Merge pull request #613 from htacg/issue_434

Issue 434
This commit is contained in:
Jim Derry 2017-09-23 19:21:20 -04:00 committed by GitHub
commit d4e6f74226
11 changed files with 320 additions and 30 deletions

View file

@ -576,6 +576,12 @@ static void GetOption(TidyDoc tdoc, /**< The tidy document. */
d->vals = "tagX, tagY, ..."; d->vals = "tagX, tagY, ...";
d->def = NULL; d->def = NULL;
break; break;
case TidyPriorityAttributes:
d->type = "Attributes Names";
d->vals = "attributeX, attributeY, ...";
d->def = NULL;
break;
case TidyCharEncoding: case TidyCharEncoding:
case TidyInCharEncoding: case TidyInCharEncoding:

View file

@ -925,7 +925,7 @@ TIDY_EXPORT TidyIterator TIDY_CALL tidyOptGetDocLinksList(TidyDoc tdoc, /**< T
/** Given a valid TidyIterator initiated with tidyOptGetDocLinksList(), returns /** Given a valid TidyIterator initiated with tidyOptGetDocLinksList(), returns
** a TidyOption instance. ** a TidyOption instance.
** @result Returns in instane of TidyOption. ** @result Returns in instance of TidyOption.
*/ */
TIDY_EXPORT TidyOption TIDY_CALL tidyOptGetNextDocLinks(TidyDoc tdoc, /**< The tidy document to query. */ TIDY_EXPORT TidyOption TIDY_CALL tidyOptGetNextDocLinks(TidyDoc tdoc, /**< The tidy document to query. */
TidyIterator* pos /**< The TidyIterator (initiated with tidyOptGetDocLinksList()) token. */ TidyIterator* pos /**< The TidyIterator (initiated with tidyOptGetDocLinksList()) token. */

2
include/tidyenum.h Normal file → Executable file
View file

@ -220,6 +220,7 @@ extern "C" {
FN(MISSING_ENDTAG_FOR) \ FN(MISSING_ENDTAG_FOR) \
FN(MISSING_IMAGEMAP) \ FN(MISSING_IMAGEMAP) \
FN(MISSING_QUOTEMARK) \ FN(MISSING_QUOTEMARK) \
FN(MISSING_QUOTEMARK_OPEN) \
FN(MISSING_SEMICOLON_NCR) \ FN(MISSING_SEMICOLON_NCR) \
FN(MISSING_SEMICOLON) \ FN(MISSING_SEMICOLON) \
FN(MISSING_STARTTAG) \ FN(MISSING_STARTTAG) \
@ -607,6 +608,7 @@ typedef enum
TidyPPrintTabs, /**< Indent using tabs istead of spaces */ TidyPPrintTabs, /**< Indent using tabs istead of spaces */
TidyPreserveEntities, /**< Preserve entities */ TidyPreserveEntities, /**< Preserve entities */
TidyPreTags, /**< Declared pre tags */ TidyPreTags, /**< Declared pre tags */
TidyPriorityAttributes, /**< Attributes to place first in an element */
#if SUPPORT_ASIAN_ENCODINGS #if SUPPORT_ASIAN_ENCODINGS
TidyPunctWrap, /**< consider punctuation and breaking spaces for wrapping */ TidyPunctWrap, /**< consider punctuation and breaking spaces for wrapping */
#else #else

