Add support for writing non-canonicalized XML

Specifically, don't sort attributes and close tags immediately. Make
this mode the default.

2 files changed, 173 insertions, 55 deletions

diff --git a/genx.c b/genx.c
index 91b0e50..8a1e8e3 100644
--- a/genx.c
+++ b/genx.c
@@ -99,6 +99,7 @@ struct genxAttribute_rec
   collector     value;
   int           provided;   /* provided for current element? */
   attrType      atype;
+  genxAttribute next;       /* Attribute order chain if not canonical. */
 };
 
 /*******************************
@@ -122,7 +123,7 @@ struct genxWriter_rec
   plist           	   attributes;
   plist                    prefixes;
   plist           	   stack;
-  struct genxAttribute_rec arec;
+  struct genxAttribute_rec arec;       /* Dummy attribute used for lookup. */
   char *                   etext[100];
   genxAlloc                alloc;
   genxDealloc              dealloc;
@@ -131,6 +132,13 @@ struct genxWriter_rec
   int                      ppIndent;
   int                      ppDepth;
   Boolean                  ppSimple;
+
+  /* Canonicalization. */
+  Boolean                  canonical;
+
+  /* Attrbute order when not canonical. */
+  genxAttribute            firstAttribute;
+  genxAttribute            lastAttribute;
 };
 
 /*******************************
@@ -605,7 +613,10 @@ genxWriter genxNew(genxAlloc alloc, genxDealloc dealloc, void * userData)
   xml->declaration = xml->defaultDecl;
 
   w->ppIndent = 0; /* Pretty-printing is disabled by default. */
+  w->canonical = False; /* No canonicalization by default. */
 
+  w->firstAttribute = NULL;
+  w->lastAttribute = NULL;
   return w;
 }
 
@@ -626,6 +637,23 @@ genxStatus genxReset (genxWriter w)
     ((genxNamespace) w->namespaces.pointers[i])->baroque = False;
   }
 
+  /* Clear provided attributes. */
+  for (i = 0; i < w->attributes.count; i++)
+    ((genxAttribute) w->attributes.pointers[i])->provided = False;
+
+  /* Clear attribute list. */
+  if (!w->canonical)
+  {
+    while (w->firstAttribute != NULL)
+    {
+      genxAttribute t = w->firstAttribute->next;
+      w->firstAttribute->next = NULL;
+      w->firstAttribute = t;
+    }
+
+    w->lastAttribute = NULL;
+  }
+
   w->status = GENX_SUCCESS;
   w->sequence = SEQUENCE_NO_DOC;
 
@@ -664,6 +692,24 @@ int genxGetPrettyPrint(genxWriter w)
 }
 
 /*
+ * get/set canonicalization.
+ */
+genxStatus genxSetCanonical(genxWriter w, int flag)
+{
+  if (w->sequence == SEQUENCE_NO_DOC)
+    w->canonical = flag;
+  else
+    w->status = GENX_SEQUENCE_ERROR;
+
+  return w->status;
+}
+
+int genxGetCanonical(genxWriter w)
+{
+  return w->canonical;
+}
+
+/*
  * get/set allocator
  */
 void genxSetAlloc(genxWriter w, genxAlloc alloc)
@@ -1089,6 +1135,7 @@ static genxAttribute declareAttribute(genxWriter w, genxNamespace ns,
   a->ns = ns;
   a->provided = False;
   a->atype = w->arec.atype;
+  a->next = NULL;
 
   if ((a->name = copy(w, name)) == NULL)
   {
@@ -1150,7 +1197,7 @@ static genxStatus sendxBounded(genxWriter w, constUtf8 start, constUtf8 end)
     return GENX_IO_ERROR;
 }
 
-#define SendCheck(w,s) if ((w->status=sendx(w,(utf8)s))!=GENX_SUCCESS) return w->status;
+#define SendCheck(w,s) if ((w->status=sendx(w,(constUtf8)s))!=GENX_SUCCESS) return w->status
 
