@BibMartin @mariusvniekerk @michelleful @ocefpaf @mnfienen I'd like to put this on master soon. Please let me know if there are any objections or corrections that need to be made.

Issue

I often work with a distant IPython notebook / jupyter server ; in that case, there are cross-origin issues in the browser, as leaflet.css and leaflet.js cannot be loaded.

Fix

I fix it in using urllib2 and asking python to download and serve these files. There may be cleaner ways to do that ; especially in adapting the template.

Bonus

And I take the occasion to put the rendered figure in an iframe, so that it's javascript does not interact with the notebook's. This also enables to create an iframe with the same size as the original matplotlib figure.

mplleaflet seems to respect most of the options you can set on a matplotlib plot, except for one crucial one, which is opacity. That is, at least AFAIK.

Am I wrong (in that case, what's the parameter name!)? If not, then I believe that such a control should be made a feature.

Hi. I've recently downloaded the package and have been messing around with some basic examples.

The following snippet runs without error,

%matplotlib inline

import mplleaflet
import matplotlib.pyplot as plt

plt.plot([0,45], [0,45])

mplleaflet.show()

but changing mplleaflet.show() to mplleaflet.display() returns the following error

---------------------------------------------------------------------------
LookupError                               Traceback (most recent call last)
<ipython-input-6-a8022d0ab3fb> in <module>()
      6 plt.plot([0,45], [0,45])
      7 
----> 8 mplleaflet.display()

C:\Program Files\Miniconda3\lib\site-packages\mplleaflet\_display.py in display(fig, closefig, **kwargs)
    153     # We embed everything in an iframe.
    154     iframe_html = '<iframe src="data:text/html;base64,{html}" width="{width}" height="{height}"></iframe>'\
--> 155     .format(html = html.encode('base64'),
    156             width = '100%',
    157             height= int(60.*fig.get_figheight()),

LookupError: 'base64' is not a text encoding; use codecs.encode() to handle arbitrary codecs

I'm running Python 3.4.3, iPython 4.0.0 through Jupyter 4.0.6, matplotlib 1.4.3 and mplleaflet 0.0.4. OS is Win8.1 (64bit).

Cheers, Eilam

I have created a map of a GPS-track. If i open the HTML-File with Chrome (99.0.4844.51) or Edge ( 99.0.1150.39) i only see my track on a blank area without the tile-data. With Firefox the map is displayed.

Motivation: the fig_to_html function embeds a geojson that can be very heavy when there are a lot of coordinates and features. According to this most of the time extra float precision on the coordinates doesn't have a visible effect.

With the PR you can add float_precision to the fig_to_html function and when the geojson is created the floats will be formatted.

Example:

import mplleaflet
import matplotlib.pyplot as plt

fig, ax = plt.subplots()
ax.plot((1,2.365485622,36.254685), (1,2,3.3))

If we save the fig to html in the usual way we get:

mplleaflet.fig_to_html(fig)

We get

...
var gjData = {
    "type": "FeatureCollection", 
    "features": [{
        "type": "Feature", 
        "geometry": {
            "type": "LineString", 
            "coordinates": [[1.0, 1.0], [2.365485622, 2.0], [36.254685, 3.3]]
        }, 
        "properties": {"color": "#1F77B4", "weight": 1.5, "opacity": 1, "fillOpacity": 1}
    }]
};
...

If we use the new keyword:

mplleaflet.fig_to_html(fig, float_precision=3)

We would get:

...
var gjData = {
    "type": "FeatureCollection", 
    "features": [{
        "type": "Feature", 
        "geometry": {
            "type": "LineString", 
            "coordinates": [[1.000, 1.000], [2.365, 2.000], [36.255, 3.300]]
        }, 
        "properties": {"color": "#1F77B4", "weight": 1.500, "opacity": 1, "fillOpacity": 1}
    }]
};
...

The example above is not very representative but if you are using contours, for instance, the difference is very important.

I added several tiling options to maptiles.py. These include: Thunderforest Landscape, Esri Aerial, Esri National Geographic, and Stamen Watercolor.

I hope you will consider pulling them in.

