Array support works, at least for INTEGERS.

Author: fogzot

ChangeLog

@@ -1,3 +1,15 @@
+2005-03-23  Federico Di Gregorio  <fog@debian.org>
+
+	* psycopg/typecast_basic.c: all the basic casters now respect the
+	passed string length.
+
+	* psycopg/typecast.c (typecast_cast): set curs->caster to self
+	during the type-casting.
+
+	* psycopg/cursor_type.c: added "typecaster" attribute to the
+	cursor (this is safe, cursors can't be shared among threads and
+	the attribute is RO.)
+
 2005-03-22  Federico Di Gregorio  <fog@debian.org>
 
 	* psycopg/typecast_array.c: added some more structure to implement

psycopg/cursor.h

@@ -56,6 +56,7 @@ typedef struct {
     Oid         lastoid;   /* last oid from an insert or InvalidOid */
 
     PyObject *casts;      /* an array (tuple) of typecast functions */
+    PyObject *caster;     /* the current typecaster object */
 
     PyObject *copyfile;   /* file-like used during COPY TO/FROM ops */
     long int  copysize;   /* size of the copy buffer during COPY TO/FROM ops */

psycopg/cursor_type.c

@@ -1217,6 +1217,7 @@ static struct PyMemberDef cursorObject_members[] = {
     {"query", T_STRING, OFFSETOF(query), RO},
     {"row_factory", T_OBJECT, OFFSETOF(tuple_factory), 0},
     {"tzinfo_factory", T_OBJECT, OFFSETOF(tzinfo_factory), 0},
+    {"typecaster", T_OBJECT, OFFSETOF(caster), RO},
 #endif
     {NULL}
 };

psycopg/typecast.c

@@ -154,6 +154,25 @@ typecast_init(PyObject *dict)
     return 0;
 }
 
+/* typecast_get_by_name - get a type object by name (slow!) */
+
+static PyObject*
+typecast_get_by_name(unsigned char *name)
+{
+    PyObject *value, *res = NULL;
+    int ppos = 0;
+
+    while (PyDict_Next(psyco_types, &ppos, NULL, &value)) {
+        if (strcmp(PyString_AsString(((typecastObject*)value)->name),
+                   name) == 0) {
+            res = value;
+            break;
+        }
+    }
+
+    /* borrowed reference */
+    return res;
+}
 
 /* typecast_add - add a type object to the dictionary */
 int
@@ -179,6 +198,8 @@ typecast_add(PyObject *obj, int binary)
         }
     }
 
+    Dprintf("typecast_add:     base caster: %p", type->bcast);
+
     return 0;
 }
 
@@ -196,7 +217,7 @@ static struct memberlist typecastObject_memberlist[] = {
 /* numeric methods */
 
 static PyObject *
-typecast_new(PyObject *name, PyObject *values, PyObject *cast);
+typecast_new(PyObject *name, PyObject *values, PyObject *cast, PyObject *base);
 
 static int
 typecast_coerce(PyObject **pv, PyObject **pw)
@@ -207,7 +228,7 @@ typecast_coerce(PyObject **pv, PyObject **pw)
             args = PyTuple_New(1);
             Py_INCREF(*pw);
             PyTuple_SET_ITEM(args, 0, *pw);
-            coer = typecast_new(NULL, args, NULL);
+            coer = typecast_new(NULL, args, NULL, NULL);
             *pw = coer;
             Py_DECREF(args);
             Py_INCREF(*pv);
@@ -347,7 +368,7 @@ PyTypeObject typecastType = {
 };
 
 static PyObject *
-typecast_new(PyObject *name, PyObject *values, PyObject *cast)
+typecast_new(PyObject *name, PyObject *values, PyObject *cast, PyObject *base)
 {
     typecastObject *obj;
 
@@ -368,7 +389,11 @@ typecast_new(PyObject *name, PyObject *values, PyObject *cast)
 
     obj->pcast = NULL;
     obj->ccast = NULL;
-    
+    obj->bcast = base;
+
+    if (obj->bcast) Py_INCREF(obj->bcast);
+
+    /* FIXME: raise an exception when None is passed as Python caster */
     if (cast && cast != Py_None) {
         Py_INCREF(cast);
         obj->pcast = cast;
@@ -382,27 +407,36 @@ typecast_new(PyObject *name, PyObject *values, PyObject *cast)
 PyObject *
 typecast_from_python(PyObject *self, PyObject *args, PyObject *keywds)
 {
-    PyObject *v, *name, *cast = NULL;
+    PyObject *v, *name, *cast = NULL, *base = NULL;
 
-    static char *kwlist[] = {"values", "name", "castobj", NULL};
+    static char *kwlist[] = {"values", "name", "castobj", "baseobj", NULL};
 
-    if (!PyArg_ParseTupleAndKeywords(args, keywds, "O!|O!O", kwlist, 
+    if (!PyArg_ParseTupleAndKeywords(args, keywds, "O!|O!OO", kwlist, 
                                      &PyTuple_Type, &v,
                                      &PyString_Type, &name,
-                                     &cast)) {
+                                     &cast, &base)) {
         return NULL;
     }
 
