X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=util%2Fmarcdisp.c;h=9156281d8dd8214f7c454f9b0b20561194d6835f;hb=c39a893dfdae5f792139177132e7e7a70e010aa7;hp=b28c88e52a85176ca36cfb365fe266fc5700b5aa;hpb=a57b3c893b5eac42f8351f39331c7188e3abcd99;p=yaz-moved-to-github.git

diff --git a/util/marcdisp.c b/util/marcdisp.c
index b28c88e..9156281 100644
--- a/util/marcdisp.c
+++ b/util/marcdisp.c
@@ -1,58 +1,8 @@
 /*
- * Copyright (c) 1995-2001, Index Data
+ * Copyright (c) 1995-2003, Index Data
  * See the file LICENSE for details.
  *
- * $Log: marcdisp.c,v $
- * Revision 1.16  2002-01-22 10:54:46  adam
- * MARC decode fix. Attribute set fix for scan in server. Prox logging.
- *
- * Revision 1.15  2001/10/29 09:17:19  adam
- * New function marc_display_exl - used by YAZ client. Server returns
- * bad record on position 98 (for testing).
- *
- * Revision 1.14  2001/10/23 21:00:20  adam
- * Old Z39.50 codecs gone. Added ZOOM. WRBUF MARC display util.
- *
- * Revision 1.13  2001/10/15 19:36:48  adam
- * New function marc_display_wrbuf.
- *
- * Revision 1.12  2000/10/02 11:07:44  adam
- * Added peer_name member for bend_init handler. Changed the YAZ
- * client so that tcp: can be avoided in target spec.
- *
- * Revision 1.11  2000/02/29 13:44:55  adam
- * Check for config.h (currently not generated).
- *
- * Revision 1.10  2000/02/05 10:47:19  adam
- * Identifier-length and indicator-lenght no longer set to 2 (forced).
- *
- * Revision 1.9  1999/12/21 16:24:48  adam
- * More robust ISO2709 handling (in case of real bad formats).
- *
- * Revision 1.8  1999/11/30 13:47:12  adam
- * Improved installation. Moved header files to include/yaz.
- *
- * Revision 1.7  1997/09/24 13:29:40  adam
- * Added verbose option -v to marcdump utility.
- *
- * Revision 1.6  1997/09/04 07:52:27  adam
- * Moved atoi_n function to separate source file.
- *
- * Revision 1.5  1997/05/01 15:08:15  adam
- * Added log_mask_str_x routine.
- *
- * Revision 1.4  1995/09/29 17:12:34  quinn
- * Smallish
- *
- * Revision 1.3  1995/09/27  15:03:03  quinn
- * Modified function heads & prototypes.
- *
- * Revision 1.2  1995/05/16  08:51:12  quinn
- * License, documentation, and memory fixes
- *
- * Revision 1.1  1995/04/10  10:28:46  quinn
- * Added copy of CCL and MARC display
- *
+ * $Id: marcdisp.c,v 1.29 2003-02-12 22:02:47 adam Exp $
  */
 
 #if HAVE_CONFIG_H
@@ -66,8 +16,85 @@
 #include <yaz/wrbuf.h>
 #include <yaz/yaz-util.h>
 
