[Refdb-cvs] CVS: refdb/src refdbdnote.c,1.13,1.14 refdbdref.c,1.50,1.51

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

Update of /cvsroot/refdb/refdb/src
In directory sc8-pr-cvs1:/tmp/cvs-serv25105

Modified Files:
	refdbdnote.c refdbdref.c 
Log Message:
retrieve database encoding for XML import

Index: refdbdnote.c
===================================================================
RCS file: /cvsroot/refdb/refdb/src/refdbdnote.c,v
retrieving revision 1.13
retrieving revision 1.14
diff -u -U2 -r1.13 -r1.14
--- refdbdnote.c	4 Jan 2004 01:57:31 -0000	1.13
+++ refdbdnote.c	17 Jan 2004 23:07:41 -0000	1.14
@@ -25,4 +25,5 @@
 #include <dbi/dbi.h>
 #include <expat.h> /* the xml parser header file */
+#include <iconv.h>
 
 #include "refdb.h"
@@ -68,6 +69,8 @@
                       added ID values
 
+  char* input_encoding encoding of the input data
+
   ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++*/
-int addnote(struct CLIENT_REQUEST* ptr_clrequest, char* set_owner, struct ADDRESULT* ptr_addresult, int replace_note, Lilid* ptr_sentinel) {
+int addnote(struct CLIENT_REQUEST* ptr_clrequest, char* set_owner, struct ADDRESULT* ptr_addresult, int replace_note, Lilid* ptr_sentinel, char* input_encoding) {
   dbi_conn conn;
   dbi_conn conn_refdb = NULL;
@@ -79,4 +82,5 @@
   char *the_user; /* name (either set_owner or username) */
   char *return_msg; /* string to hold a return message for the client */
+  const char *db_encoding; /* encoding of database */
   const char *drivername; /* name of the libdbi driver */
   char old_db[DBNAME_LENGTH+1];
@@ -119,4 +123,24 @@
   }
 
+  /* get the database encoding */
+  db_encoding = dbi_conn_get_encoding(conn);
+
+  /* if we need to convert, create a conversion descriptor for iconv() */
+  if (db_encoding && strcmp(db_encoding, "UTF-8")) {
+    andata.conv_descriptor = iconv_open(!strcmp(db_encoding, "US-ASCII") ? "ASCII" : db_encoding, "UTF-8");
+    if (andata.conv_descriptor == (iconv_t)(-1)) {
+      andata.conv_descriptor = NULL;
+      LOG_PRINT(LOG_WARNING, "cannot set output encoding");
+    }
+    else {
+      LOG_PRINT(LOG_DEBUG, "encoding is now:");
+      LOG_PRINT(LOG_DEBUG, db_encoding);
+    }
+  }
+  else {
+    andata.conv_descriptor = NULL;
+    LOG_PRINT(LOG_DEBUG, "encoding is UTF-8, no conversion required");
+  }
+
   while (1) { /* leave with break */
     /* create the parser instance */
@@ -217,4 +241,8 @@
   delete_all_lilimem(&sentinel);
 
+  if (andata.conv_descriptor) {
+    iconv_close(andata.conv_descriptor);
+  }
+
   return success;
 }

Index: refdbdref.c
===================================================================
RCS file: /cvsroot/refdb/refdb/src/refdbdref.c,v
retrieving revision 1.50
retrieving revision 1.51
diff -u -U2 -r1.50 -r1.51
--- refdbdref.c	10 Jan 2004 01:26:39 -0000	1.50
+++ refdbdref.c	17 Jan 2004 23:07:41 -0000	1.51
@@ -27,4 +27,5 @@
 #include <time.h>
 #include <expat.h> /* the xml parser header file */
+#include <iconv.h>
 #include <dbi/dbi.h>
 
@@ -59,5 +60,5 @@
 
 /* forward declaration of local functions */
-static unsigned long long read_ris_data(struct CLIENT_REQUEST* ptr_clrequest, struct ADDRESULT* ptr_addresult, dbi_conn conn, dbi_conn conn_refdb, int replace_ref, int n_keep_id, const char* the_user, Lilid* ptr_sentinel);
+static unsigned long long read_ris_data(struct CLIENT_REQUEST* ptr_clrequest, struct ADDRESULT* ptr_addresult, dbi_conn conn, dbi_conn conn_refdb, int replace_ref, int n_keep_id, const char* the_user, Lilid* ptr_sentinel, iconv_t conv_descriptor);
 static int real_run_keyword_scan(struct CLIENT_REQUEST* ptr_clrequest, Lilid* ptr_sentinel, int mode, int* ptr_insert);
 
@@ -270,6 +271,8 @@
                       added ID values
 
+  char* input_encoding encoding of the input data
+
   ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++*/
-int addref(struct CLIENT_REQUEST* ptr_clrequest, char* set_owner, struct ADDRESULT* ptr_addresult, int replace_ref, int n_keep_id, const char* reftype, Lilid* ptr_sentinel) {
+int addref(struct CLIENT_REQUEST* ptr_clrequest, char* set_owner, struct ADDRESULT* ptr_addresult, int replace_ref, int n_keep_id, const char* reftype, Lilid* ptr_sentinel, char* input_encoding) {
   dbi_conn conn;
   dbi_conn conn_refdb = NULL;
@@ -281,4 +284,5 @@
   char *the_user; /* name (either set_owner or username) */
   char *return_msg; /* string to hold a return message for the client */
+  const char *db_encoding; /* encoding of database */
   const char *drivername; /* name of the libdbi driver */
   char old_db[DBNAME_LENGTH+1];
@@ -328,4 +332,24 @@
   }
 
+  /* get the database encoding */
+  db_encoding = dbi_conn_get_encoding(conn);
+
+  /* if we need to convert, create a conversion descriptor for iconv() */
+  if (db_encoding && strcmp(db_encoding, "UTF-8")) {
+    ardata.conv_descriptor = iconv_open(!strcmp(db_encoding, "US-ASCII") ? "ASCII" : db_encoding, "UTF-8");
+    if (ardata.conv_descriptor == (iconv_t)(-1)) {
+      ardata.conv_descriptor = NULL;
+      LOG_PRINT(LOG_WARNING, "cannot set output encoding");
+    }
+    else {
+      LOG_PRINT(LOG_DEBUG, "encoding is now:");
+      LOG_PRINT(LOG_DEBUG, db_encoding);
+    }
+  }
+  else {
+    ardata.conv_descriptor = NULL;
+    LOG_PRINT(LOG_DEBUG, "encoding is UTF-8, no conversion required");
+  }
+
   if (!strcmp(my_dbi_conn_get_cap(conn, "multiple_db"), "f")) {
     /* pgsql, sqlite need another connection to retrieve values from
@@ -341,4 +365,7 @@
       dbi_conn_close(conn);
       delete_all_lilimem(&sentinel);
+      if (ardata.conv_descriptor) {
+	iconv_close(ardata.conv_descriptor);
+      }
       LOG_PRINT(LOG_ERR, connerr.text);
       return 0;
@@ -450,5 +477,5 @@
     }
 
-    ullresult = read_ris_data(ptr_clrequest, ptr_addresult, conn, conn_refdb, replace_ref, n_keep_id, the_user, ptr_sentinel);
+    ullresult = read_ris_data(ptr_clrequest, ptr_addresult, conn, conn_refdb, replace_ref, n_keep_id, the_user, ptr_sentinel, ardata.conv_descriptor);
   }
 
@@ -466,4 +493,8 @@
   free_authorinfo(ardata.ptr_ainfo);
 
+  if (ardata.conv_descriptor) {
+    iconv_close(ardata.conv_descriptor);
+  }
+
   return success;
 }
@@ -496,6 +527,8 @@
                       added ID values
 
+  iconv_t conv_descriptor conversion descriptor for iconv()
+
   ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++*/
-static unsigned long long read_ris_data(struct CLIENT_REQUEST* ptr_clrequest, struct ADDRESULT* ptr_addresult, dbi_conn conn, dbi_conn conn_refdb, int replace_ref, int n_keep_id, const char* the_user, Lilid* ptr_sentinel) {
+static unsigned long long read_ris_data(struct CLIENT_REQUEST* ptr_clrequest, struct ADDRESULT* ptr_addresult, dbi_conn conn, dbi_conn conn_refdb, int replace_ref, int n_keep_id, const char* the_user, Lilid* ptr_sentinel, iconv_t conv_descriptor) {
   int numbyte;
   int success = 1; /* be optimistic */
@@ -574,4 +607,6 @@
       }
 
+      /* ToDo: run iconv here */
+
       /* split reference into lines and squeeze contents into database */
       result = process_ris_set(ris_set, conn, conn_refdb, replace_ref, (char*)the_user, ptr_clrequest, n_keep_id, ptr_sentinel);