* mu-str: add mu_str_guess_first_name, mu_str_guess_last_name and

mu_str_guess_nick, + unit tests

  these function do a simple guess for the first/lastname and nick, so we can
prefill	the results of 'mu cfind' with some first guesses.
This commit is contained in:
Dirk-Jan C. Binnema
2011-03-06 12:14:26 +02:00
parent 17f2e3bdf9
commit 9a600e71c9
3 changed files with 216 additions and 0 deletions

View File

@ -425,3 +425,95 @@ mu_str_escape_c_literal (const gchar* str)
return g_string_free (tmp, FALSE);
}
gchar*
mu_str_guess_last_name (const char *name)
{
const gchar *lastsp;
if (!name)
return g_strdup ("");
lastsp = g_strrstr (name, " ");
return g_strdup (lastsp ? lastsp + 1 : "");
}
gchar*
mu_str_guess_first_name (const char *name)
{
const gchar *lastsp;
if (!name)
return g_strdup ("");
lastsp = g_strrstr (name, " ");
if (lastsp)
return g_strndup (name, lastsp - name);
else
return g_strdup (name);
}
static gchar*
cleanup_str (const char* str)
{
gchar *s;
const gchar *cur;
unsigned i;
if (mu_str_is_empty(str))
return g_strdup ("");
s = g_new0 (char, strlen(str) + 1);
for (cur = str, i = 0; *cur; ++cur) {
if (ispunct(*cur) || isspace(*cur))
continue;
else
s[i++] = *cur;
}
return s;
}
gchar*
mu_str_guess_nick (const char* name)
{
gchar *fname, *lname, *nick;
gchar initial[7];
fname = mu_str_guess_first_name (name);
lname = mu_str_guess_last_name (name);
/* if there's no last name, use first name as the nick */
if (mu_str_is_empty(fname) || mu_str_is_empty(lname)) {
g_free (lname);
nick = fname;
goto leave;
}
memset (initial, 0, sizeof(initial));
/* couldn't we get an initial for the last name? */
if (g_unichar_to_utf8 (g_utf8_get_char (lname), initial) == 0) {
g_free (lname);
nick = fname;
goto leave;
}
nick = g_strdup_printf ("%s%s", fname, initial);
g_free (fname);
leave:
{
gchar *tmp;
tmp = cleanup_str (nick);
g_free (nick);
nick = tmp;
}
return nick;
}

View File

@ -261,6 +261,48 @@ char* mu_str_escape_c_literal (const gchar* str)
*/
#define mu_str_is_empty(S) ((!(S)||!(S)[0])?TRUE:FALSE)
/**
* guess some nick name for the given name; if we can determine an
* first name, last name, the nick will be first name + the first char
* of the last name. otherwise, it's just the first name. clearly,
* this is just a rough guess for setting an initial value for nicks.
*
* @param name a name
*
* @return the guessed nick, as a newly allocated string (free with g_free)
*/
gchar* mu_str_guess_nick (const char* name)
G_GNUC_WARN_UNUSED_RESULT;
/**
* guess the first name for the given name; clearly,
* this is just a rough guess for setting an initial value.
*
* @param name a name
*
* @return the first name, as a newly allocated string (free with
* g_free)
*/
gchar* mu_str_guess_first_name (const char* name)
G_GNUC_WARN_UNUSED_RESULT;
/**
* guess the last name for the given name; clearly,
* this is just a rough guess for setting an initial value.
*
* @param name a name
*
* @return the last name, as a newly allocated string (free with
* g_free)
*/
gchar* mu_str_guess_last_name (const char* name)
G_GNUC_WARN_UNUSED_RESULT;
G_END_DECLS
#endif /*__MU_STR_H__*/

View File

@ -251,6 +251,82 @@ test_mu_str_date_parse_hdwmy (void)
}
static void
test_mu_str_guess_first_name (void)
{
int i;
struct {
char *src, *exp;
} tests[] = {
{ "Richard M. Stallman", "Richard M." },
{ "John Rambo", "John" },
{ "Ivanhoe", "Ivanhoe" },
{ "", "" }
};
for (i = 0; i != G_N_ELEMENTS(tests); ++i) {
gchar *s;
s = mu_str_guess_first_name (tests[i].src);
g_assert_cmpstr (s, ==, tests[i].exp);
g_free (s);
}
}
static void
test_mu_str_guess_last_name (void)
{
int i;
struct {
char *src, *exp;
} tests[] = {
{ "Richard M. Stallman", "Stallman" },
{ "John Rambo", "Rambo" },
{ "Ivanhoe", "" },
{ "", "" }
};
for (i = 0; i != G_N_ELEMENTS(tests); ++i) {
gchar *s;
s = mu_str_guess_last_name (tests[i].src);
g_assert_cmpstr (s, ==, tests[i].exp);
g_free (s);
}
}
static void
test_mu_str_guess_nick (void)
{
int i;
struct {
char *src, *exp;
} tests[] = {
{ "Richard M. Stallman", "RichardMS" },
{ "John Rambo", "JohnR" },
{ "Ivanhoe", "Ivanhoe" },
{ "", "" }
};
for (i = 0; i != G_N_ELEMENTS(tests); ++i) {
gchar *s;
s = mu_str_guess_nick (tests[i].src);
g_assert_cmpstr (s, ==, tests[i].exp);
g_free (s);
}
}
int
main (int argc, char *argv[])
@ -289,6 +365,12 @@ main (int argc, char *argv[])
g_test_add_func ("/mu-str/mu-str_date_parse_hdwmy",
test_mu_str_date_parse_hdwmy);
g_test_add_func ("/mu-str/mu-str_guess_first_name",
test_mu_str_guess_first_name);
g_test_add_func ("/mu-str/mu-str_guess_last_name",
test_mu_str_guess_last_name);
g_test_add_func ("/mu-str/mu-str_guess_nick",
test_mu_str_guess_nick);
/* FIXME: add tests for mu_str_flags; but note the
* function simply calls mu_msg_field_str */