- migrating to unicode

git-svn-id: http://galileo.dmi.unict.it/svn/relational/trunk@329 014f5005-505e-4b48-8d0a-63407b615a7c

CHANGELOG

@@ -7,6 +7,7 @@
 - Can use both PySide and PyQt
 - Removed buttons for adding and deleting tuples
 - Can edit relations within the GUI
+- Migrating to unicode (instead of strings)
 
 1.0
 - Adds history in the GUI

relational/optimizations.py

@@ -36,6 +36,21 @@ from tokenize import generate_tokens
 
 sel_op=('//=','**=','and','not','in','//','**','<<','>>','==','!=','>=','<=','+=','-=','*=','/=','%=','or','+','-','*','/','&','|','^','~','<','>','%','=','(',')',',','[',']')
 
+PRODUCT=parser.PRODUCT
+DIFFERENCE=parser.DIFFERENCE
+UNION=parser.UNION
+INTERSECTION=parser.INTERSECTION
+DIVISION=parser.DIVISION
+JOIN=parser.JOIN
+JOIN_LEFT=parser.JOIN_LEFT
+JOIN_RIGHT=parser.JOIN_RIGHT
+JOIN_FULL=parser.JOIN_FULL
+PROJECTION=parser.PROJECTION
+SELECTION=parser.SELECTION
+RENAME=parser.RENAME
+ARROW=parser.ARROW
+
+
 def replace_node(replace,replacement):
     '''This function replaces "replace" node with the node "with",
     the father of the node will now point to the with node'''
@@ -86,7 +101,7 @@ def duplicated_select(n):
     in and
     '''
     changes=0
-    if n.name=='σ' and n.child.name=='σ':        
+    if n.name==SELECTION and n.child.name==SELECTION:
         if n.prop != n.child.prop: #Nested but different, joining them
             n.prop = n.prop + " and " + n.child.prop
             
@@ -114,30 +129,30 @@ def futile_union_intersection_subtraction(n):
     changes=0
     
     #Union and intersection of the same thing
-    if n.name in ('ᑌ','ᑎ') and n.left==n.right:
+    if n.name in (UNION,INTERSECTION) and n.left==n.right:
         changes=1
         replace_node(n,n.left)
     
     #selection and union of the same thing
-    elif (n.name == 'ᑌ'):
-        if n.left.name=='σ' and n.left.child==n.right:
+    elif (n.name == UNION):
+        if n.left.name==SELECTION and n.left.child==n.right:
             changes=1
             replace_node(n,n.right)
-        elif n.right.name=='σ' and n.right.child==n.left:
+        elif n.right.name==SELECTION and n.right.child==n.left:
             changes=1
             replace_node(n,n.left)
             
     #selection and intersection of the same thing
-    elif n.name == 'ᑎ':
-        if n.left.name=='σ' and n.left.child==n.right:
+    elif n.name == INTERSECTION:
+        if n.left.name==SELECTION and n.left.child==n.right:
             changes=1
             replace_node(n,n.left)
-        elif n.right.name=='σ' and n.right.child==n.left:
+        elif n.right.name==SELECTION and n.right.child==n.left:
             changes=1
             replace_node(n,n.right)
             
     #Subtraction and selection of the same thing
-    elif (n.name == '-' and (n.right.name=='σ' and n.right.child==n.left)): #Subtraction of two equal things, but one has a selection
+    elif (n.name == DIFFERENCE and (n.right.name==SELECTION and n.right.child==n.left)): #Subtraction of two equal things, but one has a selection
         n.name=n.right.name
         n.kind=n.right.kind
         n.child=n.right.child
@@ -145,10 +160,10 @@ def futile_union_intersection_subtraction(n):
         n.left=n.right=None
     
