mERGE branch 'next' into issue-456

README/OPTIONS.md

@@ -38,7 +38,7 @@ struct _tidy_option
     TidyOptionType      type;       /* string, int or bool */
     ulong               dflt;       /* default for TidyInteger and TidyBoolean */
     ParseProperty*      parser;     /* parsing method, read-only if NULL */
-    const ctmbstr*      pickList;   /* pick list */
+    PickListItems*      pickList;   /* pick list */
     ctmbstr             pdflt;      /* default for TidyString */
 };
 ~~~
@@ -78,7 +78,7 @@ Care, each of these enumeration strings have been equated to two uppercase lette
 
 The next item is the `default` value for a boolean, tristate or integer. Note tidy set `no=0` and `yes=1` as its own `Bool` enumeration.
 
-There are a number of `parser` for the options. Likewise a number of `pickList`. Find another option similar to your new option and use the same values.
+There are a number of `parser` for the options. Likewise a number of `pickList`. Find another option similar to your new option and use the same values. The `parser` is the function that parses config file or command line text input, and the `picklist` constitutes the canonical values for the option. Some types of values logically don't have picklists, such as strings or pure integers.
 
 Presently no options have the final `default` string, and it is left out of the table. The compiler will add a NULL.
 

console/tidy.c

@@ -473,26 +473,20 @@ static tmbstr get_escaped_name( ctmbstr name )
  */
 
 
-/** Utility to determine if an option is an AutoBool.
+/** Utility to determine if an option has a picklist.
  ** @param topt The option to check.
- ** @result Returns a Bool indicating whether the option is an Autobool or not.
+ ** @result Returns a Bool indicating whether the option has a picklist or not.
  */
-static Bool isAutoBool( TidyOption topt )
+static Bool hasPickList( TidyOption topt )
 {
     TidyIterator pos;
-    ctmbstr def;
+    
-
     if ( tidyOptGetType( topt ) != TidyInteger)
         return no;
-
+    
     pos = tidyOptGetPickList( topt );
-    while ( pos )
+    
-    {
+    return tidyOptGetNextPick( topt, &pos ) != NULL;
-        def = tidyOptGetNextPick( topt, &pos );
-        if (0==strcmp(def,"yes"))
-            return yes;
-    }
-    return no;
 }
 
 /** Returns the configuration category name for the specified configuration
@@ -555,16 +549,6 @@ static void GetOption(TidyDoc tdoc,    /**< The tidy document. */
     /* Handle special cases first. */
     switch ( optId )
     {
-        case TidyDuplicateAttrs:
-        case TidySortAttributes:
-        case TidyNewline:
-        case TidyAccessibilityCheckLevel:
-        case TidyUseCustomTags:
-            d->type = "enum";
-            d->vals = NULL;
-            d->def = tidyOptGetCurrPick( tdoc, optId );
-            break;
-
         case TidyDoctype:
             d->type = "DocType";
             d->vals = NULL;
@@ -576,16 +560,16 @@ static void GetOption(TidyDoc tdoc,    /**< The tidy document. */
             d->def = sdef;
         }
             break;
-
+            
         case TidyInlineTags:
         case TidyBlockTags:
         case TidyEmptyTags:
         case TidyPreTags:
-            d->type = "Tag names";
+            d->type = "Tag Names";
             d->vals = "tagX, tagY, ...";
             d->def = NULL;
             break;
-
+            
         case TidyCharEncoding:
         case TidyInCharEncoding:
         case TidyOutCharEncoding:
@@ -602,15 +586,13 @@ static void GetOption(TidyDoc tdoc,    /**< The tidy document. */
         {
             case TidyBoolean:
                 d->type = "Boolean";
-                d->vals = "y/n, yes/no, t/f, true/false, 1/0";
                 d->def = tidyOptGetCurrPick( tdoc, optId );
                 break;
-
+                
             case TidyInteger:
-                if (isAutoBool(topt))
+                if (hasPickList(topt))
                 {
-                    d->type = "AutoBool";
+                    d->type = "Enum";
-                    d->vals = "auto, y/n, yes/no, t/f, true/false, 1/0";
                     d->def = tidyOptGetCurrPick( tdoc, optId );
                 }
                 else
@@ -621,13 +603,13 @@ static void GetOption(TidyDoc tdoc,    /**< The tidy document. */
                         d->vals = "0 (no wrapping), 1, 2, ...";
                     else
                         d->vals = "0, 1, 2, ...";
-
+                    
                     idef = tidyOptGetInt( tdoc, optId );
                     sprintf(d->tempdefs, "%u", idef);
                     d->def = d->tempdefs;
                 }
                 break;
-
+                
             case TidyString:
                 d->type = "String";
                 d->vals = NULL;

include/tidyenum.h

@@ -726,6 +726,36 @@ typedef enum
     TidyCR          /**< Use Macintosh style: CR */
 } TidyLineEnding;
 
+    
+/** TidyEncodingOptions option values specify the input and/or output encoding.
+ ** @remark This enum's starting value is guaranteed to remain stable.
+ */
+typedef enum
+{
+    TidyEncRaw = 0,
+    TidyEncAscii,
+    TidyEncLatin0,
+    TidyEncLatin1,
+    TidyEncUtf8,
+#ifndef NO_NATIVE_ISO2022_SUPPORT
+    TidyEncIso2022,
+#endif
+    TidyEncMac,
+    TidyEncWin1252,
+    TidyEncIbm858,
+
+#if SUPPORT_UTF16_ENCODINGS
+    TidyEncUtf16le,
+    TidyEncUtf16be,
+    TidyEncUtf16,
+#endif
+
+#if SUPPORT_ASIAN_ENCODINGS
+    TidyEncBig5,
+    TidyEncShiftjis
+#endif
+} TidyEncodingOptions;
+
 
 /** Mode controlling treatment of doctype
  ** @remark This enum's starting value is guaranteed to remain stable.
@@ -758,6 +788,16 @@ typedef enum
     TidySortAttrAlpha      /**< Sort attributes alphabetically */
 } TidyAttrSortStrategy;
 
+/** Mode controlling capitalization of things, such as attributes.
+ ** @remark This enum's starting value is guaranteed to remain stable.
+ */
+typedef enum
+{
+    TidyUppercaseNo = 0,   /**< Don't uppercase. */
+    TidyUppercaseYes,      /**< Do uppercase. */
+    TidyUppercasePreserve  /**< Preserve case. */
+} TidyUppercase;
+
 
 /** @}
  ** @name Document Tree

localize/translations/language_en_gb.po

@@ -5,7 +5,7 @@ msgstr ""
 "Plural-Forms: nplurals=2; plural=n != 1;\n"
 "X-Generator: HTML Tidy poconvert.rb\n"
 "Project-Id-Version: \n"
-"PO-Revision-Date: 2017-03-22 15:54:52\n"
+"PO-Revision-Date: 2017-05-13 21:04:45\n"
 "Last-Translator: jderry\n"
 "Language-Team: \n"
 
@@ -299,32 +299,6 @@ msgctxt "TidyDropEmptyParas"
 msgid "This option specifies if Tidy should discard empty paragraphs. "
 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 "TidyDropFontTags"
-msgid ""
-"Deprecated; <em>do not use</em>. This option is destructive to "
-"<code>&lt;font&gt;</code> tags, and it will be removed from future "
-"versions of Tidy. Use the <code>clean</code> option instead. "
-"<br/>"
-"If you do set this option despite the warning it will perform "
-"as <code>clean</code> except styles will be inline instead of put into "
-"a CSS class. <code>&lt;font&gt;</code> tags will be dropped completely "
-"and their styles will not be preserved. "
-"<br/>"
-"If both <code>clean</code> and this option are enabled, "
-"<code>&lt;font&gt;</code> tags will still be dropped completely, and "
-"other styles will be preserved in a CSS class instead of inline. "
-"<br/>"
-"See <code>clean</code> for more information. "
-msgstr ""
-
 #. Important notes for translators:
 #. - Use only <code></code>, <var></var>, <em></em>, <strong></strong>, and
 #. <br/>.
@@ -557,18 +531,6 @@ msgctxt "TidyHideComments"
 msgid "This option specifies if Tidy should print out comments. "
 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 "TidyHideEndTags"
-msgid "This option is an alias for <code>omit-optional-tags</code>. "
-msgstr ""
-
 #. Important notes for translators:
 #. - Use only <code></code>, <var></var>, <em></em>, <strong></strong>, and
 #. <br/>.
@@ -1301,10 +1263,10 @@ msgstr ""
 msgctxt "TidyStrictTagsAttr"
 msgid ""
 "This options ensures that tags and attributes are applicable for the "
-"version of HTML that Tidy outputs. When set to <var>yes</var> (the "
+"version of HTML that Tidy outputs. When set to <var>yes</var> and the "
-"default) and the output document type is a strict doctype, then Tidy "
+"output document type is a strict doctype, then Tidy will report "
-"will report errors. If the output document type is a loose or "
+"errors. If the output document type is a loose or transitional "
-"transitional doctype, then Tidy will report warnings. "
+"doctype, then Tidy will report warnings. "
 "<br/>"
 "Additionally if <code>drop-proprietary-attributes</code> is enabled, "
 "then not applicable attributes will be dropped, too. "
@@ -1406,6 +1368,18 @@ msgid ""
 "characters."
 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 "TidyWarnPropAttrs"
+msgid "This option specifies if Tidy should warn on proprietary attributes."
+msgstr ""
+
 #. Important notes for translators:
 #. - Use only <code></code>, <var></var>, <em></em>, <strong></strong>, and
 #. <br/>.
@@ -1745,6 +1719,11 @@ msgctxt "FILE_CANT_OPEN"
 msgid "Can't open \"%s\"\n"
 msgstr ""
 
+#, c-format
+msgctxt "FILE_NOT_FILE"
+msgid "\"%s\" is not a file!\n"
+msgstr ""
+
 #, c-format
 msgctxt "LINE_COLUMN_STRING"
 msgid "line %d column %d - "
@@ -1880,16 +1859,6 @@ msgid ""
 "        TD cells that set the axis attribute are also treated as header cells.\n"
 msgstr ""
 
-#. This console output should be limited to 78 characters per line.
-msgctxt "TEXT_WINDOWS_CHARS"
-msgid ""
-"Characters codes for the Microsoft Windows fonts in the range\n"
-"128 - 159 may not be recognized on other platforms. You are\n"
-"instead recommended to use named entities, e.g. &trade; rather\n"
-"than Windows character code 153 (0x2122 in Unicode). Note that\n"
-"as of February 1998 few browsers support the new entities.\n"
-msgstr ""
-
 #. This console output should be limited to 78 characters per line.
 #. - %s represents a string-encoding name which may be localized in your language.
 #, c-format
@@ -2145,6 +2114,11 @@ msgctxt "ID_NAME_MISMATCH"
 msgid "%s id and name attribute value mismatch"
 msgstr ""
 
+#, c-format
+msgctxt "ILLEGAL_URI_CODEPOINT"
+msgid "%s illegal characters found in URI"
+msgstr ""
+
 #, c-format
 msgctxt "ILLEGAL_URI_REFERENCE"
 msgid "%s improperly escaped URI reference"

localize/translations/language_es.po

@@ -5,7 +5,7 @@ msgstr ""
 "Plural-Forms: nplurals=2; plural=n != 1;\n"
 "X-Generator: HTML Tidy poconvert.rb\n"
 "Project-Id-Version: \n"
-"PO-Revision-Date: 2017-03-22 15:54:52\n"
+"PO-Revision-Date: 2017-05-13 21:04:45\n"
 "Last-Translator: jderry\n"
 "Language-Team: \n"
 
@@ -299,32 +299,6 @@ msgctxt "TidyDropEmptyParas"
 msgid "This option specifies if Tidy should discard empty paragraphs. "
 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 "TidyDropFontTags"
-msgid ""
-"Deprecated; <em>do not use</em>. This option is destructive to "
-"<code>&lt;font&gt;</code> tags, and it will be removed from future "
-"versions of Tidy. Use the <code>clean</code> option instead. "
-"<br/>"
-"If you do set this option despite the warning it will perform "
-"as <code>clean</code> except styles will be inline instead of put into "
-"a CSS class. <code>&lt;font&gt;</code> tags will be dropped completely "
-"and their styles will not be preserved. "
-"<br/>"
-"If both <code>clean</code> and this option are enabled, "
-"<code>&lt;font&gt;</code> tags will still be dropped completely, and "
-"other styles will be preserved in a CSS class instead of inline. "
-"<br/>"
-"See <code>clean</code> for more information. "
-msgstr ""
-
 #. Important notes for translators:
 #. - Use only <code></code>, <var></var>, <em></em>, <strong></strong>, and
 #. <br/>.
@@ -557,18 +531,6 @@ msgctxt "TidyHideComments"
 msgid "This option specifies if Tidy should print out comments. "
 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 "TidyHideEndTags"
-msgid "This option is an alias for <code>omit-optional-tags</code>. "
-msgstr ""
-
 #. Important notes for translators:
 #. - Use only <code></code>, <var></var>, <em></em>, <strong></strong>, and
 #. <br/>.
@@ -1282,10 +1244,10 @@ msgstr ""
 msgctxt "TidyStrictTagsAttr"
 msgid ""
 "This options ensures that tags and attributes are applicable for the "
-"version of HTML that Tidy outputs. When set to <var>yes</var> (the "
+"version of HTML that Tidy outputs. When set to <var>yes</var> and the "
-"default) and the output document type is a strict doctype, then Tidy "
+"output document type is a strict doctype, then Tidy will report "
-"will report errors. If the output document type is a loose or "
+"errors. If the output document type is a loose or transitional "
-"transitional doctype, then Tidy will report warnings. "
+"doctype, then Tidy will report warnings. "
 "<br/>"
 "Additionally if <code>drop-proprietary-attributes</code> is enabled, "
 "then not applicable attributes will be dropped, too. "
@@ -1387,6 +1349,18 @@ msgid ""
 "characters."
 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 "TidyWarnPropAttrs"
+msgid "This option specifies if Tidy should warn on proprietary attributes."
+msgstr ""
+
 #. Important notes for translators:
 #. - Use only <code></code>, <var></var>, <em></em>, <strong></strong>, and
 #. <br/>.
@@ -1726,6 +1700,11 @@ msgctxt "FILE_CANT_OPEN"
 msgid "Can't open \"%s\"\n"
 msgstr ""
 
+#, c-format
+msgctxt "FILE_NOT_FILE"
+msgid "\"%s\" is not a file!\n"
+msgstr ""
+
 #, c-format
 msgctxt "LINE_COLUMN_STRING"
 msgid "line %d column %d - "
@@ -1861,16 +1840,6 @@ msgid ""
 "        TD cells that set the axis attribute are also treated as header cells.\n"
 msgstr ""
 
-#. This console output should be limited to 78 characters per line.
-msgctxt "TEXT_WINDOWS_CHARS"
-msgid ""
-"Characters codes for the Microsoft Windows fonts in the range\n"
-"128 - 159 may not be recognized on other platforms. You are\n"
-"instead recommended to use named entities, e.g. &trade; rather\n"
-"than Windows character code 153 (0x2122 in Unicode). Note that\n"
-"as of February 1998 few browsers support the new entities.\n"
-msgstr ""
-
 #. This console output should be limited to 78 characters per line.
 #. - %s represents a string-encoding name which may be localized in your language.
 #, c-format
@@ -2122,6 +2091,11 @@ msgctxt "ID_NAME_MISMATCH"
 msgid "%s id and name attribute value mismatch"
 msgstr ""
 
+#, c-format
+msgctxt "ILLEGAL_URI_CODEPOINT"
+msgid "%s illegal characters found in URI"
+msgstr ""
+
 #, c-format
 msgctxt "ILLEGAL_URI_REFERENCE"
 msgid "%s improperly escaped URI reference"

localize/translations/language_es_mx.po

@@ -5,7 +5,7 @@ msgstr ""
 "Plural-Forms: nplurals=2; plural=n != 1;\n"
 "X-Generator: HTML Tidy poconvert.rb\n"
 "Project-Id-Version: \n"
-"PO-Revision-Date: 2017-03-22 15:54:52\n"
+"PO-Revision-Date: 2017-05-13 21:04:45\n"
 "Last-Translator: jderry\n"
 "Language-Team: \n"
 
@@ -299,32 +299,6 @@ msgctxt "TidyDropEmptyParas"
 msgid "This option specifies if Tidy should discard empty paragraphs. "
 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 "TidyDropFontTags"
-msgid ""
-"Deprecated; <em>do not use</em>. This option is destructive to "
-"<code>&lt;font&gt;</code> tags, and it will be removed from future "
-"versions of Tidy. Use the <code>clean</code> option instead. "
-"<br/>"
-"If you do set this option despite the warning it will perform "
-"as <code>clean</code> except styles will be inline instead of put into "
-"a CSS class. <code>&lt;font&gt;</code> tags will be dropped completely "
-"and their styles will not be preserved. "
-"<br/>"
-"If both <code>clean</code> and this option are enabled, "
-"<code>&lt;font&gt;</code> tags will still be dropped completely, and "
-"other styles will be preserved in a CSS class instead of inline. "
-"<br/>"
-"See <code>clean</code> for more information. "
-msgstr ""
-
 #. Important notes for translators:
 #. - Use only <code></code>, <var></var>, <em></em>, <strong></strong>, and
 #. <br/>.
@@ -557,18 +531,6 @@ msgctxt "TidyHideComments"
 msgid "This option specifies if Tidy should print out comments. "
 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 "TidyHideEndTags"
-msgid "This option is an alias for <code>omit-optional-tags</code>. "
-msgstr ""
-
 #. Important notes for translators:
 #. - Use only <code></code>, <var></var>, <em></em>, <strong></strong>, and
 #. <br/>.
@@ -1277,10 +1239,10 @@ msgstr ""
 msgctxt "TidyStrictTagsAttr"
 msgid ""
 "This options ensures that tags and attributes are applicable for the "
-"version of HTML that Tidy outputs. When set to <var>yes</var> (the "
+"version of HTML that Tidy outputs. When set to <var>yes</var> and the "
-"default) and the output document type is a strict doctype, then Tidy "
+"output document type is a strict doctype, then Tidy will report "
-"will report errors. If the output document type is a loose or "
+"errors. If the output document type is a loose or transitional "
-"transitional doctype, then Tidy will report warnings. "
+"doctype, then Tidy will report warnings. "
 "<br/>"
 "Additionally if <code>drop-proprietary-attributes</code> is enabled, "
 "then not applicable attributes will be dropped, too. "
@@ -1382,6 +1344,18 @@ msgid ""
 "characters."
 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 "TidyWarnPropAttrs"
+msgid "This option specifies if Tidy should warn on proprietary attributes."
+msgstr ""
+
 #. Important notes for translators:
 #. - Use only <code></code>, <var></var>, <em></em>, <strong></strong>, and
 #. <br/>.
@@ -1721,6 +1695,11 @@ msgctxt "FILE_CANT_OPEN"
 msgid "Can't open \"%s\"\n"
 msgstr ""
 
+#, c-format
+msgctxt "FILE_NOT_FILE"
+msgid "\"%s\" is not a file!\n"
+msgstr ""
+
 #, c-format
 msgctxt "LINE_COLUMN_STRING"
 msgid "line %d column %d - "
@@ -1856,16 +1835,6 @@ msgid ""
 "        TD cells that set the axis attribute are also treated as header cells.\n"
 msgstr ""
 
-#. This console output should be limited to 78 characters per line.
-msgctxt "TEXT_WINDOWS_CHARS"
-msgid ""
-"Characters codes for the Microsoft Windows fonts in the range\n"
-"128 - 159 may not be recognized on other platforms. You are\n"
-"instead recommended to use named entities, e.g. &trade; rather\n"
-"than Windows character code 153 (0x2122 in Unicode). Note that\n"
-"as of February 1998 few browsers support the new entities.\n"
-msgstr ""
-
 #. This console output should be limited to 78 characters per line.
