff bash$ diff -b --ignore-all-space originalfile formattedfile
Generate the object code from the original input source code using the compiler -
g++ -c myprogram.cpp
Here g++ is GNU C++ compiler. This will create object output myprogram.o
Save this file -
mv myprogram.o myprogram_orig.o
Now run bcpp -
bcpp myprogram.cpp
This will create the formatted output program file myprogram.cpp and move the original file to myprogram.cpp.orig. Compile the new file with -
g++ -c myprogram.cpp
Now use the unix 'diff' command to compare the two object files -
diff myprogram.o myprogram_orig.o
Both these files MUST BE IDENTICAL . This verifies that bcpp is working perfectly. On DOS or Windows 95 you may want to use the free Cygnus Cygwin 'diff' or 'MKS' utilities.
If for some reason you are not able to diff the object files then you MUST use the assembly output as described below.
You can use the assembler output instead of object output from the C++ compiler for doing the comparison. Like -
g++ -S myprogram.cpp
This creates myprogram.s. Verify with -
diff myprogram.s myprogram_orig.s
This step gives 100[percnt] guarantee that your valuable source code is intact and bcpp is JUST doing ONLY formatting and is NOT changing or damaging your code in any way. This method gives you 100[percnt] quality assurance and life term or long term WARRANTY on beautifier programs like 'bcpp', 'cb' or 'indent'.
It is strongly recommended that you do these two steps every time you run beautifier programs like bcpp , indent or cb .
Since you cannot compile the Java source code to machine code and you can compile Java source to byte-codes you cannot use the technique given in Method 2 above. When you do diff on Java class files it will always be different.
In this method, a different technique will be given which can be used to validate any beautifier program for Java. Also this method is quite powerful and can be used to validate any beautifier program for any language like C, C++, PERL, SQL, HTML or Java. Since all beautifier program simply rearrange or insert whitespaces , you can strip all the whitespaces from original source file and dump it to a file called verify1.out and strip all the whitespaces from beautified source file and dump it to a file called verify2.out. Now, do a diff on verify1.out and verify2.out. If there is no difference, then beautifier program is working properly. The method is not 100[percnt] perfect and can catch atleast 98[percnt] of the errors/bugs in the beautifier program. Use this method in conjunction with other methods. But this method is better than not having a verification at all and blindly trusting the beautifier program!!
Note: A whitespace can be one of following - blank space ' ', form-feed '\f', newline '\n', carriage return '\r', horizontal tab '\t' or vertical tab '\v'.
bash$ java StripWhitespaces sample.java > verify1.out bash$ java StripWhitespaces sample_beutified.java > verify2.out bash$ diff verify1.out verify2.out bash$ java StripWhitespaces sample.cpp > verify1.out bash$ java StripWhitespaces sample_beutified.cpp > verify2.out bash$ diff verify1.out verify2.out bash$ java StripWhitespaces sample.sql > verify1.out bash$ java StripWhitespaces sample_beutified.sql > verify2.out bash$ diff verify1.out verify2.out
The source code of StripWhitespaces Java program is not given here. It is left as an exercise for students (you) to write a small program in Java which will simply strip whitespaces from the input text file and output to standard console output. Students are also urged to write this small program (StripWhitespaces) in C, PERL, Unix shell script (Korn, Bourne) and AWK script. Students can see howto the same task can be accomplished in these five different languages and can do comparison of ease of programing. You should put a newline '\n' character after every 50 characters while generating verify1.out and verify2.out so that when you do a diff you can see on which lines differences are coming up. Otherwise, verify*.out files will just contain one line and it will be difficult to pin-point where exactly the beautifier program is failing (got this point ???).
This is a Korn shell script to verify beautifier program. Requires "pdksh*.rpm" from Linux 'contrib' cdrom. Save this file as 'text' file and chmod a+rx on it. You can re-write this shell script in PERL so that you can use it on Window 95/NT or MSDOS. Uncomment the PRGM variable to point to bcpp , cb or indent
#!/bin/ksh # Verification program to check C++ Beautifiers 'bcpp', 'indent' or cb ############################################################ # Copyright # The copyright policy is GNU/GPL. # Author: Al Dev (Alavoor Vasudevan) alavoor[AT]yahoo.com ############################################################ check_beautify_now() { # Remove all the temp files.... \rm -f ${TMP_FILE} \rm -f ${TMP_CPPFILE}*.* FNAME=$1 if [ ! -f ${FNAME} ]; then print "\nError: The file ${FNAME} does not exist!!. Aborting now ...." exit fi \cp -f ${FNAME} ${TMP_CPPFILE}.cpp ${COMPILER} -c ${TMP_CPPFILE}.cpp if [ ! -f ${TMP_CPPFILE}.o ]; then print "Fatal Error: Failed to compile ${FNAME}. Aborting now... " exit fi \mv -f ${TMP_CPPFILE}.o ${TMP_CPPFILE}_orig.o aa=`basename $PRGM` print "\nRunning, verifying $aa on ${FNAME}" ${PRGM} ${TMP_CPPFILE}.cpp ${COMPILER} -c ${TMP_CPPFILE}.cpp \rm -f $TMP_FILE diff ${TMP_CPPFILE}.o ${TMP_CPPFILE}_orig.o 1> $TMP_FILE 2>> $TMP_FILE result="" result=`wc -c $TMP_FILE | awk '{print $1}' ` if [ "$result" = "0" ]; then print "Success!! Beautifier $aa is working properly!!\n" else print "Fatal Error: Something wrong!! Beautifier is not working!!" exit fi # ${COMPILER} -S ${TMP_CPPFILE}.cpp # diff ${TMP_CPPFILE}.s ${TMP_CPPFILE}_orig.s # Remove all the temp files.... \rm -f ${TMP_FILE} \rm -f ${TMP_CPPFILE}*.* } ########## Main of program begins here ##################3 #PRGM=/usr/bin/bcpp #PRGM=/usr/bin/cb PRGM=/usr/bin/indent COMPILER=/usr/bin/g++ TMP_FILE=beautify.tmp TMP_CPPFILE=beautify-tmp_cppfile print -n "Enter the C++ file name <default is *.cpp> : " read ans if [ "$ans" = "" -o "$ans" = " " ]; then ans="ALL" else FILENAME=$ans fi # Remove all the temp files.... \rm -f ${TMP_FILE} \rm -f ${TMP_CPPFILE}*.* if [ "$ans" != "ALL" ]; then check_beautify_now ${FILENAME} else ls *.cpp | while read FILENAME do check_beautify_now ${FILENAME} done fi