From 5ba347e100fc48df4db79f4e388203b42b67c6a9 Mon Sep 17 00:00:00 2001
From: Neil Kindlon <nek3d@d-128-109-67.bootp.Virginia.EDU>
Date: Fri, 7 Nov 2014 13:56:31 -0500
Subject: [PATCH] Fixed bug 100, allowing scientific notation for coords, not
returning INT_MIN for invalid numbers
---
src/utils/general/ParseTools.cpp | 29 ++++++++++++++++++++++++++---
test/merge/expFormat.bed | 1 +
test/merge/test-merge.sh | 10 ++++++++++
3 files changed, 37 insertions(+), 3 deletions(-)
create mode 100644 test/merge/expFormat.bed
diff --git a/src/utils/general/ParseTools.cpp b/src/utils/general/ParseTools.cpp
index 04782a71..e85065ad 100644
--- a/src/utils/general/ParseTools.cpp
+++ b/src/utils/general/ParseTools.cpp
@@ -2,6 +2,9 @@
#include <climits>
#include <cctype>
#include <cstring>
+#include <cstdio>
+#include <cstdlib>
+#include <sstream>
//This functions recognizes only numbers with digits, plus sign, minus sign, decimal point, e, or E. Hexadecimal and pointers not currently supported.
bool isNumeric(const QuickString &str) {
@@ -19,12 +22,26 @@ int str2chrPos(const QuickString &str) {
}
int str2chrPos(const char *str, size_t ulen) {
+
if (ulen == 0) {
ulen = strlen(str);
}
+
+ //first test for exponents / scientific notation
+ bool hasExponent = false;
+ for (size_t i=0; i < ulen; i++) {
+ if (str[i] == 'e' || str[i] == 'E') {
+ std::istringstream ss(str);
+ double retVal;
+ ss >> retVal;
+ return (int)retVal;
+ }
+ }
+
int len=(int)ulen;
if (len < 1 || len > 10) {
- return INT_MIN; //can't do more than 9 digits and a minus sign
+ fprintf(stderr, "***** ERROR: too many digits/characters for integer conversion in string %s. Exiting...\n", str);
+ exit(1);
}
register int sum=0;
@@ -39,9 +56,15 @@ int str2chrPos(const char *str, size_t ulen) {
for (int i=startPos; i < len; i++) {
char currChar = str[i];
+ if (currChar == 'e' || currChar == 'E') {
+ //default to atoi for scientific notation
+ return atoi(str);
+ }
if (!isdigit(currChar)) {
- return INT_MIN;
+ fprintf(stderr, "***** ERROR: illegal character '%c' found in integer conversion of string %s. Exiting...\n", currChar, str);
+ exit(1);
}
+
int dig = currChar - 48; //ascii code for zero.
int power = len -i -1;
@@ -77,7 +100,7 @@ int str2chrPos(const char *str, size_t ulen) {
sum += dig *1000000000;
break;
default:
- return INT_MIN;
+ return 0;
break;
}
}
diff --git a/test/merge/expFormat.bed b/test/merge/expFormat.bed
new file mode 100644
index 00000000..d7bd5182
--- /dev/null
+++ b/test/merge/expFormat.bed
@@ -0,0 +1 @@
+chr1 8e02 830
diff --git a/test/merge/test-merge.sh b/test/merge/test-merge.sh
index 586940bf..c6c80495 100644
--- a/test/merge/test-merge.sh
+++ b/test/merge/test-merge.sh
@@ -518,3 +518,13 @@ chr1 30 100" > exp
$BT merge -i a.bed -iobuf 8192 > obs
check exp obs
rm exp obs
+
+###########################################################
+# Test that scientific notation is allowed for coordinates
+###########################################################
+echo " merge.t43...\c"
+echo \
+"chr1 800 830" > exp
+$BT merge -i expFormat.bed > obs
+check exp obs
+rm obs exp
--
GitLab