View file

@ -5,7 +5,7 @@ msgstr ""
"Plural-Forms: nplurals=2; plural=n != 1;\n" "Plural-Forms: nplurals=2; plural=n != 1;\n"
"X-Generator: HTML Tidy poconvert.rb\n" "X-Generator: HTML Tidy poconvert.rb\n"
"Project-Id-Version: \n" "Project-Id-Version: \n"
"POT-Creation-Date: 2017-09-19 14:00:03\n" "POT-Creation-Date: 2017-09-23 07:53:22\n"
"Last-Translator: jderry\n" "Last-Translator: jderry\n"
"Language-Team: \n" "Language-Team: \n"
@ -1055,6 +1055,24 @@ msgid ""
"This option is ignored in XML mode. " "This option is ignored in XML mode. "
msgstr "" msgstr ""
#. Important notes for translators:
#. - Use only <code></code>, <var></var>, <em></em>, <strong></strong>, and
#. <br/>.
#. - Entities, tags, attributes, etc., should be enclosed in <code></code>.
#. - Option values should be enclosed in <var></var>.
#. - It's very important that <br/> be self-closing!
#. - The strings "Tidy" and "HTML Tidy" are the program name and must not
#. be translated.
msgctxt "TidyPriorityAttributes"
msgid ""
"This option allows prioritizing the writing of attributes in tidied "
"documents, allowing them to written before the other attributes of an "
"element. For example, you might specify that <var>id</var> and "
"<var>name</var> are written before every other attribute. "
"<br/>"
"This option takes a space or comma separated list of attribute names. "
msgstr ""
#. Important notes for translators: #. Important notes for translators:
#. - Use only <code></code>, <var></var>, <em></em>, <strong></strong>, and #. - Use only <code></code>, <var></var>, <em></em>, <strong></strong>, and
#. <br/>. #. <br/>.
@ -1226,6 +1244,10 @@ msgid ""
"This option specifies that Tidy should sort attributes within an element " "This option specifies that Tidy should sort attributes within an element "
"using the specified sort algorithm. If set to <var>alpha</var>, the " "using the specified sort algorithm. If set to <var>alpha</var>, the "
"algorithm is an ascending alphabetic sort. " "algorithm is an ascending alphabetic sort. "
"<br/>"
"When used while sorting with <code>priority-attributes</code>, any "
"attribute sorting will take place after the priority attributes have "
"been output. "
msgstr "" msgstr ""
#. Important notes for translators: #. Important notes for translators:
@ -2329,6 +2351,11 @@ msgctxt "MISSING_QUOTEMARK"
msgid "%s attribute with missing trailing quote mark" msgid "%s attribute with missing trailing quote mark"
msgstr "" msgstr ""
#, c-format
msgctxt "MISSING_QUOTEMARK_OPEN"
msgid "value for attribute \"%s\" missing quote marks"
msgstr ""
#, c-format #, c-format
msgctxt "MISSING_SEMICOLON_NCR" msgctxt "MISSING_SEMICOLON_NCR"
msgid "numeric character reference \"%s\" doesn't end in ';'" msgid "numeric character reference \"%s\" doesn't end in ';'"

