Merge pull request #112 from scossu/development

Switch contributing branch

Dockerfile

@@ -1,15 +1,33 @@
-FROM        python:3.7
-MAINTAINER  Michael B. Klein <michael.klein@northwestern.edu>
-RUN         mkdir -p /usr/local /data
-WORKDIR     /usr/local
-ADD         . lakesuperior
-WORKDIR     /usr/local/lakesuperior
-RUN         git submodule update --init
-RUN         [[ -f lakesuperior/model/base.c ]] || \
-                pip install -r requirements_dev.txt
+FROM        python:3.7-alpine
+LABEL       "maintainer"="Michael B. Klein <michael.klein@northwestern.edu>" \
+            "maintainer"="Stefano Cossu <scossu@getty.edu>"
+
+RUN         apk add --no-cache build-base git
+RUN         pip3 install cython==0.29.6 cymem
+
+RUN         mkdir -p /data
+WORKDIR     /usr/local/lsup/src
+COPY        .git ./.git
+COPY        ext ./ext
+COPY        bin ./bin
+COPY        lakesuperior ./lakesuperior
+COPY        setup.py README.rst ./
+
+RUN         git submodule update --init ext
 RUN         pip install -e .
-RUN         cp ./docker/etc/* ./lakesuperior/etc.defaults/
-CMD         ./docker/docker_entrypoint
+COPY        ./docker/etc ./lakesuperior/etc.defaults
+
+# Clean up build-required packages & dirs.
+RUN         apk del build-base git
+RUN         rm -rf .git
+
+RUN         [ -f /data/ldprs_store/data.mdb ] || \
+                echo yes | lsup-admin bootstrap
+
+VOLUME      /data
+
 EXPOSE      8000
-HEALTHCHECK --interval=30s --timeout=5s \
-  CMD curl -X OPTIONS -f http://localhost:8000/ || exit 1
+
+ENTRYPOINT  ["gunicorn", "-c", "python:lakesuperior.wsgi", \
+            "lakesuperior.server:fcrepo"]
+#ENTRYPOINT  ["./bin/fcrepo"]

bin/fcrepo

@@ -1,3 +1,3 @@
-#!/bin/bash
+#!/bin/sh
 
 gunicorn -c python:lakesuperior.wsgi lakesuperior.server:fcrepo

docker-compose.yml

@@ -1,11 +1,12 @@
-version: '2'
+version: '3'
 
 volumes:
   lakesuperior:
 
 services:
   lakesuperior:
-    image: scossu/lakesuperior:stable
+    build:
+      context: .
     volumes:
       - lakesuperior:/data
     ports:

docker/etc/application.yml

@@ -10,7 +10,7 @@
 #   One of ``prod``, ``test`` or ``dev``.
 #   ``prod`` is normal running mode. 'test' is used for running test suites.
 #   ``dev`` is similar to normal mode but with reload and debug enabled.
-app_mode: 'dev'
+app_mode: 'prod'
 
 ###
 #   Base data directory.
@@ -23,7 +23,7 @@ app_mode: 'dev'
 #   the individual subdirectories can be mounted on different file systems.
 #
 #   If unset, it will default to <lakesuperior package root>/data.
-data_dir:
+data_dir: /data
 
 ###
 # Configuration for binary path and fixity check generation. The hash is a

docs/contributing.rst

@@ -65,11 +65,10 @@ displaying and linking correctly.
 
 Likewise, please add mindful testing to new fatures or bug fixes.
 
-Development is done on the ``development`` branch. If you have any suggested
-addition to the code, please fork the repo, create a new branch for your topic
-and open a pull request against development. In case you find a critical bug,
-a hotfix can be proposed against master if agreed in the related issue
-discussion.
+Development is done on the ``master`` branch. If you have an addition to the
+code that you have previously discussed by opening an issue, please fork the
+repo, create a new branch for your topic and open a pull request against
+master.
 