 /*******************************
  * XML writing routines.  The semantics of the externally-facing ones are
@@ -1191,6 +1238,32 @@ static genxStatus writeIndentation(genxWriter w)
 }
 
 /*
+ * Output attribute.
+ */
+static genxStatus writeAttribute(genxAttribute a)
+{
+  genxWriter w = a->writer;
+
+  if (a->ns && a->ns->baroque && a->ns->declaration == w->xmlnsEquals)
+    return w->status = GENX_ATTRIBUTE_IN_DEFAULT_NAMESPACE;
+
+  SendCheck(w, " ");
+
+  if (a->ns)
+  {
+    SendCheck(w, a->ns->declaration->name + STRLEN_XMLNS_COLON);
+    SendCheck(w, ":");
+  }
+
+  SendCheck(w, a->name);
+  SendCheck(w, "=\"");
+  SendCheck(w, a->value.buf);
+  SendCheck(w, "\"");
+
+  return w->status;
+}
+
+/*
  * Write out the attributes we've been gathering up for an element.  We save
  *  them until we've gathered them all so they can be writen in canonical
  *  order.
@@ -1199,7 +1272,7 @@ static genxStatus writeIndentation(genxWriter w)
  *  we build it, then as each attribute is added, we fill in its value and
  *  mark the fact that it's been added, in the "provided" field.
  */
-static genxStatus writeStartTag(genxWriter w)
+static genxStatus writeStartTag(genxWriter w, Boolean close)
 {
   int i;
   genxAttribute * aa = (genxAttribute *) w->attributes.pointers;
@@ -1233,27 +1306,42 @@ static genxStatus writeStartTag(genxWriter w)
   }
   SendCheck(w, e->type);
 
-  for (i = 0; i < w->attributes.count; i++)
+  /* If we are canonicalizing, then write sorted attributes. Otherwise
+     write them in the order specified. */
+  if (w->canonical)
   {
-    if (aa[i]->provided)
+    for (i = 0; i < w->attributes.count; i++)
     {
-      if (aa[i]->ns && aa[i]->ns->baroque &&
-	  aa[i]->ns->declaration == w->xmlnsEquals)
-	return w->status = GENX_ATTRIBUTE_IN_DEFAULT_NAMESPACE;
-
-      SendCheck(w, " ");
-
-      if (aa[i]->ns)
+      if (aa[i]->provided)
       {
-	SendCheck(w, aa[i]->ns->declaration->name + STRLEN_XMLNS_COLON)
-	SendCheck(w, ":");
+        if (writeAttribute (aa[i]) != GENX_SUCCESS)
+          return w->status;
+
+        aa[i]->provided = False;
       }
-      SendCheck(w, aa[i]->name);
-      SendCheck(w, "=\"");
-      SendCheck(w, aa[i]->value.buf);
-      SendCheck(w, "\"");
     }
   }
+  else
+  {
+    /* Keep the chain consistent even if we bail out mid way because of
+       an error. This way we will still be able to clear it in reset().*/
+    while (w->firstAttribute != NULL)
+    {
+      genxAttribute t = w->firstAttribute->next;
+
+      if (writeAttribute (w->firstAttribute) != GENX_SUCCESS)
+        return w->status;
+
+      w->firstAttribute->provided = False;
+      w->firstAttribute->next = NULL;
+      w->firstAttribute = t;
+    }
+
+    w->lastAttribute = NULL;
+  }
+
+  if (close)
+    SendCheck(w, "/");
   SendCheck(w, ">");
   return GENX_SUCCESS;
 }
@@ -1325,7 +1413,6 @@ genxStatus genxUnsetDefaultNamespace(genxWriter w)
 genxStatus genxStartElement(genxElement e)
 {
   genxWriter w = e->writer;
-  int i;
 
   switch (w->sequence)
   {
@@ -1335,7 +1422,7 @@ genxStatus genxStartElement(genxElement e)
     return w->status = GENX_SEQUENCE_ERROR;
   case SEQUENCE_START_TAG:
   case SEQUENCE_ATTRIBUTES:
-    if ((w->status = writeStartTag(w)) != GENX_SUCCESS)
+    if ((w->status = writeStartTag(w, False)) != GENX_SUCCESS)
       return w->status;
     break;
   case SEQUENCE_PRE_DOC:
@@ -1345,10 +1432,6 @@ genxStatus genxStartElement(genxElement e)
 
   w->sequence = SEQUENCE_START_TAG;
 
