'+ title + '
' + summary +'
80 |
81 |
82 |
88 | 404
83 | 84 |Page not found
85 | 86 | 87 |Page not found
85 | 86 | 87 |Here is a python implementation of my accretion disc and emission line lightcurve-fitting code (previously CREAM). This guide briefly covers generating synthetic data and calling a new pycecream object to ingest and fit the accretion disc model (or emission line model) to a set of input light curves. I also demonstrate how to access the output of the pycecream fit. The output includes the fitted light curves, any new light curve data points after merging, fitted response functions and parameter MCMC chain histories for the disc and/or tophat response parameters.
119 |Most of these features are used in some form or another from a previous f90 version of this code (CREAM) in the following literature
120 |Please send questions to ds207@st-andrews.ac.uk.
127 |I use Gfortran. If you have an alternate (e.g ifort),
131 | please indicate the standard command used to call
132 | the fortran compiler using the fortran_caller argument
133 | (default fortran_caller = gfortran).
pip install pycecream
136 | In this example we generate 4 disk light curves and 2 emission-line light curves modelled as a top-hat with a 20-day lag. The code below generates a list where each index contains an Nx3 numpy array for each light curve. The 3 vertical axis for each light curve are the time, flux and noise respectively (query synthetic_data['echo lightcurves'][0] for an example of the format required when inputting your own light curve data).
140 |The example below combines continuum and line light curves and illustrates a case in which you may have two of the same emission line (and so want to fit with the same response function model) but observed from different telescopes that require seperate noise models.
141 |import astropy_stark.myfake as mf
142 | import matplotlib.pylab as plt
143 |
144 | '''
145 | mf.myfake arguments are
146 |
147 | wavelengths: enter the wavelengths (-1 indicates an emission line light curve modelled with a top-hat response),
148 |
149 | snr: set the signal-to-noise relative to light curve rms
150 |
151 | cadence:set the mean cadence
152 |
153 | top hat centroid: set the centroid for the top-hat (I think thats what this does but the line lag
154 | thing is still newish so Im used to just making continuum light curve)
155 | '''
156 |
157 |
158 | synthetic_data = mf.myfake(
159 | [4000.0,5000.0,5000.0,7000.0,-1.0,-1.0],
160 | [50.0,50.0,10.0,50.0,50,10.],
161 | [1.0,1.0,2.0,1.0,1.0,3.0],
162 | thcent = 20.0
163 | )
164 |
165 | '''This recovers the synthetic data'''
166 | dat = synthetic_data['echo light curves']
167 | import pycecream
171 |
172 | #instantiate a pycecream object
173 | a = pycecream.pycecream()
174 |
175 | '''
176 | If you use a fortran compiler other than gfortran please indicate here.
177 | I just re-enter gfortran here for demonstration purposes even though
178 | this is unecassary as gfortran is the default argument.
179 | '''
180 | a.fortran_caller = 'gfortran'
181 |
182 |
183 |
184 | '''Choose an output directory in which to save the results.
185 | This will be a new directory that you have not previously created (pycecream will make it automatically).
186 |
187 | NOTE: Each new cream simulation must have a new name for "output_directory argument below
188 | otherwise an excpetion is raised. This is to prevent accidentally overwriting previous simulations.
189 | I might change this in a future version
190 | '''
191 | a.project_folder = 'fit_synthetic_lightcurves'
192 |
193 |
194 |
195 | '''
196 | Add the light curves to be modeled. Inputs should be a 3 column numpy
197 | array of `time`, `flux`, `error`.
198 | In this case we are using the "dat" output
199 | from the synthetic data above.
200 | '''
201 | a.add_lc(dat[0], name = 'continuum 4000')
202 | a.add_lc(dat[1], name = 'continuum 5000')
203 | a.add_lc(dat[2], name = 'continuum 5000 (b)')
204 | a.add_lc(dat[3], name = 'continuum 7000')
205 |
206 | #If adding a line light curve, must indicate using the "kind" argument
207 | a.add_lc(dat[4],name='test line 1',kind='line')
208 |
209 | #If we want the same line response function model, set "share_previous_lag"=True
210 | a.add_lc(dat[5],name='test line 1 (shared)',kind='line',share_previous_lag=True)
211 |
212 |
213 |
214 | '''
215 | specify the numnber of MCMC iterations. Normally at least several thousand are necessary but shorter numbers
216 | can be used just to check everything is working is done here.
217 | '''
218 | a.N_iterations=40
219 |
220 | '''
221 | specify the step sizes for the fit parameters.
222 | Here we are setting the accretion rate step size to vary by ~ 0.1 solar masses per year.
223 | '''
224 | a.p_accretion_rate_step = 0.1
225 |
226 | '''
227 | Check the input settings are ok prior to running
228 | '''
229 | print(a.lightcurve_input_params)
230 |
231 | '''
232 | RUN! specify ncores (default = 1) to parallelise with 1 chain per core
233 | '''
234 | a.run(ncores = 4)
235 | The main outputs are stored in two dataframes.
239 |This a dictionary of 3 data frames.
241 |1.1) output_lightcurves['model']: standard time, model, error envelope for each file
242 |
243 | 1.2) output_lightcurves['merged model'] AS above but with the error bars, vertical and horrizontal scalings applied relative to the reference model. Not sure but I think the reference model defaults to the first occurence of a particular wavelength in the order that it was added in self.add_lc
244 |
245 | 1.3) output_lightcurves['merged data'] DICTIONARY (since the input data light curves can be different sizes) The same transformations but applied to the input light curve data. useful if using cream only to merge the orriginal light curves from different telescopes to a new scale for further study elsewhere
246 | These are the MCMC chains for each parameter.
249 |'''
250 | Get the mcmc chains and output fits.
251 | Each of these arguments come with a "location" argument where you can point to a
252 | previous simulation and recover the outputs.
253 | If this is left blank we default to the current simulation
254 | '''
255 | output_chains = a.get_MCMC_chains(location = None)
256 | output_lightcurves = a.get_light_curve_fits(location = None)
257 |
258 |
259 | '''
260 | NEW: 11/12/2019 Now the fourier chains are available as a pandas
261 | dataframe.
262 | Stats on the sine and cosine parameters are also available for each
263 | freuqency accessible in the `fourier_stats` dictionary element of this
264 | `get_MCMC_fourier_chains` function.
265 | '''
266 | output_fourier_chains = a.get_MCMC_fourier_chains(location=None)
267 | fourier_chains = output_fourier_chains['fourier_chains']
268 | fourier_stats = output_fourier_chains['fourier_stats']
269 |
270 |
271 | '''
272 | make figures of the fit, posterior, light curves etc. file prefix tells the code where you want to save the output.
273 | The figure plotting is somewhat primitive and is a relic of when I still used cream. You may prefer to use your own
274 | output figures with the output of the "get_MCMC_chains" and "get_light_curve_fits" functions above.
275 | '''
276 | a.plot_results(file_prefix='fit_figures')
277 |
278 |
279 |
280 |
281 | '''
282 | figures can also be made on an indivdual basis with axes objects returned from python plotting functions
283 | '''
284 | #plot the fitted light curves.
285 | a.plot_lightcurves()
286 | plt.show()
287 |
288 |
289 | #plot the driving light curve
290 | a.plot_driver()
291 | plt.show()
292 |
293 |
294 | #plot the parameter trace plots
295 | a.plot_trace()
296 | plt.show()
297 |
298 |
299 | #plot the covariance parameter plot for the disc parameters
300 | a.plot_posterior()
301 | plt.show()
302 |
303 |
304 | ' + summary +'
No results found
"); 41 | } 42 | } 43 | 44 | function doSearch () { 45 | var query = document.getElementById('mkdocs-search-query').value; 46 | if (query.length > 2) { 47 | if (!window.Worker) { 48 | displayResults(search(query)); 49 | } else { 50 | searchWorker.postMessage({query: query}); 51 | } 52 | } else { 53 | // Clear results for short queries 54 | displayResults([]); 55 | } 56 | } 57 | 58 | function initSearch () { 59 | var search_input = document.getElementById('mkdocs-search-query'); 60 | if (search_input) { 61 | search_input.addEventListener("keyup", doSearch); 62 | } 63 | var term = getSearchTermFromLocation(); 64 | if (term) { 65 | search_input.value = term; 66 | doSearch(); 67 | } 68 | } 69 | 70 | function onWorkerMessage (e) { 71 | if (e.data.allowSearch) { 72 | initSearch(); 73 | } else if (e.data.results) { 74 | var results = e.data.results; 75 | displayResults(results); 76 | } 77 | } 78 | 79 | if (!window.Worker) { 80 | console.log('Web Worker API not supported'); 81 | // load index in main thread 82 | $.getScript(joinUrl(base_url, "search/worker.js")).done(function () { 83 | console.log('Loaded worker'); 84 | init(); 85 | window.postMessage = function (msg) { 86 | onWorkerMessage({data: msg}); 87 | }; 88 | }).fail(function (jqxhr, settings, exception) { 89 | console.error('Could not load worker.js'); 90 | }); 91 | } else { 92 | // Wrap search in a web worker 93 | var searchWorker = new Worker(joinUrl(base_url, "search/worker.js")); 94 | searchWorker.postMessage({init: true}); 95 | searchWorker.onmessage = onWorkerMessage; 96 | } 97 | -------------------------------------------------------------------------------- /docs/_build/site/search/search_index.json: -------------------------------------------------------------------------------- 1 | {"config":{"lang":["en"],"prebuild_index":false,"separator":"[\\s\\-]+"},"docs":[{"location":"","text":"PyceCREAM Here is a python implementation of my accretion disc and emission line lightcurve-fitting code (previously CREAM). This guide briefly covers generating synthetic data and calling a new pycecream object to ingest and fit the accretion disc model (or emission line model) to a set of input light curves. I also demonstrate how to access the output of the pycecream fit. The output includes the fitted light curves, any new light curve data points after merging, fitted response functions and parameter MCMC chain histories for the disc and/or tophat response parameters. Most of these features are used in some form or another from a previous f90 version of this code (CREAM) in the following literature Grier et al in prep Grier et al 2018 https://iopscience.iop.org/article/10.3847/1538-4357/aa98dc/pdf Starkey et al 2017 https://ui.adsabs.harvard.edu/#abs/arXiv:1611.06051 Starkey et al 2016 https://ui.adsabs.harvard.edu/#abs/arXiv:1511.06162 Please send questions to ds207@st-andrews.ac.uk. Docs available here Requirements & Installation Install Fortran compiler (see macports / wget) I use Gfortran. If you have an alternate (e.g ifort), please indicate the standard command used to call the fortran compiler using the fortran_caller argument (default fortran_caller = gfortran ). Install Pycecream pip install pycecream Section 1 (Optional): Generate test synthetic light curves In this example we generate 4 disk light curves and 2 emission-line light curves modelled as a top-hat with a 20-day lag. The code below generates a list where each index contains an Nx3 numpy array for each light curve. The 3 vertical axis for each light curve are the time, flux and noise respectively (query synthetic_data['echo lightcurves'][0] for an example of the format required when inputting your own light curve data). The example below combines continuum and line light curves and illustrates a case in which you may have two of the same emission line (and so want to fit with the same response function model) but observed from different telescopes that require seperate noise models. import astropy_stark.myfake as mf import matplotlib.pylab as plt ''' mf.myfake arguments are wavelengths: enter the wavelengths (-1 indicates an emission line light curve modelled with a top-hat response), snr: set the signal-to-noise relative to light curve rms cadence:set the mean cadence top hat centroid: set the centroid for the top-hat (I think thats what this does but the line lag thing is still newish so Im used to just making continuum light curve) ''' synthetic_data = mf.myfake( [4000.0,5000.0,5000.0,7000.0,-1.0,-1.0], [50.0,50.0,10.0,50.0,50,10.], [1.0,1.0,2.0,1.0,1.0,3.0], thcent = 20.0 ) '''This recovers the synthetic data''' dat = synthetic_data['echo light curves'] Section 2: Set-up and run PyceCREAM import pycecream #instantiate a pycecream object a = pycecream.pycecream() ''' If you use a fortran compiler other than gfortran please indicate here. I just re-enter gfortran here for demonstration purposes even though this is unecassary as gfortran is the default argument. ''' a.fortran_caller = 'gfortran' '''Choose an output directory in which to save the results. This will be a new directory that you have not previously created (pycecream will make it automatically). NOTE: Each new cream simulation must have a new name for \"output_directory argument below otherwise an excpetion is raised. This is to prevent accidentally overwriting previous simulations. I might change this in a future version ''' a.project_folder = 'fit_synthetic_lightcurves' ''' Add the light curves to be modeled. Inputs should be a 3 column numpy array of `time`, `flux`, `error`. In this case we are using the \"dat\" output from the synthetic data above. ''' a.add_lc(dat[0], name = 'continuum 4000') a.add_lc(dat[1], name = 'continuum 5000') a.add_lc(dat[2], name = 'continuum 5000 (b)') a.add_lc(dat[3], name = 'continuum 7000') #If adding a line light curve, must indicate using the \"kind\" argument a.add_lc(dat[4],name='test line 1',kind='line') #If we want the same line response function model, set \"share_previous_lag\"=True a.add_lc(dat[5],name='test line 1 (shared)',kind='line',share_previous_lag=True) ''' specify the numnber of MCMC iterations. Normally at least several thousand are necessary but shorter numbers can be used just to check everything is working is done here. ''' a.N_iterations=40 ''' specify the step sizes for the fit parameters. Here we are setting the accretion rate step size to vary by ~ 0.1 solar masses per year. ''' a.p_accretion_rate_step = 0.1 ''' Check the input settings are ok prior to running ''' print(a.lightcurve_input_params) ''' RUN! specify ncores (default = 1) to parallelise with 1 chain per core ''' a.run(ncores = 4) Section 3: Outputs The main outputs are stored in two dataframes. 1) output_lightcurves = a.get_light_curve_fits(): This a dictionary of 3 data frames. 1.1) output_lightcurves['model']: standard time, model, error envelope for each file 1.2) output_lightcurves['merged model'] AS above but with the error bars, vertical and horrizontal scalings applied relative to the reference model. Not sure but I think the reference model defaults to the first occurence of a particular wavelength in the order that it was added in self.add_lc 1.3) output_lightcurves['merged data'] DICTIONARY (since the input data light curves can be different sizes) The same transformations but applied to the input light curve data. useful if using cream only to merge the orriginal light curves from different telescopes to a new scale for further study elsewhere 2) output_chains = a.get_MCMC_chains(): These are the MCMC chains for each parameter. ''' Get the mcmc chains and output fits. Each of these arguments come with a \"location\" argument where you can point to a previous simulation and recover the outputs. If this is left blank we default to the current simulation ''' output_chains = a.get_MCMC_chains(location = None) output_lightcurves = a.get_light_curve_fits(location = None) ''' NEW: 11/12/2019 Now the fourier chains are available as a pandas dataframe. Stats on the sine and cosine parameters are also available for each freuqency accessible in the `fourier_stats` dictionary element of this `get_MCMC_fourier_chains` function. ''' output_fourier_chains = a.get_MCMC_fourier_chains(location=None) fourier_chains = output_fourier_chains['fourier_chains'] fourier_stats = output_fourier_chains['fourier_stats'] ''' make figures of the fit, posterior, light curves etc. file prefix tells the code where you want to save the output. The figure plotting is somewhat primitive and is a relic of when I still used cream. You may prefer to use your own output figures with the output of the \"get_MCMC_chains\" and \"get_light_curve_fits\" functions above. ''' a.plot_results(file_prefix='fit_figures') ''' figures can also be made on an indivdual basis with axes objects returned from python plotting functions ''' #plot the fitted light curves. a.plot_lightcurves() plt.show() #plot the driving light curve a.plot_driver() plt.show() #plot the parameter trace plots a.plot_trace() plt.show() #plot the covariance parameter plot for the disc parameters a.plot_posterior() plt.show()","title":"Home"},{"location":"#pycecream","text":"Here is a python implementation of my accretion disc and emission line lightcurve-fitting code (previously CREAM). This guide briefly covers generating synthetic data and calling a new pycecream object to ingest and fit the accretion disc model (or emission line model) to a set of input light curves. I also demonstrate how to access the output of the pycecream fit. The output includes the fitted light curves, any new light curve data points after merging, fitted response functions and parameter MCMC chain histories for the disc and/or tophat response parameters. Most of these features are used in some form or another from a previous f90 version of this code (CREAM) in the following literature Grier et al in prep Grier et al 2018 https://iopscience.iop.org/article/10.3847/1538-4357/aa98dc/pdf Starkey et al 2017 https://ui.adsabs.harvard.edu/#abs/arXiv:1611.06051 Starkey et al 2016 https://ui.adsabs.harvard.edu/#abs/arXiv:1511.06162 Please send questions to ds207@st-andrews.ac.uk.","title":"PyceCREAM"},{"location":"#docs-available-here","text":"","title":"Docs available here"},{"location":"#requirements-installation","text":"","title":"Requirements & Installation"},{"location":"#install-fortran-compiler-see-macports-wget","text":"I use Gfortran. If you have an alternate (e.g ifort), please indicate the standard command used to call the fortran compiler using the fortran_caller argument (default fortran_caller = gfortran ).","title":"Install Fortran compiler (see macports / wget)"},{"location":"#install-pycecream","text":"pip install pycecream","title":"Install Pycecream"},{"location":"#section-1-optional-generate-test-synthetic-light-curves","text":"In this example we generate 4 disk light curves and 2 emission-line light curves modelled as a top-hat with a 20-day lag. The code below generates a list where each index contains an Nx3 numpy array for each light curve. The 3 vertical axis for each light curve are the time, flux and noise respectively (query synthetic_data['echo lightcurves'][0] for an example of the format required when inputting your own light curve data). The example below combines continuum and line light curves and illustrates a case in which you may have two of the same emission line (and so want to fit with the same response function model) but observed from different telescopes that require seperate noise models. import astropy_stark.myfake as mf import matplotlib.pylab as plt ''' mf.myfake arguments are wavelengths: enter the wavelengths (-1 indicates an emission line light curve modelled with a top-hat response), snr: set the signal-to-noise relative to light curve rms cadence:set the mean cadence top hat centroid: set the centroid for the top-hat (I think thats what this does but the line lag thing is still newish so Im used to just making continuum light curve) ''' synthetic_data = mf.myfake( [4000.0,5000.0,5000.0,7000.0,-1.0,-1.0], [50.0,50.0,10.0,50.0,50,10.], [1.0,1.0,2.0,1.0,1.0,3.0], thcent = 20.0 ) '''This recovers the synthetic data''' dat = synthetic_data['echo light curves']","title":"Section 1 (Optional): Generate test synthetic light curves"},{"location":"#section-2-set-up-and-run-pycecream","text":"import pycecream #instantiate a pycecream object a = pycecream.pycecream() ''' If you use a fortran compiler other than gfortran please indicate here. I just re-enter gfortran here for demonstration purposes even though this is unecassary as gfortran is the default argument. ''' a.fortran_caller = 'gfortran' '''Choose an output directory in which to save the results. This will be a new directory that you have not previously created (pycecream will make it automatically). NOTE: Each new cream simulation must have a new name for \"output_directory argument below otherwise an excpetion is raised. This is to prevent accidentally overwriting previous simulations. I might change this in a future version ''' a.project_folder = 'fit_synthetic_lightcurves' ''' Add the light curves to be modeled. Inputs should be a 3 column numpy array of `time`, `flux`, `error`. In this case we are using the \"dat\" output from the synthetic data above. ''' a.add_lc(dat[0], name = 'continuum 4000') a.add_lc(dat[1], name = 'continuum 5000') a.add_lc(dat[2], name = 'continuum 5000 (b)') a.add_lc(dat[3], name = 'continuum 7000') #If adding a line light curve, must indicate using the \"kind\" argument a.add_lc(dat[4],name='test line 1',kind='line') #If we want the same line response function model, set \"share_previous_lag\"=True a.add_lc(dat[5],name='test line 1 (shared)',kind='line',share_previous_lag=True) ''' specify the numnber of MCMC iterations. Normally at least several thousand are necessary but shorter numbers can be used just to check everything is working is done here. ''' a.N_iterations=40 ''' specify the step sizes for the fit parameters. Here we are setting the accretion rate step size to vary by ~ 0.1 solar masses per year. ''' a.p_accretion_rate_step = 0.1 ''' Check the input settings are ok prior to running ''' print(a.lightcurve_input_params) ''' RUN! specify ncores (default = 1) to parallelise with 1 chain per core ''' a.run(ncores = 4)","title":"Section 2: Set-up and run PyceCREAM"},{"location":"#section-3-outputs","text":"The main outputs are stored in two dataframes.","title":"Section 3: Outputs"},{"location":"#1-output_lightcurves-aget_light_curve_fits","text":"This a dictionary of 3 data frames. 1.1) output_lightcurves['model']: standard time, model, error envelope for each file 1.2) output_lightcurves['merged model'] AS above but with the error bars, vertical and horrizontal scalings applied relative to the reference model. Not sure but I think the reference model defaults to the first occurence of a particular wavelength in the order that it was added in self.add_lc 1.3) output_lightcurves['merged data'] DICTIONARY (since the input data light curves can be different sizes) The same transformations but applied to the input light curve data. useful if using cream only to merge the orriginal light curves from different telescopes to a new scale for further study elsewhere","title":"1) output_lightcurves = a.get_light_curve_fits():"},{"location":"#2-output_chains-aget_mcmc_chains","text":"These are the MCMC chains for each parameter. ''' Get the mcmc chains and output fits. Each of these arguments come with a \"location\" argument where you can point to a previous simulation and recover the outputs. If this is left blank we default to the current simulation ''' output_chains = a.get_MCMC_chains(location = None) output_lightcurves = a.get_light_curve_fits(location = None) ''' NEW: 11/12/2019 Now the fourier chains are available as a pandas dataframe. Stats on the sine and cosine parameters are also available for each freuqency accessible in the `fourier_stats` dictionary element of this `get_MCMC_fourier_chains` function. ''' output_fourier_chains = a.get_MCMC_fourier_chains(location=None) fourier_chains = output_fourier_chains['fourier_chains'] fourier_stats = output_fourier_chains['fourier_stats'] ''' make figures of the fit, posterior, light curves etc. file prefix tells the code where you want to save the output. The figure plotting is somewhat primitive and is a relic of when I still used cream. You may prefer to use your own output figures with the output of the \"get_MCMC_chains\" and \"get_light_curve_fits\" functions above. ''' a.plot_results(file_prefix='fit_figures') ''' figures can also be made on an indivdual basis with axes objects returned from python plotting functions ''' #plot the fitted light curves. a.plot_lightcurves() plt.show() #plot the driving light curve a.plot_driver() plt.show() #plot the parameter trace plots a.plot_trace() plt.show() #plot the covariance parameter plot for the disc parameters a.plot_posterior() plt.show()","title":"2) output_chains = a.get_MCMC_chains():"},{"location":"pycecream/","text":"pycecream","title":"pycecream"}]} -------------------------------------------------------------------------------- /docs/_build/site/search/worker.js: -------------------------------------------------------------------------------- 1 | var base_path = 'function' === typeof importScripts ? '.' : '/search/'; 2 | var allowSearch = false; 3 | var index; 4 | var documents = {}; 5 | var lang = ['en']; 6 | var data; 7 | 8 | function getScript(script, callback) { 9 | console.log('Loading script: ' + script); 10 | $.getScript(base_path + script).done(function () { 11 | callback(); 12 | }).fail(function (jqxhr, settings, exception) { 13 | console.log('Error: ' + exception); 14 | }); 15 | } 16 | 17 | function getScriptsInOrder(scripts, callback) { 18 | if (scripts.length === 0) { 19 | callback(); 20 | return; 21 | } 22 | getScript(scripts[0], function() { 23 | getScriptsInOrder(scripts.slice(1), callback); 24 | }); 25 | } 26 | 27 | function loadScripts(urls, callback) { 28 | if( 'function' === typeof importScripts ) { 29 | importScripts.apply(null, urls); 30 | callback(); 31 | } else { 32 | getScriptsInOrder(urls, callback); 33 | } 34 | } 35 | 36 | function onJSONLoaded () { 37 | data = JSON.parse(this.responseText); 38 | var scriptsToLoad = ['lunr.js']; 39 | if (data.config && data.config.lang && data.config.lang.length) { 40 | lang = data.config.lang; 41 | } 42 | if (lang.length > 1 || lang[0] !== "en") { 43 | scriptsToLoad.push('lunr.stemmer.support.js'); 44 | if (lang.length > 1) { 45 | scriptsToLoad.push('lunr.multi.js'); 46 | } 47 | for (var i=0; i < lang.length; i++) { 48 | if (lang[i] != 'en') { 49 | scriptsToLoad.push(['lunr', lang[i], 'js'].join('.')); 50 | } 51 | } 52 | } 53 | loadScripts(scriptsToLoad, onScriptsLoaded); 54 | } 55 | 56 | function onScriptsLoaded () { 57 | console.log('All search scripts loaded, building Lunr index...'); 58 | if (data.config && data.config.separator && data.config.separator.length) { 59 | lunr.tokenizer.separator = new RegExp(data.config.separator); 60 | } 61 | if (data.index) { 62 | index = lunr.Index.load(data.index); 63 | data.docs.forEach(function (doc) { 64 | documents[doc.location] = doc; 65 | }); 66 | console.log('Lunr pre-built index loaded, search ready'); 67 | } else { 68 | index = lunr(function () { 69 | if (lang.length === 1 && lang[0] !== "en" && lunr[lang[0]]) { 70 | this.use(lunr[lang[0]]); 71 | } else if (lang.length > 1) { 72 | this.use(lunr.multiLanguage.apply(null, lang)); // spread operator not supported in all browsers: https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Operators/Spread_operator#Browser_compatibility 73 | } 74 | this.field('title'); 75 | this.field('text'); 76 | this.ref('location'); 77 | 78 | for (var i=0; i < data.docs.length; i++) { 79 | var doc = data.docs[i]; 80 | this.add(doc); 81 | documents[doc.location] = doc; 82 | } 83 | }); 84 | console.log('Lunr index built, search ready'); 85 | } 86 | allowSearch = true; 87 | postMessage({allowSearch: allowSearch}); 88 | } 89 | 90 | function init () { 91 | var oReq = new XMLHttpRequest(); 92 | oReq.addEventListener("load", onJSONLoaded); 93 | var index_path = base_path + '/search_index.json'; 94 | if( 'function' === typeof importScripts ){ 95 | index_path = 'search_index.json'; 96 | } 97 | oReq.open("GET", index_path); 98 | oReq.send(); 99 | } 100 | 101 | function search (query) { 102 | if (!allowSearch) { 103 | console.error('Assets for search still loading'); 104 | return; 105 | } 106 | 107 | var resultDocuments = []; 108 | var results = index.search(query); 109 | for (var i=0; i < results.length; i++){ 110 | var result = results[i]; 111 | doc = documents[result.ref]; 112 | doc.summary = doc.text.substring(0, 200); 113 | resultDocuments.push(doc); 114 | } 115 | return resultDocuments; 116 | } 117 | 118 | if( 'function' === typeof importScripts ) { 119 | onmessage = function (e) { 120 | if (e.data.init) { 121 | init(); 122 | } else if (e.data.query) { 123 | postMessage({ results: search(e.data.query) }); 124 | } else { 125 | console.error("Worker - Unrecognized message: " + e); 126 | } 127 | }; 128 | } 129 | -------------------------------------------------------------------------------- /docs/_build/site/sitemap.xml: -------------------------------------------------------------------------------- 1 | 2 |