From: Adam Dickmeiss <adam@indexdata.dk>
Date: Mon, 13 May 2002 14:13:37 +0000 (+0000)
Subject: XML reader for data1 (EXPAT)
X-Git-Tag: YAZ.1.8.7.CCL~5
X-Git-Url: http://sru.miketaylor.org.uk/?a=commitdiff_plain;h=651ccb74d12f2968e791b3f411a6c55bca68428a;p=yaz-moved-to-github.git

XML reader for data1 (EXPAT)
---

diff --git a/CHANGELOG b/CHANGELOG
index 59cfb09..472384b 100644
--- a/CHANGELOG
+++ b/CHANGELOG
@@ -1,7 +1,10 @@
 Possible compatibility problems with earlier versions marked with '*'.
 
+
 --- 1.8.7 2002/MM/DD
 
+XML reader for data1 (EXPAT).
+
 Fixed bug in ZOOM connect that occurred on OpenBSD (maybe others). Patch
 by Oleg Kolobov.
 
diff --git a/buildconf.sh b/buildconf.sh
index 5373ba8..6c75e17 100755
--- a/buildconf.sh
+++ b/buildconf.sh
@@ -1,5 +1,5 @@
 #!/bin/sh
-# $Id: buildconf.sh,v 1.10 2002-03-16 11:46:18 adam Exp $
+# $Id: buildconf.sh,v 1.11 2002-05-13 14:13:37 adam Exp $
 aclocal 
 libtoolize --automake --force 
 automake -a 
@@ -7,4 +7,4 @@ autoconf
 if [ -f config.cache ]; then
 	rm config.cache
 fi
-util/cvs-date.tcl include/yaz/yaz-date.h
+#util/cvs-date.tcl include/yaz/yaz-date.h
diff --git a/configure.in b/configure.in
index ea5551e..c5b3a91 100644
--- a/configure.in
+++ b/configure.in
@@ -1,10 +1,11 @@
 dnl YAZ Toolkit, Index Data 1994-2001
 dnl See the file LICENSE for details.
-dnl $Id: configure.in,v 1.72 2002-05-03 13:47:57 adam Exp $
+dnl $Id: configure.in,v 1.73 2002-05-13 14:13:37 adam Exp $
 AC_INIT(include/yaz/yaz-version.h)
 AM_INIT_AUTOMAKE(yaz, 1.8.7)
 dnl
 AC_SUBST(READLINE_LIBS)
+AC_SUBST(YAZ_CONF_CFLAGS)
 dnl ------ Checking programs
 AC_PROG_CC
 AC_PROG_CPP
@@ -93,6 +94,30 @@ if test "$checkBoth" = "1"; then
 fi
 AC_CHECK_FUNC(gethostbyname, , AC_CHECK_LIB(nsl, main, [LIBS="$LIBS -lnsl"]))
 dnl
+dnl ------ EXPAT
+expat=yes
+AC_ARG_WITH(expat,   [  --with-expat[=DIR]      EXPAT library in DIR],[expat=$withval])
+if test "$expat" != "no"; then
+	xLIBS="$LIBS";
+	xCFLAGS="$CFLAGS";
+	if test "$expat" != "yes"; then
+                EXPAT_CFLAGS="-I$expat/include"
+                EXPAT_LIBS="-L$expat/lib"
+		CFLAGS="$EXPAT_CFLAGS $CFLAGS"
+		LIBS="$EXPAT_LIBS $LIBS"
+	fi
+	AC_CHECK_LIB(expat,XML_ParserCreate,[LIBS="$LIBS -lexpat"])
+	if test "$ac_cv_lib_expat_XML_ParserCreate" = "yes"; then
+		AC_CHECK_HEADERS(expat.h)
+                if test "$ac_cv_header_expat_h" = "yes"; then
+                        YAZ_CONF_CFLAGS="$YAZ_CONF_CFLAGS -DYAZ_HAVE_EXPAT=1"
+                fi
+	else
+		LIBS="$xLIBS"
+		CFLAGS="$xCFLAGS"
+	fi
+fi
+dnl
 dnl ------ Open SSL
 openssl=no
 AC_ARG_WITH(openssl, [  --with-openssl[=DIR]    OpenSSL library in DIR], [openssl=$withval])
@@ -125,6 +150,7 @@ if test "$openssl" != "no"; then
 	CFLAGS="$xCFLAGS"
 fi
 AM_CONDITIONAL(ISSSL, test $USE_SSL = "1")
+dnl
 dnl ------ GNU Readline
 READLINE_SHARED_LIBADD=""
 AC_CHECK_LIB(ncurses, tgetent, [READLINE_SHARED_LIBADD="-lncurses"],
@@ -135,7 +161,7 @@ AC_CHECK_LIB(readline, readline, [READLINE_LIBS="$READLINE_LIBS -lreadline $READ
 AC_CHECK_LIB(history, add_history, [READLINE_LIBS="$READLINE_LIBS -lhistory"])
 if test "$ac_cv_lib_readline_readline" = "yes"; then
 	AC_CHECK_HEADERS(readline/readline.h readline/history.h)
-        OLDLIBS=$LIBS
+        xLIBS=$LIBS
         LIBS="$LIBS $READLINE_LIBS"
         AC_TRY_LINK([
 	#include <stdio.h>
@@ -155,7 +181,7 @@ if test "$ac_cv_lib_readline_readline" = "yes"; then
                 rl_completion_matches (0, 0);
         }
         ],AC_DEFINE(HAVE_READLINE_RL_COMPLETION_MATCHES))
-        LIBS=$OLDLIBS
+        LIBS=$xLIBS
 fi
 dnl ------ various functions
 AC_CHECK_FUNCS(vsnprintf gettimeofday poll)
diff --git a/include/yaz/data1.h b/include/yaz/data1.h
index 769b512..430d413 100644
--- a/include/yaz/data1.h
+++ b/include/yaz/data1.h
@@ -23,7 +23,7 @@
  * LIABILITY, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE
  * OF THIS SOFTWARE.
  *
- * $Id: data1.h,v 1.9 2002-05-07 11:02:56 adam Exp $
+ * $Id: data1.h,v 1.10 2002-05-13 14:13:37 adam Exp $
  */
 
