Fix issues detected by random-test

bowtie_inspect.cpp

@@ -61,7 +61,7 @@ static void printUsage(ostream& out) {
 	<< "Options:" << endl;
 	if(wrapper == "basic-0") {
 		out << "  --large-index      force inspection of the 'large' index, even if a" << endl
-			<< "                     'small' one is present." << endl;
+		    << "                     'small' one is present." << endl;
 	}
 	out << "  -a/--across <int>  Number of characters across in FASTA output (default: 60)" << endl
 	<< "  -n/--names         Print reference sequence names only" << endl
@@ -74,9 +74,9 @@ static void printUsage(ostream& out) {
 	if(wrapper.empty()) {
 		cerr << endl
 		     << "*** Warning ***" << endl
-			 << "'boowtie-inspect' was run directly.  It is recommended "
-			 << "to use the wrapper script instead."
-			 << endl << endl;
+		     << "'bowtie-inspect' was run directly.  It is recommended "
+		     << "to use the wrapper script instead."
+		     << endl << endl;
 	}
 }
 
@@ -317,7 +317,7 @@ void print_index_sequences(
 			if (textoff_adj - last_text_off > 1)
 				curr_ref_seq += string(textoff_adj - last_text_off - 1, 'N');
 
-			curr_ref_seq.push_back(cat_ref[i]);
+			curr_ref_seq.push_back("ACGT"[(int)cat_ref[i]]);
 			last_text_off = textoff;
 			first = false;
 		}

ebwt_search_backtrack.h

@@ -1596,7 +1596,7 @@ class GreedyDFSRangeSource {
 		ASSERT_ONLY(qualTot += qual0);
 		uint32_t ci = (uint32_t)(_qlen - _mms[0] - 1);
 		// _chars[] is index in terms of RHS-relative depth
-		int c = charToDna5[(int)_chars[ci]];
+		int c = asc2dna[(int)_chars[ci]];
 		assert_lt(c, 4);
 		assert_neq(c, (int)(*_qry)[_mms[0]]);
 		// Second, append the substituted character for the position
@@ -1612,7 +1612,7 @@ class GreedyDFSRangeSource {
 			ASSERT_ONLY(qualTot += qual1);
 			ci = (uint32_t)(_qlen - _mms[1] - 1);
 			// _chars[] is index in terms of RHS-relative depth
-			c = charToDna5[(int)_chars[ci]];
+			c = asc2dna[(int)_chars[ci]];
 			assert_lt(c, 4);
 			assert_neq(c, (int)(*_qry)[_mms[1]]);
 			// Second, append the substituted character for the position
@@ -1627,7 +1627,7 @@ class GreedyDFSRangeSource {
 				ASSERT_ONLY(qualTot += qual2);
 				ci = (uint32_t)(_qlen - _mms[2] - 1);
 				// _chars[] is index in terms of RHS-relative depth
-				c = charToDna5[(int)_chars[ci]];
+				c = asc2dna[(int)_chars[ci]];
 				assert_lt(c, 4);
 				assert_neq(c, (int)(*_qry)[_mms[2]]);
 				// Second, append the substituted character for the position
@@ -1854,7 +1854,7 @@ class EbwtRangeSource : public RangeSource {
 				assert(rc == 'A' || rc == 'C' || rc == 'G' || rc == 'T');
 				ASSERT_ONLY(char oc = (char)qryBuf_[qlen_ - seedRange_.mms[i] - 1]);
 				assert_neq(rc, oc);
-				qryBuf_[qlen_ - seedRange_.mms[i] - 1] = charToDna5[(int)rc];
+				qryBuf_[qlen_ - seedRange_.mms[i] - 1] = asc2dna[(int)rc];
 				assert_neq(rc, (*qry_)[qlen_ - seedRange_.mms[i] - 1]);
 			}
 			qry_ = &qryBuf_;

filebuf.h

@@ -393,7 +393,7 @@ class FileBuf {
 			// if we see caret or EOF, break
 			if(c == '>' || c == -1) break;
 			// append and continue
-			seq[seqcur++] = charToDna5[c];
+			seq[seqcur++] = asc2dna[c];
 			get(); c = peek();
 		}
 	}

hit_set.h

@@ -257,7 +257,7 @@ struct HitSet {
 		assert_lt(sz, 1024);
 		seq.resize(sz);
 		for(size_t j = 0; j < sz; j++) {
-			seq[j] = charToDna5[fb.get()];
+			seq[j] = asc2dna[fb.get()];
 		}
                 qual.resize(sz);
 		fb.get(qual.wbuf(), sz);

pat.cpp

@@ -477,7 +477,7 @@ bool VectorPatternSource::parse(Read& ra, Read& rb, TReadId rdid) const {
 				assert_in(toupper(c), "ACGTN");
 				if(nchar++ >= this->trim5_) {
 					assert_neq(0, asc2dnacat[c]);
-					r.patFw.append(charToDna5[c]); // ascii to int
+					r.patFw.append(asc2dna[c]); // ascii to int
 				}
 			}
 			c = ra.readOrigBuf[cur++];
@@ -610,7 +610,7 @@ bool FastaPatternSource::parse(Read& r, Read& rb, TReadId rdid) const {
 		if(isalpha(c)) {
 			// If it's past the 5'-end trim point
 			if(nchar++ >= this->trim5_) {
-				r.patFw.append(charToDna5[c]);
+				r.patFw.append(asc2dna[c]);
 			}
 		}
 		assert_lt(cur, buflen);
@@ -768,7 +768,7 @@ bool FastaContinuousPatternSource::parse(
 			assert_in(toupper(c), "ACGTN");
 			if(nchar++ >= this->trim5_) {
 				assert_neq(0, asc2dnacat[c]);
-				ra.patFw.append(charToDna5[c]); // ascii to int
+				ra.patFw.append(asc2dna[c]); // ascii to int
 			}
 		}
 	}
@@ -893,7 +893,7 @@ bool FastqPatternSource::parse(Read &r, Read& rb, TReadId rdid) const {
 		if(isalpha(c)) {
 			// If it's past the 5'-end trim point
 			if(nchar++ >= this->trim5_) {
-				r.patFw.append(charToDna5[c]);
+				r.patFw.append(asc2dna[c]);
 			}
 		}
 		assert_lt(cur, buflen);
@@ -994,7 +994,17 @@ pair<bool, int> TabbedPatternSource::nextBatchFromFile(
 			readbuf[readi].readOrigBuf.append(c);
 			c = getc_wrapper();
 		}
-		while(c >= 0 && (c == '\n' || c == '\r') && readi < pt.max_buf_ - 1) {
+                if (c == '\n') {
+			readbuf[readi].readOrigBuf.append(c);
+			c = getc_wrapper();
+			if (c == '\r')
+				readbuf[readi].readOrigBuf.append(c);
+                        else {
+				ungetc_wrapper(c);
+				c = '\n'; // reset to last seen char
+                        }
+                }
+                while(c >= 0 && (c == '\n' || c == '\r') && readi < pt.max_buf_ - 1) {
 			c = getc_wrapper();
 		}
 	}
@@ -1050,7 +1060,7 @@ bool TabbedPatternSource::parse(Read& ra, Read& rb, TReadId rdid) const {
 				assert_in(toupper(c), "ACGTN");
 				if(nchar++ >= this->trim5_) {
 					assert_neq(0, asc2dnacat[c]);
-					r.patFw.append(charToDna5[c]);
+					r.patFw.append(asc2dna[c]);
 				}
 			}
 			c = ra.readOrigBuf[cur++];
@@ -1136,8 +1146,17 @@ pair<bool, int> RawPatternSource::nextBatchFromFile(
 			readbuf[readi].readOrigBuf.append(c);
 			c = getc_wrapper();
 		}
-
-	}
+                if (c == '\n') {
+			readbuf[readi].readOrigBuf.append(c);
+			c = getc_wrapper();
+			if (c == '\r')
+				readbuf[readi].readOrigBuf.append(c);
+			else {
+				ungetc_wrapper(c);
+				c = '\n'; // reset to last character seen
+			}
+                }
+        }
 	while (readi > 0 && readbuf[readi-1].readOrigBuf.length() == 0)
 		readi--;
 	return make_pair(c < 0, readi);
@@ -1161,12 +1180,11 @@ bool RawPatternSource::parse(Read& r, Read& rb, TReadId rdid) const {
 	cur--;
 	while(cur < buflen) {
 		c = r.readOrigBuf[cur++];
-		assert(c != '\r' && c != '\n');
 		if(isalpha(c)) {
 			assert_in(toupper(c), "ACGTN");
 			if(nchar++ >= this->trim5_) {
 				assert_neq(0, asc2dnacat[c]);
-				r.patFw.append(charToDna5[c]);
+				r.patFw.append(asc2dna[c]);
 			}
 		}
 	}

