extend unusedfields loplugin to find fields that can be private

and apply the results in xmlscript

Change-Id: Ib126f6e1576639abfd171e99d9561be9715ece2f

1 files changed, 46 insertions, 51 deletions

diff --git a/compilerplugins/clang/unusedfields.py b/compilerplugins/clang/unusedfields.py
index 785b7a9a1881..7bf910f62d80 100755
--- a/compilerplugins/clang/unusedfields.py
+++ b/compilerplugins/clang/unusedfields.py
@@ -5,11 +5,13 @@ import re
 import io
 
 definitionSet = set()
+protectedAndPublicDefinitionSet = set() # set of tuple(type, name)
 definitionToSourceLocationMap = dict()
 definitionToTypeMap = dict()
 callSet = set()
 readFromSet = set()
 sourceLocationSet = set()
+touchedFromOutsideSet = set()
 
 # clang does not always use exactly the same numbers in the type-parameter vars it generates
 # so I need to substitute them to ensure we can match correctly.
@@ -17,30 +19,43 @@ normalizeTypeParamsRegex = re.compile(r"type-parameter-\d+-\d+")
 def normalizeTypeParams( line ):
     return normalizeTypeParamsRegex.sub("type-parameter-?-?", line)
 
+def parseFieldInfo( tokens ):
+    if len(tokens) == 3:
+        return (normalizeTypeParams(tokens[1]), tokens[2])
+    else:
+        return (normalizeTypeParams(tokens[1]), "")
+
 # The parsing here is designed to avoid grabbing stuff which is mixed in from gbuild.
 # I have not yet found a way of suppressing the gbuild output.
 with io.open("loplugin.unusedfields.log", "rb", buffering=1024*1024) as txt:
     for line in txt:
         tokens = line.strip().split("\t")
         if tokens[0] == "definition:":
-            fieldInfo = (normalizeTypeParams(tokens[1]), tokens[2])
+            access = tokens[1]
+            fieldInfo = (normalizeTypeParams(tokens[2]), tokens[3])
+            srcLoc = tokens[5]
+            # ignore external source code
+            if (srcLoc.startswith("external/")):
+                continue
+            # ignore build folder
+            if (srcLoc.startswith("workdir/")):
+                continue
             definitionSet.add(fieldInfo)
-            definitionToTypeMap[fieldInfo] = tokens[3]
-            definitionToSourceLocationMap[fieldInfo] = tokens[4]
+            definitionToTypeMap[fieldInfo] = tokens[4]
+            if access == "protected" or access == "public":
+                protectedAndPublicDefinitionSet.add(fieldInfo)
+            definitionToSourceLocationMap[fieldInfo] = tokens[5]
         elif tokens[0] == "touch:":
-            if len(tokens) == 3:
-                callInfo = (normalizeTypeParams(tokens[1]), tokens[2])
-                callSet.add(callInfo)
-            else:
-                callInfo = (normalizeTypeParams(tokens[1]), "")
-                callSet.add(callInfo)
+            callSet.add(parseFieldInfo(tokens))
+        elif tokens[0] == "read:":
+            readFromSet.add(parseFieldInfo(tokens))
         elif tokens[0] == "read:":
-            if len(tokens) == 3:
-                readInfo = (normalizeTypeParams(tokens[1]), tokens[2])
-                readFromSet.add(readInfo)
-            else:
-                readInfo = (normalizeTypeParams(tokens[1]), "")
-                readFromSet.add(readInfo)
+            readFromSet.add(parseFieldInfo(tokens))
+        elif tokens[0] == "outside:":
+            touchedFromOutsideSet.add(parseFieldInfo(tokens))
+        else:
+            print( "unknown line: " + line)
+
 # Invert the definitionToSourceLocationMap
 # If we see more than one method at the same sourceLocation, it's being autogenerated as part of a template
 # and we should just ignore
@@ -58,24 +73,6 @@ for d in definitionSet:
     if d in callSet:
         continue
     srcLoc = definitionToSourceLocationMap[d];
-    # ignore external source code
-    if (srcLoc.startswith("external/")):
-        continue
-    # ignore build folder
-    if (srcLoc.startswith("workdir/")):
-        continue
-    # ignore our stable/URE/UNO api
-    if (srcLoc.startswith("include/com/")
-        or srcLoc.startswith("include/cppu/")
-        or srcLoc.startswith("include/cppuhelper/")
-        or srcLoc.startswith("include/osl/")
-        or srcLoc.startswith("include/rtl/")
-        or srcLoc.startswith("include/sal/")
-        or srcLoc.startswith("include/salhelper/")
-        or srcLoc.startswith("include/systools/")
-        or srcLoc.startswith("include/typelib/")
-        or srcLoc.startswith("include/uno/")):
-        continue
     # this is all representations of on-disk data structures
     if (srcLoc.startswith("sc/source/filter/inc/scflt.hxx")
         or srcLoc.startswith("sw/source/filter/ww8/")
@@ -108,24 +105,6 @@ for d in definitionSet:
     if d in readFromSet:
         continue
     srcLoc = definitionToSourceLocationMap[d];
-    # ignore external source code
-    if (srcLoc.startswith("external/")):
-        continue
-    # ignore build folder
-    if (srcLoc.startswith("workdir/")):
-        continue
-    # ignore our stable/URE/UNO api
-    if (srcLoc.startswith("include/com/")
-        or srcLoc.startswith("include/cppu/")
-        or srcLoc.startswith("include/cppuhelper/")
-        or srcLoc.startswith("include/osl/")
-        or srcLoc.startswith("include/rtl/")
-        or srcLoc.startswith("include/sal/")
-        or srcLoc.startswith("include/salhelper/")
-        or srcLoc.startswith("include/systools/")
-        or srcLoc.startswith("include/typelib/")
-        or srcLoc.startswith("include/uno/")):
-        continue
     # this is all representations of on-disk data structures
     if (srcLoc.startswith("sc/source/filter/inc/scflt.hxx")
         or srcLoc.startswith("sw/source/filter/ww8/")
@@ -140,6 +119,17 @@ for d in definitionSet:
 
     writeonlySet.add((clazz + " " + definitionToTypeMap[d], srcLoc))
 
+
+canBePrivateSet = set()
+for d in protectedAndPublicDefinitionSet:
+    clazz = d[0] + " " + d[1]
+    if d in touchedFromOutsideSet:
+        continue
+    srcLoc = definitionToSourceLocationMap[d];
+
+    canBePrivateSet.add((clazz + " " + definitionToTypeMap[d], srcLoc))
+
+
 # sort the results using a "natural order" so sequences like [item1,item2,item10] sort nicely
 def natural_sort_key(s, _nsre=re.compile('([0-9]+)')):
     return [int(text) if text.isdigit() else text.lower()
@@ -148,6 +138,7 @@ def natural_sort_key(s, _nsre=re.compile('([0-9]+)')):
 # sort results by name and line number
 tmp1list = sorted(untouchedSet, key=lambda v: natural_sort_key(v[1]))
 tmp2list = sorted(writeonlySet, key=lambda v: natural_sort_key(v[1]))
+tmp3list = sorted(canBePrivateSet, key=lambda v: natural_sort_key(v[1]))
 
 # print out the results
 with open("loplugin.unusedfields.report-untouched", "wt") as f:
@@ -158,5 +149,9 @@ with open("loplugin.unusedfields.report-writeonly", "wt") as f:
     for t in tmp2list:
         f.write( t[1] + "\n" )
         f.write( "    " + t[0] + "\n" )
+with open("loplugin.unusedfields.report-can-be-private", "wt") as f:
+    for t in tmp3list:
+        f.write( t[1] + "\n" )
+        f.write( "    " + t[0] + "\n" )