X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=src%2Fmarcdisp.c;h=edabacdd626ad9b7d69d93c103ddd99ec4f5a37b;hb=5265b9beb2ccb4208c503ad60c56daa97a9c194d;hp=a544fcf90c9244b1a1206895510326168c1c2261;hpb=80b89a67e69a53fb3dc7b58c5abc9bf99af93139;p=yaz-moved-to-github.git

diff --git a/src/marcdisp.c b/src/marcdisp.c
index a544fcf..edabacd 100644
--- a/src/marcdisp.c
+++ b/src/marcdisp.c
@@ -1,8 +1,13 @@
 /*
- * Copyright (c) 1995-2004, Index Data
+ * Copyright (C) 1995-2005, Index Data ApS
  * See the file LICENSE for details.
  *
- * $Id: marcdisp.c,v 1.5 2004-03-16 13:15:58 adam Exp $
+ * $Id: marcdisp.c,v 1.17 2005-02-25 09:37:53 adam Exp $
+ */
+
+/**
+ * \file marcdisp.c
+ * \brief Implements MARC display - and conversion utilities
  */
 
 #if HAVE_CONFIG_H
@@ -21,6 +26,8 @@ struct yaz_marc_t_ {
     int xml;
     int debug;
     yaz_iconv_t iconv_cd;
+    char subfield_str[8];
+    char endline_str[8];
 };
 
 yaz_marc_t yaz_marc_create(void)
@@ -30,9 +37,23 @@ yaz_marc_t yaz_marc_create(void)
     mt->debug = 0;
     mt->m_wr = wrbuf_alloc();
     mt->iconv_cd = 0;
+    strcpy(mt->subfield_str, " $");
+    strcpy(mt->endline_str, "\n");
     return mt;
 }
 
+void yaz_marc_subfield_str(yaz_marc_t mt, const char *s)
+{
+    strncpy(mt->subfield_str, s, sizeof(mt->subfield_str)-1);
+    mt->subfield_str[sizeof(mt->subfield_str)-1] = '\0';
+}
+
+void yaz_marc_endline_str(yaz_marc_t mt, const char *s)
+{
+    strncpy(mt->endline_str, s, sizeof(mt->endline_str)-1);
+    mt->endline_str[sizeof(mt->endline_str)-1] = '\0';
+}
+
 void yaz_marc_destroy(yaz_marc_t mt)
 {
     if (!mt)
@@ -51,6 +72,14 @@ static void marc_cdata (yaz_marc_t mt, const char *buf, size_t len, WRBUF wr)
 	wrbuf_iconv_write_cdata(wr, mt->iconv_cd, buf, len);
 }
 
+static int atoi_n_check(const char *buf, int size, int *val)
+{
+    if (!isdigit(*(const unsigned char *) buf))
+	return 0;
+    *val = atoi_n(buf, size);
+    return 1;
+}
+
 int yaz_marc_decode_wrbuf (yaz_marc_t mt, const char *buf, int bsize, WRBUF wr)
 {
     int entry_p;
@@ -61,6 +90,14 @@ int yaz_marc_decode_wrbuf (yaz_marc_t mt, const char *buf, int bsize, WRBUF wr)
     int length_data_entry;
     int length_starting;
     int length_implementation;
+    char lead[24];
+    int produce_warnings = 0;
+
+    if (mt->debug)
+	produce_warnings = 1;
+    if (mt->xml == YAZ_MARC_SIMPLEXML || mt->xml == YAZ_MARC_OAIMARC
+	|| mt->xml == YAZ_MARC_MARCXML || mt->xml == YAZ_MARC_XCHANGE)
+	produce_warnings = 1;
 
     wrbuf_rewind(wr);
 
@@ -71,27 +108,57 @@ int yaz_marc_decode_wrbuf (yaz_marc_t mt, const char *buf, int bsize, WRBUF wr)
 	{
 	    char str[40];
 	    