-  /* clear provided attributes */
-  for (i = 0; i < w->attributes.count; i++)
-    ((genxAttribute) w->attributes.pointers[i])->provided = 0;
-
   /*
    * push the stack.  We push a NULL after a pointer to this element
    *  because the stack will also contain pointers to the namespace
@@ -1575,7 +1658,17 @@ static genxStatus addAttribute(genxAttribute a, constUtf8 valuestr)
 
   if (valuestr && a->provided)
     return w->status = GENX_DUPLICATE_ATTRIBUTE;
-  a->provided = 1;
+
+  a->provided = True;
+
+  /* Add the attribute to the ordered list if not canonical. */
+  if (!w->canonical)
+  {
+    if (w->lastAttribute != NULL)
+      w->lastAttribute = w->lastAttribute->next = a;
+    else
+      w->lastAttribute = w->firstAttribute = a;
+  }
 
   return GENX_SUCCESS;
 }
@@ -1632,15 +1725,24 @@ genxStatus genxEndAttribute(genxWriter w)
   if (a->provided)
     return w->status = GENX_DUPLICATE_ATTRIBUTE;
 
-  a->provided = 1;
+  a->provided = True;
+
+  /* Add the attribute to the ordered list if not canonical. */
+  if (!w->canonical)
+  {
+    if (w->lastAttribute != NULL)
+      w->lastAttribute = w->lastAttribute->next = a;
+    else
+      w->lastAttribute = w->firstAttribute = a;
+  }
 
   return GENX_SUCCESS;
 }
 
 genxStatus genxEndElement(genxWriter w)
 {
-  genxElement e;
   int i;
+  Boolean close = True;
 
   switch (w->sequence)
   {
@@ -1651,42 +1753,51 @@ genxStatus genxEndElement(genxWriter w)
     return w->status = GENX_SEQUENCE_ERROR;
   case SEQUENCE_START_TAG:
   case SEQUENCE_ATTRIBUTES:
-    if ((w->status = writeStartTag(w)) != GENX_SUCCESS)
+    if ((w->status = writeStartTag(w, !w->canonical)) != GENX_SUCCESS)
       return w->status;
+    close = w->canonical;
     break;
   case SEQUENCE_CONTENT:
     break;
   }
 
   /*
-   * first peek into the stack to find the right namespace declaration
-   *  (if any) so we can properly prefix the end-tag.  Have to do this
-   *  before unwinding the stack because that might reset some xmlns
-   *  prefixes to the context in the parent element
+   * Output the closing tag.
    */
-  for (i = w->stack.count - 1; w->stack.pointers[i] != NULL; i -= 2)
-    ;
-  e = (genxElement) w->stack.pointers[--i];
-
-  if (w->ppIndent)
+  if (close)
   {
-    w->ppDepth--;
+    genxElement e;
 
-    if (!w->ppSimple)
-      if (writeIndentation (w) != GENX_SUCCESS)
-        return w->status;
+    /*
+     * first peek into the stack to find the right namespace declaration
+     *  (if any) so we can properly prefix the end-tag.  Have to do this
+     *  before unwinding the stack because that might reset some xmlns
+     *  prefixes to the context in the parent element
+     */
+    for (i = w->stack.count - 1; w->stack.pointers[i] != NULL; i -= 2)
+      ;
+    e = (genxElement) w->stack.pointers[--i];
 
-    w->ppSimple = False;
-  }
+    if (w->ppIndent)
+    {
+      w->ppDepth--;
 
