extractors: initial youku support

4 files changed, 218 insertions, 6 deletions

diff --git a/src/extractors/extractor.c b/src/extractors/extractor.c
index 4bfc7cb..833686e 100644
--- a/src/extractors/extractor.c
+++ b/src/extractors/extractor.c
@@ -6,9 +6,10 @@
 #include "haokan.h"
 #include "youku.h"
 
-Site_map site_map = {
-    {{"www.bilibili.com", SITE_BILIBILI}, {"haokan.baidu.com", SITE_HAOKAN}},
-    2};
+Site_map site_map = {{{"www.bilibili.com", SITE_BILIBILI},
+                      {"haokan.baidu.com", SITE_HAOKAN},
+                      {"v.youku.com", SITE_YOUKU}},
+                     3};
 
 void options_cleanup(Options *options) {
   free_and_nullify((void **)&options->URL);
@@ -26,6 +27,9 @@ int extract(void *v) {
   case SITE_HAOKAN:
     haokan_extract(options);
     break;
+  case SITE_YOUKU:
+    youku_extract(options);
+    break;
   }
   options_cleanup(options);
   return 0;
diff --git a/src/extractors/extractor.h b/src/extractors/extractor.h
index 347662b..9daf0ff 100644
--- a/src/extractors/extractor.h
+++ b/src/extractors/extractor.h
@@ -6,14 +6,14 @@
 #include <stdbool.h>
 #include <stddef.h>
 
-enum site { SITE_BILIBILI, SITE_HAOKAN };
+enum site { SITE_BILIBILI, SITE_HAOKAN, SITE_YOUKU };
 typedef enum site site_t;
 
 typedef struct site_map {
   struct {
     char domain[SHRT_MAX];
     site_t site;
-  } pairs[2];
+  } pairs[3];
   unsigned char size;
 } Site_map;
 
@@ -25,7 +25,7 @@ typedef struct options {
   char *pagedata;
 } Options;
 
-void options_cleanup(Options*);
+void options_cleanup(Options *);
 
 int extract(void *);
 
