# Find calls of rtl::O[U]String::getLength() that check for an empty string (and
# that would better use calls to rtl::O[U]String::isEmpty) in the LibreOffice
# code base.
#
# Emits GCC notes prefixed with "[EmptyLength plugin]".  The re-constructed
# original comparison and the suggested improvement should be taken with a grain
# of salt.
#
# * See <https://fedorahosted.org/gcc-python-plugin/> for details about the GCC
#   Python plugin.
#
# * Requires a recent gcc-python-plugin.git version (post v0.7) including
#   <http://git.fedorahosted.org/git/?p=gcc-python-plugin.git;a=commit;h=
#   2bb77306c37bb03a616c2554cfbf25dc02640b43> adding "fullname" support.
#
# * Can be enabled in a LibreOffice build via
#
#   ./autogen.sh ... CXX='g++ -fplugin=</path-to>/gcc-python-plugin/python.so
#       -fplugin-arg-python-script=</path-to>/emptylength.py'
#
# * Produces false positives like in (sal/rtl/source/bootstrap.cxx:237):
#
#   inline void EnsureNoFinalSlash (rtl::OUString & url)
#   {
#       sal_Int32 i = url.getLength();
#       if (i > 0 && url[i - 1] == '/') {
#           url = url.copy(0, i - 1);
#       }
#   }
#
#   sal/rtl/source/bootstrap.cxx:240:5: note: [EmptyLength plugin] replace
#   "getLength() > 0" with "!isEmpty()"
#
# * Disable ccache (beware g++ pointing to /usr/lib64/ccache/g++, etc.) if you
#   want to modify this script and re-run GCC.  Otherwise, your script might not
#   be re-run.

import gcc

verbose = False

class EmptyLength(gcc.GimplePass):
    def execute(self, fun):
        for bb in fun.cfg.basic_blocks:
            if bb.gimple:
                processBlock(bb.gimple)

def processBlock(statements):
    calls = set ()
    for stmt in statements:
        verbose and stmt.loc and gcc.inform(stmt.loc, 'XXX stmt: "%s"' % stmt)
        if isinstance(stmt, gcc.GimpleCall):
            if isSimpleVar(stmt.lhs) and stmt.fndecl \
                    and isLength(stmt.fndecl.fullname):
                verbose and gcc.inform(stmt.loc, 'XXX match call');
                calls.add(stmt.lhs.addr)
        elif isinstance(stmt, gcc.GimpleAssign):
            if len(stmt.rhs) == 1:
                if isSimpleVar(stmt.lhs) and stmt.exprcode == gcc.VarDecl \
                        and isCall(calls, stmt.rhs[0]):
                    verbose and gcc.inform(stmt.loc, 'XXX match assign');
                    calls.add(stmt.lhs.addr)
            elif len(stmt.rhs) == 2:
                verbose and gcc.inform(stmt.loc, 'XXX check assign');
                checkLength(
                    stmt.loc, calls, stmt.exprcode, stmt.rhs[0], stmt.rhs[1])
        elif isinstance(stmt, gcc.GimpleCond):
            verbose and gcc.inform(stmt.loc, 'XXX check cond');
            checkLength(stmt.loc, calls, stmt.exprcode, stmt.lhs, stmt.rhs)

def checkLength(location, calls, exprcode, lhs, rhs):
    if isCall(calls, lhs):
        if exprcode == gcc.EqExpr and isConst(rhs, 0):
            replace(location, 'getLength() == 0', 'isEmpty()')
        elif exprcode == gcc.GeExpr and isConst(rhs, 1):
            replace(location, 'getLength() >= 1', '!isEmpty()')
        elif exprcode == gcc.GtExpr and isConst(rhs, 0):
            replace(location, 'getLength() > 0', '!isEmpty()')
        elif exprcode == gcc.LeExpr and isConst(rhs, 0):
            replace(location, 'getLength() <= 0', 'isEmpty()')
        elif exprcode == gcc.LtExpr and isConst(rhs, 1):
            replace(location, 'getLength() < 1', 'isEmpty()')
        elif exprcode == gcc.NeExpr and isConst(rhs, 0):
            replace(location, 'getLength() != 0', '!isEmpty()')
    elif isCall(calls, rhs):
        if exprcode == gcc.EqExpr and isConst(lhs, 0):
            replace(location, '0 == getLength()', 'isEmpty()')
        elif exprcode == gcc.GeExpr and isConst(lhs, 0):
            replace(location, '0 >= getLength()', 'isEmpty()')
        elif exprcode == gcc.GtExpr and isConst(lhs, 1):
            replace(location, '1 > getLength()', 'isEmpty()')
        elif exprcode == gcc.LeExpr and isConst(lhs, 1):
            replace(location, '1 <= getLength()', '!isEmpty()')
        elif exprcode == gcc.LtExpr and isConst(lhs, 0):
            replace(location, '0 < getLength()', '!isEmpty()')
        elif exprcode == gcc.NeExpr and isConst(lhs, 0):
            replace(location, '0 != getLength()', '!isEmpty()')

def isLength(fullname):
    return fullname == 'sal_Int32 rtl::OString::getLength() const' \
        or fullname == 'sal_Int32 rtl::OUString::getLength() const' \
        or fullname == 'sal_Int32 rtl::OStringBuffer::getLength() const' \
        or fullname == 'sal_Int32 rtl::OUStringBuffer::getLength() const';

def isCall(calls, tree):
    return isSimpleVar(tree) and tree.addr in calls

def isSimpleVar(tree):
    return isinstance(tree, gcc.VarDecl) and tree.addr

def isConst(tree, value):
    return isinstance(tree, gcc.Constant) and tree.constant == value

def replace(location, found, better):
    gcc.inform(
        location,
        '[EmptyLength plugin] replace "%s" with "%s"' % (found, better))

ps = EmptyLength(name='EmptyLength')
ps.register_after('cfg')