External DTD support. (but wrapped in #ifdef EXTERNAL_ENTITY_SUPPORT)

1999-06-15 23:20:01 +00:00 · 1999-06-15 23:20:01 +00:00 · 7e287d56b0
--- a/expat/xmlparse/xmlparse.c
+++ b/expat/xmlparse/xmlparse.c
@ -160,6 +160,9 @@ typedef struct {
  int complete;
  int standalone;
  const XML_Char *base;
+#ifdef EXTERNAL_ENTITY_SUPPORT
+  const XML_Char *systemId;
+#endif
  PREFIX defaultPrefix;
 } DTD;

@ -270,6 +273,9 @@ typedef struct {
  XML_CharacterDataHandler characterDataHandler;
  XML_ProcessingInstructionHandler processingInstructionHandler;
  XML_CommentHandler commentHandler;
+#ifdef EXTERNAL_ENTITY_SUPPORT
+  XML_ExternalDTDLoader externalDTDLoader;
+#endif
  XML_StartCdataSectionHandler startCdataSectionHandler;
  XML_EndCdataSectionHandler endCdataSectionHandler;
  XML_DefaultHandler defaultHandler;
@ -326,6 +332,9 @@ typedef struct {
 #define characterDataHandler (((Parser *)parser)->characterDataHandler)
 #define processingInstructionHandler (((Parser *)parser)->processingInstructionHandler)
 #define commentHandler (((Parser *)parser)->commentHandler)
+#ifdef EXTERNAL_ENTITY_SUPPORT
+#define externalDTDLoader (((Parser *)parser)->externalDTDLoader)
+#endif
 #define startCdataSectionHandler (((Parser *)parser)->startCdataSectionHandler)
 #define endCdataSectionHandler (((Parser *)parser)->endCdataSectionHandler)
 #define defaultHandler (((Parser *)parser)->defaultHandler)
@ -406,6 +415,9 @@ XML_Parser XML_ParserCreate(const XML_Char *encodingName)
  characterDataHandler = 0;
  processingInstructionHandler = 0;
  commentHandler = 0;
+#ifdef EXTERNAL_ENTITY_SUPPORT
+  externalDTDLoader = 0;
+#endif
  startCdataSectionHandler = 0;
  endCdataSectionHandler = 0;
  defaultHandler = 0;
@ -515,6 +527,9 @@ XML_Parser XML_ExternalEntityParserCreate(XML_Parser oldParser,
  XML_CharacterDataHandler oldCharacterDataHandler = characterDataHandler;
  XML_ProcessingInstructionHandler oldProcessingInstructionHandler = processingInstructionHandler;
  XML_CommentHandler oldCommentHandler = commentHandler;
+#ifdef EXTERNAL_ENTITY_SUPPORT
+  XML_ExternalDTDLoader oldExternalDTDLoader = externalDTDLoader;
+#endif
  XML_StartCdataSectionHandler oldStartCdataSectionHandler = startCdataSectionHandler;
  XML_EndCdataSectionHandler oldEndCdataSectionHandler = endCdataSectionHandler;
  XML_DefaultHandler oldDefaultHandler = defaultHandler;
@ -537,6 +552,9 @@ XML_Parser XML_ExternalEntityParserCreate(XML_Parser oldParser,
  characterDataHandler = oldCharacterDataHandler;
  processingInstructionHandler = oldProcessingInstructionHandler;
  commentHandler = oldCommentHandler;
+#ifdef EXTERNAL_ENTITY_SUPPORT
+  externalDTDLoader = oldExternalDTDLoader;
+#endif
  startCdataSectionHandler = oldStartCdataSectionHandler;
  endCdataSectionHandler = oldEndCdataSectionHandler;
  defaultHandler = oldDefaultHandler;
@ -661,6 +679,14 @@ void XML_SetCommentHandler(XML_Parser parser,
  commentHandler = handler;
 }

