X-Git-Url: http://woldlab.caltech.edu/gitweb/?p=tabix.git;a=blobdiff_plain;f=TabixReader.java;h=0d359c5968ef218020b22bf3394fb049da16fcf6;hp=acaff24ef0349b91f393093d5151b6bc32ca1a9d;hb=HEAD;hpb=c70c92b7a385548d8e670638a93a083401f49e4d

diff --git a/TabixReader.java b/TabixReader.java
index acaff24..0d359c5 100644
--- a/TabixReader.java
+++ b/TabixReader.java
@@ -93,6 +93,8 @@ public class TabixReader
 
 	private static int reg2bins(final int beg, final int _end, final int[] list) {
 		int i = 0, k, end = _end;
+		if (beg >= end) return 0;
+		if (end >= 1<<29) end = 1<<29;
 		--end;
 		list[i++] = 0;
 		for (k =    1 + (beg>>26); k <=    1 + (end>>26); ++k) list[i++] = k;
@@ -215,7 +217,7 @@ public class TabixReader
 		int[] ret = new int[3];
 		colon = reg.indexOf(':'); hyphen = reg.indexOf('-');
 		chr = colon >= 0? reg.substring(0, colon) : reg;
-		ret[1] = colon >= 0? Integer.parseInt(reg.substring(colon+1, hyphen)) - 1 : 0;
+		ret[1] = colon >= 0? Integer.parseInt(reg.substring(colon+1, hyphen >= 0? hyphen : reg.length())) - 1 : 0;
 		ret[2] = hyphen >= 0? Integer.parseInt(reg.substring(hyphen+1)) : 0x7fffffff;
 		ret[0] = chr2tid(chr);
 		return ret;
@@ -224,17 +226,17 @@ public class TabixReader
 	private TIntv getIntv(final String s) {
 		TIntv intv = new TIntv();
 		int col = 0, end = 0, beg = 0;
-		while ((end = s.indexOf('\t', beg)) >= 0) {
+		while ((end = s.indexOf('\t', beg)) >= 0 || end == -1) {
 			++col;
 			if (col == mSc) {
 				intv.tid = chr2tid(s.substring(beg, end));
 			} else if (col == mBc) {
-				intv.beg = intv.end = Integer.parseInt(s.substring(beg, end));
+				intv.beg = intv.end = Integer.parseInt(s.substring(beg, end==-1?s.length():end));
 				if ((mPreset&0x10000) != 0) ++intv.end;
 				else --intv.beg;
 				if (intv.beg < 0) intv.beg = 0;
 				if (intv.end < 1) intv.end = 1;
-			} else { // FIXME: SAM/VCF supports are not tested yet
+			} else { // FIXME: SAM supports are not tested yet
 				if ((mPreset&0xffff) == 0) { // generic
 					if (col == mEc)
 						intv.end = Integer.parseInt(s.substring(beg, end));
@@ -252,22 +254,25 @@ public class TabixReader
 						intv.end = intv.beg + l;
 					}
 				} else if ((mPreset&0xffff) == 2) { // VCF
-					if (col == 5) {
-						String alt = s.substring(beg, end);
-						int i, max = 1;
-						for (i = 0; i < alt.length(); ++i) {
-							if (alt.charAt(i) == 'D') { // deletion
-								int j;
-							    for (j = i; j < alt.length() && alt.charAt(j) >= '0' && alt.charAt(j) <= '9'; ++j);
-								int l = Integer.parseInt(alt.substring(i, j));
-								if (max < l) max = l;
-								i = j - 1;
-							}
+					String alt;
+					alt = end >= 0? s.substring(beg, end) : s.substring(beg);
+					if (col == 4) { // REF
+						if (alt.length() > 0) intv.end = intv.beg + alt.length();
+					} else if (col == 8) { // INFO
+						int e_off = -1, i = alt.indexOf("END=");
+						if (i == 0) e_off = 4;
+						else if (i > 0) {
+							i = alt.indexOf(";END=");
+							if (i >= 0) e_off = i + 5;
+						}
+						if (e_off > 0) {
+							i = alt.indexOf(";", e_off);
+							intv.end = Integer.parseInt(i > e_off? alt.substring(e_off, i) : alt.substring(e_off));
 						}
-						intv.end = intv.beg + max;
 					}
 				}
 			}
+			if (end == -1) break;
 			beg = end + 1;
 		}
 		return intv;
@@ -320,7 +325,9 @@ public class TabixReader
 		TIndex idx = mIndex[tid];
 		int[] bins = new int[MAX_BIN];
 		int i, l, n_off, n_bins = reg2bins(beg, end, bins);
-		min_off = (beg>>TAD_LIDX_SHIFT >= idx.l.length)? idx.l[idx.l.length-1] : idx.l[beg>>TAD_LIDX_SHIFT];
+		if (idx.l.length > 0)
+			min_off = (beg>>TAD_LIDX_SHIFT >= idx.l.length)? idx.l[idx.l.length-1] : idx.l[beg>>TAD_LIDX_SHIFT];
+		else min_off = 0;
 		for (i = n_off = 0; i < n_bins; ++i) {
 			if ((chunks = idx.b.get(bins[i])) != null)
 				n_off += chunks.length;
@@ -332,6 +339,7 @@ public class TabixReader
 				for (int j = 0; j < chunks.length; ++j)
 					if (less64(min_off, chunks[j].v))
 						off[n_off++] = new TPair64(chunks[j]);
+		if (n_off == 0) return null;
 		Arrays.sort(off, 0, n_off);
 		// resolve completely contained adjacent blocks
 		for (i = 1, l = 0; i < n_off; ++i) {
@@ -378,7 +386,7 @@ public class TabixReader
 					System.out.println(s);
 			} else { // a region is specified; random access
 				TabixReader.Iterator iter = tr.query(args[1]); // get the iterator
-				while ((s = iter.next()) != null)
+				while (iter != null && (s = iter.next()) != null)
 					System.out.println(s);
 			}
 		} catch (IOException e) {