 #. - %s represents a string-encoding name which may be localized in your language.
 #, c-format
@@ -2117,6 +2086,11 @@ msgctxt "ID_NAME_MISMATCH"
 msgid "%s id and name attribute value mismatch"
 msgstr ""
 
+#, c-format
+msgctxt "ILLEGAL_URI_CODEPOINT"
+msgid "%s illegal characters found in URI"
+msgstr ""
+
 #, c-format
 msgctxt "ILLEGAL_URI_REFERENCE"
 msgid "%s improperly escaped URI reference"

localize/translations/language_zh_cn.po

@@ -5,7 +5,7 @@ msgstr ""
 "Plural-Forms: nplurals=1; plural=0;\n"
 "X-Generator: HTML Tidy poconvert.rb\n"
 "Project-Id-Version: \n"
-"PO-Revision-Date: 2017-03-22 15:54:52\n"
+"PO-Revision-Date: 2017-05-13 21:04:45\n"
 "Last-Translator: jderry\n"
 "Language-Team: \n"
 
@@ -299,32 +299,6 @@ msgctxt "TidyDropEmptyParas"
 msgid "This option specifies if Tidy should discard empty paragraphs. "
 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 "TidyDropFontTags"
-msgid ""
-"Deprecated; <em>do not use</em>. This option is destructive to "
-"<code>&lt;font&gt;</code> tags, and it will be removed from future "
-"versions of Tidy. Use the <code>clean</code> option instead. "
-"<br/>"
-"If you do set this option despite the warning it will perform "
-"as <code>clean</code> except styles will be inline instead of put into "
-"a CSS class. <code>&lt;font&gt;</code> tags will be dropped completely "
-"and their styles will not be preserved. "
-"<br/>"
-"If both <code>clean</code> and this option are enabled, "
-"<code>&lt;font&gt;</code> tags will still be dropped completely, and "
-"other styles will be preserved in a CSS class instead of inline. "
-"<br/>"
-"See <code>clean</code> for more information. "
-msgstr ""
-
 #. Important notes for translators:
 #. - Use only <code></code>, <var></var>, <em></em>, <strong></strong>, and
 #. <br/>.
@@ -557,18 +531,6 @@ msgctxt "TidyHideComments"
 msgid "This option specifies if Tidy should print out comments. "
 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 "TidyHideEndTags"
-msgid "This option is an alias for <code>omit-optional-tags</code>. "
-msgstr ""
-
 #. Important notes for translators:
 #. - Use only <code></code>, <var></var>, <em></em>, <strong></strong>, and
 #. <br/>.
@@ -1277,10 +1239,10 @@ msgstr ""
 msgctxt "TidyStrictTagsAttr"
 msgid ""
 "This options ensures that tags and attributes are applicable for the "
-"version of HTML that Tidy outputs. When set to <var>yes</var> (the "
+"version of HTML that Tidy outputs. When set to <var>yes</var> and the "
-"default) and the output document type is a strict doctype, then Tidy "
+"output document type is a strict doctype, then Tidy will report "
-"will report errors. If the output document type is a loose or "
+"errors. If the output document type is a loose or transitional "
-"transitional doctype, then Tidy will report warnings. "
+"doctype, then Tidy will report warnings. "
 "<br/>"
 "Additionally if <code>drop-proprietary-attributes</code> is enabled, "
 "then not applicable attributes will be dropped, too. "
@@ -1382,6 +1344,18 @@ msgid ""
 "characters."
 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 "TidyWarnPropAttrs"
+msgid "This option specifies if Tidy should warn on proprietary attributes."
+msgstr ""
+
 #. Important notes for translators:
 #. - Use only <code></code>, <var></var>, <em></em>, <strong></strong>, and
 #. <br/>.
@@ -1721,6 +1695,11 @@ msgctxt "FILE_CANT_OPEN"
 msgid "Can't open \"%s\"\n"
 msgstr "无法打开”%s”\n"
 
+#, c-format
+msgctxt "FILE_NOT_FILE"
+msgid "\"%s\" is not a file!\n"
+msgstr ""
+
 #, c-format
 msgctxt "LINE_COLUMN_STRING"
 msgid "line %d column %d - "
@@ -1854,16 +1833,6 @@ msgid ""
 "        TD cells that set the axis attribute are also treated as header cells.\n"
 msgstr ""
 
-#. This console output should be limited to 78 characters per line.
-msgctxt "TEXT_WINDOWS_CHARS"
-msgid ""
-"Characters codes for the Microsoft Windows fonts in the range\n"
-"128 - 159 may not be recognized on other platforms. You are\n"
-"instead recommended to use named entities, e.g. &trade; rather\n"
-"than Windows character code 153 (0x2122 in Unicode). Note that\n"
-"as of February 1998 few browsers support the new entities.\n"
-msgstr ""
-
 #. This console output should be limited to 78 characters per line.
 #. - %s represents a string-encoding name which may be localized in your language.
 #, c-format
@@ -2111,6 +2080,11 @@ msgctxt "ID_NAME_MISMATCH"
 msgid "%s id and name attribute value mismatch"
 msgstr ""
 
+#, c-format
+msgctxt "ILLEGAL_URI_CODEPOINT"
+msgid "%s illegal characters found in URI"
+msgstr ""
+
 #, c-format
 msgctxt "ILLEGAL_URI_REFERENCE"
 msgid "%s improperly escaped URI reference"

src/attrs.c

@@ -824,15 +824,15 @@ static const Attribute* attrsLookup(TidyDocImpl* doc,
 
 #if ATTRIBUTE_HASH_LOOKUP
     for (p = attribs->hashtab[attrsHash(atnam)]; p && p->attr; p = p->next)
-        if (TY_(tmbstrcmp)(atnam, p->attr->name) == 0)
+        if (TY_(tmbstrcasecmp)(atnam, p->attr->name) == 0)
             return p->attr;
 
     for (np = attribute_defs; np && np->name; ++np)
-        if (TY_(tmbstrcmp)(atnam, np->name) == 0)
+        if (TY_(tmbstrcasecmp)(atnam, np->name) == 0)
             return attrsInstall(doc, attribs, np);
 #else
     for (np = attribute_defs; np && np->name; ++np)
-        if (TY_(tmbstrcmp)(atnam, np->name) == 0)
+        if (TY_(tmbstrcasecmp)(atnam, np->name) == 0)
             return np;
 #endif
 
@@ -1617,7 +1617,7 @@ void TY_(CheckUrl)( TidyDocImpl* doc, Node *node, AttVal *attval)
     {
         if ( cfgBool(doc, TidyFixUri) )
             TY_(ReportAttrError)( doc, node, attval, ESCAPED_ILLEGAL_URI);
-        else
+        else if ( !(TY_(HTMLVersion)(doc) & VERS_HTML5) )
             TY_(ReportAttrError)( doc, node, attval, ILLEGAL_URI_REFERENCE);
 
         doc->badChars |= BC_INVALID_URI;

src/config.c

@@ -1,22 +1,12 @@
-/*
+/* config.c
-  config.c -- read config file and manage config properties
+ * Read configuration files and manage configuration properties.
-  
+ *
-  (c) 1998-2008 (W3C) MIT, ERCIM, Keio University
+ * Copyright (c) 1998-2017 World Wide Web Consortium (Massachusetts
-  See tidy.h for the copyright notice.
+ * Institute of Technology, European Research Consortium for Informatics
-
+ * and Mathematics, Keio University) and HTACG.
-  config files associate a property name with a value.
+ *
-
+ * See tidy.h for the copyright notice.
-  // comments can start at the beginning of a line
+ */
-  # comments can start at the beginning of a line
-  name: short values fit onto one line
-  name: a really long value that
-   continues on the next line
-
-  property names are case insensitive and should be less than
-  60 characters in length and must start at the begining of
-  the line, as whitespace at the start of a line signifies a
-  line continuation.
-*/
 
 #include "config.h"
 #include "tidy-int.h"
@@ -29,8 +19,9 @@
 #else
 #ifdef DMALLOC
 /*
-   macro for valloc() in dmalloc.h may conflict with declaration for valloc() in unistd.h -
+   macro for valloc() in dmalloc.h may conflict with declaration for valloc()
-   we don't need (debugging for) valloc() here. dmalloc.h should come last but it doesn't.
+   in unistd.h - we don't need (debugging for) valloc() here. dmalloc.h should
+  come last but it doesn't.
 */
 #ifdef valloc
 #undef valloc
@@ -56,103 +47,106 @@ void TY_(FreeConfig)( TidyDocImpl* doc )
 }
 
 
-/* Arrange so index can be cast to enum
+/* 
+   Arrange so index can be cast to enum. Note that the value field in the 
+   following structures is not currently used in code; they're present for 
+   documentation purposes currently. The arrays must be populated in enum order.
