From 1d1e1e8bafa1e7a73a41a960dfc1e3b0d314d869 Mon Sep 17 00:00:00 2001
From: Paul Gardner-Stephen <paul@servalproject.org>
Date: Wed, 24 Apr 2019 09:47:51 +0930
Subject: [PATCH] begin implementing filtering of similarities based on
 explanation of common instruction sequences

---
 src/similarity.c | 27 +++++++++++++++++++++++++--
 strings/F00320   | 11 +++++++++++
 2 files changed, 36 insertions(+), 2 deletions(-)
 create mode 100644 strings/F00320

diff --git a/src/similarity.c b/src/similarity.c
index c01432ce..0f58591a 100644
--- a/src/similarity.c
+++ b/src/similarity.c
@@ -93,11 +93,34 @@ int main(int argc,char **argv)
 	if (k>2) {
 	  matches[k]++;
 	}
-	if (k>3) {
+	if (k>2) {
+	  char similarity_name[MAX_SIZE];
+	  snprintf(similarity_name,MAX_SIZE,"strings/");
+	  int offset=strlen(similarity_name);
+	  for(int b=0;b<k;b++) {
+	    snprintf(&similarity_name[offset],MAX_SIZE-offset,"%02X",f1[i+b]);
+	    offset+=2;
+	  }
+	  FILE *f=fopen(similarity_name,"r");
+	  if (f) {
+	    // Get explanation why this match is irrelevant
+	    char line[1024]; line[0]=0;
+	    fgets(line,1024,f);
+	    while (line[0]&&line[strlen(line)-1]=='\r') line[strlen(line)-1]=0;
+	    while (line[0]&&line[strlen(line)-1]=='\n') line[strlen(line)-1]=0;
+	    fprintf(stderr,"Ignoring $%04X = $%04X + %d (%s)\n",
+		    i,j,k,line);
+	    break;
+	    fclose(f);
+	  }
+	  
 	  // Display particularly long matches
 	  printf("$%04X = $%04X :",i,j);
-	  for(int b=0;b<k;b++) printf(" %02X",f1[i+b]);
+	  for(int b=0;b<k;b++) {
+	    printf(" %02X",f1[i+b]);
+	  }
 	  printf("\n");
+	  
 	}
       }
 
diff --git a/strings/F00320 b/strings/F00320
new file mode 100644
index 00000000..b6551cbc
--- /dev/null
+++ b/strings/F00320
@@ -0,0 +1,11 @@
+Branch over following JSR instruction
+A branch with argument $03 skips the following 3 bytes.
+A JSR instruction has the initial byte value of $20.
+Therefore $F0 $XX $20 is part of a 
+Bxx +
+JSR $xxxx
+*
+
+instruction sequence.
+This is a very common construction, and the logical way to 
+conditionally call a routines, and therefore is not copyrightable.