parser: Support custom charset conversion implementations

Implement xmlCtxtSetCharEncConvImpl. I agree that the name is terrible.

include/libxml/parser.h

@@ -452,6 +452,9 @@ struct _xmlParserCtxt {
 
     xmlResourceLoader resourceLoader XML_DEPRECATED_MEMBER;
     void *resourceCtxt XML_DEPRECATED_MEMBER;
+
+    xmlCharEncConvImpl convImpl XML_DEPRECATED_MEMBER;
+    void *convCtxt XML_DEPRECATED_MEMBER;
 };
 
 /**
@@ -1438,6 +1441,10 @@ XMLPUBFUN void
 		xmlCtxtSetResourceLoader(xmlParserCtxtPtr ctxt,
 					 xmlResourceLoader loader,
 					 void *vctxt);
+XMLPUBFUN void
+		xmlCtxtSetCharEncConvImpl(xmlParserCtxtPtr ctxt,
+					 xmlCharEncConvImpl impl,
+					 void *vctxt);
 XMLPUBFUN void
 		xmlCtxtSetMaxAmplification(xmlParserCtxtPtr ctxt,
 					 unsigned maxAmpl);

include/libxml/parserInternals.h

@@ -376,7 +376,8 @@ XMLPUBFUN xmlParserInputPtr
 xmlInputCreateIO(const char *url, xmlInputReadCallback ioRead,
                  xmlInputCloseCallback ioClose, void *ioCtxt, int flags);
 XMLPUBFUN int
-xmlInputSetEncoding(xmlParserInputPtr input, const char *encoding);
+xmlInputSetEncodingHandler(xmlParserInputPtr input,
+                           xmlCharEncodingHandlerPtr handler);
 
 /**
  * Namespaces.

parserInternals.c

@@ -1078,9 +1078,34 @@ xmlCopyChar(int len ATTRIBUTE_UNUSED, xmlChar *out, int val) {
  *									*
  ************************************************************************/
 
+/**
+ * xmlCtxtSetCharEncConvImpl:
+ * @ctxt:  parser context
+ * @impl:  callback
+ * @vctxt:  user data
+ *
+ * Installs a custom implementation to convert between character
+ * encodings.
+ *
+ * This bypasses legacy feature like global encoding handlers or
+ * encoding aliases.
+ *
+ * Available since 2.14.0.
+ */
+void
+xmlCtxtSetCharEncConvImpl(xmlParserCtxtPtr ctxt, xmlCharEncConvImpl impl,
+                          void *vctxt) {
+    if (ctxt == NULL)
+        return;
+
+    ctxt->convImpl = impl;
+    ctxt->convCtxt = vctxt;
+}
+
 static int
-xmlDetectEBCDIC(xmlParserInputPtr input, xmlCharEncodingHandlerPtr *hout) {
+xmlDetectEBCDIC(xmlParserCtxtPtr ctxt, xmlCharEncodingHandlerPtr *hout) {
     xmlChar out[200];
+    xmlParserInputPtr input = ctxt->input;
     xmlCharEncodingHandlerPtr handler;
     int inlen, outlen, res, i;
 
@@ -1088,9 +1113,10 @@ xmlDetectEBCDIC(xmlParserInputPtr input, xmlCharEncodingHandlerPtr *hout) {
 
