From 01c9d6d306678fcd1737e85bbbc9e2fad2bdc808 Mon Sep 17 00:00:00 2001
From: John Kerl <kerl@fb.com>
Date: Sat, 28 Apr 2018 11:21:14 -0700
Subject: [PATCH] ssub function

---
 c/dsl/function_manager.c |  3 +++
 c/lib/mvfuncs.c          | 52 ++++++++++++++++++++++++++++++++++++++++
 c/lib/mvfuncs.h          |  2 ++
 3 files changed, 57 insertions(+)

diff --git a/c/dsl/function_manager.c b/c/dsl/function_manager.c
index e0ce8c58b..b4b57b03f 100644
--- a/c/dsl/function_manager.c
+++ b/c/dsl/function_manager.c
@@ -233,6 +233,7 @@ static function_lookup_t FUNCTION_LOOKUP_TABLE[] = {
 	{FUNC_CLASS_STRING, "gsub",     3,0, "Example: '$name=gsub($name, \"old\", \"new\")'\n(replace all)."},
 	{FUNC_CLASS_STRING, "strlen",   1,0, "String length."},
 	{FUNC_CLASS_STRING, "sub",      3,0, "Example: '$name=sub($name, \"old\", \"new\")'\n(replace once)."},
+	{FUNC_CLASS_STRING, "ssub",     3,0, "Like sub but does no regexing. No characters are special."},
 	{FUNC_CLASS_STRING, "substr",   3,0,
 		"substr(s,m,n) gives substring of s from 0-up position m to n \n"
 		"inclusive. Negative indices -len .. -1 alias to 0 .. len-1."},
@@ -1228,6 +1229,8 @@ static rval_evaluator_t* fmgr_alloc_evaluator_from_ternary_func_name(char* fnnm,
 		return rval_evaluator_alloc_from_s_sss_func(sub_no_precomp_func,  parg1, parg2, parg3);
 	} else if (streq(fnnm, "gsub")) {
 		return rval_evaluator_alloc_from_s_sss_func(gsub_no_precomp_func, parg1, parg2, parg3);
+	} else if (streq(fnnm, "ssub")) {
+		return rval_evaluator_alloc_from_s_sss_func(s_sss_ssub_func,      parg1, parg2, parg3);
 	} else if (streq(fnnm, "logifit")) {
 		return rval_evaluator_alloc_from_f_fff_func(f_fff_logifit_func,   parg1, parg2, parg3);
 	} else if (streq(fnnm, "madd")) {
diff --git a/c/lib/mvfuncs.c b/c/lib/mvfuncs.c
index 91166e8c6..2b2082dd6 100644
--- a/c/lib/mvfuncs.c
+++ b/c/lib/mvfuncs.c
@@ -192,6 +192,58 @@ mv_t gsub_precomp_func(mv_t* pval1, regex_t* pregex, string_builder_t* psb, mv_t
 	return mv_from_string(output, free_flags);
 }
 
+// ----------------------------------------------------------------
+// String-substitution with no regexes or special characters.
+// It is assumed that all inputs have already been checked to be strings.
+mv_t s_sss_ssub_func(mv_t* pmvinput, mv_t* pmvold, mv_t* pmvnew) {
+	char* pinput = pmvinput->u.strv;
+	char* pold = pmvold->u.strv;
+	char* pnew = pmvnew->u.strv;
+
+	char* pmatch = strstr(pinput, pold);
+
+	if (pmatch == NULL) {
+		mv_free(pmvold);
+		mv_free(pmvnew);
+		return *pmvinput;
+	} else {
+		// Example:
+		// input: aaaaOOObbbbb
+		// old:   OOO
+		// new:   NNNNN
+		// Output length: strlen(aaaa) + strlen(NNNNN) + strlen(bbbbb)
+
+		// Compute lengths
+		int input_length = strlen(pinput);
+		int old_length = strlen(pold);
+		int new_length = strlen(pnew);
+		int output_length = input_length - old_length + new_length + 1;
+		int pre_length = pmatch - pinput; // the "aaaa" part
+		int post_length = input_length - pre_length - old_length; // the "bbbbb" part
+
+		// Allocate output
+		char* poutput = mlr_malloc_or_die(output_length);
+		char* p = poutput;
+
+		// Populate output
+		strncpy(p, pinput, pre_length);
+		p += pre_length;
+
+		strcpy(p, pnew);
+		p += new_length;
+
+		strcpy(p, &pinput[pre_length + old_length]);
+		p += post_length;
+
+		*p = 0;
+
+		mv_free(pmvinput);
+		mv_free(pmvold);
+		mv_free(pmvnew);
+		return mv_from_string(poutput, FREE_ENTRY_VALUE);
+	}
+}
+
 // ----------------------------------------------------------------
 // https://en.wikipedia.org/wiki/Hamming_weight
 
diff --git a/c/lib/mvfuncs.h b/c/lib/mvfuncs.h
index cc49478a6..fecae7d27 100644
--- a/c/lib/mvfuncs.h
+++ b/c/lib/mvfuncs.h
@@ -193,6 +193,8 @@ mv_t sub_no_precomp_func(mv_t* pval1, mv_t* pval2, mv_t* pval3);
 mv_t sub_precomp_func(mv_t* pval1, regex_t* pregex, string_builder_t* psb, mv_t* pval3);
 mv_t gsub_no_precomp_func(mv_t* pval1, mv_t* pval2, mv_t* pval3);
 mv_t gsub_precomp_func(mv_t* pval1, regex_t* pregex, string_builder_t* psb, mv_t* pval3);
+// String-substitution with no regexes or special characters.
+mv_t s_sss_ssub_func(mv_t* pstring, mv_t* pold, mv_t* pnew);
 
 // ----------------------------------------------------------------
 mv_t s_x_sec2gmt_func(mv_t* pval1);