From 115c01abcfe01bfb7a7a03317b32ca6d88d84e69 Mon Sep 17 00:00:00 2001
From: Glenn Hickey <glenn.hickey@gmail.com>
Date: Mon, 18 Nov 2024 10:24:00 -0500
Subject: [PATCH] assert written seq strings contain only valid chars

---
 impl/fasta_chunk.c   | 7 +++++++
 impl/fasta_extract.c | 6 ++++++
 2 files changed, 13 insertions(+)

diff --git a/impl/fasta_chunk.c b/impl/fasta_chunk.c
index a11c450..7379909 100644
--- a/impl/fasta_chunk.c
+++ b/impl/fasta_chunk.c
@@ -11,6 +11,7 @@
 #include <float.h>
 #include <getopt.h>
 #include <time.h>
+#include <ctype.h>
 #include "bioioC.h"
 #include "commonC.h"
 #include "sonLib.h"
@@ -85,6 +86,12 @@ static void processSequenceToChunk(void* dest, const char *fastaHeader, const ch
         char *seq_chunk = stString_getSubString(sequence, i, j-i);
         assert(strlen(seq_chunk) == j - i);
 
+        // sanity check
+        for (int64_t k = 0; k < j-i; ++k) {
+            char c = tolower(seq_chunk[k]);
+            assert(c == 'a' || c == 'c' || c == 'g' || c == 't' || c == 'n');
+        }
+
         // print the sequence to the file
         fprintf(chunkFileHandle, "%s\n", seq_chunk);
         free(seq_chunk); // cleanup the fragment
diff --git a/impl/fasta_extract.c b/impl/fasta_extract.c
index 17594fe..3929c90 100644
--- a/impl/fasta_extract.c
+++ b/impl/fasta_extract.c
@@ -10,6 +10,7 @@
 #include <float.h>
 #include <getopt.h>
 #include <time.h>
+#include <ctype.h>
 #include "bioioC.h"
 #include "commonC.h"
 #include "sonLib.h"
@@ -37,6 +38,11 @@ static void report_interval(FILE *output, char *seq_name, int64_t start, int64_t
     int64_t seq_length = (int64_t)stHash_search(sequenceLengths, seq_name);
     assert(0 <= start); assert(start <= end); assert(end <= seq_length);
     char *s = stString_getSubString(sequence, start, end-start);
+    // sanity check
+    for (int64_t i = 0; i < end-start; ++i) {
+        char c = tolower(s[i]);
+        assert(c == 'a' || c == 'c' || c == 'g' || c == 't' || c == 'n');
+    }
     fprintf(output, ">%s|%" PRIi64 "|%" PRIi64 "\n%s\n", seq_name, seq_length, start, s);
     free(s);
 }