 Last but not least, read carefully the `Code of Conduct
 <https://github.com/scossu/lakesuperior/blob/master/code_of_conduct.md>`__.

lakesuperior/__init__.py

@@ -4,7 +4,7 @@
 
 
 version = '1.0 alpha'
-release = '1.0.0a21'
+release = '1.0.0a22'
 
 basedir = path.dirname(path.realpath(__file__))
 """

lakesuperior/api/resource.py

@@ -11,11 +11,13 @@
 from rdflib.namespace import XSD
 
 from lakesuperior.config_parser import config
+from lakesuperior.dictionaries.namespaces import ns_collection as nsc
 from lakesuperior.exceptions import (
         InvalidResourceError, ResourceNotExistsError, TombstoneError)
 from lakesuperior import env, thread_env
 from lakesuperior.globals import RES_DELETED, RES_UPDATED
 from lakesuperior.model.ldp.ldp_factory import LDP_NR_TYPE, LdpFactory
+from lakesuperior.util.toolbox import rel_uri_to_urn
 
 
 logger = logging.getLogger(__name__)
@@ -230,26 +232,30 @@ def create_or_replace(uid, **kwargs):
     :rtype: tuple(str, lakesuperior.model.ldp.ldpr.Ldpr)
     :return: A tuple of:
         1. Event type (str): whether the resource was created or updated.
-        2. Resource (lakesuperior.model.ldp.ldpr.Ldpr): The new or updated resource.
+        2. Resource (lakesuperior.model.ldp.ldpr.Ldpr): The new or updated
+            resource.
     """
     rsrc = LdpFactory.from_provided(uid, **kwargs)
     return rsrc.create_or_replace(), rsrc
 
 
 @transaction(True)
-def update(uid, update_str, is_metadata=False):
+def update(uid, update_str, is_metadata=False, handling='strict'):
     """
     Update a resource with a SPARQL-Update string.
 
     :param string uid: Resource UID.
     :param string update_str: SPARQL-Update statements.
     :param bool is_metadata: Whether the resource metadata are being updated.
+    :param str handling: How to handle server-managed triples. ``strict``
+        (the default) rejects the update with an exception if server-managed
+        triples are being changed. ``lenient`` modifies the update graph so
+        offending triples are removed and the update can be applied.
 
     :raise InvalidResourceError: If ``is_metadata`` is False and the resource
         being updated is a LDP-NR.
     """
-    # FCREPO is lenient here and Hyrax requires it.
-    rsrc = LdpFactory.from_stored(uid, handling='lenient')
+    rsrc = LdpFactory.from_stored(uid, handling=handling)
     if LDP_NR_TYPE in rsrc.ldp_types and not is_metadata:
         raise InvalidResourceError(
                 'Cannot use this method to update an LDP-NR content.')
@@ -274,6 +280,16 @@ def update_delta(uid, remove_trp, add_trp):
         add, as 3-tuples of RDFLib terms.
     """
     rsrc = LdpFactory.from_stored(uid)
+
+    # FIXME Wrong place to put this, should be at the LDP level.
+    remove_trp = {
+        (rel_uri_to_urn(s, uid), p, rel_uri_to_urn(o, uid))
+        for s, p, o in remove_trp
+    }
+    add_trp = {
+        (rel_uri_to_urn(s, uid), p, rel_uri_to_urn(o, uid))
+        for s, p, o in add_trp
+    }
     remove_trp = rsrc.check_mgd_terms(remove_trp)
     add_trp = rsrc.check_mgd_terms(add_trp)
 

lakesuperior/endpoints/ldp.py

@@ -23,7 +23,8 @@
 from lakesuperior.exceptions import (
         ChecksumValidationError, ResourceNotExistsError, TombstoneError,
         ServerManagedTermError, InvalidResourceError, SingleSubjectError,
-        ResourceExistsError, IncompatibleLdpTypeError)
+        ResourceExistsError, IncompatibleLdpTypeError,
+        IndigestibleError)
 from lakesuperior.globals import RES_CREATED
 from lakesuperior.model.ldp.ldp_factory import LdpFactory
 from lakesuperior.model.ldp.ldp_nr import LdpNr
@@ -283,18 +284,33 @@ def post_resource(parent_uid):
         kwargs['mimetype'] = mimetype
         # Check digest if requested.
         if 'digest' in request.headers:
-            kwargs['prov_cksum_algo'], kwargs['prov_cksum'] = \
+            try:
+                kwargs['prov_cksum_algo'], kwargs['prov_cksum'] = (
                     request.headers['digest'].split('=')
+                )
+            except ValueError:
+                return (
+                    f'Cannot parse digest value: {request.headers["digest"]}',
+                    400
+                )
 
