[Mumps2Py:] [38] reworked tokprepass. py so that I could fix variable names and initialization in addition to indentlevels .

[subversion@xxxxxxxxxxxxx]

Revision: 38
Author:   pgallot
Date:     2008-01-21 19:51:52 +0000 (Mon, 21 Jan 2008)

Log Message:
-----------
reworked tokprepass.py so that I could fix variable names and initialization in addition to indentlevels.

Modified Paths:
--------------
    trunk/mumps2py/tokprepass.py


Modified: trunk/mumps2py/tokprepass.py
===================================================================
--- trunk/mumps2py/tokprepass.py	2008-01-21 19:48:09 UTC (rev 37)
+++ trunk/mumps2py/tokprepass.py	2008-01-21 19:51:52 UTC (rev 38)
@@ -19,31 +19,68 @@
 tokprepass contains code for massaging the intermediate tokens before
 translating a module into Python.
 """
+import re
 
 from mumps_module import  ModuleInfo
 from tokens import *
 
-def fix_indent_levels(a_mumps_module):
+def fix_indent_levels(token, state):
     """fixing the indent levels for FOR,IF, ELSE commands"""
-    new_indentlevel = 0
-    next_indentlevel = 0
-    for token in a_mumps_module.TokenList:
-        if token.lineno: # starting a new line...
-            new_indentlevel = token.indentlevel
-            next_indentlevel = token.indentlevel
-        if token.toktype in (FORCMD, IFCMD, ELSECMD):
-            # the scope of these commands is the remainder of the line
-            # so the remainder of the line is a nested block of code.
-            # that's the theory anyway.
-            next_indentlevel = new_indentlevel + 1
+    if token.lineno: # starting a new line...
+        state['new'] = token.indentlevel
+        state['next'] = token.indentlevel
+    if token.toktype in (FORCMD, IFCMD, ELSECMD):
+        # the scope of these commands is the remainder of the line
+        # so the remainder of the line is a nested block of code.
+        # that's the theory anyway.
+        state['next'] = state['new'] + 1
 
-        token.indentlevel = new_indentlevel
-        new_indentlevel = next_indentlevel
-##        print "%s: current indent level: %d, next command indent level: %d" % \
-##              (token.dscr(),token.indentlevel, new_indentlevel)
+    token.indentlevel = state['new']
+    state['new'] = state['next']
+##  print "%s: current indent level: %d, next command indent level: %d" % \
+##        (token.dscr(),token.indentlevel, new_indentlevel)
 
+def findall_localvars(token, state):
+    """ find all subtoken LOCALVAR instances """
+    token.extract_subtokens(state['collection'], (LOCALVAR, ))
+    
 
 def prepass(a_mumps_module):
     """ Calls various functions to massage the tokens within a module."""
+    indentlevel_state = {'new':0, 'next':0}
+    localvars_state = {'collection':{} }
+    for token in a_mumps_module.TokenList:
+        fix_indent_levels(token, indentlevel_state)
+        findall_localvars(token, localvars_state)
 
-    fix_indent_levels(a_mumps_module)
+    # there are two things to be done with variables:
+    # (1)The variable names need to have any '%' characters replaced with
+    # something suitable, wherever that variable is used.
+    # (2) If any use of a variable is with indices, then all uses of that
+    # variable need to use indices.
+    
+    vars_dict = {}
+    for token in localvars_state['collection'][LOCALVAR]:
+        if vars_dict.has_key(token.varname):
+            vars_dict[token.varname].append(token)
+        else:
+            vars_dict[token.varname] = [ token, ]
+    for var in vars_dict.keys():
+        is_indexed = False
+        for token in vars_dict[var]:
+            is_indexed = is_indexed or token.__dict__.has_key('indices')
+        if is_indexed:
+            for token in vars_dict[var]:
+                if not token.__dict__.has_key('indices'):
+                    token.indices = None
+
+        if re.search(r'[%]', var):
+            if len(var) == 1:
+                new_varname = "var"
+            else:
+                new_varname = var[:]
+                re.sub(r'[%]', '_', new_varname)
+                if new_varname[0] == '_':
+                    new_varname = 'var' + new_varname
+            for token in vars_dict[var]:
+                token.varname = new_varname