Thanks! mike

When I run this example, I get the following traceback:

import mplleaflet
import matplotlib.pyplot as plt

plt.scatter([0, 10, 0, 10], [0, 0, 10, 10], c=[1, 2, 3, 4])
mplleaflet.display()

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-100-b851626ef3d4> in <module>()
      1 plt.scatter([0, 10, 0, 10], [0, 0, 10, 10], c=[1, 2, 3, 4])
----> 2 mplleaflet.display()

/Users/shoyer/miniconda/envs/rapid/lib/python2.7/site-packages/mplleaflet/_display.pyc in display(fig, closefig, **kwargs)
    136         plt.close(fig)
    137 
--> 138     html = fig_to_html(fig, template="ipynb.html", **kwargs)
    139     return HTML(html)
    140 

/Users/shoyer/miniconda/envs/rapid/lib/python2.7/site-packages/mplleaflet/_display.pyc in fig_to_html(fig, template, tiles, crs, epsg)
     70     renderer = LeafletRenderer(crs=crs, epsg=epsg)
     71     exporter = Exporter(renderer)
---> 72     exporter.run(fig)
     73 
     74     attribution = _attribution + ' | ' + tiles[1]

/Users/shoyer/miniconda/envs/rapid/lib/python2.7/site-packages/mplexporter/exporter.pyc in run(self, fig)
     49             import matplotlib.pyplot as plt
     50             plt.close(fig)
---> 51         self.crawl_fig(fig)
     52 
     53     @staticmethod

/Users/shoyer/miniconda/envs/rapid/lib/python2.7/site-packages/mplexporter/exporter.pyc in crawl_fig(self, fig)
    116                                        props=utils.get_figure_properties(fig)):
    117             for ax in fig.axes:
--> 118                 self.crawl_ax(ax)
    119 
    120     def crawl_ax(self, ax):

/Users/shoyer/miniconda/envs/rapid/lib/python2.7/site-packages/mplexporter/exporter.pyc in crawl_ax(self, ax)
    138                 self.draw_patch(ax, patch)
    139             for collection in ax.collections:
--> 140                 self.draw_collection(ax, collection)
    141             for image in ax.images:
    142                 self.draw_image(ax, image)

/Users/shoyer/miniconda/envs/rapid/lib/python2.7/site-packages/mplexporter/exporter.pyc in draw_collection(self, ax, collection, force_pathtrans, force_offsettrans)
    270                                            offset_order=offset_order,
    271                                            styles=styles,
--> 272                                            mplobj=collection)
    273 
    274     def draw_image(self, ax, image):

/Users/shoyer/miniconda/envs/rapid/lib/python2.7/site-packages/mplexporter/renderers/base.pyc in draw_path_collection(self, paths, path_coordinates, path_transforms, offsets, offset_coordinates, offset_order, styles, mplobj)
    253 
    254         for tup in self._iter_path_collection(paths, path_transforms,
--> 255                                               offsets, styles):
    256             (path, path_transform, offset, ec, lw, fc) = tup
    257             vertices, pathcodes = path

/Users/shoyer/miniconda/envs/rapid/lib/python2.7/site-packages/mplexporter/renderers/base.pyc in _iter_path_collection(paths, path_transforms, offsets, styles)
    190         N = max(len(paths), len(offsets))
    191 
--> 192         if not path_transforms:
    193             path_transforms = [np.eye(3)]
    194 

ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()

Thanks for your work on this project! I can't believe I just discovered it yesterday :).

I can't seem to get any contourf plots to work with mplleaflet. Other plot types (e.g., scatter plots) work fine.

The example runs without any errors on my machine, but it generates a map with only the background layer.

Any ideas for how to debug this?

I am not sure if this is the best way to implement it because matplotlib's alpha is defined only once, while in leaflet we have opacity for the stroke opacity and fillOpacity for the patch opacity.

However, I believe we need to override leaflet's default fillOpacity, which is only 0.2, to fix visualizations like this one:

http://nbviewer.ipython.org/github/rsignell-usgs/mwra-wq/blob/master/notebooks/wq_test.ipynb

The esri world topo maps are pretty good. They currently have better water feature coverage and building outlines in my area of interest than OSM (I should update OSM, but that is a different issue....)

Hi,

I am currently reproducing code for plotting maps of Germany (https://juanitorduz.github.io/germany_plots/).

I try to get the interactive map running:

import mplleaflet

fig, ax = plt.subplots()

berlin_df.plot(
    ax=ax,
    alpha=0.2
)

berlin_neighbourhoods_df.plot(
    ax=ax,
    column='neighbourhood_group',
    categorical=True,
    cmap='tab20',
)

mplleaflet.display(fig=fig)

(one of the last steps of the tutorial)

I already fixed the _gridOnMajor bug like suggested here: https://github.com/jwass/mplleaflet/issues/80 (thanks to @22tommibaer01).

Unfortunately, I ran into another bug with the following error log:

[...]
File ~/.pyenv/versions/3.10.1/lib/python3.10/site-packages/mplleaflet/mplexporter/exporter.py:263, in Exporter.draw_collection(self, ax, collection, force_pathtrans, force_offsettrans)
    255 styles = {'linewidth': collection.get_linewidths(),
    256           'facecolor': collection.get_facecolors(),
    257           'edgecolor': collection.get_edgecolors(),
    258           'alpha': collection._alpha,
    259           'zorder': collection.get_zorder()}
    261 offset_dict = {"data": "before",
    262                "screen": "after"}
--> 263 offset_order = offset_dict[collection.get_offset_position()]
    265 self.renderer.draw_path_collection(paths=processed_paths,
    266                                    path_coordinates=path_coords,
    267                                    path_transforms=path_transforms,
   (...)
    271                                    styles=styles,
    272                                    mplobj=collection)

AttributeError: 'PatchCollection' object has no attribute 'get_offset_position'

This seems to be related to the plotly issue described here: https://github.com/plotly/plotly.py/issues/3624, but I have no idea how to fix this.

My environment looks like this:

> pip show mplleaflet
Name: mplleaflet
Version: 0.0.5

> pip show matplotlib
Name: matplotlib
Version: 3.5.1

Help is much appreciated. Thanks!

mplleaflet attribute error with first use

Problem

• download mplleaflet v. 0.0.5 from conda-forge
• started a test script like

import matplotlib.pyplot as plt
import mplleaflet

long=[72.12,72.012,72.25,72.0254,73] #data (random)
lat=[50,50.21,51.1263,52.4174,53.21]

ax=plt.plot(long, lat) #matplotlib

mplleaflet.display() #mplleaflet

• not working; Error:

file "C:\Anaconda\lib\site-packages\mplleaflet\mplexporter\utils.py", line 241, in get_grid_style
    if axis._gridOnMajor and len(gridlines) > 0:

AttributeError: 'XAxis' object has no attribute '_gridOnMajor'

Solution

• not a big deal, just one line of code
• solution by this

Possible Solution

• go to line and file from the error message (def get_grid_style)
• change axis._gridOnMajor to axis._major_tick_kw['gridOn']
• finish

Setup

• conda 4.10.3
• python 3.8.8
• matplotlib 3.4.2
• mplleaflet 0.0.5

I have kml file with 1000+ features and I am trying to plot via matplotlib and mplleaflet . When I tried to plot the GeoSeries with 1000+ features it shows blank whereas it works perfect when features < 1000

I am using mplleaflet inside Python Notebook

Here is the code :

import mplleaflet
import maptplotlib.pyplot as plt
import geopandas as gpd

df= gpd.read_file('xyz.kml')

ax_wfs = df['geometry'][:1000].plot(figsize = (15,10) , color = 'black'  )
mplleaflet.display(fig=ax_wfs.figure , tiles='osm' )

I have also tried some stackoverflow hacks like %%capture %matplotlib inline changing alpha value to 0.1

but unfortunately it didnt work at all . @jwass could you please help me out ? This seems like a bug

Thanks in advance! Ash

Hello,

I know how to drag matplotlib patches like rectangles and polygons with mouse as in the examples on this page event_handling.

I would love to be able to do it with mplleaflet, that is to say with openstreetmaps background.

However, I wonder if it is possible given that mplleaflet displays in web browser whereas matplotlib relies on another backend (tk by default I think) to collect events.

If it is currently not possible, do you think there are any chance it could become possible ? or perhaps another library enables it

Regards