extractors: support haokan videos

4 files changed, 64 insertions, 4 deletions

diff --git a/src/extractors/extractor.c b/src/extractors/extractor.c
index 7f5e05c..ccc1ec6 100644
--- a/src/extractors/extractor.c
+++ b/src/extractors/extractor.c
@@ -1,10 +1,13 @@
 #include <stdlib.h>
 
+#include "../utils/utils.h"
 #include "bilibili.h"
 #include "extractor.h"
-#include "../utils/utils.h"
+#include "haokan.h"
 
-Site_map site_map = {{{"www.bilibili.com", SITE_BILIBILI}}, 1};
+Site_map site_map = {
+    {{"www.bilibili.com", SITE_BILIBILI}, {"haokan.baidu.com", SITE_HAOKAN}},
+    2};
 
 void options_cleanup(Options *options) {
   free_and_nullify(options->URL);
@@ -19,6 +22,9 @@ int extract(void *v) {
   case SITE_BILIBILI:
     bilibili_extract(options);
     break;
+  case SITE_HAOKAN:
+    haokan_extract(options);
+    break;
   }
   options_cleanup(options);
   return 0;
diff --git a/src/extractors/extractor.h b/src/extractors/extractor.h
index d3ebeec..347662b 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 };
+enum site { SITE_BILIBILI, SITE_HAOKAN };
 typedef enum site site_t;
 
 typedef struct site_map {
   struct {
     char domain[SHRT_MAX];
     site_t site;
-  } pairs[1];
+  } pairs[2];
   unsigned char size;
 } Site_map;
 
diff --git a/src/extractors/haokan.c b/src/extractors/haokan.c
new file mode 100644
index 0000000..d2d0379
--- /dev/null
+++ b/src/extractors/haokan.c
@@ -0,0 +1,46 @@
+#include "haokan.h"
+#include "../logger.h"
+#include "../process_url.h"
+#include <limits.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+void haokan_extract(Options *options) {
+  char *resp, *title, *videoURL;
+  get(options->URL, &resp);
+  const char *patterns_str[2] = {"<div class='ssr-video-title\'>(.*?)</div>",
+                                 "\"playurl\":\"(http.+?)\""};
+  const str_array_t patterns = {(char **)patterns_str, 2};
+  str_array_t results = create_str_array(0);
+  int r = regex_match(resp, patterns, &results);
+  // Should match exactly two results in HTML, otherwise error out.
+  if (!r && results.n == 2) {
+    // for (unsigned short i = 0; i < results.n; i++) {
+    //   DEBUG_PRINT("%s\n", results.str[i]);
+    // }
+    title = results.str[0];
+    substitute_str(results.str[1], "\\\\/", "/", &videoURL);
+    DEBUG_PRINT("title: %s\n", title);
+    DEBUG_PRINT("videoURL: %s\n", videoURL);
+
+    char *ct = NULL;
+    get_info(videoURL, NULL, &ct);
+    if (ct == NULL) {
+      goto end;
+    }
+    const char *ext = mimeType2ext(ct);
+    char *filename = malloc(strlen(title) + strlen(ct) + 2);
+    sprintf(filename, "%s.%s", title, ext);
+
+    add_url(videoURL, NULL, filename, "https://haokan.baidu.com", NULL, NULL);
+
+  end:
+    free_str_array(&results);
+    free_and_nullify(filename);
+    free_and_nullify(videoURL);
+    return;
+  }
+  LOG("haokan", "Download failed.\n");
+  free_str_array(&results);
+}
diff --git a/src/extractors/haokan.h b/src/extractors/haokan.h
new file mode 100644
index 0000000..f258a4d
--- /dev/null
+++ b/src/extractors/haokan.h
@@ -0,0 +1,8 @@
+#ifndef HAOKAN_H_
+#define HAOKAN_H_
+
+#include "extractor.h"
+
+void haokan_extract(struct options *);
+
+#endif