 #ifndef DATA1_H
@@ -43,8 +43,6 @@
 #define d1_isspace(c) strchr(" \r\n\t\f", c)
 #define d1_isdigit(c) ((c) <= '9' && (c) >= '0')
 
-#define DATA1_USING_XATTR 1
-
 YAZ_BEGIN_CDECL
 
 #define data1_matchstr(s1, s2) yaz_matchstr(s1, s2)
@@ -195,13 +193,11 @@ typedef struct data1_sub_elements {
     data1_element *elements;
 } data1_sub_elements;
 
-#if DATA1_USING_XATTR
 typedef struct data1_xattr {
     char *name;
     char *value;
     struct data1_xattr *next;
 } data1_xattr;
-#endif
 
 typedef struct data1_absyn
 {
@@ -249,9 +245,7 @@ typedef struct data1_node
 	    int get_bytes;
 	    unsigned node_selected : 1;
 	    unsigned make_variantlist : 1;
-#if DATA1_USING_XATTR
             data1_xattr *attributes;
-#endif
 	} tag;
 
 	struct
@@ -342,9 +336,11 @@ YAZ_EXPORT data1_node *data1_mk_node2(data1_handle dh, NMEM m,
                                       int type, data1_node *parent);
 
 YAZ_EXPORT data1_node *data1_mk_tag (data1_handle dh, NMEM nmem, 
-                                     const char *tag, data1_node *at);
+                                     const char *tag, const char **attr,
+                                     data1_node *at);
 YAZ_EXPORT data1_node *data1_mk_tag_n (data1_handle dh, NMEM nmem,
                                        const char *tag, size_t len,
+                                       const char **attr,
                                        data1_node *at);
 
