From 2099441f329c1c4638d8d5364d69cf12206f43b8 Mon Sep 17 00:00:00 2001
From: Nick Wellnhofer <wellnhofer@aevum.de>
Date: Mon, 13 Mar 2023 17:51:13 +0100
Subject: [PATCH] parser: Stop calling xmlParserInputShrink

Introduce xmlParserShrink which takes a parser context to simplify error
handling.
---
 HTMLparser.c             |  2 +-
 include/private/parser.h |  2 ++
 parser.c                 | 13 ++-------
 parserInternals.c        | 57 ++++++++++++++++++++++++++++++++++++++++
 4 files changed, 62 insertions(+), 12 deletions(-)

diff --git a/HTMLparser.c b/HTMLparser.c
index a4e691f0..e9084950 100644
--- a/HTMLparser.c
+++ b/HTMLparser.c
@@ -295,7 +295,7 @@ htmlNodeInfoPop(htmlParserCtxtPtr ctxt)
 
 #define SHRINK if ((ctxt->input->cur - ctxt->input->base > 2 * INPUT_CHUNK) && \
 		   (ctxt->input->end - ctxt->input->cur < 2 * INPUT_CHUNK)) \
-	xmlParserInputShrink(ctxt->input)
+	xmlParserShrink(ctxt)
 
 #define GROW if ((ctxt->progressive == 0) &&				\
 		 (ctxt->input->end - ctxt->input->cur < INPUT_CHUNK))	\
diff --git a/include/private/parser.h b/include/private/parser.h
index 22e1314a..628f5a90 100644
--- a/include/private/parser.h
+++ b/include/private/parser.h
@@ -25,5 +25,7 @@ __xmlErrEncoding(xmlParserCtxtPtr ctxt, xmlParserErrors xmlerr,
                  const xmlChar *str2) LIBXML_ATTR_FORMAT(3,0);
 XML_HIDDEN int
 xmlParserGrow(xmlParserCtxtPtr ctxt);
+XML_HIDDEN int
+xmlParserShrink(xmlParserCtxtPtr ctxt);
 
 #endif /* XML_PARSER_H_PRIVATE__ */
diff --git a/parser.c b/parser.c
index 44331d11..53289c7a 100644
--- a/parser.c
+++ b/parser.c
@@ -2065,16 +2065,7 @@ static int spacePop(xmlParserCtxtPtr ctxt) {
 #define SHRINK if ((ctxt->progressive == 0) &&				\
 		   (ctxt->input->cur - ctxt->input->base > 2 * INPUT_CHUNK) && \
 		   (ctxt->input->end - ctxt->input->cur < 2 * INPUT_CHUNK)) \
-	xmlSHRINK (ctxt);
-
-static void xmlSHRINK (xmlParserCtxtPtr ctxt) {
-    /* Don't shrink memory buffers. */
-    if ((ctxt->input->buf) &&
-        ((ctxt->input->buf->encoder) || (ctxt->input->buf->readcallback)))
-        xmlParserInputShrink(ctxt->input);
-    if (*ctxt->input->cur == 0)
-        xmlParserGrow(ctxt);
-}
+	xmlParserShrink(ctxt);
 
 #define GROW if ((ctxt->progressive == 0) &&				\
 		 (ctxt->input->end - ctxt->input->cur < INPUT_CHUNK))	\
@@ -11484,7 +11475,7 @@ xmlParseTryOrFinish(xmlParserCtxtPtr ctxt, int terminate) {
 
     if ((ctxt->input != NULL) &&
         (ctxt->input->cur - ctxt->input->base > 4096)) {
-        xmlParserInputShrink(ctxt->input);
+        xmlParserShrink(ctxt);
     }
 
     while (ctxt->instate != XML_PARSER_EOF) {
diff --git a/parserInternals.c b/parserInternals.c
index 558186ae..bcf120a9 100644
--- a/parserInternals.c
+++ b/parserInternals.c
@@ -380,6 +380,63 @@ xmlParserInputGrow(xmlParserInputPtr in, int len) {
     return(ret);
 }
 
+/**
+ * xmlParserShrink:
+ * @ctxt:  an XML parser context
+ */
+int
+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);
+
+    used = in->cur - in->base;
+    /*
+     * Do not shrink on large buffers whose only a tiny fraction
+     * was consumed
+     */
+    if (used > INPUT_CHUNK) {
+	size_t res = xmlBufShrink(buf->buffer, used - LINE_LEN);
+
+	if (res > 0) {
+            used -= res;
+            if ((res > ULONG_MAX) ||
+                (in->consumed > ULONG_MAX - (unsigned long)res))
+                in->consumed = ULONG_MAX;
+            else
+                in->consumed += res;
+	}
+    }
+
+    if (xmlBufUse(buf->buffer) < INPUT_CHUNK)
+        ret = xmlParserInputBufferGrow(buf, INPUT_CHUNK);
+
+    in->base = xmlBufContent(buf->buffer);
+    if (in->base == NULL) {
+        in->base = BAD_CAST "";
+        in->cur = in->base;
+        in->end = in->base;
+        xmlErrMemory(ctxt, NULL);
+        return(-1);
+    }
+    in->cur = in->base + used;
+    in->end = xmlBufEnd(buf->buffer);
+
+    /* TODO: Get error code from xmlParserInputBufferGrow */
+    if (ret < 0) {
+        xmlErrInternal(ctxt, "Growing input buffer", NULL);
+        ctxt->instate = XML_PARSER_EOF;
+    }
+
+    return(ret);
+}
+
 /**
  * xmlParserInputShrink:
  * @in:  an XML parser input