From b39e5d9507ae6f752466a5d1c3302d6fde05842f Mon Sep 17 00:00:00 2001
From: Jim Monte <jim.monte01@gmail.com>
Date: Mon, 9 Dec 2019 19:09:39 -0500
Subject: [PATCH] Made checking for single-character words faster. The testing
 now also uses less memory and works for 8-bit characters, which would have
 previously caused buffer overruns. It is also more modular and has additional
 documentation.

---
 src/frontend/init.c           | 11 ------
 src/frontend/parser/lexical.c | 73 ++++++++++++++++++++++++++++-------
 src/include/ngspice/cpdefs.h  |  5 ---
 3 files changed, 60 insertions(+), 29 deletions(-)

diff --git a/src/frontend/init.c b/src/frontend/init.c
index fddb48679..5e963e585 100644
--- a/src/frontend/init.c
+++ b/src/frontend/init.c
@@ -12,10 +12,6 @@ Author: 1985 Wayne A. Christopher, U. C. Berkeley CAD Group
 #include "variable.h"
 
 
-char cp_chars[128]; /* used in fcn cp_lexer() from lexical.c */
-
-static char *singlec = "<>;&";
-
 void
 cp_init(void)
 /* called from ft_cpinit() in cpitf.c.
@@ -25,13 +21,6 @@ cp_init(void)
    cp_curin, cp_curout, cp_curerr (defined in streams.c)
 */
 {
-    char *s;
-
-    memset(cp_chars, 0, 128);
-    for (s = singlec; *s; s++)
-        /* break word to right or left of characters <>;&*/
-        cp_chars[(int) *s] = (CPC_BRR | CPC_BRL);
-
     cp_vset("history", CP_NUM, &cp_maxhistlength);
 
     cp_curin = stdin;
diff --git a/src/frontend/parser/lexical.c b/src/frontend/parser/lexical.c
index 32fae3767..e5ba348c6 100644
--- a/src/frontend/parser/lexical.c
+++ b/src/frontend/parser/lexical.c
@@ -7,6 +7,7 @@ Author: 1985 Wayne A. Christopher, U. C. Berkeley CAD Group
  * Initial lexer.
  */
 
+#include "ngspice/defines.h"
 #include "ngspice/ngspice.h"
 #include "ngspice/cpdefs.h"
 
@@ -44,6 +45,23 @@ Author: 1985 Wayne A. Christopher, U. C. Berkeley CAD Group
 #include "ngspice/fteinput.h"
 #include "lexical.h"
 
+/** Constants related to characters that form their own words.
+ ** These expressions will be resolved at compile time */
+#define ID_SOLO_CHAR 1 /* Identifier for special chars */
+
+/* Largest of the special chars */
+#define MAX_SOLO_CHAR1 ('<' > '>' ? '<' : '>')
+#define MAX_SOLO_CHAR2 (MAX_SOLO_CHAR1 > ';' ? MAX_SOLO_CHAR1 : ';')
+#define MAX_SOLO_CHAR (MAX_SOLO_CHAR2 > '&' ? MAX_SOLO_CHAR2 : '&')
+
+/* Smallest of the special chars */
+#define MIN_SOLO_CHAR1 ('<' < '>' ? '<' : '>')
+#define MIN_SOLO_CHAR2 (MIN_SOLO_CHAR1 < ';' ? MIN_SOLO_CHAR1 : ';')
+#define MIN_SOLO_CHAR (MIN_SOLO_CHAR2 < '&' ? MIN_SOLO_CHAR2 : '&')
+
+/* Largest index of solo char array */
+#define MAX_INDEX_SOLO_CHAR (MAX_SOLO_CHAR - MIN_SOLO_CHAR)
+
 static void prompt(void);
 
 extern bool cp_echo;  /* For CDHW patches: defined in variable.c */
@@ -370,19 +388,48 @@ nloop:
             goto ldefault;
 
         default:
-            /* We have to remember the special case $<
-             * here
-             */
-        ldefault:
-            if ((cp_chars[c] & CPC_BRL) && (buf.i > 0))
-                if ((c != '<') || (buf.s[buf.i - 1] != '$'))
-                    newword;
-            push(&buf, c);
-            if (cp_chars[c] & CPC_BRR)
-                if ((c != '<') || (buf.i < 2) || (buf.s[buf.i - 2] != '$'))
-                    newword;
-        }
-    }
+            /* $< is a special case where the '<' is not treated
+             * as a character forming its own word */
+        ldefault: {
+            /* Lookup table for "solo" chars forming their own word */
+            static const char id_solo_chars[MAX_INDEX_SOLO_CHAR + 1] = {
+                ['<' - MIN_SOLO_CHAR] = ID_SOLO_CHAR,
+                ['>' - MIN_SOLO_CHAR] = ID_SOLO_CHAR,
+                [';' - MIN_SOLO_CHAR] = ID_SOLO_CHAR,
+                ['&' - MIN_SOLO_CHAR] = ID_SOLO_CHAR
+            };
+
+            /* Find index into solo chars table */
+            const unsigned int index_char =
+                    (unsigned int) c - (unsigned int) MIN_SOLO_CHAR;
+
+            /* Flag that the current character c is a solo character */
+            const bool f_solo_char = index_char <= MAX_INDEX_SOLO_CHAR &&
+                    id_solo_chars[index_char];
+            bool f_is_dollar_lt = FALSE;
+
+            if (f_solo_char && buf.i > 0) {
+                /* The current char is a character forming its own word,
+                 * unless it is "$<" */
+                if (c == '<' && buf.s[buf.i - 1] == '$') { /* is "$<" */
+                    f_is_dollar_lt = TRUE; /* set flag that "$<" found */
+                }
+                else {
+                    /* not "$<", so terminate current word and start
+                     * another one */
+                     newword;
+                }
+            }
+
+            push(&buf, c); /* Add the current char to the current word */
+
+            if (f_solo_char && !f_is_dollar_lt) {
+                /* Split into a new word if this char forms its own word */
+                newword;
+            }
+        } /* end of ldefault block */
+        } /* end of switch over character value */
+    } /* end of loop over characters */
 
 done:
     if (wlist->wl_word)
diff --git a/src/include/ngspice/cpdefs.h b/src/include/ngspice/cpdefs.h
index 00ba6908e..8fc78dfd8 100644
--- a/src/include/ngspice/cpdefs.h
+++ b/src/include/ngspice/cpdefs.h
@@ -84,11 +84,6 @@ struct alias {
     struct alias *al_prev;
 } ;
 
-/* The current record of what characters are special. */
-
-#define CPC_BRR  004 /* Break word to right of character. */
-#define CPC_BRL  010 /* Break word to left of character. */
-
 
 #define CT_ALIASES  1
 #define CT_LABEL    15