+#ifdef EXTERNAL_ENTITY_SUPPORT
+void XML_SetExternalDTDLoader(XML_Parser parser,
+		              XML_ExternalDTDLoader loader)
+{
+  externalDTDLoader = loader;
+}
+#endif
+
 void XML_SetCdataSectionHandler(XML_Parser parser,
 				XML_StartCdataSectionHandler start,
 			        XML_EndCdataSectionHandler end)
@ -1930,6 +1956,12 @@ prologProcessor(XML_Parser parser,
 		const char *end,
 		const char **nextPtr)
 {
+#ifdef EXTERNAL_ENTITY_SUPPORT
+  char * dtdData;
+  char * dtdNext;
+  int dtdLen;
+#endif
+
  for (;;) {
    const char *next;
    int tok = XmlPrologTok(encoding, s, end, &next);
@ -1964,8 +1996,36 @@ prologProcessor(XML_Parser parser,
      }
      break;
    case XML_ROLE_DOCTYPE_SYSTEM_ID:
+#ifdef EXTERNAL_ENTITY_SUPPORT
+      dtd.systemId = poolStoreString(&dtd.pool, encoding,
+	                              s + encoding->minBytesPerChar,
+	 		              next - encoding->minBytesPerChar);
+      if (!dtd.systemId)
+	  return XML_ERROR_NO_MEMORY;
+      poolFinish(&dtd.pool);
+#endif
      hadExternalDoctype = 1;
      break;
+#ifdef EXTERNAL_ENTITY_SUPPORT
+    case XML_ROLE_DOCTYPE_CLOSE:
+      if (hadExternalDoctype == 1) {
+        // if no loader specified, just default (non-hacked) processing
+        if (externalDTDLoader) {
+          XmlPrologStateHack(&prologState);
+          hadExternalDoctype = 0;
+
+          dtdLen = externalDTDLoader(dtd.base, dtd.systemId, &dtdData);
+          if (dtdLen < 0) {
+              return XML_ERROR_NO_MEMORY;
+          }
+          // XXX better error processing
+          dtdNext = dtdData + dtdLen;
+          prologProcessor(parser, dtdData, dtdNext, &dtdNext);
+          // XXX better error processing
+        }
+      }
+      break;
+#endif
    case XML_ROLE_DOCTYPE_PUBLIC_ID:
    case XML_ROLE_ENTITY_PUBLIC_ID:
      if (!XmlIsPublicId(encoding, s, next, &eventPtr))
--- a/expat/xmlparse/xmlparse.h
+++ b/expat/xmlparse/xmlparse.h
@ -107,6 +107,13 @@ typedef void (*XML_ProcessingInstructionHandler)(void *userData,
 /* data is 0 terminated */
 typedef void (*XML_CommentHandler)(void *userData, const XML_Char *data);

+//#define EXTERNAL_ENTITY_SUPPORT
+#ifdef EXTERNAL_ENTITY_SUPPORT
+typedef int (*XML_ExternalDTDLoader)(const XML_Char * base, 
+                                     const XML_Char * systemId,
+                                     char ** data);
+#endif
+
 typedef void (*XML_StartCdataSectionHandler)(void *userData);
 typedef void (*XML_EndCdataSectionHandler)(void *userData);

@ -271,6 +278,13 @@ void XMLPARSEAPI
 XML_SetCommentHandler(XML_Parser parser,
                      XML_CommentHandler handler);

+
+#ifdef EXTERNAL_ENTITY_SUPPORT
+void XMLPARSEAPI
+XML_SetExternalDTDLoader(XML_Parser parser,
+                         XML_ExternalDTDLoader loader);
+#endif
+
 void XMLPARSEAPI
 XML_SetCdataSectionHandler(XML_Parser parser,
 			   XML_StartCdataSectionHandler start,
--- a/expat/xmltok/xmlrole.c
+++ b/expat/xmltok/xmlrole.c
@ -1101,3 +1101,10 @@ void XmlPrologStateInit(PROLOG_STATE *state)
 {
  state->handler = prolog0;
 }
+
+#ifdef EXTERNAL_ENTITY_SUPPORT
+void XmlPrologStateHack(PROLOG_STATE *state)
+{
+  state->handler = prolog1;
+}
+#endif
--- a/expat/xmltok/xmlrole.h
+++ b/expat/xmltok/xmlrole.h
@ -91,6 +91,11 @@ typedef struct prolog_state {

 void XMLTOKAPI XmlPrologStateInit(PROLOG_STATE *);