 YAZ_EXPORT data1_node *data1_mk_text_n (data1_handle dh, NMEM mem,
@@ -414,6 +410,9 @@ YAZ_EXPORT char *data1_insert_string_n (data1_handle dh, data1_node *res,
                                         NMEM m, const char *str, size_t len);
 YAZ_EXPORT data1_node *data1_read_sgml (data1_handle dh, NMEM m,
 					const char *buf);
+YAZ_EXPORT data1_node *data1_read_xml (data1_handle dh,
+                                       int (*rf)(void *, char *, size_t),
+                                       void *fh, NMEM m);
 YAZ_EXPORT void data1_absyn_trav (data1_handle dh, void *handle,
 				  void (*fh)(data1_handle dh,
 					     void *h, data1_absyn *a));
diff --git a/retrieval/Makefile.am b/retrieval/Makefile.am
index c95ff59..3076065 100644
--- a/retrieval/Makefile.am
+++ b/retrieval/Makefile.am
@@ -1,4 +1,4 @@
-## $Id: Makefile.am,v 1.5 2002-04-15 09:44:44 adam Exp $
+## $Id: Makefile.am,v 1.6 2002-05-13 14:13:37 adam Exp $
 
 noinst_LTLIBRARIES=libret.la
 
@@ -6,7 +6,7 @@ LIBS =
 
 libret_la_SOURCES = d1_handle.c d1_read.c d1_attset.c d1_tagset.c d1_absyn.c \
   d1_grs.c d1_sutrs.c d1_varset.c d1_espec.c d1_doespec.c d1_map.c d1_marc.c \
-  d1_write.c d1_expout.c d1_sumout.c d1_soif.c d1_prtree.c d1_if.c
+  d1_write.c d1_expout.c d1_sumout.c d1_soif.c d1_prtree.c d1_if.c d1_expat.c
 