-	    sprintf (str, "Record length %d - aborting\n", record_length);
-	    wrbuf_puts (wr, str);
+	    wrbuf_printf(wr, "<!-- Record length %d - aborting -->\n",
+			    record_length);
 	}
         return -1;
     }
+    memcpy(lead, buf, 24);  /* se can modify the header for output */
+
     /* ballout if bsize is known and record_length is less than that */
     if (bsize != -1 && record_length > bsize)
 	return -1;
-    if (isdigit(buf[10]))
-        indicator_length = atoi_n (buf+10, 1);
-    else
-        indicator_length = 2;
-    if (isdigit(buf[11]))
-	identifier_length = atoi_n (buf+11, 1);
-    else
+    if (!atoi_n_check(buf+10, 1, &indicator_length))
+    {
+	if (produce_warnings)
+	    wrbuf_printf(wr, "<!-- Indicator length at offset 10 should hold a digit. Assuming 2 -->\n");
+	lead[10] = '2';
+	indicator_length = 2;
+    }
+    if (!atoi_n_check(buf+11, 1, &identifier_length))
+    {
+	if (produce_warnings)
+	    wrbuf_printf(wr, "<!-- Identifier length at offset 11 should hold a digit. Assuming 2 -->\n");
+	lead[11] = '2';
         identifier_length = 2;
-    base_address = atoi_n (buf+12, 5);
-
-    length_data_entry = atoi_n (buf+20, 1);
-    length_starting = atoi_n (buf+21, 1);
-    length_implementation = atoi_n (buf+22, 1);
+    }
+    if (!atoi_n_check(buf+12, 5, &base_address))
+    {
+	if (produce_warnings)
+	    wrbuf_printf(wr, "<!-- Base address at offsets 12..16 should hold a number. Assuming 0 -->\n");
+	base_address = 0;
+    }
+    if (!atoi_n_check(buf+20, 1, &length_data_entry))
+    {
+	if (produce_warnings)
+	    wrbuf_printf(wr, "<!-- Length data entry at offset 20 should hold a digit. Assuming 4 -->\n");
+        length_data_entry = 4;
+	lead[20] = '4';
+    }
+    if (!atoi_n_check(buf+21, 1, &length_starting))
+    {
+	if (produce_warnings)
+	    wrbuf_printf(wr, "<!-- Length starting at offset 21 should hold a digit. Assuming 5 -->\n");
+        length_starting = 5;
+	lead[21] = '5';
+    }
+    if (!atoi_n_check(buf+22, 1, &length_implementation))
+    {
+	if (produce_warnings)
+	    wrbuf_printf(wr, "<!-- Length implementation at offset 22 should hold a digit. Assuming 0 -->\n");
+	length_implementation = 0;
+	lead[22] = '0';
+    }
 
     if (mt->xml != YAZ_MARC_LINE)
     {
@@ -134,7 +201,16 @@ int yaz_marc_decode_wrbuf (yaz_marc_t mt, const char *buf, int bsize, WRBUF wr)
                 wr,
                 "<record xmlns=\"http://www.loc.gov/MARC21/slim\">\n"
                 "  <leader>");
-	    marc_cdata(mt, buf, 24, wr);
+	    lead[9] = 'a';                 /* set leader to signal unicode */
+	    marc_cdata(mt, lead, 24, wr); 
+            wrbuf_printf(wr, "</leader>\n");
+            break;
+	case YAZ_MARC_XCHANGE:
+            wrbuf_printf(
+                wr,
+                "<record xmlns=\"http://www.bs.dk/standards/MarcXchange\">\n"
+                "  <leader>");
+	    marc_cdata(mt, lead, 24, wr);
             wrbuf_printf(wr, "</leader>\n");
             break;
         }
@@ -170,10 +246,11 @@ int yaz_marc_decode_wrbuf (yaz_marc_t mt, const char *buf, int bsize, WRBUF wr)
         if (entry_p >= record_length)
             return -1;
     }
-    if (mt->debug && base_address != entry_p+1)
+    if (base_address != entry_p+1)
     {
-	wrbuf_printf (wr,"  <!-- base address not at end of directory "
-		      "base=%d end=%d -->\n", base_address, entry_p+1);
+	if (produce_warnings)
+	    wrbuf_printf (wr,"  <!-- Base address not at end of directory "
+			  "base=%d end=%d -->\n", base_address, entry_p+1);
     }
     base_address = entry_p+1;
 
@@ -200,8 +277,11 @@ int yaz_marc_decode_wrbuf (yaz_marc_t mt, const char *buf, int bsize, WRBUF wr)
 	    i = data_offset + base_address;
 	    end_offset = i+data_length-1;
 	    
