parser: Rework shrinking of input buffers

Don't try to grow the input buffer in xmlParserShrink. This makes sure
that no memory allocations are made and the function always succeeds.

Remove unnecessary invocations of SHRINK. Invoke SHRINK at the end of
DTD parsing loops.

Shrink before growing.

HTMLparser.c

@@ -3100,7 +3100,6 @@ htmlParseScript(htmlParserCtxtPtr ctxt) {
     int nbchar = 0;
     int cur,l;
 
-    SHRINK;
     cur = CUR_CHAR(l);
     while (cur != 0) {
 	if ((cur == '<') && (NXT(1) == '/')) {
@@ -3358,7 +3357,6 @@ htmlParsePI(htmlParserCtxtPtr ctxt) {
 	 * this is a Processing Instruction.
 	 */
 	SKIP(2);
-	SHRINK;
 
 	/*
 	 * Parse the target name and check for special support like
@@ -3481,7 +3479,6 @@ htmlParseComment(htmlParserCtxtPtr ctxt) {
 
     state = ctxt->instate;
     ctxt->instate = XML_PARSER_COMMENT;
-    SHRINK;
     SKIP(4);
     buf = (xmlChar *) xmlMallocAtomic(size);
     if (buf == NULL) {
@@ -4477,8 +4474,8 @@ htmlParseContent(htmlParserCtxtPtr ctxt) {
             htmlParseCharData(ctxt);
         }
 
-        GROW;
         SHRINK;
+        GROW;
     }
     if (currentNode != NULL) xmlFree(currentNode);
 }
@@ -4920,8 +4917,8 @@ htmlParseContentInternal(htmlParserCtxtPtr ctxt) {
             htmlParseCharData(ctxt);
         }
 
-        GROW;
         SHRINK;
+        GROW;
     }
     if (currentNode != NULL) xmlFree(currentNode);
 }

include/private/parser.h

@@ -27,7 +27,7 @@ XML_HIDDEN void
 xmlHaltParser(xmlParserCtxtPtr ctxt);
 XML_HIDDEN int
 xmlParserGrow(xmlParserCtxtPtr ctxt);
-XML_HIDDEN int
+XML_HIDDEN void
 xmlParserShrink(xmlParserCtxtPtr ctxt);
 
 #endif /* XML_PARSER_H_PRIVATE__ */

parser.c

@@ -4183,7 +4183,6 @@ xmlParseSystemLiteral(xmlParserCtxtPtr ctxt) {
     xmlChar stop;
     int state = ctxt->instate;
 
-    SHRINK;
     if (RAW == '"') {
         NEXT;
 	stop = '"';
@@ -4265,7 +4264,6 @@ xmlParsePubidLiteral(xmlParserCtxtPtr ctxt) {
     xmlChar stop;
     xmlParserInputState oldstate = ctxt->instate;
 
-    SHRINK;
     if (RAW == '"') {
         NEXT;
 	stop = '"';
@@ -4387,7 +4385,6 @@ xmlParseCharData(xmlParserCtxtPtr ctxt, ATTRIBUTE_UNUSED int cdata) {
     int col = ctxt->input->col;
     int ccol;
 
-    SHRINK;
     GROW;
     /*
      * Accelerated common case where input don't need to be
@@ -4533,7 +4530,6 @@ xmlParseCharDataComplex(xmlParserCtxtPtr ctxt) {
     int nbchar = 0;
     int cur, l;
 
-    SHRINK;
     cur = CUR_CHAR(l);
     while ((cur != '<') && /* checked */
            (cur != '&') &&
@@ -4629,8 +4625,6 @@ xmlChar *
 xmlParseExternalID(xmlParserCtxtPtr ctxt, xmlChar **publicID, int strict) {
     xmlChar *URI = NULL;
 
-    SHRINK;
-
     *publicID = NULL;
     if (CMP6(CUR_PTR, 'S', 'Y', 'S', 'T', 'E', 'M')) {
         SKIP(6);
@@ -4847,7 +4841,6 @@ xmlParseComment(xmlParserCtxtPtr ctxt) {
     ctxt->instate = XML_PARSER_COMMENT;
     inputid = ctxt->input->id;
     SKIP(2);
-    SHRINK;
     GROW;
 
     /*
@@ -5133,7 +5126,6 @@ xmlParsePI(xmlParserCtxtPtr ctxt) {
 	 * this is a Processing Instruction.
 	 */
 	SKIP(2);
-	SHRINK;
 
 	/*
 	 * Parse the target name and check for special support like
@@ -5272,7 +5264,6 @@ xmlParseNotationDecl(xmlParserCtxtPtr ctxt) {
 
     if (CMP8(CUR_PTR, 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N')) {
 	int inputid = ctxt->input->id;
-	SHRINK;
 	SKIP(8);
 	if (SKIP_BLANKS == 0) {
 	    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
@@ -5360,7 +5351,6 @@ xmlParseEntityDecl(xmlParserCtxtPtr ctxt) {
     /* GROW; done in the caller */
     if (CMP6(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'Y')) {
 	int inputid = ctxt->input->id;
-	SHRINK;
 	SKIP(6);
 	if (SKIP_BLANKS == 0) {
 	    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
@@ -5684,7 +5674,6 @@ xmlParseNotationType(xmlParserCtxtPtr ctxt) {
 	xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_STARTED, NULL);
 	return(NULL);
     }
-    SHRINK;
     do {
         NEXT;
 	SKIP_BLANKS;
@@ -5756,7 +5745,6 @@ xmlParseEnumerationType(xmlParserCtxtPtr ctxt) {
 	xmlFatalErr(ctxt, XML_ERR_ATTLIST_NOT_STARTED, NULL);
 	return(NULL);
     }
-    SHRINK;
     do {
         NEXT;
 	SKIP_BLANKS;
@@ -5885,7 +5873,6 @@ xmlParseEnumeratedType(xmlParserCtxtPtr ctxt, xmlEnumerationPtr *tree) {
  */
 int
 xmlParseAttributeType(xmlParserCtxtPtr ctxt, xmlEnumerationPtr *tree) {
-    SHRINK;
     if (CMP5(CUR_PTR, 'C', 'D', 'A', 'T', 'A')) {
 	SKIP(5);
 	return(XML_ATTRIBUTE_CDATA);
@@ -6070,7 +6057,6 @@ xmlParseElementMixedContentDecl(xmlParserCtxtPtr ctxt, int inputchk) {
     if (CMP7(CUR_PTR, '#', 'P', 'C', 'D', 'A', 'T', 'A')) {
 	SKIP(7);
 	SKIP_BLANKS;
-	SHRINK;
 	if (RAW == ')') {
 	    if (ctxt->input->id != inputchk) {
 		xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY,
@@ -6242,7 +6228,6 @@ xmlParseElementChildrenContentDeclPriv(xmlParserCtxtPtr ctxt, int inputchk,
 	GROW;
     }
     SKIP_BLANKS;
-    SHRINK;
     while ((RAW != ')') && (ctxt->instate != XML_PARSER_EOF)) {
         /*
 	 * Each loop we parse one separator and one element.
@@ -6787,6 +6772,7 @@ xmlParseConditionalSections(xmlParserCtxtPtr ctxt) {
             break;
 
         SKIP_BLANKS;
+        SHRINK;
         GROW;
     }
 
@@ -7018,6 +7004,7 @@ xmlParseExternalSubset(xmlParserCtxtPtr ctxt, const xmlChar *ExternalID,
             return;
         }
         SKIP_BLANKS;
+        SHRINK;
     }
 
     if (RAW != 0) {
@@ -8343,6 +8330,8 @@ xmlParseInternalSubset(xmlParserCtxtPtr ctxt) {
                 return;
             }
 	    SKIP_BLANKS;
+            SHRINK;
+            GROW;
 	}
 	if (RAW == ']') {
 	    NEXT;
@@ -9229,14 +9218,6 @@ xmlParseStartTag2(xmlParserCtxtPtr ctxt, const xmlChar **pref,
     if (RAW != '<') return(NULL);
     NEXT1;
 
-    /*
-     * NOTE: it is crucial with the SAX2 API to never call SHRINK beyond that
-     *       point since the attribute values may be stored as pointers to
-     *       the buffer and calling SHRINK would destroy them !
-     *       The Shrinking is only possible once the full set of attribute
-     *       callbacks have been done.
-     */
-    SHRINK;
     cur = ctxt->input->cur - ctxt->input->base;
     inputid = ctxt->input->id;
     nbatts = 0;
@@ -9881,8 +9862,8 @@ xmlParseContentInternal(xmlParserCtxtPtr ctxt) {
 	    xmlParseCharData(ctxt, 0);
 	}
 
-	GROW;
 	SHRINK;
+	GROW;
     }
 }
 

parserInternals.c

@@ -409,17 +409,16 @@ xmlParserInputGrow(xmlParserInputPtr in, int len) {
  * xmlParserShrink:
  * @ctxt:  an XML parser context
  */
-int
+void
 xmlParserShrink(xmlParserCtxtPtr ctxt) {
     xmlParserInputPtr in = ctxt->input;
     xmlParserInputBufferPtr buf = in->buf;
     size_t used;
-    int ret = 0;
 
     /* Don't shrink memory buffers. */
     if ((buf == NULL) ||
         ((buf->encoder == NULL) && (buf->readcallback == NULL)))
-        return(0);
+        return;
 
     used = in->cur - in->base;
     /*
@@ -439,18 +438,7 @@ xmlParserShrink(xmlParserCtxtPtr ctxt) {
 	}
     }
 
-    if (xmlBufUse(buf->buffer) < INPUT_CHUNK)
-        ret = xmlParserInputBufferGrow(buf, INPUT_CHUNK);
-
     xmlBufSetInputBaseCur(buf->buffer, in, 0, used);
-
-    /* TODO: Get error code from xmlParserInputBufferGrow */
-    if (ret < 0) {
-        xmlErrInternal(ctxt, "Growing input buffer", NULL);
-        xmlHaltParser(ctxt);
-    }
-
-    return(ret);
 }
 
 /**