-    return typecast_new(name, v, cast);
+    return typecast_new(name, v, cast, base);
 }
 
 PyObject *
 typecast_from_c(typecastObject_initlist *type)
 {
-    PyObject *tuple;
+    PyObject *tuple, *base = NULL;
     typecastObject *obj;
     int i, len = 0;
 
+    /* before doing anything else we look for the base */
+    if (type->base) {
+        base = typecast_get_by_name(type->base);
+        if (!base) {
+            PyErr_Format(Error, "typecast base not found: %s", type->base);
+            return NULL;
+        }
+    }
+
     while (type->values[len] != 0) len++;
 
     tuple = PyTuple_New(len);
@@ -411,9 +445,10 @@ typecast_from_c(typecastObject_initlist *type)
     for (i = 0; i < len ; i++) {
         PyTuple_SET_ITEM(tuple, i, PyInt_FromLong(type->values[i]));
     }
-    
+
+        
     obj = (typecastObject *)
-        typecast_new(PyString_FromString(type->name), tuple, NULL);
+        typecast_new(PyString_FromString(type->name), tuple, NULL, base);
 
     if (obj) {
         obj->ccast = type->cast;
@@ -425,18 +460,26 @@ typecast_from_c(typecastObject_initlist *type)
 PyObject *
 typecast_cast(PyObject *obj, unsigned char *str, int len, PyObject *curs)
 {
+    PyObject *old, *res = NULL;
     typecastObject *self = (typecastObject *)obj;
 
+    /* we don't incref, the caster *can't* die at this point */
+    old = ((cursorObject*)curs)->caster;
+    ((cursorObject*)curs)->caster = obj;
+    
     if (self->ccast) {
         Dprintf("typecast_call: calling C cast function");
-        return self->ccast(str, len, curs);
+        res = self->ccast(str, len, curs);
     }
     else if (self->pcast) {
         Dprintf("typecast_call: calling python callable");
-        return PyObject_CallFunction(self->pcast, "s#O", str, len, curs);
+        res = PyObject_CallFunction(self->pcast, "s#O", str, len, curs);
     }
     else {
         PyErr_SetString(Error, "internal error: no casting function found");
-        return NULL;
     }
+
+    ((cursorObject*)curs)->caster = old;
+
+    return res;
 }

psycopg/typecast.h

@@ -43,6 +43,7 @@ typedef struct {
 
     typecast_function  ccast;  /* the C casting function */
     PyObject          *pcast;  /* the python casting function */
+    PyObject          *bcast;  /* base cast, used by array typecasters */
 } typecastObject;
 
 /* the initialization values are stored here */

psycopg/typecast_array.c

@@ -20,23 +20,97 @@
  */
 
 
-/* the pointer to the datetime module API is initialized by the module init
-   code, we just need to grab it */
-extern PyObject* pyDateTimeModuleP;
-extern PyObject *pyDateTypeP;
-extern PyObject *pyTimeTypeP;
-extern PyObject *pyDateTimeTypeP;
-extern PyObject *pyDeltaTypeP;
+/** typecast_array_scan - scan a string looking for array items **/
+
+#define ASCAN_EOF   0
+#define ASCAN_BEGIN 1
+#define ASCAN_END   2
+#define ASCAN_TOKEN 3
+
+static int
+typecast_array_tokenize(unsigned char *str, int strlength,
+                        int *pos, unsigned char** token, int *length)
+{
+    int i = *pos;
+
+    Dprintf("TOKENIZE for %d, pos = %d", strlength, *pos);
+    
+    while (i < strlength) {
+        switch (str[i]) {
+        case '{':
+            *pos = i+1;
+            return ASCAN_BEGIN;
+
+        case '}':
+            *pos = i+1;
+            return ASCAN_END;
+
+        case ',':
+            *token = &str[*pos];
+            *length = i - *pos;
+            *pos = i+1;
+            Dprintf("TOKENIZE pos = %d, length = %d", *pos, *length);
+            return ASCAN_TOKEN;
+
+        default:
+            i++;
+        }
+    }
+
+    *token = &str[*pos];
+    *length = i - *pos;
+    
+    return ASCAN_EOF;
+}
+
+static int
+typecast_array_scan(unsigned char *str, int strlength,
+                    PyObject *curs, PyObject *base, PyObject *array)
+{
+    int state, length, pos = 0;
+    unsigned char *token;
+    
+    while (1) {
+        state = typecast_array_tokenize(str, strlength, &pos, &token, &length);
+        
+        if (state == ASCAN_TOKEN || state ==  ASCAN_EOF) {
+            PyObject *obj = typecast_cast(base, token, length, curs);
+            if (obj == NULL) return 0;
+            PyList_Append(array, obj);
+            Py_DECREF(obj);
+        }
+        else {
+            Dprintf("** RECURSION (not supported right now)!!");
+        }
+
+        if (state ==  ASCAN_EOF) break;
+    }
+
+    return 1;
+}
 