     try:
         rsrc = rsrc_api.create(parent_uid, slug, **kwargs)
+    except IndigestibleError:
+        return (
+            f'Unable to parse digest header: {request.headers["digest"]}'
+        ), 400
     except ResourceNotExistsError as e:
         return str(e), 404
     except (InvalidResourceError, ChecksumValidationError) as e:
         return str(e), 409
     except TombstoneError as e:
         return _tombstone_response(e, uid)
     except ServerManagedTermError as e:
+        rsp_headers['Link'] = (
+            f'<{uri}>; rel="{nsc["ldp"].constrainedBy}"; '
+            f'{g.webroot}/info/ldp_constraints"'
+        )
         return str(e), 412
 
     uri = g.tbox.uid_to_uri(rsrc.uid)
@@ -303,8 +319,9 @@ def post_resource(parent_uid):
     rsp_headers['Location'] = uri
 
     if mimetype and kwargs.get('rdf_fmt') is None:
-        rsp_headers['Link'] = (f'<{uri}/fcr:metadata>; rel="describedby"; '
-                               f'anchor="{uri}"')
+        rsp_headers['Link'] = (
+            f'<{uri}/fcr:metadata>; rel="describedby"; anchor="{uri}"'
+        )
 
     return uri, 201, rsp_headers
 
@@ -393,6 +410,8 @@ def patch_resource(uid, is_metadata=False):
     if cond_ret:
         return cond_ret
 
+    handling, _ = set_post_put_params()
+
     rsp_headers = {'Content-Type' : 'text/plain; charset=utf-8'}
     if request.mimetype != 'application/sparql-update':
         return 'Provided content type is not a valid parsable format: {}'\
@@ -401,7 +420,7 @@ def patch_resource(uid, is_metadata=False):
     update_str = request.get_data().decode('utf-8')
     local_update_str = g.tbox.localize_ext_str(update_str, nsc['fcres'][uid])
     try:
-        rsrc = rsrc_api.update(uid, local_update_str, is_metadata)
+        rsrc = rsrc_api.update(uid, local_update_str, is_metadata, handling)
     except (ServerManagedTermError, SingleSubjectError) as e:
         return str(e), 412
     except InvalidResourceError as e:

lakesuperior/endpoints/main.py

@@ -2,9 +2,10 @@
 
 from os import path
 
-from flask import Blueprint, render_template
+from flask import Blueprint, jsonify, render_template
 
 from lakesuperior import release
+from lakesuperior.dictionaries import srv_mgd_terms as smt
 
 logger = logging.getLogger(__name__)
 
@@ -21,9 +22,11 @@ def index():
     return render_template('index.html', release=release)
 
 
-@main.route('/debug', methods=['GET'])
-def debug():
-    """Debug page."""
-    raise RuntimeError()
-
-
+@main.route('/info/ldp_constraints', methods=['GET'])
+def ldp_constraints():
+    """ LDP term constraints. """
+    return jsonify({
+        'srv_mgd_subjects': [*smt.srv_mgd_subjects],
+        'srv_mgd_predicates': [*smt.srv_mgd_predicates],
+        'srv_mgd_types': [*smt.srv_mgd_types],
+    })

lakesuperior/etc.defaults/application.yml

@@ -43,7 +43,7 @@ uuid:
     #
     #   ``blake2b`` is a strong, fast cryptographic alternative to SHA2/3:
     #   https://blake2.net/
-    algo: sha1
+    algo: blake2b
 
 ###
 # Data store configuration.
@@ -77,12 +77,6 @@ store:
         #   Changes to this parameter require a full migration.
         referential_integrity: lenient
 
-        ###
-        #   Split newly minted URIs into pairtrees.
-        #
-        #   This mimics Fedora4 behavior which segments an identifier on POST.
-        legacy_ptree_split: False
-
     ###
     #   The path used to persist LDP-NR (bitstreams).
     #

lakesuperior/exceptions.py

@@ -39,11 +39,17 @@ def __init__(self, uid, prov_cksum, calc_cksum, msg=None):
 
 
     def __str__(self):
-        return self.msg or (f'Validation failed for resource {self.uid}. '
-                            f'Provided checksum: {self.prov_cksum}; '
+        return self.msg or (f'validation failed for resource {self.uid}. '
+                            f'provided checksum: {self.prov_cksum}; '
                             f'calculated checksum: {self.calc_cksum}')
 
 
+class IndigestibleError(ResourceError):
+    """
+    Raised when an unsupported digest algorithm is requested.
+    """
+    pass
+
 
 class ResourceNotExistsError(ResourceError):
     '''