Skip to content

Latest commit

 

History

History
278 lines (199 loc) · 17.3 KB

readme.md

File metadata and controls

278 lines (199 loc) · 17.3 KB

NoTex.ch: An Online Text Editor

Join the chat at https://gitter.im/hsk81/notex-v2.0

An online text editor for reStructuredText, Markdown, LaTex and more! It has integrated project management, syntax highlighting, split view and spell checking for 85+ dictionaries. Projects can be exported and imported as ZIP archives plus reStructuredText projects can be converted to PDF, HTML or LaTex. Give it a try: Visit https://notex.ch, open a sample project and press the Export as .. PDF button.

Commercial Support

Do you need help in setting up NoTex for your organization or institution? Does the described setup not correspond to your requirements? Would you like to have the application customized? Don't hesitate, and write to [email protected]: We'll provide you with an SLA (service level agreement) tailored to your needs!

Installation

  • git clone https://github.com/hsk81/notex notex.git && cd notex.git

Clone the GIT repository to the local disk and change the current working directory to the top level of the repository.

  • docker build -rm -t hsk81/notex:run .

Build a docker container image and tag it as hsk81/notex:run: If your machine or internet connection is slow then just go have lunch, or do something time consuming, since the build process will take a while. A docker version 0.11.1 or newer is recommended.

Execution: Development

  • docker run -t -p 5050:5000 hsk81/notex:run ./webed.py help

Show all available commands; execute e.g. ./webed.py run -h to see detailed options for the specified command. Don't get confused about ./webed.py, which is the internal name of NoTex.

  • docker run -t -p 5050:5000 hsk81/notex:run $(cat RUN.dev)

Run the docker container and map the internal port 5000 to the external port 5050. The $(cat RUN.dev) sub-process delivers the actual command to start the application; see the RUN.dev file for details.

  • http://localhost:5050

Navigate your browser to the above location, and enjoy! The application runs in debug mode, so don't use this approach in a production environment; see the next section for that.

INFO: Actual conversion (to PDF, HTML etc.) will not work yet, since the corresponding workers have not been started!

Execution: Production

You need to run three components to get a functional application: a frontend ntx which is connected via a queue qqq to a backend conversion worker spx-1. But before starting any of the components you first need the setup a location, which can be used to exchange data:

  • mkdir -p /var/www/webed && chown www-data:www-data /var/www/webed -R

Create /var/www/webed for sharing purposes (on the host machine), and give ownership to the www-data user and group; some GNU/Linux distributions may use http instead of www-data as the owner.

Frontend: ntx

export QUEUE=tcp://172.17.42.1 && docker run -name ntx -t -p 8080:80 -p 9418:9418 -v /var/www/webed:/var/www/webed:rw hsk81/notex:run PING_ADDRESS=$QUEUE:7070 DATA_ADDRESS=$QUEUE:9090 $(cat RUN.pro)

Export first the QUEUE environment variable which needs to contain the TCP/IP address of a queue (to be started in the next step); if you run all three components on the same host then you can use the address of docker's bridge , e.g. lxcbr0 (or similar: run ifconfig to get a listing of enabled interfaces).

Then run the frontend container named ntx and map the internal port 80 to the external port 8080; the PING_ADRESS and DATA_ADDRESS variables are set within the containers environment and tell the frontend where the ping and data channels need to connect to; finally the $(cat RUN.pro) sub-process delivers the actual command to start the application and is executed as a container process; see the RUN.pro file for details.

The command also maps the 9418 port, which belongs to a git-daemon: This allows you to clone a particular repository from your host (if you know its randomly generated name), like git clone git://...

Queue: qqq

docker run -name qqq -t -p 7070:7070 -p 9090:9090 -p 7171:7171 -p 9191:9191 hsk81/notex:run ./webed-sphinx.py queue -pfa 'tcp://*:7070' -dfa 'tcp://*:9090' -pba 'tcp://*:7171' -dba 'tcp://*:9191'

Start the queue container named qqq, map the required ports to the host machine, and ensure that the TCP/IP binding addresses for the ping and data channels are declared correctly; actually you could omit the -pfa, -dfa, -pba and -dba arguments, since in this case the default values are used anyway.

As mentioned the queue offers four binding points: two for frontend container(s) to connect to (-pfa or --ping-frontend-address, and -dfa or --data-frontend-address), plus another two for backend worker(s) to connect to (-pba or --ping-backend-address, and -dba or --data-backend-address).

The application uses the ping and data channels for different purposes: Given a conversion job, the frontend asks the queue via a "ping" through the former channel if a worker is available, and if so it sends the corresponding data through the latter channel. The queue figures in a similar way which backend converter is idle and chooses it for the job.

Backend: spx-1

export QUEUE=tcp://172.17.42.1 && docker run -name spx-1 -t -v /var/www/webed:/var/www/webed:rw hsk81/notex:run ./webed-sphinx.py converter -p $QUEUE:7171 -d $QUEUE:9191 --worker-threads 2

Run a worker container named spx-1, and connect to the queue by wiring the ping and data channels to the corresponding address and ports. The worker starts internally two threads: depending on job load and resources you can increase or decrease the number of conversion threads per worker.

You could also start another worker container by repeating the same command except by using another name, e.g. spx-2: But this does not make much sense, if the same physical host is used (increase the number of worker threads instead); if you would run the command on another host though, then you probably would need to provide a correct TCP/IP address for the QUEUE variable.

  • http://localhost:8080

Navigate your browser to the above location, and enjoy! The frontend container runs internally nginx to serve the application; depending on your needs use another external port than 8080 or proxy to localhost:8080 via an nginx (or apache etc.) instance on your host machine.

Configuration

The default configuration (for the production environment) should be adapted to your needs, since otherwise some of the services which are included in NoTex might not run as expected. Below you'll find the configuration apdaptations which are used for the NoTex.ch site itself.

Caching (webed.conf)

On NoTex.ch memcached is not run within the container, but externally (the command from RUN.pro has been adapted by omitting MEMCACHED [it could have also be set to zero: MEMCACHED=0]). Therefore the corresponding memcached_pass value had to be adapted:

diff --git a/webed.conf b/webed.conf
index 09aa27b..5c15110 100644
--- a/webed.conf
+++ b/webed.conf
@@ -15,7 +15,7 @@ server {
     location /cache {
         internal;
         set $memcached_key      $args;
-        memcached_pass          127.0.0.1:11211;
+        memcached_pass          172.17.42.1:11211;
         default_type            application/octet-stream;
         expires                 15s;
         add_header              Cache-Control private;

Caching and Admin (default.py)

External Caching

On NoTex.ch memcached and redis are not run within the container, but externally (the command from RUN.pro has been adapted by omitting MEMCACHED and REDIS [they could have also be set to zero: MEMCACHED=0 and REDIS=0]). Therefore the corresponding CACHE{0,1,2,3}_SERVERS values had to be adapted:

diff --git a/webed/config/default.py b/webed/config/default.py
index e77e80c..751fcfc 100644
--- a/webed/config/default.py
+++ b/webed/config/default.py
@@ -27,13 +27,13 @@ class DefaultConfig:
         'CACHE_DEFAULT_TIMEOUT', PERMANENT_SESSION_LIFETIME.total_seconds ()))
 
     CACHE0_KEY_PREFIX = os.getenv ('CACHE0_KEY_PREFIX', 'webed-std:')
-    CACHE0_SERVERS = eval (os.getenv ('CACHE0_SERVERS', str (['127.0.0.1'])))
+    CACHE0_SERVERS = eval (os.getenv ('CACHE0_SERVERS', str (['172.17.42.1'])))
     CACHE1_KEY_PREFIX = os.getenv ('CACHE1_KEY_PREFIX', 'webed-obj:')
-    CACHE1_SERVERS = eval (os.getenv ('CACHE1_SERVERS', str (['127.0.0.1'])))
+    CACHE1_SERVERS = eval (os.getenv ('CACHE1_SERVERS', str (['172.17.42.1'])))
     CACHE2_KEY_PREFIX = os.getenv ('CACHE2_KEY_PREFIX', 'webed-sss:')
-    CACHE2_SERVERS = eval (os.getenv ('CACHE2_SERVERS', str (['127.0.0.1'])))
+    CACHE2_SERVERS = eval (os.getenv ('CACHE2_SERVERS', str (['172.17.42.1'])))
     CACHE3_KEY_PREFIX = os.getenv ('CACHE3_KEY_PREFIX', 'webed-dbs:')
-    CACHE3_SERVERS = eval (os.getenv ('CACHE3_SERVERS', str (['127.0.0.1'])))
+    CACHE3_SERVERS = eval (os.getenv ('CACHE3_SERVERS', str (['172.17.42.1'])))
 
     LOG_FILE = os.path.join ('..', 'webed.logs', 'webed.log')
     LOG_FILE = os.getenv ('LOG_FILE', LOG_FILE)
@@ -63,10 +63,10 @@ class DefaultConfig:
     ##
 
     PRIVILEGED_ADDRESSES = eval (os.getenv (
-        'PRIVILEGED_ADDRESSES', str (['127.0.0.1'])))
+        'PRIVILEGED_ADDRESSES', str (['XXX.X.X.X'])))
 
     PROXY_FIX = eval (os.getenv (
-        'PROXY_FIX', str (False)))
+        'PROXY_FIX', str (True)))
 
     ##
     ## MIMETYPE_PATHs is a list of paths point to a `mime.types` file, which

Admin Security

In addition, the PRIVILEGED_ADDRESSES has been changed to allow secure access to the admin interface: To access it use (a) the IP address of a machine, which has a back-end proxy (like squid) running, (b) on which you can SSH tunnel to, and (c) which has a front-end proxy (like nginx) running (pointing to the NoTex instance). Further then (d) connect your browser to the local port of the SSH tunnel (change the corresponding proxy settings in your browser).

You should also make sure to run the front-end proxy via HTTPS (for which a self-signed SSL certificated should be enough). Then navigate to e.g. admin.notex.ch (or whereever the now privileged machine is running):

  • https://admin.notex.ch/admin/

Provided, you've set-up the privileged machine correctly, the admin interface should appear: By default you'll be recognized as an annonymous user. To login visit https://admin.notex.ch/admin/login and to logout https://admin.notex.ch/admin/logout: On successful login you should be recognized as an admin user, and acquire access to the corresponding actions/menus.

The security of this setup is based upon the admin user possessing the SSH key file for the privileged machine to be able to establish an SSH tunnel successfully. As long as the priviledged machine is not compromised, the access should be safe. Consider also a browser plugin which can quickly enable and disable the proxy access: As long as the proxy is disabled you should not be able to login as an admin user.

Set PROXY_FIX to True, and correctly forward the remote address of a client (in the front-end proxy via setting the X-Forwarded-For and X-Real-IP headers: see the proxy.conf file for details), since otherwise this setup will not work. Do not set PROXY_FIX if you connect the NoTex instance directly (without a front-end proxy in between) to the internet!

Security, DB, Mail, Blog and CDN (production.py)

Replace the authentication tokens marked with XXX with the corresponding values for your environment: A SECRET_KEY can be generated by simply running the production.py script; the DB is run (like the caches) externally (which means the POSTGRESQL environment variable has been omitted in RUN.pro [or set to zero: POSTGRESQL=0]); the MAIL_* and DEFAULT_MAIL_* values have been adapted to allow the contact page to work properly; the BLOG_* values have been changed to let the desired blogs appear; and finally a CDN has been setup to allow fast global access:

diff --git a/webed/config/production.py b/webed/config/production.py
index 8f5e39a..3104eec 100644
--- a/webed/config/production.py
+++ b/webed/config/production.py
@@ -23,14 +23,14 @@ import os
 ###############################################################################
 ###############################################################################
 
-SECRET_KEY = '000000000000000000000000000000000000000=' ## [!!]
+SECRET_KEY = 'XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX=' ## [**]
 SECRET_KEY = os.getenv ('SECRET_KEY', SECRET_KEY)
 
 ###############################################################################
 ###############################################################################
 
 SQLALCHEMY_DATABASE_URI = os.getenv ('SQLALCHEMY_DATABASE_URI',
-    'postgresql://webed-p:password@localhost/webed-p') ## [ok]
+    'postgresql://webed-p:[email protected]/webed-p') ## [ok]
 
 ###############################################################################
 ###############################################################################
@@ -45,26 +45,26 @@ VCS_CONF = os.getenv ('VCS_CONF', VCS_CONF)
 ###############################################################################
 ###############################################################################
 
-MAIL_SERVER = os.getenv ('MAIL_SERVER', 'SMTP.SERVER.NET') ## [??]
+MAIL_SERVER = os.getenv ('MAIL_SERVER', 'smtp.XXXXXX.XXX') ## [**]
 MAIL_PORT = int (os.getenv ('MAIL_PORT', 587)) ## [ok]
 MAIL_USE_TLS = eval (os.getenv ('MAIL_USE_TLS', str (True))) ## [ok]
-MAIL_USERNAME = os.getenv ('MAIL_USERNAME', 'USERNAME') ## [??]
-MAIL_PASSWORD = os.getenv ('MAIL_PASSWORD', 'PASSWORD') ## [??]
+MAIL_USERNAME = os.getenv ('MAIL_USERNAME', 'XXXXXXXX') ## [**]
+MAIL_PASSWORD = os.getenv ('MAIL_PASSWORD', 'XXXXXXXX') ## [**]
 
 DEFAULT_MAIL_SENDER = eval (os.getenv (
-    'DEFAULT_MAIL_SENDER', str (('SENDER', '[email protected]')))) ## [??]
+    'DEFAULT_MAIL_SENDER', str (('XXXXXX', '[email protected]')))) ## [**]
 DEFAULT_MAIL_RECEIVERS = eval (os.getenv (
-    'DEFAULT_MAIL_RECEIVERS', str (['[email protected]']))) ## [??]
+    'DEFAULT_MAIL_RECEIVERS', str (['[email protected]']))) ## [**]
 
 ###############################################################################
 ###############################################################################
 
 BLOG_ID = os.getenv ('BLOG_ID', str (
-    '0000000000000000000')) ## [??]
+    'XXXXXXXXXXXXXXXXXXX')) ## [**]
 BLOG_URL = os.getenv ('BLOG_URL', str (
     'https://www.googleapis.com/blogger/v3/blogs/%s/posts' % BLOG_ID)) ## [ok]
 BLOG_API_KEY = os.getenv ('BLOG_API_KEY', str (
-    '0000000000000--000000000000000000000000')) ## [??]
+    'XXXXXXXXXXXXX--XXXXXXXXXXXXXXXXXXXXXXXX')) ## [**]
 BLOG_HEADERS = eval (os.getenv ('BLOG_HEADERS', str (
     {'Accept-Encoding': 'gzip', 'User-Agent': 'NoTex.ch (gzip)'}))) ## [ok]
@@ -72,6 +72,7 @@ BLOG_HEADERS = eval (os.getenv ('BLOG_HEADERS', str (
 ###############################################################################
 
 FORUM_URL = os.getenv ('FORUM_URL', 'https://bbs.notex.ch/index.php') ## [ok]
+CDN = os.getenv ('CDN', 'https://XXXXXXXXXXX') ## [**]
 
 ###############################################################################
 ###############################################################################

Web frontend to GIT (gitweb.conf)

The gitweb service provides a web interface to a GIT repository; the $project_list setting should definitely be set, since it controls which projects are seen on the main view. Since such a list is not desirable - due to privacy reasons - it points to an empty/non-existent $projectroot/project.lst file:

diff --git a/gitweb.conf b/gitweb.conf
index 78c8ad1..5696f40 100644
--- a/gitweb.conf
+++ b/gitweb.conf
@@ -5,10 +5,10 @@ $git_temp = "/tmp";
 $projectroot = "/var/www/webed/acid"; 
 
 # File listing projects, or directory to be scanned for projets.
-$projects_list = "$projectroot";
+$projects_list = "$projectroot/project.lst";
 
 # Base URLs for links displayed in the web interface.
-our @git_base_url_list = qw(git://localhost);
+our @git_base_url_list = qw(git://vcs.notex.ch);
 
 # Show the author of each line in a source file.
 $feature{'blame'}{'default'} = [1];
@@ -17,10 +17,10 @@ $feature{'blame'}{'default'} = [1];
 $feature{'highlight'}{'default'} = [1];
 
 # Label for the "home link" at the top of all pages.
-$home_link_str = "NoTex";
+$home_link_str = "NoTex.ch";
 
 # Target of the home link on the top of all pages.
-$home_link = "http://localhost:8008/git/";
+$home_link = "https://notex.ch";
 
 # Name of your site or organization.
-$site_name = "NoTex - Git Web Interface";
+$site_name = "NoTex.ch - Git Web Interface";

Commercial Support

Do you need help in setting up NoTex for your organization or institution? Does the described setup not correspond to your requirements? Would you like to have the application customized? Do not hesitate, and write to [email protected]: We'll provide you with an SLA (service level agreement) tailored to your needs!