 */
-static const ctmbstr boolPicks[] = 
+static PickListItems boolPicks = {
-{
+    { "no",  TidyNoState,  { "0", "n", "f", "no",  "false", NULL } },
-  "no",
+    { "yes", TidyYesState, { "1", "y", "t", "yes", "true",  NULL } },
-  "yes",
+    { NULL }
-  NULL
 };
 
-static const ctmbstr autoBoolPicks[] = 
+static PickListItems autoBoolPicks = {
-{
+    { "no",   TidyNoState,  { "0", "n", "f", "no",  "false", NULL } },
-  "no",
+    { "yes",  TidyYesState, { "1", "y", "t", "yes", "true",  NULL } },
-  "yes",
+    { "auto", TidyYesState, { "auto",                        NULL } },
-  "auto",
+    { NULL }
-  NULL
 };
 
-static const ctmbstr repeatAttrPicks[] = 
+static PickListItems repeatAttrPicks = {
-{
+    { "keep-first", TidyNoState,  { "keep-first", NULL } },
-  "keep-first",
+    { "keep-last",  TidyYesState, { "keep-last",  NULL } },
-  "keep-last",
+    { NULL }
-  NULL
 };
 
-static const ctmbstr accessPicks[] = 
+static PickListItems accessPicks = {
-{
+    { "0 (Tidy Classic)",      0, { "0", "0 (Tidy Classic)",      NULL } },
-  "0 (Tidy Classic)",
+    { "1 (Priority 1 Checks)", 1, { "1", "1 (Priority 1 Checks)", NULL } },
-  "1 (Priority 1 Checks)",
+    { "2 (Priority 2 Checks)", 2, { "2", "2 (Priority 2 Checks)", NULL } },
-  "2 (Priority 2 Checks)",
+    { "3 (Priority 3 Checks)", 3, { "3", "3 (Priority 3 Checks)", NULL } },
-  "3 (Priority 3 Checks)",
+    { NULL }
-  NULL
 };
 
-static const ctmbstr charEncPicks[] = 
+static PickListItems charEncPicks = {
-{
+    { "raw",      TidyEncRaw,      { "raw",      NULL } },
-  "raw",
+    { "ascii",    TidyEncAscii,    { "ascii",    NULL } },
-  "ascii",
+    { "latin0",   TidyEncLatin0,   { "latin0",   NULL } },
-  "latin0",
+    { "latin1",   TidyEncLatin1,   { "latin1",   NULL } },
-  "latin1",
+    { "utf8",     TidyEncUtf8,     { "utf8",     NULL } },
-  "utf8",
 #ifndef NO_NATIVE_ISO2022_SUPPORT
-  "iso2022",
+    { "iso2022",  TidyEncIso2022,  { "iso2022",  NULL } },
 #endif
-  "mac",
+    { "mac",      TidyEncMac,      { "mac",      NULL } },
-  "win1252",
+    { "win1252",  TidyEncWin1252,  { "win1252",  NULL } },
-  "ibm858",
+    { "ibm858",   TidyEncIbm858,   { "ibm858",   NULL } },
 
 #if SUPPORT_UTF16_ENCODINGS
-  "utf16le",
+    { "utf16le",  TidyEncUtf16le,  { "utf16le",  NULL } },
-  "utf16be",
+    { "utf16be",  TidyEncUtf16be,  { "utf16be",  NULL } },
-  "utf16",
+    { "utf16",    TidyEncUtf16,    { "utf16",    NULL } },
 #endif
 
 #if SUPPORT_ASIAN_ENCODINGS
-  "big5",
+    { "big5",     TidyEncBig5,     { "big5",     NULL } },
-  "shiftjis",
+    { "shiftjis", TidyEncShiftjis, { "shiftjis", NULL } },
 #endif
 
-  NULL
+    { NULL }
 };
 
-static const ctmbstr newlinePicks[] = 
+static PickListItems newlinePicks = {
-{
+    { "LF",   TidyLF,   { "lf",   NULL } },
-  "LF",
+    { "CRLF", TidyCRLF, { "crlf", NULL } },
-  "CRLF",
+    { "CR",   TidyCR,   { "cr",   NULL } },
-  "CR",
+    { NULL }
-  NULL
 };
 
-static const ctmbstr doctypePicks[] = 
+static PickListItems doctypePicks = {
-{
+    { "html5",        TidyDoctypeHtml5,  { "html5",                 NULL } },
-  "html5",
+    { "omit",         TidyDoctypeOmit,   { "omit",                  NULL } },
-  "omit",
+    { "auto",         TidyDoctypeAuto,   { "auto",                  NULL } },
-  "auto",
+    { "strict",       TidyDoctypeStrict, { "strict",                NULL } },
-  "strict",
+    { "transitional", TidyDoctypeLoose,  { "loose", "transitional", NULL } },
-  "transitional",
+    { "user",         TidyDoctypeUser,   { "user",                  NULL } },
-  "user",
+    { NULL }
-  NULL 
 };
 
-static const ctmbstr sorterPicks[] = 
+static PickListItems sorterPicks = {
-{
+    { "none",  TidySortAttrNone,  { "none",  NULL } },
-  "none",
+    { "alpha", TidySortAttrAlpha, { "alpha", NULL } },
-  "alpha",
+    { NULL }
-  NULL
 };
 
-static const ctmbstr customTagsPicks[] =
+static PickListItems customTagsPicks = {
-{
+    {"no",         TidyCustomNo,         { "no", "n",            NULL } },
-    "no",
+    {"blocklevel", TidyCustomBlocklevel, { "blocklevel",         NULL } },
-    "blocklevel",
+    {"empty",      TidyCustomEmpty,      { "empty",              NULL } },
-    "empty",
+    {"inline",     TidyCustomInline,     { "inline", "y", "yes", NULL } },
-    "inline",
+    {"pre",        TidyCustomPre,        { "pre",                NULL } },
-    "pre",
+    { NULL }
-    NULL
 };
 
+static PickListItems attributeCasePicks = {
+    { "no",       TidyUppercaseNo,       { "0", "n", "f", "no",  "false", NULL } },
+    { "yes",      TidyUppercaseYes,      { "1", "y", "t", "yes", "true",  NULL } },
+    { "preserve", TidyUppercasePreserve, { "preserve",                    NULL } },
+    { NULL }
+};
+
+
+
 #define MU TidyMarkup
 #define DG TidyDiagnostics
 #define PP TidyPrettyPrint
@@ -171,7 +165,7 @@ static const ctmbstr customTagsPicks[] =
 
 /* If Accessibility checks not supported, make config setting read-only */
 #if SUPPORT_ACCESSIBILITY_CHECKS
-#define ParseAcc ParseInt
+#define ParseAcc ParsePickList
 #else
 #define ParseAcc NULL 
 #endif
@@ -181,13 +175,6 @@ static void AdjustConfig( TidyDocImpl* doc );
 /* parser for integer values */
 static ParseProperty ParseInt;
 
-/* parser for 't'/'f', 'true'/'false', 'y'/'n', 'yes'/'no' or '1'/'0' */
-static ParseProperty ParseBool;
-
-/* parser for 't'/'f', 'true'/'false', 'y'/'n', 'yes'/'no', '1'/'0'
-   or 'auto' */
-static ParseProperty ParseAutoBool;
-
 /* a string excluding whitespace */
 static ParseProperty ParseName;
 
@@ -200,108 +187,97 @@ static ParseProperty ParseString;
 /* a space or comma separated list of tag names */
 static ParseProperty ParseTagNames;
 
-/* alpha */
+/* RAW, ASCII, LATIN0, LATIN1, UTF8, ISO2022, MACROMAN,
-static ParseProperty ParseSorter;
-
-/* RAW, ASCII, LATIN0, LATIN1, UTF8, ISO2022, MACROMAN, 
    WIN1252, IBM858, UTF16LE, UTF16BE, UTF16, BIG5, SHIFTJIS
 */
 static ParseProperty ParseCharEnc;
-static ParseProperty ParseNewline;
 
 /* html5 | omit | auto | strict | loose | <fpi> */
 static ParseProperty ParseDocType;
 
-/* keep-first or keep-last? */
+/* 20150515 - support using tabs instead of spaces - Issue #108
-static ParseProperty ParseRepeatAttr;
+ */
-
-/*\
- * 20150515 - support using tabs instead of spaces - Issue #108
- * (a) parser for 't'/'f', 'true'/'false', 'y'/'n', 'yes'/'no' or '1'/'0' 
- * (b) sets the TidyIndentSpaces to 1 if 'yes'
- * (c) sets the indent_char to '\t' or ' '
-\*/
 static ParseProperty ParseTabs;
 
-/* Parse the value of TidyUseCustomTags */
+/* General parser for options having picklists */
-static ParseProperty ParseUseCustomTags;
+static ParseProperty ParsePickList;
 
 
 /* Ensure struct order is same order as tidyenum.h:TidyOptionId! */
 static const TidyOptionImpl option_defs[] =
 {
-    { TidyUnknownOption,           MS, "unknown!",                    IN, 0,               NULL,              NULL            },
+    { TidyUnknownOption,           MS, "unknown!",                    IN, 0,               NULL,              NULL                },
-    { TidyAccessibilityCheckLevel, DG, "accessibility-check",         IN, 0,               ParseAcc,          accessPicks     },
+    { TidyAccessibilityCheckLevel, DG, "accessibility-check",         IN, 0,               ParseAcc,          &accessPicks        },
-    { TidyAltText,                 MU, "alt-text",                    ST, 0,               ParseString,       NULL            },
+    { TidyAltText,                 MU, "alt-text",                    ST, 0,               ParseString,       NULL                },
-    { TidyAnchorAsName,            MU, "anchor-as-name",              BL, yes,             ParseBool,         boolPicks       },
+    { TidyAnchorAsName,            MU, "anchor-as-name",              BL, yes,             ParsePickList,     &boolPicks          },
-    { TidyAsciiChars,              CE, "ascii-chars",                 BL, no,              ParseBool,         boolPicks       },
+    { TidyAsciiChars,              CE, "ascii-chars",                 BL, no,              ParsePickList,     &boolPicks          },
-    { TidyBlockTags,               MU, "new-blocklevel-tags",         ST, 0,               ParseTagNames,     NULL            },
+    { TidyBlockTags,               MU, "new-blocklevel-tags",         ST, 0,               ParseTagNames,     NULL                },
-    { TidyBodyOnly,                MU, "show-body-only",              IN, no,              ParseAutoBool,     autoBoolPicks   },
+    { TidyBodyOnly,                MU, "show-body-only",              IN, no,              ParsePickList,     &autoBoolPicks      },
-    { TidyBreakBeforeBR,           PP, "break-before-br",             BL, no,              ParseBool,         boolPicks       },
+    { TidyBreakBeforeBR,           PP, "break-before-br",             BL, no,              ParsePickList,     &boolPicks          },
-    { TidyCharEncoding,            CE, "char-encoding",               IN, UTF8,            ParseCharEnc,      charEncPicks    },
+    { TidyCharEncoding,            CE, "char-encoding",               IN, UTF8,            ParseCharEnc,      &charEncPicks       },
-    { TidyCoerceEndTags,           MU, "coerce-endtags",              BL, yes,             ParseBool,         boolPicks       },
+    { TidyCoerceEndTags,           MU, "coerce-endtags",              BL, yes,             ParsePickList,     &boolPicks          },
-    { TidyCSSPrefix,               MU, "css-prefix",                  ST, 0,               ParseCSS1Selector, NULL            },