-  SendCheck(w, "</");
-  if (e->ns && e->ns->declaration != w->xmlnsEquals)
-  {
-    SendCheck(w, e->ns->declaration->name + STRLEN_XMLNS_COLON);
-    SendCheck(w, ":");
+      if (!w->ppSimple)
+        if (writeIndentation (w) != GENX_SUCCESS)
+          return w->status;
+
+      w->ppSimple = False;
+    }
+
+    SendCheck(w, "</");
+    if (e->ns && e->ns->declaration != w->xmlnsEquals)
+    {
+      SendCheck(w, e->ns->declaration->name + STRLEN_XMLNS_COLON);
+      SendCheck(w, ":");
+    }
+    SendCheck(w, e->type);
+    SendCheck(w, ">");
   }
-  SendCheck(w, e->type);
-  SendCheck(w, ">");
 
   /*
    * pop zero or more namespace declarations, then a null, then the
@@ -1807,7 +1918,7 @@ genxStatus genxAddText(genxWriter w, constUtf8 start)
   if (w->sequence == SEQUENCE_START_TAG ||
       w->sequence == SEQUENCE_ATTRIBUTES)
   {
-    if ((w->status = writeStartTag(w)) != GENX_SUCCESS)
+    if ((w->status = writeStartTag(w, False)) != GENX_SUCCESS)
       return w->status;
     w->sequence = SEQUENCE_CONTENT;
   }
@@ -1840,7 +1951,7 @@ genxStatus genxAddBoundedText(genxWriter w, constUtf8 start, constUtf8 end)
   if (w->sequence == SEQUENCE_START_TAG ||
       w->sequence == SEQUENCE_ATTRIBUTES)
   {
-    if ((w->status = writeStartTag(w)) != GENX_SUCCESS)
+    if ((w->status = writeStartTag(w, False)) != GENX_SUCCESS)
       return w->status;
     w->sequence = SEQUENCE_CONTENT;
   }
@@ -1880,7 +1991,7 @@ genxStatus genxAddCharacter(genxWriter w, int c)
   if (w->sequence == SEQUENCE_START_TAG ||
       w->sequence == SEQUENCE_ATTRIBUTES)
   {
-    if ((w->status = writeStartTag(w)) != GENX_SUCCESS)
+    if ((w->status = writeStartTag(w, False)) != GENX_SUCCESS)
       return w->status;
     w->sequence = SEQUENCE_CONTENT;
   }
@@ -2005,7 +2116,7 @@ genxStatus genxComment(genxWriter w, constUtf8 text)
   if (w->sequence == SEQUENCE_START_TAG ||
       w->sequence == SEQUENCE_ATTRIBUTES)
   {
-    if ((w->status = writeStartTag(w)) != GENX_SUCCESS)
+    if ((w->status = writeStartTag(w, False)) != GENX_SUCCESS)
       return w->status;
     w->sequence = SEQUENCE_CONTENT;
   }
@@ -2058,7 +2169,7 @@ genxStatus genxPI(genxWriter w, constUtf8 target, constUtf8 text)
   if (w->sequence == SEQUENCE_START_TAG ||
       w->sequence == SEQUENCE_ATTRIBUTES)
   {
-    if ((w->status = writeStartTag(w)) != GENX_SUCCESS)
+    if ((w->status = writeStartTag(w, False)) != GENX_SUCCESS)
       return w->status;
     w->sequence = SEQUENCE_CONTENT;
   }
diff --git a/genx.h b/genx.h
index 6369990..0d4fd48 100644
--- a/genx.h
+++ b/genx.h
@@ -116,6 +116,13 @@ genxStatus genxSetPrettyPrint(genxWriter w, int indentation);
 int genxGetPrettyPrint(genxWriter w);
 
 /*
+ * Set/get canonicalization. If true, then output explicit closing
+ * tags and sort attributes. Default is false.
+ */
+genxStatus genxSetCanonical(genxWriter w, int flag);
+int genxGetCanonical(genxWriter w);
+
+/*
  * User-provided memory allocator, if desired.  For example, if you were
  *  in an Apache module, you could arrange for genx to use ap_palloc by
  *  making the pool accessible via the userData call.