[Refdb-cvs] CVS: refdb/src noteshandler.c,1.10,1.11 risxhandler.c,1.21,1.22

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

Update of /cvsroot/refdb/refdb/src
In directory sc8-pr-cvs1:/tmp/cvs-serv21530

Modified Files:
	noteshandler.c risxhandler.c 
Log Message:
added character conversion to end_handler()

Index: noteshandler.c
===================================================================
RCS file: /cvsroot/refdb/refdb/src/noteshandler.c,v
retrieving revision 1.10
retrieving revision 1.11
diff -u -U2 -r1.10 -r1.11
--- noteshandler.c	3 Jan 2004 20:11:43 -0000	1.10
+++ noteshandler.c	17 Jan 2004 22:47:20 -0000	1.11
@@ -27,6 +27,8 @@
 #include <string.h>
 #include <time.h> /* for strftime */
+#include <errno.h>
 #include <expat.h> /* header of the XML parser */
 #include <syslog.h> /* priority levels of log messages */
+#include <iconv.h>
 #include <dbi/dbi.h>
 
@@ -623,4 +625,68 @@
 
   if (!ptr_andata->ndb_error && !ptr_andata->nmem_error) {
+    /* do a character conversion if required. expat dumps all
+       character data as UTF-8 regardless of the input encoding */
+    size_t inlength;
+    size_t outlength;
+    char* my_elvalue = NULL; /* this ptr will be modified by iconv() */
+    char* my_elvalue_start = NULL; /* records initial state of my_elvalue */
+    const char* my_instring = NULL; /* this ptr will be modified by iconv() */
+
+    if (ptr_andata->conv_descriptor && *((ptr_andata->ptr_first)->ptr_elvalue)) {
+      inlength = strlen((ptr_andata->ptr_first)->ptr_elvalue) + 1;
+      /* with the encodings supported by our database engines, the converted
+	 string can't be longer than the input string */
+      outlength = inlength;
+
+      if ((my_elvalue = (char*)malloc(outlength)) == NULL) {
+	if ((new_msgpool = mstrcat(ptr_andata->msgpool, outomem_n.text, &(ptr_andata->msgpool_len), 0)) == NULL) {
+	  return;
+	}
+	else {
+	  ptr_andata->msgpool = new_msgpool;
+	}
+	(ptr_andata->nmem_error)++;
+	return;
+      }
+
+      /* keep start of the converted string */
+      my_elvalue_start = my_elvalue;
+
+      /* variable will be modified by iconv, so don't use original */
+      my_instring = (const char*)((ptr_andata->ptr_first)->ptr_elvalue);
+
+      /* now actually do the conversion */
+      if (iconv(ptr_andata->conv_descriptor, &my_instring, &inlength, &my_elvalue, &outlength) == (size_t)(-1)) {
+	if (errno == EILSEQ) {
+	  new_msgpool = mstrcat(ptr_andata->msgpool, "iconv: invalid input character sequence\n", &(ptr_andata->msgpool_len), 0);
+	  LOG_PRINT(LOG_WARNING, "iconv: invalid input character sequence");
+	}
+	else if (errno == E2BIG) {
+	  new_msgpool = mstrcat(ptr_andata->msgpool, "iconv: output buffer too small\n", &(ptr_andata->msgpool_len), 0);
+	  LOG_PRINT(LOG_WARNING, "iconv: output buffer too small");
+	}
+	else if (errno == EINVAL) {
+	  new_msgpool = mstrcat(ptr_andata->msgpool, "iconv: incomplete input character\n", &(ptr_andata->msgpool_len), 0);
+	  LOG_PRINT(LOG_WARNING, "iconv: incomplete input character");
+	}
+
+	if (new_msgpool == NULL) {
+	  return;
+	}
+	else {
+	  ptr_andata->msgpool = new_msgpool;
+	}
+	(ptr_andata->ndb_error)++;
+	return;
+      }
+      /* else: conversion went ok. We free the original string and replace
+	 it with the converted copy */
+      if ((ptr_andata->ptr_first)->ptr_elvalue) {
+	free((ptr_andata->ptr_first)->ptr_elvalue);
+      }
+      (ptr_andata->ptr_first)->ptr_elvalue = my_elvalue_start;
+    }
+    /* else: no conversion required */
+
     /* ------------------------------------------------------------ */
     if (!strcmp(el, "title") && ptr_andata->replace_note != 2) {

