emelie/tidy-html5
1
0
Fork 0
Code Issues Pull requests Releases Wiki Activity

Issue #483 - Add ReportSurrogateError() service and connect.

Browse source
This commit is contained in:
Geoff McLane 2017-02-11 18:33:45 +01:00
parent 75bc1f06c7
commit 7f73d4f429
4 changed files with 15 additions and 11 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

6
src/language_en.h
View file

@ -336,9 +336,9 @@ static languageDefinition language_en = { whichPluralForm_en, {
{ INVALID_UTF8, 0, "%s invalid UTF-8 bytes (char. code %s)" }, /* Error */ { INVALID_UTF8, 0, "%s invalid UTF-8 bytes (char. code %s)" }, /* Error */
{ INVALID_UTF16, 0, "%s invalid UTF-16 surrogate pair (char. code %s)" }, /* Error */ { INVALID_UTF16, 0, "%s invalid UTF-16 surrogate pair (char. code %s)" }, /* Error */
{ INVALID_NCR, 0, "%s invalid numeric character reference %s" }, /* Error */ { INVALID_NCR, 0, "%s invalid numeric character reference %s" }, /* Error */
{ BAD_SURROGATE_PAIR, 0, "Have out-of-range surrogate pair U+%s:U+%s, replaced with 2 U+FFFD values."}, /* warning */ { BAD_SURROGATE_PAIR, 0, "Have out-of-range surrogate pair U+%04X:U+%04X, replaced with U+FFFD value."}, /* warning */
{ BAD_SURROGATE_TAIL, 0, "Leading (High) surrogate pair U+%s, with no trailing (Low) entity, replaced with U+FFFD." }, /* warning */ { BAD_SURROGATE_TAIL, 0, "Leading (High) surrogate pair U+%04X, with no trailing (Low) entity, replaced with U+FFFD." }, /* warning */
{ BAD_SURROGATE_LEAD, 0, "Trailing (Low) surrogate pair U+%s, with no leading (High) entity, replaced with U+FFFD." }, /* warning */ { BAD_SURROGATE_LEAD, 0, "Trailing (Low) surrogate pair U+%04X, with no leading (High) entity, replaced with U+FFFD." }, /* warning */
/* ReportEntityError */ /* ReportEntityError */
{ MISSING_SEMICOLON, 0, "entity \"%s\" doesn't end in ';'" }, /* Warning in HTML, Error in XML/XHTML */ { MISSING_SEMICOLON, 0, "entity \"%s\" doesn't end in ';'" }, /* Warning in HTML, Error in XML/XHTML */

12
src/lexer.c
View file

@ -1121,8 +1121,7 @@ static SPStatus GetSurrogatePair(TidyDocImpl* doc, Bool isXml, uint *pch)
{ {
status = SP_failed; /* is one of the 32 out-of-range pairs */ status = SP_failed; /* is one of the 32 out-of-range pairs */
*pch = 0xFFFD; /* return substitute character */ *pch = 0xFFFD; /* return substitute character */
/* SP WARNING: - BAD_SURROGATE_PAIR */ TY_(ReportSurrogateError)(doc, BAD_SURROGATE_PAIR, fch, ch); /* SP WARNING: - */
fprintf(stderr, "Warning: Have out-of-range surrogate pair U+%04X:U+%04X, replaced with U+FFFD value.\n", fch, ch);
} }
} }
} }
@ -1291,8 +1290,7 @@ static void ParseEntity( TidyDocImpl* doc, GetTokenMode mode )
status = GetSurrogatePair(doc, isXml, &ch); status = GetSurrogatePair(doc, isXml, &ch);
if (status == SP_error) if (status == SP_error)
{ {
/* SP WARNING: BAD_SURROGATE_TAIL - use substitute character */ TY_(ReportSurrogateError)(doc, BAD_SURROGATE_TAIL, c1, 0); /* SP WARNING: - using substitute character */
fprintf(stderr, "Warning: Leading(High) surrogate pair U+%04X, with no trailing(Low) entity, replaced with U+FFFD.\n", c1);
TY_(UngetChar)('&', doc->docIn); /* otherwise put it back */ TY_(UngetChar)('&', doc->docIn); /* otherwise put it back */
} }
} }
@ -1301,16 +1299,14 @@ static void ParseEntity( TidyDocImpl* doc, GetTokenMode mode )
/* put this non-entity lead char back */ /* put this non-entity lead char back */
TY_(UngetChar)(c1, doc->docIn); TY_(UngetChar)(c1, doc->docIn);
/* Have leading surrogate pair, with no tail */ /* Have leading surrogate pair, with no tail */
/* SP WARNING: BAD_SURROGATE_TAIL - use substitute character */ TY_(ReportSurrogateError)(doc, BAD_SURROGATE_TAIL, ch, 0); /* SP WARNING: - using substitute character */
fprintf(stderr, "Warning: Leading(High) surrogate pair U+%04X, with no trailing(Low) entity, replaced with U+FFFD.\n", ch);
ch = 0xFFFD; ch = 0xFFFD;
} }
} }
else if (!preserveEntities && found && TY_(IsHighSurrogate)(ch)) else if (!preserveEntities && found && TY_(IsHighSurrogate)(ch))
{ {
/* Have trailing surrogate pair, with no lead */ /* Have trailing surrogate pair, with no lead */
/* SP WARNING: - BAD_SURROGATE_LEAD - - use substitute character */ TY_(ReportSurrogateError)(doc, BAD_SURROGATE_LEAD, ch, 0); /* SP WARNING: - using substitute character */
fprintf(stderr, "Warning: Trailing (Low) surrogate pair U+%04X, with no leading (High) entity, replaced with U+FFFD.\n", ch);
ch = 0xFFFD; ch = 0xFFFD;
} }

7
src/message.c
View file

@ -530,6 +530,13 @@ void TY_(ReportEntityError)( TidyDocImpl* doc, uint code, ctmbstr entity,
messageLexer( doc, TidyWarning, code, fmt, entityname ); messageLexer( doc, TidyWarning, code, fmt, entityname );
} }
void TY_(ReportSurrogateError)(TidyDocImpl* doc, uint code, uint c1, uint c2)
{
ctmbstr fmt = tidyLocalizedString(code);
if (fmt)
messageLexer(doc, TidyWarning, code, fmt, c1, c2);
}
void TY_(ReportAttrError)(TidyDocImpl* doc, Node *node, AttVal *av, uint code) void TY_(ReportAttrError)(TidyDocImpl* doc, Node *node, AttVal *av, uint code)
{ {
char const *name = "NULL", *value = "NULL"; char const *name = "NULL", *value = "NULL";

1
src/message.h
View file

@ -46,6 +46,7 @@ void TY_(ReportEncodingError)(TidyDocImpl* doc, uint code, uint c, Bool discarde
void TY_(ReportEntityError)( TidyDocImpl* doc, uint code, ctmbstr entity, int c ); void TY_(ReportEntityError)( TidyDocImpl* doc, uint code, ctmbstr entity, int c );
void TY_(ReportAttrError)( TidyDocImpl* doc, Node* node, AttVal* av, uint code ); void TY_(ReportAttrError)( TidyDocImpl* doc, Node* node, AttVal* av, uint code );
void TY_(ReportMissingAttr)( TidyDocImpl* doc, Node* node, ctmbstr name ); void TY_(ReportMissingAttr)( TidyDocImpl* doc, Node* node, ctmbstr name );
void TY_(ReportSurrogateError)(TidyDocImpl* doc, uint code, uint c1, uint c2);
#if SUPPORT_ACCESSIBILITY_CHECKS #if SUPPORT_ACCESSIBILITY_CHECKS