rsdos: initial_packet field is now returned correctly

 * avoid passing parsedString struct into functions, as the
   scope change can completely break the contents if intentional
   null bytes are present
 * change parseString to allocate memory for string contents
   directly and copy the string into it
 * null characters are appending to the end of the string in
   read_string() itself
 * added parameter to read_string() to force null termination to
   NOT be added to the read string
 * ensure entire initial packet contents are returned by
   getRsdosPacketString(), not just up to the first null byte

src/pyavro_stardust/baseavro.pxd

@@ -1,4 +1,7 @@
 from libcpp.vector cimport vector
+from cpython cimport array
+import array
+
 
 cdef struct parsedString:
     unsigned int toskip
@@ -14,7 +17,7 @@ cdef struct parsedNumericArrayBlock:
 cdef (unsigned int, long) read_long(const unsigned char[:] buf,
         const unsigned int maxlen)
 cdef parsedString read_string(const unsigned char[:] buf,
-        const unsigned int maxlen)
+        const unsigned int maxlen, int addNullTerm=*)
 cdef parsedNumericArrayBlock read_numeric_array(const unsigned char[:] buf,
         const unsigned int maxlen)
 

src/pyavro_stardust/baseavro.pyx

@@ -5,6 +5,7 @@ from cpython.mem cimport PyMem_Malloc, PyMem_Free, PyMem_Realloc
 import zlib, wandio, sys
 cimport cython
 
+
 cdef (unsigned int, long) read_long(const unsigned char[:] buf,
         const unsigned int maxlen):
     cdef unsigned int longlen = 0
@@ -31,7 +32,7 @@ cdef (unsigned int, long) read_long(const unsigned char[:] buf,
     return (longlen + 1, (n >> 1) ^ -(n & 1))
 
 cdef parsedString read_string(const unsigned char[:] buf,
-        const unsigned int maxlen):
+        const unsigned int maxlen, int addNullTerm=True):
     cdef unsigned int skip
     cdef long strlen
     cdef parsedString s
@@ -45,7 +46,15 @@ cdef parsedString read_string(const unsigned char[:] buf,
 
     s.toskip = skip
     s.strlen = strlen
-    s.start = <unsigned char *>&(buf[skip])
+
+    if addNullTerm:
+        s.start = <unsigned char *>PyMem_Malloc(strlen + 1)
+        memcpy(s.start, &(buf[skip]), strlen)
+        s.start[strlen] = b'\x00'
+    else:
+        s.start = <unsigned char *>PyMem_Malloc(strlen)
+        memcpy(s.start, &(buf[skip]), strlen)
+
     return s
 
 cdef parsedNumericArrayBlock read_numeric_array(const unsigned char[:] buf,
@@ -184,17 +193,13 @@ cdef class AvroRecord:
         if attrind < 0 or <unsigned int>attrind >= self.stringcount:
             return 0
 
-        astr = read_string(buf, maxlen)
+        astr = read_string(buf, maxlen, True)
 
         if astr.toskip == 0:
             return 0
 
         self.sizeinbuf += astr.toskip + astr.strlen
-        self.attributes_s[attrind] = <char *>PyMem_Malloc(sizeof(char) * astr.strlen + 1)
-
-        memcpy(self.attributes_s[attrind], astr.start, astr.strlen)
-        self.attributes_s[attrind][astr.strlen] = b'\x00'
-
+        self.attributes_s[attrind] = <char *>astr.start
         return astr.toskip + astr.strlen
 
     @cython.wraparound(False)

src/pyavro_stardust/rsdos.pxd

@@ -1,4 +1,4 @@
-import cython
+cimport cython
 from pyavro_stardust.baseavro cimport AvroRecord, AvroReader, parsedString
 
 cpdef enum RsdosAttribute:
@@ -19,21 +19,20 @@ cpdef enum RsdosAttribute:
     ATTR_RSDOS_LATEST_TIME_USEC = 14
     ATTR_RSDOS_LAST_ATTRIBUTE = 15
 