diff --git a/src/extractors/youku.c b/src/extractors/youku.c
new file mode 100644
index 0000000..4e5bc47
--- /dev/null
+++ b/src/extractors/youku.c
@@ -0,0 +1,179 @@
+#include "youku.h"
+#include "../logger.h"
+#include "../process_url.h"
+#include "../utils/ffmpeg.h"
+#include "../utils/utils.h"
+#include <cjson/cJSON.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+static callback_struct_t callback_struct;
+
+static int generate_api(Youku_options *youku_options) {
+  char params[UCHAR_MAX];
+  snprintf(params, sizeof(params),
+           "vid=%s&ccode=%s&client_ip=192.168.1.1&client_ts=%d&utid=%s",
+           youku_options->vid, youku_options->ccode,
+           (int)(youku_options->client_ts / 1000), youku_options->utid);
+  youku_options->api = malloc(strlen(YOUKU_API) + strlen(params) + 1);
+  strcpy(youku_options->api, YOUKU_API);
+  strcat(youku_options->api, params);
+  return 0;
+}
+
+static int parse_jsondata(cJSON *json, Youku_data *youku_data) {
+  cJSON *data_obj = cJSON_GetObjectItem(json, "data");
+  if (!data_obj) {
+    LOG("cJSON", "Parse API data failed.\n");
+    return 1;
+  }
+
+  cJSON *video_obj = cJSON_GetObjectItem(data_obj, "video");
+  cJSON *streams_obj = cJSON_GetObjectItem(data_obj, "stream");
+  if (!video_obj || !streams_obj) {
+    LOG("cJSON", "Parse API data child object failed.\n");
+    return 1;
+  }
+  cJSON *title_obj = cJSON_GetObjectItem(video_obj, "title");
+  if (!title_obj) {
+    LOG("cJSON", "Parse API data.video.title failed.\n");
+    return 1;
+  }
+  youku_data->title = title_obj->valuestring;
+  DEBUG_PRINT("data.video.title: %s\n", youku_data->title);
+
+  cJSON *e;
+  int max_size = 0;
+  cJSON_ArrayForEach(e, streams_obj) {
+    cJSON *size_obj = cJSON_GetObjectItem(e, "size");
+    cJSON *width_obj = cJSON_GetObjectItem(e, "width");
+    cJSON *height_obj = cJSON_GetObjectItem(e, "height");
+    cJSON *m3u8_url_obj = cJSON_GetObjectItem(e, "m3u8_url");
+    if (!size_obj || !width_obj || !height_obj || !m3u8_url_obj) {
+      LOG("cJSON", "Parse API data.stream child object failed.\n");
+      return 1;
+    }
+    if (size_obj->valueint > max_size) {
+      max_size = size_obj->valueint;
+      youku_data->width = width_obj->valueint;
+      youku_data->height = height_obj->valueint;
+      youku_data->m3u8_url = m3u8_url_obj->valuestring;
+    }
+  }
+  DEBUG_PRINT("width: %d, height: %d\n", youku_data->width, youku_data->height);
+
+  return 0;
+}
+
+static void youku_options_cleanup(Youku_options *youku_options) {
+  free_and_nullify((void **)&youku_options->vid);
+  free_and_nullify((void **)&youku_options->utid);
+  free_and_nullify((void **)&youku_options->api);
+  cJSON_Delete(youku_options->json);
+}
+
+static int youku_convert(void *v) {
+  callback_struct_t *callback_struct = (callback_struct_t *)v;
+  const char *m3u8fn = get_str_element(callback_struct, 0);
+  const char *filename = get_str_element(callback_struct, 1);
+  DEBUG_PRINT("m3u8fn: %s\n", m3u8fn);
+  DEBUG_PRINT("filename: %s\n", filename);
+
+  int r = remux(m3u8fn, filename);
+  LOG("Youku", "All done!\n");
+  free_str_array(callback_struct);
+  return r;
+}
+
+void youku_extract(Options *options) {
+  Youku_options youku_options = {0};
+  str_array_t results = create_str_array(0);
+  {
+    const char *patterns_str[2] = {"id_(.+?)\\.html", "id_(.+)"};
+    const str_array_t patterns = {(char **)patterns_str, 2};
+    int r = regex_match(options->URL, patterns, &results);
+    if (r) {
+      LOG("Youku", "Parse URL failed.\n");
+      goto end;
+    }
+    // for (unsigned short i = 0; i < results.n; i++) {
+    //   DEBUG_PRINT("%s\n", results.str[i]);
+    // }
+    if (!results.str[0]) {
+      LOG("Youku", "Get vid failed.\n");
+      goto end;
+    }
+    youku_options.vid = malloc(strlen(results.str[0]) + 1);
+    strcpy(youku_options.vid, results.str[0]);
+    DEBUG_PRINT("vid: %s\n", youku_options.vid);
+  }
+
+  {
+    resize_str_array(&results, 0);
+    char *cookie = NULL;
+    set_referer("https://v.youku.com");
+    get_info("http://log.mmstat.com/eg.js", NULL, NULL, &cookie);
+    if (!cookie) {
+      goto end;
+    }
+    const char *patterns_str[1] = {"cna=(.+?);"};
+    const str_array_t patterns = {(char **)patterns_str, 1};
+    int r = regex_match(cookie, patterns, &results);
+    if (r) {
+      LOG("Youku", "Parse cookie failed.\n");
+      goto end;
+    }
+    // for (unsigned short i = 0; i < results.n; i++) {
+    //   DEBUG_PRINT("%s\n", results.str[i]);
+    // }
+    if (!results.str[0]) {
+      LOG("Youku", "Get utid failed.\n");
+      goto end;
+    }
+    youku_options.utid = malloc(strlen(results.str[0]) + 1);
+    strcpy(youku_options.utid, results.str[0]);
+    DEBUG_PRINT("utid: %s\n", youku_options.utid);
+  }
+
+  youku_options.ccode = "0524";
+  youku_options.client_ip = "192.168.1.1";
+  youku_options.client_ts = time(NULL);
+  if (generate_api(&youku_options)) {
+    LOG("Youku", "Generate API failed.\n");
+    goto end;
+  }
+  DEBUG_PRINT("Generated API: %s\n", youku_options.api);
+  char *resp = NULL;
+  if (get(youku_options.api, &resp)) {
+    LOG("Youku", "Get API data failed.\n");
+    free_and_nullify((void **)&resp);
+    goto end;
+  }
+  // DEBUG_PRINT("resp: %s\n", resp);
+
+  youku_options.json = cJSON_Parse(resp);
+  free_and_nullify((void **)&resp);
+  Youku_data youku_data = {0};
+  if (parse_jsondata(youku_options.json, &youku_data)) {
+    LOG("Youku", "Parse jsondata failed.\n");
+    goto end;
+  }
+
+  callback_struct = create_str_array(2);
+  char *m3u8fn = malloc(strlen(youku_data.title) + 18);
+  sprintf(m3u8fn, "%s-[%dx%d].m3u8", youku_data.title, youku_data.width,
+          youku_data.height);
+  set_str_element(&callback_struct, 0, m3u8fn);
+  char *filename = malloc(strlen(youku_data.title) + 17);
+  sprintf(filename, "%s-[%dx%d].mp4", youku_data.title, youku_data.width,
+          youku_data.height);
+  set_str_element(&callback_struct, 1, filename);
+  add_url(youku_data.m3u8_url, NULL, m3u8fn, youku_convert, &callback_struct);
+  free_and_nullify((void **)&m3u8fn);
+  free_and_nullify((void **)&filename);
+
+end:
+  free_str_array(&results);
+  youku_options_cleanup(&youku_options);
+}
diff --git a/src/extractors/youku.h b/src/extractors/youku.h
new file mode 100644
index 0000000..721ff75
--- /dev/null
+++ b/src/extractors/youku.h
@@ -0,0 +1,29 @@
+#ifndef YOUKU_H_
+#define YOUKU_H_
+
+#include <time.h>
+
+#include "extractor.h"
+
+#define YOUKU_API "https://ups.youku.com/ups/get.json?"
+
+typedef struct youku_options {
+    char *api;
+    char *vid;
+    char *utid;
+    char *ccode;
+    const char *client_ip;
+    time_t client_ts;
+    cJSON *json;
+} Youku_options;
+
+typedef struct youku_data {
+    char *title;
+    int width;
+    int height;
+    char *m3u8_url;
+} Youku_data;
+
+void youku_extract(Options *options);
+
+#endif