From f14cecb6523bd4275c2858c2d54167c94db25ede Mon Sep 17 00:00:00 2001
From: Charles Plessy <plessy@debian.org>
Date: Thu, 19 Aug 2010 15:49:52 +0900
Subject: [PATCH] Imported Upstream version 0.2.2

---
 NEWS             | 13 +++++++++++++
 TabixReader.java | 31 +++++++++++++++++--------------
 index.c          | 26 ++++++++++++++------------
 main.c           |  4 ++--
 4 files changed, 46 insertions(+), 28 deletions(-)

diff --git a/NEWS b/NEWS
index eac654c..a6f3c70 100644
--- a/NEWS
+++ b/NEWS
@@ -1,3 +1,16 @@
+Beta Release 0.2.2 (28 June, 2010)
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Notable changes:
+
+ * Dropped the VCF3 support. Added VCF4 support.
+
+ * Avoided the function name collision with samtools.
+
+(0.2.2: 28 June 2010, r603)
+
+
+
 Beta Release 0.2.1 (3 June, 2010)
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
diff --git a/TabixReader.java b/TabixReader.java
index 752f71c..40d3951 100644
--- a/TabixReader.java
+++ b/TabixReader.java
@@ -226,7 +226,7 @@ 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));
@@ -236,7 +236,7 @@ public class TabixReader
 				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));
@@ -254,22 +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;
diff --git a/index.c b/index.c
index e5b227c..8eddcec 100644
--- a/index.c
+++ b/index.c
@@ -19,7 +19,7 @@ typedef struct {
 } pair64_t;
 
 #define pair64_lt(a,b) ((a).u < (b).u)
-KSORT_INIT(off, pair64_t, pair64_lt)
+KSORT_INIT(offt, pair64_t, pair64_lt)
 
 typedef struct {
 	uint32_t m, n;
@@ -192,17 +192,19 @@ static int get_intv(ti_index_t *idx, kstring_t *str, ti_intv_t *intv)
 					}
 				} else if ((idx->conf.preset&0xffff) == TI_PRESET_VCF) {
 					// FIXME: the following is NOT tested and is likely to be buggy
-					if (id == 5) { // ALT
-						char *t;
-						int max = 1;
-						for (s = str->s + b; s < str->s + i;) {
-							if (s[i] == 'D') {
-								long x = strtol(s + 1, &t, 10);
-								if (x > max) max = x;
-								s = t + 1;
-							} else ++s;
+					if (id == 4) {
+						if (b < i) intv->end = intv->beg + (i - b);
+					} else if (id == 8) { // look for "END="
+						int c = str->s[i];
+						str->s[i] = 0;
+						s = strstr(str->s + b, "END=");
+						if (s == str->s + b) s += 4;
+						else if (s) {
+							s = strstr(str->s + b, ";END=");
+							if (s) s += 5;
 						}
-						intv->end = intv->beg + max;
+						if (s) intv->end = strtol(s, &s, 0);
+						str->s[i] = c;
 					}
 				}
 			}
@@ -804,7 +806,7 @@ ti_iter_t ti_iter_query(const ti_index_t *idx, int tid, int beg, int end)
 	free(bins);
 	{
 		int l;
-		ks_introsort(off, n_off, off);
+		ks_introsort(offt, n_off, off);
 		// resolve completely contained adjacent blocks
 		for (i = 1, l = 0; i < n_off; ++i)
 			if (off[l].v < off[i].v)
diff --git a/main.c b/main.c
index 79c3708..84e5cfc 100644
--- a/main.c
+++ b/main.c
@@ -6,7 +6,7 @@
 #include "bgzf.h"
 #include "tabix.h"
 
-#define PACKAGE_VERSION "0.2.1 (r582)"
+#define PACKAGE_VERSION "0.2.2 (r603)"
 
 int main(int argc, char *argv[])
 {
@@ -21,7 +21,7 @@ int main(int argc, char *argv[])
 			if (strcmp(optarg, "gff") == 0) conf = ti_conf_gff;
 			else if (strcmp(optarg, "bed") == 0) conf = ti_conf_bed;
 			else if (strcmp(optarg, "sam") == 0) conf = ti_conf_sam;
-			else if (strcmp(optarg, "vcf") == 0) conf = ti_conf_vcf;
+			else if (strcmp(optarg, "vcf") == 0 || strcmp(optarg, "vcf4") == 0) conf = ti_conf_vcf;
 			else if (strcmp(optarg, "psltbl") == 0) conf = ti_conf_psltbl;
 			else {
 				fprintf(stderr, "[main] unrecognized preset '%s'\n", optarg);
-- 
2.30.2