From 59a9469483054a8b11a88a5e5683b152afd3ac34 Mon Sep 17 00:00:00 2001
From: Gordon Williams <gw@pur3.co.uk>
Date: Tue, 29 Oct 2024 10:55:45 +0000
Subject: [PATCH] RegExp: add optimisation for RegExp that is simply checking
 if a string ends with something Speeds up .hash on Bangle.js

---
 ChangeLog              |  1 +
 README_BuildProcess.md |  1 +
 src/jswrap_regexp.c    | 47 ++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 49 insertions(+)

diff --git a/ChangeLog b/ChangeLog
index b0b5bb844..b13e1cd3b 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -53,6 +53,7 @@
             E.nativeCall also checks to see if getting data as a flat string failed (could have caused segfault)
             Fix for regression after #2547 fix (left hand operand of maths with a valueOf method)
             X.on now allocates a new array for each new handler added, stops new handlers being called for the event that's currently being handled (#2559)
+            RegExp: add optimisation for RegExp that is simply checking if a string ends with something
             
      2v24 : Bangle.js2: Add 'Bangle.touchRd()', 'Bangle.touchWr()'
             Bangle.js2: After Bangle.showTestScreen, put Bangle.js into a hard off state (not soft off)
diff --git a/README_BuildProcess.md b/README_BuildProcess.md
index 23f3e3363..f15536444 100644
--- a/README_BuildProcess.md
+++ b/README_BuildProcess.md
@@ -157,6 +157,7 @@ There are some specifically that are useful for cutting a few bytes out of the b
 * `ESPR_NO_SOFTWARE_I2C` - don't build in software I2C support
 * `ESPR_NO_BLUETOOTH_MESSAGES` - don't include text versions of Bluetooth error messages (just the error number)
 * `ESPR_LIMIT_DATE_RANGE` - limits the acceptable range for Date years (saves a few hundred bytes)
+* `ESPR_NO_REGEX_OPTIMISE` - strips out some speed optimisations from the regex library
 
 These are set automatically when `SAVE_ON_FLASH` is set (see `jsutils.h`)
 
diff --git a/src/jswrap_regexp.c b/src/jswrap_regexp.c
index e5307413f..f7e8564e9 100644
--- a/src/jswrap_regexp.c
+++ b/src/jswrap_regexp.c
@@ -317,6 +317,35 @@ JsVar *jswrap_regexp_constructor(JsVar *str, JsVar *flags) {
       jsvObjectSetChild(r, "flags", flags);
   }
   jsvObjectSetChildAndUnLock(r, "lastIndex", jsvNewFromInteger(0));
+#ifndef ESPR_NO_REGEX_OPTIMISE
+  /* Quick shortcut - if we were using regex just to find the end of a string (eg just
+  normal chars and then $ at the end), do it with a single string compare which is faster.
+  We save `endsWith` here and then check in jswrap_regexp_exec */
+  JsvStringIterator it;
+  jsvStringIteratorNew(&it,str,0);
+  char ch = 0;
+  bool noControlChars = true;
+  char buf[32];
+  size_t bufc = 0;
+  while (jsvStringIteratorHasChar(&it)) {
+    bool wasSlash = ch=='\\';
+    if (ch && strchr(".[]()|^*+$",ch)) noControlChars = false;
+    ch = jsvStringIteratorGetCharAndNext(&it);
+    if (wasSlash) {
+      if (!strchr(".\\",ch)) {
+        noControlChars = false;
+        ch = 0;
+      }
+    }
+    if (ch && (wasSlash || ch!='\\') && bufc<sizeof(buf)) buf[bufc++]=ch;
+    if (wasSlash) ch = 0;
+  }
+  jsvStringIteratorFree(&it);
+  if (ch=='$' && noControlChars && bufc<sizeof(buf)) {
+    bufc--; // remove '$'
+    jsvObjectSetChildAndUnLock(r, "endsWith", jsvNewStringOfLength((unsigned int)bufc,buf));
+  }
+#endif
   return r;
 }
 
@@ -359,6 +388,24 @@ Or with groups `/W(o)rld/.exec("Hello World")` returns:
 JsVar *jswrap_regexp_exec(JsVar *parent, JsVar *arg) {
   JsVar *str = jsvAsString(arg);
   JsVarInt lastIndex = jsvObjectGetIntegerChild(parent, "lastIndex");
+#ifndef ESPR_NO_REGEX_OPTIMISE
+  /* Quick shortcut - if we were using regex just to find the end of a string,
+  do it with a single string compare which is faster */
+  JsVar *endsWith = jsvObjectGetChildIfExists(parent, "endsWith");
+  if (endsWith) {
+    int idx = (int)jsvGetStringLength(arg) - (int)jsvGetStringLength(endsWith);
+    if ((lastIndex <= idx) && jsvCompareString(arg, endsWith, (size_t)idx,0,true)==0) {
+      JsVar *rmatch = jsvNewEmptyArray();
+      jsvSetArrayItem(rmatch, 0, endsWith);
+      jsvObjectSetChildAndUnLock(rmatch, "index", jsvNewFromInteger(idx));
+      jsvObjectSetChildAndUnLock(rmatch, "input", str);
+      jsvUnLock(endsWith);
+      return rmatch;
+    }
+    jsvUnLock(endsWith);
+  }
+#endif
+  // Otherwise do proper regex
   JsVar *regex = jsvObjectGetChildIfExists(parent, "source");
   if (!jsvIsString(regex) || lastIndex>(JsVarInt)jsvGetStringLength(str)) {
     jsvUnLock2(str,regex);