Merge branch '32bit'

nilmdb/client/httpclient.py

@@ -123,14 +123,36 @@ class HTTPClient(object):
         """
         (response, isjson) = self._do_req(method, url, query, body,
                                           stream = True, headers = headers)
+
+        # Like the iter_lines function in Requests, but only splits on
+        # the specified line ending.
+        def lines(source, ending):
+            pending = None
+            for chunk in source:
+                if pending is not None:
+                    chunk = pending + chunk
+                tmp = chunk.split(ending)
+                lines = tmp[:-1]
+                if chunk.endswith(ending):
+                    pending = None
+                else:
+                    pending = tmp[-1]
+                for line in lines:
+                    yield line
+            if pending is not None: # pragma: no cover (missing newline)
+                yield pending
+
+        # Yield the chunks or lines as requested
         if binary:
             for chunk in response.iter_content(chunk_size = 65536):
                 yield chunk
         elif isjson:
-            for line in response.iter_lines():
+            for line in lines(response.iter_content(chunk_size = 1),
+                              ending = '\r\n'):
                 yield json.loads(line)
         else:
-            for line in response.iter_lines():
+            for line in lines(response.iter_content(chunk_size = 65536),
+                              ending = '\n'):
                 yield line
 
     def get_gen(self, url, params = None, binary = False):

nilmdb/cmdline/cmdline.py

@@ -72,10 +72,16 @@ class Complete(object): # pragma: no cover
         path = parsed_args.path
         if not path:
             return []
-        return ( self.escape(k + '=' + v)
-                 for (k,v) in client.stream_get_metadata(path).iteritems()
-                 if k.startswith(prefix) )
-
+        results = []
+        # prefix comes in as UTF-8, but results need to be Unicode,
+        # weird.  Still doesn't work in all cases, but that's bugs in
+        # argcomplete.
+        prefix = nilmdb.utils.unicode.decode(prefix)
+        for (k,v) in client.stream_get_metadata(path).iteritems():
+            kv = self.escape(k + '=' + v)
+            if kv.startswith(prefix):
+                results.append(kv)
+        return results
 
 class Cmdline(object):
 

nilmdb/cmdline/metadata.py

@@ -41,10 +41,10 @@ def cmd_metadata(self):
     if self.args.set is not None or self.args.update is not None:
         # Either set, or update
         if self.args.set is not None:
-            keyvals = self.args.set
+            keyvals = map(nilmdb.utils.unicode.decode, self.args.set)
             handler = self.client.stream_set_metadata
         else:
-            keyvals = self.args.update
+            keyvals = map(nilmdb.utils.unicode.decode, self.args.update)
             handler = self.client.stream_update_metadata
 
         # Extract key=value pairs
@@ -62,7 +62,9 @@ def cmd_metadata(self):
             self.die("error setting/updating metadata: %s", str(e))
     elif self.args.delete is not None:
         # Delete (by setting values to empty strings)
-        keys = self.args.delete or None
+        keys = None
+        if self.args.delete:
+            keys = map(nilmdb.utils.unicode.decode, self.args.delete)
         try:
             data = self.client.stream_get_metadata(self.args.path, keys)
             for key in data:
@@ -72,7 +74,9 @@ def cmd_metadata(self):
             self.die("error deleting metadata: %s", str(e))
     else:
         # Get (or unspecified)
-        keys = self.args.get or None
+        keys = None
+        if self.args.get:
+            keys = map(nilmdb.utils.unicode.decode, self.args.get)
         try:
             data = self.client.stream_get_metadata(self.args.path, keys)
         except nilmdb.client.ClientError as e:
@@ -81,4 +85,6 @@ def cmd_metadata(self):
             # Print nonexistant keys as having empty value
             if value is None:
                 value = ""
-            printf("%s=%s\n", key, value)
+            printf("%s=%s\n",
+                   nilmdb.utils.unicode.encode(key),
+                   nilmdb.utils.unicode.encode(value))

nilmdb/server/rocket.c

@@ -5,6 +5,9 @@
 #include <ctype.h>
 #include <stdint.h>
 
+#define __STDC_FORMAT_MACROS
+#include <inttypes.h>
+
 /* Values missing from stdint.h */
 #define UINT8_MIN 0
 #define UINT16_MIN 0
@@ -19,13 +22,6 @@
 
 typedef int64_t timestamp_t;
 
