X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=util%2Fmarcdisp.c;h=b7288727df874a332867779ff9287773078c0b41;hb=4b3d1e197d75fcb2607be8d86d244f20c4180000;hp=f728d12a699ac2754929a71365ac07c661fc648e;hpb=08e514143cd7d16db15c1c3d4d15d687f561c047;p=yaz-moved-to-github.git

diff --git a/util/marcdisp.c b/util/marcdisp.c
index f728d12..b728872 100644
--- a/util/marcdisp.c
+++ b/util/marcdisp.c
@@ -1,8 +1,8 @@
 /*
- * Copyright (c) 1995-2002, Index Data
+ * Copyright (c) 1995-2003, Index Data
  * See the file LICENSE for details.
  *
- * $Id: marcdisp.c,v 1.18 2002-02-28 13:21:16 adam Exp $
+ * $Id: marcdisp.c,v 1.30 2003-07-14 12:59:23 adam Exp $
  */
 
 #if HAVE_CONFIG_H
@@ -16,7 +16,91 @@
 #include <yaz/wrbuf.h>
 #include <yaz/yaz-util.h>
 
-int yaz_marc_decode (const char *buf, WRBUF wr, int debug, int bsize, int xml)
+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;
+	    case '"':
+		wrbuf_puts(wr, "&quot;");
+		break;
+	    case '\'':
+		wrbuf_puts(wr, "&apos;");
+		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;
@@ -27,10 +111,12 @@ int yaz_marc_decode (const char *buf, WRBUF wr, int debug, int bsize, int xml)
     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];
 	    
@@ -56,9 +142,54 @@ int yaz_marc_decode (const char *buf, WRBUF wr, int debug, int bsize, int xml)
     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);
@@ -73,7 +204,10 @@ int yaz_marc_decode (const char *buf, WRBUF wr, int debug, int bsize, int xml)
 	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;
@@ -93,19 +227,6 @@ int yaz_marc_decode (const char *buf, WRBUF wr, int debug, int bsize, int xml)
         memcpy (tag, buf+entry_p, 3);
 	entry_p += 3;
         tag[3] = '\0';
-        if (xml)
-        {
-            wrbuf_puts (wr, "<field name=\"");
-            wrbuf_puts (wr, tag);
-            wrbuf_puts (wr, "\"");
-        }
-        else
-        {
-            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);
@@ -121,23 +242,53 @@ int yaz_marc_decode (const char *buf, WRBUF wr, int debug, int bsize, int xml)
         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)
 	{
-            if (debug)
-                wrbuf_puts (wr, " Ind: ");
             for (j = 0; j<indicator_length; j++, i++)
             {
-                if (xml)
+                switch(mt->xml)
                 {
-                    char nostr[30];
-                    sprintf (nostr, " indicator%d=\"%c\"", j+1, buf[i]);
-                    wrbuf_puts (wr, nostr);
-                }
-                else
+                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 (xml)
+        if (mt->xml)
         {
             wrbuf_puts (wr, ">");
             if (identifier_flag)
@@ -145,84 +296,172 @@ int yaz_marc_decode (const char *buf, WRBUF wr, int debug, int bsize, int xml)
         }
         else
         {
-            if (debug)
+            if (mt->debug && !mt->xml)
                 wrbuf_puts (wr, " Fields: ");
         }
-	while (buf[i] != ISO2709_RS && buf[i] != ISO2709_FS && i < end_offset)
-	{
-            if (identifier_flag)
-	    {
-	        i++;
-                if (xml)
+        if (identifier_flag)
+        {
+            while (buf[i] != ISO2709_RS && buf[i] != ISO2709_FS && i < end_offset)
+            {
+                int i0;
+                i++;
+                switch(mt->xml)
                 {
-                    wrbuf_puts (wr, "  <field name=\"");
+                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, "\">");
-                }
-                else
-                {
-                    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_putc (wr, ' ');
+                    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;
                 }
-	        while (buf[i] != ISO2709_RS && buf[i] != ISO2709_IDFS &&
-	               buf[i] != ISO2709_FS && i < end_offset)
-		{
-		    wrbuf_putc (wr, buf[i]);
-		    i++;
-		}
-                if (xml)
-                    wrbuf_puts (wr, "</field>\n");
-	    }
-	    else
-	    {
-		wrbuf_putc (wr, buf[i]);
-		i++;
-	    }
+                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);
 	}
-        if (!xml)
+        if (!mt->xml)
             wrbuf_putc (wr, '\n');
 	if (i < end_offset)
 	    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");
-        if (xml)
+        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_wrbuf (const char *buf, WRBUF wr, int debug,
-                        int bsize)
+int yaz_marc_decode_buf (yaz_marc_t mt, const char *buf, int bsize,
+                         char **result, int *rsize)
 {
-    return yaz_marc_decode (buf, wr, debug, bsize, 0);
+    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;
 }
 
-int marc_display_exl (const char *buf, FILE *outf, int debug, int length)
+void yaz_marc_xml(yaz_marc_t mt, int xmlmode)
 {
-    int record_length;
+    if (mt)
+        mt->xml = xmlmode;
+}
 
-    WRBUF wrbuf = wrbuf_alloc ();
-    record_length = marc_display_wrbuf (buf, wrbuf, debug, length);
+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;
+}
+
+/* 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);
 }
 
-