From 512a1ce0698d370c313bb561bbf078935fa0342e Mon Sep 17 00:00:00 2001
From: Mitch Curtis <mitch.curtis@digia.com>
Date: Thu, 7 Nov 2013 09:36:29 +0100
Subject: [PATCH 147/192] Disallow deep or widely nested entity references.

Nested references with a depth of 2 or greater will fail. References
that partially expand to greater than 1024 characters will also fail.

This is a backport of 46a8885ae486e238a39efa5119c2714f328b08e4.

Change-Id: I0c2e1fa13d6ccb5f88641dae2ed3f28bfdeaf609
Reviewed-by: Richard J. Moore <rich@kde.org>
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
---
 src/xml/sax/qxml.cpp                               | 51 +++++++++++++++++++
 .../auto/qxmlsimplereader/tst_qxmlsimplereader.cpp | 58 ++++++++++++++++++++++
 .../xmldocs/1-levels-nested-dtd.xml                | 12 +++++
 .../xmldocs/2-levels-nested-dtd.xml                | 13 +++++
 .../internal-entity-polynomial-attribute.xml       | 13 +++++
 5 files changed, 147 insertions(+)
 create mode 100644 tests/auto/qxmlsimplereader/xmldocs/1-levels-nested-dtd.xml
 create mode 100644 tests/auto/qxmlsimplereader/xmldocs/2-levels-nested-dtd.xml
 create mode 100644 tests/auto/qxmlsimplereader/xmldocs/internal-entity-polynomial-attribute.xml

diff --git a/src/xml/sax/qxml.cpp b/src/xml/sax/qxml.cpp
index a1777c5..3904632 100644
--- a/src/xml/sax/qxml.cpp
+++ b/src/xml/sax/qxml.cpp
@@ -424,6 +424,10 @@ private:
     int     stringValueLen;
     QString emptyStr;
 
+    // The limit to the amount of times the DTD parsing functions can be called
+    // for the DTD currently being parsed.
+    int dtdRecursionLimit;
+
     const QString &string();
     void stringClear();
     void stringAddC(QChar);
@@ -492,6 +496,7 @@ private:
     void unexpectedEof(ParseFunction where, int state);
     void parseFailed(ParseFunction where, int state);
     void pushParseState(ParseFunction function, int state);
+    bool isPartiallyExpandedEntityValueTooLarge(QString *errorMessage);
 
     Q_DECLARE_PUBLIC(QXmlSimpleReader)
     QXmlSimpleReader *q_ptr;
@@ -2759,6 +2764,7 @@ QXmlSimpleReaderPrivate::QXmlSimpleReaderPrivate(QXmlSimpleReader *reader)
     useNamespacePrefixes = false;
     reportWhitespaceCharData = true;
     reportEntities = false;
+    dtdRecursionLimit = 2;
 }
 
 QXmlSimpleReaderPrivate::~QXmlSimpleReaderPrivate()
@@ -5018,6 +5024,11 @@ bool QXmlSimpleReaderPrivate::parseDoctype()
                 }
                 break;
             case Mup:
+                if (dtdRecursionLimit > 0 && parameterEntities.size() > dtdRecursionLimit) {
+                    reportParseError(QString::fromLatin1(
+                        "DTD parsing exceeded recursion limit of %1.").arg(dtdRecursionLimit));
+                    return false;
+                }
                 if (!parseMarkupdecl()) {
                     parseFailed(&QXmlSimpleReaderPrivate::parseDoctype, state);
                     return false;
@@ -6627,6 +6638,37 @@ bool QXmlSimpleReaderPrivate::parseChoiceSeq()
     return false;
 }
 
+bool QXmlSimpleReaderPrivate::isPartiallyExpandedEntityValueTooLarge(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();
+            }
+        }
+    }
+
+    foreach (QString entityName, referencedEntityCounts.keys()) {
+        const int timesReferenced = referencedEntityCounts[entityName];
+        const QString entityValue = entities[entityName];
+        if (entityValue.size() * timesReferenced > 1024) {
+            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);
+            }
+            return true;
+        }
+    }
+    return false;
+}
+
 /*
   Parse a EntityDecl [70].
 
@@ -6721,6 +6763,15 @@ bool QXmlSimpleReaderPrivate::parseEntityDecl()
         switch (state) {
             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.
+                        reportParseError(errorMessage);
+                        return false;
+                    }
+
                     entities.insert(name(), string());
                     if (declHnd) {
                         if (!declHnd->internalEntityDecl(name(), string())) {
-- 
1.8.4.2