+    { TidyCSSPrefix,               MU, "css-prefix",                  ST, 0,               ParseCSS1Selector, NULL                },
-    { TidyCustomTags,              IR, "new-custom-tags",             ST, 0,               ParseTagNames,     NULL            }, /* 20170309 - Issue #119 */
+    { TidyCustomTags,              IR, "new-custom-tags",             ST, 0,               ParseTagNames,     NULL                }, /* 20170309 - Issue #119 */
-    { TidyDecorateInferredUL,      MU, "decorate-inferred-ul",        BL, no,              ParseBool,         boolPicks       },
+    { TidyDecorateInferredUL,      MU, "decorate-inferred-ul",        BL, no,              ParsePickList,     &boolPicks          },
-    { TidyDoctype,                 MU, "doctype",                     ST, 0,               ParseDocType,      doctypePicks    },
+    { TidyDoctype,                 MU, "doctype",                     ST, 0,               ParseDocType,      &doctypePicks       },
 #ifndef DOXYGEN_SHOULD_SKIP_THIS
-    { TidyDoctypeMode,             IR, "doctype-mode",                IN, TidyDoctypeAuto, NULL,              doctypePicks    },
+    { TidyDoctypeMode,             IR, "doctype-mode",                IN, TidyDoctypeAuto, NULL,              &doctypePicks       },
 #endif
-    { TidyDropEmptyElems,          MU, "drop-empty-elements",         BL, yes,             ParseBool,         boolPicks       },
+    { TidyDropEmptyElems,          MU, "drop-empty-elements",         BL, yes,             ParsePickList,     &boolPicks          },
-    { TidyDropEmptyParas,          MU, "drop-empty-paras",            BL, yes,             ParseBool,         boolPicks       },
+    { TidyDropEmptyParas,          MU, "drop-empty-paras",            BL, yes,             ParsePickList,     &boolPicks          },
-    { TidyDropPropAttrs,           MU, "drop-proprietary-attributes", BL, no,              ParseBool,         boolPicks       },
+    { TidyDropPropAttrs,           MU, "drop-proprietary-attributes", BL, no,              ParsePickList,     &boolPicks          },
-    { TidyDuplicateAttrs,          MU, "repeated-attributes",         IN, TidyKeepLast,    ParseRepeatAttr,   repeatAttrPicks },
+    { TidyDuplicateAttrs,          MU, "repeated-attributes",         IN, TidyKeepLast,    ParsePickList,     &repeatAttrPicks    },
-    { TidyEmacs,                   MS, "gnu-emacs",                   BL, no,              ParseBool,         boolPicks       },
+    { TidyEmacs,                   MS, "gnu-emacs",                   BL, no,              ParsePickList,     &boolPicks          },
 #ifndef DOXYGEN_SHOULD_SKIP_THIS
-    { TidyEmacsFile,               IR, "gnu-emacs-file",              ST, 0,               ParseString,       NULL            },
+    { TidyEmacsFile,               IR, "gnu-emacs-file",              ST, 0,               ParseString,       NULL                },
 #endif
-    { TidyEmptyTags,               MU, "new-empty-tags",              ST, 0,               ParseTagNames,     NULL            },
+    { TidyEmptyTags,               MU, "new-empty-tags",              ST, 0,               ParseTagNames,     NULL                },
-    { TidyEncloseBlockText,        MU, "enclose-block-text",          BL, no,              ParseBool,         boolPicks       },
+    { TidyEncloseBlockText,        MU, "enclose-block-text",          BL, no,              ParsePickList,     &boolPicks          },
-    { TidyEncloseBodyText,         MU, "enclose-text",                BL, no,              ParseBool,         boolPicks       },
+    { TidyEncloseBodyText,         MU, "enclose-text",                BL, no,              ParsePickList,     &boolPicks          },
-    { TidyErrFile,                 MS, "error-file",                  ST, 0,               ParseString,       NULL            },
+    { TidyErrFile,                 MS, "error-file",                  ST, 0,               ParseString,       NULL                },
-    { TidyEscapeCdata,             MU, "escape-cdata",                BL, no,              ParseBool,         boolPicks       },
+    { TidyEscapeCdata,             MU, "escape-cdata",                BL, no,              ParsePickList,     &boolPicks          },
-    { TidyEscapeScripts,           PP, "escape-scripts",              BL, yes,             ParseBool,         boolPicks       }, /* 20160227 - Issue #348 */
+    { TidyEscapeScripts,           PP, "escape-scripts",              BL, yes,             ParsePickList,     &boolPicks          }, /* 20160227 - Issue #348 */
-    { TidyFixBackslash,            MU, "fix-backslash",               BL, yes,             ParseBool,         boolPicks       },
+    { TidyFixBackslash,            MU, "fix-backslash",               BL, yes,             ParsePickList,     &boolPicks          },
-    { TidyFixComments,             MU, "fix-bad-comments",            BL, yes,             ParseBool,         boolPicks       },
+    { TidyFixComments,             MU, "fix-bad-comments",            BL, yes,             ParsePickList,     &boolPicks          },
-    { TidyFixUri,                  MU, "fix-uri",                     BL, yes,             ParseBool,         boolPicks       },
+    { TidyFixUri,                  MU, "fix-uri",                     BL, yes,             ParsePickList,     &boolPicks          },
-    { TidyForceOutput,             MS, "force-output",                BL, no,              ParseBool,         boolPicks       },
+    { TidyForceOutput,             MS, "force-output",                BL, no,              ParsePickList,     &boolPicks          },
-    { TidyGDocClean,               MU, "gdoc",                        BL, no,              ParseBool,         boolPicks       },
+    { TidyGDocClean,               MU, "gdoc",                        BL, no,              ParsePickList,     &boolPicks          },
-    { TidyHideComments,            MU, "hide-comments",               BL, no,              ParseBool,         boolPicks       },
+    { TidyHideComments,            MU, "hide-comments",               BL, no,              ParsePickList,     &boolPicks          },
-    { TidyHtmlOut,                 MU, "output-html",                 BL, no,              ParseBool,         boolPicks       },
+    { TidyHtmlOut,                 MU, "output-html",                 BL, no,              ParsePickList,     &boolPicks          },
-    { TidyInCharEncoding,          CE, "input-encoding",              IN, UTF8,            ParseCharEnc,      charEncPicks    },
+    { TidyInCharEncoding,          CE, "input-encoding",              IN, UTF8,            ParseCharEnc,      &charEncPicks       },
-    { TidyIndentAttributes,        PP, "indent-attributes",           BL, no,              ParseBool,         boolPicks       },
+    { TidyIndentAttributes,        PP, "indent-attributes",           BL, no,              ParsePickList,     &boolPicks          },
-    { TidyIndentCdata,             MU, "indent-cdata",                BL, no,              ParseBool,         boolPicks       },
+    { TidyIndentCdata,             MU, "indent-cdata",                BL, no,              ParsePickList,     &boolPicks          },
-    { TidyIndentContent,           PP, "indent",                      IN, TidyNoState,     ParseAutoBool,     autoBoolPicks   },
+    { TidyIndentContent,           PP, "indent",                      IN, TidyNoState,     ParsePickList,     &autoBoolPicks      },
-    { TidyIndentSpaces,            PP, "indent-spaces",               IN, 2,               ParseInt,          NULL            },
+    { TidyIndentSpaces,            PP, "indent-spaces",               IN, 2,               ParseInt,          NULL                },
-    { TidyInlineTags,              MU, "new-inline-tags",             ST, 0,               ParseTagNames,     NULL            },
+    { TidyInlineTags,              MU, "new-inline-tags",             ST, 0,               ParseTagNames,     NULL                },
-    { TidyJoinClasses,             MU, "join-classes",                BL, no,              ParseBool,         boolPicks       },
+    { TidyJoinClasses,             MU, "join-classes",                BL, no,              ParsePickList,     &boolPicks          },
-    { TidyJoinStyles,              MU, "join-styles",                 BL, yes,             ParseBool,         boolPicks       },
+    { TidyJoinStyles,              MU, "join-styles",                 BL, yes,             ParsePickList,     &boolPicks          },
-    { TidyKeepFileTimes,           MS, "keep-time",                   BL, no,              ParseBool,         boolPicks       },
+    { TidyKeepFileTimes,           MS, "keep-time",                   BL, no,              ParsePickList,     &boolPicks          },
-    { TidyLiteralAttribs,          MU, "literal-attributes",          BL, no,              ParseBool,         boolPicks       },
+    { TidyLiteralAttribs,          MU, "literal-attributes",          BL, no,              ParsePickList,     &boolPicks          },
-    { TidyLogicalEmphasis,         MU, "logical-emphasis",            BL, no,              ParseBool,         boolPicks       },
+    { TidyLogicalEmphasis,         MU, "logical-emphasis",            BL, no,              ParsePickList,     &boolPicks          },
-    { TidyLowerLiterals,           MU, "lower-literals",              BL, yes,             ParseBool,         boolPicks       },
+    { TidyLowerLiterals,           MU, "lower-literals",              BL, yes,             ParsePickList,     &boolPicks          },
