Changeset 723 for team

Timestamp:

12/09/07 16:22:53 (13 months ago)

Author:

michiel

Message:

import now copies wp-upload user uploaded stuff and replaces the links to them in the content.
Re #37

Location:

team/michiel/import_blogs/common

Files:

• mvblog_common.php (modified) (1 diff)
• mvblog_import_wordpress.php (modified) (4 diffs)

team/michiel/import_blogs/common/mvblog_common.php

@@ -806 +806 @@
         }
         /* }}} */
-
+        /* filesystem methods */
+        /* copy_files {{{ */
+        /**
+         * Copies files and directories recursive to dirs under site_images
+         *
+         * @param string $src The sourcefile or dir
+         * @param string $dst The dir inside site_images where the files/folders should be copied to
+         * @param int $overwrite_dest if set will overwrite the destination if it already exists
+         */
+        public function copy_files($srcbase, $src, $dest, $overwrite_dest = 0) {
+                // calculate where the site_images dir is
+                $basepath = realpath(dirname(__FILE__)."/../site_images/");
+                $dst = sprintf("%s/%s", $basepath, $dest);
+                //check if the dst dir is there
+                if (!is_dir($dst))
+                        if (!mkdir($dst))
+                                die("could not create directory $dst");
+                if ($handle = opendir(sprintf("%s/%s", $srcbase, $src))) { //open source directory
+                        while (false !== ($file = readdir($handle))) { //loop through all items in the opened source dir
+                                if ($file != "." && $file != "..") { //we wont copy the special 'current dir' and 'one dir up' items
+                                        if ($src)
+                                                $path = sprintf("%s/%s", $src, $file);
+                                        else
+                                                $path = $file;
+                                                $s = sprintf("%s/%s", $srcbase, $path);
+                                                $d = sprintf("%s/%s", $dst, $path);
+                                        if (is_file($s)) {
+                                                if (is_file($d) || $overwrite_dest) {
+                                                        if (!copy($s, $d))
+                                                                echo "Something went wrong while trying to copy $s to $d<br>";
+                                                }
+                                        } elseif (is_dir($s)) {
+                                                if (!is_dir($d))
+                                                        if (!mkdir($d))
+                                                                echo "Something went wrong while trying to create dir $d<br>\n";
+                                                $this->copy_files($srcbase, $path, $dest, 1);
+                                        }
+                                }
+                        }
+                        closedir($handle);
+                }
+        }
+        /* }}} */
         /* output methods */
         /* html_header: {{{ */

team/michiel/import_blogs/common/mvblog_import_wordpress.php

@@ -56 +56 @@
                 echo "</tr><tr>\n";
                 echo "<td>WordPress Table prefix:</td><td><input type=\"text\" style=\"width: 200px;\" name=\"dbprefix\" /></td>\n";
+                echo "</tr><tr>\n";
+                echo "<td>WordPress Base URL:</td><td><input type=\"text\" style=\"width: 200px;\" name=\"orig_url\" /></td>\n";
+                echo "</tr><tr>\n";
+                echo "<td>WordPress Filesystem Location:</td><td><input type=\"text\" style=\"width: 200px;\" name=\"fs_location\" /></td>\n";
                 echo "</tr></table>\n";
                 echo "<input type=\"submit\" value=\"".gettext("import")."\" />\n";
                 echo "</form>\n";
         }
+
         public function run_import($data) {
+                // first try to copy all files
+                $this->copy_files($data["fs_location"]."wp-content/uploads", "", "wp-uploads", 1);
                 $category = array();
                 // find out if we have enough info for a database connection
@@ -72 +79 @@
                 $db = mysql_connect($data["dbserver"], $data["dbuser"], $data["dbpass"]);
                 mysql_select_db($data["dbname"], $db);
+                //grab WP db version as we only support 2.3 and newer
+                $sql = sprintf("SELECT option_value FROM %soptions WHERE option_name='db_version'", $data["dbprefix"]);
+                $res = mysql_query($sql);
+                $wpdbversion = mysql_result($res, 0);
+                if ($wpdbversion < 5539)
+                        die("You need at least WP 2.3.0");
                 //import categories and maintain array with oldid=>newid so we can link the articles to the correct ids
-                $sql = sprintf("SELECT cat_ID, cat_name, category_description FROM %swp_categories", $data["dbprefix"]);
-                $res = mysql_query($sql);
+                /*
+                 * WP categories are stored in the table <prefix>terms
+                 * The table <prefix>term_taxonomy holds a count for them
+                 * The table <prefix>term_relationships links articles and term_taxonomy records together
+                 * <posts> -> <term_relationships> -> <term_taxonomy> -> <terms>
+                 */
+                $sql = sprintf("SELECT %1\$sterms.term_id, %1\$sterms.name, %1\$sterm_taxonomy.term_taxonomy_id from %1\$sterms LEFT JOIN %1\$sterm_taxonomy ON %1\$sterm_taxonomy.term_id = %1\$sterms.term_id", $data["dbprefix"]);
+                $res = mysql_query($sql) or die(mysql_error());
                 while ($row = mysql_fetch_assoc($res)) {
-                        $cat_sql = sprintf("INSERT INTO categories (name, `desc`, public, active) VALUES ('%s', '%s', 1, 1)",
-                                $row["cat_name"], $row["category_description"]);
+                        $cat_sql = sprintf("INSERT INTO categories (name, `desc`, public, active) VALUES ('%1\$s', '%1\$s', 1, 1)",
+                                $row["name"]);
                         $r = $GLOBALS["admin"]->db->query($cat_sql);
-                        $category[$row["cat_ID"]] = $GLOBALS["admin"]->db->lastInsertID("categories", "id");
+                        $category[$row["term_taxonomy_id"]] = $GLOBALS["admin"]->db->lastInsertID("categories", "id");
                 }
                 //select all articles
-                $sql = sprintf("SELECT * FROM %swp_posts WHERE post_type='post'", $data["dbprefix"]);
+                $sql = sprintf("SELECT * FROM %sposts WHERE post_type='post'", $data["dbprefix"]);
                 $res = mysql_query($sql);
                 while ($row = mysql_fetch_assoc($res)) {
@@ -98 +117 @@
                                 $last_modified = 0;
                         //get categories
-                        $q = sprintf("SELECT category_id FROM %swp_post2cat WHERE post_id = %d", $data["dbprefix"], $row["ID"]);
+                        $q = sprintf("SELECT term_taxonomy_id FROM %sterm_relationships WHERE object_id=%d", $data["dbprefix"], $row["ID"]);
                         $r = mysql_query($q);
                         $categories = array();
                         while ($catr = mysql_fetch_assoc($r)) {
-                                $categories[] = $category[$catr["category_id"]];
+                                $categories[] = $category[$catr["term_taxonomy_id"]];
                         }
-                        $data["oldsite"] = "http://www.electricmonk.nl/log/";
+                        $data["oldsite"] = $data["orig_url"];
                         if (array_key_exists("oldsite", $data) && $data["oldsite"]) {
-                                //find all locally hosted images and put them into mvblog space
+                                $wpuploadurl = sprintf("%s%s", $data["oldsite"], "wp-content/uploads/");
                                 //find all links pointing to internal links and replace them with the new structure
-                                preg_match_all("/<[(img)|a][^>]*[.]*>/si", $row["post_content"], $matches);
+                                preg_match_all("/<[img|a][^>]*[.]*>/si", $row["post_content"], $matches);
                                 foreach ($matches[0] as $match) {
-                                        //get the src or href attribute
+                                        //get the href attribute
+                                        preg_match_all("/href=[\"|\'](.+?)[\"|\']/si", $match, $m);
+                                        $filtermatch = $m[1];
+                                        foreach ($filtermatch as $v) {
+                                                if (strstr($v, $data["oldsite"]) && strstr($v, $wpuploadurl)) {
+                                                        $row["post_content"] = str_replace($v, "site_images/wp-uploads/".str_replace($wpuploadurl, "", $v), $row["post_content"]);
+                                                }
+                                        }
+                                        unset($m);
+                                        //get the src attribute
                                         preg_match_all("/src=[\"|\'](.+?)[\"|\']/si", $match, $m);
-                                        print_r($match);
-                                        print_r($m);
-                                        preg_match_all("/href=[\"|\'](.+?)[\"|\']/si", $match, $m);
-                                        print_r($m);
-                                        echo "======================\n";
+                                        $filtermatch = $m[1];
+                                        foreach ($filtermatch as $v) {
+                                                if (strstr($v, $data["oldsite"]) && strstr($v, $wpuploadurl)) {
+                                                        $row["post_content"] = str_replace($v, "site_images/wp-uploads/".str_replace($wpuploadurl, "", $v), $row["post_content"]);
+                                                }
+                                        }
                                 }
                         }
@@ -127 +156 @@
                         }
                 }
+                echo "All done !";
         }
 }