-int marc_display_wrbuf (const char *buf, WRBUF wr, int debug,
-			int bsize)
+struct yaz_marc_t_ {
+    WRBUF m_wr;
+    int xml;
+    int debug;
+};
+
+yaz_marc_t yaz_marc_create(void)
+{
+    yaz_marc_t mt = (yaz_marc_t) xmalloc(sizeof(*mt));
+    mt->xml = YAZ_MARC_LINE;
+    mt->debug = 0;
+    mt->m_wr = wrbuf_alloc();
+    return mt;
+}
+
+void yaz_marc_destroy(yaz_marc_t mt)
+{
+    if (!mt)
+        return ;
+    wrbuf_free (mt->m_wr, 1);
+    xfree (mt);
+}
+
+static void marc_cdata (yaz_marc_t mt, const char *buf, size_t len, WRBUF wr)
+{
+    size_t i;
+    for (i = 0; i<len; i++)
+    {
+        if (mt->xml)
+        {
+            switch (buf[i]) {
+            case '<':
+                wrbuf_puts(wr, "&lt;");
+                break;
+            case '>':
+                wrbuf_puts(wr, "&gt;");
+                break;
+            case '&':
+                wrbuf_puts(wr, "&amp;");
+                break;
+            default:
+                wrbuf_putc(wr, buf[i]);
+            }
+        }
+        else
+            wrbuf_putc(wr, buf[i]);
+    }
+}
+
+#if 0
+static void marc_cdata (yaz_marc_t mt, const char *buf, size_t len)
+{
+    if (!mt->cd)
+        marc_cdata2 (mt, buf, len);
+    else
+    {
+        char outbuf[12];
+        size_t inbytesleft = len;
+        const char *inp = buf;
+        
+        while (inbytesleft)
+        {
+            size_t outbytesleft = sizeof(outbuf);
+            char *outp = outbuf;
+            size_t r = yaz_iconv (mt->cd, (char**) &inp, &inbytesleft, 
+                                  &outp, &outbytesleft);
+            if (r == (size_t) (-1))
+            {
+                int e = yaz_iconv_error(mt->cd);
+                if (e != YAZ_ICONV_E2BIG)
+                    break;
+            }
+            marc_cdata2 (mt, outbuf, outp - outbuf);
+        }
+    }
+}
+#endif
+
+int yaz_marc_decode_wrbuf (yaz_marc_t mt, const char *buf, int bsize, WRBUF wr)
 {
     int entry_p;
     int record_length;
@@ -78,10 +105,12 @@ int marc_display_wrbuf (const char *buf, WRBUF wr, int debug,
     int length_starting;
     int length_implementation;
 
+    wrbuf_rewind(wr);
+
     record_length = atoi_n (buf, 5);
     if (record_length < 25)
     {
-	if (debug)
+	if (mt->debug)
 	{
 	    char str[40];
 	    
@@ -107,9 +136,54 @@ int marc_display_wrbuf (const char *buf, WRBUF wr, int debug,
     length_starting = atoi_n (buf+21, 1);
     length_implementation = atoi_n (buf+22, 1);
 
-    if (debug)
+    if (mt->xml)
+    {
+        char str[80];
+        int i;
+        switch(mt->xml)
+        {
+        case YAZ_MARC_SIMPLEXML:
+            wrbuf_puts (wr, "<iso2709\n");
+            sprintf (str, " RecordStatus=\"%c\"\n", buf[5]);
+            wrbuf_puts (wr, str);
+            sprintf (str, " TypeOfRecord=\"%c\"\n", buf[6]);
+            wrbuf_puts (wr, str);
+            for (i = 1; i<=19; i++)
+            {
+                sprintf (str, " ImplDefined%d=\"%c\"\n", i, buf[6+i]);
+                wrbuf_puts (wr, str);
+            }
+            wrbuf_puts (wr, ">\n");
+            break;
+        case YAZ_MARC_OAIMARC:
+            wrbuf_puts(
+                wr,
+                "<oai_marc xmlns=\"http://www.openarchives.org/OIA/oai_marc\""
+                "\n"
+                " xmlns:xsi=\"http://www.w3.org/2000/10/XMLSchema-instance\""
+                "\n"
+                " xsi:schemaLocation=\"http://www.openarchives.org/OAI/oai_marc.xsd\""
+                "\n"
+                );
+            
+            sprintf (str, " status=\"%c\" type=\"%c\" catForm=\"%c\">\n",
+                     buf[5], buf[6], buf[7]);
+            wrbuf_puts (wr, str);
+            break;
+        case YAZ_MARC_MARCXML:
+            wrbuf_printf(
+                wr,
+                "<record xmlns=\"http://www.loc.gov/MARC21/slim\">\n"
+                "  <leader>%.24s</leader>\n", buf);
+            break;
+        }
+    }
+    if (mt->debug)
     {
 	char str[40];
+
+        if (mt->xml)
+            wrbuf_puts (wr, "<!--\n");
 	sprintf (str, "Record length         %5d\n", record_length);
 	wrbuf_puts (wr, str);
 	sprintf (str, "Indicator length      %5d\n", indicator_length);
@@ -124,7 +198,10 @@ int marc_display_wrbuf (const char *buf, WRBUF wr, int debug,
 	wrbuf_puts (wr, str);
 	sprintf (str, "Length implementation %5d\n", length_implementation);
 	wrbuf_puts (wr, str);
+        if (mt->xml)
+            wrbuf_puts (wr, "-->\n");
     }
+
     for (entry_p = 24; buf[entry_p] != ISO2709_FS; )
     {
         entry_p += 3+length_data_entry+length_starting;
@@ -139,84 +216,246 @@ int marc_display_wrbuf (const char *buf, WRBUF wr, int debug,
 	int end_offset;
 	int i, j;
 	char tag[4];
+        int identifier_flag = 1;
 
