From 11432f309c7b35dbffb0f23b65b65b2cb0e8ee81 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Lars=20B=C3=A4rring?= <lars.barring@smhi.se>
Date: Wed, 22 Oct 2025 16:16:10 +0200
Subject: [PATCH] New lexer rules and udunits2 decodeInput function

---
 lib/scanner.l   | 10 +++++++++-
 prog/udunits2.c | 42 +++++++++++++++++++++++++++++++++++++-----
 2 files changed, 46 insertions(+), 6 deletions(-)

diff --git a/lib/scanner.l b/lib/scanner.l
index 456fe02..1c28a0f 100644
--- a/lib/scanner.l
+++ b/lib/scanner.l
@@ -140,7 +140,9 @@ from                    [Ff][Rr][Oo][Mm]
 since                   [Ss][Ii][Nn][Cc][Ee]
 ref                     [Rr][Ee][Ff]
 per                     [Pp][Ee][Rr]
-
+nanspell     ([nN][aA][nN](\([^()]*\))?)
+infspell     ([iI][nN][fF]([iI][nN][iI][tT][yY])?)
+idchar       [A-Za-z0-9_]
 %Start		ID_SEEN SHIFT_SEEN DATE_SEEN CLOCK_SEEN
 
 %%
@@ -149,6 +151,12 @@ per                     [Pp][Ee][Rr]
 	_restartScanner = 0;
     }
 
+<INITIAL,SHIFT_SEEN>{sign}?{nanspell}{idchar} { yyless(0);}
+<INITIAL,SHIFT_SEEN>{sign}?{infspell}{idchar} { yyless(0);}
+
+<INITIAL,SHIFT_SEEN>{sign}?{nanspell} { yyerror("NaN is not allowed in unit expressions."); return 0; }
+<INITIAL,SHIFT_SEEN>{sign}?{infspell} { yyerror("Infinity is not allowed in unit expressions."); return 0; }
+
 <INITIAL,ID_SEEN>{space}*(@|{after}|{from}|{ref}|{since}){space}* {
     BEGIN SHIFT_SEEN;
     return SHIFT;
diff --git a/prog/udunits2.c b/prog/udunits2.c
index 94c0e0c..dd55942 100644
--- a/prog/udunits2.c
+++ b/prog/udunits2.c
@@ -27,6 +27,7 @@
 #include <stdarg.h>
 #include <stdio.h>
 #include <stdlib.h>
+#include <math.h>
 #include <string.h>
 #ifndef _MSC_VER
 #include <strings.h>
@@ -423,11 +424,42 @@ decodeInput(
 
     ut_free(_haveUnit);
 
-    int nbytes;
-    if (sscanf(input, "%lg %n", &_haveUnitAmount, &nbytes) == 1) {
-        input += nbytes;
-    }
-    else {
+    int nbytes = 0;
+    double amt;
+    const char *p = input;
+    char *endp = NULL;
+
+    errno = 0;
+    amt = strtod(p, &endp);
+
+    if (endp != p) {
+        /* We did consume something that looks like a number? */
+        int next = (unsigned char)*endp;
+
+        /* If the numeric value is non-finite (NaN/Inf) ... */
+        if (!isfinite(amt)) {
+            /* ... but the next char continues an identifier, then that was not an amount.
+               Example: "nanosecond" -> "n" "a" "n" then "o" (identifier char).*/
+            if (isalpha(next) || next == '_') {
+                /* Treat as: no leading number; let ut_parse() see the whole string. */
+                _haveUnitAmount = 1;
+                /* DO NOT advance input. */
+            } else {
+                /* Truly a standalone non-finite amount (e.g. "nan m", "inf s") -> reject */
+                errMsg("NaN or Infinity is not allowed in unit expressions.");
+                return 0;
+            }
+        } else {
+            /* Finite amount is OK: Accept and advance to the remainder. */
+            _haveUnitAmount = amt;
+            input = endp;
+            /* Optional: skip a single ASCII space if present (traditional behavior). */
+            while (*input && isspace((unsigned char)*input)) {
+                ++input;
+            }
+        }
+    } else {
+        /* No leading number parsed at all. */
         _haveUnitAmount = 1;
     }