lib-fts, fts: fts_language_detect() - Return error string instead of logging it

diff --git a/src/lib-fts/fts-language.c b/src/lib-fts/fts-language.c
index fb5c0ea9a8d62807c4ec2158a061808967ed9ecd..2f1d11228f22329e58086895d109e4170bbfaf3d 100644 (file)
--- a/src/lib-fts/fts-language.c
+++ b/src/lib-fts/fts-language.c
@@ -26,7 +26,7 @@ struct fts_language_list {
        const char *textcat_config;
        const char *textcat_datadir;
        void *textcat_handle;
-       bool textcat_failed;
+       const char *textcat_failed;
 };
 
 pool_t fts_languages_pool;
@@ -224,7 +224,8 @@ static bool fts_language_match_lists(struct fts_language_list *list,
 #endif
 
 #ifdef HAVE_FTS_EXTTEXTCAT
-static int fts_language_textcat_init(struct fts_language_list *list)
+static int fts_language_textcat_init(struct fts_language_list *list,
+                                    const char **error_r)
 {
        const char *config_path;
        const char *data_dir;
@@ -232,8 +233,10 @@ static int fts_language_textcat_init(struct fts_language_list *list)
        if (list->textcat_handle != NULL)
                return 0;
 
-       if (list->textcat_failed)
+       if (list->textcat_failed != NULL) {
+               *error_r = list->textcat_failed;
                return -1;
+       }
            
        config_path = list->textcat_config != NULL ? list->textcat_config :
                TEXTCAT_DATADIR"/fpdb.conf";
@@ -241,9 +244,9 @@ static int fts_language_textcat_init(struct fts_language_list *list)
                TEXTCAT_DATADIR"/";
        list->textcat_handle = special_textcat_Init(config_path, data_dir);
        if (list->textcat_handle == NULL) {
-               i_error("special_textcat_Init(%s, %s) failed",
+               *error_r = list->textcat_failed = p_strdup_printf(list->pool,
+                       "special_textcat_Init(%s, %s) failed",
                        config_path, data_dir);
-               list->textcat_failed = TRUE;
                return -1;
        }
        /* The textcat minimum document size could be set here. It
@@ -256,14 +259,15 @@ static enum fts_language_result
 fts_language_detect_textcat(struct fts_language_list *list ATTR_UNUSED,
                            const unsigned char *text ATTR_UNUSED,
                            size_t size ATTR_UNUSED,
-                           const struct fts_language **lang_r ATTR_UNUSED)
+                           const struct fts_language **lang_r ATTR_UNUSED,
+                           const char **error_r ATTR_UNUSED)
 {
 #ifdef HAVE_FTS_EXTTEXTCAT
        candidate_t *candp; /* textcat candidate result array pointer */
        int cnt;
        bool match = FALSE;
 
-       if (fts_language_textcat_init(list) < 0)
+       if (fts_language_textcat_init(list, error_r) < 0)
                return FTS_LANGUAGE_RESULT_ERROR;
 
        candp = textcat_GetClassifyFullOutput(list->textcat_handle);
@@ -301,7 +305,8 @@ enum fts_language_result
 fts_language_detect(struct fts_language_list *list,
                    const unsigned char *text ATTR_UNUSED,
                    size_t size ATTR_UNUSED,
-                    const struct fts_language **lang_r)
+                   const struct fts_language **lang_r,
+                   const char **error_r)
 {
        i_assert(array_count(&list->languages) > 0);
 
@@ -312,5 +317,5 @@ fts_language_detect(struct fts_language_list *list,
                *lang_r = *langp;
                return FTS_LANGUAGE_RESULT_OK;
        }
-       return fts_language_detect_textcat(list, text, size, lang_r);
+       return fts_language_detect_textcat(list, text, size, lang_r, error_r);
 }

diff --git a/src/lib-fts/fts-language.h b/src/lib-fts/fts-language.h
index 85b2bc22dfb96d1a9349f0fd4ca052eb6dc85414..884998f07f43ed006faa8573e3c8b88d80bad3ed 100644 (file)
--- a/src/lib-fts/fts-language.h
+++ b/src/lib-fts/fts-language.h
@@ -62,10 +62,11 @@ fts_language_list_get_first(struct fts_language_list *list);
 
 /* If text was detected to be one of the languages in the list,
    returns FTS_LANGUAGE_RESULT_OK and (a pointer to) the language (in
-   the list). */
+   the list). error_r is set for FTS_LANGUAGE_RESULT_ERROR. */
 enum fts_language_result
 fts_language_detect(struct fts_language_list *list,
                    const unsigned char *text, size_t size,
-                    const struct fts_language **lang_r);
+                   const struct fts_language **lang_r,
+                   const char **error_r);
 
