#!/usr/bin/env python
# coding: utf-8

# # Notebook support in Sherpa
# 
# A number of objects have been updated to support HTML output when displayed in a Jupyter notebook. Let's take a quick tour!
# 
# ## Data1D, Data1DInt, and Data2D
# 
# First we have the data objects:

# In[1]:


import numpy as np

from sherpa.data import Data1D, Data1DInt, Data2D

x = np.arange(100, 200, 20)
y = [120, 240, 30, 95, 130]

d1 = Data1D('oned', x, y)
d1i = Data1DInt('onedint', x[:-1], x[1:], y[:-1])
               
x0 = [150, 250, 100]
x1 = [250, 200, 200]
y2 = [50, 40, 70]
d2 = Data2D('twod', x0, x1, y2)      


# Each can be displayed with `print`, which shows a textual representation of attribute and values:

# In[2]:


print(d1)


# Or they can be displayed as-is which, **in a Jupyter notebook**, will display either a plot or a HTML table. The `Data1D` and `Data1DInt` classes will display a plot (if the `pylab` plotting backend is selected), and the `Data2D` class a table.

# In[3]:


d1


# In[4]:


print(d1i)


# In[5]:


d1i


# As mentioned, the `Data2D` class just gets a fancy HTML table but no plot:

# In[6]:


print(d2)


# In[7]:


d2


# ## DataPHA, DataARF, and DataRMF
# 
# The Astronomy-specific PHA, ARF, and RMF data classes can also be displayed. These (when you have `pylab` selected) display both the data and a table of information.

# In[8]:


from sherpa.astro import io

pha = io.read_pha('../sherpa-test-data/sherpatest/9774.pi')
arf = io.read_arf('../sherpa-test-data/sherpatest/9774.arf')
rmf = io.read_rmf('../sherpa-test-data/sherpatest/9774.rmf')


# In[9]:


print(pha)


# In[10]:


pha


# The PHA object will change the display based on the data - that is, if you change the filtering and grouping you will see a different plot,
# and the table will also change:

# In[11]:


pha.notice(0.3, 7)
pha.group_counts(20, tabStops=~pha.mask)

pha


# It will also change if you change the analysis setting:

# In[12]:


pha.set_analysis('wave')

pha


# The ARF and RMF objects do not change based on their settings:

# In[13]:


print(arf)


# In[14]:


arf


# In[15]:


print(rmf)


# For the RMF, five energies are selected that span the response of the instrument, and the response to these monochromatic energies are displayed.

# In[16]:


rmf


# ## DataIMG
# 
# For images with little metadata, and no WCS information, we just get an image:

# In[17]:


img = io.read_image('../sherpa-test-data/sherpatest/img.fits')


# In[18]:


img


# If the image contains WCS information, or some basic metadata, then we will get extra tables
# (unfortunately this test image doesn't display particularly wonderfully as the source
# is faint!).

# In[19]:


img2 = io.read_image('../sherpa-test-data/sherpatest/acisf08478_000N001_r0043_regevt3_srcimg.fits')

img2


# As with the PHA object, we can change the display slightly, such as changing the `coord` setting and spatially filtering the data:

# In[20]:


img2.set_coord('physical')
img2.notice2d('circle(3150, 4520, 20)')

img2


# ## Models and parameters
# 
# Models and parameters can also be displayed directly as HTML tables, mirroring their `print` output.

# In[21]:


from sherpa.models.basic import Gauss2D, Const2D

mgauss = Gauss2D()
mconst = Const2D()

mgauss.xpos = 3150
mgauss.ypos = 4520

mdl = mgauss + mconst


# We can compare the model output (this also works with a single component, such as `mgauss` and `mconst`):

# In[22]:


print(mdl)


# In[23]:


mdl


# We can have a display for parameters (I chose this model since we can see the minimum and maximum columns get displayed as units of $\pi$ in the notebook-display version):

# In[24]:


print(mgauss.theta)


# In[25]:


mgauss.theta


# ## Fitting data
# 
# Various objects related to fitting will also display in Jupyter notebooks. I fit a simple model (the model we just created, in fact) to the last image we were looking at. For this example I use the `sherpa.astro.ui` layer to fit, rather than creating the fit object manually.

# In[26]:


from sherpa.astro import ui

ui.set_data(img2)
ui.set_source(mdl)

ui.set_stat('cash')
ui.set_method('simplex')


# The output of the fit call is still just text:

# In[27]:


ui.fit()


# However, we can see the model results (as shown above):

# In[28]:


ui.get_source()


# We can also display the fit results directly (I am dropping the comparison to the `print` output in part to show you can just call routines like `ui.get_git_results` and see the display without needing to call `print`, at least in a Jupyter notebook):

# In[29]:


ui.get_fit_results()


# Similarly, the output of `conf` (or `covar`) is just text, but the results can be accessed directly with `ui.get_conf_results` or `ui.get_covar_results`:

# In[30]:


ui.conf()


# In[31]:


ui.get_conf_results()


# The `get_stat_info` call returns information for each dataset. As this is a list, the overall output just gets displayed as text, but if you access an individual element you will get a HTML table: 

# In[32]:


ui.get_stat_info()


# In[33]:


ui.get_stat_info()[0]


# Once you have created an interval- or region-projection plot, such as this comparison of the x and y centers of the gaussian, you can display the results with the relevant `get` call (in this case `ui.get_reg_proj`).

# In[34]:


ui.reg_proj(mgauss.xpos, mgauss.ypos)


# In[35]:


ui.get_reg_proj()