hooks migrate - cleaning up markdown after conversion

content-migration/hooks/migrate-hooks

@@ -7,35 +7,54 @@ $cache_dir = __DIR__ . '/wiki_cache';
 $root_dir = dirname(dirname(__DIR__));
 $hooks_dir = "$root_dir/docs/hooks";
 
+
+function clean_markdown($markdown) {
+  // add domain name to all hyperlinks that point to the wiki
+  $pattern = '@\]\((/confluence[^)]*)\)@';
+  $replace = '](https://wiki.civicrm.org$1)';
+  $markdown = preg_replace($pattern,$replace,$markdown);
+
+  // set all headings of level 3 and above to level 2
+  $pattern = '@^(##)(#+) (.*)$@m';
+  $replace = '## $3';
+  $markdown = preg_replace($pattern,$replace,$markdown);
+
+  return $markdown;
+}
+
+
+
 # create hooks directory if needed
 if ( !is_dir($hooks_dir) ) {
   mkdir($hooks_dir);
 }
-
 chdir($hooks_dir);
-
 foreach ($hooks_by_category as $category => $hooks) {
   foreach ($hooks as $hook) {
     $hook_name = $hook['name'];
-    $hook_file = "$hook_name.md";
+    $markdown_file = "$hook_name.md";
     $html = "$cache_dir/$hook_name";
-    if ( file_exists($html) && !file_exists($hook_file) ) {
+    if ( file_exists($html) && !file_exists($markdown_file) ) {
       echo "converting $hook_name ... ";
-      $conversion_status = 1;
-      system("webpage2md $html > $hook_file", $conversion_status);
-      // TODO fix links that point to wiki using only a slash
-      if( $conversion_status == 0 ) {
+      $conv_output_array = array();
+      $conv_status = 1;
+      exec("webpage2md $html", $conv_output_array, $conv_status);
+      if( $conv_status == 0 ) {
+        $conv_output = implode("\n",$conv_output_array);
+        $conv_output = clean_markdown($conv_output);
+        file_put_contents($markdown_file, $conv_output);
         echo "done" . PHP_EOL;
       }
       else {
         echo "ERROR CONVERTING $hook_name" . PHP_EOL;
+        echo "$conv_output";
       }
     }
     else if ( !file_exists($html) ) {
       echo "WARNING: $hook_name not yet fetched" . PHP_EOL;
     }
-    else if ( file_exists($hook_file) ) {
-      echo "ignoring $hook_name (hook file already exists)" . PHP_EOL;
+    else if ( file_exists($markdown_file) ) {
+      echo "ignoring $hook_name (markdown file already exists)" . PHP_EOL;
     }
   }
 }