 AM_CPPFLAGS=-I$(top_srcdir)/include 
 
diff --git a/retrieval/d1_expat.c b/retrieval/d1_expat.c
new file mode 100644
index 0000000..099a922
--- /dev/null
+++ b/retrieval/d1_expat.c
@@ -0,0 +1,100 @@
+/*
+ * Copyright (c) 2002, Index Data.
+ * See the file LICENSE for details.
+ *
+ * $Id: d1_expat.c,v 1.1 2002-05-13 14:13:37 adam Exp $
+ */
+
+#if HAVE_EXPAT_H
+
+#include <assert.h>
+#include <stdio.h>
+#include <stdlib.h>
+
+#include <yaz/xmalloc.h>
+#include <yaz/log.h>
+#include <yaz/data1.h>
+
+#include <expat.h>
+
+struct user_info {
+    data1_node *d1_stack[256];
+    int level;
+    data1_handle dh;
+    NMEM nmem;
+};
+
+static void cb_start (void *user, const char *el, const char **attr)
+{
+    struct user_info *ui = (struct user_info*) user;
+    if (ui->level)
+    {
+        ui->d1_stack[ui->level] = data1_mk_tag (ui->dh, ui->nmem, el, attr,
+                                                ui->d1_stack[ui->level-1]);
+    }
+    else
+    {
+        ui->d1_stack[0] = data1_mk_root (ui->dh, ui->nmem, el);
+    }
+    ui->level++;
+}
+
+static void cb_end (void *user, const char *el)
+{
+    struct user_info *ui = (struct user_info*) user;
+
+    ui->level--;
+}
+
+static void cb_chardata (void *user, const char *s, int len)
+{
+    struct user_info *ui = (struct user_info*) user;
+    ui->d1_stack[ui->level] = data1_mk_text_n (ui->dh, ui->nmem, s, len,
+                                               ui->d1_stack[ui->level -1]);
+}
+
+#define XML_CHUNK 1024
+
+data1_node *data1_read_xml (data1_handle dh,
+                            int (*rf)(void *, char *, size_t), void *fh,
+                            NMEM m)
+{
+    XML_Parser parser;
+    struct user_info uinfo;
+    int done = 0;
+
+    uinfo.level = 0;
+    uinfo.dh = dh;
+    uinfo.d1_stack[0] = 0;
+    uinfo.nmem = m;
+
+    parser = XML_ParserCreate (0 /* encoding */);
+    
+    XML_SetElementHandler (parser, cb_start, cb_end);
+    XML_SetCharacterDataHandler (parser, cb_chardata);
+    XML_SetUserData (parser, &uinfo);
+
+    while (!done)
+    {
+        int r;
+        void *buf = XML_GetBuffer (parser, XML_CHUNK);
+        if (!buf)
+        {
+            /* error */
+            return 0;
+        }
+        r = (*rf)(fh, buf, XML_CHUNK);
+        if (r < 0)
+        {
+            /* error */
+            return 0;
+        }
+        else if (r == 0)
+            done = 1;
+        XML_ParseBuffer (parser, r, done);
+    }
+    XML_ParserFree (parser);
+    return uinfo.d1_stack[0];
+}
+
+#endif
diff --git a/retrieval/d1_grs.c b/retrieval/d1_grs.c
index 882758b..774b188 100644
--- a/retrieval/d1_grs.c
+++ b/retrieval/d1_grs.c
@@ -4,7 +4,10 @@
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: d1_grs.c,v $
- * Revision 1.19  2002-04-15 09:06:30  adam
+ * Revision 1.20  2002-05-13 14:13:37  adam
+ * XML reader for data1 (EXPAT)
+ *
+ * Revision 1.19  2002/04/15 09:06:30  adam
  * Fix explain tags for XML writer
  *
  * Revision 1.18  2002/04/12 14:40:07  adam
@@ -204,7 +207,7 @@ static Z_ElementData *nodetoelementdata(data1_handle dh, data1_node *n,
 		break;
 	    case DATA1I_text:
 	        toget = n->u.data.len;
-		if (p->u.tag.get_bytes > 0 && p->u.tag.get_bytes < toget)
+		if (p && p->u.tag.get_bytes > 0 && p->u.tag.get_bytes < toget)
 		    toget = p->u.tag.get_bytes;
 	    	res->which = Z_ElementData_string;
 		res->u.string = (char *)odr_malloc(o, toget+1);
diff --git a/retrieval/d1_prtree.c b/retrieval/d1_prtree.c
index e5df6fe..6f04412 100644
--- a/retrieval/d1_prtree.c
+++ b/retrieval/d1_prtree.c
@@ -3,7 +3,7 @@
  * See the file LICENSE for details.
  * Sebastian Hammer, Adam Dickmeiss
  *
- * $Id: d1_prtree.c,v 1.7 2002-01-26 19:20:25 adam Exp $
+ * $Id: d1_prtree.c,v 1.8 2002-05-13 14:13:37 adam Exp $
  */
 
 #include <yaz/log.h>
@@ -27,12 +27,11 @@ static void pr_tree (data1_handle dh, data1_node *n, FILE *out, int level)
     fprintf (out, "%*s", level, "");
     switch (n->which)
     {
-     case DATA1N_root:
-         fprintf (out, "root abstract syntax=%s\n", n->u.root.type);
-         break;
+    case DATA1N_root:
+        fprintf (out, "root abstract syntax=%s\n", n->u.root.type);
+        break;
     case DATA1N_tag:
 	fprintf (out, "tag type=%s\n", n->u.tag.tag);
-#if DATA1_USING_XATTR
         if (n->u.tag.attributes)
         {
             data1_xattr *xattr = n->u.tag.attributes;
@@ -41,7 +40,6 @@ static void pr_tree (data1_handle dh, data1_node *n, FILE *out, int level)
                 fprintf (out, " %s=%s ", xattr->name, xattr->value);
             fprintf (out, "\n");
         }
-#endif
 	break;
     case DATA1N_data:
 	fprintf (out, "data type=");
diff --git a/retrieval/d1_read.c b/retrieval/d1_read.c
index 21290c4..ac19a9f 100644
--- a/retrieval/d1_read.c
+++ b/retrieval/d1_read.c
@@ -3,7 +3,7 @@
  * See the file LICENSE for details.
  * Sebastian Hammer, Adam Dickmeiss
  *
- * $Id: d1_read.c,v 1.41 2002-05-07 11:02:56 adam Exp $
+ * $Id: d1_read.c,v 1.42 2002-05-13 14:13:37 adam Exp $
  */
 