     /*
      * To detect the EBCDIC code page, we convert the first 200 bytes
-     * to EBCDIC-US and try to find the encoding declaration.
+     * to IBM037 (EBCDIC-US) and try to find the encoding declaration.
      */
-    res = xmlLookupCharEncodingHandler(XML_CHAR_ENCODING_EBCDIC, &handler);
+    res = xmlCreateCharEncodingHandler("IBM037", /* output */ 0,
+            ctxt->convImpl, ctxt->convCtxt, &handler);
     if (res != 0)
         return(res);
     outlen = sizeof(out) - 1;
@@ -1133,8 +1159,9 @@ xmlDetectEBCDIC(xmlParserInputPtr input, xmlCharEncodingHandlerPtr *hout) {
                 break;
             out[i] = 0;
             xmlCharEncCloseFunc(handler);
-            res = xmlOpenCharEncodingHandler((char *) out + start,
-                                             /* output */ 0, &handler);
+            res = xmlCreateCharEncodingHandler((char *) out + start,
+                    /* output */ 0, ctxt->convImpl, ctxt->convCtxt,
+                    &handler);
             if (res != 0)
                 return(res);
             *hout = handler;
@@ -1147,7 +1174,8 @@ done:
      * Encoding handlers are stateful, so we have to recreate them.
      */
     xmlCharEncCloseFunc(handler);
-    res = xmlLookupCharEncodingHandler(XML_CHAR_ENCODING_EBCDIC, &handler);
+    res = xmlCreateCharEncodingHandler("IBM037", /* output */ 0,
+            ctxt->convImpl, ctxt->convCtxt, &handler);
     if (res != 0)
         return(res);
     *hout = handler;
@@ -1184,7 +1212,7 @@ xmlSwitchEncoding(xmlParserCtxtPtr ctxt, xmlCharEncoding enc)
             res = 0;
             break;
         case XML_CHAR_ENCODING_EBCDIC:
-            res = xmlDetectEBCDIC(ctxt->input, &handler);
+            res = xmlDetectEBCDIC(ctxt, &handler);
             break;
         default:
             res = xmlLookupCharEncodingHandler(enc, &handler);
@@ -1224,7 +1252,8 @@ xmlSwitchInputEncodingName(xmlParserCtxtPtr ctxt, xmlParserInputPtr input,
     if (encoding == NULL)
         return(-1);
 
-    res = xmlOpenCharEncodingHandler(encoding, /* output */ 0, &handler);
+    res = xmlCreateCharEncodingHandler(encoding, /* output */ 0,
+            ctxt->convImpl, ctxt->convCtxt, &handler);
     if (res != 0) {
         xmlFatalErr(ctxt, res, encoding);
         return(-1);
@@ -1267,7 +1296,7 @@ xmlSwitchEncodingName(xmlParserCtxtPtr ctxt, const char *encoding) {
  *
  * Returns an xmlParserErrors code.
  */
-static int
+int
 xmlInputSetEncodingHandler(xmlParserInputPtr input,
                            xmlCharEncodingHandlerPtr handler) {
     int nbchars;
@@ -1341,33 +1370,6 @@ xmlInputSetEncodingHandler(xmlParserInputPtr input,
     return(XML_ERR_OK);
 }
 
-/**
- * xmlInputSetEncoding:
- * @input:  the input stream
- * @encoding:  the encoding name
- *
- * Use specified encoding to decode input data. This overrides the
- * encoding found in the XML declaration.
- *
- * Available since 2.14.0.
- *
- * Returns an xmlParserErrors code.
- */
-int
-xmlInputSetEncoding(xmlParserInputPtr input, const char *encoding) {
-    xmlCharEncodingHandlerPtr handler;
-    int res;
-
-    if (encoding == NULL)
-        return(XML_ERR_ARGUMENT);
-
-    res = xmlOpenCharEncodingHandler(encoding, /* output */ 0, &handler);
-    if (res != 0)
-        return(res);
-
-    return(xmlInputSetEncodingHandler(input, handler));
-}
-
 /**
  * xmlSwitchInputEncoding:
  * @ctxt:  the parser context, only for error reporting
@@ -2260,8 +2262,15 @@ xmlCheckHTTPInputInternal(xmlParserInputPtr input) {
     if ((xmlStrstr(BAD_CAST mime, BAD_CAST "/xml")) ||
         (xmlStrstr(BAD_CAST mime, BAD_CAST "+xml"))) {
         encoding = xmlNanoHTTPEncoding(input->buf->context);
-        if (encoding != NULL)
-            xmlInputSetEncoding(input, encoding);
+        if (encoding != NULL) {
+            xmlCharEncodingHandlerPtr handler;
+            int res;
+
+            res = xmlOpenCharEncodingHandler(encoding, /* output */ 0,
+                                             &handler);
+            if (res == 0)
+                xmlInputSetEncodingHandler(input, handler);
+        }
     }
 
     redir = xmlNanoHTTPRedir(input->buf->context);

