From: jmarshallnz <jcmarsha@gmail.com>
Date: Thu, 28 Nov 2013 04:31:15 +0000 (-0800)
Subject: Merge pull request #3667 from Karlson2k/pcre_reports_more_01
X-Git-Tag: Gotham_alpha10~25
X-Git-Url: http://code.vuplus.com/gitweb/?a=commitdiff_plain;h=bd01e1e6333c67f0cdcc7c75091598338e07c22e;p=vuplus_xbmc

Merge pull request #3667 from Karlson2k/pcre_reports_more_01

Put more data to log on RegExp UTF-8 errors
---

bd01e1e6333c67f0cdcc7c75091598338e07c22e
diff --cc xbmc/utils/RegExp.cpp
index 7db40df,c3383a5..459deb2
--- a/xbmc/utils/RegExp.cpp
+++ b/xbmc/utils/RegExp.cpp
@@@ -237,13 -239,26 +239,26 @@@ int CRegExp::PrivateRegFind(size_t buff
  
  #ifdef PCRE_ERROR_SHORTUTF8 
      case PCRE_ERROR_SHORTUTF8:
+       {
+         const size_t startPos = (m_subject.length() > fragmentLen) ? CUtf8Utils::RFindValidUtf8Char(m_subject, m_subject.length() - fragmentLen) : 0;
+         if (startPos != std::string::npos)
+           CLog::Log(LOGERROR, "PCRE: Bad UTF-8 character at the end of string. Text before bad character: \"%s\"", m_subject.substr(startPos).c_str());
+         else
+           CLog::Log(LOGERROR, "PCRE: Bad UTF-8 character at the end of string");
+         return -1;
+       }
  #endif
      case PCRE_ERROR_BADUTF8:
-       CLog::Log(LOGERROR, "PCRE: Bad UTF-8 character");
-       return -1;
- 
+       {
+         const size_t startPos = (m_iOvector[0] > fragmentLen) ? CUtf8Utils::RFindValidUtf8Char(m_subject, m_iOvector[0] - fragmentLen) : 0;
+         if (m_iOvector[0] >= 0 && startPos != std::string::npos)
+           CLog::Log(LOGERROR, "PCRE: Bad UTF-8 character, error code: %d, position: %d. Text before bad char: \"%s\"", m_iOvector[0], m_iOvector[1], m_subject.substr(startPos, m_iOvector[0] - startPos + 1).c_str());
+         else
+           CLog::Log(LOGERROR, "PCRE: Bad UTF-8 character, error code: %d, position: %d", m_iOvector[0], m_iOvector[1]);
+         return -1;
+       }
      case PCRE_ERROR_BADUTF8_OFFSET:
 -      CLog::Log(LOGERROR, "PCRE: Offset (%d) is pointing to the middle of UTF-8 character", startoffset);
 +      CLog::Log(LOGERROR, "PCRE: Offset is pointing to the middle of UTF-8 character");
        return -1;
  
      default: