From 01f6d85371b09aad249a02c53461b1f066a76955 Mon Sep 17 00:00:00 2001
From: Timothy Chavez <timothy.chavez@hp.com>
Date: Wed, 19 Nov 2014 12:42:42 -0600
Subject: [PATCH] Restore pad to new location at a given revision

This script gives an admin with shell access the ability to restore a
pad at a given revision by essentially rebuilding it at a new location
with data associated with the original pad.  The upsides to creating a
new pad vs. changing the original are: 1) avoiding service disruptions
(no deletes, no moving targets - builds from previous revision); and 2)
preservation of data (no deletes, no overwriting of the source pad).
The most obvious downside is the pad has a new ID which could require
folks to update their links, bookmarks, etc. to point at the new
location.
---
 bin/rebuildPad.js | 131 ++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 131 insertions(+)
 create mode 100644 bin/rebuildPad.js

diff --git a/bin/rebuildPad.js b/bin/rebuildPad.js
new file mode 100644
index 00000000..b5d9b7e3
--- /dev/null
+++ b/bin/rebuildPad.js
@@ -0,0 +1,131 @@
+/*
+  This is a repair tool. It rebuilds an old pad at a new pad location up to a
+  known "good" revision.
+*/
+
+if(process.argv.length != 4) {
+  console.error("Use: node bin/repairPad.js $PADID $REV");
+  process.exit(1);
+}
+
+var npm = require("../src/node_modules/npm");
+var async = require("../src/node_modules/async");
+var ueberDB = require("../src/node_modules/ueberDB");
+
+var padId = process.argv[2];
+var newRevHead = process.argv[3];
+var newPadId = padId + "-rebuilt";
+
+var db, pad, newPad, settings;
+var AuthorManager, ChangeSet, PadManager;
+
+async.series([
+  function(callback) {
+    npm.load({}, function(err) {
+      if(err) {
+        console.error("Could not load NPM: " + err)
+        process.exit(1);
+      } else {
+        callback();
+      }
+    })
+  },
+  function(callback) {
+    // Get a handle into the database
+    db = require('../src/node/db/DB');
+    db.init(callback);
+  }, function(callback) {
+     // Get references to the original pad and to a newly created pad
+     // HACK: This is a standalone script, so we want to write everything
+     // out to the database immediately.  The only problem with this is
+     // that a driver (like the mysql driver) hardcodes these values.
+     db.db.db.settings = {cache: 0, writeInterval: 0, json: true};
+     PadManager = require('../src/node/db/PadManager');
+     PadManager.getPad(padId, function(err, _pad)  {
+       pad = _pad;
+       PadManager.getPad(newPadId, function(err, _newPad) {
+         newPad = _newPad;
+         callback();
+       });
+     });
+  }, function(callback) {
+    // Clone all Chat revisions
+    var chatHead = pad.chatHead;
+    for(var i = 0; i <= chatHead; i++) {
+      db.db.get("pad:" + padId + ":chat:" + i, function (err, chat) {
+        db.db.set("pad:" + newPadId + ":chat:" + i, chat);
+        console.log("Created: Chat Revision: pad:" + newPadId + ":chat:" + i)
+      });
+    }
+    callback();
+  }, function(callback) {
+    // Rebuild Pad from revisions up to and including the new revision head
+    AuthorManager = require("../src/node/db/AuthorManager");
+    Changeset = require("ep_etherpad-lite/static/js/Changeset");
+    // Author attributes are derived from changesets, but there can also be
+    // non-author attributes with specific mappings that changesets depend on
+    // and, AFAICT, cannot be recreated any other way
+    newPad.pool.numToAttrib = pad.pool.numToAttrib;
+    for(var i = 1; i <= newRevHead; i++) {
+      db.db.get("pad:" + padId + ":revs:" + i, function(err, rev) {
+        var author = rev.meta.author;
+        var timestamp = rev.meta.timestamp;
+        var changeset = rev.changeset;
+ 
+        var newAText = Changeset.applyToAText(changeset, newPad.atext, newPad.pool);
+        Changeset.copyAText(newAText, newPad.atext);
+
+        var newRev = ++newPad.head;
+
+        var newRevData = {};
+        newRevData.changeset = changeset;
+        newRevData.meta = {};
+        newRevData.meta.author = author;
+        newRevData.meta.timestamp = timestamp;
+
+        newPad.pool.putAttrib(['author', author || '']);
+
+        if(newRev % 100 == 0)
+        {
+          newRevData.meta.atext = newPad.atext;
+        }
+
+        db.db.set("pad:"+newPad.id+":revs:"+newRev, newRevData);
+        console.log("Created: Revision: pad:" + newPad.id + ":revs:" + newRev);
+
+        if(author)
+          AuthorManager.addPad(author, newPad.id);
+
+        if (newRev == newRevHead) {
+          callback();
+        }
+      });
+    }
+  }, function(callback) {
+    // Add saved revisions up to the new revision head
+    console.log(newPad.head);
+    var newSavedRevisions = [];
+    for(var i in pad.savedRevisions) {
+      savedRev = pad.savedRevisions[i]
+      if (savedRev.revNum <= newRevHead) {
+        newSavedRevisions.push(savedRev);
+        console.log("Added: Saved Revision: " + savedRev.revNum);
+      }
+    }
+    newPad.savedRevisions = newSavedRevisions;
+    callback();
+  }, function(callback) {
+    // Save the source pad
+    db.db.set("pad:"+newPadId, newPad, function(err) {
+      console.log("Created: Source Pad: pad:" + newPadId);
+      newPad.saveToDatabase();
+      callback();
+    });
+  }
+], function (err) {
+  if(err) throw err;
+  else {
+    console.info("finished");
+    process.exit(0);
+  }
+});