diff options
| author | frosty <gabriel@bwaaa.monster> | 2026-08-16 23:45:43 -0400 |
|---|---|---|
| committer | frosty <gabriel@bwaaa.monster> | 2026-08-16 23:45:43 -0400 |
| commit | e352971bc59c356886f57d97d01ce195c3b0d952 (patch) | |
| tree | ebadb7b28802a2c8396935fe8756931d0038127f /src/Routes/Search.c | |
| parent | 499bb9b1268cd422619efdc46889960425462aae (diff) | |
| parent | 12f046669e6a9ebd5e9d47ccf0ddd265fcdcdd18 (diff) | |
| download | omnisearch-e352971bc59c356886f57d97d01ce195c3b0d952.tar.gz | |
Merge branch 'indev'
Diffstat (limited to 'src/Routes/Search.c')
| -rw-r--r-- | src/Routes/Search.c | 457 |
1 files changed, 171 insertions, 286 deletions
diff --git a/src/Routes/Search.c b/src/Routes/Search.c index b580c10..7fcdac8 100644 --- a/src/Routes/Search.c +++ b/src/Routes/Search.c @@ -140,29 +140,31 @@ static InfoBox fetch_currency_wrapper(char *query) { return fetch_currency_data(query); } char *get_base_url(const char *input) { - if (!input) return NULL; + if (!input) + return NULL; - const char *start = input; + const char *start = input; - const char *protocol_pos = strstr(input, "://"); - if (protocol_pos) { - start = protocol_pos + 3; - } + const char *protocol_pos = strstr(input, "://"); + if (protocol_pos) { + start = protocol_pos + 3; + } - const char *end = start; - while (*end && *end != '/' && *end != '?' && *end != '#') { - end++; - } + const char *end = start; + while (*end && *end != '/' && *end != '?' && *end != '#') { + end++; + } - size_t len = end - start; + size_t len = end - start; - char *domain = (char *)malloc(len + 1); - if (!domain) return NULL; + char *domain = (char *)malloc(len + 1); + if (!domain) + return NULL; - strncpy(domain, start, len); - domain[len] = '\0'; + strncpy(domain, start, len); + domain[len] = '\0'; - return domain; + return domain; } static int is_calculator_query(const char *query) { if (!query) @@ -249,88 +251,65 @@ static void *infobox_thread_func(void *arg) { return NULL; } -static int add_infobox_to_collection(InfoBox *infobox, char ****collection, - int **inner_counts, int current_count) { - *collection = - (char ***)realloc(*collection, sizeof(char **) * (current_count + 1)); - *inner_counts = - (int *)realloc(*inner_counts, sizeof(int) * (current_count + 1)); - - (*collection)[current_count] = - (char **)malloc(sizeof(char *) * INFOBOX_FIELD_COUNT); - (*collection)[current_count][0] = - infobox->title ? strdup(infobox->title) : NULL; - (*collection)[current_count][1] = - infobox->thumbnail_url ? strdup(infobox->thumbnail_url) : NULL; - (*collection)[current_count][2] = - infobox->extract ? strdup(infobox->extract) : NULL; - (*collection)[current_count][3] = infobox->url ? strdup(infobox->url) : NULL; - (*collection)[current_count][4] = infobox->url ? strdup(infobox->url) : NULL; - (*inner_counts)[current_count] = INFOBOX_FIELD_COUNT; - - return current_count + 1; +static int add_infobox_to_collection(InfoBox *infobox, + StringMatrix *collection) { + const char *values[INFOBOX_FIELD_COUNT] = { + infobox->title, infobox->thumbnail_url, infobox->extract, + infobox->url, infobox->url, + }; + return string_matrix_append(collection, values, INFOBOX_FIELD_COUNT); } static int add_warning_to_collection(const char *engine_name, const char *warning_message, - char ****collection, int **inner_counts, - int current_count) { - char ***new_collection = - (char ***)malloc(sizeof(char **) * (current_count + 1)); - int *new_inner_counts = - (int *)malloc(sizeof(int) * (current_count + 1)); + StringMatrix *collection) { + const char *values[] = {engine_name, warning_message}; + return string_matrix_append(collection, values, 2); +} - if (!new_collection || !new_inner_counts) { - free(new_collection); - free(new_inner_counts); - return current_count; - } +static void free_user_engine_list(char **user_engines, int user_engine_count) { + for (int i = 0; i < user_engine_count; i++) + free(user_engines[i]); + free(user_engines); +} - if (*collection && current_count > 0) { - memcpy(new_collection, *collection, sizeof(char **) * current_count); - } - if (*inner_counts && current_count > 0) { - memcpy(new_inner_counts, *inner_counts, sizeof(int) * current_count); +static void free_search_results(SearchResult **all_results, ScrapeJob *jobs, + int job_count) { + for (int i = 0; i < job_count; i++) { + for (int j = 0; all_results[i] && j < jobs[i].results_count; j++) { + free(all_results[i][j].url); + free(all_results[i][j].title); + free(all_results[i][j].snippet); + } + free(all_results[i]); } +} - free(*collection); - free(*inner_counts); - - *collection = new_collection; - *inner_counts = new_inner_counts; - - (*collection)[current_count] = (char **)malloc(sizeof(char *) * 2); - if (!(*collection)[current_count]) - return current_count; - - (*collection)[current_count][0] = strdup(engine_name ? engine_name : ""); - (*collection)[current_count][1] = - strdup(warning_message ? warning_message : ""); - - if (!(*collection)[current_count][0] || !(*collection)[current_count][1]) { - free((*collection)[current_count][0]); - free((*collection)[current_count][1]); - free((*collection)[current_count]); - return current_count; +static void free_infobox_results(InfoBoxThreadData *data) { + for (int i = 0; i < HANDLER_COUNT; i++) { + free_infobox(&data[i].result); } - - (*inner_counts)[current_count] = 2; - return current_count + 1; } -static const char *warning_message_for_job(const ScrapeJob *job, const char *locale) { +static const char *warning_message_for_job(const ScrapeJob *job, + const char *locale) { switch (job->status) { case SCRAPE_STATUS_FETCH_ERROR: { const char *msg = beaker_get_locale_value(locale, "warning_fetch_error"); - return msg ? msg : "request failed before OmniSearch could read search results."; + return msg ? msg + : "request failed before OmniSearch could read search results."; } case SCRAPE_STATUS_PARSE_MISMATCH: { const char *msg = beaker_get_locale_value(locale, "warning_parse_mismatch"); - return msg ? msg : "returned search results in a format OmniSearch could not parse."; + return msg ? msg + : "returned search results in a format OmniSearch could not " + "parse."; } case SCRAPE_STATUS_BLOCKED: { const char *msg = beaker_get_locale_value(locale, "warning_blocked"); - return msg ? msg : "returned a captcha or another blocking page instead of search results."; + return msg ? msg + : "returned a captcha or another blocking page instead of " + "search results."; } default: return NULL; @@ -384,8 +363,8 @@ static int engine_allowed_for_user(const SearchEngine *eng, char **user_ids, static char *build_search_href(const char *query, const char *engine_id, int page) { const char *safe_query = query ? query : ""; - int use_engine = engine_id && engine_id[0] != '\0' && - !engine_id_matches(engine_id, "all"); + int use_engine = + engine_id && engine_id[0] != '\0' && !engine_id_matches(engine_id, "all"); size_t needed = strlen("/search?q=") + strlen(safe_query) + 1; if (use_engine) @@ -437,13 +416,18 @@ int results_handler(UrlParams *params) { extern Config global_config; TemplateContext ctx = new_context(); char *raw_query = ""; - const char *selected_engine_id = "all"; + char selected_engine_id[32]; + char *default_engine = get_default_search_engine(); + snprintf(selected_engine_id, sizeof(selected_engine_id), "%s", + default_engine); + free(default_engine); int page = 1; int btnI = 0; char **user_engines = NULL; int user_engine_count = 0; - int has_user_pref = (get_user_engines(&user_engines, &user_engine_count) == 0); + int has_user_pref = + (get_user_engines(&user_engines, &user_engine_count) == 0); if (params) { for (int i = 0; i < params->count; i++) { @@ -454,7 +438,8 @@ int results_handler(UrlParams *params) { if (parsed > 1) page = parsed; } else if (strcmp(params->params[i].key, "engine") == 0) { - selected_engine_id = params->params[i].value; + snprintf(selected_engine_id, sizeof(selected_engine_id), "%s", + params->params[i].value); } else if (strcmp(params->params[i].key, "btnI") == 0) { btnI = atoi(params->params[i].value); } @@ -471,9 +456,11 @@ int results_handler(UrlParams *params) { beaker_set_locale(&ctx, locale); const char *rate_limit_msg = beaker_get_locale_value(locale, "rate_limit"); - if (!rate_limit_msg) rate_limit_msg = "Slow down! Too many searches from you!"; + if (!rate_limit_msg) + rate_limit_msg = "Slow down! Too many searches from you!"; const char *no_results_msg = beaker_get_locale_value(locale, "no_results"); - if (!no_results_msg) no_results_msg = "No results found"; + if (!no_results_msg) + no_results_msg = "No results found"; char page_str[16]; snprintf(page_str, sizeof(page_str), "%d", page); @@ -491,18 +478,24 @@ int results_handler(UrlParams *params) { if (!raw_query || strlen(raw_query) == 0) { send_redirect("/"); - if (has_user_pref) { - for (int i = 0; i < user_engine_count; i++) - free(user_engines[i]); - free(user_engines); - } + free(locale); + free_user_engine_list(user_engines, user_engine_count); free_context(&ctx); return -1; } const SearchEngine *selected_engine = find_enabled_engine(selected_engine_id); - if (!selected_engine) - selected_engine_id = "all"; + if (selected_engine && + !engine_allowed_for_user(selected_engine, user_engines, + user_engine_count, has_user_pref)) { + selected_engine = NULL; + } + if (selected_engine) { + snprintf(selected_engine_id, sizeof(selected_engine_id), "%s", + selected_engine->id); + } else { + snprintf(selected_engine_id, sizeof(selected_engine_id), "all"); + } context_set(&ctx, "selected_engine", selected_engine_id); char *search_href = build_search_href(raw_query, selected_engine_id, 1); @@ -528,13 +521,6 @@ int results_handler(UrlParams *params) { infobox_data[i].result = (InfoBox){NULL}; } - if (page == 1) { - for (int i = 0; i < HANDLER_COUNT; i++) { - pthread_create(&infobox_threads[i], NULL, infobox_thread_func, - &infobox_data[i]); - } - } - ScrapeJob jobs[ENGINE_COUNT]; SearchResult *all_results[ENGINE_COUNT]; @@ -592,11 +578,8 @@ int results_handler(UrlParams *params) { snprintf(response, sizeof(response), "<h1>%s</h1>", rate_limit_msg); send_response(response); free(request_cache_key); - if (has_user_pref) { - for (int i = 0; i < user_engine_count; i++) - free(user_engines[i]); - free(user_engines); - } + free(locale); + free_user_engine_list(user_engines, user_engine_count); free_context(&ctx); return -1; } @@ -615,15 +598,14 @@ int results_handler(UrlParams *params) { } if (filter_engine_count > 1) { - char ***filter_matrix = NULL; - int *filter_inner_counts = NULL; - int filter_count = 0; + StringMatrix filter_matrix; + string_matrix_init(&filter_matrix); char *all_href = build_search_href(raw_query, "all", 1); - - filter_count = add_link_to_collection( + const char *all_values[LINK_FIELD_COUNT] = { all_href, "All", - selected_engine ? "engine-filter" : "engine-filter active", - &filter_matrix, &filter_inner_counts, filter_count); + selected_engine ? "engine-filter" : "engine-filter active"}; + + string_matrix_append(&filter_matrix, all_values, LINK_FIELD_COUNT); free(all_href); for (int i = 0; i < ENGINE_COUNT; i++) { @@ -639,22 +621,24 @@ int results_handler(UrlParams *params) { ? "engine-filter active" : "engine-filter"; - filter_count = add_link_to_collection(filter_href, ENGINE_REGISTRY[i].name, - filter_class, &filter_matrix, - &filter_inner_counts, filter_count); + const char *filter_values[LINK_FIELD_COUNT] = { + filter_href, ENGINE_REGISTRY[i].name, filter_class}; + string_matrix_append(&filter_matrix, filter_values, LINK_FIELD_COUNT); free(filter_href); } - if (filter_count > 0) { - context_set_array_of_arrays(&ctx, "engine_filters", filter_matrix, - filter_count, filter_inner_counts); - for (int i = 0; i < filter_count; i++) { - for (int j = 0; j < LINK_FIELD_COUNT; j++) - free(filter_matrix[i][j]); - free(filter_matrix[i]); - } - free(filter_matrix); - free(filter_inner_counts); + if (filter_matrix.count > 0) { + context_set_array_of_arrays(&ctx, "engine_filters", filter_matrix.rows, + filter_matrix.count, + filter_matrix.field_counts); + } + string_matrix_free(&filter_matrix); + } + + if (page == 1) { + for (int i = 0; i < HANDLER_COUNT; i++) { + pthread_create(&infobox_threads[i], NULL, infobox_thread_func, + &infobox_data[i]); } } @@ -672,27 +656,12 @@ int results_handler(UrlParams *params) { for (int i = 0; i < engine_idx; i++) { if (jobs[i].results_count > 0 && all_results[i][0].url) { char *redirect_url = strdup(all_results[i][0].url); - for (int j = 0; j < enabled_engine_count; j++) { - for (int k = 0; k < jobs[j].results_count; k++) { - free(all_results[j][k].url); - free(all_results[j][k].title); - free(all_results[j][k].snippet); - } - free(all_results[j]); - } - if (page == 1) { - for (int j = 0; j < HANDLER_COUNT; j++) { - if (infobox_data[j].success) { - free_infobox(&infobox_data[j].result); - } - } - } + free_search_results(all_results, jobs, engine_idx); + if (page == 1) + free_infobox_results(infobox_data); free(request_cache_key); - if (has_user_pref) { - for (int i = 0; i < user_engine_count; i++) - free(user_engines[i]); - free(user_engines); - } + free(locale); + free_user_engine_list(user_engines, user_engine_count); free_context(&ctx); if (redirect_url) { send_redirect(redirect_url); @@ -701,54 +670,37 @@ int results_handler(UrlParams *params) { return 0; } } - for (int i = 0; i < enabled_engine_count; i++) { - free(all_results[i]); - } - if (page == 1) { - for (int i = 0; i < HANDLER_COUNT; i++) { - if (infobox_data[i].success) { - free_infobox(&infobox_data[i].result); - } - } - } + free_search_results(all_results, jobs, engine_idx); + if (page == 1) + free_infobox_results(infobox_data); free(request_cache_key); - if (has_user_pref) { - for (int i = 0; i < user_engine_count; i++) - free(user_engines[i]); - free(user_engines); - } + free(locale); + free_user_engine_list(user_engines, user_engine_count); free_context(&ctx); char no_results_html[128]; - snprintf(no_results_html, sizeof(no_results_html), "<h1>%s</h1>", no_results_msg); + snprintf(no_results_html, sizeof(no_results_html), "<h1>%s</h1>", + no_results_msg); send_response(no_results_html); return 0; } - - char ***infobox_matrix = NULL; - int *infobox_inner_counts = NULL; - int infobox_count = 0; + + StringMatrix infobox_matrix; + string_matrix_init(&infobox_matrix); if (page == 1) { for (int i = 0; i < HANDLER_COUNT; i++) { if (infobox_data[i].success) { - infobox_count = - add_infobox_to_collection(&infobox_data[i].result, &infobox_matrix, - &infobox_inner_counts, infobox_count); + add_infobox_to_collection(&infobox_data[i].result, &infobox_matrix); } } } - if (infobox_count > 0) { - context_set_array_of_arrays(&ctx, "infoboxes", infobox_matrix, - infobox_count, infobox_inner_counts); - for (int i = 0; i < infobox_count; i++) { - for (int j = 0; j < INFOBOX_FIELD_COUNT; j++) - free(infobox_matrix[i][j]); - free(infobox_matrix[i]); - } - free(infobox_matrix); - free(infobox_inner_counts); + if (infobox_matrix.count > 0) { + context_set_array_of_arrays(&ctx, "infoboxes", infobox_matrix.rows, + infobox_matrix.count, + infobox_matrix.field_counts); } + string_matrix_free(&infobox_matrix); int warning_count = 0; for (int i = 0; i < enabled_engine_count; i++) { @@ -757,35 +709,25 @@ int results_handler(UrlParams *params) { } if (warning_count > 0) { - char ***warning_matrix = NULL; - int *warning_inner_counts = NULL; - int warning_index = 0; + StringMatrix warning_matrix; + string_matrix_init(&warning_matrix); for (int i = 0; i < enabled_engine_count; i++) { const char *warning_message = warning_message_for_job(&jobs[i], locale); if (!warning_message) continue; - warning_index = add_warning_to_collection( - jobs[i].engine->name, warning_message, &warning_matrix, - &warning_inner_counts, warning_index); + add_warning_to_collection(jobs[i].engine->name, warning_message, + &warning_matrix); } - if (warning_index > 0) { - context_set_array_of_arrays(&ctx, "engine_warnings", warning_matrix, - warning_index, warning_inner_counts); + if (warning_matrix.count > 0) { + context_set_array_of_arrays(&ctx, "engine_warnings", warning_matrix.rows, + warning_matrix.count, + warning_matrix.field_counts); } - if (warning_matrix) { - for (int i = 0; i < warning_index; i++) { - free(warning_matrix[i][0]); - free(warning_matrix[i][1]); - free(warning_matrix[i]); - } - free(warning_matrix); - } - if (warning_inner_counts) - free(warning_inner_counts); + string_matrix_free(&warning_matrix); } int total_results = 0; @@ -794,33 +736,8 @@ int results_handler(UrlParams *params) { } if (total_results > 0) { - char ***results_matrix = (char ***)malloc(sizeof(char **) * total_results); - int *results_inner_counts = (int *)malloc(sizeof(int) * total_results); - if (!results_matrix || !results_inner_counts) { - char *html = render_template("results.html", &ctx); - if (html) { - send_response(html); - free(html); - } - for (int i = 0; i < enabled_engine_count; i++) - free(all_results[i]); - if (page == 1) { - for (int i = 0; i < HANDLER_COUNT; i++) { - if (infobox_data[i].success) { - free_infobox(&infobox_data[i].result); - } - } - } - free(request_cache_key); - if (has_user_pref) { - for (int i = 0; i < user_engine_count; i++) - free(user_engines[i]); - free(user_engines); - } - free_context(&ctx); - return 0; - } - int unique_count = 0; + StringMatrix results_matrix; + string_matrix_init(&results_matrix); UrlHashTable url_table; url_hash_init(&url_table); @@ -837,61 +754,45 @@ int results_handler(UrlParams *params) { url_hash_insert(&url_table, display_url); - results_matrix[unique_count] = - (char **)malloc(sizeof(char *) * RESULT_FIELD_COUNT); - if (!results_matrix[unique_count]) { - free(all_results[i][j].url); - free(all_results[i][j].title); - free(all_results[i][j].snippet); - continue; - } char *pretty_url = pretty_display_url(display_url); char *base_url = get_base_url(display_url); - - results_matrix[unique_count][0] = strdup(display_url); - results_matrix[unique_count][1] = strdup(pretty_url); - results_matrix[unique_count][2] = all_results[i][j].title - ? strdup(all_results[i][j].title) - : strdup("Untitled"); - results_matrix[unique_count][3] = - all_results[i][j].snippet ? strdup(all_results[i][j].snippet) - : strdup(""); - results_matrix[unique_count][4] = strdup(base_url ? base_url : ""); - results_matrix[unique_count][5] = strdup(""); - - results_inner_counts[unique_count] = RESULT_FIELD_COUNT; + const char *values[RESULT_FIELD_COUNT] = { + display_url, + pretty_url, + all_results[i][j].title ? all_results[i][j].title : "Untitled", + all_results[i][j].snippet ? all_results[i][j].snippet : "", + base_url, + "", + }; + int append_result = + string_matrix_append(&results_matrix, values, RESULT_FIELD_COUNT); free(pretty_url); free(base_url); free(all_results[i][j].url); free(all_results[i][j].title); free(all_results[i][j].snippet); - - unique_count++; + if (append_result != 0) { + continue; + } } free(all_results[i]); } - context_set_array_of_arrays(&ctx, "results", results_matrix, unique_count, - results_inner_counts); + context_set_array_of_arrays(&ctx, "results", results_matrix.rows, + results_matrix.count, + results_matrix.field_counts); - char ***pager_matrix = NULL; - int *pager_inner_counts = NULL; - SearchHrefData href_data = { .query = raw_query, .engine_id = selected_engine_id }; - int pager_count = build_pagination(page, search_href_builder, - &href_data, &pager_matrix, - &pager_inner_counts); + StringMatrix pager_matrix; + SearchHrefData href_data = {.query = raw_query, + .engine_id = selected_engine_id}; + int pagination_result = string_matrix_build_pagination( + page, search_href_builder, &href_data, &pager_matrix); - if (pager_count > 0) { - context_set_array_of_arrays(&ctx, "pagination_links", pager_matrix, - pager_count, pager_inner_counts); - for (int i = 0; i < pager_count; i++) { - for (int j = 0; j < LINK_FIELD_COUNT; j++) - free(pager_matrix[i][j]); - free(pager_matrix[i]); - } - free(pager_matrix); - free(pager_inner_counts); + if (pagination_result == 0 && pager_matrix.count > 0) { + context_set_array_of_arrays(&ctx, "pagination_links", pager_matrix.rows, + pager_matrix.count, + pager_matrix.field_counts); } char *html = render_template("results.html", &ctx); @@ -900,13 +801,8 @@ int results_handler(UrlParams *params) { free(html); } - for (int i = 0; i < unique_count; i++) { - for (int j = 0; j < RESULT_FIELD_COUNT; j++) - free(results_matrix[i][j]); - free(results_matrix[i]); - } - free(results_matrix); - free(results_inner_counts); + string_matrix_free(&pager_matrix); + string_matrix_free(&results_matrix); url_hash_free(&url_table); } else { char *html = render_template("results.html", &ctx); @@ -915,26 +811,15 @@ int results_handler(UrlParams *params) { free(html); } - for (int i = 0; i < enabled_engine_count; i++) { - free(all_results[i]); - } + free_search_results(all_results, jobs, engine_idx); } free(request_cache_key); - if (page == 1) { - for (int i = 0; i < HANDLER_COUNT; i++) { - if (infobox_data[i].success) { - free_infobox(&infobox_data[i].result); - } - } - } + if (page == 1) + free_infobox_results(infobox_data); free(locale); - if (has_user_pref) { - for (int i = 0; i < user_engine_count; i++) - free(user_engines[i]); - free(user_engines); - } + free_user_engine_list(user_engines, user_engine_count); free_context(&ctx); return 0; |
