From cecceb0cdd87482124a73ecf537f3445d68be13e Mon Sep 17 00:00:00 2001
From: Mitch Curtis <mitch.curtis@digia.com>
Date: Tue, 12 Nov 2013 13:44:56 +0100
Subject: [PATCH 162/192] Fully expand entities to ensure deep or widely nested
 ones fail parsing

With 512a1ce0698d370c313bb561bbf078935fa0342e, we failed when parsing
entities whose partially expanded size was greater than 1024
characters. That was not enough, so now we fully expand all entities.

This is a backport of f1053d94f59f053ce4acad9320df14f1fbe4faac.

Change-Id: I41dd6f4525c63e82fd320a22d19248169627f7e0
Reviewed-by: Richard J. Moore <rich@kde.org>
---
 src/xml/sax/qxml.cpp                               | 61 +++++++++++++---------
 .../auto/qxmlsimplereader/tst_qxmlsimplereader.cpp |  2 +-
 2 files changed, 37 insertions(+), 26 deletions(-)

diff --git a/src/xml/sax/qxml.cpp b/src/xml/sax/qxml.cpp
index 3904632..befa801 100644
--- a/src/xml/sax/qxml.cpp
+++ b/src/xml/sax/qxml.cpp
@@ -426,7 +426,9 @@ private:
 
     // The limit to the amount of times the DTD parsing functions can be called
     // for the DTD currently being parsed.
-    int dtdRecursionLimit;
+    static const int dtdRecursionLimit = 2;
+    // The maximum amount of characters an entity value may contain, after expansion.
+    static const int entityCharacterLimit = 1024;
 
     const QString &string();
     void stringClear();
@@ -496,7 +498,7 @@ private:
     void unexpectedEof(ParseFunction where, int state);
     void parseFailed(ParseFunction where, int state);
     void pushParseState(ParseFunction function, int state);
-    bool isPartiallyExpandedEntityValueTooLarge(QString *errorMessage);
+    bool isExpandedEntityValueTooLarge(QString *errorMessage);
 
     Q_DECLARE_PUBLIC(QXmlSimpleReader)
     QXmlSimpleReader *q_ptr;
@@ -2764,7 +2766,6 @@ QXmlSimpleReaderPrivate::QXmlSimpleReaderPrivate(QXmlSimpleReader *reader)
     useNamespacePrefixes = false;
     reportWhitespaceCharData = true;
     reportEntities = false;
-    dtdRecursionLimit = 2;
 }
 
 QXmlSimpleReaderPrivate::~QXmlSimpleReaderPrivate()
@@ -6638,30 +6639,43 @@ bool QXmlSimpleReaderPrivate::parseChoiceSeq()
     return false;
 }
 
-bool QXmlSimpleReaderPrivate::isPartiallyExpandedEntityValueTooLarge(QString *errorMessage)
+bool QXmlSimpleReaderPrivate::isExpandedEntityValueTooLarge(QString *errorMessage)
 {
-    const QString value = string();
-    QMap<QString, int> referencedEntityCounts;
-    foreach (QString entityName, entities.keys()) {
-        for (int i = 0; i < value.size() && i != -1; ) {
-            i = value.indexOf(entityName, i);
-            if (i != -1) {
-                // The entityName we're currently trying to find
-                // was matched in this string; increase our count.
-                ++referencedEntityCounts[entityName];
-                i += entityName.size();
+    QMap<QString, int> literalEntitySizes;
+    // The entity at (QMap<QString,) referenced the entities at (QMap<QString,) (int>) times.
+    QMap<QString, QMap<QString, int> > referencesToOtherEntities;
+    QMap<QString, int> expandedSizes;
+
+    // For every entity, check how many times all entity names were referenced in its value.
+    foreach (QString toSearch, entities.keys()) {
+        // The amount of characters that weren't entity names, but literals, like 'X'.
+        QString leftOvers = entities.value(toSearch);
+        // How many times was entityName referenced by toSearch?
+        foreach (QString entityName, entities.keys()) {
+            for (int i = 0; i < leftOvers.size() && i != -1; ) {
+                i = leftOvers.indexOf(QString::fromLatin1("&%1;").arg(entityName), i);
+                if (i != -1) {
+                    leftOvers.remove(i, entityName.size() + 2);
+                    // The entityName we're currently trying to find was matched in this string; increase our count.
+                    ++referencesToOtherEntities[toSearch][entityName];
+                }
             }
         }
+        literalEntitySizes[toSearch] = leftOvers.size();
     }
 
-    foreach (QString entityName, referencedEntityCounts.keys()) {
-        const int timesReferenced = referencedEntityCounts[entityName];
-        const QString entityValue = entities[entityName];
-        if (entityValue.size() * timesReferenced > 1024) {
+    foreach (QString entity, referencesToOtherEntities.keys()) {
+        expandedSizes[entity] = literalEntitySizes[entity];
+        foreach (QString referenceTo, referencesToOtherEntities.value(entity).keys()) {
+            const int references = referencesToOtherEntities.value(entity).value(referenceTo);
+            // The total size of an entity's value is the expanded size of all of its referenced entities, plus its literal size.
+            expandedSizes[entity] += expandedSizes[referenceTo] * references + literalEntitySizes[referenceTo] * references;
+        }
+
+        if (expandedSizes[entity] > entityCharacterLimit) {
             if (errorMessage) {
-                *errorMessage = QString::fromLatin1("The XML entity \"%1\""
-                    "expands too a string that is too large to process when "
-                    "referencing \"%2\" %3 times.").arg(entityName).arg(entityName).arg(timesReferenced);
+                *errorMessage = QString::fromLatin1("The XML entity \"%1\" expands too a string that is too large to process (%2 characters > %3).");
+                *errorMessage = (*errorMessage).arg(entity).arg(expandedSizes[entity]).arg(entityCharacterLimit);
             }
             return true;
         }
@@ -6764,10 +6778,7 @@ bool QXmlSimpleReaderPrivate::parseEntityDecl()
             case EValue:
                 if ( !entityExist(name())) {
                     QString errorMessage;
-                    if (isPartiallyExpandedEntityValueTooLarge(&errorMessage)) {
-                        // The entity at entityName is entityValue.size() characters
-                        // long in its unexpanded form, and was mentioned timesReferenced times,
-                        // resulting in a string that would be greater than 1024 characters.
+                    if (isExpandedEntityValueTooLarge(&errorMessage)) {
                         reportParseError(errorMessage);
                         return false;
                     }
-- 
1.8.4.2