From bddc64c57e116d8738f5e527e486e406002e6e16 Mon Sep 17 00:00:00 2001
From: Sherlock <119709043+agwekixj@users.noreply.github.com>
Date: Wed, 17 Sep 2025 11:24:43 -0700
Subject: [PATCH] Bug 799662 - Importing accounts from UTF8 csv file does not
 work correctly

Modifies the imports of Accounts from CSV, Bills & Invoices, and Customer & Vendors, to only call g_locale_to_utf8() if the file is determined to be not XML UTF-8 compliant.

Removes the potential XML-prohibited codepoints.
---
 .../import-export/bi-import/dialog-bi-import.c | 17 ++++++++++++-----
 .../import-export/csv-imp/csv-account-import.c | 18 ++++++++++++++++--
 .../customer-import/dialog-customer-import.c   | 17 ++++++++++++-----
 3 files changed, 40 insertions(+), 12 deletions(-)

diff --git a/gnucash/import-export/bi-import/dialog-bi-import.c b/gnucash/import-export/bi-import/dialog-bi-import.c
index 5a382028ba..f29743b933 100644
--- a/gnucash/import-export/bi-import/dialog-bi-import.c
+++ b/gnucash/import-export/bi-import/dialog-bi-import.c
@@ -36,6 +36,7 @@
 #include <glib.h>
 #include <glib/gstdio.h>
 
+#include "gnc-glib-utils.h"
 #include "gnc-date.h"
 #include "gnc-ui.h"
 #include "gnc-ui-util.h"
@@ -170,8 +171,15 @@ gnc_bi_import_read_file (const gchar * filename, const gchar * parser_regexp,
         if ((l > 0) && (line[l - 1] == '\n'))
             line[l - 1] = 0;
 
-        // convert line from locale into utf8
-        line_utf8 = g_locale_to_utf8 (line, -1, NULL, NULL, NULL);
+        // if the line doesn't conform to UTF-8, try a default charcter set
+        // conversion based on locale
+        if (g_utf8_validate(line, -1, NULL))
+            line_utf8 = line;
+        else
+            line_utf8 = g_locale_to_utf8 (line, -1, NULL, NULL, NULL);
+
+        // Remove the potential XML-prohibited codepoints from the UTF-8 compliant string
+        gnc_utf8_strip_invalid(line_utf8);
 
         // parse the line
         match_info = NULL;	// it seems, that in contrast to documentation, match_info is not always set -> g_match_info_free will segfault
@@ -216,9 +224,8 @@ gnc_bi_import_read_file (const gchar * filename, const gchar * parser_regexp,
         }
 
         g_match_info_free (match_info);
-        match_info = 0;
-        g_free (line_utf8);
-        line_utf8 = 0;
+        if (line_utf8 != line)
+            g_free (line_utf8);
     }
     g_free (line);
     line = 0;
diff --git a/gnucash/import-export/csv-imp/csv-account-import.c b/gnucash/import-export/csv-imp/csv-account-import.c
index 867cf8c479..f575fee7dd 100644
--- a/gnucash/import-export/csv-imp/csv-account-import.c
+++ b/gnucash/import-export/csv-imp/csv-account-import.c
@@ -30,6 +30,7 @@
 #include <glib/gi18n.h>
 #include <glib/gstdio.h>
 
+#include "gnc-glib-utils.h"
 #include "gnc-ui-util.h"
 #include <regex.h>
 #include "Account.h"
@@ -98,8 +99,21 @@ csv_import_read_file (GtkWindow *window, const gchar *filename,
         return RESULT_OPEN_FAILED;
     }
 
-    contents = g_locale_to_utf8 (locale_cont, -1, NULL, NULL, NULL);
-    g_free (locale_cont);
+    // if the contents don't conform to UTF-8, try a default charcter set
+    // conversion based on locale
+    if (g_utf8_validate(locale_cont, -1, NULL))
+    {
+        contents = locale_cont;
+    }
+    else
+    {
+        contents = g_locale_to_utf8 (locale_cont, -1, NULL, NULL, NULL);
+        g_free (locale_cont);
+    }
+
+    // Remove the potential XML-prohibited codepoints from the UTF-8 compliant content
+    gnc_utf8_strip_invalid(contents);
+
 
     // compile the regular expression and check for errors
     err = NULL;
diff --git a/gnucash/import-export/customer-import/dialog-customer-import.c b/gnucash/import-export/customer-import/dialog-customer-import.c
index 793ae5ae06..2991be1756 100644
--- a/gnucash/import-export/customer-import/dialog-customer-import.c
+++ b/gnucash/import-export/customer-import/dialog-customer-import.c
@@ -34,6 +34,7 @@
 #include <glib.h>
 #include <glib/gstdio.h>
 
+#include "gnc-glib-utils.h"
 #include "gnc-ui.h"
 #include "gnc-ui-util.h"
 #include "gnc-gui-query.h"
@@ -135,8 +136,15 @@ gnc_customer_import_read_file (const gchar *filename, const gchar *parser_regexp
         if ((l > 0) && (line[l - 1] == '\n'))
             line[l - 1] = 0;
 
-        // convert line from locale into utf8
-        line_utf8 = g_locale_to_utf8 (line, -1, NULL, NULL, NULL);
+        // if the line doesn't conform to UTF-8, try a default charcter set
+        // conversion based on locale
+        if (g_utf8_validate(line, -1, NULL))
+            line_utf8 = line;
+        else
+            line_utf8 = g_locale_to_utf8 (line, -1, NULL, NULL, NULL);
+
+        // Remove the potential XML-prohibited codepoints from the UTF-8 compliant string
+        gnc_utf8_strip_invalid(line_utf8);
 
         // parse the line
         match_info = NULL; // it seems, that in contrast to documentation, match_info is not always set -> g_match_info_free will segfault
@@ -176,9 +184,8 @@ gnc_customer_import_read_file (const gchar *filename, const gchar *parser_regexp
         }
 
         g_match_info_free (match_info);
-        match_info = 0;
-        g_free (line_utf8);
-        line_utf8 = 0;
+        if (line_utf8 != line)
+            g_free (line_utf8);
     }
     g_free (line);
     line = 0;