various fixes and improvements to sscanf

2 files changed, 48 insertions, 9 deletions

diff --git a/src/library.js b/src/library.js
index 290f71f2..f2f1a3cb 100644
--- a/src/library.js
+++ b/src/library.js
@@ -2244,6 +2244,15 @@ LibraryManager.library = {
   // TODO: Document.
   _scanString__deps: ['_isFloat'],
   _scanString: function(format, get, unget, varargs) {
+    if (!__scanString.whiteSpace) {
+      __scanString.whiteSpace = {};
+      __scanString.whiteSpace[' '.charCodeAt(0)] = 1;
+      __scanString.whiteSpace['\t'.charCodeAt(0)] = 1;
+      __scanString.whiteSpace['\n'.charCodeAt(0)] = 1;
+      __scanString.whiteSpace[' '] = 1;
+      __scanString.whiteSpace['\t'] = 1;
+      __scanString.whiteSpace['\n'] = 1;
+    }
     // Supports %x, %4x, %d.%d, %s, %f, %lf.
     // TODO: Support all format specifiers.
     format = Pointer_stringify(format);
@@ -2251,6 +2260,7 @@ LibraryManager.library = {
     var argsi = 0;
     var fields = 0;
     var argIndex = 0;
+    var next;
     for (var formatIndex = 0; formatIndex < format.length; formatIndex++) {
       if (next <= 0) return fields;
       var next = get();
@@ -2266,11 +2276,14 @@ LibraryManager.library = {
         if (formatIndex != maxSpecifierStart) {
           max_ = parseInt(format.slice(maxSpecifierStart, formatIndex), 10);
         }
-        // TODO: Handle type size modifier.
         var long_ = false;
+        var half = false;
         if (format[formatIndex] == 'l') {
           long_ = true;
           formatIndex++;
+        } else if (format[formatIndex] == 'h') {
+          half = true;
+          formatIndex++;
         }
         var type = format[formatIndex];
         formatIndex++;
@@ -2290,13 +2303,16 @@ LibraryManager.library = {
             buffer.pop();
             unget();
           }
+          unget();
+          next = get();
         } else {
           while ((curr < max_ || isNaN(max_)) && next > 0) {
-            if ((type === 'd' && next >= '0'.charCodeAt(0) && next <= '9'.charCodeAt(0)) ||
-                (type === 'x' && (next >= '0'.charCodeAt(0) && next <= '9'.charCodeAt(0) ||
-                                  next >= 'a'.charCodeAt(0) && next <= 'f'.charCodeAt(0) ||
-                                  next >= 'A'.charCodeAt(0) && next <= 'F'.charCodeAt(0))) ||
-                (type === 's' && (next != ' '.charCodeAt(0) && next != '\t'.charCodeAt(0) && next != '\n'.charCodeAt(0))) &&
+            if (!(next in __scanString.whiteSpace) && // stop on whitespace
+                ((type == 's' ||
+                 ((type === 'd' || type == 'u') && next >= '0'.charCodeAt(0) && next <= '9'.charCodeAt(0)) ||
+                 (type === 'x' && (next >= '0'.charCodeAt(0) && next <= '9'.charCodeAt(0) ||
+                                   next >= 'a'.charCodeAt(0) && next <= 'f'.charCodeAt(0) ||
+                                   next >= 'A'.charCodeAt(0) && next <= 'F'.charCodeAt(0))))) &&
                 (formatIndex >= format.length || next !== format[formatIndex].charCodeAt(0))) { // Stop when we read something that is coming up
               buffer.push(String.fromCharCode(next));
               next = get();
@@ -2311,8 +2327,12 @@ LibraryManager.library = {
         var argPtr = {{{ makeGetValue('varargs', 'argIndex', 'void*') }}};
         argIndex += Runtime.getNativeFieldSize('void*');
         switch (type) {
-          case 'd':
-            {{{ makeSetValue('argPtr', 0, 'parseInt(text, 10)', 'i32') }}}
+          case 'd': case 'u':
+            if (half) {
+              {{{ makeSetValue('argPtr', 0, 'parseInt(text, 10)', 'i16') }}};
+            } else {
+              {{{ makeSetValue('argPtr', 0, 'parseInt(text, 10)', 'i32') }}};
+            }
             break;
           case 'x':
             {{{ makeSetValue('argPtr', 0, 'parseInt(text, 16)', 'i32') }}}
@@ -2332,6 +2352,12 @@ LibraryManager.library = {
             break;
         }
         fields++;
+      } else if (format[formatIndex] in __scanString.whiteSpace) {
+        while (next in __scanString.whiteSpace) {
+          next = get();
+          if (next <= 0) return fields;  // End of input.
+        }
+        unget();
       } else {
         // Not a specifier.
         if (format[formatIndex].charCodeAt(0) !== next) {
diff --git a/tests/runner.py b/tests/runner.py
index ef779783..17acdbd9 100755
--- a/tests/runner.py
+++ b/tests/runner.py
@@ -3808,10 +3808,23 @@ at function.:blag
           printf("|%s|\n", buffy);
           sscanf("cheez somethingmoar\tyet more\n", "cheez %s", buffy);
           printf("|%s|\n", buffy);
+
+          int numverts = -1;
+          printf("%d\n", sscanf("	numverts 1499\n", " numverts %d", &numverts)); // white space is the same, even if tab vs space
+          printf("%d\n", numverts);
+
+          int index;
+          float u, v;
+          short start, count;
+          printf("%d\n", sscanf("	vert 87 ( 0.481565 0.059481 ) 0 1\n", " vert %d ( %f %f ) %hu %hu", &index, &u, &v, &start, &count));
+          printf("%d,%.6f,%.6f,%hu,%hu\n", index, u, v, start, count);
+
           return 0;
         }
         '''
-      self.do_run(src, 'en-us : 2\nen-r : 99\nen : 3\n1.234567, 0.000000\n-3.0300\n|some|\n|something|\n|somethingmoar|')
+      self.do_run(src, 'en-us : 2\nen-r : 99\nen : 3\n1.234567, 0.000000\n-3.0300\n|some|\n|something|\n|somethingmoar|\n' +
+                       '1\n1499\n' +
+                       '5\n87,0.481565,0.059481,0,1\n')
 
     def test_sscanf_2(self):
       # doubles