-	    while (buf[i] != ISO2709_RS && buf[i] != ISO2709_FS &&
-		   i < end_offset)
+	    if (data_length <= 0 || data_offset < 0 || end_offset >= record_length)
+		return -1;
+        
+	    while (i < end_offset &&
+		    buf[i] != ISO2709_RS && buf[i] != ISO2709_FS)
 		i++;
 	    sz1 = 1+i - (data_offset + base_address);
 	    if (mt->iconv_cd)
@@ -218,9 +298,9 @@ int yaz_marc_decode_wrbuf (yaz_marc_t mt, const char *buf, int bsize, WRBUF wr)
 	}
 	wrbuf_putc(wr_dir, ISO2709_FS);
 	wrbuf_printf(wr_head, "%05d", data_p+1 + base_address);
-	wrbuf_write(wr_head, buf+5, 7);
+	wrbuf_write(wr_head, lead+5, 7);
 	wrbuf_printf(wr_head, "%05d", base_address);
-	wrbuf_write(wr_head, buf+17, 7);
+	wrbuf_write(wr_head, lead+17, 7);
 
 	wrbuf_write(wr, wrbuf_buf(wr_head), 24);
 	wrbuf_write(wr, wrbuf_buf(wr_dir), wrbuf_len(wr_dir));
@@ -236,10 +316,12 @@ int yaz_marc_decode_wrbuf (yaz_marc_t mt, const char *buf, int bsize, WRBUF wr)
 	int end_offset;
 	int i, j;
 	char tag[4];
-        int identifier_flag = 1;
+        int identifier_flag = 0;
+	int entry_p0;
 
         memcpy (tag, buf+entry_p, 3);
 	entry_p += 3;
+	entry_p0 = entry_p;
         tag[3] = '\0';
 	data_length = atoi_n (buf+entry_p, length_data_entry);
 	entry_p += length_data_entry;
@@ -247,15 +329,39 @@ int yaz_marc_decode_wrbuf (yaz_marc_t mt, const char *buf, int bsize, WRBUF wr)
 	entry_p += length_starting;
 	i = data_offset + base_address;
 	end_offset = i+data_length-1;
+
+	if (data_length <= 0 || data_offset < 0 || end_offset >= record_length)
+	{
+	    if (produce_warnings)
+	        wrbuf_printf (wr,"  <!-- Bad data-offset=%d or "
+				    "data-length=%d -->\n",
+			                  data_length, data_offset);
+	    break;
+	}
         
-        if (indicator_length < 4 && indicator_length > 0)
+	if (mt->debug)
+	{
+	    wrbuf_printf(wr, "<!-- offset=%d data dlength=%d doffset=%d -->\n",
+		    entry_p0, data_length, data_offset);
+	}
+        
+        if (memcmp (tag, "00", 2))
+            identifier_flag = 1;  /* if not 00X assume subfields */
+        else if (indicator_length < 4 && indicator_length > 0)
         {
-            if (buf[i + indicator_length] != ISO2709_IDFS)
-                identifier_flag = 0;
+	    /* Danmarc 00X have subfields */
+	    if (buf[i + indicator_length] == ISO2709_IDFS)
+		identifier_flag = 1;
+	    else if (buf[i + indicator_length + 1] == ISO2709_IDFS)
+		identifier_flag = 2;
         }
-        else if (!memcmp (tag, "00", 2))
-            identifier_flag = 0;
-        
+
+	if (mt->debug)
+	{
+	    wrbuf_printf(wr, "<!-- identifier_flag = %d -->\n",
+			 identifier_flag);
+	} 
+       
         switch(mt->xml)
         {
         case YAZ_MARC_LINE:
@@ -265,23 +371,31 @@ int yaz_marc_decode_wrbuf (yaz_marc_t mt, const char *buf, int bsize, WRBUF wr)
             wrbuf_puts (wr, " ");
             break;
         case YAZ_MARC_SIMPLEXML:
-            wrbuf_printf (wr, "<field tag=\"%s\"", tag);
+            wrbuf_printf (wr, "<field tag=\"");
+	    marc_cdata(mt, tag, strlen(tag), wr);
+	    wrbuf_printf(wr, "\"");
             break;
         case YAZ_MARC_OAIMARC:
             if (identifier_flag)
-                wrbuf_printf (wr, "  <varfield id=\"%s\"", tag);
+                wrbuf_printf (wr, "  <varfield id=\"");
             else
-                wrbuf_printf (wr, "  <fixfield id=\"%s\"", tag);
+                wrbuf_printf (wr, "  <fixfield id=\"");
+	    marc_cdata(mt, tag, strlen(tag), wr);
+	    wrbuf_printf(wr, "\"");
             break;
         case YAZ_MARC_MARCXML:
+        case YAZ_MARC_XCHANGE:
             if (identifier_flag)
-                wrbuf_printf (wr, "  <datafield tag=\"%s\"", tag);
+                wrbuf_printf (wr, "  <datafield tag=\"");
             else
-                wrbuf_printf (wr, "  <controlfield tag=\"%s\"", tag);
+                wrbuf_printf (wr, "  <controlfield tag=\"");
+	    marc_cdata(mt, tag, strlen(tag), wr);
+	    wrbuf_printf(wr, "\"");
         }
         
