From d248e7e9234927e32fd4cac77d743ece53086965 Mon Sep 17 00:00:00 2001
From: Sebastian Pipping <sebastian@pipping.org>
Date: Thu, 28 Sep 2023 15:56:42 +0200
Subject: [PATCH] xmlwf: Support custom buffer size for XML_GetBuffer and read

---
 expat/xmlwf/xmlfile.c        | 10 +++-------
 expat/xmlwf/xmlfile.h        |  2 ++
 expat/xmlwf/xmlwf.c          | 21 +++++++++++++++++++++
 expat/xmlwf/xmlwf_helpgen.py |  1 +
 4 files changed, 27 insertions(+), 7 deletions(-)

diff --git a/expat/xmlwf/xmlfile.c b/expat/xmlwf/xmlfile.c
index c55e3417..0cb7e1a2 100644
--- a/expat/xmlwf/xmlfile.c
+++ b/expat/xmlwf/xmlfile.c
@@ -71,11 +71,7 @@
 #  endif
 #endif
 
-#ifdef _DEBUG
-#  define READ_SIZE 16
-#else
-#  define READ_SIZE (1024 * 8)
-#endif
+int g_read_size_bytes = 1024 * 8;
 
 typedef struct {
   XML_Parser parser;
@@ -195,7 +191,7 @@ processStream(const XML_Char *filename, XML_Parser parser) {
   }
   for (;;) {
     int nread;
-    char *buf = (char *)XML_GetBuffer(parser, READ_SIZE);
+    char *buf = (char *)XML_GetBuffer(parser, g_read_size_bytes);
     if (! buf) {
       if (filename != NULL)
         close(fd);
@@ -203,7 +199,7 @@ processStream(const XML_Char *filename, XML_Parser parser) {
                filename != NULL ? filename : T("xmlwf"));
       return 0;
     }
-    nread = read(fd, buf, READ_SIZE);
+    nread = read(fd, buf, g_read_size_bytes);
     if (nread < 0) {
       tperror(filename != NULL ? filename : T("STDIN"));
       if (filename != NULL)
diff --git a/expat/xmlwf/xmlfile.h b/expat/xmlwf/xmlfile.h
index d75dda22..014f1319 100644
--- a/expat/xmlwf/xmlfile.h
+++ b/expat/xmlwf/xmlfile.h
@@ -42,5 +42,7 @@
 #  define XML_FMT_INT_MOD "l"
 #endif
 
+extern int g_read_size_bytes;
+
 extern int XML_ProcessFile(XML_Parser parser, const XML_Char *filename,
                            unsigned flags);
diff --git a/expat/xmlwf/xmlwf.c b/expat/xmlwf/xmlwf.c
index 276ddbb5..c84fcd77 100644
--- a/expat/xmlwf/xmlwf.c
+++ b/expat/xmlwf/xmlwf.c
@@ -899,6 +899,7 @@ usage(const XML_Char *prog, int rc) {
       T("  -e ENCODING   override any in-document [e]ncoding declaration\n")
       T("  -w            enable support for [W]indows code pages\n")
       T("  -r            disable memory-mapping and use [r]ead calls instead\n")
+      T("  -g BYTES      buffer size to request per call pair to XML_[G]etBuffer and read (default: 8 KiB)\n")
       T("  -k            when processing multiple files, [k]eep processing after first file with error\n")
       T("\n")
       T("output control arguments:\n")
@@ -1042,6 +1043,26 @@ tmain(int argc, XML_Char **argv) {
     case T('v'):
       showVersion(argv[0]);
       return 0;
+    case T('g'): {
+      const XML_Char *valueText = NULL;
+      XMLWF_SHIFT_ARG_INTO(valueText, argc, argv, i, j);
+
+      errno = 0;
+      XML_Char *afterValueText = (XML_Char *)valueText;
+      const long long read_size_bytes_candidate
+          = tcstoull(valueText, &afterValueText, 10);
+      if ((errno != 0) || (afterValueText[0] != T('\0'))
+          || (read_size_bytes_candidate < 1)
+          || (read_size_bytes_candidate > (INT_MAX / 2 + 1))) {
+        // This prevents tperror(..) from reporting misleading "[..]: Success"
+        errno = ERANGE;
+        tperror(T("invalid buffer size") T(
+            " (needs an integer from 1 to INT_MAX/2+1 i.e. 1,073,741,824 on most platforms)"));
+        exit(XMLWF_EXIT_USAGE_ERROR);
+      }
+      g_read_size_bytes = (int)read_size_bytes_candidate;
+      break;
+    }
     case T('k'):
       continueOnError = 1;
       j++;
diff --git a/expat/xmlwf/xmlwf_helpgen.py b/expat/xmlwf/xmlwf_helpgen.py
index 706b3c57..c9ac9a1e 100755
--- a/expat/xmlwf/xmlwf_helpgen.py
+++ b/expat/xmlwf/xmlwf_helpgen.py
@@ -63,6 +63,7 @@ input_related.add_argument('-x', action='store_true', help='enable processing of
 input_related.add_argument('-e', action='store', metavar='ENCODING', help='override any in-document [e]ncoding declaration')
 input_related.add_argument('-w', action='store_true', help='enable support for [W]indows code pages')
 input_related.add_argument('-r', action='store_true', help='disable memory-mapping and use [r]ead calls instead')
+input_related.add_argument('-g', metavar='BYTES', help='buffer size to request per call pair to XML_[G]etBuffer and read (default: 8 KiB)')
 input_related.add_argument('-k', action='store_true', help='when processing multiple files, [k]eep processing after first file with error')
 
 output_related = parser.add_argument_group('output control arguments')