-    { TidyMakeBare,                MU, "bare",                        BL, no,              ParseBool,         boolPicks       },
+    { TidyMakeBare,                MU, "bare",                        BL, no,              ParsePickList,     &boolPicks          },
-    { TidyMakeClean,               MU, "clean",                       BL, no,              ParseBool,         boolPicks       },
+    { TidyMakeClean,               MU, "clean",                       BL, no,              ParsePickList,     &boolPicks          },
-    { TidyMark,                    MS, "tidy-mark",                   BL, yes,             ParseBool,         boolPicks       },
+    { TidyMark,                    MS, "tidy-mark",                   BL, yes,             ParsePickList,     &boolPicks          },
-    { TidyMergeDivs,               MU, "merge-divs",                  IN, TidyAutoState,   ParseAutoBool,     autoBoolPicks   },
+    { TidyMergeDivs,               MU, "merge-divs",                  IN, TidyAutoState,   ParsePickList,     &autoBoolPicks      },
-    { TidyMergeEmphasis,           MU, "merge-emphasis",              BL, yes,             ParseBool,         boolPicks       },
+    { TidyMergeEmphasis,           MU, "merge-emphasis",              BL, yes,             ParsePickList,     &boolPicks          },
-    { TidyMergeSpans,              MU, "merge-spans",                 IN, TidyAutoState,   ParseAutoBool,     autoBoolPicks   },
+    { TidyMergeSpans,              MU, "merge-spans",                 IN, TidyAutoState,   ParsePickList,     &autoBoolPicks      },
 #if SUPPORT_ASIAN_ENCODINGS
-    { TidyNCR,                     MU, "ncr",                         BL, yes,             ParseBool,         boolPicks       },
+    { TidyNCR,                     MU, "ncr",                         BL, yes,             ParsePickList,     &boolPicks          },
 #endif
-    { TidyNewline,                 CE, "newline",                     IN, DLF,             ParseNewline,      newlinePicks    },
+    { TidyNewline,                 CE, "newline",                     IN, DLF,             ParsePickList,     &newlinePicks       },
-    { TidyNumEntities,             MU, "numeric-entities",            BL, no,              ParseBool,         boolPicks       },
+    { TidyNumEntities,             MU, "numeric-entities",            BL, no,              ParsePickList,     &boolPicks          },
-    { TidyOmitOptionalTags,        MU, "omit-optional-tags",          BL, no,              ParseBool,         boolPicks       },
+    { TidyOmitOptionalTags,        MU, "omit-optional-tags",          BL, no,              ParsePickList,     &boolPicks          },
-    { TidyOutCharEncoding,         CE, "output-encoding",             IN, UTF8,            ParseCharEnc,      charEncPicks    },
+    { TidyOutCharEncoding,         CE, "output-encoding",             IN, UTF8,            ParseCharEnc,      &charEncPicks       },
-    { TidyOutFile,                 MS, "output-file",                 ST, 0,               ParseString,       NULL            },
+    { TidyOutFile,                 MS, "output-file",                 ST, 0,               ParseString,       NULL                },
 #if SUPPORT_UTF16_ENCODINGS
-    { TidyOutputBOM,               CE, "output-bom",                  IN, TidyAutoState,   ParseAutoBool,     autoBoolPicks   },
+    { TidyOutputBOM,               CE, "output-bom",                  IN, TidyAutoState,   ParsePickList,     &autoBoolPicks      },
 #endif
-    { 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,              ParseBool,         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                },
 #if SUPPORT_ASIAN_ENCODINGS
-    { TidyPunctWrap,               PP, "punctuation-wrap",            BL, no,              ParseBool,         boolPicks       },
+    { TidyPunctWrap,               PP, "punctuation-wrap",            BL, no,              ParsePickList,     &boolPicks          },
 #endif
     { TidyQuiet,                   MS, "quiet",                       BL, no,              ParseBool,         boolPicks       },
     { TidyQuoteAmpersand,          MU, "quote-ampersand",             BL, yes,             ParseBool,         boolPicks       },
@@ -340,6 +316,7 @@ static const TidyOptionImpl option_defs[] =
     { N_TIDY_OPTIONS,              XX, NULL,                          XY, 0,               NULL,              NULL            }
 };
 
+
 /* Should only be called by options set by name
 ** thus, it is cheaper to do a few scans than set
 ** up every option in a hash table.
@@ -625,7 +602,7 @@ TidyTriState    TY_(_cfgGetAutoBool)( TidyDocImpl* doc, TidyOptionId optId )
   ulong val = TY_(_cfgGet)( doc, optId );
   const TidyOptionImpl* opt = &option_defs[ optId ];
   assert( opt && opt->type == TidyInteger
-          && opt->parser == ParseAutoBool );
+          && opt->parser == ParsePickList );
   return (TidyTriState) val;
 }
 
@@ -1058,7 +1035,7 @@ void AdjustConfig( TidyDocImpl* doc )
     {
         TY_(SetOptionBool)( doc, TidyXmlOut, yes );
         TY_(SetOptionBool)( doc, TidyUpperCaseTags, no );
-        TY_(SetOptionBool)( doc, TidyUpperCaseAttrs, no );
+        TY_(SetOptionInt)( doc, TidyUpperCaseAttrs, no );
         /* TY_(SetOptionBool)( doc, TidyXmlPIs, yes ); */
     }
 
@@ -1122,76 +1099,6 @@ Bool ParseInt( TidyDocImpl* doc, const TidyOptionImpl* entry )
     return digits;
 }
 
-/* true/false or yes/no or 0/1 or "auto" only looks at 1st char */
-static Bool ParseTriState( TidyTriState theState, TidyDocImpl* doc,
-                           const TidyOptionImpl* entry, ulong* flag )
-{
-    TidyConfigImpl* cfg = &doc->config;
-    tchar c = SkipWhite( cfg );
-
-    if (c == 't' || c == 'T' || c == 'y' || c == 'Y' || c == '1')
-        *flag = yes;
-    else if (c == 'f' || c == 'F' || c == 'n' || c == 'N' || c == '0')
-        *flag = no;
-    else if (theState == TidyAutoState && (c == 'a' || c =='A'))
-        *flag = TidyAutoState;
-    else
-    {
-        TY_(ReportBadArgument)( doc, entry->name );
-        return no;
-    }
-
-    return yes;
-}
-
-/* cr, lf or crlf */
-Bool ParseNewline( TidyDocImpl* doc, const TidyOptionImpl* entry )
-{
-    int nl = -1;
-    tmbchar work[ 16 ] = {0};
-    tmbstr cp = work, end = work + sizeof(work);
-    TidyConfigImpl* cfg = &doc->config;
-    tchar c = SkipWhite( cfg );
-
-    while ( c!=EndOfStream && cp < end && !TY_(IsWhite)(c) && c != '\r' && c != '\n' )
-    {
-        *cp++ = (tmbchar) c;
-        c = AdvanceChar( cfg );
-    }
-    *cp = 0;
-
-    if ( TY_(tmbstrcasecmp)(work, "lf") == 0 )
-        nl = TidyLF;
-    else if ( TY_(tmbstrcasecmp)(work, "crlf") == 0 )
-        nl = TidyCRLF;
-    else if ( TY_(tmbstrcasecmp)(work, "cr") == 0 )
-        nl = TidyCR;
-
-    if ( nl < TidyLF || nl > TidyCR )
-        TY_(ReportBadArgument)( doc, entry->name );
-    else
-        TY_(SetOptionInt)( doc, entry->id, nl );
-    return ( nl >= TidyLF && nl <= TidyCR );
-}
-
-Bool ParseBool( TidyDocImpl* doc, const TidyOptionImpl* entry )
-{
-    ulong flag = 0;
-    Bool status = ParseTriState( TidyNoState, doc, entry, &flag );
-    if ( status )
-        TY_(SetOptionBool)( doc, entry->id, flag != 0 );
-    return status;
-}
-
-Bool ParseAutoBool( TidyDocImpl* doc, const TidyOptionImpl* entry )
-{
-    ulong flag = 0;
-    Bool status = ParseTriState( TidyAutoState, doc, entry, &flag );
-    if ( status )
-        TY_(SetOptionInt)( doc, entry->id, flag );
-    return status;
-}
-
 /* a string excluding whitespace */
 Bool FUNC_UNUSED ParseName( TidyDocImpl* doc, const TidyOptionImpl* option )
 {
@@ -1243,15 +1150,77 @@ Bool ParseCSS1Selector( TidyDocImpl* doc, const TidyOptionImpl* option )
     return yes;
 }
 
+/* A general parser for anything using pick lists. This provides the engine to
+   determine the proper option value, and can be used by parsers in addition to
+   ParsePickList that require special handling.
+ */
+Bool GetParsePickListValue( TidyDocImpl* doc, const TidyOptionImpl* entry, uint *result )
+{
+    TidyConfigImpl* cfg = &doc->config;
+    tchar c = SkipWhite( cfg );
+    tmbchar work[ 16 ] = {0};
+    tmbstr cp = work, end = work + sizeof(work);
+    const PickListItem *item = NULL;
+    uint ix = 0;
+    
+    while ( c!=EndOfStream && cp < end && !TY_(IsWhite)(c) && c != '\r' && c != '\n' )
+    {
+        *cp++ = (tmbchar) c;
+        c = AdvanceChar( cfg );
+    }
+    
+    while ( (item = &(*entry->pickList)[ ix ]) && item->label )
+    {
+        ctmbstr input;
+        uint i = 0;
+        while ( ( input = &(*item->inputs[i]) ) )
+        {
+            if (TY_(tmbstrcasecmp)(work, input) == 0 )
+            {
+                *result = ix;
+                return yes;
+            }
+            ++i;
+        }
+        ++ix;
+    }
+    
+    TY_(ReportBadArgument)( doc, entry->name );
+    return no;
+}
+
+
+/* A general parser for anything using pick lists that don't require special
+   handling.
+ */
+Bool ParsePickList( TidyDocImpl* doc, const TidyOptionImpl* entry )
+{
+    uint value;
+    
+    if ( GetParsePickListValue( doc, entry, &value ) )
+    {
+        if ( entry->type == TidyBoolean )
+            TY_(SetOptionBool)( doc, entry->id, value );
+        else if ( entry->type == TidyInteger )
+            TY_(SetOptionInt)( doc, entry->id, value );
+        return yes;
+    }
+  
+    TY_(ReportBadArgument)( doc, entry->name );
+    return no;
+}
+
+
 /*\
  * 20150515 - support using tabs instead of spaces - Issue #108
  * Sets the indent character to a tab if on, and set indent space count to 1
  * and sets indent character to a space if off.
