Updates for Stencila

docs/source/config_files.rst

@@ -107,6 +107,22 @@ You also need to have a ``runtime.txt`` file that is formatted as
 ``r-<YYYY>-<MM>-<DD>``, where YYYY-MM-DD is a snapshot of MRAN that will be
 used for your R installation.
 
+``manifest.xml`` - Install Stencila
+===================================
+
+`Stencila <https://stenci.la/>`_ is an open source office suite for reproducible research.
+It is powered by the open file format `Dar <https://github.com/substance/dar>`_.
+
+If your repository contains a Stencila document, repo2docker detects it based on the file ``manifest.xml``.
+The required `execution contexts <https://stenci.la/learn/intro.html>` are extracted from a Dar article (i.e.
+files named ``*.jats.xml``).
+
+You may also have a ``runtime.txt`` and/or an ``install.R`` to manually configure the your R installation.
+
+To see example repositories, visit our
+`Stencila with R <https://github.com/binder-examples/stencila-r/>`_ and 
+`Stencila with Python <https://github.com/binder-examples/stencila-py>`_ examples.
+
 .. _postBuild:
 
 ``postBuild`` - Run code after installing the environment

repo2docker/buildpacks/base.py

@@ -271,16 +271,54 @@ def stencila_manifest_dir(self):
         # ${STENCILA_ARCHIVE_DIR}/${STENCILA_ARCHIVE}/manifest.xml
 
         self._stencila_manifest_dir = None
+
+        if os.path.exists('binder'):
+            return self._stencila_manifest_dir
+
         for root, dirs, files in os.walk("."):
             if "manifest.xml" in files:
+                self.log.debug("Found a manifest.xml at %s", root)
                 self._stencila_manifest_dir = root.split(os.path.sep, 1)[1]
                 self.log.info(
-                    "Found stencila manifest.xml in %s",
+                    "Using stencila manifest.xml in %s",
                     self._stencila_manifest_dir,
                 )
                 break
         return self._stencila_manifest_dir
 
+    @property
+    def stencila_manifest_contexts(self):
+        """Find the stencila manifest contexts if it exists"""
+        if hasattr(self, '_stencila_manifest_contexts'):
+            return self._stencila_manifest_contexts
+
+        # look at the content of the jats.xml file to extract the required execution contexts
+
+        self._stencila_manifest_contexts = None
+
+        if os.path.exists('binder'):
+            return self._stencila_manifest_contexts
+
+        for root, dirs, files in os.walk("."):
+            for filename in files:
+                if filename.endswith(".jats.xml"):
+                    self.log.debug("Found a .jats.xml: %s", filename)
+                    self._stencila_manifest_contexts = set()
+
+                    # extract code languages from file
+                    with open(os.path.join(root, filename)) as f:
+                        for line in f:
+                            match = re.match('.*language="(.+?)"', line)
+                            if match:
+                                self._stencila_manifest_contexts.add(match.group(1))
+
+                    self.log.info(
+                        "Using stencila executions contexts %s",
+                        self._stencila_manifest_contexts,
+                    )
+                    break
+        return self._stencila_manifest_contexts
+
     def get_build_scripts(self):
         """
         Ordered list of shell script snippets to build the base image.
@@ -518,14 +556,27 @@ def get_assemble_scripts(self):
             ))
         except FileNotFoundError:
             pass
+        if self.stencila_manifest_contexts:
+            if {'py', 'pyjp'}.intersection(self.stencila_manifest_contexts):
+                assemble_scripts.extend(
+                    [
+                        (
+                            "${NB_USER}",
+                            r"""
+                            ${KERNEL_PYTHON_PREFIX}/bin/pip install --no-cache https://github.com/stencila/py/archive/f1260796.tar.gz && \
+                            ${KERNEL_PYTHON_PREFIX}/bin/python -m stencila register
+                            """,
+                        )
+                    ]
+                )
+            #if {'r'}.intersection(self.stencila_manifest_contexts):
+                # handled in RBuildPack
         if self.stencila_manifest_dir:
             assemble_scripts.extend(
                 [
                     (
                         "${NB_USER}",
                         r"""
-                        ${KERNEL_PYTHON_PREFIX}/bin/pip install --no-cache https://github.com/stencila/py/archive/f6a245fd.tar.gz && \
-                        ${KERNEL_PYTHON_PREFIX}/bin/python -m stencila register && \
                         ${NB_PYTHON_PREFIX}/bin/pip install --no-cache nbstencilaproxy==0.1.1 && \
                         jupyter serverextension enable --sys-prefix --py nbstencilaproxy && \
                         jupyter nbextension install    --sys-prefix --py nbstencilaproxy && \

repo2docker/buildpacks/r.py

@@ -19,7 +19,14 @@ class RBuildPack(PythonBuildPack):
        date snapshot of https://mran.microsoft.com/timemachine
        from which libraries are to be installed.
 