View file

@ -929,6 +929,49 @@ AttVal* TY_(RepairAttrValue)(TidyDocImpl* doc, Node* node, ctmbstr name, ctmbstr
return TY_(AddAttribute)(doc, node, name, value); return TY_(AddAttribute)(doc, node, name, value);
} }
void TY_(FreeAttrPriorityList)( TidyDocImpl* doc )
{
PriorityAttribs *priorities = &(doc->attribs.priorityAttribs);
if ( priorities->list )
{
uint i = 0;
while ( priorities->list[i] != NULL )
{
TidyFree( doc->allocator, priorities->list[i] );
i++;
}
TidyFree( doc->allocator, priorities->list );
}
}
void TY_(DefinePriorityAttribute)(TidyDocImpl* doc, ctmbstr name)
{
enum { capacity = 10 };
PriorityAttribs *priorities = &(doc->attribs.priorityAttribs);
if ( !priorities->list )
{
priorities->list = TidyAlloc(doc->allocator, sizeof(ctmbstr) * capacity );
priorities->list[0] = NULL;
priorities->capacity = capacity;
priorities->count = 0;
}
if ( priorities->count >= priorities->capacity )
{
priorities->capacity = priorities->capacity * 2;
priorities->list = realloc( priorities->list, sizeof(tmbstr) * priorities->capacity + 1 );
}
priorities->list[priorities->count] = TY_(tmbstrdup)( doc->allocator, name);
priorities->count++;
priorities->list[priorities->count] = NULL;
}
static Bool CheckAttrType( TidyDocImpl* doc, static Bool CheckAttrType( TidyDocImpl* doc,
ctmbstr attrname, AttrCheck type ) ctmbstr attrname, AttrCheck type )
{ {
@ -2174,15 +2217,15 @@ void CheckType( TidyDocImpl* doc, Node *node, AttVal *attval)
} }
static static
AttVal *SortAttVal( AttVal* list, TidyAttrSortStrategy strat ); AttVal *SortAttVal( TidyDocImpl* doc, AttVal* list, TidyAttrSortStrategy strat );
void TY_(SortAttributes)(Node* node, TidyAttrSortStrategy strat) void TY_(SortAttributes)(TidyDocImpl* doc, Node* node, TidyAttrSortStrategy strat)
{ {
while (node) while (node)
{ {
node->attributes = SortAttVal( node->attributes, strat ); node->attributes = SortAttVal( doc, node->attributes, strat );
if (node->content) if (node->content)
TY_(SortAttributes)(node->content, strat); TY_(SortAttributes)(doc, node->content, strat);
node = node->next; node = node->next;
} }
} }
@ -2219,25 +2262,88 @@ void TY_(SortAttributes)(Node* node, TidyAttrSortStrategy strat)
* SOFTWARE. * SOFTWARE.
*/ */
typedef int(*ptAttValComparator)(AttVal *one, AttVal *two); typedef int(*ptAttValComparator)(AttVal *one, AttVal *two, ctmbstr *list);
/* Comparison function for TidySortAttrAlpha */ /* Returns the index of the item in the array, or -1 if not in the array */
static static
int AlphaComparator(AttVal *one, AttVal *two) int indexof( ctmbstr item, ctmbstr *list )
{ {
if ( list )
{
uint i = 0;
while ( list[i] != NULL ) {
if ( TY_(tmbstrcasecmp)(item, list[i]) == 0 )
return i;
i++;
}
}
return -1;
}
/* Comparison function for TidySortAttrAlpha. Will also consider items in
the passed in list as higher-priority, and will group them first.
*/
static
int AlphaComparator(AttVal *one, AttVal *two, ctmbstr *list)
{
int oneIndex = indexof( one->attribute, list );
int twoIndex = indexof( two->attribute, list );
/* If both on the list, the lower index has priority. */
if ( oneIndex >= 0 && twoIndex >= 0 )
return oneIndex < twoIndex ? -1 : 1;
/* If A on the list but B not on the list, then A has priority. */
if ( oneIndex >= 0 && twoIndex == -1 )
return -1;
/* If A not on the list but B is on the list, then B has priority. */
if ( oneIndex == -1 && twoIndex >= 0 )
return 1;
/* Otherwise nothing is on the list, so just compare strings. */
return TY_(tmbstrcmp)(one->attribute, two->attribute); return TY_(tmbstrcmp)(one->attribute, two->attribute);
} }
/* Comparison function for prioritizing list items. It doesn't otherwise
sort.
*/
static
int PriorityComparator(AttVal *one, AttVal *two, ctmbstr *list)
{
int oneIndex = indexof( one->attribute, list );
int twoIndex = indexof( two->attribute, list );
/* If both on the list, the lower index has priority. */
if ( oneIndex >= 0 && twoIndex >= 0 )
return oneIndex < twoIndex ? -1 : 1;
/* If A on the list but B not on the list, then A has priority. */
if ( oneIndex >= 0 && twoIndex == -1 )
return -1;
/* If A not on the list but B is on the list, then B has priority. */
if ( oneIndex == -1 && twoIndex >= 0 )
return 1;
/* Otherwise nothing is on the list, so just mark them as the same. */
return 0;
}
/* The "factory method" that returns a pointer to the comparator function */ /* The "factory method" that returns a pointer to the comparator function */
static static
ptAttValComparator GetAttValComparator(TidyAttrSortStrategy strat) ptAttValComparator GetAttValComparator(TidyAttrSortStrategy strat, ctmbstr *list)
{ {
switch (strat) switch (strat)
{ {
case TidySortAttrAlpha: case TidySortAttrAlpha:
return AlphaComparator; return AlphaComparator;
case TidySortAttrNone: case TidySortAttrNone:
if ( list && list[0] )
return PriorityComparator;
break; break;
} }
return 0; return 0;
@ -2245,9 +2351,14 @@ ptAttValComparator GetAttValComparator(TidyAttrSortStrategy strat)
/* The sort routine */ /* The sort routine */
static static
AttVal *SortAttVal( AttVal *list, TidyAttrSortStrategy strat) AttVal *SortAttVal( TidyDocImpl* doc, AttVal *list, TidyAttrSortStrategy strat)
{ {
ptAttValComparator ptComparator = GetAttValComparator(strat); /* Get the list from the passed-in tidyDoc. */
// ctmbstr* priorityList = (ctmbstr*)doc->attribs.priorityAttribs.list;
// ctmbstr priorityList[] = { "id", NULL };
ctmbstr* priorityList = (ctmbstr*)doc->attribs.priorityAttribs.list;
ptAttValComparator ptComparator = GetAttValComparator(strat, priorityList);
AttVal *p, *q, *e, *tail; AttVal *p, *q, *e, *tail;
int insize, nmerges, psize, qsize, i; int insize, nmerges, psize, qsize, i;
@ -2258,6 +2369,10 @@ AttVal *SortAttVal( AttVal *list, TidyAttrSortStrategy strat)
if (!list) if (!list)
return NULL; return NULL;
/* If no comparator, return the list as-is */
if (ptComparator == 0)
return list;
insize = 1; insize = 1;
while (1) { while (1) {
@ -2291,7 +2406,7 @@ AttVal *SortAttVal( AttVal *list, TidyAttrSortStrategy strat)
} else if (qsize == 0 || !q) { } else if (qsize == 0 || !q) {
/* q is empty; e must come from p. */ /* q is empty; e must come from p. */
e = p; p = p->next; psize--; e = p; p = p->next; psize--;
} else if (ptComparator(p,q) <= 0) { } else if (ptComparator(p,q, priorityList) <= 0) {
/* First element of p is lower (or same); /* First element of p is lower (or same);
* e must come from p. */ * e must come from p. */
e = p; p = p->next; psize--; e = p; p = p->next; psize--;

View file

@ -60,6 +60,13 @@ enum
ANCHOR_HASH_SIZE=1021u ANCHOR_HASH_SIZE=1021u
}; };
/* Keeps a list of attributes that are sorted ahead of the others. */
typedef struct _priorityAttribs {
tmbstr* list;
uint count;
uint capacity;
} PriorityAttribs;
struct _TidyAttribImpl struct _TidyAttribImpl
{ {
/* anchor/node lookup */ /* anchor/node lookup */
@ -68,6 +75,9 @@ struct _TidyAttribImpl
/* Declared literal attributes */ /* Declared literal attributes */
Attribute* declared_attr_list; Attribute* declared_attr_list;
/* Prioritized list of attributes to write */
PriorityAttribs priorityAttribs;
#if ATTRIBUTE_HASH_LOOKUP #if ATTRIBUTE_HASH_LOOKUP
AttrHash* hashtab[ATTRIBUTE_HASH_SIZE]; AttrHash* hashtab[ATTRIBUTE_HASH_SIZE];
#endif #endif
@ -93,6 +103,9 @@ AttVal* TY_(AddAttribute)( TidyDocImpl* doc,
AttVal* TY_(RepairAttrValue)(TidyDocImpl* doc, Node* node, ctmbstr name, ctmbstr value); AttVal* TY_(RepairAttrValue)(TidyDocImpl* doc, Node* node, ctmbstr name, ctmbstr value);
/* Add an item to the list of priority attributes to write first. */
void TY_(DefinePriorityAttribute)(TidyDocImpl* doc, ctmbstr name);
Bool TY_(IsUrl)( TidyDocImpl* doc, ctmbstr attrname ); Bool TY_(IsUrl)( TidyDocImpl* doc, ctmbstr attrname );
/* Bool IsBool( TidyDocImpl* doc, ctmbstr attrname ); */ /* Bool IsBool( TidyDocImpl* doc, ctmbstr attrname ); */
@ -132,13 +145,15 @@ void TY_(FreeAnchors)( TidyDocImpl* doc );
void TY_(InitAttrs)( TidyDocImpl* doc ); void TY_(InitAttrs)( TidyDocImpl* doc );
void TY_(FreeAttrTable)( TidyDocImpl* doc ); void TY_(FreeAttrTable)( TidyDocImpl* doc );
void TY_(FreeAttrPriorityList)( TidyDocImpl* doc );
void TY_(AppendToClassAttr)( TidyDocImpl* doc, AttVal *classattr, ctmbstr classname ); void TY_(AppendToClassAttr)( TidyDocImpl* doc, AttVal *classattr, ctmbstr classname );
/* /*
the same attribute name can't be used the same attribute name can't be used
more than once in each element more than once in each element
*/ */
void TY_(RepairDuplicateAttributes)( TidyDocImpl* doc, Node* node, Bool isXml ); void TY_(RepairDuplicateAttributes)( TidyDocImpl* doc, Node* node, Bool isXml );
void TY_(SortAttributes)(Node* node, TidyAttrSortStrategy strat); void TY_(SortAttributes)(TidyDocImpl* doc, Node* node, TidyAttrSortStrategy strat);
Bool TY_(IsBoolAttribute)( AttVal* attval ); Bool TY_(IsBoolAttribute)( AttVal* attval );
Bool TY_(attrIsEvent)( AttVal* attval ); Bool TY_(attrIsEvent)( AttVal* attval );

View file

@ -175,6 +175,9 @@ static void AdjustConfig( TidyDocImpl* doc );
/* parser for integer values */ /* parser for integer values */
static ParseProperty ParseInt; static ParseProperty ParseInt;
/* a space or comma separated list */
static ParseProperty ParseList;
/* a string excluding whitespace */ /* a string excluding whitespace */
static ParseProperty ParseName; static ParseProperty ParseName;
@ -277,6 +280,7 @@ static const TidyOptionImpl option_defs[] =
{ TidyPPrintTabs, PP, "indent-with-tabs", BL, no, ParseTabs, &boolPicks }, /* 20150515 - Issue #108 */ { TidyPPrintTabs, PP, "indent-with-tabs", BL, no, ParseTabs, &boolPicks }, /* 20150515 - Issue #108 */
{ TidyPreserveEntities, MU, "preserve-entities", BL, no, ParsePickList, &boolPicks }, { TidyPreserveEntities, MU, "preserve-entities", BL, no, ParsePickList, &boolPicks },
{ TidyPreTags, MU, "new-pre-tags", ST, 0, ParseTagNames, NULL }, { TidyPreTags, MU, "new-pre-tags", ST, 0, ParseTagNames, NULL },
{ TidyPriorityAttributes, MU, "priority-attributes", ST, 0, ParseList, NULL },
#if SUPPORT_ASIAN_ENCODINGS #if SUPPORT_ASIAN_ENCODINGS
{ TidyPunctWrap, PP, "punctuation-wrap", BL, no, ParsePickList, &boolPicks }, { TidyPunctWrap, PP, "punctuation-wrap", BL, no, ParsePickList, &boolPicks },
#endif #endif
@ -1087,6 +1091,106 @@ void AdjustConfig( TidyDocImpl* doc )
} }
} }
/* Coordinates Config update and Attributes data for priority attributes, as
a service to ParseList().
*/
void TY_(DeclarePriorityAttrib)( TidyDocImpl* doc, TidyOptionId optId, ctmbstr name )
{
ctmbstr prvval = cfgStr( doc, optId );
tmbstr catval = NULL;
ctmbstr theval = name;
if ( prvval )
{
uint len = TY_(tmbstrlen)(name) + TY_(tmbstrlen)(prvval) + 3;
catval = TY_(tmbstrndup)( doc->allocator, prvval, len );
TY_(tmbstrcat)( catval, ", " );
TY_(tmbstrcat)( catval, name );
theval = catval;
}
TY_(DefinePriorityAttribute)( doc, name );
SetOptionValue( doc, optId, theval );
if ( catval )
TidyDocFree( doc, catval );
}
/* a space or comma separated list of attribute names */
Bool ParseList( TidyDocImpl* doc, const TidyOptionImpl* option )
{
TidyConfigImpl* cfg = &doc->config;
tmbchar buf[1024];
uint i = 0, nItems = 0;
uint c = SkipWhite( cfg );
SetOptionValue( doc, option->id, NULL );
do
{
if (c == ' ' || c == '\t' || c == ',')
{
c = AdvanceChar( cfg );
continue;
}
if ( c == '\r' || c == '\n' )
{
uint c2 = AdvanceChar( cfg );
if ( c == '\r' && c2 == '\n' )
c = AdvanceChar( cfg );
else
c = c2;
if ( !TY_(IsWhite)(c) )
{
buf[i] = 0;
TY_(UngetChar)( c, cfg->cfgIn );
TY_(UngetChar)( '\n', cfg->cfgIn );
break;
}
}
while ( i < sizeof(buf)-2 && c != EndOfStream && !TY_(IsWhite)(c) && c != ',' )
{
buf[i++] = (tmbchar) c;
c = AdvanceChar( cfg );
}
buf[i] = '\0';
if (i == 0) /* Skip empty attribute definition. Possible when */
continue; /* there is a trailing space on the line. */
/* add attribute to array */
switch ( option->id )
{
case TidyPriorityAttributes:
TY_(DeclarePriorityAttrib)( doc, option->id, buf );
break;
default:
break;
}
i = 0;
++nItems;
}
while ( c != EndOfStream );
if ( i > 0 )
switch ( option->id )
{
case TidyPriorityAttributes:
TY_(DeclarePriorityAttrib)( doc, option->id, buf );
break;
default:
break;
}
return ( nItems > 0 );
}
/* unsigned integers */ /* unsigned integers */
Bool ParseInt( TidyDocImpl* doc, const TidyOptionImpl* entry ) Bool ParseInt( TidyDocImpl* doc, const TidyOptionImpl* entry )
{ {
@ -1351,9 +1455,9 @@ Bool ParseTagNames( TidyDocImpl* doc, const TidyOptionImpl* option )
return ( nTags > 0 ); return ( nTags > 0 );
} }
/* a string including whitespace */ /* a string including whitespace */
/* munges whitespace sequences */ /* munges whitespace sequences */
Bool ParseString( TidyDocImpl* doc, const TidyOptionImpl* option ) Bool ParseString( TidyDocImpl* doc, const TidyOptionImpl* option )
{ {
TidyConfigImpl* cfg = &doc->config; TidyConfigImpl* cfg = &doc->config;

21
src/language_en.h Normal file → Executable file
View file

@ -1005,6 +1005,22 @@ static languageDefinition language_en = { whichPluralForm_en, {
"<br/>" "<br/>"
"This option is ignored in XML mode. " "This option is ignored in XML mode. "
}, },
{/* Important notes for translators:
- Use only <code></code>, <var></var>, <em></em>, <strong></strong>, and
<br/>.
- Entities, tags, attributes, etc., should be enclosed in <code></code>.
- Option values should be enclosed in <var></var>.
- It's very important that <br/> be self-closing!
- The strings "Tidy" and "HTML Tidy" are the program name and must not
be translated. */
TidyPriorityAttributes, 0,
"This option allows prioritizing the writing of attributes in tidied "
"documents, allowing them to written before the other attributes of an "
"element. For example, you might specify that <var>id</var> and "
"<var>name</var> are written before every other attribute. "
"<br/>"
"This option takes a space or comma separated list of attribute names. "
},
#if SUPPORT_ASIAN_ENCODINGS #if SUPPORT_ASIAN_ENCODINGS
{/* Important notes for translators: {/* Important notes for translators:
- Use only <code></code>, <var></var>, <em></em>, <strong></strong>, and - Use only <code></code>, <var></var>, <em></em>, <strong></strong>, and
@ -1156,6 +1172,10 @@ static languageDefinition language_en = { whichPluralForm_en, {
"This option specifies that Tidy should sort attributes within an element " "This option specifies that Tidy should sort attributes within an element "
"using the specified sort algorithm. If set to <var>alpha</var>, the " "using the specified sort algorithm. If set to <var>alpha</var>, the "
"algorithm is an ascending alphabetic sort. " "algorithm is an ascending alphabetic sort. "
"<br/>"
"When used while sorting with <code>priority-attributes</code>, any "
"attribute sorting will take place after the priority attributes have "
"been output. "
}, },
{/* Important notes for translators: {/* Important notes for translators:
- Use only <code></code>, <var></var>, <em></em>, <strong></strong>, and - Use only <code></code>, <var></var>, <em></em>, <strong></strong>, and
@ -1877,6 +1897,7 @@ static languageDefinition language_en = { whichPluralForm_en, {
{ MISSING_ENDTAG_FOR, 0, "missing </%s>" }, { MISSING_ENDTAG_FOR, 0, "missing </%s>" },
{ MISSING_IMAGEMAP, 0, "%s should use client-side image map" }, { MISSING_IMAGEMAP, 0, "%s should use client-side image map" },
{ MISSING_QUOTEMARK, 0, "%s attribute with missing trailing quote mark" }, { MISSING_QUOTEMARK, 0, "%s attribute with missing trailing quote mark" },
{ MISSING_QUOTEMARK_OPEN, 0, "value for attribute \"%s\" missing quote marks" },
{ MISSING_SEMICOLON_NCR, 0, "numeric character reference \"%s\" doesn't end in ';'" }, { MISSING_SEMICOLON_NCR, 0, "numeric character reference \"%s\" doesn't end in ';'" },
{ MISSING_SEMICOLON, 0, "entity \"%s\" doesn't end in ';'" }, { MISSING_SEMICOLON, 0, "entity \"%s\" doesn't end in ';'" },
{ MISSING_STARTTAG, 0, "missing <%s>" }, { MISSING_STARTTAG, 0, "missing <%s>" },

13
src/lexer.c Normal file → Executable file
View file

@ -3621,8 +3621,9 @@ static Node *ParsePhp( TidyDocImpl* doc )
} }
/* consumes the '>' terminating start tags */ /* consumes the '>' terminating start tags */
/* @TODO: float the errors back to the calling method */
static tmbstr ParseAttribute( TidyDocImpl* doc, Bool *isempty, static tmbstr ParseAttribute( TidyDocImpl* doc, Bool *isempty,
Node **asp, Node **php) Node **asp, Node **php )
{ {
Lexer* lexer = doc->lexer; Lexer* lexer = doc->lexer;
int start, len = 0; int start, len = 0;
@ -3984,8 +3985,6 @@ static tmbstr ParseValue( TidyDocImpl* doc, ctmbstr name,
{ {
uint q = c; uint q = c;
TY_(ReportAttrError)( doc, lexer->token, NULL, UNEXPECTED_QUOTEMARK );
/* handle <input onclick=s("btn1")> and <a title=foo""">...</a> */ /* handle <input onclick=s("btn1")> and <a title=foo""">...</a> */
/* this doesn't handle <a title=foo"/> which browsers treat as */ /* this doesn't handle <a title=foo"/> which browsers treat as */
/* 'foo"/' nor <a title=foo" /> which browser treat as 'foo"' */ /* 'foo"/' nor <a title=foo" /> which browser treat as 'foo"' */
@ -4156,7 +4155,7 @@ static tmbstr ParseValue( TidyDocImpl* doc, ctmbstr name,
value = NULL; value = NULL;
/* note delimiter if given */ /* note delimiter if given */
*pdelim = (delim ? delim : '"'); *pdelim = delim;
return value; return value;
} }
@ -4273,11 +4272,13 @@ static AttVal* ParseAttrs( TidyDocImpl* doc, Bool *isempty )
(cfgBool(doc, TidyXmlTags) && IsValidXMLAttrName(attribute)))) (cfgBool(doc, TidyXmlTags) && IsValidXMLAttrName(attribute))))
{ {
av = TY_(NewAttribute)(doc); av = TY_(NewAttribute)(doc);
av->delim = delim; av->delim = delim ? delim : '"';
av->attribute = attribute; av->attribute = attribute;
av->value = value; av->value = value;
av->dict = TY_(FindAttribute)( doc, av ); av->dict = TY_(FindAttribute)( doc, av );
AddAttrToList( &list, av ); AddAttrToList( &list, av );
if ( !delim && value )
TY_(ReportAttrError)( doc, lexer->token, av, MISSING_QUOTEMARK_OPEN);
} }
else else
{ {

View file

@ -309,6 +309,7 @@ static struct _dispatchTable {
{ MISSING_ENDTAG_FOR, TidyWarning, formatStandard }, { MISSING_ENDTAG_FOR, TidyWarning, formatStandard },
{ MISSING_IMAGEMAP, TidyWarning, formatAttributeReport }, { MISSING_IMAGEMAP, TidyWarning, formatAttributeReport },
{ MISSING_QUOTEMARK, TidyWarning, formatAttributeReport }, { MISSING_QUOTEMARK, TidyWarning, formatAttributeReport },
{ MISSING_QUOTEMARK_OPEN, TidyInfo, formatAttributeReport },
{ MISSING_SEMICOLON_NCR, TidyWarning, formatStandard }, { MISSING_SEMICOLON_NCR, TidyWarning, formatStandard },
{ MISSING_SEMICOLON, TidyWarning, formatStandard }, { MISSING_SEMICOLON, TidyWarning, formatStandard },
{ MISSING_STARTTAG, TidyWarning, formatStandard }, { MISSING_STARTTAG, TidyWarning, formatStandard },
@ -531,6 +532,9 @@ TidyMessageImpl *formatAttributeReport(TidyDocImpl* doc, Node *element, Node *no
switch (code) switch (code)
{ {
case MISSING_QUOTEMARK_OPEN:
return TY_(tidyMessageCreateWithNode)(doc, node, code, level, name );
case BACKSLASH_IN_URI: case BACKSLASH_IN_URI:
case ESCAPED_ILLEGAL_URI: case ESCAPED_ILLEGAL_URI:
case FIXED_BACKSLASH: case FIXED_BACKSLASH:
@ -546,14 +550,12 @@ TidyMessageImpl *formatAttributeReport(TidyDocImpl* doc, Node *element, Node *no
case UNEXPECTED_QUOTEMARK: case UNEXPECTED_QUOTEMARK:
case WHITE_IN_URI: case WHITE_IN_URI:
return TY_(tidyMessageCreateWithNode)(doc, node, code, level, tagdesc ); return TY_(tidyMessageCreateWithNode)(doc, node, code, level, tagdesc );
break;
case ATTRIBUTE_IS_NOT_ALLOWED: case ATTRIBUTE_IS_NOT_ALLOWED:
case JOINING_ATTRIBUTE: case JOINING_ATTRIBUTE:
case MISSING_ATTR_VALUE: case MISSING_ATTR_VALUE:
case PROPRIETARY_ATTRIBUTE: case PROPRIETARY_ATTRIBUTE:
return TY_(tidyMessageCreateWithNode)(doc, node, code, level, tagdesc, name ); return TY_(tidyMessageCreateWithNode)(doc, node, code, level, tagdesc, name );
break;
case ATTRIBUTE_VALUE_REPLACED: case ATTRIBUTE_VALUE_REPLACED:
case BAD_ATTRIBUTE_VALUE: case BAD_ATTRIBUTE_VALUE:
@ -561,30 +563,25 @@ TidyMessageImpl *formatAttributeReport(TidyDocImpl* doc, Node *element, Node *no
case INSERTING_AUTO_ATTRIBUTE: case INSERTING_AUTO_ATTRIBUTE:
case INVALID_ATTRIBUTE: case INVALID_ATTRIBUTE:
return TY_(tidyMessageCreateWithNode)(doc, node, code, level, tagdesc, name, value ); return TY_(tidyMessageCreateWithNode)(doc, node, code, level, tagdesc, name, value );
break;
case MISMATCHED_ATTRIBUTE_ERROR: case MISMATCHED_ATTRIBUTE_ERROR:
case MISMATCHED_ATTRIBUTE_WARN: case MISMATCHED_ATTRIBUTE_WARN:
return TY_(tidyMessageCreateWithNode)(doc, node, code, level, tagdesc, name, HTMLVersion(doc)); return TY_(tidyMessageCreateWithNode)(doc, node, code, level, tagdesc, name, HTMLVersion(doc));
break;
case ANCHOR_NOT_UNIQUE: case ANCHOR_NOT_UNIQUE:
case ATTR_VALUE_NOT_LCASE: case ATTR_VALUE_NOT_LCASE:
case PROPRIETARY_ATTR_VALUE: case PROPRIETARY_ATTR_VALUE:
case XML_ID_SYNTAX: case XML_ID_SYNTAX:
return TY_(tidyMessageCreateWithNode)(doc, node, code, level, tagdesc, value ); return TY_(tidyMessageCreateWithNode)(doc, node, code, level, tagdesc, value );
break;
case REPEATED_ATTRIBUTE: case REPEATED_ATTRIBUTE:
return TY_(tidyMessageCreateWithNode)(doc, node, code, level, tagdesc, value, name ); return TY_(tidyMessageCreateWithNode)(doc, node, code, level, tagdesc, value, name );
break;
case UNEXPECTED_END_OF_FILE_ATTR: case UNEXPECTED_END_OF_FILE_ATTR:
/* on end of file adjust reported position to end of input */ /* on end of file adjust reported position to end of input */
doc->lexer->lines = doc->docIn->curline; doc->lexer->lines = doc->docIn->curline;
doc->lexer->columns = doc->docIn->curcol; doc->lexer->columns = doc->docIn->curcol;
return TY_(tidyMessageCreateWithLexer)(doc, code, level, tagdesc ); return TY_(tidyMessageCreateWithLexer)(doc, code, level, tagdesc );
break;
} }
return NULL; return NULL;
@ -1427,6 +1424,7 @@ static const TidyOptionId TidyNumEntitiesLinks[] = { TidyDoctype, TidyPreser
static const TidyOptionId TidyOutCharEncodingLinks[] = { TidyCharEncoding, TidyUnknownOption }; static const TidyOptionId TidyOutCharEncodingLinks[] = { TidyCharEncoding, TidyUnknownOption };
static const TidyOptionId TidyOutFileLinks[] = { TidyErrFile, TidyUnknownOption }; static const TidyOptionId TidyOutFileLinks[] = { TidyErrFile, TidyUnknownOption };
static const TidyOptionId TidyPreTagsLinks[] = { TidyBlockTags, TidyEmptyTags, TidyInlineTags, TidyUseCustomTags, TidyUnknownOption }; static const TidyOptionId TidyPreTagsLinks[] = { TidyBlockTags, TidyEmptyTags, TidyInlineTags, TidyUseCustomTags, TidyUnknownOption };
static const TidyOptionId TidySortAttributesLinks[] = { TidyPriorityAttributes, TidyUnknownOption };
static const TidyOptionId TidyUseCustomTagsLinks[] = { TidyBlockTags, TidyEmptyTags, TidyInlineTags, TidyPreTags, TidyUnknownOption }; static const TidyOptionId TidyUseCustomTagsLinks[] = { TidyBlockTags, TidyEmptyTags, TidyInlineTags, TidyPreTags, TidyUnknownOption };
static const TidyOptionId TidyWrapAttValsLinks[] = { TidyWrapScriptlets, TidyLiteralAttribs, TidyUnknownOption }; static const TidyOptionId TidyWrapAttValsLinks[] = { TidyWrapScriptlets, TidyLiteralAttribs, TidyUnknownOption };
static const TidyOptionId TidyWrapScriptletsLinks[] = { TidyWrapAttVals, TidyUnknownOption }; static const TidyOptionId TidyWrapScriptletsLinks[] = { TidyWrapAttVals, TidyUnknownOption };
@ -1454,6 +1452,7 @@ static const TidyOptionDoc docs_xrefs[] =
{ TidyOutCharEncoding, TidyOutCharEncodingLinks }, { TidyOutCharEncoding, TidyOutCharEncodingLinks },
{ TidyOutFile, TidyOutFileLinks }, { TidyOutFile, TidyOutFileLinks },
{ TidyPreTags, TidyPreTagsLinks }, { TidyPreTags, TidyPreTagsLinks },
{ TidySortAttributes, TidySortAttributesLinks },
{ TidyUseCustomTags, TidyUseCustomTagsLinks }, { TidyUseCustomTags, TidyUseCustomTagsLinks },
{ TidyWrapAttVals, TidyWrapAttValsLinks }, { TidyWrapAttVals, TidyWrapAttValsLinks },
{ TidyWrapScriptlets, TidyWrapScriptletsLinks }, { TidyWrapScriptlets, TidyWrapScriptletsLinks },

View file

@ -181,6 +181,7 @@ void tidyDocRelease( TidyDocImpl* doc )
TY_(FreeConfig)( doc ); TY_(FreeConfig)( doc );
TY_(FreeAttrTable)( doc ); TY_(FreeAttrTable)( doc );
TY_(FreeAttrPriorityList)( doc );
TY_(FreeTags)( doc ); TY_(FreeTags)( doc );
/*\ /*\
* Issue #186 - Now FreeNode depend on the doctype, so the lexer is needed * Issue #186 - Now FreeNode depend on the doctype, so the lexer is needed
@ -2215,8 +2216,7 @@ int tidyDocSaveStream( TidyDocImpl* doc, StreamOut* out )
else else
TY_(ReplacePreformattedSpaces)(doc, &doc->root); TY_(ReplacePreformattedSpaces)(doc, &doc->root);
if ( sortAttrStrat != TidySortAttrNone ) TY_(SortAttributes)(doc, &doc->root, sortAttrStrat);
TY_(SortAttributes)(&doc->root, sortAttrStrat);
if ( showMarkup && (doc->errors == 0 || forceOutput) ) if ( showMarkup && (doc->errors == 0 || forceOutput) )
{ {