+//#define EXTERNAL_ENTITY_SUPPORT
+#ifdef EXTERNAL_ENTITY_SUPPORT
+void XMLTOKAPI XmlPrologStateHack(PROLOG_STATE *);
+#endif
+
 #define XmlTokenRole(state, tok, ptr, end, enc) \
 (((state)->handler)(state, tok, ptr, end, enc))

--- a/htmlparser/src/nsExpatTokenizer.cpp
+++ b/htmlparser/src/nsExpatTokenizer.cpp
@ -30,6 +30,11 @@
 #include "nsIParser.h"
 #include "prlog.h"

+#ifdef EXTERNAL_ENTITY_SUPPORT
+#include "nsINetService.h"
+#include "nsIServiceManager.h"
+#endif
+
 /************************************************************************
  And now for the main class -- nsExpatTokenizer...
 ************************************************************************/
@ -115,6 +120,9 @@ void nsExpatTokenizer::SetupExpatCallbacks(void) {
    XML_SetUnparsedEntityDeclHandler(mExpatParser, HandleUnparsedEntityDecl);
    XML_SetNotationDeclHandler(mExpatParser, HandleNotationDecl);
    XML_SetExternalEntityRefHandler(mExpatParser, HandleExternalEntityRef);
+#ifdef EXTERNAL_ENTITY_SUPPORT
+    XML_SetExternalDTDLoader(mExpatParser, LoadExternalDTD);
+#endif
    XML_SetUnknownEncodingHandler(mExpatParser, HandleUnknownEncoding, NULL);    
  }
 }
@ -449,6 +457,54 @@ int nsExpatTokenizer::HandleExternalEntityRef(XML_Parser parser,
  return result;
 }

+#ifdef EXTERNAL_ENTITY_SUPPORT
+static NS_DEFINE_IID(kNetServiceCID, NS_NETSERVICE_CID);
+static NS_DEFINE_IID(kINetServiceIID, NS_INETSERVICE_IID);
+
+int nsExpatTokenizer::LoadExternalDTD(const XML_Char * base, 
+                                      const XML_Char * systemId,
+                                      char ** data)
+{
+  // XXX free resources when not needed anymore & generally clean this code
+
+  nsINetService * netService = nsnull;
+  nsresult res = nsServiceManager::GetService(kNetServiceCID,
+    kINetServiceIID, (nsISupports**) &netService);
+  if (NS_FAILED(res)) {
+    return -1;
+  }
+
+  // XXX try to compose this using base url
+  nsIURL * url = nsnull;
+  nsString * s = new nsString((PRUnichar *)systemId);
+  res = netService->CreateURL(&url, *s);
+  if (NS_FAILED(res)) {
+    return -1;
+  }
+
+  nsIInputStream * in = nsnull;
+  res = netService->OpenBlockingStream(url, nsnull, &in);
+  if (NS_FAILED(res)) {
+    NS_RELEASE(url);
+    return -1;
+  }
+
+  // XXX yuck!
+  char * buff = (char *) malloc(1000);
+  PRUnichar * buff2 = (PRUnichar *) malloc(1000);
+  PRUint32 len = 0;
+  PRUint32 j;
+  res = in->Read(buff, 1000, &len);
+  for (j=0 ; j<len; j++) buff2[j] = (PRUnichar) buff[j];
+
+  NS_RELEASE(url);
+  NS_RELEASE(in);
+
+  *data = (char *)buff2;
+  return len*2;
+}
+#endif
+
 int nsExpatTokenizer::HandleUnknownEncoding(void *encodingHandlerData,
                                       const XML_Char *name,
                                       XML_Encoding *info) {
--- a/htmlparser/src/nsExpatTokenizer.h
+++ b/htmlparser/src/nsExpatTokenizer.h
@ -104,6 +104,12 @@ protected:
      const XML_Char *base,
      const XML_Char *systemId,
      const XML_Char *publicId);
+//#define EXTERNAL_ENTITY_SUPPORT
+#ifdef EXTERNAL_ENTITY_SUPPORT
+    static int LoadExternalDTD(const XML_Char * base, 
+      const XML_Char * systemId, 
+      char ** data);
+#endif
    static int HandleUnknownEncoding(void *encodingHandlerData,
      const XML_Char *name,
      XML_Encoding *info);
--- a/parser/expat/lib/xmlparse.c
+++ b/parser/expat/lib/xmlparse.c
@ -160,6 +160,9 @@ typedef struct {
  int complete;
  int standalone;
  const XML_Char *base;
+#ifdef EXTERNAL_ENTITY_SUPPORT
+  const XML_Char *systemId;
+#endif
  PREFIX defaultPrefix;
 } DTD;