-/* This code probably needs to be double-checked for the case where
-   sizeof(long) != 8, so enforce that here with something that will
-   fail at build time.  We assume that the python integer type can
-   hold an int64_t. */
-const static char __long_ok[1 - 2*!(sizeof(int64_t) ==
-				    sizeof(long int))] = { 0 };
-
 /* Somewhat arbitrary, just so we can use fixed sizes for strings
    etc. */
 static const int MAX_LAYOUT_COUNT = 1024;
@@ -58,7 +54,7 @@ static PyObject *raise_str(int line, int col, int code, const char *string)
 static PyObject *raise_int(int line, int col, int code, int64_t num)
 {
 	PyObject *o;
-	o = Py_BuildValue("(iiil)", line, col, code, num);
+	o = Py_BuildValue("(iiiL)", line, col, code, (long long)num);
 	if (o != NULL) {
 		PyErr_SetObject(ParseError, o);
 		Py_DECREF(o);
@@ -249,11 +245,11 @@ static PyObject *Rocket_get_file_size(Rocket *self)
 /****
  * Append from string
  */
-static inline long int strtol10(const char *nptr, char **endptr) {
-	return strtol(nptr, endptr, 10);
+static inline long int strtoll10(const char *nptr, char **endptr) {
+	return strtoll(nptr, endptr, 10);
 }
-static inline long int strtoul10(const char *nptr, char **endptr) {
-	return strtoul(nptr, endptr, 10);
+static inline long int strtoull10(const char *nptr, char **endptr) {
+	return strtoull(nptr, endptr, 10);
 }
 
 /* .append_string(count, data, offset, linenum, start, end, last_timestamp) */
@@ -264,6 +260,7 @@ static PyObject *Rocket_append_string(Rocket *self, PyObject *args)
 	int offset;
 	const char *linestart;
 	int linenum;
+        long long ll1, ll2, ll3;
 	timestamp_t start;
 	timestamp_t end;
 	timestamp_t last_timestamp;
@@ -280,10 +277,13 @@ static PyObject *Rocket_append_string(Rocket *self, PyObject *args)
 	   but we need the null termination for strto*.  If we had
 	   strnto* that took a length, we could use t# and not require
 	   a copy. */
-	if (!PyArg_ParseTuple(args, "isiilll:append_string", &count,
+	if (!PyArg_ParseTuple(args, "isiiLLL:append_string", &count,
 			      &data, &offset, &linenum,
-			      &start, &end, &last_timestamp))
+			      &ll1, &ll2, &ll3))
 		return NULL;
+        start = ll1;
+        end = ll2;
+        last_timestamp = ll3;
 
 	/* Skip spaces, but don't skip over a newline. */
 #define SKIP_BLANK(buf) do {			\
@@ -372,14 +372,14 @@ static PyObject *Rocket_append_string(Rocket *self, PyObject *args)
 				goto extra_data_on_line;		\
 			break
 
-			CS(INT8,   strtol10,  t64.i, t8.i,  t8.u,         , 1);
-			CS(UINT8,  strtoul10, t64.u, t8.u,  t8.u,         , 1);
-			CS(INT16,  strtol10,  t64.i, t16.i, t16.u, le16toh, 2);
-			CS(UINT16, strtoul10, t64.u, t16.u, t16.u, le16toh, 2);
-			CS(INT32,  strtol10,  t64.i, t32.i, t32.u, le32toh, 4);
-			CS(UINT32, strtoul10, t64.u, t32.u, t32.u, le32toh, 4);
-			CS(INT64,  strtol10,  t64.i, t64.i, t64.u, le64toh, 8);
-			CS(UINT64, strtoul10, t64.u, t64.u, t64.u, le64toh, 8);
+			CS(INT8,   strtoll10,  t64.i, t8.i,  t8.u,         , 1);
+			CS(UINT8,  strtoull10, t64.u, t8.u,  t8.u,         , 1);
+			CS(INT16,  strtoll10,  t64.i, t16.i, t16.u, le16toh, 2);
+			CS(UINT16, strtoull10, t64.u, t16.u, t16.u, le16toh, 2);
+			CS(INT32,  strtoll10,  t64.i, t32.i, t32.u, le32toh, 4);
+			CS(UINT32, strtoull10, t64.u, t32.u, t32.u, le32toh, 4);
+			CS(INT64,  strtoll10,  t64.i, t64.i, t64.u, le64toh, 8);
+			CS(UINT64, strtoull10, t64.u, t64.u, t64.u, le64toh, 8);
 			CS(FLOAT32, strtod,   t64.d, t32.f, t32.u, le32toh, 4);
 			CS(FLOAT64, strtod,   t64.d, t64.d, t64.u, le64toh, 8);
 #undef CS
@@ -397,7 +397,8 @@ static PyObject *Rocket_append_string(Rocket *self, PyObject *args)
 	/* Build return value and return */
 	offset = buf - data;
 	PyObject *o;
-	o = Py_BuildValue("(iili)", written, offset, last_timestamp, linenum);
+	o = Py_BuildValue("(iiLi)", written, offset,
+                          (long long)last_timestamp, linenum);
 	return o;
 err:
 	PyErr_SetFromErrno(PyExc_OSError);
@@ -431,14 +432,18 @@ static PyObject *Rocket_append_binary(Rocket *self, PyObject *args)
         int data_len;
         int linenum;
 	int offset;
+        long long ll1, ll2, ll3;
 	timestamp_t start;
 	timestamp_t end;
 	timestamp_t last_timestamp;
 
-	if (!PyArg_ParseTuple(args, "it#iilll:append_binary",
+	if (!PyArg_ParseTuple(args, "it#iiLLL:append_binary",
                               &count, &data, &data_len, &offset,
-                              &linenum, &start, &end, &last_timestamp))
+                              &linenum, &ll1, &ll2, &ll3))
 		return NULL;
+        start = ll1;
+        end = ll2;
+        last_timestamp = ll3;
 
         /* Advance to offset */
         if (offset > data_len)
@@ -476,8 +481,8 @@ static PyObject *Rocket_append_binary(Rocket *self, PyObject *args)
 
 	/* Build return value and return */
 	PyObject *o;
-	o = Py_BuildValue("(iili)", rows, offset + rows * self->binary_size,
-                          last_timestamp, linenum);
+	o = Py_BuildValue("(iiLi)", rows, offset + rows * self->binary_size,
+                          (long long)last_timestamp, linenum);
 	return o;
 }
 
@@ -534,7 +539,7 @@ static PyObject *Rocket_extract_string(Rocket *self, PyObject *args)
 		if (fread(&t64.u, 8, 1, self->file) != 1)
 			goto err;
 		t64.u = le64toh(t64.u);
-		ret = sprintf(&str[len], "%ld", t64.i);
+		ret = sprintf(&str[len], "%" PRId64, t64.i);
 		if (ret <= 0)
 			goto err;
 		len += ret;
@@ -556,14 +561,14 @@ static PyObject *Rocket_extract_string(Rocket *self, PyObject *args)
 				len += ret;				\
 			}						\
 			break
-			CASE(INT8,   "%hhd",   t8.i,  t8.u,         , 1);
-			CASE(UINT8,  "%hhu",   t8.u,  t8.u,         , 1);
-			CASE(INT16,  "%hd",    t16.i, t16.u, le16toh, 2);
-			CASE(UINT16, "%hu",    t16.u, t16.u, le16toh, 2);
-			CASE(INT32,  "%d",     t32.i, t32.u, le32toh, 4);
-			CASE(UINT32, "%u",     t32.u, t32.u, le32toh, 4);
-			CASE(INT64,  "%ld",    t64.i, t64.u, le64toh, 8);
-			CASE(UINT64, "%lu",    t64.u, t64.u, le64toh, 8);
+			CASE(INT8,   "%" PRId8,  t8.i,  t8.u,         , 1);
+			CASE(UINT8,  "%" PRIu8,  t8.u,  t8.u,         , 1);
+			CASE(INT16,  "%" PRId16, t16.i, t16.u, le16toh, 2);
+			CASE(UINT16, "%" PRIu16, t16.u, t16.u, le16toh, 2);
+			CASE(INT32,  "%" PRId32, t32.i, t32.u, le32toh, 4);
+			CASE(UINT32, "%" PRIu32, t32.u, t32.u, le32toh, 4);
+			CASE(INT64,  "%" PRId64, t64.i, t64.u, le64toh, 8);
+			CASE(UINT64, "%" PRIu64, t64.u, t64.u, le64toh, 8);
 			/* These next two are a bit debatable.  floats
 			   are 6-9 significant figures, so we print 7.
 			   Doubles are 15-19, so we print 17.  This is
@@ -653,7 +658,7 @@ static PyObject *Rocket_extract_timestamp(Rocket *self, PyObject *args)
 
 	/* Convert and return */
 	t64.u = le64toh(t64.u);
-	return Py_BuildValue("l", t64.i);
+	return Py_BuildValue("L", (long long)t64.i);
 }
 
 /****

nilmdb/utils/__init__.py

@@ -14,3 +14,4 @@ import nilmdb.utils.iterator
 import nilmdb.utils.interval
 import nilmdb.utils.lock
 import nilmdb.utils.sort
+import nilmdb.utils.unicode

nilmdb/utils/unicode.py

@@ -0,0 +1,22 @@
+def encode(u):
+    """Try to encode something from Unicode to a string using the
+    default encoding.  If it fails, try encoding as UTF-8."""
+    if not isinstance(u, unicode):
+        return u
+    try:
+        return u.encode()
+    except UnicodeEncodeError:
+        return u.encode("utf-8")
+
+def decode(s):
+    """Try to decode someting from string to Unicode using the
+    default encoding.  If it fails, try decoding as UTF-8."""
+    if isinstance(s, unicode):
+        return s
+    try:
+        return s.decode()
+    except UnicodeDecodeError:
+        try:
+            return s.decode("utf-8")
+        except UnicodeDecodeError:
+            return s # best we can do

tests/test_cmdline.py

@@ -88,7 +88,7 @@ class TestCmdline(object):
                 sys.exit(0)
             except SystemExit as e:
                 exitcode = e.code
-        captured = outfile.getvalue()
+        captured = nilmdb.utils.unicode.decode(outfile.getvalue())
         self.captured = captured
         self.exitcode = exitcode