Fixed buf_siz and added/mod unit tests to stress stream parsing.

Author: pykler

tests/test_yajl.py

@@ -123,3 +123,9 @@ def test_ctxIsPassedToAllCallbacks(self):
             "Called self.content_handler.yajl_end_map(%(ctx)s)\n"
             % {'ctx': repr(yajl.addressof(ctx))}
         )
+
+    def test_bufSizeLessThanZeroRaisesException(self):
+        for buf_siz in range(-5, 1):
+            self.failUnlessRaises(
+                yajl.YajlConfigError,
+                yajl.YajlParser, self.content_handler, buf_siz=buf_siz)

tests/test_yajl_c.py

@@ -47,18 +47,26 @@ def assertSameAsGold(self, filename):
         got = self.out.read() #.splitlines(1)
         self.failUnlessEqual(expected, got)
 
+    def resetOutput(self):
+        self.out.seek(0)
+        self.out.truncate()
+
 def _make_test(filename, testname):
     def test(self):
         kwargs = {}
         if testname.startswith('dc_'):
             kwargs['allow_comments'] = False
-        parser = yajl.YajlParser(self.content_handler, **kwargs)
-        with open(filename) as f:
-            try:
-                parser.parse(f)
-            except yajl.YajlError, e:
-                self.out.write('%s\n' %str(e).splitlines()[0])
-        self.assertSameAsGold(filename)
+        for buf_siz in range(32):
+            # try out a range of buffer_sizes to stress stream parsing
+            kwargs['buf_siz'] = buf_siz + 1
+            parser = yajl.YajlParser(self.content_handler, **kwargs)
+            with open(filename) as f:
+                try:
+                    parser.parse(f)
+                except yajl.YajlError, e:
+                    self.out.write('%s\n' %str(e).splitlines()[0])
+            self.assertSameAsGold(filename)
+            self.resetOutput()
     return test
 
 def _add_methods():

yajl/yajl_parse.py

@@ -126,6 +126,9 @@ def __init__(self, content_handler, allow_comments=True, check_utf8=True, buf_si
         `check_utf8` specifies whether utf8 charachters are allowed in the document
         `buf_siz` the number of bytes to process from the input stream at a time
         '''
+        # input validation
+        if buf_siz <= 0:
+            raise YajlConfigError('Buffer Size (buf_siz) must be set > 0')
         c_funcs = (
             YAJL_NULL, YAJL_BOOL, YAJL_INT, YAJL_DBL, YAJL_NUM,
             YAJL_STR, YAJL_SDCT, YAJL_DCTK, YAJL_EDCT, YAJL_SARR,
@@ -189,7 +192,6 @@ def parse(self, f=sys.stdin, ctx=None):
         '''Function to parse a JSON stream.
         Parameters:
           `f`        : file stream to read from
-          `buf_size` : size in bytes of read buffer
           `ctx`      : A ctypes pointer that will be passed to
                        all callback functions as the first param
 
@@ -201,7 +203,7 @@ def parse(self, f=sys.stdin, ctx=None):
         hand = yajl.yajl_alloc( byref(self.callbacks), byref(self.cfg), None, ctx)
         try:
             while 1:
-                fileData = f.read(self.buf_siz-1)
+                fileData = f.read(self.buf_siz)
                 if not fileData:
                     stat = yajl.yajl_parse_complete(hand)
                 else: