diff options
Diffstat (limited to 'src/Routes/Images.c')
| -rw-r--r-- | src/Routes/Images.c | 277 |
1 files changed, 277 insertions, 0 deletions
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 <curl/curl.h> +#include <libxml/HTMLparser.h> +#include <libxml/xpath.h> +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <time.h> + +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("<h1>No query provided</h1>"); + 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("<h1>Error fetching images</h1>"); + 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("<h1>Error rendering image results</h1>"); + } + + 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; +} |