ref_read.cpp

@@ -114,7 +114,7 @@ RefRecord fastaRefReadSize(FileBuf& in,
 			// Output it
 			if(bpout != NULL) {
 				// output nucleotide
-				bpout->write(charToDna5[c]);
+				bpout->write(asc2dna[c]);
 			}
 		} else if(cat == 2) {
 			// It's an N or a gap

ref_read.h

@@ -231,8 +231,8 @@ static RefRecord fastaRefReadAppend(FileBuf& in,
 			// Consume it
 			len++;
 			// Add it to referenece buffer
-			dst.set(charToDna5[c], dstoff++);
-			assert_lt(charToDna5[(int)dst[dst.length()-1]], 4);
+			dst.set(asc2dna[c], dstoff++);
+			assert_lt((int)dst[dstoff-1], 4);
 		}
 		c = in.get();
 		if(rparms.nsToAs && dna4Cat[c] == 2) c = 'A';
@@ -248,8 +248,7 @@ static RefRecord fastaRefReadAppend(FileBuf& in,
 	// ilen = length of buffer before this last sequence was appended.
 	if(rparms.reverse == REF_READ_REVERSE_EACH) {
 		// Find limits of the portion we just appended
-		size_t nlen = dstoff;
-		dst.reverseWindow(ilen, nlen);
+		dst.reverseWindow(ilen, len);
 	}
 	return RefRecord((TIndexOffU)off, (TIndexOffU)len, first);
 }

reference.h

@@ -323,7 +323,7 @@ class BitPairReference {
 						if((*infiles)[i].at(0) == '\\') {
 							(*infiles)[i].erase(0, 1);
 						}
-						osv.push_back(BTRefString((*infiles)[i].c_str()));
+						osv.push_back(BTRefString((*infiles)[i].c_str(), true));
 					}
 				} else {
 					readSequenceFiles(*infiles, osv);

sstring.h

@@ -2619,6 +2619,12 @@ std::ostream& operator<< (std::ostream& os, const SStringFixed<T, S>& str) {
 	return os;
 }
 
+template<typename T>
+std::ostream& operator<<(std::ostream& os, const SString<T>& str) {
+	os << str.toZBuf();
+	return os;
+}
+
 extern uint8_t asc2dna[];
 extern uint8_t asc2col[];
 
@@ -3488,7 +3494,7 @@ class SDnaMaskString : public SStringExpandable<char, S, M> {
 
 typedef SStringExpandable<char, 1024, 2> BTString;
 typedef SDnaStringExpandable<1024, 2>    BTDnaString;
-typedef SDnaStringExpandable<1024, 2>    BTRefString;
 typedef SDnaMaskString<32, 2>            BTDnaMask;
+typedef SString<char>                    BTRefString;
 
 #endif /* SSTRING_H_ */