-\*/
+ \*/
 Bool ParseTabs( TidyDocImpl* doc, const TidyOptionImpl* entry )
 {
-    ulong flag = 0;
+    uint flag = 0;
-    Bool status = ParseTriState( TidyNoState, doc, entry, &flag );
+    Bool status = GetParsePickListValue( doc, entry, &flag );
+    
     if ( status ) {
         Bool tabs = flag != 0 ? yes : no;
         TY_(SetOptionBool)( doc, entry->id, tabs );
@@ -1264,52 +1233,6 @@ Bool ParseTabs( TidyDocImpl* doc, const TidyOptionImpl* entry )
     return status;
 }
 
-/* Parse the value of TidyUseCustomTags. Like other option values, we will
- * look for the first character only, of no, blocklevel, empty, inline, pre.
- */
-Bool ParseUseCustomTags( TidyDocImpl* doc, const TidyOptionImpl* entry )
-{
-    uint value;
-    TidyConfigImpl* cfg = &doc->config;
-    tchar c = SkipWhite( cfg );
-    
-    switch (c)
-    {
-        case 'n':
-        case 'N':
-            value = TidyCustomNo;
-            break;
-            
-        case 'b':
-            value = TidyCustomBlocklevel;
-            break;
-            
-        case 'e':
-        case 'E':
-            value = TidyCustomEmpty;
-            break;
-        
-        case 'y':
-        case 'Y':
-        case 'i':
-        case 'I':
-            value = TidyCustomInline;
-            break;
-
-        case 'p':
-        case 'P':
-            value = TidyCustomPre;
-            break;
-            
-        default:
-            TY_(ReportBadArgument)( doc, entry->name );
-            return no;
-    }
-    
-    TY_(SetOptionInt)( doc, TidyUseCustomTags, value );
-    return yes;
-}
-
 
 /* Coordinates Config update and Tags data */
 void TY_(DeclareUserTag)( TidyDocImpl* doc, TidyOptionId optId,
@@ -1545,11 +1468,8 @@ ctmbstr TY_(CharEncodingOptName)( int encoding )
 */
 Bool ParseDocType( TidyDocImpl* doc, const TidyOptionImpl* option )
 {
-    tmbchar buf[ 32 ] = {0};
-    uint i = 0;
     Bool status = yes;
-    TidyDoctypeModes dtmode = TidyDoctypeAuto;
+    uint value;
-
     TidyConfigImpl* cfg = &doc->config;
     tchar c = SkipWhite( cfg );
 
@@ -1559,94 +1479,21 @@ Bool ParseDocType( TidyDocImpl* doc, const TidyOptionImpl* option )
     {
         status = ParseString(doc, option);
         if (status)
+        {
             TY_(SetOptionInt)( doc, TidyDoctypeMode, TidyDoctypeUser );
-
+        }
         return status;
     }
-
+    
-    /* read first word */
+    if ( (status = GetParsePickListValue( doc, option, &value ) ) )
-    while ( i < sizeof(buf)-1 && c != EndOfStream && !TY_(IsWhite)(c) )
     {
-        buf[i++] = (tmbchar) c;
+        TY_(SetOptionInt)( doc, TidyDoctypeMode, value );
-        c = AdvanceChar( cfg );
     }
-    buf[i] = '\0';
-
-    if ( TY_(tmbstrcasecmp)(buf, "auto") == 0 )
-        dtmode = TidyDoctypeAuto;
-    else if ( TY_(tmbstrcasecmp)(buf, "html5") == 0 )
-        dtmode = TidyDoctypeHtml5;
-    else if ( TY_(tmbstrcasecmp)(buf, "omit") == 0 )
-        dtmode = TidyDoctypeOmit;
-    else if ( TY_(tmbstrcasecmp)(buf, "strict") == 0 )
-        dtmode = TidyDoctypeStrict;
-    else if ( TY_(tmbstrcasecmp)(buf, "loose") == 0 ||
-              TY_(tmbstrcasecmp)(buf, "transitional") == 0 )
-        dtmode = TidyDoctypeLoose;
     else
     {
         TY_(ReportBadArgument)( doc, option->name );
-        status = no;
-    }
-     
-    if ( status )
-        TY_(SetOptionInt)( doc, TidyDoctypeMode, dtmode );
-    return status;
-}
-
-Bool ParseRepeatAttr( TidyDocImpl* doc, const TidyOptionImpl* option )
-{
-    Bool status = yes;
-    tmbchar buf[64] = {0};
-    uint i = 0;
-
-    TidyConfigImpl* cfg = &doc->config;
-    tchar c = SkipWhite( cfg );
-
-    while (i < sizeof(buf)-1 && c != EndOfStream && !TY_(IsWhite)(c))
-    {
-        buf[i++] = (tmbchar) c;
-        c = AdvanceChar( cfg );
-    }
-    buf[i] = '\0';
-
-    if ( TY_(tmbstrcasecmp)(buf, "keep-first") == 0 )
-        cfg->value[ TidyDuplicateAttrs ].v = TidyKeepFirst;
-    else if ( TY_(tmbstrcasecmp)(buf, "keep-last") == 0 )
-        cfg->value[ TidyDuplicateAttrs ].v = TidyKeepLast;
-    else
-    {
-        TY_(ReportBadArgument)( doc, option->name );
-        status = no;
-    }
-    return status;
-}
-
-Bool ParseSorter( TidyDocImpl* doc, const TidyOptionImpl* option )
-{
-    Bool status = yes;
-    tmbchar buf[64] = {0};
-    uint i = 0;
-
-    TidyConfigImpl* cfg = &doc->config;
-    tchar c = SkipWhite( cfg );
-
-    while (i < sizeof(buf)-1 && c != EndOfStream && !TY_(IsWhite)(c))
-    {
-        buf[i++] = (tmbchar) c;
-        c = AdvanceChar( cfg );
-    }
-    buf[i] = '\0';
-
-    if ( TY_(tmbstrcasecmp)(buf, "alpha") == 0 )
-        cfg->value[ TidySortAttributes ].v = TidySortAttrAlpha;
-    else if ( TY_(tmbstrcasecmp)(buf, "none") == 0)
-        cfg->value[ TidySortAttributes ].v = TidySortAttrNone;
-    else
-    {
-        TY_(ReportBadArgument)( doc, option->name );
-        status = no;
     }
+    
     return status;
 }
 
@@ -1692,12 +1539,22 @@ ctmbstr      TY_(getNextOptionPick)( const TidyOptionImpl* option,
 {
     size_t ix;
     ctmbstr val = NULL;
+    const PickListItem *item= NULL;
     assert( option!=NULL && iter != NULL );
 
     ix = (size_t) *iter;
-    if ( ix > 0 && ix < 16 && option->pickList )
+    
-        val = option->pickList[ ix-1 ];
+    if ( option->pickList )
-    *iter = (TidyIterator) ( val && option->pickList[ix] ? ix + 1 : (size_t)0 );
+    {
+        if ( ix > 0 && ix < TIDY_PL_SIZE && option->pickList )
+        {
+            item = &(*option->pickList)[ ix-1 ];
+            val = item->label;
+        }
+        item = &(*option->pickList)[ ix ];
+        *iter = (TidyIterator) ( val && item->label ? ix + 1 : (size_t)0 );
+    }
+    
     return val;
 }
 
@@ -1731,12 +1588,19 @@ static int  WriteOptionBool( const TidyOptionImpl* option, Bool bval, StreamOut*
 
 static int  WriteOptionPick( const TidyOptionImpl* option, uint ival, StreamOut* out )
 {
-    uint ix;
+    uint ix = 0;
-    const ctmbstr* val = option->pickList;
+    const PickListItem *item = NULL;
-    for ( ix=0; val[ix] && ix<ival; ++ix )
+    
-        /**/;
+    if ( option-> pickList )
-    if ( ix==ival && val[ix] )
+    {
-        return WriteOptionString( option, val[ix], out );
+        while ( (item = &(*option->pickList)[ ix ]) && item->label && ix<ival )
+        {
+            ++ix;
+        }
+        if ( ix==ival && item->label )
+            return WriteOptionString( option, item->label, out );
+    }
+    
     return -1;
 }
 
@@ -1796,7 +1660,7 @@ static int  SaveConfigToStream( TidyDocImpl* doc, StreamOut* out )
           else
             rc = WriteOptionPick( option, dtmode, out );
         }
-        else if ( option->pickList )
+        else if ( option->pickList)
           rc = WriteOptionPick( option, val->v, out );
         else
         {

src/config.h

@@ -1,30 +1,65 @@
 #ifndef __CONFIG_H__
 #define __CONFIG_H__
 
-/* config.h -- read config file and manage config properties
+/**************************************************************************//**
-  
+ * @file
-  (c) 1998-2006 (W3C) MIT, ERCIM, Keio University
+ * Read configuration files and manage configuration properties.
-  See tidy.h for the copyright notice.
+ *
-
+ * Config files associate a property name with a value.
-  config files associate a property name with a value.
+ *
-
+ * // comments can start at the beginning of a line
-  // comments can start at the beginning of a line
+ * # comments can start at the beginning of a line
-  # comments can start at the beginning of a line
+ * name: short values fit onto one line
-  name: short values fit onto one line
+ * name: a really long value that
-  name: a really long value that
+ *  continues on the next line
-   continues on the next line
+ *
-
+ * Property names are case insensitive and should be less than 60 characters
-  property names are case insensitive and should be less than
+ * in length, and must start at the begining of the line, as whitespace at
-  60 characters in length and must start at the begining of
+ * the start of a line signifies a line continuation.
-  the line, as whitespace at the start of a line signifies a
+ *
-  line continuation.
+ * @author  HTACG, et al (consult git log)
-
+ * 
-*/
+ * @copyright
+ *     Copyright (c) 1998-2017 World Wide Web Consortium (Massachusetts
+ *     Institute of Technology, European Research Consortium for Informatics
+ *     and Mathematics, Keio University) and HTACG.
+ * @par
+ *     All Rights Reserved.
+ * @par
+ *     See `tidy.h` for the complete license.
+ *
+ * @date Additional updates: consult git log
+ *
+ ******************************************************************************/
 
 #include "forward.h"
 #include "tidy.h"
 #include "streamio.h"
 
+/** PickLists may have up to 16 items. For some reason,
+ ** this limit has always been hard-coded into Tidy.