 #include <assert.h>
@@ -69,9 +69,7 @@ data1_node *data1_mk_node2 (data1_handle dh, NMEM m, int type,
 	r->u.tag.node_selected = 0;
 	r->u.tag.make_variantlist = 0;
 	r->u.tag.get_bytes = -1;
-#if DATA1_USING_XATTR
 	r->u.tag.attributes = 0;
-#endif
 	break;
     case DATA1N_root:
 	r->u.root.type = 0;
@@ -124,11 +122,13 @@ data1_node *data1_mk_root (data1_handle dh, NMEM nmem,  const char *name)
 }
 
 data1_node *data1_mk_tag_n (data1_handle dh, NMEM nmem, 
-                            const char *tag, size_t len, data1_node *at)
+                            const char *tag, size_t len, const char **attr,
+                            data1_node *at)
 {
     data1_node *partag = get_parent_tag(dh, at);
     data1_node *res = data1_mk_node2 (dh, nmem, DATA1N_tag, at);
     data1_element *e = NULL;
+    data1_xattr **p;
     
     res->u.tag.tag = data1_insert_string_n (dh, res, nmem, tag, len);
     
@@ -137,13 +137,22 @@ data1_node *data1_mk_tag_n (data1_handle dh, NMEM nmem,
     res->u.tag.element =
 	data1_getelementbytagname (dh, at->root->u.root.absyn,
 				   e, res->u.tag.tag);
+    p = &res->u.tag.attributes;
+    while (attr && *attr)
+    {
+        *p = (data1_xattr*) nmem_malloc (nmem, sizeof(**p));
+        (*p)->name = nmem_strdup (nmem, attr[0]);
+        (*p)->value = nmem_strdup (nmem, attr[1]);
+        p = &(*p)->next;
+    }
+    *p = 0;
     return res;
 }
 
 data1_node *data1_mk_tag (data1_handle dh, NMEM nmem,
-                          const char *tag, data1_node *at) 
+                          const char *tag, const char **attr, data1_node *at) 
 {
-    return data1_mk_tag_n (dh, nmem, tag, strlen(tag), at);
+    return data1_mk_tag_n (dh, nmem, tag, strlen(tag), attr, at);
 }
 
 data1_node *data1_search_tag (data1_handle dh, data1_node *n,
@@ -163,7 +172,7 @@ data1_node *data1_mk_tag_uni (data1_handle dh, NMEM nmem,
 {
     data1_node *node = data1_search_tag (dh, at->child, tag);
     if (!node)
-	node = data1_mk_tag (dh, nmem, tag, at);
+	node = data1_mk_tag (dh, nmem, tag, 0 /* attr */, at);
     else
         node->child = node->last_child = 0;
     return node;
@@ -341,7 +350,6 @@ data1_node *data1_mk_tag_data_text_uni (data1_handle dh, data1_node *at,
 }
 
 
-#if DATA1_USING_XATTR
 data1_xattr *data1_read_xattr (data1_handle dh, NMEM m,
 			       int (*get_byte)(void *fh), void *fh,
 			       WRBUF wrbuf, int *ch)
@@ -405,7 +413,6 @@ data1_xattr *data1_read_xattr (data1_handle dh, NMEM m,
     *ch = c;
     return p_first;
 }
-#endif
 
