refactor: put HTTP and XML logic into reusable modules

src/Utility/HttpClient.c

@@ -0,0 +1,81 @@
+#include "HttpClient.h"
+#include "../Proxy/Proxy.h"
+#include <stdlib.h>
+#include <string.h>
+
+static size_t write_callback(void *contents, size_t size, size_t nmemb,
+                             void *userp) {
+  size_t realsize = size * nmemb;
+  HttpResponse *mem = (HttpResponse *)userp;
+
+  if (mem->size + realsize + 1 > mem->capacity) {
+    size_t new_cap = mem->capacity == 0 ? 16384 : mem->capacity * 2;
+    while (new_cap < mem->size + realsize + 1)
+      new_cap *= 2;
+
+    char *ptr = realloc(mem->memory, new_cap);
+    if (!ptr) {
+      return 0;
+    }
+    mem->memory = ptr;
+    mem->capacity = new_cap;
+  }
+
+  memcpy(&(mem->memory[mem->size]), contents, realsize);
+  mem->size += realsize;
+  mem->memory[mem->size] = 0;
+
+  return realsize;
+}
+
+HttpResponse http_get(const char *url, const char *user_agent) {
+  HttpResponse resp = {.memory = NULL, .size = 0, .capacity = 0};
+
+  if (!url) {
+    return resp;
+  }
+
+  resp.memory = malloc(16384);
+  if (!resp.memory) {
+    return resp;
+  }
+  resp.capacity = 16384;
+
+  CURL *curl = curl_easy_init();
+  if (!curl) {
+    free(resp.memory);
+    resp.memory = NULL;
+    return resp;
+  }
+
+  curl_easy_setopt(curl, CURLOPT_URL, url);
+  curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, write_callback);
+  curl_easy_setopt(curl, CURLOPT_WRITEDATA, &resp);
+  curl_easy_setopt(curl, CURLOPT_USERAGENT,
+                   user_agent ? user_agent : "libcurl-agent/1.0");
+  curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 1L);
+  curl_easy_setopt(curl, CURLOPT_TIMEOUT, 15L);
+  apply_proxy_settings(curl);
+
+  CURLcode res = curl_easy_perform(curl);
+  curl_easy_cleanup(curl);
+
+  if (res != CURLE_OK) {
+    free(resp.memory);
+    resp.memory = NULL;
+    resp.size = 0;
+    resp.capacity = 0;
+  }
+
+  return resp;
+}
+
+void http_response_free(HttpResponse *resp) {
+  if (!resp) {
+    return;
+  }
+  free(resp->memory);
+  resp->memory = NULL;
+  resp->size = 0;
+  resp->capacity = 0;
+}

src/Utility/HttpClient.h

@@ -0,0 +1,16 @@
+#ifndef HTTPCLIENT_H
+#define HTTPCLIENT_H
+
+#include <curl/curl.h>
+#include <stddef.h>
+
+typedef struct {
+  char *memory;
+  size_t size;
+  size_t capacity;
+} HttpResponse;
+
+HttpResponse http_get(const char *url, const char *user_agent);
+void http_response_free(HttpResponse *resp);
+
+#endif

src/Utility/XmlHelper.c

@@ -0,0 +1,65 @@
+#include "XmlHelper.h"
+#include <stdlib.h>
+#include <string.h>
+
+SearchResult *xml_result_alloc(int count, int max_results) {
+  if (count <= 0 || max_results <= 0) {
+    return NULL;
+  }
+  int actual = (count < max_results) ? count : max_results;
+  return (SearchResult *)calloc(actual, sizeof(SearchResult));
+}
+
+void xml_result_free(SearchResult *results, int count) {
+  if (!results) {
+    return;
+  }
+  for (int i = 0; i < count; i++) {
+    free(results[i].url);
+    free(results[i].title);
+    free(results[i].snippet);
+  }
+  free(results);
+}
+
+xmlXPathObjectPtr xml_xpath_eval(xmlXPathContextPtr ctx, const char *xpath) {
+  if (!ctx || !xpath) {
+    return NULL;
+  }
+  return xmlXPathEvalExpression((const xmlChar *)xpath, ctx);
+}
+
+char *xml_node_content(xmlNodePtr node) {
+  if (!node) {
+    return NULL;
+  }
+  char *content = (char *)xmlNodeGetContent(node);
+  return content;
+}
+
+char *xpath_text(xmlDocPtr doc, const char *xpath) {
+  if (!doc || !xpath) {
+    return NULL;
+  }
+
+  xmlXPathContextPtr ctx = xmlXPathNewContext(doc);
+  if (!ctx) {
+    return NULL;
+  }
+
+  xmlXPathObjectPtr obj = xmlXPathEvalExpression((const xmlChar *)xpath, ctx);
+  xmlXPathFreeContext(ctx);
+
+  if (!obj || !obj->nodesetval || obj->nodesetval->nodeNr == 0) {
+    if (obj)
+      xmlXPathFreeObject(obj);
+    return NULL;
+  }
+
+  xmlChar *content = xmlNodeGetContent(obj->nodesetval->nodeTab[0]);
+  char *result = content ? strdup((char *)content) : NULL;
+  if (content)
+    xmlFree(content);
+  xmlXPathFreeObject(obj);
+  return result;
+}

src/Utility/XmlHelper.h

@@ -0,0 +1,14 @@
+#ifndef XMLHELPER_H
+#define XMLHELPER_H
+
+#include "../Scraping/Scraping.h"
+#include <libxml/xpath.h>
+
+SearchResult *xml_result_alloc(int count, int max_results);
+void xml_result_free(SearchResult *results, int count);
+
+xmlXPathObjectPtr xml_xpath_eval(xmlXPathContextPtr ctx, const char *xpath);
+char *xml_node_content(xmlNodePtr node);
+char *xpath_text(xmlDocPtr doc, const char *xpath);
+
+#endif