     #Subtraction of the same thing or with selection on the left child
-    elif (n.name=='-' and ((n.left==n.right) or (n.left.name=='σ' and n.left.child==n.right)) ):#Empty relation
+    elif (n.name==DIFFERENCE and ((n.left==n.right) or (n.left.name==SELECTION and n.left.child==n.right)) ):#Empty relation
         changes=1
         n.kind=parser.UNARY
-        n.name='σ'
+        n.name=SELECTION
         n.prop='False'
         n.child=n.left.get_left_leaf()
         #n.left=n.right=None
@@ -161,8 +176,8 @@ def down_to_unions_subtractions_intersections(n):
     σ i==2 (c) ᑌ σ i==2(d).   
     '''
     changes=0
-    _o=('ᑌ','-','ᑎ')
-    if n.name=='σ' and n.child.name in _o:
+    _o=(UNION,DIFFERENCE,INTERSECTION)
+    if n.name==SELECTION and n.child.name in _o:
         
         left=parser.node()
         left.prop=n.prop
@@ -190,7 +205,7 @@ def duplicated_projection(n):
     them with π i (R)'''
     changes=0
         
-    if n.name=='π' and n.child.name=='π':
+    if n.name==PROJECTION and n.child.name==PROJECTION:
         n.child=n.child.child
         changes+=1
     
@@ -201,13 +216,13 @@ def selection_inside_projection(n):
     converts them into π k(σ j (R))'''    
     changes=0
     
-    if n.name=='σ' and n.child.name=='π':
+    if n.name==SELECTION and n.child.name==PROJECTION:
         changes=1
         temp=n.prop
         n.prop=n.child.prop
         n.child.prop=temp
-        n.name='π'
-        n.child.name='σ'
+        n.name=PROJECTION
+        n.child.name=SELECTION
     
     return changes+recoursive_scan(selection_inside_projection,n)
 
@@ -219,15 +234,15 @@ def swap_union_renames(n):
     Does the same with subtraction and intersection'''
     changes=0
     
-    if n.name in ('-','ᑌ','ᑎ') and n.left.name==n.right.name and n.left.name=='ρ':
+    if n.name in (DIFFERENCE,UNION,INTERSECTION) and n.left.name==n.right.name and n.left.name==RENAME:
         l_vars={}
         for i in n.left.prop.split(','):
-            q=i.split('➡')
+            q=i.split(ARROW)
             l_vars[q[0].strip()]=q[1].strip()
             
         r_vars={}
         for i in n.right.prop.split(','):
-            q=i.split('➡')
+            q=i.split(ARROW)
             r_vars[q[0].strip()]=q[1].strip()
             
         if r_vars==l_vars:
@@ -240,7 +255,7 @@ def swap_union_renames(n):
             q.left=n.left.child
             q.right=n.right.child
             
-            n.name='ρ'
+            n.name=RENAME
             n.kind=parser.UNARY
             n.child=q
             n.prop=n.left.prop
@@ -252,14 +267,14 @@ def futile_renames(n):
     '''This function purges renames like id->id'''
     changes=0
     
-    if n.name=='ρ':
+    if n.name==RENAME:
         #Located two nested renames.
         changes=1
 
         #Creating a dictionary with the attributes
         _vars={}
         for i in n.prop.split(','):
-            q=i.split('➡')
+            q=i.split(ARROW)
             _vars[q[0].strip()]=q[1].strip()
         #Scans dictionary to locate things like "a->b,b->c" and replace them with "a->c"
         for key in list(_vars.keys()):
@@ -290,7 +305,7 @@ def subsequent_renames(n):
     futile_renames(n) 
     changes=0
     
-    if n.name=='ρ' and n.child.name==n.name:
+    if n.name==RENAME and n.child.name==n.name:
         #Located two nested renames.
         changes=1
         #Joining the attribute into one
@@ -300,7 +315,7 @@ def subsequent_renames(n):
         #Creating a dictionary with the attributes
         _vars={}
         for i in n.prop.split(','):
-            q=i.split('➡')
+            q=i.split(ARROW)
             _vars[q[0].strip()]=q[1].strip()
         #Scans dictionary to locate things like "a->b,b->c" and replace them with "a->c"
         for key in list(_vars.keys()):
@@ -338,7 +353,7 @@ def tokenize_select(expression):
     selection. The expression can contain parenthesis.
     It will use a subclass of str with the attribute level, which
     will specify the nesting level of the token into parenthesis.'''
-    g=generate_tokens(StringIO(expression).readline)
+    g=generate_tokens(StringIO(str(expression)).readline)
     l=list(token[1] for token in g)
     
     l.remove('')
@@ -375,13 +390,13 @@ def swap_rename_projection(n):
     '''
     changes=0
     