Index: risxhandler.c
===================================================================
RCS file: /cvsroot/refdb/refdb/src/risxhandler.c,v
retrieving revision 1.21
retrieving revision 1.22
diff -u -U2 -r1.21 -r1.22
--- risxhandler.c	7 Jan 2004 19:43:03 -0000	1.21
+++ risxhandler.c	17 Jan 2004 22:47:20 -0000	1.22
@@ -30,6 +30,8 @@
 #include <stdlib.h>
 #include <string.h>
+#include <errno.h>
 #include <expat.h> /* header of the XML parser */
 #include <syslog.h> /* priority levels of log messages */
+#include <iconv.h>
 #include <dbi/dbi.h>
 
@@ -636,4 +638,68 @@
 
   if (!ptr_ardata->ndb_error && !ptr_ardata->nmem_error) {
+    /* do a character conversion if required. expat dumps all
+       character data as UTF-8 regardless of the input encoding */
+    size_t inlength;
+    size_t outlength;
+    char* my_elvalue = NULL; /* this ptr will be modified by iconv() */
+    char* my_elvalue_start = NULL; /* records initial state of my_elvalue */
+    const char* my_instring = NULL; /* this ptr will be modified by iconv() */
+
+    if (ptr_ardata->conv_descriptor && *((ptr_ardata->ptr_first)->ptr_elvalue)) {
+      inlength = strlen((ptr_ardata->ptr_first)->ptr_elvalue) + 1;
+      /* with the encodings supported by our database engines, the converted
+	 string can't be longer than the input string */
+      outlength = inlength;
+
+      if ((my_elvalue = (char*)malloc(outlength)) == NULL) {
+	if ((new_msgpool = mstrcat(ptr_ardata->msgpool, outomem_n.text, &(ptr_ardata->msgpool_len), 0)) == NULL) {
+	  return;
+	}
+	else {
+	  ptr_ardata->msgpool = new_msgpool;
+	}
+	(ptr_ardata->nmem_error)++;
+	return;
+      }
+
+      /* keep start of the converted string */
+      my_elvalue_start = my_elvalue;
+
+      /* variable will be modified by iconv, so don't use original */
+      my_instring = (const char*)((ptr_ardata->ptr_first)->ptr_elvalue);
+
+      /* now actually do the conversion */
+      if (iconv(ptr_ardata->conv_descriptor, &my_instring, &inlength, &my_elvalue, &outlength) == (size_t)(-1)) {
+	if (errno == EILSEQ) {
+	  new_msgpool = mstrcat(ptr_ardata->msgpool, "iconv: invalid input character sequence\n", &(ptr_ardata->msgpool_len), 0);
+	  LOG_PRINT(LOG_WARNING, "iconv: invalid input character sequence");
+	}
+	else if (errno == E2BIG) {
+	  new_msgpool = mstrcat(ptr_ardata->msgpool, "iconv: output buffer too small\n", &(ptr_ardata->msgpool_len), 0);
+	  LOG_PRINT(LOG_WARNING, "iconv: output buffer too small");
+	}
+	else if (errno == EINVAL) {
+	  new_msgpool = mstrcat(ptr_ardata->msgpool, "iconv: incomplete input character\n", &(ptr_ardata->msgpool_len), 0);
+	  LOG_PRINT(LOG_WARNING, "iconv: incomplete input character");
+	}
+
+	if (new_msgpool == NULL) {
+	  return;
+	}
+	else {
+	  ptr_ardata->msgpool = new_msgpool;
+	}
+	(ptr_ardata->ndb_error)++;
+	return;
+      }
+      /* else: conversion went ok. We free the original string and replace
+	 it with the converted copy */
+      if ((ptr_ardata->ptr_first)->ptr_elvalue) {
+	free((ptr_ardata->ptr_first)->ptr_elvalue);
+      }
+      (ptr_ardata->ptr_first)->ptr_elvalue = my_elvalue_start;
+    }
+    /* else: no conversion required */
+
     /* ------------------------------------------------------------ */
     if (!strcmp(el, "title") && ptr_ardata->replace_ref != 2) {