113498266Sopenharmony_ci/***************************************************************************
213498266Sopenharmony_ci *                                  _   _ ____  _
313498266Sopenharmony_ci *  Project                     ___| | | |  _ \| |
413498266Sopenharmony_ci *                             / __| | | | |_) | |
513498266Sopenharmony_ci *                            | (__| |_| |  _ <| |___
613498266Sopenharmony_ci *                             \___|\___/|_| \_\_____|
713498266Sopenharmony_ci *
813498266Sopenharmony_ci * Copyright (C) Daniel Stenberg, <daniel@haxx.se>, et al.
913498266Sopenharmony_ci *
1013498266Sopenharmony_ci * This software is licensed as described in the file COPYING, which
1113498266Sopenharmony_ci * you should have received as part of this distribution. The terms
1213498266Sopenharmony_ci * are also available at https://curl.se/docs/copyright.html.
1313498266Sopenharmony_ci *
1413498266Sopenharmony_ci * You may opt to use, copy, modify, merge, publish, distribute and/or sell
1513498266Sopenharmony_ci * copies of the Software, and permit persons to whom the Software is
1613498266Sopenharmony_ci * furnished to do so, under the terms of the COPYING file.
1713498266Sopenharmony_ci *
1813498266Sopenharmony_ci * This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
1913498266Sopenharmony_ci * KIND, either express or implied.
2013498266Sopenharmony_ci *
2113498266Sopenharmony_ci * SPDX-License-Identifier: curl
2213498266Sopenharmony_ci *
2313498266Sopenharmony_ci ***************************************************************************/
2413498266Sopenharmony_ci/* <DESC>
2513498266Sopenharmony_ci * Stream-parse a document using the streaming Expat parser.
2613498266Sopenharmony_ci * </DESC>
2713498266Sopenharmony_ci */
2813498266Sopenharmony_ci/* Written by David Strauss
2913498266Sopenharmony_ci *
3013498266Sopenharmony_ci * Expat => https://libexpat.github.io/
3113498266Sopenharmony_ci *
3213498266Sopenharmony_ci * gcc -Wall -I/usr/local/include xmlstream.c -lcurl -lexpat -o xmlstream
3313498266Sopenharmony_ci *
3413498266Sopenharmony_ci */
3513498266Sopenharmony_ci
3613498266Sopenharmony_ci#include <stdio.h>
3713498266Sopenharmony_ci#include <stdlib.h>
3813498266Sopenharmony_ci#include <string.h>
3913498266Sopenharmony_ci#include <assert.h>
4013498266Sopenharmony_ci
4113498266Sopenharmony_ci#include <expat.h>
4213498266Sopenharmony_ci#include <curl/curl.h>
4313498266Sopenharmony_ci
4413498266Sopenharmony_cistruct MemoryStruct {
4513498266Sopenharmony_ci  char *memory;
4613498266Sopenharmony_ci  size_t size;
4713498266Sopenharmony_ci};
4813498266Sopenharmony_ci
4913498266Sopenharmony_cistruct ParserStruct {
5013498266Sopenharmony_ci  int ok;
5113498266Sopenharmony_ci  size_t tags;
5213498266Sopenharmony_ci  size_t depth;
5313498266Sopenharmony_ci  struct MemoryStruct characters;
5413498266Sopenharmony_ci};
5513498266Sopenharmony_ci
5613498266Sopenharmony_cistatic void startElement(void *userData, const XML_Char *name,
5713498266Sopenharmony_ci                         const XML_Char **atts)
5813498266Sopenharmony_ci{
5913498266Sopenharmony_ci  struct ParserStruct *state = (struct ParserStruct *) userData;
6013498266Sopenharmony_ci  state->tags++;
6113498266Sopenharmony_ci  state->depth++;
6213498266Sopenharmony_ci
6313498266Sopenharmony_ci  /* Get a clean slate for reading in character data. */
6413498266Sopenharmony_ci  free(state->characters.memory);
6513498266Sopenharmony_ci  state->characters.memory = NULL;
6613498266Sopenharmony_ci  state->characters.size = 0;
6713498266Sopenharmony_ci}
6813498266Sopenharmony_ci
6913498266Sopenharmony_cistatic void characterDataHandler(void *userData, const XML_Char *s, int len)
7013498266Sopenharmony_ci{
7113498266Sopenharmony_ci  struct ParserStruct *state = (struct ParserStruct *) userData;
7213498266Sopenharmony_ci  struct MemoryStruct *mem = &state->characters;
7313498266Sopenharmony_ci
7413498266Sopenharmony_ci  char *ptr = realloc(mem->memory, mem->size + len + 1);
7513498266Sopenharmony_ci  if(!ptr) {
7613498266Sopenharmony_ci    /* Out of memory. */
7713498266Sopenharmony_ci    fprintf(stderr, "Not enough memory (realloc returned NULL).\n");
7813498266Sopenharmony_ci    state->ok = 0;
7913498266Sopenharmony_ci    return;
8013498266Sopenharmony_ci  }
8113498266Sopenharmony_ci
8213498266Sopenharmony_ci  mem->memory = ptr;
8313498266Sopenharmony_ci  memcpy(&(mem->memory[mem->size]), s, len);
8413498266Sopenharmony_ci  mem->size += len;
8513498266Sopenharmony_ci  mem->memory[mem->size] = 0;
8613498266Sopenharmony_ci}
8713498266Sopenharmony_ci
8813498266Sopenharmony_cistatic void endElement(void *userData, const XML_Char *name)