         if (identifier_flag)
 	{
+	    i += identifier_flag-1;
             for (j = 0; j<indicator_length; j++, i++)
             {
                 switch(mt->xml)
@@ -295,18 +409,25 @@ int yaz_marc_decode_wrbuf (yaz_marc_t mt, const char *buf, int bsize, WRBUF wr)
                     wrbuf_putc(wr, buf[i]);
                     break;
                 case YAZ_MARC_SIMPLEXML:
-                    wrbuf_printf(wr, " Indicator%d=\"%c\"", j+1, buf[i]);
+                    wrbuf_printf(wr, " Indicator%d=\"", j+1);
+		    marc_cdata(mt, buf+i, 1, wr);
+                    wrbuf_printf(wr, "\"");
                     break;
                 case YAZ_MARC_OAIMARC:
-                    wrbuf_printf(wr, " i%d=\"%c\"", j+1, buf[i]);
+                    wrbuf_printf(wr, " i%d=\"", j+1);
+		    marc_cdata(mt, buf+i, 1, wr);
+                    wrbuf_printf(wr, "\"");
                     break;
                 case YAZ_MARC_MARCXML:
-                    wrbuf_printf(wr, " ind%d=\"%c\"", j+1, buf[i]);
+                case YAZ_MARC_XCHANGE:
+                    wrbuf_printf(wr, " ind%d=\"", j+1);
+		    marc_cdata(mt, buf+i, 1, wr);
+                    wrbuf_printf(wr, "\"");
                 }
             }
 	}
         if (mt->xml == YAZ_MARC_SIMPLEXML || mt->xml == YAZ_MARC_MARCXML
-	    || mt->xml == YAZ_MARC_OAIMARC)
+	    || mt->xml == YAZ_MARC_OAIMARC || mt->xml == YAZ_MARC_XCHANGE)
         {
             wrbuf_puts (wr, ">");
             if (identifier_flag)
@@ -319,7 +440,8 @@ int yaz_marc_decode_wrbuf (yaz_marc_t mt, const char *buf, int bsize, WRBUF wr)
         }
         if (identifier_flag)
         {
-            while (buf[i] != ISO2709_RS && buf[i] != ISO2709_FS && i < end_offset)
+            while (i < end_offset &&
+		    buf[i] != ISO2709_RS && buf[i] != ISO2709_FS)
             {
                 int i0;
                 i++;
@@ -332,33 +454,35 @@ int yaz_marc_decode_wrbuf (yaz_marc_t mt, const char *buf, int bsize, WRBUF wr)
 		    i += identifier_length;
 		    break;
                 case YAZ_MARC_LINE: 
-                    wrbuf_puts (wr, " $"); 
-                    for (j = 1; j<identifier_length; j++, i++)
-                        wrbuf_putc (wr, buf[i]);
+                    wrbuf_puts (wr, mt->subfield_str); 
+		    marc_cdata(mt, buf+i, identifier_length-1, wr);
+		    i = i+identifier_length-1;
                     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]);
+		    marc_cdata(mt, buf+i, identifier_length-1, wr);
+		    i = i+identifier_length-1;
                     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]);
+		    marc_cdata(mt, buf+i, identifier_length-1, wr);
+		    i = i+identifier_length-1;
                     wrbuf_puts (wr, "\">");
                     break;
                 case YAZ_MARC_MARCXML:
+                case YAZ_MARC_XCHANGE:
                     wrbuf_puts (wr, "    <subfield code=\"");
-                    for (j = 1; j<identifier_length; j++, i++)
-                        wrbuf_putc (wr, buf[i]);
+		    marc_cdata(mt, buf+i, identifier_length-1, wr);
+		    i = i+identifier_length-1;
                     wrbuf_puts (wr, "\">");
                     break;
                 }
                 i0 = i;
-                while (buf[i] != ISO2709_RS && buf[i] != ISO2709_IDFS &&
-                       buf[i] != ISO2709_FS && i < end_offset)
+                while (i < end_offset &&
+			buf[i] != ISO2709_RS && buf[i] != ISO2709_IDFS &&
+                        buf[i] != ISO2709_FS)
                     i++;
                 marc_cdata(mt, buf + i0, i - i0, wr);
 