 /*
  * Ugh. Sometimes functions just grow and grow on you. This one reads a
@@ -436,9 +443,8 @@ data1_node *data1_read_nodex (data1_handle dh, NMEM m,
 	
 	if (c == '<') /* beginning of tag */
 	{
-#if DATA1_USING_XATTR
 	    data1_xattr *xattr;
-#endif
+
 	    char tag[64];
 	    char args[256];
 	    int null_tag = 0;
@@ -466,17 +472,8 @@ data1_node *data1_read_nodex (data1_handle dh, NMEM m,
 		c = (*get_byte)(fh);
 	    }
 	    tag[i] = '\0';
-#if DATA1_USING_XATTR
 	    xattr = data1_read_xattr (dh, m, get_byte, fh, wrbuf, &c);
 	    args[0] = '\0';
-#else
-	    while (d1_isspace(c))
-		c = (*get_byte)(fh);
-	    for (i = 0; c && c != '>' && c != '/'; c = (*get_byte)(fh))
-		if (i < (sizeof(args)-1))
-		    args[i++] = c;
-	    args[i] = '\0';
-#endif
 	    if (c == '/')
 	    {    /* <tag attrs/> or <tag/> */
 		null_tag = 1;
@@ -568,8 +565,12 @@ data1_node *data1_read_nodex (data1_handle dh, NMEM m,
 			data1_insert_string (dh, res, m, args + val_offset);
 		}
 	    }
-	    else /* tag.. acquire our element in the abstract syntax */
-                res = data1_mk_tag (dh, m, tag, parent);
+	    else 
+            {
+                /* tag.. acquire our element in the abstract syntax */
+                res = data1_mk_tag (dh, m, tag, 0 /* attr */, parent);
+                res->u.tag.attributes = xattr;
+            }
 	    d1_stack[level] = res;
 	    d1_stack[level+1] = 0;
 	    if (level < 250 && !null_tag)
diff --git a/retrieval/d1_write.c b/retrieval/d1_write.c
index 51f2ff3..bf8b8fe 100644
--- a/retrieval/d1_write.c
+++ b/retrieval/d1_write.c
@@ -3,7 +3,7 @@
  * See the file LICENSE for details.
  * Sebastian Hammer, Adam Dickmeiss
  *
- * $Id: d1_write.c,v 1.10 2002-04-15 09:06:30 adam Exp $
+ * $Id: d1_write.c,v 1.11 2002-05-13 14:13:37 adam Exp $
  */
 
 #include <string.h>
@@ -43,13 +43,11 @@ static int nodetoidsgml(data1_node *n, int select, WRBUF b, int col)
 	    }
 	    else
 	    {
-#if DATA1_USING_XATTR
 		data1_xattr *p;
-#endif
+
 		sprintf (line, "%*s<", col, "");
 		wrbuf_puts (b, line);
 		wrbuf_puts (b, tag);
-#if DATA1_USING_XATTR
 		for (p = c->u.tag.attributes; p; p = p->next)
 		{
 		    wrbuf_putc (b, ' ');
@@ -59,7 +57,6 @@ static int nodetoidsgml(data1_node *n, int select, WRBUF b, int col)
 		    wrbuf_puts (b, p->value);
 		    wrbuf_putc (b, '"');
 		}
-#endif
 		wrbuf_puts(b, ">\n");
 		if (nodetoidsgml(c, select, b, (col > 40) ? 40 : col+2) < 0)
 		    return -1;
diff --git a/yaz-config.in b/yaz-config.in
index 2c5f686..f7117be 100644
--- a/yaz-config.in
+++ b/yaz-config.in
@@ -1,5 +1,5 @@
 #!/bin/sh
-# $Id: yaz-config.in,v 1.12 2002-03-18 12:22:00 adam Exp $
+# $Id: yaz-config.in,v 1.13 2002-05-13 14:13:37 adam Exp $
 yazprefix=@prefix@
 yaz_echo_cflags=no
 yaz_echo_libs=no
@@ -137,6 +137,8 @@ else
     fi
 fi
 
+YAZINC="$YAZINC @YAZ_CONF_CFLAGS@"
+
 if test "$lib_thread" = "yes"; then
     YAZINC="$YAZINC @CFLAGSTHREADS@"
 fi