+ */
+#define TIDY_PL_SIZE 16
+
+/** Structs of this type contain information needed in order to present pick lists,
+ ** relate pick list entries to public enum values, and parse strings that are
+ ** accepted in order to assign the value.
+ */
+typedef struct PickListItem {
+    ctmbstr label;      /**< PickList label for this item. */
+    const int value;    /**< The option value represented by this label. */
+    ctmbstr inputs[10]; /**< String values that can select this value. */
+} PickListItem;
+
+/** An array of PickListItems, fixed in size for in-code declarations. 
+ ** Arrays must be populated in 0 to 10 order, as the option value is assigned
+ ** based on this index and *not* on the structures' value field. It remains
+ ** a best practice, however, to assign a public enum value with the proper
+ ** index value.
+ */
+typedef const PickListItem PickListItems[TIDY_PL_SIZE];
+
+
 struct _tidy_option;
 typedef struct _tidy_option TidyOptionImpl;
 
@@ -33,13 +68,13 @@ typedef Bool (ParseProperty)( TidyDocImpl* doc, const TidyOptionImpl* opt );
 struct _tidy_option
 {
     TidyOptionId        id;
-    TidyConfigCategory  category;   /* put 'em in groups */
+    TidyConfigCategory  category;        /* put 'em in groups */
-    ctmbstr             name;       /* property name */
+    ctmbstr             name;            /* property name */
-    TidyOptionType      type;       /* string, int or bool */
+    TidyOptionType      type;            /* string, int or bool */
-    ulong               dflt;       /* default for TidyInteger and TidyBoolean */
+    ulong               dflt;            /* default for TidyInteger and TidyBoolean */
-    ParseProperty*      parser;     /* parsing method, read-only if NULL */
+    ParseProperty*      parser;          /* parsing method, read-only if NULL */
-    const ctmbstr*      pickList;   /* pick list */
+    PickListItems*      pickList;        /* new style pick list */
-    ctmbstr             pdflt;      /* default for TidyString */
+    ctmbstr             pdflt;           /* default for TidyString */
 };
 
 typedef union

src/language_en.h

@@ -1202,8 +1202,12 @@ static languageDefinition language_en = { whichPluralForm_en, {
         "This option specifies if Tidy should output attribute names in upper "
         "case. "
         "<br/>"
-        "The default is <var>no</var>, which results in lower case attribute "
+        "When set to <var>no</var>, attribute names will be written in lower "
-        "names, except for XML input, where the original case is preserved. "
+        "case. Specifying <var>yes</var> will output attribute names in upper "
+        "case, and <var>preserve</var> can used to leave attribute names "
+        "untouched. "
+        "<br/>"
+        "When using XML input, the original case is always preserved. "
     },
     {/* Important notes for translators:
       - Use only <code></code>, <var></var>, <em></em>, <strong></strong>, and

src/language_en_gb.h

@@ -28,7 +28,7 @@
  *
  * Orginating PO file metadata:
  *   PO_LAST_TRANSLATOR=jderry
- *   PO_REVISION_DATE=2017-03-22 15:54:52
+ *   PO_REVISION_DATE=2017-05-13 21:04:45
  */
 
 #ifdef _MSC_VER

src/language_es.h

@@ -28,7 +28,7 @@
  *
  * Orginating PO file metadata:
  *   PO_LAST_TRANSLATOR=jderry
- *   PO_REVISION_DATE=2017-03-22 15:54:52
+ *   PO_REVISION_DATE=2017-05-13 21:04:45
  */
 
 #ifdef _MSC_VER

src/language_es_mx.h

@@ -28,7 +28,7 @@
  *
  * Orginating PO file metadata:
  *   PO_LAST_TRANSLATOR=jderry
- *   PO_REVISION_DATE=2017-03-22 15:54:52
+ *   PO_REVISION_DATE=2017-05-13 21:04:45
  */
 
 #ifdef _MSC_VER

src/language_zh_cn.h

@@ -28,7 +28,7 @@
  *
  * Orginating PO file metadata:
  *   PO_LAST_TRANSLATOR=jderry
- *   PO_REVISION_DATE=2017-03-22 15:54:52
+ *   PO_REVISION_DATE=2017-05-13 21:04:45
  */
 
 #ifdef _MSC_VER

src/lexer.c

@@ -3723,11 +3723,30 @@ static tmbstr  ParseAttribute( TidyDocImpl* doc, Bool *isempty,
         if (TY_(IsWhite)(c))
             break;
 
+        if (c == '/') /* Issue #395 - potential self closing tag */
+        {
+            c = TY_(ReadChar)(doc->docIn);  /* read next */
+            if (c == '>')
+            {
+                /* got a self closing tag - put is back and continue... */
+                TY_(UngetChar)(c, doc->docIn);
+                break;
+            }
+            else
+            {
+                /* Not '/>' - put it back */
+                TY_(UngetChar)(c, doc->docIn);
+            }
+        }
+
         /* what should be done about non-namechar characters? */
         /* currently these are incorporated into the attr name */
 
-        if ( !cfgBool(doc, TidyXmlTags) && TY_(IsUpper)(c) )
+        if ( cfg(doc, TidyUpperCaseAttrs) != TidyUppercasePreserve )
-            c = TY_(ToLower)(c);
+        {
+            if ( !cfgBool(doc, TidyXmlTags) && TY_(IsUpper)(c) )
+                c = TY_(ToLower)(c);
+        }
 
         TY_(AddCharToLexer)( lexer, c );
         lastc = c;

src/message.c

@@ -243,6 +243,7 @@ void TY_(ReportNotice)(TidyDocImpl* doc, Node *element, Node *node, uint code)
                     tagtype = tidyLocalizedString( TIDYCUSTOMINLINE_STRING );
                     break;
                 case TidyCustomPre:
+                default:
                     tagtype = tidyLocalizedString( TIDYCUSTOMPRE_STRING );
                     break;
             }

src/parser.c

@@ -4833,7 +4833,7 @@ void TY_(ParseDocument)(TidyDocImpl* doc)
                 if ( !htmlOut )
                 {
                     TY_(SetOptionBool)( doc, TidyUpperCaseTags, no );
-                    TY_(SetOptionBool)( doc, TidyUpperCaseAttrs, no );
+                    TY_(SetOptionInt)( doc, TidyUpperCaseAttrs, no );
                 }
             }
         }

src/pprint.c

@@ -1238,7 +1238,7 @@ static void PPrintAttribute( TidyDocImpl* doc, uint indent,
     Bool xmlOut    = cfgBool( doc, TidyXmlOut );
     Bool xhtmlOut  = cfgBool( doc, TidyXhtmlOut );
     Bool wrapAttrs = cfgBool( doc, TidyWrapAttVals );
-    Bool ucAttrs   = cfgBool( doc, TidyUpperCaseAttrs );
+    uint ucAttrs   = cfg( doc, TidyUpperCaseAttrs );
     Bool indAttrs  = cfgBool( doc, TidyIndentAttributes );
     uint xtra      = AttrIndent( doc, node, attr );
     Bool first     = AttrNoIndentFirst( /*doc,*/ node, attr );
@@ -1287,7 +1287,7 @@ static void PPrintAttribute( TidyDocImpl* doc, uint indent,
 
         if (c > 0x7F)
             name += TY_(GetUTF8)(name, &c);
-        else if (ucAttrs)
+        else if (ucAttrs == TidyUppercaseYes)
             c = TY_(ToUpper)(c);
 
         AddChar(pprint, c);
@@ -1734,8 +1734,8 @@ static void PPrintXmlDecl( TidyDocImpl* doc, uint indent, Node *node )
     saveWrap = WrapOff( doc );
 
     /* no case translation for XML declaration pseudo attributes */
-    ucAttrs = cfgBool(doc, TidyUpperCaseAttrs);
+    ucAttrs = cfg(doc, TidyUpperCaseAttrs);
-    TY_(SetOptionBool)(doc, TidyUpperCaseAttrs, no);
+    TY_(SetOptionInt)(doc, TidyUpperCaseAttrs, no);
 
     AddString( pprint, "<?xml" );
 
@@ -1749,7 +1749,7 @@ static void PPrintXmlDecl( TidyDocImpl* doc, uint indent, Node *node )
       PPrintAttribute( doc, indent, node, att );
 
     /* restore old config value */
-    TY_(SetOptionBool)(doc, TidyUpperCaseAttrs, ucAttrs);
+    TY_(SetOptionInt)(doc, TidyUpperCaseAttrs, ucAttrs);
 
     if ( node->end <= 0 || doc->lexer->lexbuf[node->end - 1] != '?' )
         AddChar( pprint, '?' );
@@ -1935,6 +1935,8 @@ static Bool TY_(nodeIsTextLike)( Node *node )
         return yes;
     if ( node->type == AspTag )
         return yes;
+    if (node->type == PhpTag)
+        return yes; /* Issue #392 */
     /* add other text like nodes... */
     return no;
 }

src/tidylib.c

@@ -476,15 +476,22 @@ ctmbstr TIDY_CALL       tidyOptGetEncName( TidyDoc tdoc, TidyOptionId optId )
 ctmbstr TIDY_CALL       tidyOptGetCurrPick( TidyDoc tdoc, TidyOptionId optId )
 {
     const TidyOptionImpl* option = TY_(getOption)( optId );
+
     if ( option && option->pickList )
     {
-        uint ix, pick = tidyOptGetInt( tdoc, optId );
+        uint ix = 0;
-        const ctmbstr* pL = option->pickList;
+        uint pick = tidyOptGetInt( tdoc, optId );
-        for ( ix=0; *pL && ix < pick; ++ix )
+        const PickListItem *item = NULL;
-            ++pL;
+        
-        if ( *pL )
+        // loop through the picklist until index matches the value
-            return *pL;
+        while ( (item = &(*option->pickList)[ ix ]) && item->label && ix<pick )
+        {
+            ++ix;
+        }
+        if ( ix==pick && item->label )
+            return item->label;
     }
+    
     return NULL;
 }
 

version.txt

@@ -1,2 +1,2 @@
-5.5.21
+5.5.29
-2017.05.07
+2017.05.27