/* 
 * OpenXML (docx) to text conversion.
 * Steve Simon,   Oct 2012
 */
#include <u.h>
#include <libc.h>
#include <bio.h>
#include <xml.h>
#include <ctype.h>

static void para(Biobuf *bp, Elem *ep);

static void
object(Biobuf *bp, Elem *ep)
{
	char *v, *p;

	for(; ep; ep = ep->next)
		if(strcmp(ep->name, "o:OLEObject") == 0)
			if((v = xmlvalue(ep, "ProgID")) != nil){
				/* trim cruft */
				if((p = strrchr(v, '.')) != nil && atoi(p+1) != 0)
					*p = 0;
				Bprint(bp, "%s", v);
			}

}
	
static void
run(Biobuf *bp, Elem *ep)
{
	for(; ep; ep = ep->next){
		if(strcmp(ep->name, "w:drawing") == 0)
			Bprint(bp, "[drawing]");

		if(strcmp(ep->name, "w:pict") == 0)
			Bprint(bp, "[picture]");

		if(strcmp(ep->name, "w:object") == 0 && ep->child){
			Bprint(bp, "[object: ");
			object(bp, ep->child);
			Bprint(bp, "]");
		}

		if(strcmp(ep->name, "w:t") == 0 && ep->pcdata)
			Bprint(bp, "%s", ep->pcdata);

		if(strcmp(ep->name, "w:tab") == 0)
			Bprint(bp, "\t");
	}
}


static void
cell(Biobuf *bp, Elem *ep)
{
	int first;

	first = 1;
	for(; ep; ep = ep->next)
		if(strcmp(ep->name, "w:p") == 0 && ep->child){
			if(! first)
				Bprint(bp, " ");
			first = 0;
			para(bp, ep->child);
		}
}

static void
row(Biobuf *bp, Elem *ep)
{
	for(; ep; ep = ep->next){
		if(strcmp(ep->name, "w:tc") == 0){
			Bprint(bp, "\t");
			cell(bp, ep->child);
		}
	}
	Bprint(bp, "\n");
}

static void
table(Biobuf *bp, Elem *ep)
{
	for(; ep; ep = ep->next)
		if(strcmp(ep->name, "w:tr") == 0 && ep->child)
			row(bp, ep->child);
}

static void
paraprops(Biobuf *bp, Elem *ep)
{
	for(; ep; ep = ep->next)
		if(strcmp(ep->name, "w:numPr") == 0 && ep->child)
			Bprint(bp, "    • ");
}
	
static void
para(Biobuf *bp, Elem *ep)
{
	for(; ep; ep = ep->next){
		if(strcmp(ep->name, "w:pPr") == 0 && ep->child)
			paraprops(bp, ep->child);

		if(strcmp(ep->name, "w:r") == 0 && ep->child)
			run(bp, ep->child);

		if(strcmp(ep->name, "w:hyperlink") == 0 && ep->child)
			para(bp, ep->child);

		/* autogenerated fields, e.g. auto numbered Figures */
		if(strcmp(ep->name, "w:fldSimple") == 0 && ep->child)
			para(bp, ep->child);

	}

}

static void
body(Biobuf *bp, Elem *ep)
{
	for(; ep; ep = ep->next){
		if(strcmp(ep->name, "w:p") == 0 && ep->child){
			para(bp, ep->child);
			Bprint(bp, "\n\n");
		}
		if(strcmp(ep->name, "w:tbl") == 0 && ep->child){
			Bprint(bp, "\n");
			table(bp, ep->child);
		}
	}
}

static void
usage(void)
{
	fprint(2, "usage: %s [-n] [file.xml]\n", argv0);
	exits("usage");
}

void
main(int argc, char *argv[])
{
	int fd;	
	Xml *xp;
	Elem *ep;
	char *err;
	Biobuf bout;

	err = nil;
	ARGBEGIN{
	case 'd':
		xmldebug++;
		break;
	default:
		usage();
	}ARGEND;

	if(argc == 0){
		if((xp = xmlparse(0, 8192, 0)) == nil)
			sysfatal("stdin: %r\n");
	}
	else{
		if((fd = open(argv[0], OREAD)) == -1)
			sysfatal("%s cannot open\n", argv[0]);
		if((xp = xmlparse(fd, 8192, 0)) == nil)
			sysfatal("%s: %r\n", argv[0]);
		close(fd);
	}


	Binit(&bout, 1, OWRITE);
	if((ep = xmllook(xp->root, "/w:document/w:body", nil, nil)) == nil || ep->child == nil)
		err = "bad xml format";
	body(&bout, ep->child);
	Bterm(&bout);

	xmlfree(xp);
	exits(err);
}