         memcpy (tag, buf+entry_p, 3);
 	entry_p += 3;
         tag[3] = '\0';
-	if (debug)
-	    wrbuf_puts (wr, "Tag: ");
-	wrbuf_puts (wr, tag);
-	wrbuf_puts (wr, " ");
 	data_length = atoi_n (buf+entry_p, length_data_entry);
 	entry_p += length_data_entry;
 	data_offset = atoi_n (buf+entry_p, length_starting);
 	entry_p += length_starting;
 	i = data_offset + base_address;
 	end_offset = i+data_length-1;
-	if (debug)
-	    wrbuf_puts (wr, " Ind: ");
-        if (memcmp (tag, "00", 2) && indicator_length)
+        
+        if (indicator_length < 4 && indicator_length > 0)
+        {
+            if (buf[i + indicator_length] != ISO2709_IDFS)
+                identifier_flag = 0;
+        }
+        else if (!memcmp (tag, "00", 2))
+            identifier_flag = 0;
+        
+        switch(mt->xml)
+        {
+        case YAZ_MARC_LINE:
+            if (mt->debug)
+                wrbuf_puts (wr, "Tag: ");
+            wrbuf_puts (wr, tag);
+            wrbuf_puts (wr, " ");
+            break;
+        case YAZ_MARC_SIMPLEXML:
+            wrbuf_printf (wr, "<field tag=\"%s\"", tag);
+            break;
+        case YAZ_MARC_OAIMARC:
+            if (identifier_flag)
+                wrbuf_printf (wr, "  <varfield id=\"%s\"", tag);
+            else
+                wrbuf_printf (wr, "  <fixfield id=\"%s\"", tag);
+            break;
+        case YAZ_MARC_MARCXML:
+            if (identifier_flag)
+                wrbuf_printf (wr, "  <datafield tag=\"%s\"", tag);
+            else
+                wrbuf_printf (wr, "  <controlfield tag=\"%s\"", tag);
+        }
+        
+        if (identifier_flag)
 	{
             for (j = 0; j<indicator_length; j++, i++)
-		wrbuf_putc (wr, buf[i]);
+            {
+                switch(mt->xml)
+                {
+                case YAZ_MARC_LINE:
+                    if (mt->debug)
+                        wrbuf_puts (wr, " Ind: ");
+                    wrbuf_putc (wr, buf[i]);
+                    break;
+                case YAZ_MARC_SIMPLEXML:
+                    wrbuf_printf (wr, " Indicator%d=\"%c\"", j+1, buf[i]);
+                    break;
+                case YAZ_MARC_OAIMARC:
+                    wrbuf_printf (wr, " i%d=\"%c\"", j+1, buf[i]);
+                    break;
+                case YAZ_MARC_MARCXML:
+                    wrbuf_printf (wr, " ind%d=\"%c\"", j+1, buf[i]);
+                }
+            }
 	}
-	if (debug)
-	    wrbuf_puts (wr, " Fields: ");
-	while (buf[i] != ISO2709_RS && buf[i] != ISO2709_FS && i < end_offset)
-	{
-            if (memcmp (tag, "00", 2) && identifier_length)
-	    {
-	        i++;
-		wrbuf_puts (wr, " $"); 
-                for (j = 1; j<identifier_length; j++, i++)
-		    wrbuf_putc (wr, buf[i]);
-		wrbuf_putc (wr, ' ');
-	        while (buf[i] != ISO2709_RS && buf[i] != ISO2709_IDFS &&
-	               buf[i] != ISO2709_FS && i < end_offset)
-		{
-		    wrbuf_putc (wr, buf[i]);
-		    i++;
-		}
-	    }
-	    else
-	    {
-		wrbuf_putc (wr, buf[i]);
-		i++;
-	    }
+        if (mt->xml)
+        {
+            wrbuf_puts (wr, ">");
+            if (identifier_flag)
+                wrbuf_puts (wr, "\n");
+        }
+        else
+        {
+            if (mt->debug && !mt->xml)
+                wrbuf_puts (wr, " Fields: ");
+        }
+        if (identifier_flag)
+        {
