From 7f73d4f429cfe68a0c7384c7f9e577e13500d7fe Mon Sep 17 00:00:00 2001
From: Geoff McLane <ubuntu@geoffair.info>
Date: Sat, 11 Feb 2017 18:33:45 +0100
Subject: [PATCH] Issue #483 - Add ReportSurrogateError() service and connect.

---
 src/language_en.h |  6 +++---
 src/lexer.c       | 12 ++++--------
 src/message.c     |  7 +++++++
 src/message.h     |  1 +
 4 files changed, 15 insertions(+), 11 deletions(-)

diff --git a/src/language_en.h b/src/language_en.h
index b048c73..ab191e0 100644
--- a/src/language_en.h
+++ b/src/language_en.h
@@ -336,9 +336,9 @@ static languageDefinition language_en = { whichPluralForm_en, {
     { INVALID_UTF8,                 0,   "%s invalid UTF-8 bytes (char. code %s)"                                  }, /* Error */
     { INVALID_UTF16,                0,   "%s invalid UTF-16 surrogate pair (char. code %s)"                        }, /* Error */
     { INVALID_NCR,                  0,   "%s invalid numeric character reference %s"                               }, /* Error */
-    { BAD_SURROGATE_PAIR,           0,   "Have out-of-range surrogate pair U+%s:U+%s, replaced with 2 U+FFFD values."}, /* warning */
-    { BAD_SURROGATE_TAIL,           0,   "Leading (High) surrogate pair U+%s, with no trailing (Low) entity, replaced with U+FFFD." }, /* warning */
-    { BAD_SURROGATE_LEAD,           0,   "Trailing (Low) surrogate pair U+%s, with no leading (High) entity, replaced with U+FFFD." }, /* warning */
+    { BAD_SURROGATE_PAIR,           0,   "Have out-of-range surrogate pair U+%04X:U+%04X, replaced with U+FFFD value."}, /* warning */
+    { BAD_SURROGATE_TAIL,           0,   "Leading (High) surrogate pair U+%04X, with no trailing (Low) entity, replaced with U+FFFD." }, /* warning */
+    { BAD_SURROGATE_LEAD,           0,   "Trailing (Low) surrogate pair U+%04X, with no leading (High) entity, replaced with U+FFFD." }, /* warning */
 
     /* ReportEntityError */
     { MISSING_SEMICOLON,            0,   "entity \"%s\" doesn't end in ';'"                                        }, /* Warning in HTML, Error in XML/XHTML */
diff --git a/src/lexer.c b/src/lexer.c
index e3fa267..210c9cf 100644
--- a/src/lexer.c
+++ b/src/lexer.c
@@ -1121,8 +1121,7 @@ static SPStatus GetSurrogatePair(TidyDocImpl* doc, Bool isXml, uint *pch)
             {
                 status = SP_failed; /* is one of the 32 out-of-range pairs */
                 *pch = 0xFFFD;  /* return substitute character */
-                /* SP WARNING: - BAD_SURROGATE_PAIR */
-                fprintf(stderr, "Warning: Have out-of-range surrogate pair U+%04X:U+%04X, replaced with U+FFFD value.\n", fch, ch);
+                TY_(ReportSurrogateError)(doc, BAD_SURROGATE_PAIR, fch, ch); /* SP WARNING: -  */
             }
         }
     }
@@ -1291,8 +1290,7 @@ static void ParseEntity( TidyDocImpl* doc, GetTokenMode mode )
             status = GetSurrogatePair(doc, isXml, &ch);
             if (status == SP_error)
             {
-                /* SP WARNING: BAD_SURROGATE_TAIL - use substitute character */
-                fprintf(stderr, "Warning: Leading(High) surrogate pair U+%04X, with no trailing(Low) entity, replaced with U+FFFD.\n", c1);
+                TY_(ReportSurrogateError)(doc, BAD_SURROGATE_TAIL, c1, 0); /* SP WARNING: - using substitute character */
                 TY_(UngetChar)('&', doc->docIn);  /* otherwise put it back */
             }
         }
@@ -1301,16 +1299,14 @@ static void ParseEntity( TidyDocImpl* doc, GetTokenMode mode )
             /* put this non-entity lead char back */
             TY_(UngetChar)(c1, doc->docIn);
             /* Have leading surrogate pair, with no tail */
-            /* SP WARNING: BAD_SURROGATE_TAIL - use substitute character */
-            fprintf(stderr, "Warning: Leading(High) surrogate pair U+%04X, with no trailing(Low) entity, replaced with U+FFFD.\n", ch);
+            TY_(ReportSurrogateError)(doc, BAD_SURROGATE_TAIL, ch, 0); /* SP WARNING: - using substitute character */
             ch = 0xFFFD;
         }
     } 
     else if (!preserveEntities && found && TY_(IsHighSurrogate)(ch))
     {
         /* Have trailing surrogate pair, with no lead */
-        /* SP WARNING: - BAD_SURROGATE_LEAD -  - use substitute character */
-        fprintf(stderr, "Warning: Trailing (Low) surrogate pair U+%04X, with no leading (High) entity, replaced with U+FFFD.\n", ch);
+        TY_(ReportSurrogateError)(doc, BAD_SURROGATE_LEAD, ch, 0); /* SP WARNING: - using substitute character */
         ch = 0xFFFD;
     }
 
diff --git a/src/message.c b/src/message.c
index 2948deb..31bbb42 100755
--- a/src/message.c
+++ b/src/message.c
@@ -530,6 +530,13 @@ void TY_(ReportEntityError)( TidyDocImpl* doc, uint code, ctmbstr entity,
         messageLexer( doc, TidyWarning, code, fmt, entityname );
 }
 
+void TY_(ReportSurrogateError)(TidyDocImpl* doc, uint code, uint c1, uint c2)
+{
+    ctmbstr fmt = tidyLocalizedString(code);
+    if (fmt)
+        messageLexer(doc, TidyWarning, code, fmt, c1, c2);
+}
+
 void TY_(ReportAttrError)(TidyDocImpl* doc, Node *node, AttVal *av, uint code)
 {
     char const *name = "NULL", *value = "NULL";
diff --git a/src/message.h b/src/message.h
index a1fc129..affc5af 100644
--- a/src/message.h
+++ b/src/message.h
@@ -46,6 +46,7 @@ void TY_(ReportEncodingError)(TidyDocImpl* doc, uint code, uint c, Bool discarde
 void TY_(ReportEntityError)( TidyDocImpl* doc, uint code, ctmbstr entity, int c );
 void TY_(ReportAttrError)( TidyDocImpl* doc, Node* node, AttVal* av, uint code );
 void TY_(ReportMissingAttr)( TidyDocImpl* doc, Node* node, ctmbstr name );
+void TY_(ReportSurrogateError)(TidyDocImpl* doc, uint code, uint c1, uint c2);
 
 #if SUPPORT_ACCESSIBILITY_CHECKS