I know, I know, GIGO, and why am I using -k? (I am using -k because some of the problems that force using -k will still produce perfectly useful output. I suspect fixing this would require re-evaluating the severity of quite a few errors; using -k is way easier.))
I have several poorly-extracted MIBs where an I-D header ended up in the middle of the IMPORTS clause, causing a syntax error and several smiImport data structures with a null module ID. Since lastModule==NULL was used as a flag for "did I ever print anything", and strcmp has unpredictable handling for being passed NULL, I just skipped the inside of the loop if the current import's module is NULL.
Diff follows.
Bill
--- dump-perl.c (revision 6635) +++ dump-perl.c (working copy) @@ -617,11 +617,12 @@ for (i = 0, smiImport = smiGetFirstImport(smiModule); smiImport; i++, - lastModule = smiImport->module, smiImport = smiGetNextImport(smiImport)) { if (i == 0) { fprintSegment(f, INDENT, ""imports" => [\n", 0); } + if (smiImport->module == NULL) + continue;
if ( lastModule == NULL || strcmp(lastModule, smiImport->module)) { if ( lastModule ) { @@ -636,6 +637,7 @@ } fprintSegment(f, 4 * INDENT, "", 0); fprint(f, ""%s",\n", smiImport->name); + lastModule = smiImport->module; } if (lastModule) { fprintSegment(f, 3 * INDENT, "],\n", 0);