XML parser segfaults when given an obsContext block

[joemasiero]

digest2/digest2/d2ades.c

Lines 257 to 280 in 7ebad07

	tracklet *parse_ades(const char *filepath) {
	/* Load XML document */
	xmlDocPtr doc = xmlParseFile(filepath);
	if (doc == NULL) {
	fprintf(stderr, "Error: unable to parse file \"%s\"\n", filepath);
	return NULL;
	}
	xmlXPathContextPtr xpathCtx = xmlXPathNewContext(doc);
	xmlXPathObjectPtr optical_nodes = xmlXPathEvalExpression("//ades/optical", xpathCtx);
	if (optical_nodes == NULL) {
	fprintf(stderr, "Error: Nothing found...");
	xmlXPathFreeContext(xpathCtx);
	xmlFreeDoc(doc);
	return NULL;
	}
	tracklet *tk = parse_nodes(optical_nodes);
	xmlXPathFreeContext(xpathCtx);
	xmlFreeDoc(doc);
	return tk;

This parser works find when using the sample.xml data file provided. However, when using an ADES file formatted with (an otherwise valid) obsContext block, it reports:

> ./digest2 -p $DIGEST2 test.xml     
Desig.    RMS Int NEO N22 N18 Other Possibilities
zsh: segmentation fault  ./digest2 -p $DIGEST2 test.xml

removing the <obsData>, <obsBlock> and <obsContext> structures and leaving only the <optical> blocks directly under <ades> (instead of inside <obsData> and <obsBlock>) allows this to work as expected.

[joemasiero]

An additional issue once I overcome the above: If I put in data from a spacecraft (e.g. C58) it parses the tracklets but defines everything as Int=100, NEO=100. However, if I convert the same file to MPC80 and run it again, I get a much more realistic distribution of scores. Perhaps the s/c position isn't being parsed right in the ADES reader? I'll note I'm using geocentric s/c positions for both files

[joemasiero]

So, if you update this line:

digest2/digest2/d2ades.c

Line 267 in 7ebad07

xmlXPathObjectPtr optical_nodes = xmlXPathEvalExpression("//ades/optical", xpathCtx);

to
xmlXPathObjectPtr optical_nodes = xmlXPathEvalExpression("//ades/obsBlock/obsData/optical", xpathCtx);
the code will now parse valid ADES files correctly. Still having the 100-100 problem when reading a valid ADES file

[joemasiero]

This should be closed by PR #2