 #endif

diff --git a/src/lib-fts/test-fts-language.c b/src/lib-fts/test-fts-language.c
index d7afc6b928eb0024446f42a2ba1e016731b290fd..e5968e9bdcdede39190d3c8f12c26e9e848adc38 100644 (file)
--- a/src/lib-fts/test-fts-language.c
+++ b/src/lib-fts/test-fts-language.c
@@ -28,7 +28,7 @@ static void test_fts_language_detect_finnish(void)
        test_begin("fts language detect Finnish");
        test_assert(fts_language_list_init(settings, &lp, &error) == 0);
        test_assert(fts_language_list_add_names(lp, names, &unknown) == TRUE);
-       test_assert(fts_language_detect(lp, finnish, sizeof(finnish)-1, &lang_r)
+       test_assert(fts_language_detect(lp, finnish, sizeof(finnish)-1, &lang_r, &error)
                    == FTS_LANGUAGE_RESULT_OK);
        test_assert(strcmp(lang_r->name, "fi") == 0);
        fts_language_list_deinit(&lp);
@@ -55,7 +55,7 @@ static void test_fts_language_detect_english(void)
        test_begin("fts language detect English");
        test_assert(fts_language_list_init(settings, &lp, &error) == 0);
        test_assert(fts_language_list_add_names(lp, names, &unknown) == TRUE);
-       test_assert(fts_language_detect(lp, english, sizeof(english)-1, &lang_r)
+       test_assert(fts_language_detect(lp, english, sizeof(english)-1, &lang_r, &error)
                    == FTS_LANGUAGE_RESULT_OK);
        test_assert(strcmp(lang_r->name, "en") == 0);
        fts_language_list_deinit(&lp);
@@ -90,7 +90,7 @@ static void test_fts_language_detect_french(void)
        test_begin("fts language detect French");
        test_assert(fts_language_list_init(settings, &lp, &error) == 0);
        test_assert(fts_language_list_add_names(lp, names, &unknown) == TRUE);
-       test_assert(fts_language_detect(lp, french, sizeof(french)-1, &lang_r)
+       test_assert(fts_language_detect(lp, french, sizeof(french)-1, &lang_r, &error)
                    == FTS_LANGUAGE_RESULT_OK);
        test_assert(strcmp(lang_r->name, "fr") == 0);
        fts_language_list_deinit(&lp);
@@ -127,7 +127,7 @@ static void test_fts_language_detect_german(void)
        test_begin("fts language detect German");
        test_assert(fts_language_list_init(settings, &lp, &error) == 0);
        test_assert(fts_language_list_add_names(lp, names, &unknown) == TRUE);
-       test_assert(fts_language_detect(lp, german, sizeof(german)-1, &lang_r)
+       test_assert(fts_language_detect(lp, german, sizeof(german)-1, &lang_r, &error)
                    == FTS_LANGUAGE_RESULT_OK);
        test_assert(strcmp(lang_r->name, "de") == 0);
        fts_language_list_deinit(&lp);
@@ -153,7 +153,7 @@ static void test_fts_language_detect_swedish(void)
        test_begin("fts language detect Swedish");
        test_assert(fts_language_list_init(settings, &lp, &error) == 0);
        test_assert(fts_language_list_add_names(lp, names, &unknown) == TRUE);
-       test_assert(fts_language_detect(lp, swedish, sizeof(swedish)-1, &lang_r)
+       test_assert(fts_language_detect(lp, swedish, sizeof(swedish)-1, &lang_r, &error)
                    == FTS_LANGUAGE_RESULT_OK);
        test_assert(strcmp(lang_r->name, "sv") == 0);
        fts_language_list_deinit(&lp);
@@ -177,7 +177,7 @@ static void test_fts_language_detect_bokmal(void)
        test_begin("fts language detect Bokmal as Norwegian");
        test_assert(fts_language_list_init(settings, &lp, &error) == 0);
        test_assert(fts_language_list_add_names(lp, names, &unknown) == TRUE);
-       test_assert(fts_language_detect(lp, bokmal, sizeof(bokmal)-1, &lang_r)
+       test_assert(fts_language_detect(lp, bokmal, sizeof(bokmal)-1, &lang_r, &error)
                    == FTS_LANGUAGE_RESULT_OK);
        test_assert(strcmp(lang_r->name, "no") == 0);
        fts_language_list_deinit(&lp);
@@ -201,7 +201,7 @@ static void test_fts_language_detect_nynorsk(void)
        test_begin("fts language detect Nynorsk as Norwegian");
        test_assert(fts_language_list_init(settings, &lp, &error) == 0);
        test_assert(fts_language_list_add_names(lp, names, &unknown) == TRUE);
-       test_assert(fts_language_detect(lp, nynorsk, sizeof(nynorsk)-1, &lang_r)
+       test_assert(fts_language_detect(lp, nynorsk, sizeof(nynorsk)-1, &lang_r, &error)
                    == FTS_LANGUAGE_RESULT_OK);
        test_assert(strcmp(lang_r->name, "no") == 0);
        fts_language_list_deinit(&lp);
@@ -226,7 +226,7 @@ static void test_fts_language_detect_finnish_as_english(void)
        test_begin("fts language detect Finnish as English");
        test_assert(fts_language_list_init(settings, &lp, &error) == 0);
        test_assert(fts_language_list_add_names(lp, names, &unknown) == TRUE);
-       test_assert(fts_language_detect(lp, finnish, sizeof(finnish)-1, &lang_r)
+       test_assert(fts_language_detect(lp, finnish, sizeof(finnish)-1, &lang_r, &error)
                    == FTS_LANGUAGE_RESULT_OK);
        test_assert(strcmp(lang_r->name, "en") == 0);
        fts_language_list_deinit(&lp);
@@ -253,7 +253,7 @@ static void test_fts_language_detect_na(void)
        test_begin("fts language detect not available");
        test_assert(fts_language_list_init(settings, &lp, &error) == 0);
        test_assert(fts_language_list_add_names(lp, names, &unknown) == TRUE);
-       test_assert(fts_language_detect(lp, english, sizeof(english)-1, &lang_r)
+       test_assert(fts_language_detect(lp, english, sizeof(english)-1, &lang_r, &error)
                    == FTS_LANGUAGE_RESULT_UNKNOWN);
        fts_language_list_deinit(&lp);
        test_end();
@@ -273,7 +273,7 @@ static void test_fts_language_detect_unknown(void)
        test_begin("fts language detect unknown");
        test_assert(fts_language_list_init(settings, &lp, &error) == 0);
        test_assert(fts_language_list_add_names(lp, names, &unknown) == TRUE);
-       test_assert(fts_language_detect(lp, klingon, sizeof(klingon), &lang_r)
+       test_assert(fts_language_detect(lp, klingon, sizeof(klingon), &lang_r, &error)
                    == FTS_LANGUAGE_RESULT_UNKNOWN);
        fts_language_list_deinit(&lp);
        test_end();

diff --git a/src/plugins/fts/doveadm-fts.c b/src/plugins/fts/doveadm-fts.c
index 697e3d07b7f63c975debb231c848b469b0800091..1b902a1ee69c2efdcac4624d6d23e13438f768e7 100644 (file)
--- a/src/plugins/fts/doveadm-fts.c
+++ b/src/plugins/fts/doveadm-fts.c
@@ -190,10 +190,11 @@ cmd_fts_tokenize_run(struct doveadm_mail_cmd_context *_ctx,
                struct fts_language_list *lang_list =
                        fts_user_get_language_list(user);
                enum fts_language_result result;
+               const char *error;
 
                result = fts_language_detect(lang_list,
                    (const unsigned char *)ctx->tokens, strlen(ctx->tokens),
-                    &lang);
+                    &lang, &error);
                if (lang == NULL)
                        lang = fts_language_list_get_first(lang_list);
                switch (result) {
@@ -206,7 +207,7 @@ cmd_fts_tokenize_run(struct doveadm_mail_cmd_context *_ctx,
                case FTS_LANGUAGE_RESULT_OK:
                        break;
                case FTS_LANGUAGE_RESULT_ERROR:
-                       i_error("Language detection library initialization failed");
+                       i_error("Language detection library initialization failed: %s", error);
                        _ctx->exit_code = EX_CONFIG;
                        return -1;
                default:

diff --git a/src/plugins/fts/fts-build-mail.c b/src/plugins/fts/fts-build-mail.c
index e088e7397b7713267a5194b923c559034075d39a..ec353380a589852d1a2e89ba22f00f8ca9fe4b58 100644 (file)
--- a/src/plugins/fts/fts-build-mail.c
+++ b/src/plugins/fts/fts-build-mail.c
@@ -298,8 +298,9 @@ fts_detect_language(struct fts_mail_build_context *ctx,
        struct mail_user *user = ctx->update_ctx->backend->ns->user;
        struct fts_language_list *lang_list = fts_user_get_language_list(user);
        const struct fts_language *lang;
+       const char *error;
 
-       switch (fts_language_detect(lang_list, data, size, &lang)) {
+       switch (fts_language_detect(lang_list, data, size, &lang, &error)) {
        case FTS_LANGUAGE_RESULT_SHORT:
                /* save the input so far and try again later */
                buffer_append(ctx->pending_input, data, size);
@@ -319,6 +320,7 @@ fts_detect_language(struct fts_mail_build_context *ctx,
        case FTS_LANGUAGE_RESULT_ERROR:
                /* internal language detection library failure
                   (e.g. invalid config). don't index anything. */
+               i_error("Language detection library initialization failed: %s", error);
                return -1;
        default:
                i_unreached();