From f3aa7ca0bc2ef7c286609e8f87d07cc2568093af Mon Sep 17 00:00:00 2001 From: frosty Date: Tue, 6 Jan 2026 23:46:24 -0500 Subject: rebase(d) --- src/Routes/Home.c | 14 +++ src/Routes/Home.h | 8 ++ src/Routes/Images.c | 277 ++++++++++++++++++++++++++++++++++++++++++++++++++++ src/Routes/Images.h | 8 ++ src/Routes/Search.c | 273 +++++++++++++++++++++++++++++++++++++++++++++++++++ src/Routes/Search.h | 8 ++ 6 files changed, 588 insertions(+) create mode 100644 src/Routes/Home.c create mode 100644 src/Routes/Home.h create mode 100644 src/Routes/Images.c create mode 100644 src/Routes/Images.h create mode 100644 src/Routes/Search.c create mode 100644 src/Routes/Search.h (limited to 'src/Routes') diff --git a/src/Routes/Home.c b/src/Routes/Home.c new file mode 100644 index 0000000..81370ba --- /dev/null +++ b/src/Routes/Home.c @@ -0,0 +1,14 @@ +#include "Home.h" +#include + +int home_handler(UrlParams *params) { + (void)params; + TemplateContext ctx = new_context(); + char *rendered_html = render_template("home.html", &ctx); + send_response(rendered_html); + + free(rendered_html); + free_context(&ctx); + + return 0; +} diff --git a/src/Routes/Home.h b/src/Routes/Home.h new file mode 100644 index 0000000..5d01ab3 --- /dev/null +++ b/src/Routes/Home.h @@ -0,0 +1,8 @@ +#ifndef HOME_H +#define HOME_H + +#include + +int home_handler(UrlParams *params); + +#endif diff --git a/src/Routes/Images.c b/src/Routes/Images.c new file mode 100644 index 0000000..47e3a72 --- /dev/null +++ b/src/Routes/Images.c @@ -0,0 +1,277 @@ +#include "Images.h" +#include "../Utility/Unescape.h" + +#include +#include +#include +#include +#include +#include +#include + +static void get_image_timestamp(char *buffer, size_t size) { + time_t now = time(NULL); + struct tm *t = localtime(&now); + if (t) { + strftime(buffer, size, "%Y-%m-%d %H:%M:%S", t); + } +} + +#define IMG_LOG_INFO(msg, ...) \ + { \ + char ts[20]; \ + get_image_timestamp(ts, sizeof(ts)); \ + fprintf(stderr, "[%s] INFO [ImagesHandler] " msg "\n", ts, \ + ##__VA_ARGS__); \ + } + +#define IMG_LOG_ERROR(msg, ...) \ + { \ + char ts[20]; \ + get_image_timestamp(ts, sizeof(ts)); \ + fprintf(stderr, "[%s] ERROR [ImagesHandler] " msg "\n", ts, \ + ##__VA_ARGS__); \ + } + +struct MemoryBlock { + char *response; + size_t size; +}; + +static size_t ImageWriteCallback(void *data, size_t size, size_t nmemb, + void *userp) { + size_t realsize = size * nmemb; + struct MemoryBlock *mem = (struct MemoryBlock *)userp; + char *ptr = (char *)realloc(mem->response, mem->size + realsize + 1); + if (ptr == NULL) { + IMG_LOG_ERROR("Realloc failed in WriteCallback (out of memory)"); + return 0; + } + mem->response = ptr; + memcpy(&(mem->response[mem->size]), data, realsize); + mem->size += realsize; + mem->response[mem->size] = 0; + return realsize; +} + +static char *fetch_images_html(const char *url) { + CURL *curl_handle; + struct MemoryBlock chunk = {.response = malloc(1), .size = 0}; + if (!chunk.response) { + IMG_LOG_ERROR("Initial malloc failed for fetch_images_html"); + return NULL; + } + + IMG_LOG_INFO("Initializing cURL handle for URL: %s", url); + curl_handle = curl_easy_init(); + if (!curl_handle) { + IMG_LOG_ERROR("curl_easy_init() failed"); + free(chunk.response); + return NULL; + } + + curl_easy_setopt(curl_handle, CURLOPT_URL, url); + curl_easy_setopt(curl_handle, CURLOPT_WRITEFUNCTION, ImageWriteCallback); + curl_easy_setopt(curl_handle, CURLOPT_WRITEDATA, (void *)&chunk); + curl_easy_setopt( + curl_handle, CURLOPT_USERAGENT, + "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36"); + curl_easy_setopt(curl_handle, CURLOPT_FOLLOWLOCATION, 1L); + curl_easy_setopt(curl_handle, CURLOPT_TIMEOUT, 10L); + + CURLcode res = curl_easy_perform(curl_handle); + if (res != CURLE_OK) { + IMG_LOG_ERROR("curl_easy_perform() failed: %s", curl_easy_strerror(res)); + free(chunk.response); + curl_easy_cleanup(curl_handle); + return NULL; + } + + IMG_LOG_INFO("Successfully fetched %zu bytes from Yahoo Images", chunk.size); + curl_easy_cleanup(curl_handle); + return chunk.response; +} + +static char *get_json_field_internal(const char *json, const char *key) { + if (!json) return NULL; + char search_key[64]; + snprintf(search_key, sizeof(search_key), "\"%s\":\"", key); + char *start = strstr(json, search_key); + if (!start) return NULL; + start += strlen(search_key); + char *end = strchr(start, '\"'); + if (!end) return NULL; + + size_t len = end - start; + char *val = (char *)malloc(len + 1); + if (!val) return NULL; + + size_t j = 0; + for (size_t i = 0; i < len; i++) { + if (start[i] == '\\' && i + 1 < len && start[i + 1] == '/') { + val[j++] = '/'; + i++; + } else { + val[j++] = start[i]; + } + } + val[j] = '\0'; + return val; +} + +int images_handler(UrlParams *params) { + IMG_LOG_INFO("Start images_handler request processing"); + TemplateContext ctx = new_context(); + char *raw_query = ""; + + if (params) { + for (int i = 0; i < params->count; i++) { + if (strcmp(params->params[i].key, "q") == 0) { + raw_query = params->params[i].value; + break; + } + } + } + + char *encoded_query = strdup(raw_query); + + char *display_query = url_decode_query(raw_query); + context_set(&ctx, "query", display_query); + + if (!encoded_query || strlen(encoded_query) == 0) { + IMG_LOG_INFO("Empty search query received, returning early warning"); + send_response("

No query provided

"); + if (encoded_query) free(encoded_query); + if (display_query) free(display_query); + free_context(&ctx); + return -1; + } + + char url[1024]; + snprintf(url, sizeof(url), + "https://images.search.yahoo.com/search/images?p=%s", encoded_query); + + IMG_LOG_INFO("Requesting external HTML from Yahoo Images..."); + char *html = fetch_images_html(url); + if (!html) { + IMG_LOG_ERROR("Failed to fetch image search results from Yahoo"); + send_response("

Error fetching images

"); + free(encoded_query); + free(display_query); + free_context(&ctx); + return -1; + } + + IMG_LOG_INFO("Parsing HTML with libxml2..."); + htmlDocPtr doc = htmlReadMemory(html, (int)strlen(html), NULL, NULL, + HTML_PARSE_RECOVER | HTML_PARSE_NOERROR); + if (!doc) { + IMG_LOG_ERROR("htmlReadMemory failed to create document pointer"); + free(html); + free(encoded_query); + free(display_query); + free_context(&ctx); + return -1; + } + + xmlXPathContextPtr xpathCtx = xmlXPathNewContext(doc); + + if (!xpathCtx) { + IMG_LOG_ERROR("xmlXPathNewContext failed"); + xmlFreeDoc(doc); + free(html); + free(encoded_query); + free(display_query); + free_context(&ctx); + return -1; + } + + IMG_LOG_INFO("Executing XPath expression: //li[@data]"); + xmlXPathObjectPtr xpathObj = + xmlXPathEvalExpression((const xmlChar *)"//li[@data]", xpathCtx); + + int image_count = 0; + char ***image_matrix = NULL; + int *inner_counts = NULL; + + if (xpathObj && xpathObj->nodesetval) { + int nodes = xpathObj->nodesetval->nodeNr; + IMG_LOG_INFO("XPath found %d potential image nodes", nodes); + + int max_images = (nodes < 32) ? nodes : 32; + image_matrix = malloc(sizeof(char **) * max_images); + inner_counts = malloc(sizeof(int) * max_images); + + for (int i = 0; i < nodes; i++) { + if (image_count >= 32) break; + + xmlNodePtr node = xpathObj->nodesetval->nodeTab[i]; + xmlChar *data_attr = xmlGetProp(node, (const xmlChar *)"data"); + if (data_attr) { + char *iurl = get_json_field_internal((char *)data_attr, "iurl"); + char *title = get_json_field_internal((char *)data_attr, "alt"); + char *rurl = get_json_field_internal((char *)data_attr, "rurl"); + + if (iurl && strlen(iurl) > 0) { + image_matrix[image_count] = malloc(sizeof(char *) * 3); + image_matrix[image_count][0] = strdup(iurl); + image_matrix[image_count][1] = strdup(title ? title : "Image"); + image_matrix[image_count][2] = strdup(rurl ? rurl : "#"); + inner_counts[image_count] = 3; + image_count++; + } + + if (iurl) free(iurl); + if (title) free(title); + if (rurl) free(rurl); + xmlFree(data_attr); + } + } + IMG_LOG_INFO("Successfully parsed %d valid image results (capped at 32)", + image_count); + } else { + IMG_LOG_INFO("No image nodes found in the HTML document"); + } + + IMG_LOG_INFO("Setting image array in template context..."); + context_set_array_of_arrays(&ctx, "images", image_matrix, image_count, + inner_counts); + + IMG_LOG_INFO("Rendering images.html template..."); + char *rendered = render_template("images.html", &ctx); + if (rendered) { + IMG_LOG_INFO("Sending rendered template to client (%zu bytes)", + strlen(rendered)); + send_response(rendered); + free(rendered); + } else { + IMG_LOG_ERROR("render_template returned NULL for images.html"); + send_response("

Error rendering image results

"); + } + + IMG_LOG_INFO("Beginning memory cleanup..."); + + if (image_matrix) { + for (int i = 0; i < image_count; i++) { + for (int j = 0; j < 3; j++) { + free(image_matrix[i][j]); + } + free(image_matrix[i]); + } + free(image_matrix); + } + if (inner_counts) { + free(inner_counts); + } + + if (xpathObj) xmlXPathFreeObject(xpathObj); + if (xpathCtx) xmlXPathFreeContext(xpathCtx); + if (doc) xmlFreeDoc(doc); + free(html); + free(encoded_query); + free(display_query); + free_context(&ctx); + + IMG_LOG_INFO("Images request cycle complete"); + return 0; +} diff --git a/src/Routes/Images.h b/src/Routes/Images.h new file mode 100644 index 0000000..86f4a31 --- /dev/null +++ b/src/Routes/Images.h @@ -0,0 +1,8 @@ +#ifndef IMAGES_HANDLER_H +#define IMAGES_HANDLER_H + +#include + +int images_handler(UrlParams *params); + +#endif diff --git a/src/Routes/Search.c b/src/Routes/Search.c new file mode 100644 index 0000000..110e6f7 --- /dev/null +++ b/src/Routes/Search.c @@ -0,0 +1,273 @@ +#include "Search.h" +#include "../Infobox/Wikipedia.h" +#include "../Infobox/Calculator.h" +#include "../Scraping/Scraping.h" +#include "../Utility/Display.h" +#include "../Utility/Unescape.h" +#include +#include +#include +#include +#include +#include + +typedef struct { + const SearchEngine *engine; + const char *query; + SearchResult *results; + int count; +} EngineThreadData; + +static void *scrape_thread_func(void *arg) { + EngineThreadData *data = (EngineThreadData *)arg; + data->count = scrape_engine(data->engine, data->query, &data->results, 10); + return NULL; +} + +typedef struct { + const char *query; + InfoBox result; + int success; +} InfoBoxThreadData; + +static void *wiki_thread_func(void *arg) { + InfoBoxThreadData *data = (InfoBoxThreadData *)arg; + char *dynamic_url = construct_wiki_url(data->query); + if (dynamic_url) { + data->result = fetch_wiki_data(dynamic_url); + data->success = + (data->result.title != NULL && data->result.extract != NULL && + strlen(data->result.extract) > 10); + free(dynamic_url); + } else { + data->success = 0; + } + return NULL; +} + +static int is_calculator_query(const char *query) { + if (!query) return 0; + + int has_digit = 0; + int has_operator = 0; + + for (const char *p = query; *p; p++) { + if (isdigit(*p) || *p == '.') { + has_digit = 1; + } + if (*p == '+' || *p == '-' || *p == '*' || *p == '/' || *p == '=' || + *p == '^') { + has_operator = 1; + } + } + + return has_digit && (has_operator || strchr(query, '.')); +} + +static void *calc_thread_func(void *arg) { + InfoBoxThreadData *data = (InfoBoxThreadData *)arg; + + if (is_calculator_query(data->query)) { + data->result = fetch_calc_data((char *)data->query); + data->success = + (data->result.title != NULL && data->result.extract != NULL); + } else { + data->success = 0; + } + + 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 *) * 4); + (*collection)[current_count][0] = infobox->title; + (*collection)[current_count][1] = infobox->thumbnail_url; + (*collection)[current_count][2] = infobox->extract; + (*collection)[current_count][3] = infobox->url; + (*inner_counts)[current_count] = 4; + + return current_count + 1; +} + +int results_handler(UrlParams *params) { + TemplateContext ctx = new_context(); + char *raw_query = ""; + + if (params) { + for (int i = 0; i < params->count; i++) { + if (strcmp(params->params[i].key, "q") == 0) { + raw_query = params->params[i].value; + break; + } + } + } + + char *encoded_query = strdup(raw_query); + + char *display_query = url_decode_query(raw_query); + LOG_INFO("Processing search request for query: '%s'", display_query); + context_set(&ctx, "query", display_query); + + if (!encoded_query || strlen(encoded_query) == 0) { + LOG_ERROR("Empty search query provided."); + send_response("

No query provided

"); + if (encoded_query) free(encoded_query); + if (display_query) free(display_query); + free_context(&ctx); + return -1; + } + + pthread_t wiki_tid, calc_tid; + InfoBoxThreadData wiki_data = {.query = display_query, .success = 0}; + InfoBoxThreadData calc_data = {.query = display_query, .success = 0}; + + pthread_create(&wiki_tid, NULL, wiki_thread_func, &wiki_data); + pthread_create(&calc_tid, NULL, calc_thread_func, &calc_data); + + pthread_t engine_tids[ENGINE_COUNT]; + EngineThreadData engine_data[ENGINE_COUNT]; + + for (int i = 0; i < ENGINE_COUNT; i++) { + engine_data[i].engine = &ENGINE_REGISTRY[i]; + engine_data[i].query = encoded_query; + + engine_data[i].results = NULL; + engine_data[i].count = 0; + pthread_create(&engine_tids[i], NULL, scrape_thread_func, &engine_data[i]); + } + + pthread_join(wiki_tid, NULL); + pthread_join(calc_tid, NULL); + + char ***infobox_matrix = NULL; + int *infobox_inner_counts = NULL; + int infobox_count = 0; + + if (calc_data.success) { + infobox_count = + add_infobox_to_collection(&calc_data.result, &infobox_matrix, + &infobox_inner_counts, infobox_count); + } + + if (wiki_data.success) { + infobox_count = + add_infobox_to_collection(&wiki_data.result, &infobox_matrix, + &infobox_inner_counts, infobox_count); + } + + if (infobox_count > 0) { + context_set_array_of_arrays(&ctx, "infoboxes", infobox_matrix, + infobox_count, infobox_inner_counts); + free(infobox_matrix); + free(infobox_inner_counts); + } else { + context_set_array_of_arrays(&ctx, "infoboxes", NULL, 0, NULL); + } + + int total_results = 0; + for (int i = 0; i < ENGINE_COUNT; i++) { + pthread_join(engine_tids[i], NULL); + total_results += engine_data[i].count; + } + + if (total_results > 0) { + char ***results_matrix = (char ***)malloc(sizeof(char **) * total_results); + int *results_inner_counts = (int *)malloc(sizeof(int) * total_results); + char **seen_urls = (char **)malloc(sizeof(char *) * total_results); + int unique_count = 0; + + for (int i = 0; i < ENGINE_COUNT; i++) { + for (int j = 0; j < engine_data[i].count; j++) { + char *raw_url = engine_data[i].results[j].url; + char *clean_url = unescape_search_url(raw_url); + char *display_url = clean_url ? clean_url : raw_url; + + int is_duplicate = 0; + for (int k = 0; k < unique_count; k++) { + if (strcmp(seen_urls[k], display_url) == 0) { + is_duplicate = 1; + break; + } + } + + if (is_duplicate) { + if (clean_url) free(clean_url); + free(engine_data[i].results[j].url); + free(engine_data[i].results[j].title); + free(engine_data[i].results[j].snippet); + continue; + } + + seen_urls[unique_count] = strdup(display_url); + results_matrix[unique_count] = (char **)malloc(sizeof(char *) * 4); + char *pretty_url = pretty_display_url(display_url); + + results_matrix[unique_count][0] = strdup(display_url); + results_matrix[unique_count][1] = strdup(pretty_url); + results_matrix[unique_count][2] = + engine_data[i].results[j].title + ? strdup(engine_data[i].results[j].title) + : strdup("Untitled"); + results_matrix[unique_count][3] = + engine_data[i].results[j].snippet + ? strdup(engine_data[i].results[j].snippet) + : strdup(""); + + results_inner_counts[unique_count] = 4; + + free(pretty_url); + free(engine_data[i].results[j].url); + free(engine_data[i].results[j].title); + free(engine_data[i].results[j].snippet); + if (clean_url) free(clean_url); + + unique_count++; + } + free(engine_data[i].results); + } + + context_set_array_of_arrays(&ctx, "results", results_matrix, unique_count, + results_inner_counts); + + char *html = render_template("results.html", &ctx); + if (html) { + send_response(html); + free(html); + } + + for (int i = 0; i < unique_count; i++) { + for (int j = 0; j < 4; j++) free(results_matrix[i][j]); + free(results_matrix[i]); + free(seen_urls[i]); + } + free(seen_urls); + free(results_matrix); + free(results_inner_counts); + } else { + char *html = render_template("results.html", &ctx); + if (html) { + send_response(html); + free(html); + } + } + + if (wiki_data.success) { + free_infobox(&wiki_data.result); + } + + if (calc_data.success) { + free_infobox(&calc_data.result); + } + + free(encoded_query); + free(display_query); + free_context(&ctx); + + return 0; +} diff --git a/src/Routes/Search.h b/src/Routes/Search.h new file mode 100644 index 0000000..c6bc146 --- /dev/null +++ b/src/Routes/Search.h @@ -0,0 +1,8 @@ +#ifndef SEARCH_HANDLER_H +#define SEARCH_HANDLER_H + +#include + +int results_handler(UrlParams *params); + +#endif -- cgit v1.2.3