+            while (buf[i] != ISO2709_RS && buf[i] != ISO2709_FS && i < end_offset)
+            {
+                int i0;
+                i++;
+                switch(mt->xml)
+                {
+                case YAZ_MARC_LINE: 
+                    wrbuf_puts (wr, " $"); 
+                    for (j = 1; j<identifier_length; j++, i++)
+                        wrbuf_putc (wr, buf[i]);
+                    wrbuf_putc (wr, ' ');
+                    break;
+                case YAZ_MARC_SIMPLEXML:
+                    wrbuf_puts (wr, "  <subfield code=\"");
+                    for (j = 1; j<identifier_length; j++, i++)
+                        wrbuf_putc (wr, buf[i]);
+                    wrbuf_puts (wr, "\">");
+                    break;
+                case YAZ_MARC_OAIMARC:
+                    wrbuf_puts (wr, "    <subfield label=\"");
+                    for (j = 1; j<identifier_length; j++, i++)
+                        wrbuf_putc (wr, buf[i]);
+                    wrbuf_puts (wr, "\">");
+                    break;
+                case YAZ_MARC_MARCXML:
+                    wrbuf_puts (wr, "    <subfield code=\"");
+                    for (j = 1; j<identifier_length; j++, i++)
+                        wrbuf_putc (wr, buf[i]);
+                    wrbuf_puts (wr, "\">");
+                    break;
+                }
+                i0 = i;
+                while (buf[i] != ISO2709_RS && buf[i] != ISO2709_IDFS &&
+                       buf[i] != ISO2709_FS && i < end_offset)
+                    i++;
+                marc_cdata(mt, buf + i0, i - i0, wr);
+                
+                if (mt->xml)
+                    wrbuf_puts (wr, "</subfield>\n");
+            }
+        }
+        else
+        {
+            int i0 = i;
+            while (buf[i] != ISO2709_RS && buf[i] != ISO2709_FS && i < end_offset)
+                i++;
+            marc_cdata(mt, buf + i0, i - i0, wr);
 	}
-	wrbuf_putc (wr, '\n');
+        if (!mt->xml)
+            wrbuf_putc (wr, '\n');
 	if (i < end_offset)
-	    wrbuf_puts (wr, "-- separator but not at end of field\n");
+	    wrbuf_puts (wr, "  <!-- separator but not at end of field -->\n");
 	if (buf[i] != ISO2709_RS && buf[i] != ISO2709_FS)
-	    wrbuf_puts (wr, "-- no separator at end of field\n");
+	    wrbuf_puts (wr, "  <!-- no separator at end of field -->\n");
+        switch(mt->xml)
+        {
+        case YAZ_MARC_SIMPLEXML:
+            wrbuf_puts (wr, "</field>\n");
+            break;
+        case YAZ_MARC_OAIMARC:
+            if (identifier_flag)
+                wrbuf_puts (wr, "  </varfield>\n");
+            else
+                wrbuf_puts (wr, "  </fixfield>\n");
+            break;
+        case YAZ_MARC_MARCXML:
+            if (identifier_flag)
+                wrbuf_puts (wr, "  </datafield>\n");
+            else
+                wrbuf_puts (wr, "  </controlfield>\n");
+            break;
+        }
+    }
+    switch (mt->xml)
+    {
+    case YAZ_MARC_LINE:
+        wrbuf_puts (wr, "");
+        break;
+    case YAZ_MARC_SIMPLEXML:
+        wrbuf_puts (wr, "</iso2709>\n");
+        break;
+    case YAZ_MARC_OAIMARC:
+        wrbuf_puts (wr, "</oai_marc>\n");
+        break;
+    case YAZ_MARC_MARCXML:
+        wrbuf_puts (wr, "</record>\n");
+        break;
     }
-    wrbuf_puts (wr, "");
     return record_length;
 }
 
-int marc_display_exl (const char *buf, FILE *outf, int debug, int length)
+int yaz_marc_decode_buf (yaz_marc_t mt, const char *buf, int bsize,
+                         char **result, int *rsize)
 {
-    int record_length;
+    int r = yaz_marc_decode_wrbuf(mt, buf, bsize, mt->m_wr);
+    if (r > 0)
+    {
+        if (result)
+            *result = wrbuf_buf(mt->m_wr);
+        if (rsize)
+            *rsize = wrbuf_len(mt->m_wr);
+    }
+    return r;
+}
+
+void yaz_marc_xml(yaz_marc_t mt, int xmlmode)
+{
+    if (mt)
+        mt->xml = xmlmode;
+}
+
+void yaz_marc_debug(yaz_marc_t mt, int level)
+{
+    if (mt)
+        mt->debug = level;
+}
+
+/* depricated */
+int yaz_marc_decode(const char *buf, WRBUF wr, int debug, int bsize, int xml)
+{
+    yaz_marc_t mt = yaz_marc_create();
+    int r;
+
+    mt->debug = debug;
+    mt->xml = xml;
+    r = yaz_marc_decode_wrbuf(mt, buf, bsize, wr);
+    yaz_marc_destroy(mt);
+    return r;
+}
 
-    WRBUF wrbuf = wrbuf_alloc ();
-    record_length = marc_display_wrbuf (buf, wrbuf, debug, length);
+/* depricated */
+int marc_display_wrbuf (const char *buf, WRBUF wr, int debug, int bsize)
+{
+    return yaz_marc_decode(buf, wr, debug, bsize, 0);
+}
+
+/* depricated */
+int marc_display_exl (const char *buf, FILE *outf, int debug, int bsize)
+{
+    yaz_marc_t mt = yaz_marc_create();
+    int r;
+
+    mt->debug = debug;
+    r = yaz_marc_decode_wrbuf (mt, buf, bsize, mt->m_wr);
     if (!outf)
 	outf = stdout;
-    if (record_length > 0)
-	fwrite (wrbuf_buf(wrbuf), 1, wrbuf_len(wrbuf), outf);
-    wrbuf_free (wrbuf, 1);
-    return record_length;
+    if (r > 0)
+	fwrite (wrbuf_buf(mt->m_wr), 1, wrbuf_len(mt->m_wr), outf);
+    yaz_marc_destroy(mt);
+    return r;
 }
 
-
+/* depricated */
 int marc_display_ex (const char *buf, FILE *outf, int debug)
 {
     return marc_display_exl (buf, outf, debug, -1);
 }
 
+/* depricated */
 int marc_display (const char *buf, FILE *outf)
 {
     return marc_display_ex (buf, outf, 0);
 }
 
-