-    2. An optional `install.R` file that will be executed at build time,
+    2. A `DESCRIPTION` file signaling an R package, or
+       a Stencila document (*.jats.xml) with R code chunks (i.e. language="r")
+       is found:
+
+       MRAN snapshot is set to latest date that is guaranteed to exist
+       across timezones.
+
+    3. (Optional) An `install.R` file that will be executed at build time,
        and can be used for installing packages from both MRAN and GitHub.
 
     The `r-base` package from Ubuntu apt repositories is used to install
@@ -64,21 +71,22 @@ def detect(self):
         unless a `requirements.txt` is present and we do not want to require the
         presence of a `requirements.txt` to use R.
 
-        Instead we just check if runtime.txt contains a string of the form
-        `r-<YYYY>-<MM>-<DD>`
+        Instead we check the options described in the class comment above.
         """
         # If no date is found, then self.checkpoint_date will be False
         # Otherwise, it'll be a date object, which will evaluate to True
         if self.checkpoint_date:
             return True
+
         description_R = 'DESCRIPTION'
-        if not os.path.exists('binder') and os.path.exists(description_R):
-            if not self.checkpoint_date:
-                # no R snapshot date set through runtime.txt
-                # set the R runtime to the latest date that is guaranteed to be on MRAN across timezones
-                self._checkpoint_date = datetime.date.today() - datetime.timedelta(days=2)
-                self._runtime = "r-{}".format(str(self._checkpoint_date))
-            return True
+        if not os.path.exists('binder'): 
+            if os.path.exists(description_R) or (self.stencila_manifest_contexts and "r" in self.stencila_manifest_contexts):
+                if not self.checkpoint_date:
+                    # no R snapshot date set through runtime.txt
+                    # set the R runtime to the latest date that is guaranteed to be on MRAN across timezones
+                    self._checkpoint_date = datetime.date.today() - datetime.timedelta(days=2)
+                    self._runtime = "r-{}".format(str(self._checkpoint_date))
+                return True
 
     def get_path(self):
         """
@@ -133,6 +141,7 @@ def get_build_scripts(self):
         - R's devtools package, at a particular frozen version (determined by MRAN)
         - IRKernel
         - nbrsessionproxy (to access RStudio via Jupyter Notebook)
+        - stencila R package (if Stencila document with R code chunks detected)
         """
         rstudio_url = 'https://download2.rstudio.org/rstudio-server-1.1.419-amd64.deb'
         # This is MD5, because that is what RStudio download page provides!
@@ -148,7 +157,7 @@ def get_build_scripts(self):
         # IRKernel version - specified as a tag in the IRKernel repository
         irkernel_version = '0.8.11'
 
-        return super().get_build_scripts() + [
+        scripts = [
             (
                 "root",
                 r"""
@@ -226,6 +235,22 @@ def get_build_scripts(self):
             ),
         ]
 
+        if self.stencila_manifest_contexts:
+            if "r" in self.stencila_manifest_contexts:
+                scripts += [
+                (
+                    "${NB_USER}",
+                    # Install and register stencila library
+                    r"""
+                    R --quiet -e "source('https://bioconductor.org/biocLite.R'); biocLite('graph')" && \
+                    R --quiet -e "devtools::install_github('stencila/r', ref = '361bbf560f3f0561a8612349bca66cd8978f4f24')" && \
+                    R --quiet -e "stencila::register()"
+                    """
+                ),
+            ]
+
+        return super().get_build_scripts() + scripts
+
     def get_assemble_scripts(self):
         """
         Return series of build-steps specific to this repository

tests/stencila/pyjp/verify

@@ -0,0 +1,5 @@
+#!/bin/sh
+
+jupyter serverextension list 2>&1 | grep nbstencilaproxy
+jupyter nbextension list 2>&1 | grep nbstencilaproxy
+python3 -c "import stencila"

tests/stencila/basic/verify → tests/stencila/r/verify

@@ -2,3 +2,4 @@
 
 jupyter serverextension list 2>&1 | grep nbstencilaproxy
 jupyter nbextension list 2>&1 | grep nbstencilaproxy
+R -e "library('stencila');"