 /** LONGINTEGERARRAY and INTEGERARRAY - cast integers arrays **/
 
 static PyObject *
 typecast_INTEGERARRAY_cast(unsigned char *str, int len, PyObject *curs)
 {
-    PyObject* obj = NULL;
-     
+    PyObject *obj = NULL;
+    PyObject *base = ((typecastObject*)((cursorObject*)curs)->caster)->bcast;
+    
     if (str == NULL) {Py_INCREF(Py_None); return Py_None;}
+    if (str[0] != '{') {
+        PyErr_SetString(Error, "array does not start with '{'");
+        return NULL;
+    }
+    
+    obj = PyList_New(0);
 
+    /* scan the array skipping the first level of {} */
+    if (typecast_array_scan(&str[1], len-2, curs, base, obj) == 0) {
+        Py_DECREF(obj);
+        obj = NULL;
+    }
+    
     return obj;
 }
 

psycopg/typecast_basic.c

@@ -27,16 +27,22 @@
 static PyObject *
 typecast_INTEGER_cast(unsigned char *s, int len, PyObject *curs)
 {
+    unsigned char buffer[12];
+    
     if (s == NULL) {Py_INCREF(Py_None); return Py_None;}
-    return PyInt_FromString(s, NULL, 0);
+    strncpy(buffer, s, len); buffer[len] = '\0';
+    return PyInt_FromString(buffer, NULL, 0);
 }
 
 /** LONGINTEGER - cast long integers (8 bytes) to python long **/
 
 static PyObject *
 typecast_LONGINTEGER_cast(unsigned char *s, int len, PyObject *curs)
 {
+    unsigned char buffer[24];
+    
     if (s == NULL) {Py_INCREF(Py_None); return Py_None;}
+    strncpy(buffer, s, len); buffer[len] = '\0';
     return PyLong_FromString(s, NULL, 0);
 }
 
@@ -45,7 +51,10 @@ typecast_LONGINTEGER_cast(unsigned char *s, int len, PyObject *curs)
 static PyObject *
 typecast_FLOAT_cast(unsigned char *s, int len, PyObject *curs)
 {
+    unsigned char buffer[64];
+    
     if (s == NULL) {Py_INCREF(Py_None); return Py_None;}
+    strncpy(buffer, s, len); buffer[len] = '\0';
     return PyFloat_FromDouble(atof(s));
 }
 
@@ -70,7 +79,7 @@ typecast_UNICODE_cast(unsigned char *s, int len, PyObject *curs)
     enc = PyDict_GetItemString(psycoEncodings,
                                ((cursorObject*)curs)->conn->encoding);
     if (enc) {
-        return PyUnicode_Decode(s, strlen(s), PyString_AsString(enc), NULL);
+        return PyUnicode_Decode(s, len, PyString_AsString(enc), NULL);
     }
     else {
        PyErr_Format(InterfaceError,
@@ -135,13 +144,17 @@ static PyObject *
 typecast_BINARY_cast(unsigned char *s, int l, PyObject *curs)
 {
     PyObject *res;
-    unsigned char *str;
+    unsigned char *str, saved;
     size_t len;
 
     if (s == NULL) {Py_INCREF(Py_None); return Py_None;}
-   
+
+    /* PQunescapeBytea absolutely wants a 0-terminated string and we don't
+       want to copy the whole buffer, right? Wrong... :/ */
+    saved = s[l]; s[l] = '\0';
     str = PQunescapeBytea(s, &len);
     Dprintf("typecast_BINARY_cast: unescaped %d bytes", len);
+    s[l] = saved;
 
     /* TODO: using a PyBuffer would make this a zero-copy operation but we'll
        need to define our own buffer-derived object to keep a reference to the
@@ -178,8 +191,17 @@ typecast_BOOLEAN_cast(unsigned char *s, int len, PyObject *curs)
 static PyObject *
 typecast_DECIMAL_cast(unsigned char *s, int len, PyObject *curs)
 {
+    PyObject *res = NULL;
+    unsigned char *buffer;
+    
     if (s == NULL) {Py_INCREF(Py_None); return Py_None;}
-    return PyObject_CallFunction(decimalType, "s", s);
+
+    buffer = PyMem_Malloc(len+1);
+    strncpy(buffer, s, len); buffer[len] = '\0';
+    res = PyObject_CallFunction(decimalType, "s", buffer);
+    PyMem_Free(buffer);
+
+    return res;
 }
 #else
 #define typecast_DECIMAL_cast  typecast_FLOAT_cast

setup.cfg

@@ -1,5 +1,5 @@
 [build_ext]
-define=PSYCOPG_EXTENSIONS,PSYCOPG_DISPLAY_SIZE,HAVE_PQFREEMEM,HAVE_PQPROTOCOL3
+define=PSYCOPG_EXTENSIONS,PSYCOPG_DISPLAY_SIZE,HAVE_PQFREEMEM,HAVE_PQPROTOCOL3,PSYCOPG_DEBUG
 # PSYCOPG_DEBUG can be added to enable verbose debug information
 # PSYCOPG_OWN_QUOTING can be added above but it is deprecated