-@cython.final
 cdef class AvroRsdos(AvroRecord):
 
     cdef unsigned char *packetcontent
-    cdef public int pktcontentlen
+    cdef public unsigned int pktcontentlen
 
     cpdef dict asDict(self)
     cpdef void resetRecord(self)
-    cdef void setRsdosPacketString(self, parsedString astr)
     cpdef bytes getRsdosPacketString(self)
+    cpdef int setRsdosPacketString(self, const unsigned char[:] buf,
+            const unsigned int maxlen)
 
-@cython.final
 cdef class AvroRsdosReader(AvroReader):
     cdef int _parseNextRecord(self, const unsigned char[:] buf,
-            const int maxlen)
+            const unsigned int maxlen)
 
 
 # vim: set sw=4 tabstop=4 softtabstop=4 expandtab :

src/pyavro_stardust/rsdos.pyx

@@ -1,6 +1,7 @@
 
 # cython: language_level=3
 cimport cython
+from cpython.mem cimport PyMem_Free
 from pyavro_stardust.baseavro cimport AvroRecord, read_long, read_string, \
         AvroReader, parsedString
 
@@ -33,7 +34,7 @@ cdef class AvroRsdos(AvroRecord):
         if self.pktcontentlen == 0:
             initpkt = None
         else:
-            initpkt = <bytes>self.packetcontent
+            initpkt = self.getRsdosPacketString()
 
         return {
             "timestamp": self.attributes_l[<int>ATTR_RSDOS_TIMESTAMP],
@@ -57,14 +58,25 @@ cdef class AvroRsdos(AvroRecord):
 
     cpdef void resetRecord(self):
         self.pktcontentlen = 0
+        if self.packetcontent != NULL:
+            PyMem_Free(self.packetcontent)
         super(AvroRsdos, self).resetRecord()
 
-    cdef void setRsdosPacketString(self, parsedString astr):
+    cpdef bytes getRsdosPacketString(self):
+        return <bytes>self.packetcontent[:self.pktcontentlen]
+
+    cpdef int setRsdosPacketString(self, const unsigned char[:] buf,
+            const unsigned int maxlen):
+
+        cdef parsedString astr
+
+        astr = read_string(buf, maxlen, addNullTerm=False)
+        if astr.toskip == 0:
+            return 0
         self.packetcontent = astr.start
         self.pktcontentlen = astr.strlen
-
-    cpdef bytes getRsdosPacketString(self):
-        return <bytes>self.packetcontent
+        self.sizeinbuf += astr.toskip + astr.strlen
+        return 1
 
 @cython.final
 cdef class AvroRsdosReader(AvroReader):
@@ -73,12 +85,11 @@ cdef class AvroRsdosReader(AvroReader):
         super().__init__(filepath)
         self.currentrec = AvroRsdos()
 
-    cdef int _parseNextRecord(self,  const unsigned char[:] buf,
-            const int maxlen):
+    cdef int _parseNextRecord(self, const unsigned char[:] buf,
+            const unsigned int maxlen):
 
-        cdef int offset, offinc
+        cdef unsigned int offset, offinc
         cdef RsdosAttribute i
-        cdef parsedString astr
 
         if maxlen == 0:
             return 0
@@ -89,17 +100,12 @@ cdef class AvroRsdosReader(AvroReader):
         for i in range(0, ATTR_RSDOS_LATEST_TIME_USEC + 1):
             offinc = self.currentrec.parseNumeric(buf[offset:],
                     maxlen - offset, i)
-            if offinc <= 0:
+            if offinc == 0:
                 return 0
             offset += offinc
 
-        astr = read_string(buf[offset:], maxlen - offset)
-        if astr.toskip == 0:
-            return 0
-
-        self.currentrec.setRsdosPacketString(astr)
-        self.currentrec.sizeinbuf += astr.toskip + astr.strlen
-        return 1
+        return self.currentrec.setRsdosPacketString(buf[offset:],
+                maxlen - offset);
 
 
 # vim: set sw=4 tabstop=4 softtabstop=4 expandtab :