testapi.c

@@ -15876,38 +15876,6 @@ test_xmlInputCreateUrl(void) {
 static int
 test_xmlInputSetEncoding(void) {
     int test_ret = 0;
-
-    int mem_base;
-    int ret_val;
-    xmlParserInputPtr input; /* the input stream */
-    int n_input;
-    const char * encoding; /* the encoding name */
-    int n_encoding;
-
-    for (n_input = 0;n_input < gen_nb_xmlParserInputPtr;n_input++) {
-    for (n_encoding = 0;n_encoding < gen_nb_const_char_ptr;n_encoding++) {
-        mem_base = xmlMemBlocks();
-        input = gen_xmlParserInputPtr(n_input, 0);
-        encoding = gen_const_char_ptr(n_encoding, 1);
-
-        ret_val = xmlInputSetEncoding(input, encoding);
-        desret_int(ret_val);
-        call_tests++;
-        des_xmlParserInputPtr(n_input, input, 0);
-        des_const_char_ptr(n_encoding, encoding, 1);
-        xmlResetLastError();
-        if (mem_base != xmlMemBlocks()) {
-            printf("Leak of %d blocks found in xmlInputSetEncoding",
-	           xmlMemBlocks() - mem_base);
-	    test_ret++;
-            printf(" %d", n_input);
-            printf(" %d", n_encoding);
-            printf("\n");
-        }
-    }
-    }
-    function_tests++;
-
     return(test_ret);
 }
 

testparser.c

@@ -518,6 +518,89 @@ testBuildRelativeUri(void) {
     return err;
 }
 
+static int charEncConvImplError;
+
+static int
+rot13Convert(void *vctxt, unsigned char *out, int *outlen,
+             const unsigned char *in, int *inlen) {
+    int *ctxt = vctxt;
+    int inSize = *inlen;
+    int outSize = *outlen;
+    int rot, i;
+
+    rot = *ctxt;
+
+    for (i = 0; i < inSize && i < outSize; i++) {
+        int c = in[i];
+
+        if (c >= 'A' && c <= 'Z')
+            c = 'A' + (c - 'A' + rot) % 26;
+        else if (c >= 'a' && c <= 'z')
+            c = 'a' + (c - 'a' + rot) % 26;
+
+        out[i] = c;
+    }
+
+    *inlen = i;
+    *outlen = i;
+
+    return XML_ENC_ERR_SUCCESS;
+}
+
+static void
+rot13ConvCtxtDtor(void *vctxt) {
+    xmlFree(vctxt);
+}
+
+static int
+rot13ConvImpl(void *vctxt ATTRIBUTE_UNUSED, const char *name,
+              xmlCharEncConverter *conv) {
+    int *inputCtxt;
+
+    if (strcmp(name, "rot13") != 0) {
+        fprintf(stderr, "rot13ConvImpl received wrong name\n");
+        charEncConvImplError = 1;
+
+        return XML_ERR_UNSUPPORTED_ENCODING;
+    }
+
+    conv->convert = rot13Convert;
+    conv->ctxtDtor = rot13ConvCtxtDtor;
+    
+    inputCtxt = xmlMalloc(sizeof(*inputCtxt));
+    *inputCtxt = 13;
+    conv->inputCtxt = inputCtxt;
+
+    return XML_ERR_OK;
+}
+
+static int
+testCharEncConvImpl(void) {
+    xmlParserCtxtPtr ctxt;
+    xmlDocPtr doc;
+    xmlNodePtr root;
+    int err = 0;
+
+    ctxt = xmlNewParserCtxt();
+    xmlCtxtSetCharEncConvImpl(ctxt, rot13ConvImpl, NULL);
+    charEncConvImplError = 0;
+    doc = xmlCtxtReadDoc(ctxt, BAD_CAST "<?kzy irefvba='1.0'?><qbp/>", NULL,
+                         "rot13", 0);
+    if (charEncConvImplError)
+        err = 1;
+    xmlFreeParserCtxt(ctxt);
+
+    root = xmlDocGetRootElement(doc);
+    if (root == NULL || strcmp((char *) root->name, "doc") != 0) {
+        fprintf(stderr, "testCharEncConvImpl failed\n");
+        err = 1;
+    }
+
+    xmlFreeDoc(doc);
+
+    return err;
+}
+
 int
 main(void) {
     int err = 0;
@@ -546,6 +629,7 @@ main(void) {
     err |= testWriterClose();
 #endif
     err |= testBuildRelativeUri();
+    err |= testCharEncConvImpl();
 
     return err;
 }