8913498266Sopenharmony_ci{
9013498266Sopenharmony_ci  struct ParserStruct *state = (struct ParserStruct *) userData;
9113498266Sopenharmony_ci  state->depth--;
9213498266Sopenharmony_ci
9313498266Sopenharmony_ci  printf("%5lu   %10lu   %s\n", state->depth, state->characters.size, name);
9413498266Sopenharmony_ci}
9513498266Sopenharmony_ci
9613498266Sopenharmony_cistatic size_t parseStreamCallback(void *contents, size_t length, size_t nmemb,
9713498266Sopenharmony_ci                                  void *userp)
9813498266Sopenharmony_ci{
9913498266Sopenharmony_ci  XML_Parser parser = (XML_Parser) userp;
10013498266Sopenharmony_ci  size_t real_size = length * nmemb;
10113498266Sopenharmony_ci  struct ParserStruct *state = (struct ParserStruct *) XML_GetUserData(parser);
10213498266Sopenharmony_ci
10313498266Sopenharmony_ci  /* Only parse if we are not already in a failure state. */
10413498266Sopenharmony_ci  if(state->ok && XML_Parse(parser, contents, real_size, 0) == 0) {
10513498266Sopenharmony_ci    int error_code = XML_GetErrorCode(parser);
10613498266Sopenharmony_ci    fprintf(stderr, "Parsing response buffer of length %lu failed"
10713498266Sopenharmony_ci            " with error code %d (%s).\n",
10813498266Sopenharmony_ci            real_size, error_code, XML_ErrorString(error_code));
10913498266Sopenharmony_ci    state->ok = 0;
11013498266Sopenharmony_ci  }
11113498266Sopenharmony_ci
11213498266Sopenharmony_ci  return real_size;
11313498266Sopenharmony_ci}
11413498266Sopenharmony_ci
11513498266Sopenharmony_ciint main(void)
11613498266Sopenharmony_ci{
11713498266Sopenharmony_ci  CURL *curl_handle;
11813498266Sopenharmony_ci  CURLcode res;
11913498266Sopenharmony_ci  XML_Parser parser;
12013498266Sopenharmony_ci  struct ParserStruct state;
12113498266Sopenharmony_ci
12213498266Sopenharmony_ci  /* Initialize the state structure for parsing. */
12313498266Sopenharmony_ci  memset(&state, 0, sizeof(struct ParserStruct));
12413498266Sopenharmony_ci  state.ok = 1;
12513498266Sopenharmony_ci
12613498266Sopenharmony_ci  /* Initialize a namespace-aware parser. */
12713498266Sopenharmony_ci  parser = XML_ParserCreateNS(NULL, '\0');
12813498266Sopenharmony_ci  XML_SetUserData(parser, &state);
12913498266Sopenharmony_ci  XML_SetElementHandler(parser, startElement, endElement);
13013498266Sopenharmony_ci  XML_SetCharacterDataHandler(parser, characterDataHandler);
13113498266Sopenharmony_ci
13213498266Sopenharmony_ci  /* Initialize a libcurl handle. */
13313498266Sopenharmony_ci  curl_global_init(CURL_GLOBAL_DEFAULT);
13413498266Sopenharmony_ci  curl_handle = curl_easy_init();
13513498266Sopenharmony_ci  curl_easy_setopt(curl_handle, CURLOPT_URL,
13613498266Sopenharmony_ci                   "https://www.w3schools.com/xml/simple.xml");
13713498266Sopenharmony_ci  curl_easy_setopt(curl_handle, CURLOPT_WRITEFUNCTION, parseStreamCallback);
13813498266Sopenharmony_ci  curl_easy_setopt(curl_handle, CURLOPT_WRITEDATA, (void *)parser);
13913498266Sopenharmony_ci
14013498266Sopenharmony_ci  printf("Depth   Characters   Closing Tag\n");
14113498266Sopenharmony_ci
14213498266Sopenharmony_ci  /* Perform the request and any follow-up parsing. */
14313498266Sopenharmony_ci  res = curl_easy_perform(curl_handle);
14413498266Sopenharmony_ci  if(res != CURLE_OK) {
14513498266Sopenharmony_ci    fprintf(stderr, "curl_easy_perform() failed: %s\n",
14613498266Sopenharmony_ci            curl_easy_strerror(res));
14713498266Sopenharmony_ci  }
14813498266Sopenharmony_ci  else if(state.ok) {
14913498266Sopenharmony_ci    /* Expat requires one final call to finalize parsing. */
15013498266Sopenharmony_ci    if(XML_Parse(parser, NULL, 0, 1) == 0) {
15113498266Sopenharmony_ci      int error_code = XML_GetErrorCode(parser);
15213498266Sopenharmony_ci      fprintf(stderr, "Finalizing parsing failed with error code %d (%s).\n",
15313498266Sopenharmony_ci              error_code, XML_ErrorString(error_code));
15413498266Sopenharmony_ci    }
15513498266Sopenharmony_ci    else {
15613498266Sopenharmony_ci      printf("                     --------------\n");
15713498266Sopenharmony_ci      printf("                     %lu tags total\n", state.tags);
15813498266Sopenharmony_ci    }
15913498266Sopenharmony_ci  }
16013498266Sopenharmony_ci
16113498266Sopenharmony_ci  /* Clean up. */
16213498266Sopenharmony_ci  free(state.characters.memory);
16313498266Sopenharmony_ci  XML_ParserFree(parser);
16413498266Sopenharmony_ci  curl_easy_cleanup(curl_handle);
16513498266Sopenharmony_ci  curl_global_cleanup();
16613498266Sopenharmony_ci
16713498266Sopenharmony_ci  return 0;
16813498266Sopenharmony_ci}
169