@ -270,6 +273,9 @@ typedef struct {
  XML_CharacterDataHandler characterDataHandler;
  XML_ProcessingInstructionHandler processingInstructionHandler;
  XML_CommentHandler commentHandler;
+#ifdef EXTERNAL_ENTITY_SUPPORT
+  XML_ExternalDTDLoader externalDTDLoader;
+#endif
  XML_StartCdataSectionHandler startCdataSectionHandler;
  XML_EndCdataSectionHandler endCdataSectionHandler;
  XML_DefaultHandler defaultHandler;
@ -326,6 +332,9 @@ typedef struct {
 #define characterDataHandler (((Parser *)parser)->characterDataHandler)
 #define processingInstructionHandler (((Parser *)parser)->processingInstructionHandler)
 #define commentHandler (((Parser *)parser)->commentHandler)
+#ifdef EXTERNAL_ENTITY_SUPPORT
+#define externalDTDLoader (((Parser *)parser)->externalDTDLoader)
+#endif
 #define startCdataSectionHandler (((Parser *)parser)->startCdataSectionHandler)
 #define endCdataSectionHandler (((Parser *)parser)->endCdataSectionHandler)
 #define defaultHandler (((Parser *)parser)->defaultHandler)
@ -406,6 +415,9 @@ XML_Parser XML_ParserCreate(const XML_Char *encodingName)
  characterDataHandler = 0;
  processingInstructionHandler = 0;
  commentHandler = 0;
+#ifdef EXTERNAL_ENTITY_SUPPORT
+  externalDTDLoader = 0;
+#endif
  startCdataSectionHandler = 0;
  endCdataSectionHandler = 0;
  defaultHandler = 0;
@ -515,6 +527,9 @@ XML_Parser XML_ExternalEntityParserCreate(XML_Parser oldParser,
  XML_CharacterDataHandler oldCharacterDataHandler = characterDataHandler;
  XML_ProcessingInstructionHandler oldProcessingInstructionHandler = processingInstructionHandler;
  XML_CommentHandler oldCommentHandler = commentHandler;
+#ifdef EXTERNAL_ENTITY_SUPPORT
+  XML_ExternalDTDLoader oldExternalDTDLoader = externalDTDLoader;
+#endif
  XML_StartCdataSectionHandler oldStartCdataSectionHandler = startCdataSectionHandler;
  XML_EndCdataSectionHandler oldEndCdataSectionHandler = endCdataSectionHandler;
  XML_DefaultHandler oldDefaultHandler = defaultHandler;
@ -537,6 +552,9 @@ XML_Parser XML_ExternalEntityParserCreate(XML_Parser oldParser,
  characterDataHandler = oldCharacterDataHandler;
  processingInstructionHandler = oldProcessingInstructionHandler;
  commentHandler = oldCommentHandler;
+#ifdef EXTERNAL_ENTITY_SUPPORT
+  externalDTDLoader = oldExternalDTDLoader;
+#endif
  startCdataSectionHandler = oldStartCdataSectionHandler;
  endCdataSectionHandler = oldEndCdataSectionHandler;
  defaultHandler = oldDefaultHandler;
@ -661,6 +679,14 @@ void XML_SetCommentHandler(XML_Parser parser,
  commentHandler = handler;
 }

+#ifdef EXTERNAL_ENTITY_SUPPORT
+void XML_SetExternalDTDLoader(XML_Parser parser,
+		              XML_ExternalDTDLoader loader)
+{
+  externalDTDLoader = loader;
+}
+#endif
+
 void XML_SetCdataSectionHandler(XML_Parser parser,
 				XML_StartCdataSectionHandler start,
 			        XML_EndCdataSectionHandler end)
@ -1930,6 +1956,12 @@ prologProcessor(XML_Parser parser,
 		const char *end,
 		const char **nextPtr)
 {
+#ifdef EXTERNAL_ENTITY_SUPPORT
+  char * dtdData;
+  char * dtdNext;
+  int dtdLen;
+#endif
+
  for (;;) {
    const char *next;
    int tok = XmlPrologTok(encoding, s, end, &next);
@ -1964,8 +1996,36 @@ prologProcessor(XML_Parser parser,
      }
      break;
    case XML_ROLE_DOCTYPE_SYSTEM_ID:
+#ifdef EXTERNAL_ENTITY_SUPPORT
+      dtd.systemId = poolStoreString(&dtd.pool, encoding,
+	                              s + encoding->minBytesPerChar,
+	 		              next - encoding->minBytesPerChar);
+      if (!dtd.systemId)
+	  return XML_ERROR_NO_MEMORY;
+      poolFinish(&dtd.pool);
+#endif
      hadExternalDoctype = 1;
      break;
+#ifdef EXTERNAL_ENTITY_SUPPORT
+    case XML_ROLE_DOCTYPE_CLOSE:
+      if (hadExternalDoctype == 1) {
+        // if no loader specified, just default (non-hacked) processing
+        if (externalDTDLoader) {
+          XmlPrologStateHack(&prologState);
+          hadExternalDoctype = 0;
+
+          dtdLen = externalDTDLoader(dtd.base, dtd.systemId, &dtdData);
+          if (dtdLen < 0) {
+              return XML_ERROR_NO_MEMORY;
+          }
+          // XXX better error processing
+          dtdNext = dtdData + dtdLen;
+          prologProcessor(parser, dtdData, dtdNext, &dtdNext);
+          // XXX better error processing
+        }
+      }
+      break;
+#endif
    case XML_ROLE_DOCTYPE_PUBLIC_ID:
    case XML_ROLE_ENTITY_PUBLIC_ID:
      if (!XmlIsPublicId(encoding, s, next, &eventPtr))
--- a/parser/expat/lib/xmlparse.h
+++ b/parser/expat/lib/xmlparse.h
@ -107,6 +107,13 @@ typedef void (*XML_ProcessingInstructionHandler)(void *userData,
 /* data is 0 terminated */
 typedef void (*XML_CommentHandler)(void *userData, const XML_Char *data);

+//#define EXTERNAL_ENTITY_SUPPORT
+#ifdef EXTERNAL_ENTITY_SUPPORT
+typedef int (*XML_ExternalDTDLoader)(const XML_Char * base, 
+                                     const XML_Char * systemId,
+                                     char ** data);
+#endif
+
 typedef void (*XML_StartCdataSectionHandler)(void *userData);
 typedef void (*XML_EndCdataSectionHandler)(void *userData);

@ -271,6 +278,13 @@ void XMLPARSEAPI
 XML_SetCommentHandler(XML_Parser parser,
                      XML_CommentHandler handler);

+
+#ifdef EXTERNAL_ENTITY_SUPPORT
+void XMLPARSEAPI
+XML_SetExternalDTDLoader(XML_Parser parser,
+                         XML_ExternalDTDLoader loader);
+#endif
+
 void XMLPARSEAPI
 XML_SetCdataSectionHandler(XML_Parser parser,
 			   XML_StartCdataSectionHandler start,
--- a/parser/expat/lib/xmlrole.c
+++ b/parser/expat/lib/xmlrole.c
@ -1101,3 +1101,10 @@ void XmlPrologStateInit(PROLOG_STATE *state)
 {
  state->handler = prolog0;
 }
+
+#ifdef EXTERNAL_ENTITY_SUPPORT
+void XmlPrologStateHack(PROLOG_STATE *state)
+{
+  state->handler = prolog1;
+}
+#endif
--- a/parser/expat/lib/xmlrole.h
+++ b/parser/expat/lib/xmlrole.h
@ -91,6 +91,11 @@ typedef struct prolog_state {

 void XMLTOKAPI XmlPrologStateInit(PROLOG_STATE *);

+//#define EXTERNAL_ENTITY_SUPPORT
+#ifdef EXTERNAL_ENTITY_SUPPORT
+void XMLTOKAPI XmlPrologStateHack(PROLOG_STATE *);
+#endif
+
 #define XmlTokenRole(state, tok, ptr, end, enc) \
 (((state)->handler)(state, tok, ptr, end, enc))

--- a/parser/htmlparser/src/nsExpatTokenizer.cpp
+++ b/parser/htmlparser/src/nsExpatTokenizer.cpp
@ -30,6 +30,11 @@
 #include "nsIParser.h"
 #include "prlog.h"

+#ifdef EXTERNAL_ENTITY_SUPPORT
+#include "nsINetService.h"
+#include "nsIServiceManager.h"
+#endif
+
 /************************************************************************
  And now for the main class -- nsExpatTokenizer...
 ************************************************************************/
@ -115,6 +120,9 @@ void nsExpatTokenizer::SetupExpatCallbacks(void) {
    XML_SetUnparsedEntityDeclHandler(mExpatParser, HandleUnparsedEntityDecl);
    XML_SetNotationDeclHandler(mExpatParser, HandleNotationDecl);
    XML_SetExternalEntityRefHandler(mExpatParser, HandleExternalEntityRef);
+#ifdef EXTERNAL_ENTITY_SUPPORT
+    XML_SetExternalDTDLoader(mExpatParser, LoadExternalDTD);
+#endif
    XML_SetUnknownEncodingHandler(mExpatParser, HandleUnknownEncoding, NULL);    
  }
 }
@ -449,6 +457,54 @@ int nsExpatTokenizer::HandleExternalEntityRef(XML_Parser parser,
  return result;
 }

+#ifdef EXTERNAL_ENTITY_SUPPORT
+static NS_DEFINE_IID(kNetServiceCID, NS_NETSERVICE_CID);
+static NS_DEFINE_IID(kINetServiceIID, NS_INETSERVICE_IID);
+
+int nsExpatTokenizer::LoadExternalDTD(const XML_Char * base, 
+                                      const XML_Char * systemId,
+                                      char ** data)
+{
+  // XXX free resources when not needed anymore & generally clean this code
+
+  nsINetService * netService = nsnull;
+  nsresult res = nsServiceManager::GetService(kNetServiceCID,
+    kINetServiceIID, (nsISupports**) &netService);
+  if (NS_FAILED(res)) {
+    return -1;
+  }
+
+  // XXX try to compose this using base url
+  nsIURL * url = nsnull;
+  nsString * s = new nsString((PRUnichar *)systemId);
+  res = netService->CreateURL(&url, *s);
+  if (NS_FAILED(res)) {
+    return -1;
+  }
+
+  nsIInputStream * in = nsnull;
+  res = netService->OpenBlockingStream(url, nsnull, &in);
+  if (NS_FAILED(res)) {
+    NS_RELEASE(url);
+    return -1;
+  }
+
+  // XXX yuck!
+  char * buff = (char *) malloc(1000);
+  PRUnichar * buff2 = (PRUnichar *) malloc(1000);
+  PRUint32 len = 0;
+  PRUint32 j;
+  res = in->Read(buff, 1000, &len);
+  for (j=0 ; j<len; j++) buff2[j] = (PRUnichar) buff[j];
+
+  NS_RELEASE(url);
+  NS_RELEASE(in);
+
+  *data = (char *)buff2;
+  return len*2;
+}
+#endif
+
 int nsExpatTokenizer::HandleUnknownEncoding(void *encodingHandlerData,
                                       const XML_Char *name,
                                       XML_Encoding *info) {
--- a/parser/htmlparser/src/nsExpatTokenizer.h
+++ b/parser/htmlparser/src/nsExpatTokenizer.h
@ -104,6 +104,12 @@ protected:
      const XML_Char *base,
      const XML_Char *systemId,
      const XML_Char *publicId);
+//#define EXTERNAL_ENTITY_SUPPORT
+#ifdef EXTERNAL_ENTITY_SUPPORT
+    static int LoadExternalDTD(const XML_Char * base, 
+      const XML_Char * systemId, 
+      char ** data);
+#endif
    static int HandleUnknownEncoding(void *encodingHandlerData,
      const XML_Char *name,
      XML_Encoding *info);