-    if n.name=='π' and n.child.name=='ρ':
+    if n.name==PROJECTION and n.child.name==RENAME:
         changes=1
         
         #π index,name(ρ id➡index(R))
         _vars={}
         for i in n.child.prop.split(','):
-            q=i.split('➡')
+            q=i.split(ARROW)
             _vars[q[1].strip()]=q[0].strip()
         
         _pr=n.prop.split(',')
@@ -401,7 +416,7 @@ def swap_rename_projection(n):
             n.prop+='%s➡%s,' % (_vars[i],i)
         n.prop=n.prop[:-1]
         
-        n.child.name='π'
+        n.child.name=PROJECTION
         n.child.prop=''
         for i in _pr:
             n.child.prop+=i+','
@@ -416,12 +431,12 @@ def swap_rename_select(n):
     selection, so the operation is still valid.'''
     changes=0
     
-    if n.name=='σ' and n.child.name=='ρ':
+    if n.name==SELECTION and n.child.name==RENAME:
         changes=1
         #Dictionary containing attributes of rename
         _vars={}
         for i in n.child.prop.split(','):
-            q=i.split('➡')
+            q=i.split(ARROW)
             _vars[q[1].strip()]=q[0].strip()
         
         #tokenizes expression in select
@@ -437,8 +452,8 @@ def swap_rename_select(n):
                     _tokens[i]=_vars[_tokens[i].split('.')[0]]+'.'+splitted[1]
         
         #Swapping operators
-        n.name='ρ'
-        n.child.name='σ'
+        n.name=RENAME
+        n.child.name=SELECTION
         
         n.prop=n.child.prop
         n.child.prop=''
@@ -452,16 +467,16 @@ def select_union_intersect_subtract(n):
     and replaces them with σ (i OR q) (a)
     Removing a O(n²) operation like the union'''
     changes=0
-    if n.name in ('ᑌ', 'ᑎ', '-') and n.left.name=='σ' and n.right.name=='σ' and n.left.child==n.right.child:
+    if n.name in (UNION, INTERSECTION, DIFFERENCE) and n.left.name==SELECTION and n.right.name==SELECTION and n.left.child==n.right.child:
         cahnges=1
         
-        d={'ᑌ':'or', 'ᑎ':'and', '-':'and not'}
+        d={UNION:'or', INTERSECTION:'and', DIFFERENCE:'and not'}
         op=d[n.name]
         
         newnode=parser.node()
         
         newnode.prop='((%s) %s (%s))' % (n.left.prop,op,n.right.prop)
-        newnode.name='σ'
+        newnode.name=SELECTION
         newnode.child=n.left.child
         newnode.kind=parser.UNARY
         replace_node(n,newnode)
@@ -474,12 +489,11 @@ def selection_and_product(n,rels):
     i contains attributes belonging to Q and l contains attributes belonging to both'''
     changes=0
     
-    if n.name=='σ' and n.child.name in ('*','ᐅᐊ','ᐅLEFTᐊ','ᐅRIGHTᐊ','ᐅFULLᐊ'):
+    if n.name==SELECTION and n.child.name in (PRODUCT,JOIN,JOIN_LEFT,JOIN_RIGHT,JOIN_FULL):
         l_attr=n.child.left.result_format(rels)
         r_attr=n.child.right.result_format(rels)
         
         tokens=tokenize_select(n.prop)
-        
         groups=[]
         temp=[]
         
