Merge branch 'master' of github.com:pywren/pywren-ibm-cloud

Author: gilv

install_pywren.sh

@@ -1,9 +1,16 @@
 #!/bin/sh
-LAST_RELEASE=`wget -q --no-check-certificate https://api.github.com/repos/pywren/pywren-ibm-cloud/releases/latest -O - | grep '"tag_name":' |  sed -E 's/.*"([^"]+)".*/\1/'`
-echo "Installing PyWren for IBM Cloud Functions - Release $LAST_RELEASE ..."
+if [ -z "$1" ]
+then
+    # If no version provided, get last release
+    RELEASE=`wget -q --no-check-certificate https://api.github.com/repos/pywren/pywren-ibm-cloud/releases/latest -O - | grep '"tag_name":' |  sed -E 's/.*"([^"]+)".*/\1/'`
+else
+    RELEASE=$1
+fi
+
+echo "Installing PyWren for IBM Cloud Functions - Release $RELEASE ..."
 rm -rf pywren-ibm-cloud* > /dev/null
-wget --no-check-certificate 'https://github.com/pywren/pywren-ibm-cloud/archive/'$LAST_RELEASE'.zip' -O pywren-ibm-cloud.zip 2> /dev/null
+wget --no-check-certificate 'https://github.com/pywren/pywren-ibm-cloud/archive/'$RELEASE'.zip' -O pywren-ibm-cloud.zip 2> /dev/null
 unzip  pywren-ibm-cloud.zip > /dev/null
-mv pywren-ibm-cloud-$LAST_RELEASE pywren-ibm-cloud
+mv pywren-ibm-cloud-$RELEASE pywren-ibm-cloud
 cd pywren-ibm-cloud/pywren; pip install -U . > /dev/null
 echo "done!"

pywren/pywren_ibm_cloud/partitioner.py

@@ -228,6 +228,8 @@ def __init__(self, sb, size, byterange):
         self.eof = False
 
     def read(self, n=None):
+        if self.eof:
+            raise EOFError()
         # Data always contain one byte from the previous chunk,
         # so l'ets check if it is a \n or not
         self.first_byte = self.sb.read(self.plusbytes)
@@ -238,24 +240,22 @@ def read(self, n=None):
 
         self.pos += len(retval)
 
-        first_row_end_pos = 0
+        first_row_start_pos = 0
         if self.first_byte != b'\n' and self.plusbytes != 0:
             logger.debug('Discarding first partial row')
             # Previous byte is not \n
             # This means that we have to discard first row because it is cut
-            first_row_end_pos = retval.find(b'\n')
+            first_row_start_pos = retval.find(b'\n')+1
 
         last_row_end_pos = self.pos
         # Find end of the line in threshold
         if self.pos > self.size:
-            buf = io.BytesIO(retval)
-            while not self.eof:
-                buf.readline()
-                if buf.tell() > self.size:
-                    last_row_end_pos = buf.tell()
-                    self.eof = True
-
-        return retval[first_row_end_pos+self.plusbytes:last_row_end_pos]
+            buf = io.BytesIO(retval[self.size:])
+            buf.readline()
+            last_row_end_pos = self.size+buf.tell()
+            self.eof = True
+
+        return retval[first_row_start_pos:last_row_end_pos]
 
     def readline(self):
         if self.eof: