Attachment #255474 for bug #282865




PORTNAME=	rxp
PORTVERSION=	1.5.2
PORTREVISION=	4
CATEGORIES=	textproc
MASTER_SITES=	https://www.inf.ed.ac.uk/research/isddarch/admin/

MAINTAINER=	ports@virtual-estates.net
COMMENT=	Validating XML parser written in C


LICENSE=	GPLv2

USES=	libtool
DEPRECATED=	Unfetchable and outdated
EXPIRATION_DATE=2024-12-31

USES=	uidfix

USE_LDCONFIG=	yes
-WRKSRC=                ${WRKDIR}
-MAKE_ARGS+=    VPATH=${PATCH_WRKSRC}
MAKE_ARGS+=	VPATH=${PATCH_WRKSRC}

OPTIONS_SINGLE=	CHAR_SIZE
OPTIONS_SINGLE_CHAR_SIZE=8 16
OPTIONS_DEFAULT=8
8_DESC=		8-bit characters (required by speech_tools, festival)
16_DESC=	16-bit characters (upstream's default)

#MAKE_ARGS+=	TOP=${PATCH_WRKSRC}

CFLAGS+=	-Wno-empty-body
GNU_CONFIGURE=	yes
INSTALL_TARGET=	install-strip
EXTRACT_AFTER_ARGS=	--exclude win32.h
MAKE_ENV+=	AM_LDFLAGS="-version-info 2:0:0"
.endfor
	${PRINTF} "\n.include <bsd.subdir.mk>\n" >> ${WRKDIR}/Makefile

post-patch:
	@${REINPLACE_CMD} -E -e 's|include/win32.h *||' \
	    -e 's,CHAR_SIZE=[0-9]+,CHAR_SIZE=${PORT_OPTIONS:C/[^0-9]//g},g' \
		${WRKSRC}/Makefile*

.include <bsd.port.mk>

Lines 1-2 Link Here

(-)b/textproc/rxp/distinfo (-2 / +3 lines)
1	SHA256 (rxp-1.4.4.tar.gz) = 96564e1ae1dd971415409585b697d62caac80fafaba00af8b8965128b4814208	1	TIMESTAMP = 1705901987
2	SIZE (rxp-1.4.4.tar.gz) = 153946	2	SHA256 (rxp-1.5.2.tar.gz) = fa641b4a5185d0a1d9610c8245b9d5aff5972c1a28a0da94f3e38d69f6c14513
		3	SIZE (rxp-1.5.2.tar.gz) = 476952




LIB=	rxp
SHLIB_MAJOR=	1
SHLIB_MINOR=	4
MK_PROFILE=	no

CFLAGS+=	-DCHAR_SIZE=${CHAR_SIZE} -DHAVE_LIBZ
CFLAGS+=	-Werror
WARNS=		6
LDADD=		-lz

SRCS=	catalog.c catutil.c resolve.c entityopener.c xmlparser.c	\
	url.c charset.c ctype16.c dtd.c input.c stdio16.c string16.c	\
	system.c hash.c version.c namespaces.c http.c			\
	nf16check.c nf16data.c
INCS=	charset.h ctype16.h dtd.h hash.h input.h namespaces.h \
	rxputil.h stdio16.h string16.h system.h url.h version.h \
	xmlparser.h

LIBDIR=	${PREFIX}/lib
INCSDIR=${PREFIX}/include/rxp

.include <bsd.lib.mk>

.if ${CHAR_SIZE} == 8
CFLAGS+=	-Wno-unused-label
.endif




PROG=	rxp

SRCS=	rxp.c infoset-print.c
CFLAGS+=	-DCHAR_SIZE=${CHAR_SIZE} -DHAVE_LIBZ
CFLAGS+=	-Werror
WARNS=		6

BINDIR=	${PREFIX}/bin
MANDIR=	${PREFIX}/share/man/man
LDADD=	-L${.CURDIR:H}/lib -lrxp

.include <bsd.prog.mk>




Restore the option to build with CHAR_SIZE being 8, which upstream
seems to have abandoned. We want it to be used with Festival, among
other things.

	-mi	(November 2024)

--- include/ctype16.h	2021-06-14 13:27:08.000000000 -0400
+++ include/ctype16.h	2024-11-19 13:24:35.494673000 -0500
@@ -28,7 +28,10 @@
 #define is_xml_namechar(c,map) (map[(int)(c) & 0xff] & xml_namechar)
 #define is_xml_whitespace(c) (xml_char_map[(int)(c) & 0xff] & xml_whitespace)
+#define is_xml_whitespace_16 is_xml_whitespace
+#define is_xml_namestart_16 is_xml_namestart
+#define is_xml_namechar_16 is_xml_namechar
 
-#define xml_char_map_11 0
-#define xml_char_map_105 0
+#define xml_char_map_11 xml_char_map
+#define xml_char_map_105 xml_char_map
 
 #else




--- catalog/catalog_dtd.c	2003-06-24 09:15:55.000000000 -0400
+++ catalog/catalog_dtd.c	2021-02-25 15:07:41.462399000 -0500
@@ -23,9 +23,9 @@
  */
 

+static const char xml_catalog_dtd[] =
 "<!ENTITY % pubIdChars \"CDATA\">\n"
 "<!ENTITY % publicIdentifier \"%pubIdChars;\">\n"
--- catalog/catalog.c	2004-10-11 08:28:31.000000000 -0400
+++ catalog/catalog.c	2021-02-25 15:07:41.463274000 -0500
@@ -20,5 +20,5 @@ static Parser OpenXMLDocument(char *uri)
 static void CloseXMLDocument(Parser p);
 static int SkipElement(XBit bit, Parser p);

+CatalogEntryFile GetCatalogEntryFile(Catalog catalog, const char *catalog_uri)
 {
     int i;
--- catalog/catalog.h	2003-08-28 12:13:19.000000000 -0400
+++ catalog/catalog.h	2021-02-25 15:07:41.463639000 -0500
@@ -48,13 +48,13 @@ extern void FreeCatalogEntry(CatalogEntr
 
 extern CatalogEntryFile catalog_resource_error;

+extern const char *ResolveURI(Catalog catalog, const char *uri);
 
 extern void CatalogEnable(Parser p);
--- catalog/catutil.c	2003-06-24 11:55:27.000000000 -0400
+++ catalog/catutil.c	2021-02-25 15:07:41.464062000 -0500
@@ -35,5 +35,5 @@ char *norm_pub(const char8 *public8, con
 {
     int len = public8 ? strlen(public8) : strlen16(public16);

+    static const char *p = "urn:publicid:";
     int i;
 
--- src/dtd.c	2004-11-02 13:20:20.000000000 -0500
+++ src/dtd.c	2021-02-25 15:07:41.475773000 -0500
@@ -49,13 +49,13 @@ extern STD_API unsigned char xml_char_ma
 
 #define is_xml_namestart(c,map) \
-  (c < 0x10000 ? (map[c] & xml_namestart) : (map[c >> 16] & xml_nameblock))
+  ((sizeof(c) <= 2 || c < 0x10000) ? (map[c] & xml_namestart) : (map[c >> 16] & xml_nameblock))
 
 #define is_xml_namechar(c,map) \
-  (c < 0x10000 ? (map[c] & xml_namechar)  : (map[c >> 16] & xml_nameblock))
+  ((sizeof(c) <= 2 || c < 0x10000) ? (map[c] & xml_namechar)  : (map[c >> 16] & xml_nameblock))
 
 /* NB whitespace map is the same for 1.0 and 1.1 */
 
 #define is_xml_whitespace(c) \
-  (c < 0x10000 && (xml_char_map[c] & xml_whitespace))
+  ((sizeof(c) <= 2 || c < 0x10000) && (xml_char_map[c] & xml_whitespace))
 
 #endif
--- dtd.c	2004-11-02 13:20:20.000000000 -0500
+++ dtd.c	2021-02-25 15:07:41.475773000 -0500
@@ -173,12 +173,11 @@ void FreeDtd(Dtd dtd)
  */
 

+				    char8 *publicid, char8 *systemid,
 				    Entity parent)
 {
--- include/dtd.h	2004-11-02 12:46:07.000000000 -0500
+++ include/dtd.h	2021-02-25 15:07:41.465041000 -0500
@@ -48,11 +48,11 @@ struct entity {
     /* All entities */
 

+				  char8 *publicid, char8 *systemid,
 					    Entity parent);
 XML_API NotationDefinition FindNotationN(Dtd dtd, const Char *name, int namelen);
--- catalog/entityopener.c	2003-11-27 09:06:38.000000000 -0500
+++ catalog/entityopener.c	2021-02-25 15:07:41.465352000 -0500
@@ -9,5 +9,5 @@
 #include "dtd.h"
 

+static InputSource catalog_entity_open(Entity ent, void *arg)
 {
     Catalog catalog = arg;
--- src/http.c	2001-05-01 10:29:34.000000000 -0400
+++ src/http.c	2021-02-25 15:33:16.444375000 -0500
@@ -139,5 +139,5 @@ FILE16 *http_open(const char *url,
     int server_port;
     char buf[100];

+	    /* fall through */
 	default:
 	    if(!VectorPush(text, c))
--- src/infoset-print.c	2021-06-14 13:42:14.000000000 -0400
+++ src/infoset-print.c	2024-01-22 00:58:08.195742000 -0500
@@ -23,5 +23,5 @@
 		     HashTable id_table);
 static void notations(FILE16 *f, int level, Dtd dtd);
-static void unparsed_entities(FILE16 *f, int level, Dtd dtd, Entity docent);
+static void unparsed_entities(FILE16 *f, int level, Dtd dtd);
 static void baseURI(FILE16 *f, int level, const char8 *uri);
 static void standalone(FILE16 *f, int level, StandaloneDeclaration sd);
@@ -33,11 +33,11 @@
 		    HashTable id_table);
 static void pi(FILE16 *f, int level, Dtd dtd, XBit bit);
-static void cdsect(FILE16 *f, int level, Dtd dtd, XBit bit);

+static void character(FILE16 *f, int level, const char *ecw, int c);
 static void attributes(FILE16 *f, int level, Dtd dtd, XBit bit, 
 		       HashTable id_table);
@@ -46,5 +46,5 @@
 static void namespace_attributes(FILE16 *f, int level, Dtd dtd, XBit bit,
 				 HashTable id_table);
-static void inscope_namespaces(FILE16 *f, int level, Dtd dtd, XBit bit);
+static void inscope_namespaces(FILE16 *f, int level, XBit bit);
 #if 0
 static void internal_entity(FILE16 *f, int level, Entity entity);
@@ -53,7 +53,7 @@
 static void unparsed_entity(FILE16 *f, int level, Entity entity);
 
-static void simple(FILE16 *f, int level, char *name, const char *value);

+static void pointer(FILE16 *f, int level, const char *name, Char *id);
 static Char *make_id(const char *type, const Char *name, int count);
 static void indent(FILE16 *f, int level);
@@ -67,5 +67,5 @@
 		     int *counter);
 
-struct xbit bogus_bit;
+static struct xbit bogus_bit;
 
 static Char xmlns_ns[] = {'h','t','t','p',':','/','/','w','w','w','.','w', '3',
@@ -93,5 +93,5 @@
     notations(f, 1, dtd);
 
-    unparsed_entities(f, 1, dtd, p->document_entity);
+    unparsed_entities(f, 1, dtd);
 
     baseURI(f, 1, EntityBaseURL(p->document_entity));
@@ -172,5 +172,5 @@
 }
 
-static void unparsed_entities(FILE16 *f, int level, Dtd dtd, Entity docent)
+static void unparsed_entities(FILE16 *f, int level, Dtd dtd)
 {
     Entity e;
@@ -280,5 +280,5 @@
 }
 
-static void simple(FILE16 *f, int level, char *name, const char *value)
+static void simple(FILE16 *f, int level, const char *name, const char *value)
 {
     indent(f, level);
@@ -294,5 +294,5 @@
 }
 
-static void Simple(FILE16 *f, int level, char *name, const Char *value)
+static void Simple(FILE16 *f, int level, const char *name, const Char *value)
 {
     indent(f, level);
@@ -308,5 +308,5 @@
 }
 
-static void pointer(FILE16 *f, int level, char *name, Char *id)
+static void pointer(FILE16 *f, int level, const char *name, Char *id)
 {
     indent(f, level);
@@ -342,11 +342,11 @@
 	    break;
 	case XBIT_cdsect:
-	    cdsect(f, level, dtd, bit);

+	    comment(f, level, bit);
 	    break;
 	default:
@@ -396,5 +396,5 @@
     namespace_attributes(f, level+1, dtd, bit, id_table);
 
-    inscope_namespaces(f, level+1, dtd, bit);
+    inscope_namespaces(f, level + 1, bit);
 
     baseURI(f, 1, EntityBaseURL(bit->entity)); /* XXX xml:base */
@@ -433,5 +433,5 @@
 }
 
-static void cdsect(FILE16 *f, int level, Dtd dtd, XBit bit)
+static void cdsect(FILE16 *f, int level, XBit bit)
 {
     Char *p;
@@ -443,5 +443,5 @@
 
     for(p=bit->pcdata_chars; *p; p++)
-	character(f, level, dtd, "false", *p);
+	character(f, level, "false", *p);
 
 #if 0
@@ -451,8 +451,8 @@
 }
 
-static void pcdata(FILE16 *f, int level, Dtd dtd, XBit bit)

+    const char *ecw;
 
     ecw = !bit->parent->element_definition->declared ? 0 :
@@ -461,8 +461,8 @@
 
     for(p=bit->pcdata_chars; *p; p++)
-	character(f, level, dtd, is_xml_whitespace_16(*p) ? ecw : "false", *p);
+	character(f, level, is_xml_whitespace_16(*p) ? ecw : "false", *p);
 }
 
-static void character(FILE16 *f, int level, Dtd dtd, char *ecw, int c)
+static void character(FILE16 *f, int level, const char *ecw, int c)
 {
     indent(f, level);
@@ -478,5 +478,5 @@
 }
 
-static void comment(FILE16 *f, int level, Dtd dtd, XBit bit)
+static void comment(FILE16 *f, int level, XBit bit)
 {
     indent(f, level);
@@ -615,4 +615,5 @@
 	free(token);
 	free(id);
+	/* FALLTHROUGH */
     default:
 	simple(f, level+1, "references", 0);
@@ -640,5 +641,5 @@
 }
 
-static void inscope_namespaces(FILE16 *f, int level, Dtd dtd, XBit bit)
+static void inscope_namespaces(FILE16 *f, int level, XBit bit)
 {
     NamespaceBinding nsb, nsb2;
--- src/input.c	2005-01-14 11:10:00.000000000 -0500
+++ src/input.c	2021-02-25 15:07:41.467694000 -0500
@@ -79,5 +79,5 @@ InputSource EntityOpen(Entity e)
     else
     {

-
     END_OF_LINE;
 }
--- src/namespaces.c	2003-06-02 12:56:35.000000000 -0400
+++ src/namespaces.c	2021-02-25 15:07:41.468014000 -0500
@@ -25,5 +25,5 @@ static void FreeNSElementDefinition(NSEl
 static void FreeNSAttributeDefinition(NSAttributeDefinition attribute);
 

+    Free(attribute->name);
     Free(attribute);
 }
--- include/namespaces.h	2003-06-02 12:56:35.000000000 -0400
+++ include/namespaces.h	2021-02-25 15:07:41.468292000 -0500
@@ -30,5 +30,5 @@ struct RXP_NAMESPACE {
 
 struct ns_element_definition {

+    Char *name;
     int attrnum;
 };
--- src/nf16check.c	2003-09-02 10:02:45.000000000 -0400
+++ src/nf16check.c	2021-02-25 15:07:41.468701000 -0500
@@ -7,4 +7,6 @@
 
 #include <stdlib.h>

 /* variant of nf16check, s_length gives length of s */
@@ -226,4 +228,5 @@ nf16res nf16checkL (NF16Checker checker,
     int          lastclass   = checker->lastclass;
     unsigned int c;		/* changed from char16 (RMT) */
+    uint32_t     C;
     flag         f;
     int          class;

+            else if (C < 0xE01F0)  f = simp;
             else                   f = NoNo;
             goto NEWFLAG;   /* start again with switch */
--- catalog/resolve.c	2003-08-28 12:13:12.000000000 -0400
+++ catalog/resolve.c	2021-02-25 15:07:41.469245000 -0500
@@ -8,11 +8,11 @@
 #include "string.h"
 

+    return strlen((*(const CatalogEntry *)b)->match) - 
+	   strlen((*(const CatalogEntry *)a)->match);
 }
--- src/rxp.c	2019-10-24 14:23:25.000000000 -0400
+++ src/rxp.c	2024-10-20 22:09:28.202835000 -0400
@@ -33,40 +33,39 @@
 #include "catalog.h"
 
-int attr_compare(const void *a, const void *b);

-void print_ns_attrs(NamespaceBinding ns, int count);
-void print_namespaces(NamespaceBinding ns);
-void print_attrs(ElementDefinition e, Attribute a);
-void print_text(Char *text, int is_attr);
-int printable(int c);
-void print_special(int c);
-void print_text_bit(Char *text);

+static void print_ns_attrs(NamespaceBinding ns, int count);
+static void print_namespaces(NamespaceBinding ns);
+static void print_attrs(Attribute a);
+static void print_text(const Char *text, int is_attr);
+static int printable(int c);
+static void print_special(int c);
+static void print_text_bit(Char *text);

+static int verbose = 0, expand = 1, nsgml = 0,
     attr_defaults = 0, merge = 0, strict_xml = 0, tree = 0, validate = 0,
     xml_space = 0, namespaces = 0, simple_error = 0, experiment = 0,
     read_dtd = 0, unicode_check = 0, xml_id = 0, pre_105 = 0;
-enum {o_unspec, o_none, o_bits, o_plain, o_can1, o_can2, o_can3, o_infoset, o_diff, o_diff2} output_format = o_unspec;
-char *enc_name = 0, *base_uri = 0, *my_dtd_name, *my_dtd_sysid = 0;
-CharacterEncoding encoding = CE_unknown;
-InputSource source = 0;
-int need_canonical_dtd = 0;
-int xml_version;
+static enum {o_unspec, o_none, o_bits, o_plain, o_can1, o_can2, o_can3, o_infoset, o_diff, o_diff2} output_format = o_unspec;
+static char *enc_name = NULL, *base_uri = NULL, *my_dtd_name, *my_dtd_sysid = NULL;
+static CharacterEncoding encoding = CE_unknown;
+static InputSource source = NULL;

+StaticVector(XBit, dtd_bits);
 
 int main(int argc, char **argv)
@@ -581,5 +580,5 @@
 {
     const char *sys, *pub;
-    char *ws[] = {"u", "d", "p"};
+    const char *ws[] = {"u", "d", "p"};
 
     if(output_format == o_none && bit->type != XBIT_error)
@@ -617,5 +616,5 @@
 	    if(xml_space)
 		Printf("(ws=%s) ", ws[bit->wsm]);
-	    print_attrs(0, bit->attributes);
+	    print_attrs(bit->attributes);
 	    print_namespaces(bit->ns_dict);
 	    Printf("\n");
@@ -630,5 +629,5 @@
 	    if(xml_space)
 		Printf("(ws=%s) ", ws[bit->wsm]);
-	    print_attrs(0, bit->attributes);
+	    print_attrs(bit->attributes);
 	    print_namespaces(bit->ns_dict);
 	    Printf("\n");
@@ -701,5 +700,5 @@
 		print_canonical_dtd(p, bit->element_definition->name);
 	    Printf("<%S", bit->element_definition->name);
-	    print_attrs(bit->element_definition, bit->attributes);
+	    print_attrs(bit->attributes);
 	    print_ns_attrs(bit->ns_dict, bit->nsc);
 	    if(bit->type == XBIT_start)
@@ -769,9 +768,9 @@
 int attr_compare(const void *a, const void *b)
 {
-    return Strcmp((*(Attribute *)a)->definition->name,

+void print_attrs(Attribute a)
 {
     Attribute b;
@@ -826,5 +825,5 @@
 }
 	
-void dtd_cb2(XBit bit, void *arg)
+void dtd_cb2(XBit bit, void *arg __unused)
 {
     XBit copy;
@@ -839,7 +838,7 @@
 }
 	
-void print_text(Char *text, int is_attr)
+void print_text(const Char *text, int is_attr)
 {
-    Char *pc, *last;
+    const Char *pc, *last;
     
     if(output_format == o_bits  || !expand)
@@ -875,5 +874,5 @@
 	   (xml_version > XV_1_0 && type == 3))
 	{
-	    Char *end = (c > 0xffff ? pc-1 : pc); /* if it's a surrogate pair, before the first  */
+	    const Char *end = (c > 0xffff ? pc-1 : pc); /* if it's a surrogate pair, before the first  */
 	    if(end > last)
 		Printf("%.*S", end - last, last);
@@ -949,25 +948,4 @@
 }
 
-InputSource entity_open(Entity ent, void *arg)

-
 void print_ns_attrs(NamespaceBinding ns, int count)
 {
@@ -1017,13 +995,13 @@
 }
 
-int notation_compare(const void *a, const void *b)

+    return Strcmp((*(const Entity *)a)->name, (*(const Entity *)b)->name);
 }
 
@@ -1147,5 +1125,5 @@
 
 #ifdef TIME_LIMIT
-void time_exceeded(int sig)
+void time_exceeded(int sig __unused)
 {
     fprintf(stderr, "CPU time limit (%d seconds) exceeded, sorry\n", 
--- include/rxputil.h	2003-11-27 09:06:57.000000000 -0500
+++ include/rxputil.h	2021-02-25 15:07:41.470232000 -0500
@@ -6,4 +6,8 @@
     type *name
 

+
 #define VectorInit(v) \
     ((v##_count) = (v##_alloc) = 0, (v) = 0)
--- src/stdio16.c	2005-01-14 11:10:54.000000000 -0500
+++ src/stdio16.c	2021-02-25 15:32:25.156652000 -0500
@@ -15,4 +15,5 @@
 #include <string.h>
 #include <stdarg.h>

+static int GzipFlush(FILE16 *file __unused)
 {
     return 0;
--- include/stdio16.h	2002-10-03 10:37:44.000000000 -0400
+++ include/stdio16.h	2021-02-25 15:28:59.692044000 -0500
@@ -5,5 +5,5 @@
 #include <stdio.h>
 #ifdef HAVE_LIBZ

+#endif
 
 STD_API int Fprintf(FILE16 *file, const char *format, ...);
--- src/string16.c	2004-10-11 12:18:29.000000000 -0400
+++ src/string16.c	2021-02-25 15:07:41.471817000 -0500
@@ -171,9 +171,9 @@ size_t strlen16(const char16 *s)
 }
 

     return 0;
 }
-
--- include/string16.h	1999-05-28 08:08:38.000000000 -0400
+++ include/string16.h	2021-02-25 15:07:41.472166000 -0500
@@ -23,5 +23,5 @@ STD_API int strncasecmp8(const char8 *, 
 
 STD_API char16 *strdup16(const char16 *s);

+STD_API const char16 *strstr16(const char16 *, const char16 *);
 
 STD_API void translate_latin1_utf16(const char8 *from, char16 *to);
--- include/system.h	2003-05-19 20:17:27.000000000 -0400
+++ include/system.h	2021-02-25 15:10:55.417342000 -0500
@@ -1,3 +1,5 @@
+#ifndef _RXP_SYSTEM_H
+#define _RXP_SYSTEM_H
 #define HAVE_LONG_LONG
+#define HAVE_LONG_DOUBLE
 
 #define SOCKETS_IMPLEMENTED
@@ -11,2 +14,4 @@ void *Malloc(int bytes);

 void Free(void *mem);
+
+#endif
--- src/url.c	2004-06-11 07:39:17.000000000 -0400
+++ src/url.c	2021-02-25 15:07:41.472694000 -0500
@@ -80,10 +80,10 @@ static int hexval(int hex);
 /* Mapping of scheme names to opening functions */
 

+	    *r = '/';
 	}
 }
--- src/version.c	2008-05-29 06:51:54.000000000 -0400
+++ src/version.c	2024-01-22 01:11:34.603161000 -0500
@@ -1,4 +1,5 @@
 #include "config.h"
+#include "version.h"
 
-char *rxp_version_string = 
+const char *rxp_version_string = 
     "RXP " PACKAGE_VERSION " Copyright Richard Tobin, LTG, HCRC, University of Edinburgh";
--- include/version.h	1999-02-17 09:43:19.000000000 -0500
+++ include/version.h	2021-02-25 15:07:41.473163000 -0500
@@ -1 +1 @@
-extern char *rxp_version_string;
+extern const char *rxp_version_string;
--- src/xmlparser.c	2021-06-14 14:02:03.000000000 -0400
+++ src/xmlparser.c	2024-01-22 00:53:35.001099000 -0500
@@ -4,7 +4,6 @@
 #define DEBUG_FSM 0
 
-#ifndef lint
-static char vcid[] = "$Id: xmlparser.c,v 1.137 2021/06/14 18:02:03 richard Exp $";
-#endif /* lint */
+#include <sys/cdefs.h>
+__RCSID("$Id: xmlparser.c,v 1.137 2021/06/14 18:02:03 richard Exp $");
+__RCSID("$Id: xmlparser.c,v 1.131 2004/11/03 13:47:31 richard Exp $");
 
 /* 
@@ -142,6 +141,6 @@
 };
 static int parse_string(Parser p, const char8 *where, enum literal_type type, int *normalised);
-static int parse_pi(Parser p, Entity ent);

+static int parse_comment(Parser p, int skip);
 static int parse_pcdata(Parser p);
 static int parse_starttag(Parser p);
@@ -150,5 +149,5 @@
 			     AttributeDefinition d, const Char *value);
 static int parse_attribute(Parser p);
-static WhiteSpaceMode process_xml_space(Parser p, const Char *value);
+static WhiteSpaceMode process_xml_space(const Char *value);
 static int parse_endtag(Parser p);
 static int parse_markup(Parser p);
@@ -159,5 +158,7 @@
 static HashMapRetType check_id(const HashEntryStruct *id_entry, void *p);
 static int validate_attribute(Parser p, AttributeDefinition a, ElementDefinition e, const Char *value);
+#if 0

+#endif
 static int check_attribute_syntax(Parser p, AttributeDefinition a, ElementDefinition e, const Char *value, const char *message, int real_use);
 static int check_attribute_token(Parser p, AttributeDefinition a, ElementDefinition e, const Char *value, int length, const char *message, int real_use);
@@ -167,5 +168,5 @@
 
 static struct element_definition pcdata_element;
-const ElementDefinition Epsilon = 0, PCDataElement = &pcdata_element;
+static const ElementDefinition Epsilon = 0, PCDataElement = &pcdata_element;
 
 static FSM NewFSM(void);
@@ -1207,8 +1208,8 @@
 	{
 	    if(ParserGetFlag(p, ReturnComments))
-		return parse_comment(p, 0, 0);

+		require(parse_comment(p, 1));
 		/* XXX avoid recursion here */
 		return parse(p);
@@ -1228,5 +1229,5 @@
 
     case '?':
-	return parse_pi(p, 0);
+	return parse_pi(p);
 
     case BADCHAR:
@@ -1326,5 +1327,5 @@
 static int check_qualname_syntax(Parser p, const Char *name, const char *type)
 {
-    Char *t;
+    const Char *t;
 
     t = Strchr(name, ':'); 
@@ -1358,5 +1359,8 @@
 static int parse_starttag(Parser p)
 {
-    int c, is_top_level = 0;

+#endif
     ElementDefinition e;
     AttributeDefinition d;
@@ -1374,5 +1378,7 @@
 	    require(validate_dtd(p));
 	}
+#if not_yet

+#endif
     }
 
@@ -1545,7 +1551,4 @@
 	/* check for required attributes */
 
-	AttributeDefinition d;

-
 	for(d=NextAttributeDefinition(e, 0);
 	    d;
@@ -1662,5 +1665,5 @@
 		if(a->definition == d)
 		{
-		    p->xbit.wsm = process_xml_space(p, a->value);
+		    p->xbit.wsm = process_xml_space(a->value);
 		    goto done;
 		}
@@ -1668,5 +1671,5 @@
 	    if(d->default_type == DT_none || d->default_type == DT_fixed)
 	    {
-		p->xbit.wsm = process_xml_space(p, d->default_value);
+		p->xbit.wsm = process_xml_space(d->default_value);
 		goto done;
 	    }
@@ -2090,9 +2093,8 @@
 }
 
-static WhiteSpaceMode process_xml_space(Parser p, const Char *value)

-    Char buf[9];
     const Char *v;
     int i;
@@ -2106,7 +2108,5 @@
 	if(!v[i] || is_xml_whitespace_16(v[i]))
 	    break;
-	buf[i] = v[i];
     }
-    buf[i] = '\0';
     for(; v[i]; i++)
 	if(!is_xml_whitespace_16(v[i]))
@@ -2208,5 +2208,5 @@
 	    {
 		s->next = next + 3;
-		require(parse_comment(p, 1, 0));
+		require(parse_comment(p, 1));
                 NF16StartCheck(p);
 		buflen = s->line_length;
@@ -2352,5 +2352,5 @@
 /* Called after reading '<!--'.  Won't go over an entity end. */
 
-static int parse_comment(Parser p, int skip, Entity ent)
+static int parse_comment(Parser p, int skip)
 {
     InputSource s = p->source;
@@ -2421,5 +2421,5 @@
 }
 
-static int parse_pi(Parser p, Entity ent)
+static int parse_pi(Parser p)
 {
     InputSource s = p->source;
@@ -3002,5 +3002,5 @@
     char8 *value;
     CharacterEncoding enc = CE_unknown;
-    Char c;
+    int c;
 
     /* 
@@ -3344,5 +3344,5 @@
 	else if(looking_at(p, "?"))
 	{
-	    require(parse_pi(p, cur_ent));
+	    require(parse_pi(p));
 	    if(p->dtd_callback)
 		p->dtd_callback(&p->xbit, p->dtd_callback_arg);
@@ -3355,5 +3355,5 @@
 	    if(ParserGetFlag(p, ReturnComments))
 	    {
-		require(parse_comment(p, 0, cur_ent));
+		require(parse_comment(p, 0));
 		if(p->dtd_callback)
 		    p->dtd_callback(&p->xbit, p->dtd_callback_arg);
@@ -3363,5 +3363,5 @@
 	    }
 	    else
-		return parse_comment(p, 1, cur_ent);
+		return parse_comment(p, 1);
 	}
 	else if(p->state == PS_error)	/* looking_at may have set it */
@@ -5387,17 +5387,19 @@
     Parser p = (Parser)pp;
 
-    if(!(int)hash_get_value(id_entry))
+    if (hash_get_value(id_entry) != NULL)
 	validity_error(p,
 		       "The ID %.*S was referred to but never defined",
@@ -5351,17 +5352,19 @@ static int validate_attribute(Parser p, 
 	}
 
+#if 0

-#if 0
     const Char *t;
 
@@ -5440,7 +5442,7 @@
     /* Not a validity error since erratum 73 */
     warn(p, "Dubious xml:lang attribute for element %S", e->name);
-#endif

+#endif
 
 /* Check an attribute matches Name[s] or Nmtoken[s].
@@ -5518,5 +5521,5 @@ static int check_attribute_token(Parser 
 	if(!found)
 	{
-	    hash_set_value(id_entry, (void *)(a->type == AT_id));
+	    hash_set_value(id_entry, (void *)(intptr_t)(a->type == AT_id));
 	    if(ParserGetFlag(p, XMLNamespaces))
 		for(i=0; i<length; i++)
@@ -5528,5 +5531,5 @@ static int check_attribute_token(Parser 
 	else if(a->type == AT_id)
 	{
-	    int idinfo = (int)hash_get_value(id_entry);
+	    intptr_t idinfo = (intptr_t)hash_get_value(id_entry);
 	    if(idinfo & 1)
 	    {

Lines 1-13 Link Here

(-)b/textproc/rxp/pkg-plist (-2 / +8 lines)
1	bin/rxp	1	bin/rxp
2	lib/librxp.a	2	lib/librxp.a
3	lib/librxp.so	3	lib/librxp.so
4	lib/librxp.so.1	4	lib/librxp.so.2
		5	lib/librxp.so.2.0.0
		6	include/rxp/catalog.h
		7	include/rxp/catutil.h
5	include/rxp/charset.h	8	include/rxp/charset.h
6	include/rxp/ctype16.h	9	include/rxp/ctype16.h
7	include/rxp/dtd.h	10	include/rxp/dtd.h
8	include/rxp/hash.h	11	include/rxp/hash.h
		12	include/rxp/http.h
		13	include/rxp/infoset-print.h
9	include/rxp/input.h	14	include/rxp/input.h
10	include/rxp/namespaces.h	15	include/rxp/namespaces.h
		16	include/rxp/nf16check.h
		17	include/rxp/nf16data.h
11	include/rxp/rxputil.h	18	include/rxp/rxputil.h
12	include/rxp/stdio16.h	19	include/rxp/stdio16.h
13	include/rxp/string16.h	20	include/rxp/string16.h
14	-

Return to bug 282865

Lines 1-8 Link Here

(-)b/textproc/rxp/Makefile (-23 / +17 lines)
1	PORTNAME= rxp	1	PORTNAME= rxp
2	PORTVERSION= 1.4.4	2	PORTVERSION= 1.5.2
3	PORTREVISION= 4
4	CATEGORIES= textproc	3	CATEGORIES= textproc
5	MASTER_SITES= ftp://ftp.cogsci.ed.ac.uk/pub/richard/	4	MASTER_SITES= https://www.inf.ed.ac.uk/research/isddarch/admin/
6		5
7	MAINTAINER= ports@virtual-estates.net	6	MAINTAINER= ports@virtual-estates.net
8	COMMENT= Validating XML parser written in C	7	COMMENT= Validating XML parser written in C
Lines 10-43 WWW= https://www.cogsci.ed.ac.uk/~richard/rxp.html Link Here
10		9
11	LICENSE= GPLv2	10	LICENSE= GPLv2
12		11
13	BROKEN= Unfetchable	12	USES= libtool
14	DEPRECATED= Unfetchable and outdated
15	EXPIRATION_DATE=2024-12-31
16
17	USES= uidfix
18		13
19	USE_LDCONFIG= yes	14	USE_LDCONFIG= yes
20	PATCH_WRKSRC= ${WRKSRC}/${PORTNAME}-${PORTVERSION}	15	-WRKSRC= ${WRKDIR}
21	WRKSRC= ${WRKDIR}	16	-MAKE_ARGS+= VPATH=${PATCH_WRKSRC}
22	MAKE_ARGS+= VPATH=${PATCH_WRKSRC}
23		17
24	OPTIONS_SINGLE= CHAR_SIZE	18	OPTIONS_SINGLE= CHAR_SIZE
25	OPTIONS_SINGLE_CHAR_SIZE=8 16	19	OPTIONS_SINGLE_CHAR_SIZE=8 16
26	OPTIONS_DEFAULT=8	20	OPTIONS_DEFAULT=8
27	8_DESC= 8-bit characters (required by speech_tools, festival)	21	8_DESC= 8-bit characters (required by speech_tools, festival)
28	16_DESC= 16-bit characters	22	16_DESC= 16-bit characters (upstream's default)
29		23
30	MAKE_ARGS+= CHAR_SIZE=${PORT_OPTIONS:C/[^0-9]//g}	24	#MAKE_ARGS+= TOP=${PATCH_WRKSRC}
31		25
32	do-configure:	26	CFLAGS+= -Wno-empty-body
33	.for s in lib rxp	27	GNU_CONFIGURE= yes
34	${MKDIR} ${WRKDIR}/$s	28	INSTALL_TARGET= install-strip
35	${LN} -s ${FILESDIR}/BSDmakefile.$s ${WRKDIR}/$s/Makefile	29	EXTRACT_AFTER_ARGS= --exclude win32.h
36	${PRINTF} "SUBDIR+=\t%s\n" $s >> ${WRKDIR}/Makefile	30	MAKE_ENV+= AM_LDFLAGS="-version-info 2:0:0"
37	.endfor
38	${PRINTF} "\n.include <bsd.subdir.mk>\n" >> ${WRKDIR}/Makefile
39		31
40	post-build:	32	post-patch:
41	${MKDIR} ${STAGEDIR}${PREFIX}/include/rxp	33	@${REINPLACE_CMD} -E -e 's\|include/win32.h *\|\|' \
		34	-e 's,CHAR_SIZE=[0-9]+,CHAR_SIZE=${PORT_OPTIONS:C/[^0-9]//g},g' \
		35	${WRKSRC}/Makefile*
42		36
43	.include <bsd.port.mk>	37	.include <bsd.port.mk>

Removed Link Here

(-)a/textproc/rxp/files/BSDmakefile.lib (-26 lines)
1	LIB= rxp
2	SHLIB_MAJOR= 1
3	SHLIB_MINOR= 4
4	MK_PROFILE= no
5
6	CFLAGS+= -DCHAR_SIZE=${CHAR_SIZE} -DHAVE_LIBZ
7	CFLAGS+= -Werror
8	WARNS= 6
9	LDADD= -lz
10
11	SRCS= catalog.c catutil.c resolve.c entityopener.c xmlparser.c \
12	url.c charset.c ctype16.c dtd.c input.c stdio16.c string16.c \
13	system.c hash.c version.c namespaces.c http.c \
14	nf16check.c nf16data.c
15	INCS= charset.h ctype16.h dtd.h hash.h input.h namespaces.h \
16	rxputil.h stdio16.h string16.h system.h url.h version.h \
17	xmlparser.h
18
19	LIBDIR= ${PREFIX}/lib
20	INCSDIR=${PREFIX}/include/rxp
21
22	.include <bsd.lib.mk>
23
24	.if ${CHAR_SIZE} == 8
25	CFLAGS+= -Wno-unused-label
26	.endif

Removed Link Here

(-)a/textproc/rxp/files/BSDmakefile.rxp (-12 lines)
1	PROG= rxp
2
3	SRCS= rxp.c infoset-print.c
4	CFLAGS+= -DCHAR_SIZE=${CHAR_SIZE} -DHAVE_LIBZ
5	CFLAGS+= -Werror
6	WARNS= 6
7
8	BINDIR= ${PREFIX}/bin
9	MANDIR= ${PREFIX}/share/man/man
10	LDADD= -L${.CURDIR:H}/lib -lrxp
11
12	.include <bsd.prog.mk>

Added Link Here

(-)b/textproc/rxp/files/patch-char-size-8 (+21 lines)
1	Restore the option to build with CHAR_SIZE being 8, which upstream
2	seems to have abandoned. We want it to be used with Festival, among
3	other things.
4
5	-mi (November 2024)
6
7	--- include/ctype16.h 2021-06-14 13:27:08.000000000 -0400
8	+++ include/ctype16.h 2024-11-19 13:24:35.494673000 -0500
9	@@ -28,7 +28,10 @@
10	#define is_xml_namechar(c,map) (map[(int)(c) & 0xff] & xml_namechar)
11	#define is_xml_whitespace(c) (xml_char_map[(int)(c) & 0xff] & xml_whitespace)
12	+#define is_xml_whitespace_16 is_xml_whitespace
13	+#define is_xml_namestart_16 is_xml_namestart
14	+#define is_xml_namechar_16 is_xml_namechar
15
16	-#define xml_char_map_11 0
17	-#define xml_char_map_105 0
18	+#define xml_char_map_11 xml_char_map
19	+#define xml_char_map_105 xml_char_map
20
21	#else