Merge branch 'master' of github.com:zuds-survey/zuds-pipeline

README.md

@@ -1,4 +1,4 @@
-# ZUDS Image Processing Pipeline
+# ZUDS Image Processing Pipeline [WIP]
 
 An object-oriented pipeline for ZTF image processing. Current capabilities include:
 
@@ -14,31 +14,40 @@ An object-oriented pipeline for ZTF image processing. Current capabilities inclu
 
 ## Installation
 
-### Prerequisites
+To install the package you can simply do 
 
-The pipeline requires that you install the following on your own:
-
-  * [SExtractor >= 2.19.5](https://github.com/astromatic/sextractor)
+    pip install zuds
+
+However, there are some external tools you need for the image processing steps to work. The sections below present some methods for how you can install everything needed to run the pipeline. 
+
+Prerequisites:
+
+  * [SExtractor >= 2.25.0](https://github.com/astromatic/sextractor)
   * [SWarp >= 2.38.0](https://github.com/astromatic/swarp)
-  * [hotpants >= 5.1.11](https://github.com/zuds-survey/hotpants)
   * [postgresql >= 10](https://www.postgresql.org/)
   * [cfitsio](https://heasarc.gsfc.nasa.gov/fitsio/)
-
-If you have `conda` installed on your machine, everything above but hotpants can be installed by executing
+  * [hotpants >= 5.1.11](https://github.com/zuds-survey/hotpants)
 
-     bash build.conda.sh
-     
-from within this repository. This should work on OSX, Linux, and Windows. 
 
-To build hotpants, first install the other dependencies above. Clone the zuds-survey hotpants repository linked above, cd into the directory, and type `make`. It should build an executable called `hotpants`. Copy that executable into your `PATH`, and you should be good to go.
+### Recommended method (`conda`)
 
-### Installing the python package
+This approach requires the `conda` executable to be installed on your system and used to manage your python and python packages. 
 
-Once you have the prerequisites, do 
+Clone this repository, then run 
 
-    pip install zuds
+    bash build.conda.sh
 
-to install the python package. `build.conda.sh` will do this for you, if you use it.
+to install `zuds`, sextractor, swarp, postgres, and cfitsio.
+
+After you have completed this step, to install hotpants, cd into the `hotpants` directory and type `make`. Then copy the `hotpants` executable to your `PATH`. It should then be available to the `zuds` library. 
+
+
+### Via docker-compose
+
+A complete setup of the pipeline+database is available via `docker-compose`. Clone this repository, then run `docker-compose up`. A container running a jupyter notebook with the `zuds` pipeline and all dependencies installed should spin up, as well as a separate container for the database. Navigate to `localhost:8174` and open up `demo/demo.ipynb` to start running a demo of the pipeline in a jupyter notebook. 
 
+Note: This approach is not recommended for Mac OSX users due to [resource issues with docker on Mac](https://github.com/docker/for-mac/issues/178). 
 
+### Roll your own
 
+You can download and build all of the prerequisite packages manually, then do `pip install zuds` to install the pipeline.

nersc/kernel.json

@@ -0,0 +1,14 @@
+{
+    "argv": [
+      "shifter",
+      "--image=dannygoldstein/zuds:0.1dev",
+      "--volume=HOMEDIR:/home/zuds",
+      "/opt/conda/bin/python",
+      "-m",
+      "ipykernel_launcher",
+      "-f",
+      "{connection_file}"
+    ],
+  "display_name": "zuds0.1dev",
+  "language": "python"
+}

nersc/setup_nersc.sh

@@ -0,0 +1,8 @@
+#!/bin/bash
+
+path=$HOME/.local/share/jupyter/kernels/zuds:0.1dev
+echo $path
+mkdir -p $path
+cp kernel.json $path
+sed -i "s|HOMEDIR|$HOME|g" $path/kernel.json
+