@@ -367,6 +491,7 @@ int yaz_marc_decode_wrbuf (yaz_marc_t mt, const char *buf, int bsize, WRBUF wr)
 
 		if (mt->xml == YAZ_MARC_SIMPLEXML || 
 		    mt->xml == YAZ_MARC_MARCXML ||
+		    mt->xml == YAZ_MARC_XCHANGE ||
 		    mt->xml == YAZ_MARC_OAIMARC)
                     wrbuf_puts (wr, "</subfield>\n");
             }
@@ -374,18 +499,19 @@ int yaz_marc_decode_wrbuf (yaz_marc_t mt, const char *buf, int bsize, WRBUF wr)
         else
         {
             int i0 = i;
-            while (buf[i] != ISO2709_RS && buf[i] != ISO2709_FS && i < end_offset)
+            while (i < end_offset && 
+	        buf[i] != ISO2709_RS && buf[i] != ISO2709_FS)
                 i++;
 	    marc_cdata(mt, buf + i0, i - i0, wr);
 	    if (mt->xml == YAZ_MARC_ISO2709)
 		marc_cdata(mt, buf + i, 1, wr);
 	}
         if (mt->xml == YAZ_MARC_LINE)
-            wrbuf_putc (wr, '\n');
+            wrbuf_puts (wr, mt->endline_str);
 	if (i < end_offset)
-	    wrbuf_puts (wr, "  <!-- separator but not at end of field -->\n");
+	    wrbuf_printf(wr, "  <!-- separator but not at end of field length=%d-->\n", data_length);
 	if (buf[i] != ISO2709_RS && buf[i] != ISO2709_FS)
-	    wrbuf_puts (wr, "  <!-- no separator at end of field -->\n");
+	    wrbuf_printf(wr, "  <!-- no separator at end of field length=%d-->\n", data_length);
         switch(mt->xml)
         {
         case YAZ_MARC_SIMPLEXML:
@@ -393,15 +519,16 @@ int yaz_marc_decode_wrbuf (yaz_marc_t mt, const char *buf, int bsize, WRBUF wr)
             break;
         case YAZ_MARC_OAIMARC:
             if (identifier_flag)
-                wrbuf_puts (wr, "  </varfield>\n");
+                wrbuf_puts (wr, "</varfield>\n");
             else
-                wrbuf_puts (wr, "  </fixfield>\n");
+                wrbuf_puts (wr, "</fixfield>\n");
             break;
         case YAZ_MARC_MARCXML:
+        case YAZ_MARC_XCHANGE:
             if (identifier_flag)
                 wrbuf_puts (wr, "  </datafield>\n");
             else
-                wrbuf_puts (wr, "  </controlfield>\n");
+                wrbuf_puts (wr, "</controlfield>\n");
             break;
         }
     }
@@ -417,6 +544,7 @@ int yaz_marc_decode_wrbuf (yaz_marc_t mt, const char *buf, int bsize, WRBUF wr)
         wrbuf_puts (wr, "</oai_marc>\n");
         break;
     case YAZ_MARC_MARCXML:
+    case YAZ_MARC_XCHANGE:
         wrbuf_puts (wr, "</record>\n");
         break;
     case YAZ_MARC_ISO2709: