[easytag] Improve scanner functions and fix tests

[David King <davidk src gnome org>]

commit f43c0a65e1404535c5a11a0caf7576a69a576f70
Author: Abhinav <abhijangda hotmail com>
Date:   Mon Apr 7 21:08:59 2014 +0100

    Improve scanner functions and fix tests
    
    https://bugzilla.gnome.org/show_bug.cgi?id=726108

 src/easytag.c     |   31 +++++------
 src/scan.c        |  152 +++++++++++++++++++++++------------------------------
 src/scan.h        |    8 ++--
 src/scan_dialog.c |   33 ++++++++++--
 tests/test-scan.c |   28 ++++++----
 5 files changed, 129 insertions(+), 123 deletions(-)
---
diff --git a/src/easytag.c b/src/easytag.c
index 3f29ed3..a03478b 100644
--- a/src/easytag.c
+++ b/src/easytag.c
@@ -4309,21 +4309,23 @@ Convert_All_Uppercase (GtkWidget *entry)
 static void
 Convert_All_Lowercase (GtkWidget *entry)
 {
-    gchar *string = g_strdup(gtk_entry_get_text(GTK_ENTRY(entry)));
+    gchar *res;
+    const gchar *string = gtk_entry_get_text (GTK_ENTRY (entry));
 
-    Scan_Process_Fields_All_Downcase(string);
-    gtk_entry_set_text(GTK_ENTRY(entry),string);
-    g_free(string);
+    res = Scan_Process_Fields_All_Uppercase (string);
+    gtk_entry_set_text (GTK_ENTRY (entry), res);
+    g_free (res);
 }
 
 static void
 Convert_Letter_Uppercase (GtkWidget *entry)
 {
-    gchar *string = g_strdup(gtk_entry_get_text(GTK_ENTRY(entry)));
+    gchar *res;
+    const gchar *string = gtk_entry_get_text (GTK_ENTRY (entry));
 
-    Scan_Process_Fields_Letter_Uppercase(string);
-    gtk_entry_set_text(GTK_ENTRY(entry),string);
-    g_free(string);
+    res = Scan_Process_Fields_Letter_Uppercase (string);
+    gtk_entry_set_text (GTK_ENTRY (entry), res);
+    g_free (res);
 }
 
 static void
@@ -4349,15 +4351,12 @@ Convert_Remove_Space (GtkWidget *entry)
 static void
 Convert_Insert_Space (GtkWidget *entry)
 {
-    // FIX ME : we suppose that it will not grow more than 2 times its size...
-    guint string_length = 2 * strlen(gtk_entry_get_text(GTK_ENTRY(entry)));
-    gchar *string       = g_malloc(string_length+1);
-    strncpy(string,gtk_entry_get_text(GTK_ENTRY(entry)),string_length);
-    string[string_length]='\0';
+    gchar *res;
+    const gchar *string = (gtk_entry_get_text (GTK_ENTRY (entry)));
 
-    Scan_Process_Fields_Insert_Space(&string);
-    gtk_entry_set_text(GTK_ENTRY(entry),string);
-    g_free(string);
+    res = Scan_Process_Fields_Insert_Space (string);
+    gtk_entry_set_text (GTK_ENTRY (entry), res);
+    g_free (res);
 }
 
 static void
diff --git a/src/scan.c b/src/scan.c
index 10419ca..cc74e04 100644
--- a/src/scan.c
+++ b/src/scan.c
@@ -84,39 +84,36 @@ Scan_Process_Fields_Remove_Space (gchar *string)
 }
 
 /*
- * The function inserts a space before an uppercase letter
- * It is needed to realloc the memory!
+ * Scan_Process_Fields_Insert_Space:
+ * @string: Input string
+ *
+ * This function will insert space before every uppercase character.
+ *
+ * Returns: A newly allocated string.
  */
-void
-Scan_Process_Fields_Insert_Space (gchar **string)
+gchar *
+Scan_Process_Fields_Insert_Space (const gchar *string)
 {
     gchar *iter;
     gunichar c;
-    gint j;
-    guint string_length;
-    gchar *string1;
-
-    // FIX ME : we suppose that it will not grow more than 2 times its size...
-    string_length = 2 * strlen(*string);
-    //string1 = g_realloc(*string, string_length+1);
-    string1       = g_malloc(string_length+1);
-    strncpy(string1,*string,string_length);
-    string1[string_length]='\0';
-    g_free(*string);
-    *string = string1;
-
-    for (iter = g_utf8_next_char(*string); *iter; iter = g_utf8_next_char(iter)) // At start : 
g_utf8_next_char to not consider first "uppercase" letter
+    GString *string1;
+
+    string1 = g_string_new ("");
+    g_string_append_c (string1, *string);
+
+    for (iter = g_utf8_next_char (string); *iter; iter = g_utf8_next_char (iter))
     {
-        c = g_utf8_get_char(iter);
+        c = g_utf8_get_char (iter);
 
-        if (g_unichar_isupper(c))
+        if (g_unichar_isupper (c))
         {
-            for (j = strlen(iter); j > 0; j--)
-                *(iter + j) = *(iter + j - 1);
-            *iter = ' ';
-            iter++;
+            g_string_append_c (string1, ' ');
         }
+
+        g_string_append_unichar (string1, c);
     }
+
+    return g_string_free (string1, FALSE);
 }
 
 /*
@@ -166,86 +163,69 @@ Scan_Remove_Spaces (gchar *string)
   }
 }
 
-void
-Scan_Process_Fields_All_Uppercase (gchar *string)
+/* Returns a newly-allocated string. */
+gchar *
+Scan_Process_Fields_All_Uppercase (const gchar *string)
 {
-    gchar *temp;
-    gchar temp2[6]; // Must have at least 6 bytes of space
-    gunichar c;
-
-    for (temp = string; *temp; temp = g_utf8_next_char(temp))
-    {
-        c = g_utf8_get_char(temp);
-        if (g_unichar_islower(c))
-            strncpy(temp, temp2, g_unichar_to_utf8(g_unichar_toupper(c), temp2));
-    }
+    return g_utf8_strup (string, -1);
 }
 
-void
-Scan_Process_Fields_All_Downcase (gchar *string)
+/* Returns a newly-allocated string. */
+gchar *
+Scan_Process_Fields_All_Downcase (const gchar *string)
 {
-    gchar *temp;
-    gchar temp2[6];
-    gunichar c;
-
-    for (temp = string; *temp; temp = g_utf8_next_char(temp))
-    {
-        c = g_utf8_get_char(temp);
-        if (g_unichar_isupper(c))
-            strncpy(temp, temp2, g_unichar_to_utf8(g_unichar_tolower(c), temp2));
-    }
+    return g_utf8_strdown (string, -1);
 }
 
-void
-Scan_Process_Fields_Letter_Uppercase (gchar *string)
+/* Returns a newly-allocated string. */
+gchar *
+Scan_Process_Fields_Letter_Uppercase (const gchar *string)
 {
-    gchar *temp;
+    const gchar *temp;
     gchar temp2[6];
     gboolean set_to_upper_case = TRUE;
     gunichar c;
-    gchar utf8_character[6];
-    gchar *word, *word1, *word2;
+    GString *string1;
+
+    string1 = g_string_new ("");
 
-    for (temp = string; *temp; temp = g_utf8_next_char(temp))
+    for (temp = string; *temp; temp = g_utf8_next_char (temp))
     {
-        c = g_utf8_get_char(temp);
-        if (set_to_upper_case && g_unichar_islower(c))
-            strncpy(temp, temp2, g_unichar_to_utf8(g_unichar_toupper(c), temp2));
-        else if (!set_to_upper_case && g_unichar_isupper(c))
-            strncpy(temp, temp2, g_unichar_to_utf8(g_unichar_tolower(c), temp2));
-        set_to_upper_case = FALSE; // After the first time, all will be down case
-    }
+        gchar *temp3;
+        int l;
 
-    temp = string;
+        c = g_utf8_get_char (temp);
+        l = g_unichar_to_utf8 (c, temp2);
 
-    // Uppercase again the word 'I' in english
-    while ( temp )
-    {
-        word = temp; // Needed if there is only one word
-        word1 = g_utf8_strchr(temp,-1,' ');
-        word2 = g_utf8_strchr(temp,-1,'_');
-
-        // Take the first string found (near beginning of string)
-        if (word1 && word2)
-            word = MIN(word1,word2);
-        else if (word1)
-            word = word1;
-        else if (word2)
-            word = word2;
+        if (set_to_upper_case && g_unichar_islower(c))
+        {
+            temp3 = g_utf8_strup (temp2, l);
+            g_string_append (string1, temp3);
+            g_free (temp3);
+        }
+        else if (!set_to_upper_case && g_unichar_isupper(c))
+        {
+            temp3 = g_utf8_strdown (temp2, l);
+            g_string_append (string1, temp3);
+            g_free (temp3);
+        }
         else
-            // Last word of the string
-            break;
-
-        // Go to first character of the word (char. after ' ' or '_')
-        word = word+1;
+        {
+            g_string_append_len (string1, temp2, l);
+        }
 
-        // Set uppercase word 'I'
-        if (g_ascii_strncasecmp("I ", word, strlen("I ")) == 0)
+        /* Uppercase the word 'I' in english */
+        if (!set_to_upper_case &&
+            (*(temp - 1) == ' ' || *(temp - 1) == '_') &&
+            (*temp == 'i' || *temp == 'I') &&
+            (*(temp + 1) == ' ' || *(temp + 1) == '_'))
         {
-            c = g_utf8_get_char(word);
-            strncpy(word, utf8_character, g_unichar_to_utf8(g_unichar_toupper(c), utf8_character));
+            string1->str [string1->len - 1] = 'I';
         }
 
-        temp = word;
+        /* After the first time, all will be lower case. */
+        set_to_upper_case = FALSE;
     }
+
+    return g_string_free (string1, FALSE);
 }
diff --git a/src/scan.h b/src/scan.h
index addd242..8c54686 100644
--- a/src/scan.h
+++ b/src/scan.h
@@ -25,11 +25,11 @@ void Scan_Convert_Underscore_Into_Space (gchar *string);
 void Scan_Convert_P20_Into_Space (gchar *string);
 void Scan_Convert_Space_Into_Underscore (gchar *string);
 void Scan_Process_Fields_Remove_Space (gchar *string);
-void Scan_Process_Fields_Insert_Space (gchar **string);
+gchar* Scan_Process_Fields_Insert_Space (const gchar *string);
 void Scan_Process_Fields_Keep_One_Space (gchar *string);
 void Scan_Remove_Spaces (gchar *string);
-void Scan_Process_Fields_All_Uppercase (gchar *string);
-void Scan_Process_Fields_All_Downcase (gchar *string);
-void Scan_Process_Fields_Letter_Uppercase (gchar *string);
+gchar* Scan_Process_Fields_All_Uppercase (const gchar *string);
+gchar* Scan_Process_Fields_All_Downcase (const gchar *string);
+gchar* Scan_Process_Fields_Letter_Uppercase (const gchar *string);
 
 #endif /* !ET_SCAN_H_ */
diff --git a/src/scan_dialog.c b/src/scan_dialog.c
index 8ce1e04..7b2c4b4 100644
--- a/src/scan_dialog.c
+++ b/src/scan_dialog.c
@@ -1363,7 +1363,12 @@ Scan_Process_Fields_Functions (gchar **string)
         Scan_Convert_Space_Into_Underscore (*string);
 
     if (gtk_toggle_button_get_active(GTK_TOGGLE_BUTTON(ProcessFieldsInsertSpace)))
-        Scan_Process_Fields_Insert_Space(string);
+    {
+        gchar *res;
+        res = Scan_Process_Fields_Insert_Space (*string);
+        g_free (*string);
+        *string = res;
+    }
 
     if (gtk_toggle_button_get_active(GTK_TOGGLE_BUTTON(ProcessFieldsOnlyOneSpace)))
         Scan_Process_Fields_Keep_One_Space(*string);
@@ -1372,13 +1377,28 @@ Scan_Process_Fields_Functions (gchar **string)
         Scan_Convert_Character(string);
 
     if (gtk_toggle_button_get_active(GTK_TOGGLE_BUTTON(ProcessFieldsAllUppercase)))
-        Scan_Process_Fields_All_Uppercase(*string);
+    {
+        gchar *res;
+        res = Scan_Process_Fields_All_Uppercase (*string);
+        g_free (*string);
+        *string = res;
+    }
 
     if (gtk_toggle_button_get_active(GTK_TOGGLE_BUTTON(ProcessFieldsAllDowncase)))
-        Scan_Process_Fields_All_Downcase(*string);
+    {
+        gchar *res;
+        res = Scan_Process_Fields_All_Downcase (*string);
+        g_free (*string);
+        *string = res;
+    }
 
     if (gtk_toggle_button_get_active(GTK_TOGGLE_BUTTON(ProcessFieldsFirstLetterUppercase)))
-         Scan_Process_Fields_Letter_Uppercase(*string);
+    {
+        gchar *res;
+        res = Scan_Process_Fields_Letter_Uppercase (*string);
+        g_free (*string);
+        *string = res;
+    }
 
     if (gtk_toggle_button_get_active(GTK_TOGGLE_BUTTON(ProcessFieldsFirstLettersUppercase)))
         Scan_Process_Fields_First_Letters_Uppercase(*string);
@@ -1452,7 +1472,10 @@ void Scan_Process_Fields_First_Letters_Uppercase (gchar *string)
     {
         exempt[0] = NULL;
     }
-    Scan_Process_Fields_All_Downcase(string);
+
+    temp = Scan_Process_Fields_All_Downcase (string);
+    g_free (string);
+    string = temp;
 
     if (!g_utf8_validate(string,-1,NULL))
     {
diff --git a/tests/test-scan.c b/tests/test-scan.c
index 30ff8ba..34f3235 100644
--- a/tests/test-scan.c
+++ b/tests/test-scan.c
@@ -102,13 +102,14 @@ scan_insert_space (void)
 
     for (i = 0; i < G_N_ELEMENTS (cases); i++)
     {
-        gchar *string;
+        gchar *string, *res;
 
         string = g_strdup (cases[i]);
-        Scan_Process_Fields_Insert_Space (&string);
-        check_string (string, results[i]);
+        res = Scan_Process_Fields_Insert_Space (string);
+        check_string (res, results[i]);
 
         g_free (string);
+        g_free (res);
     }
 }
 
@@ -123,13 +124,14 @@ scan_all_uppercase (void)
 
     for (i = 0; i < G_N_ELEMENTS (cases); i++)
     {
-        gchar *string;
+        gchar *string, *res;
 
         string = g_strdup (cases[i]);
-        Scan_Process_Fields_All_Uppercase (string);
-        check_string (string, results[i]);
+        res = Scan_Process_Fields_All_Uppercase (string);
+        check_string (res, results[i]);
 
         g_free (string);
+        g_free (res);
     }
 }
 
@@ -144,13 +146,14 @@ scan_all_lowercase (void)
 
     for (i = 0; i < G_N_ELEMENTS (cases); i++)
     {
-        gchar *string;
+        gchar *string, *res;
 
         string = g_strdup (cases[i]);
-        Scan_Process_Fields_All_Downcase (string);
-        check_string (string, results[i]);
+        res = Scan_Process_Fields_All_Downcase (string);
+        check_string (res, results[i]);
 
         g_free (string);
+        g_free (res);
     }
 }
 
@@ -167,13 +170,14 @@ scan_letter_uppercase (void)
 
     for (i = 0; i < G_N_ELEMENTS (cases); i++)
     {
-        gchar *string;
+        gchar *string, *res;
 
         string = g_strdup (cases [i]);
-        Scan_Process_Fields_Letter_Uppercase (string);
-        check_string (string, results [i]);
+        res = Scan_Process_Fields_Letter_Uppercase (string);
+        check_string (res, results [i]);
 
         g_free (string);
+        g_free (res);
     }
 }