@@ -521,7 +535,7 @@ def selection_and_product(n,rels):
         if len(left)>0:
             changes=1
             l_node=parser.node()
-            l_node.name='σ'
+            l_node.name=SELECTION
             l_node.kind=parser.UNARY
             l_node.child=n.child.left
             l_node.prop=''
@@ -539,7 +553,7 @@ def selection_and_product(n,rels):
         if len(right)>0:
             changes=1
             r_node=parser.node()
-            r_node.name='σ'
+            r_node.name=SELECTION
             r_node.prop=''
             r_node.kind=parser.UNARY
             r_node.child=n.child.right

relational/optimizer.py

@@ -33,6 +33,8 @@ import parser
 RELATION=parser.RELATION
 UNARY=parser.UNARY
 BINARY=parser.BINARY
+
+
 b_operators=parser.b_operators
 u_operators=parser.u_operators
 op_functions=parser.op_functions
@@ -53,12 +55,12 @@ def optimize_all(expression,rels,specific=True,general=True,debug=None):
         steps.
     
     Return value: this will return an optimized version of the expression'''
-    if isinstance(expression,str):
+    if isinstance(expression,unicode):
         n=tree(expression) #Gets the tree
     elif isinstance(expression,node):
         n=expression
     else:
-        raise (TypeError("expression must be a string or a node"))
+        raise (TypeError("expression must be a unicode string or a node"))
     
     if isinstance(debug,list):
         dbg=True
@@ -118,7 +120,7 @@ if __name__=="__main__":
     rels["D1"]= relation.relation("/home/salvo/dev/relational/trunk/samples/dates.csv")
     rels["S1"]= relation.relation("/home/salvo/dev/relational/trunk/samples/skillo.csv")
     print rels'''
-    n=tree("π indice,qq,name (ρ age➡qq,id➡indice (P1-P2))")
+    n=tree(u"π indice,qq,name (ρ age➡qq,id➡indice (P1-P2))")
     #n=tree("σ id==3 and indice==2 and name==5 or name<2(P1 * S1)")
     print n
     print n.toPython()

relational/parser.py

@@ -45,10 +45,25 @@
 RELATION=0
 UNARY=1
 BINARY=2
-b_operators=('*','-','ᑌ','ᑎ','÷','ᐅᐊ','ᐅLEFTᐊ','ᐅRIGHTᐊ','ᐅFULLᐊ') # List of binary operators
-u_operators=('π','σ','ρ') # List of unary operators
 
-op_functions={'*':'product','-':'difference','ᑌ':'union','ᑎ':'intersection','÷':'division','ᐅᐊ':'join','ᐅLEFTᐊ':'outer_left','ᐅRIGHTᐊ':'outer_right','ᐅFULLᐊ':'outer','π':'projection','σ':'selection','ρ':'rename'} # Associates operator with python method
+PRODUCT=u'*'
+DIFFERENCE=u'-'
+UNION=u'ᑌ'
+INTERSECTION=u'ᑎ'
+DIVISION=u'÷'
+JOIN=u'ᐅᐊ'
+JOIN_LEFT=u'ᐅLEFTᐊ'
+JOIN_RIGHT=u'ᐅRIGHTᐊ'
+JOIN_FULL=u'ᐅFULLᐊ'
+PROJECTION=u'π'
+SELECTION=u'σ'
+RENAME=u'ρ'
+ARROW=u'➡'
+
+b_operators=(u'*',u'-',u'ᑌ',u'ᑎ',u'÷',u'ᐅᐊ',u'ᐅLEFTᐊ',u'ᐅRIGHTᐊ',u'ᐅFULLᐊ') # List of binary operators
+u_operators=(u'π',u'σ',u'ρ') # List of unary operators
+
+op_functions={u'*':'product',u'-':'difference',u'ᑌ':'union',u'ᑎ':'intersection',u'÷':'division',u'ᐅᐊ':'join',u'ᐅLEFTᐊ':'outer_left',u'ᐅRIGHTᐊ':'outer_right',u'ᐅFULLᐊ':'outer',u'π':'projection',u'σ':'selection',u'ρ':'rename'} # Associates operator with python method
 
 class node (object):
     '''This class is a node of a relational expression. Leaves are relations and internal nodes are operations.
@@ -77,11 +92,11 @@ class node (object):
                 expression=expression[0]
         
         #The list contains only 1 string. Means it is the name of a relation
-        if len(expression)==1 and isinstance(expression[0],str): 
+        if len(expression)==1 and isinstance(expression[0],unicode): 
             self.kind=RELATION
             self.name=expression[0]
             return
-            
+        
         '''Expression from right to left, searching for binary operators
         this means that binary operators have lesser priority than
         unary operators.
@@ -93,7 +108,7 @@ class node (object):
         within sub-lists, they won't be found here, ensuring that they will
         have highest priority.'''
         for i in range(len(expression)-1,-1,-1): 
-            if expression[i] in b_operators: #Binary operator              
+            if expression[i] in b_operators: #Binary operator   
                 self.kind=BINARY
                 self.name=expression[i]
                 self.left=node(expression[:i]) 
@@ -118,9 +133,9 @@ class node (object):
             prop =self.prop
             
             #Converting parameters
-            if self.name=='π':#Projection
+            if self.name==u'π':#Projection
                 prop='\"%s\"' %  prop.replace(' ','').replace(',','\",\"')
-            elif self.name=="ρ": #Rename
+            elif self.name==u"ρ": #Rename
                 prop='{\"%s\"}' % prop.replace(',','\",\"').replace('➡','\":\"').replace(' ','')
             else: #Selection
                 prop='\"%s\"' %  prop
@@ -234,6 +249,9 @@ def tokenize(expression):
     '''This function converts an expression into a list where
     every token of the expression is an item of a list. Expressions into
     parenthesis will be converted into sublists.'''
+    if not isinstance(expression,unicode):
+        raise Exception('expected unicode')
+    
     items=[] #List for the tokens
     
     '''This is a state machine. Initial status is determined by the starting of the
@@ -266,9 +284,9 @@ def tokenize(expression):
             #Removes the entire parentesis and content from the expression
             expression=expression[end+1:].strip()
         
-        elif expression.startswith("σ") or expression.startswith("π") or expression.startswith("ρ"): #Unary 2 bytes
-            items.append(expression[0:2]) #Adding operator in the top of the list
-            expression=expression[2:].strip() #Removing operator from the expression
+        elif expression.startswith(u"σ") or expression.startswith(u"π") or expression.startswith(u"ρ"): #Unary 2 bytes
+            items.append(expression[0:1]) #Adding operator in the top of the list
+            expression=expression[1:].strip() #Removing operator from the expression
             
             if expression.startswith('('): #Expression with parenthesis, so adding what's between open and close without tokenization
                 par=expression.find('(',_find_matching_parenthesis(expression)) 
@@ -281,18 +299,18 @@ def tokenize(expression):
             items.append(expression[0])
             expression=expression[1:].strip() #1 char from the expression
             state=4
-        elif expression.startswith("ᑎ") or expression.startswith("ᑌ"): #Binary short 3 bytes
-            items.append(expression[0:3]) #Adding operator in the top of the list
-            expression=expression[3:].strip() #Removing operator from the expression
+        elif expression.startswith(u"ᑎ") or expression.startswith(u"ᑌ"): #Binary short 3 bytes
+            items.append(expression[0:1]) #Adding operator in the top of the list
+            expression=expression[1:].strip() #Removing operator from the expression
             state=4
-        elif expression.startswith("÷"): #Binary short 2 bytes
-            items.append(expression[0:2]) #Adding operator in the top of the list
-            expression=expression[2:].strip() #Removing operator from the expression
+        elif expression.startswith(u"÷"): #Binary short 2 bytes
+            items.append(expression[0:1]) #Adding operator in the top of the list
+            expression=expression[1:].strip() #Removing operator from the expression
             state=4
-        elif expression.startswith("ᐅ"): #Binary long
-            i=expression.find("ᐊ")
-            items.append(expression[:i+3])
-            expression=expression[i+3:].strip()
+        elif expression.startswith(u"ᐅ"): #Binary long
+            i=expression.find(u"ᐊ")
+            items.append(expression[:i+1])
+            expression=expression[i+1:].strip()
             
             state=4
         else: #Relation (hopefully)
@@ -347,6 +365,10 @@ def parse(expr):
     return tree(expr).toPython()
     
 if __name__=="__main__":
-    while True:
-        e=raw_input("Expression: ")
-        print parse(e)
+    #while True:
+    #    e=raw_input("Expression: ")
+    #    print parse(e)
+    b=u"σ age>1 and skill=='C' (peopleᐅᐊskills)"
+    print b[0]
+    parse(b)
+    

relational_gui/about.py

@@ -136,10 +136,7 @@ class Ui_Dialog(object):
         self.label_3.setText(QtGui.QApplication.translate("Dialog", "Version "+version, None, QtGui.QApplication.UnicodeUTF8))
         self.label_3.setTextInteractionFlags(QtCore.Qt.LinksAccessibleByMouse|QtCore.Qt.TextSelectableByMouse)
         self.groupBox_3.setTitle(QtGui.QApplication.translate("Dialog", "Author", None, QtGui.QApplication.UnicodeUTF8))
-        if os.name=='nt': #Differentiates acknowledgements depending on the system
-            self.label_2.setText(QtGui.QApplication.translate("Dialog", "Salvo \"LtWorf\" Tomaselli &lt;<a href=\"mailto:tiposchi@tiscali.it\">tiposchi@tiscali.it</a>&gt;<br>Emilio Di Prima &lt;emiliodiprima [at] msn [dot] com&gt; (For the windows version)", None, QtGui.QApplication.UnicodeUTF8))
-        else:
-            self.label_2.setText(QtGui.QApplication.translate("Dialog", "Salvo \"LtWorf\" Tomaselli &lt;<a href=\"mailto:tiposchi@tiscali.it\">tiposchi@tiscali.it</a>&gt;", None, QtGui.QApplication.UnicodeUTF8))
+        self.label_2.setText(QtGui.QApplication.translate("Dialog", "Salvo \"LtWorf\" Tomaselli &lt;<a href=\"mailto:tiposchi@tiscali.it\">tiposchi@tiscali.it</a>&gt;<br>Emilio Di Prima &lt;emiliodiprima[at]msn[dot]com&gt; (For the windows version)", None, QtGui.QApplication.UnicodeUTF8))
         self.label_2.setOpenExternalLinks (True)
         self.label_2.setTextInteractionFlags(QtCore.Qt.LinksAccessibleByMouse|QtCore.Qt.TextSelectableByMouse)
         self.groupBox_2.setTitle(QtGui.QApplication.translate("Dialog", "Links", None, QtGui.QApplication.UnicodeUTF8))

relational_gui/compatibility.py

@@ -31,8 +31,8 @@ except:
 def get_py_str(a):
     '''Returns a python string out of a QString'''
     if pyqt:
-        return str(a.toUtf8())
-    return str(a.encode("utf-8")) #Already a python string in PySide
+        return unicode(a.toUtf8(),'utf-8')
+    return unicode(a.encode("utf-8")) #Already a python string in PySide
 
 def set_utf8_text(component,text):
     if not pyqt:

relational_gui/guihandler.py

@@ -81,13 +81,14 @@ class relForm(QtGui.QMainWindow):
         '''Executes the query'''
         
         query=compatibility.get_py_str(self.ui.txtQuery.text())
-        
+        print query.__class__
         res_rel=compatibility.get_py_str(self.ui.txtResult.text())#result relation's name
         
         if not rtypes.is_valid_relation_name(res_rel):
             QtGui.QMessageBox.information(self,QtGui.QApplication.translate("Form", "Error"),QtGui.QApplication.translate("Form", "Wrong name for destination relation."))
             return
         
+        expr=parser.parse(query)#Converting expression to python code
         try:
             #Converting string to utf8 and then from qstring to normal string
             expr=parser.parse(query)#Converting expression to python code
@@ -105,7 +106,7 @@ class relForm(QtGui.QMainWindow):
 
         #Adds to history
         item='%s = %s' % (compatibility.get_py_str(self.ui.txtResult.text()),compatibility.get_py_str(self.ui.txtQuery.text()))
-        item=unicode(item.decode('utf-8'))
+        #item=item.decode('utf-8'))
         compatibility.add_list_item(self.ui.lstHistory,item)
         
         self.qcounter+=1