diff options
author | Michael Stahl <Michael.Stahl@cib.de> | 2019-10-28 14:31:23 +0100 |
---|---|---|
committer | Andras Timar <andras.timar@collabora.com> | 2021-05-05 13:16:04 +0200 |
commit | cf2e6be466017f495218c0c9f47d0e343f9670cb (patch) | |
tree | 488bffb425e2e3f69d41c8fc564fd1d8bafa7ee2 | |
parent | 8abcac4432850a9694f9d2cd89c975bc8dd6d80b (diff) |
svl: HTMLParser: stop inserting control character garbage into Writer
E.g. rhbz433940-1.html contains literal ^G characters that are inserted
as-is into SwTextNodes.
This now triggers assert about CH_TXT_ATR_FIELDSTART in
SwSubFont::GetTextSize_() that was added in
19a559b0ec9b806519c405651d6d2b2e14712b4a.
Change-Id: I6aa7de41a04069e15b40865fd57894dae0fc10db
Reviewed-on: https://gerrit.libreoffice.org/81606
Reviewed-by: Michael Stahl <michael.stahl@cib.de>
Tested-by: Michael Stahl <michael.stahl@cib.de>
(cherry picked from commit 35d248cab1f0d4800f72abb5cb6afb56f40d9083)
-rw-r--r-- | svtools/source/svhtml/parhtml.cxx | 19 |
1 files changed, 15 insertions, 4 deletions
diff --git a/svtools/source/svhtml/parhtml.cxx b/svtools/source/svhtml/parhtml.cxx index be3167d30053..dad59b28a2ab 100644 --- a/svtools/source/svhtml/parhtml.cxx +++ b/svtools/source/svhtml/parhtml.cxx @@ -29,6 +29,7 @@ #include <tools/datetime.hxx> #include <unotools/datetime.hxx> #include <svl/inettype.hxx> +#include <svl/lngmisc.hxx> #include <com/sun/star/beans/PropertyAttribute.hpp> #include <com/sun/star/document/XDocumentProperties.hpp> @@ -454,8 +455,12 @@ HtmlTokenId HTMLParser::ScanText( const sal_Unicode cBreak ) else nNextCh = 0U; - if ( ! rtl::isUnicodeCodePoint( cChar ) ) + if (!rtl::isUnicodeCodePoint(cChar) + || (linguistic::IsControlChar(cChar) + && cChar != '\r' && cChar != '\n' && cChar != '\t')) + { cChar = '?'; + } } else if( rtl::isAsciiAlpha( nNextCh ) ) { @@ -743,8 +748,11 @@ HtmlTokenId HTMLParser::ScanText( const sal_Unicode cBreak ) else { do { + if (!linguistic::IsControlChar(nNextCh)) + { // All remaining characters make their way into the text. - sTmpBuffer.appendUtf32( nNextCh ); + sTmpBuffer.appendUtf32( nNextCh ); + } if( MAX_LEN == sTmpBuffer.getLength() ) { aToken += sTmpBuffer.makeStringAndClear(); @@ -971,8 +979,11 @@ HtmlTokenId HTMLParser::GetNextRawToken() } SAL_FALLTHROUGH; default: - // all remaining characters are appended to the buffer - sTmpBuffer.appendUtf32( nNextCh ); + if (!linguistic::IsControlChar(nNextCh) || nNextCh == '\t') + { + // all remaining characters are appended to the buffer + sTmpBuffer.appendUtf32( nNextCh ); + } break; } |