refdb-cvs

[Markus H. <mho...@us...>]

Update of /cvsroot/refdb/refdb/src
In directory sc8-pr-cvs1:/tmp/cvs-serv28035

Modified Files:
	refdbdbib.c 
Log Message:
added character encoding conversion

Index: refdbdbib.c
===================================================================
RCS file: /cvsroot/refdb/refdb/src/refdbdbib.c,v
retrieving revision 1.29
retrieving revision 1.30
diff -u -U2 -r1.29 -r1.30
--- refdbdbib.c	3 Jan 2004 20:11:43 -0000	1.29
+++ refdbdbib.c	18 Jan 2004 21:17:37 -0000	1.30
@@ -34,4 +34,5 @@
 #include <arpa/inet.h>
 #include <expat.h>
+#include <iconv.h>
 #include <dbi/dbi.h>
 
@@ -96,5 +97,5 @@
 
   ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++*/
-int getbib(struct CLIENT_REQUEST* ptr_clrequest, struct bibinfo* biblio_info, char *table_name, int n_ref_format, int n_send_stylespec, struct ADDRESULT* ptr_addresult) {
+int getbib(struct CLIENT_REQUEST* ptr_clrequest, struct bibinfo* ptr_biblio_info, char *table_name, int n_ref_format, int n_send_stylespec, struct ADDRESULT* ptr_addresult) {
   XML_Parser p;
   struct getbib_data gbdata;
@@ -156,6 +157,8 @@
   const char* item;
   const char* item1;
+  const char* db_encoding;
   struct BIBCONNS bibconns;
   struct lilimem sentinel;
+  iconv_t conv_descriptor;
 
   sentinel.ptr_mem = NULL;
@@ -163,6 +166,6 @@
   sentinel.varname[0] = '\0';
 
-  biblio_info->entry_id = NULL;
-  biblio_info->xreflabel = xreflabel;
+  ptr_biblio_info->entry_id = NULL;
+  ptr_biblio_info->xreflabel = xreflabel;
 
   bibconns.conn = NULL; /* connection to the default reference database */
@@ -170,5 +173,5 @@
   bibconns.conn_source = NULL; /* connection to source reference database */
 
-  nref_counter = biblio_info->n_startnumber - 1; /* numbering starts here */
+  nref_counter = ptr_biblio_info->n_startnumber - 1; /* numbering starts here */
 
   sql_command_len = 4096;
@@ -254,9 +257,42 @@
     }
       
-    myjournal = strdup(biblio_info->format_string);
+
+
+    /* get the database encoding */
+    db_encoding = dbi_conn_get_encoding(bibconns.conn);
+
+    /* if we need to convert, create a conversion descriptor for iconv() */
+    if (db_encoding && *(ptr_biblio_info->encoding) && strcmp(db_encoding, ptr_biblio_info->encoding)) {
+      conv_descriptor = iconv_open(!strcmp(ptr_biblio_info->encoding, "US-ASCII") ? "ASCII" : ptr_biblio_info->encoding, !strcmp(db_encoding, "US-ASCII") ? "ASCII" : db_encoding);
+      if (conv_descriptor == (iconv_t)(-1)) {
+	LOG_PRINT(LOG_WARNING, "cannot set conversion descriptor");
+	iwrite(ptr_clrequest->fd, "cannot set conversion descriptor\n", 33);
+	delete_all_lilimem(&sentinel);
+	close_dbi_connections(&bibconns, drivername);
+	return 0;
+      }
+      else {
+	LOG_PRINT(LOG_DEBUG, "database encoding is:");
+	LOG_PRINT(LOG_DEBUG, db_encoding);
+      }
+    }
+    else {
+      conv_descriptor = NULL;
+      LOG_PRINT(LOG_DEBUG, "no character encoding conversion required");
+    }
+    
+    LOG_PRINT(LOG_DEBUG, "output encoding is:");
+    LOG_PRINT(LOG_DEBUG, ptr_biblio_info->encoding);
+
+
+
+    myjournal = strdup(ptr_biblio_info->format_string);
     if (!myjournal || insert_lilimem(&sentinel, (void**)&myjournal, NULL)) {
       LOG_PRINT(LOG_WARNING, outomem.text);
       delete_all_lilimem(&sentinel);
       close_dbi_connections(&bibconns, drivername);
+      if (conv_descriptor) {
+	iconv_close(conv_descriptor);
+      }
       return 0;
     }
@@ -277,4 +313,7 @@
 	close_dbi_connections(&bibconns, drivername);
 	LOG_PRINT(LOG_WARNING, "could not write to client");
+	if (conv_descriptor) {
+	  iconv_close(conv_descriptor);
+	}
 	return 0;
       }
@@ -289,4 +328,7 @@
 	  close_dbi_connections(&bibconns, drivername);
 	  delete_all_lilimem(&sentinel);
+	  if (conv_descriptor) {
+	    iconv_close(conv_descriptor);
+	  }
 	  return 0;
 	}
@@ -294,4 +336,7 @@
 	close_dbi_connections(&bibconns, drivername);
 	delete_all_lilimem(&sentinel);
+	if (conv_descriptor) {
+	  iconv_close(conv_descriptor);
+	}
 	return 0;
       }
@@ -301,4 +346,7 @@
 	close_dbi_connections(&bibconns, drivername);
 	delete_all_lilimem(&sentinel);
+	if (conv_descriptor) {
+	  iconv_close(conv_descriptor);
+	}
 	return 0;
       }
@@ -307,4 +355,7 @@
 	close_dbi_connections(&bibconns, drivername);
 	delete_all_lilimem(&sentinel);
+	if (conv_descriptor) {
+	  iconv_close(conv_descriptor);
+	}
 	return 0;
       }
@@ -379,4 +430,7 @@
       close_dbi_connections(&bibconns, drivername);
       delete_all_lilimem(&sentinel);
+      if (conv_descriptor) {
+	iconv_close(conv_descriptor);
+      }
       return 0;
     }
@@ -398,4 +452,7 @@
       close_dbi_connections(&bibconns, drivername);
       delete_all_lilimem(&sentinel);
+      if (conv_descriptor) {
+	iconv_close(conv_descriptor);
+      }
       return 0;
     }
@@ -425,4 +482,7 @@
       close_dbi_connections(&bibconns, drivername);
       delete_all_lilimem(&sentinel);
+      if (conv_descriptor) {
+	iconv_close(conv_descriptor);
+      }
       return 0;
     }
@@ -443,4 +503,7 @@
       delete_all_lilimem(&sentinel);
       close_dbi_connections(&bibconns, drivername);
+      if (conv_descriptor) {
+	iconv_close(conv_descriptor);
+      }
       return 0;
     }
@@ -450,4 +513,7 @@
       delete_all_lilimem(&sentinel);
       close_dbi_connections(&bibconns, drivername);
+      if (conv_descriptor) {
+	iconv_close(conv_descriptor);
+      }
       return 0;
     }      
@@ -460,4 +526,7 @@
       delete_all_lilimem(&sentinel);
       close_dbi_connections(&bibconns, drivername);
+      if (conv_descriptor) {
+	iconv_close(conv_descriptor);
+      }
       return 0;
     }      
@@ -477,4 +546,7 @@
       delete_all_lilimem(&sentinel);
       close_dbi_connections(&bibconns, drivername);
+      if (conv_descriptor) {
+	iconv_close(conv_descriptor);
+      }
       return 0;
     }
@@ -485,4 +557,7 @@
       delete_all_lilimem(&sentinel);
       close_dbi_connections(&bibconns, drivername);
+      if (conv_descriptor) {
+	iconv_close(conv_descriptor);
+      }
       return 0;
     }
@@ -520,5 +595,5 @@
     /* write doctype line etc */
     
-    prepare_render_dbib(outbuffer, biblio_info, bibconns.conn_refdb, n_ref_format);
+    prepare_render_dbib(outbuffer, ptr_biblio_info, bibconns.conn_refdb, n_ref_format);
 
     bibsequence = dbi_result_get_string(dbires, "BIBLIOSEQUENCE");
@@ -637,5 +712,5 @@
     
     if (!nis_numeric) { /* only useful for author/year style */
-      if (fill_in_authoryear_info(&bibconns, dbires, dbires_cit, ptr_clrequest, ptr_addresult, biblio_info, table_name, intext_string, current_mset, &nref_counter)) {
+      if (fill_in_authoryear_info(&bibconns, dbires, dbires_cit, ptr_clrequest, ptr_addresult, ptr_biblio_info, table_name, intext_string, current_mset, &nref_counter)) {
 	LOG_PRINT(LOG_DEBUG, "fill in author/year info failed");
 	dbi_result_free(dbires_cit);
@@ -769,6 +844,6 @@
       }
       
-      biblio_info->entry_id = my_dbi_result_get_string_copy(dbires, "entry_id");
-      if (!biblio_info->entry_id || !strcmp(biblio_info->entry_id, "ERROR") || insert_lilimem(&sentinel, (void**)&(biblio_info->entry_id), "id_string")) {
+      ptr_biblio_info->entry_id = my_dbi_result_get_string_copy(dbires, "entry_id");
+      if (!ptr_biblio_info->entry_id || !strcmp(ptr_biblio_info->entry_id, "ERROR") || insert_lilimem(&sentinel, (void**)&(ptr_biblio_info->entry_id), "id_string")) {
 	iwrite(ptr_clrequest->fd, "", 1);
 	if (!assemble_return_msg(ptr_addresult, notempselectcolon.text, bibconns.conn)) {
@@ -780,5 +855,5 @@
 	goto Finish;
       }
-      biblio_info->year_unique_suffix = dbi_result_get_string(dbires, "year_uni_suffix");
+      ptr_biblio_info->year_unique_suffix = dbi_result_get_string(dbires, "year_uni_suffix");
 
       /* compare current string with the string of the previous iteration */
@@ -801,8 +876,8 @@
       if (item && *item && prev_authorconcat && *prev_authorconcat
 	  && strcmp(item, prev_authorconcat) == 0) {
-	biblio_info->is_subseq = 1;
+	ptr_biblio_info->is_subseq = 1;
       }
       else {
-	biblio_info->is_subseq = 0;
+	ptr_biblio_info->is_subseq = 0;
       }
 
@@ -817,5 +892,5 @@
       item = dbi_result_get_string(dbires, "dbname");
 
-      if ((new_outbuffer = render_dbib(&outbuffer, &outbuffer_len, &bibconns, dbires1 /* reference info */, item /* database */, biblio_info, ptr_clrequest->username, "", nref_counter)) == NULL) {
+      if ((new_outbuffer = render_dbib(&outbuffer, &outbuffer_len, &bibconns, dbires1 /* reference info */, item /* database */, ptr_biblio_info, ptr_clrequest->username, "", nref_counter)) == NULL) {
 	iwrite(ptr_clrequest->fd, "", 1);
 	nfailed++;
@@ -833,4 +908,61 @@
       }
 
+      /* todo: run iconv here */
+
+
+      /* run a character encoding conversion if required */
+      if (conv_descriptor && *outbuffer) {
+	size_t inlength;
+	size_t outlength;
+	char* my_outbuffer = NULL; /* this ptr will be modified by iconv() */
+	char* my_outbuffer_start = NULL; /* records initial state of my_elvalue */
+	const char* my_instring = NULL; /* this ptr will be modified by iconv() */
+	inlength = strlen(outbuffer) + 1;
+	/* with the encodings supported by our database engines, the converted
+	   string can't be longer than three times the input string */
+	/* todo: is this assumption correct? */
+	outlength = 3*inlength;
+
+	if ((my_outbuffer = (char*)malloc(outlength)) == NULL) {
+	  iwrite(ptr_clrequest->fd, outomem_n.text, outomem_n.length);
+	  /* 	dbi_result_free(rendinfo.dbires); */
+	  dbi_result_free(dbires);
+	  close_dbi_connections(&bibconns, drivername);
+	  goto Finish;
+	}
+
+	/* keep start of the converted string */
+	my_outbuffer_start = my_outbuffer;
+	
+	/* variable will be modified by iconv, so don't use original */
+	my_instring = (const char*)outbuffer;
+
+	/* now actually do the conversion */
+	if (iconv(conv_descriptor, &my_instring, &inlength, &my_outbuffer, &outlength) == (size_t)(-1)) {
+	  if (errno == EILSEQ) {
+	    iwrite(ptr_clrequest->fd, "iconv: invalid input character sequence\n", 40);
+	    LOG_PRINT(LOG_WARNING, "iconv: invalid input character sequence");
+	  }
+	  else if (errno == E2BIG) {
+	    iwrite(ptr_clrequest->fd, "iconv: output buffer too small\n", 31);
+	    LOG_PRINT(LOG_WARNING, "iconv: output buffer too small");
+	  }
+	  else if (errno == EINVAL) {
+	    iwrite(ptr_clrequest->fd, "iconv: incomplete input character\n", 34);
+	    LOG_PRINT(LOG_WARNING, "iconv: incomplete input character");
+	  }
+	  
+	  close_dbi_connections(&bibconns, drivername);
+	  goto Finish;
+	}
+	/* else: conversion went ok. We free the original string and replace
+	   it with the converted copy */
+	if (outbuffer) {
+	  free(outbuffer);
+	}
+	outbuffer = my_outbuffer_start;
+      }
+      /* else: no conversion required */
+
       iwrite(ptr_clrequest->fd, outbuffer, strlen(outbuffer));
       outbuffer[0] = '\0';
@@ -847,5 +979,5 @@
     } /* end loop over all distinct references */
 
-    if ((new_outbuffer = finish_render_dbib(&outbuffer, &outbuffer_len, &bibconns, biblio_info, table_name, ptr_clrequest)) == NULL) {
+    if ((new_outbuffer = finish_render_dbib(&outbuffer, &outbuffer_len, &bibconns, ptr_biblio_info, table_name, ptr_clrequest)) == NULL) {
       iwrite(ptr_clrequest->fd, "", 1);
       nfailed++;
@@ -887,5 +1019,5 @@
   
  Finish:
-  ptr_addresult->success = nref_counter - biblio_info->n_startnumber + 1;
+  ptr_addresult->success = nref_counter - ptr_biblio_info->n_startnumber + 1;
   ptr_addresult->failure = nfailed;
   numbyte = tread(ptr_clrequest->fd, inbuffer, COMMAND_INBUF_LEN);
@@ -899,4 +1031,8 @@
   }
 
+  if (conv_descriptor) {
+    iconv_close(conv_descriptor);
+  }
+
   return (nfailed) ? 1:0;
 }