encoding: Support input-only and output-only converters

Make it possible to open an encoding handler only for input or output.
This avoids the creation of unnecessary converters.

Should also fix #863.

doc/libxml2-api.xml

@@ -349,7 +349,10 @@
      <exports symbol='XML_ENC_ERR_MEMORY' type='enum'/>
      <exports symbol='XML_ENC_ERR_SPACE' type='enum'/>
      <exports symbol='XML_ENC_ERR_SUCCESS' type='enum'/>
+     <exports symbol='XML_ENC_INPUT' type='enum'/>
+     <exports symbol='XML_ENC_OUTPUT' type='enum'/>
      <exports symbol='xmlCharEncError' type='typedef'/>
+     <exports symbol='xmlCharEncFlags' type='typedef'/>
      <exports symbol='xmlCharEncoding' type='typedef'/>
      <exports symbol='xmlCharEncodingHandler' type='typedef'/>
      <exports symbol='xmlCharEncodingHandlerPtr' type='typedef'/>
@@ -4528,6 +4531,8 @@ and not by parsing an instance'/>
     <enum name='XML_ENC_ERR_MEMORY' file='encoding' value='-4' type='xmlCharEncError'/>
     <enum name='XML_ENC_ERR_SPACE' file='encoding' value='-3' type='xmlCharEncError'/>
     <enum name='XML_ENC_ERR_SUCCESS' file='encoding' value='0' type='xmlCharEncError'/>
+    <enum name='XML_ENC_INPUT' file='encoding' value='1' type='xmlCharEncFlags'/>
+    <enum name='XML_ENC_OUTPUT' file='encoding' value='2' type='xmlCharEncFlags'/>
     <enum name='XML_ENTITY_DECL' file='tree' value='17' type='xmlElementType'/>
     <enum name='XML_ENTITY_NODE' file='tree' value='6' type='xmlElementType' info='unused'/>
     <enum name='XML_ENTITY_REF_NODE' file='tree' value='5' type='xmlElementType'/>
@@ -5721,6 +5726,7 @@ crash if you try to modify the tree)'/>
       <info>This is a basic byte in an UTF-8 encoded string. It&apos;s unsigned allowing to pinpoint case where char * are assigned to xmlChar * (possibly making serialization back impossible).</info>
     </typedef>
     <typedef name='xmlCharEncError' file='encoding' type='enum'/>
+    <typedef name='xmlCharEncFlags' file='encoding' type='enum'/>
     <typedef name='xmlCharEncoding' file='encoding' type='enum'/>
     <struct name='xmlCharEncodingHandler' file='encoding' type='struct _xmlCharEncodingHandler'>
       <field name='name' type='char *'/>
@@ -8117,11 +8123,11 @@ crash if you try to modify the tree)'/>
       <arg name='flush' type='int' info='end of input'/>
     </functype>
     <functype name='xmlCharEncConvImpl' file='encoding' module='encoding'>
-      <info>If this function returns XML_ERR_OK, it must fill the @out pointer with an encoding handler. The handler can be obtained from xmlCharEncNewCustomHandler.</info>
+      <info>If this function returns XML_ERR_OK, it must fill the @out pointer with an encoding handler. The handler can be obtained from xmlCharEncNewCustomHandler.  @flags can contain XML_ENC_INPUT, XML_ENC_OUTPUT or both.</info>
       <return type='int' info='an xmlParserErrors code.'/>
       <arg name='vctxt' type='void *' info='user data'/>
       <arg name='name' type='const char *' info='encoding name'/>
-      <arg name='output' type='int' info='true if output encoding, false if input'/>
+      <arg name='flags' type='xmlCharEncFlags' info='bit mask of flags'/>
       <arg name='out' type='xmlCharEncodingHandler **' info='pointer to resulting handler'/>
     </functype>
     <function name='xmlCharEncFirstLine' file='encoding' module='encoding'>
@@ -8368,10 +8374,10 @@ crash if you try to modify the tree)'/>
       <arg name='cur' type='xmlAttrPtr' info='the first attribute'/>
     </function>
     <function name='xmlCreateCharEncodingHandler' file='encoding' module='encoding'>
-      <info>Find or create a handler matching the encoding. The following converters are looked up in order:  - Built-in handler (UTF-8, UTF-16, ISO-8859-1, ASCII) - Custom implementation if provided - User-registered global handler (deprecated) - iconv if enabled - ICU if enabled  The handler must be closed with xmlCharEncCloseFunc.  If the encoding is UTF-8, a NULL handler and no error code will be returned.  Available since 2.14.0.</info>
+      <info>Find or create a handler matching the encoding. The following converters are looked up in order:  - Built-in handler (UTF-8, UTF-16, ISO-8859-1, ASCII) - Custom implementation if provided - User-registered global handler (deprecated) - iconv if enabled - ICU if enabled  The handler must be closed with xmlCharEncCloseFunc.  If the encoding is UTF-8, a NULL handler and no error code will be returned.  @flags can contain XML_ENC_INPUT, XML_ENC_OUTPUT or both.  Available since 2.14.0.</info>
       <return type='int' info='XML_ERR_OK, XML_ERR_UNSUPPORTED_ENCODING or another xmlParserErrors error code.'/>
       <arg name='name' type='const char *' info='a string describing the char encoding.'/>
-      <arg name='output' type='int' info='boolean, use handler for output'/>
+      <arg name='flags' type='xmlCharEncFlags' info='bit mask of flags'/>
       <arg name='impl' type='xmlCharEncConvImpl' info='a conversion implementation (optional)'/>
       <arg name='implCtxt' type='void *' info='user data for conversion implementation (optional)'/>
       <arg name='out' type='xmlCharEncodingHandler **' info='pointer to result'/>

encoding.c

@@ -239,12 +239,14 @@ static int nbCharEncodingHandler = 0;
 
 #ifdef LIBXML_ICONV_ENABLED
 static int
-xmlCharEncIconv(const char *name, xmlCharEncodingHandler **out);
+xmlCharEncIconv(const char *name, xmlCharEncFlags flags,
+                xmlCharEncodingHandler **out);
 #endif
 
 #ifdef LIBXML_ICU_ENABLED
 static int
-xmlCharEncUconv(const char *name, xmlCharEncodingHandler **out);
+xmlCharEncUconv(const char *name, xmlCharEncFlags flags,
+                xmlCharEncodingHandler **out);
 #endif
 
 /************************************************************************
@@ -794,7 +796,7 @@ free_handler:
  * xmlFindExtraHandler:
  * @norig:  name of the char encoding
  * @name:  potentially aliased name of the encoding
- * @output:  boolean, use handler for output
+ * @flags:  bit mask of flags
  * @impl:  a conversion implementation (optional)
  * @implCtxt:  user data for conversion implementation (optional)
  * @out:  pointer to resulting handler
@@ -804,7 +806,7 @@ free_handler:
  * Returns an xmlParserErrors error code.
  */
 static int
-xmlFindExtraHandler(const char *norig, const char *name, int output,
+xmlFindExtraHandler(const char *norig, const char *name, xmlCharEncFlags flags,
                     xmlCharEncConvImpl impl, void *implCtxt,
                     xmlCharEncodingHandler **out) {
     /*
@@ -814,7 +816,7 @@ xmlFindExtraHandler(const char *norig, const char *name, int output,
      * alias resolution.
      */
     if (impl != NULL)
-        return(impl(implCtxt, norig, output, out));
+        return(impl(implCtxt, norig, flags, out));
 
     /*
      * Deprecated
@@ -827,7 +829,8 @@ xmlFindExtraHandler(const char *norig, const char *name, int output,
 
             if (!xmlStrcasecmp((const xmlChar *) name,
                                (const xmlChar *) h->name)) {
-                if ((output ? h->output.func : h->input.func) != NULL) {
+                if ((((flags & XML_ENC_INPUT) == 0) || (h->input.func)) &&
+                    (((flags & XML_ENC_OUTPUT) == 0) || (h->output.func))) {
                     *out = h;
                     return(XML_ERR_OK);
                 }
@@ -837,7 +840,7 @@ xmlFindExtraHandler(const char *norig, const char *name, int output,
 
 #ifdef LIBXML_ICONV_ENABLED
     {
-        int ret = xmlCharEncIconv(name, out);
+        int ret = xmlCharEncIconv(name, flags, out);
 
         if (ret == XML_ERR_OK)
             return(XML_ERR_OK);
@@ -848,7 +851,7 @@ xmlFindExtraHandler(const char *norig, const char *name, int output,
 
 #ifdef LIBXML_ICU_ENABLED
     {
-        int ret = xmlCharEncUconv(name, out);
+        int ret = xmlCharEncUconv(name, flags, out);
 
         if (ret == XML_ERR_OK)
             return(XML_ERR_OK);
@@ -906,9 +909,15 @@ xmlLookupCharEncodingHandler(xmlCharEncoding enc,
         return(XML_ERR_OK);
     }
 
-    if (handler->name != NULL)
+    if (handler->name != NULL) {
-        return(xmlFindExtraHandler(handler->name, handler->name, 0,
+        xmlCharEncFlags flags = XML_ENC_INPUT;
+
+#ifdef LIBXML_OUTPUT_ENABLED
+        flags |= XML_ENC_OUTPUT;
+#endif
+        return(xmlFindExtraHandler(handler->name, handler->name, flags,
                                    NULL, NULL, out));
+    }
 
     return(XML_ERR_UNSUPPORTED_ENCODING);
 }
@@ -934,7 +943,7 @@ xmlGetCharEncodingHandler(xmlCharEncoding enc) {
 /**
  * xmlCreateCharEncodingHandler:
  * @name:  a string describing the char encoding.
- * @output:  boolean, use handler for output
+ * @flags:  bit mask of flags
  * @impl:  a conversion implementation (optional)
  * @implCtxt:  user data for conversion implementation (optional)
  * @out:  pointer to result
@@ -953,13 +962,15 @@ xmlGetCharEncodingHandler(xmlCharEncoding enc) {
  * If the encoding is UTF-8, a NULL handler and no error code will
  * be returned.
  *
+ * @flags can contain XML_ENC_INPUT, XML_ENC_OUTPUT or both.
+ *
  * Available since 2.14.0.
  *
  * Returns XML_ERR_OK, XML_ERR_UNSUPPORTED_ENCODING or another
  * xmlParserErrors error code.
  */
 int
-xmlCreateCharEncodingHandler(const char *name, int output,
+xmlCreateCharEncodingHandler(const char *name, xmlCharEncFlags flags,
                              xmlCharEncConvImpl impl, void *implCtxt,
                              xmlCharEncodingHandler **out) {
     const xmlCharEncodingHandler *handler;
@@ -970,7 +981,7 @@ xmlCreateCharEncodingHandler(const char *name, int output,
         return(XML_ERR_ARGUMENT);
     *out = NULL;
 
-    if (name == NULL)
+    if ((name == NULL) || (flags == 0))
         return(XML_ERR_ARGUMENT);
 
     norig = name;
@@ -986,13 +997,14 @@ xmlCreateCharEncodingHandler(const char *name, int output,
 
     if ((enc > 0) && ((size_t) enc < NUM_DEFAULT_HANDLERS)) {
         handler = &defaultHandlers[enc];
-        if ((output ? handler->output.func : handler->input.func) != NULL) {
+        if ((((flags & XML_ENC_INPUT) == 0) || (handler->input.func)) &&
+            (((flags & XML_ENC_OUTPUT) == 0) || (handler->output.func))) {
             *out = (xmlCharEncodingHandler *) handler;
             return(XML_ERR_OK);
         }
     }
 
-    return(xmlFindExtraHandler(norig, name, output, impl, implCtxt, out));
+    return(xmlFindExtraHandler(norig, name, flags, impl, implCtxt, out));
 }
 
 /**
@@ -1022,7 +1034,9 @@ xmlCreateCharEncodingHandler(const char *name, int output,
 int
 xmlOpenCharEncodingHandler(const char *name, int output,
                            xmlCharEncodingHandler **out) {
-    return(xmlCreateCharEncodingHandler(name, output, NULL, NULL, out));
+    xmlCharEncFlags flags = output ? XML_ENC_OUTPUT : XML_ENC_INPUT;
+
+    return(xmlCreateCharEncodingHandler(name, flags, NULL, NULL, out));
 }
 
 /**
@@ -1041,6 +1055,7 @@ xmlOpenCharEncodingHandler(const char *name, int output,
 xmlCharEncodingHandlerPtr
 xmlFindCharEncodingHandler(const char *name) {
     xmlCharEncodingHandler *ret;
+    xmlCharEncFlags flags;
 
     /*
      * This handler shouldn't be used, but we must return a non-NULL
@@ -1051,7 +1066,11 @@ xmlFindCharEncodingHandler(const char *name) {
         return((xmlCharEncodingHandlerPtr)
                 &defaultHandlers[XML_CHAR_ENCODING_UTF8]);
 
-    xmlOpenCharEncodingHandler(name, 0, &ret);
+    flags = XML_ENC_INPUT;
+#ifdef LIBXML_OUTPUT_ENABLED
+    flags |= XML_ENC_OUTPUT;
+#endif
+    xmlCreateCharEncodingHandler(name, flags, NULL, NULL, &ret);
     return(ret);
 }
 
@@ -1131,6 +1150,9 @@ static void
 xmlIconvFree(void *vctxt) {
     xmlIconvCtxt *ctxt = vctxt;
 
+    if (ctxt == NULL)
+        return;
+
     if (ctxt->cd != (iconv_t) -1)
         iconv_close(ctxt->cd);
 
@@ -1164,7 +1186,9 @@ xmlEncodingMatch(const char *name1, const char *name2) {
 #endif /* FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION */
 
 static int
-xmlCharEncIconv(const char *name, xmlCharEncodingHandler **out) {
+xmlCharEncIconv(const char *name, xmlCharEncFlags flags,
+                xmlCharEncodingHandler **out) {
+    xmlCharEncConvFunc inFunc = NULL, outFunc = NULL;
     xmlIconvCtxt *inputCtxt = NULL, *outputCtxt = NULL;
     iconv_t icv_in;
     iconv_t icv_out;
@@ -1217,6 +1241,7 @@ xmlCharEncIconv(const char *name, xmlCharEncodingHandler **out) {
     }
 #endif
 
+    if (flags & XML_ENC_INPUT) {
         inputCtxt = xmlMalloc(sizeof(xmlIconvCtxt));
         if (inputCtxt == NULL) {
             ret = XML_ERR_NO_MEMORY;
@@ -1236,6 +1261,10 @@ xmlCharEncIconv(const char *name, xmlCharEncodingHandler **out) {
         }
         inputCtxt->cd = icv_in;
 
+        inFunc = xmlIconvConvert;
+    }
+
+    if (flags & XML_ENC_OUTPUT) {
         outputCtxt = xmlMalloc(sizeof(xmlIconvCtxt));
         if (outputCtxt == NULL) {
             ret = XML_ERR_NO_MEMORY;
@@ -1255,9 +1284,11 @@ xmlCharEncIconv(const char *name, xmlCharEncodingHandler **out) {
         }
         outputCtxt->cd = icv_out;
 
-    return(xmlCharEncNewCustomHandler(name, xmlIconvConvert, xmlIconvConvert,
+        outFunc = xmlIconvConvert;
-                                      xmlIconvFree, inputCtxt, outputCtxt,
+    }
-                                      out));
+
+    return(xmlCharEncNewCustomHandler(name, inFunc, outFunc, xmlIconvFree,
+                                      inputCtxt, outputCtxt, out));
 
 error:
     if (inputCtxt != NULL)
@@ -1444,21 +1475,29 @@ xmlUconvFree(void *vctxt) {
 }
 
 static int
-xmlCharEncUconv(const char *name, xmlCharEncodingHandler **out) {
+xmlCharEncUconv(const char *name, xmlCharEncFlags flags,
+                xmlCharEncodingHandler **out) {
+    xmlCharEncConvFunc inFunc = NULL, outFunc = NULL;
     xmlUconvCtxt *ucv_in = NULL;
     xmlUconvCtxt *ucv_out = NULL;
     int ret;
 
+    if (flags & XML_ENC_INPUT) {
         ret = openIcuConverter(name, 1, &ucv_in);
         if (ret != 0)
             goto error;
+        inFunc = xmlUconvConvert;
+    }
+
+    if (flags & XML_ENC_OUTPUT) {
         ret = openIcuConverter(name, 0, &ucv_out);
         if (ret != 0)
             goto error;
+        outFunc = xmlUconvConvert;
+    }
 
-    return(xmlCharEncNewCustomHandler(name, xmlUconvConvert, xmlUconvConvert,
+    return(xmlCharEncNewCustomHandler(name, inFunc, outFunc, xmlUconvFree,
-                                      xmlUconvFree, ucv_in, ucv_out,
+                                      ucv_in, ucv_out, out));
-                                      out));
 
 error:
     if (ucv_in != NULL)

example/icu.c

@@ -171,22 +171,29 @@ icuConvCtxtDtor(void *vctxt) {
 }
 
 static int
-icuConvImpl(void *vctxt, const char *name, int output,
+icuConvImpl(void *vctxt, const char *name, xmlCharEncFlags flags,
-            xmlCharEncodingHandler **out) {
+            xmlCharEncodingHandler **result) {
+    xmlCharEncConvFunc inFunc = NULL, outFunc = NULL;
     myConvCtxt *inputCtxt = NULL;
     myConvCtxt *outputCtxt = NULL;
     int ret;
 
+    if (flags & XML_ENC_INPUT) {
         ret = icuOpen(name, 1, &inputCtxt);
         if (ret != 0)
             goto error;
+        inFunc = icuConvert;
+    }
+
+    if (flags & XML_ENC_OUTPUT) {
         ret = icuOpen(name, 0, &outputCtxt);
         if (ret != 0)
             goto error;
+        outFunc = icuConvert;
+    }
 
-    return xmlCharEncNewCustomHandler(name, icuConvert, icuConvert,
+    return xmlCharEncNewCustomHandler(name, inFunc, outFunc, icuConvCtxtDtor,
-                                      icuConvCtxtDtor, inputCtxt, outputCtxt,
+                                      inputCtxt, outputCtxt, result);
-                                      out);
 
 error:
     if (inputCtxt != NULL)

include/libxml/encoding.h

@@ -83,6 +83,11 @@ typedef enum {
     XML_CHAR_ENCODING_8859_16=	30 /* ISO-8859-16 */
 } xmlCharEncoding;
 
+typedef enum {
+    XML_ENC_INPUT = (1 << 0),
+    XML_ENC_OUTPUT = (1 << 1)
+} xmlCharEncFlags;
+
 /**
  * xmlCharEncodingInputFunc:
  * @out:  a pointer to an array of bytes to store the UTF-8 result
@@ -179,17 +184,19 @@ struct _xmlCharEncodingHandler {
  * xmlCharEncConvImpl:
  * @vctxt:  user data
  * @name:  encoding name
- * @output:  true if output encoding, false if input
+ * @flags:  bit mask of flags
  * @out:  pointer to resulting handler
  *
  * If this function returns XML_ERR_OK, it must fill the @out
  * pointer with an encoding handler. The handler can be obtained
  * from xmlCharEncNewCustomHandler.
  *
+ * @flags can contain XML_ENC_INPUT, XML_ENC_OUTPUT or both.
+ *
  * Returns an xmlParserErrors code.
  */
 typedef int
-(*xmlCharEncConvImpl)(void *vctxt, const char *name, int output,
+(*xmlCharEncConvImpl)(void *vctxt, const char *name, xmlCharEncFlags flags,
                       xmlCharEncodingHandler **out);
 
 /*
@@ -212,7 +219,7 @@ XMLPUBFUN int
 					 xmlCharEncodingHandlerPtr *out);
 XMLPUBFUN int
 	xmlCreateCharEncodingHandler	(const char *name,
-					 int output,
+					 xmlCharEncFlags flags,
 					 xmlCharEncConvImpl impl,
 					 void *implCtxt,
 					 xmlCharEncodingHandlerPtr *out);

parserInternals.c

@@ -1142,7 +1142,7 @@ xmlDetectEBCDIC(xmlParserCtxtPtr ctxt, xmlCharEncodingHandlerPtr *hout) {
      * To detect the EBCDIC code page, we convert the first 200 bytes
      * to IBM037 (EBCDIC-US) and try to find the encoding declaration.
      */
-    res = xmlCreateCharEncodingHandler("IBM037", /* output */ 0,
+    res = xmlCreateCharEncodingHandler("IBM037", XML_ENC_INPUT,
             ctxt->convImpl, ctxt->convCtxt, &handler);
     if (res != 0)
         return(res);
@@ -1188,7 +1188,7 @@ xmlDetectEBCDIC(xmlParserCtxtPtr ctxt, xmlCharEncodingHandlerPtr *hout) {
             out[i] = 0;
             xmlCharEncCloseFunc(handler);
             res = xmlCreateCharEncodingHandler((char *) out + start,
-                    /* output */ 0, ctxt->convImpl, ctxt->convCtxt,
+                    XML_ENC_INPUT, ctxt->convImpl, ctxt->convCtxt,
                     &handler);
             if (res != 0)
                 return(res);
@@ -1202,7 +1202,7 @@ done:
      * Encoding handlers are stateful, so we have to recreate them.
      */
     xmlCharEncCloseFunc(handler);
-    res = xmlCreateCharEncodingHandler("IBM037", /* output */ 0,
+    res = xmlCreateCharEncodingHandler("IBM037", XML_ENC_INPUT,
             ctxt->convImpl, ctxt->convCtxt, &handler);
     if (res != 0)
         return(res);
@@ -1265,7 +1265,7 @@ xmlSwitchInputEncodingName(xmlParserCtxtPtr ctxt, xmlParserInputPtr input,
     if (encoding == NULL)
         return(-1);
 
-    res = xmlCreateCharEncodingHandler(encoding, /* output */ 0,
+    res = xmlCreateCharEncodingHandler(encoding, XML_ENC_INPUT,
             ctxt->convImpl, ctxt->convCtxt, &handler);
     if (res == XML_ERR_UNSUPPORTED_ENCODING) {
         xmlWarningMsg(ctxt, XML_ERR_UNSUPPORTED_ENCODING,
@@ -1618,7 +1618,7 @@ xmlSetDeclaredEncoding(xmlParserCtxtPtr ctxt, xmlChar *encoding) {
          * declaration.
          */
         res = xmlCreateCharEncodingHandler((const char *) encoding,
-                /* output */ 0, ctxt->convImpl, ctxt->convCtxt, &handler);
+                XML_ENC_INPUT, ctxt->convImpl, ctxt->convCtxt, &handler);
         if (res != XML_ERR_OK) {
             xmlFatalErr(ctxt, res, (const char *) encoding);
             xmlFree(encoding);

testparser.c

@@ -1064,20 +1064,20 @@ rot13ConvCtxtDtor(void *vctxt) {
 }
 
 static int
-rot13ConvImpl(void *vctxt ATTRIBUTE_UNUSED, const char *name, int output,
+rot13ConvImpl(void *vctxt ATTRIBUTE_UNUSED, const char *name,
-              xmlCharEncodingHandler **out) {
+              xmlCharEncFlags flags, xmlCharEncodingHandler **out) {
     int *inputCtxt;
 
     if (strcmp(name, "rot13") != 0)
-        return xmlCreateCharEncodingHandler(name, output, NULL, NULL, out);
+        return xmlCreateCharEncodingHandler(name, flags, NULL, NULL, out);
 
-    if (output)
+    if (flags & XML_ENC_OUTPUT)
         return XML_ERR_UNSUPPORTED_ENCODING;
 
     inputCtxt = xmlMalloc(sizeof(*inputCtxt));
     *inputCtxt = 13;
 
-    return xmlCharEncNewCustomHandler(name, rot13Convert, rot13Convert,
+    return xmlCharEncNewCustomHandler(name, rot13Convert, NULL,
                                       rot13ConvCtxtDtor, inputCtxt, NULL,
                                       out);
 }

tools/gentest.py

@@ -523,6 +523,8 @@ for enum in enums:
     #
     if (name == None) or ((name not in argtypes) and (name not in rettypes)):
         continue;
+    if name == 'xmlCharEncFlags':
+        continue
     define = 0
 
     if (name in argtypes) and is_known_param_type(name) == 0: