From a5c4a6efe77f6dd6e0a092db9357b21602eedd31 Mon Sep 17 00:00:00 2001
From: Nick Wellnhofer <wellnhofer@aevum.de>
Date: Fri, 28 Mar 2025 16:31:14 +0100
Subject: [PATCH] parser: Fix XML_PARSE_NOBLANKS dropping non-whitespace text

Regressed with 1f5b5371.

Fixes #884.
---
 parser.c     | 13 +++++++------
 testparser.c | 34 ++++++++++++++++++++++++++++++++++
 2 files changed, 41 insertions(+), 6 deletions(-)

diff --git a/parser.c b/parser.c
index aacaf1f8..d8d590ff 100644
--- a/parser.c
+++ b/parser.c
@@ -4778,7 +4778,8 @@ static const unsigned char test_char_data[256] = {
 };
 
 static void
-xmlCharacters(xmlParserCtxtPtr ctxt, const xmlChar *buf, int size) {
+xmlCharacters(xmlParserCtxtPtr ctxt, const xmlChar *buf, int size,
+              int isBlank) {
     int checkBlanks;
 
     if ((ctxt->sax == NULL) || (ctxt->disableSAX))
@@ -4793,7 +4794,7 @@ xmlCharacters(xmlParserCtxtPtr ctxt, const xmlChar *buf, int size) {
      * essentially unusable.
      */
     if ((checkBlanks) &&
-        (areBlanks(ctxt, buf, size, 1))) {
+        (areBlanks(ctxt, buf, size, isBlank))) {
         if ((ctxt->sax->ignorableWhitespace != NULL) &&
             (ctxt->keepBlanks))
             ctxt->sax->ignorableWhitespace(ctxt->userData, buf, size);
@@ -4855,7 +4856,7 @@ get_more_space:
                 const xmlChar *tmp = ctxt->input->cur;
                 ctxt->input->cur = in;
 
-                xmlCharacters(ctxt, tmp, nbchar);
+                xmlCharacters(ctxt, tmp, nbchar, 1);
             }
             return;
         }
@@ -4891,7 +4892,7 @@ get_more:
             const xmlChar *tmp = ctxt->input->cur;
             ctxt->input->cur = in;
 
-            xmlCharacters(ctxt, tmp, nbchar);
+            xmlCharacters(ctxt, tmp, nbchar, 0);
 
             line = ctxt->input->line;
             col = ctxt->input->col;
@@ -4958,7 +4959,7 @@ xmlParseCharDataComplex(xmlParserCtxtPtr ctxt, int partial) {
 	if (nbchar >= XML_PARSER_BIG_BUFFER_SIZE) {
 	    buf[nbchar] = 0;
 
-            xmlCharacters(ctxt, buf, nbchar);
+            xmlCharacters(ctxt, buf, nbchar, 0);
 	    nbchar = 0;
             SHRINK;
 	}
@@ -4967,7 +4968,7 @@ xmlParseCharDataComplex(xmlParserCtxtPtr ctxt, int partial) {
     if (nbchar != 0) {
         buf[nbchar] = 0;
 
-        xmlCharacters(ctxt, buf, nbchar);
+        xmlCharacters(ctxt, buf, nbchar, 0);
     }
     /*
      * cur == 0 can mean
diff --git a/testparser.c b/testparser.c
index 5cca7b67..85fee9b4 100644
--- a/testparser.c
+++ b/testparser.c
@@ -255,6 +255,39 @@ testCtxtParseContent(void) {
 
     return err;
 }
+
+static int
+testNoBlanks(void) {
+    const xmlChar xml[] =
+        "<refentry>\n"
+        "  <refsect1>\n"
+        "    <para>\n"
+        "      Run <command>tester --help</command> for more options.\n"
+        "    </para>\n"
+        "  </refsect1>\n"
+        "</refentry>\n";
+    const xmlChar expect[] =
+        "<?xml version=\"1.0\"?>\n"
+        "<refentry><refsect1><para>\n"
+        "      Run <command>tester --help</command> for more options.\n"
+        "    </para></refsect1></refentry>\n";
+    xmlDocPtr doc;
+    xmlChar *out;
+    int size;
+    int err = 0;
+
+    doc = xmlReadDoc(xml, NULL, NULL, XML_PARSE_NOBLANKS);
+    xmlDocDumpMemory(doc, &out, &size);
+    xmlFreeDoc(doc);
+
+    if (!xmlStrEqual(out, expect)) {
+        fprintf(stderr, "parsing with XML_PARSE_NOBLANKS failed\n");
+        err = 1;
+    }
+    xmlFree(out);
+
+    return err;
+}
 #endif /* LIBXML_OUTPUT_ENABLED */
 
 #ifdef LIBXML_SAX1_ENABLED
@@ -1123,6 +1156,7 @@ main(void) {
 #endif
 #ifdef LIBXML_OUTPUT_ENABLED
     err |= testCtxtParseContent();
+    err |= testNoBlanks();
 #endif
 #ifdef LIBXML_SAX1_ENABLED
     err |= testBalancedChunk();