catalog: determine language from the filename

3 files changed, 74 insertions, 8 deletions

diff --git a/src/journal/catalog.c b/src/journal/catalog.c
index 90ca008064..2be81aa2f7 100644
--- a/src/journal/catalog.c
+++ b/src/journal/catalog.c
@@ -125,7 +125,10 @@ static int finish_item(
                 return log_oom();
 
         i->id = id;
-        strscpy(i->language, sizeof(i->language), language);
+        if (language) {
+                assert(strlen(language) > 1 && strlen(language) < 32);
+                strcpy(i->language, language);
+        }
         i->offset = htole64((uint64_t) offset);
 
         r = hashmap_put(h, i, i);
@@ -139,12 +142,34 @@ static int finish_item(
         return 0;
 }
 
+int catalog_file_lang(const char* filename, char **lang) {
+        char *beg, *end, *_lang;
+
+        end = endswith(filename, ".catalog");
+        if (!end)
+                return 0;
+
+        beg = end - 1;
+        while (beg > filename && *beg != '.')
+                beg --;
+
+        if (beg <= filename || end <= beg + 1 || end - beg > 32)
+                return 0;
+
+        _lang = strndup(beg + 1, end - beg - 1);
+        if (!_lang)
+                return -ENOMEM;
+
+        *lang = _lang;
+        return 1;
+}
+
 int catalog_import_file(Hashmap *h, struct strbuf *sb, const char *path) {
         _cleanup_fclose_ FILE *f = NULL;
         _cleanup_free_ char *payload = NULL;
         unsigned n = 0;
         sd_id128_t id;
-        char language[32];
+        _cleanup_free_ char *deflang = NULL, *lang = NULL;
         bool got_id = false, empty_line = true;
         int r;
 
@@ -158,6 +183,12 @@ int catalog_import_file(Hashmap *h, struct strbuf *sb, const char *path) {
                 return -errno;
         }
 
+        r = catalog_file_lang(path, &deflang);
+        if (r < 0)
+                log_error("Failed to determine language for file %s: %m", path);
+        if (r == 1)
+                log_debug("File %s has language %s.", path, deflang);
+
         for (;;) {
                 char line[LINE_MAX];
                 size_t a, b, c;
@@ -201,9 +232,12 @@ int catalog_import_file(Hashmap *h, struct strbuf *sb, const char *path) {
                         if (sd_id128_from_string(line + 2 + 1, &jd) >= 0) {
 
                                 if (got_id) {
-                                        r = finish_item(h, sb, id, language, payload);
+                                        r = finish_item(h, sb, id, lang ?: deflang, payload);
                                         if (r < 0)
                                                 return r;
+
+                                        free(lang);
+                                        lang = NULL;
                                 }
 
                                 if (with_language) {
@@ -214,14 +248,21 @@ int catalog_import_file(Hashmap *h, struct strbuf *sb, const char *path) {
                                                 log_error("[%s:%u] Language too short.", path, n);
                                                 return -EINVAL;
                                         }
-                                        if (c > sizeof(language) - 1) {
+                                        if (c > 31) {
                                                 log_error("[%s:%u] language too long.", path, n);
                                                 return -EINVAL;
                                         }
 
-                                        strscpy(language, sizeof(language), t);
-                                } else
-                                        language[0] = '\0';
+                                        if (deflang) {
+                                                log_warning("[%s:%u] language %s", path, n,
+                                                            streq(t, deflang) ?
+                                                            "specified unnecessarily" :
+                                                            "differs from default for file");
+                                                lang = strdup(t);
+                                                if (!lang)
+                                                        return -ENOMEM;
+                                        }
+                                }
 
                                 got_id = true;
                                 empty_line = false;
@@ -264,7 +305,7 @@ int catalog_import_file(Hashmap *h, struct strbuf *sb, const char *path) {
         }
 
         if (got_id) {
-                r = finish_item(h, sb, id, language, payload);
+                r = finish_item(h, sb, id, lang ?: deflang, payload);
                 if (r < 0)
                         return r;
         }
diff --git a/src/journal/catalog.h b/src/journal/catalog.h
index 24a2d0b553..5e15b99ac0 100644
--- a/src/journal/catalog.h
+++ b/src/journal/catalog.h
@@ -34,4 +34,5 @@ int catalog_update(const char* database, const char* root, const char* const* di
 int catalog_get(const char* database, sd_id128_t id, char **data);
 int catalog_list(FILE *f, const char* database, bool oneline);
 int catalog_list_items(FILE *f, const char* database, bool oneline, char **items);
+int catalog_file_lang(const char *filename, char **lang);
 extern const char * const catalog_file_dirs[];
diff --git a/src/journal/test-catalog.c b/src/journal/test-catalog.c
index 5db5bed8de..863b9f4b51 100644
--- a/src/journal/test-catalog.c
+++ b/src/journal/test-catalog.c
@@ -120,6 +120,28 @@ static void test_catalog_update(void) {
         assert(r >= 0);
 }
 
+static void test_catalog_file_lang(void) {
+        _cleanup_free_ char *lang = NULL, *lang2 = NULL, *lang3 = NULL;
+
+        assert_se(catalog_file_lang("systemd.de_DE.catalog", &lang) == 1);
+        assert_se(streq(lang, "de_DE"));
+
+        assert_se(catalog_file_lang("systemd..catalog", &lang2) == 0);
+        assert_se(lang2 == NULL);
+
+        assert_se(catalog_file_lang("systemd.fr.catalog", &lang2) == 1);
+        assert_se(streq(lang2, "fr"));
+
+        assert_se(catalog_file_lang("systemd.fr.catalog.gz", &lang3) == 0);
+        assert_se(lang3 == NULL);
+
+        assert_se(catalog_file_lang("systemd.01234567890123456789012345678901.catalog", &lang3) == 0);
+        assert_se(lang3 == NULL);
+
+        assert_se(catalog_file_lang("systemd.0123456789012345678901234567890.catalog", &lang3) == 1);
+        assert_se(streq(lang3, "0123456789012345678901234567890"));
+}
+
 int main(int argc, char *argv[]) {
         _cleanup_free_ char *text = NULL;
         int r;
@@ -144,5 +166,7 @@ int main(int argc, char *argv[]) {
         if (database)
                 unlink(database);
 
+        test_catalog_file_lang();
+
         return 0;
 }