Update PR from my fork with the "builders" feature.

pyessv/__init__.py

@@ -11,7 +11,7 @@
 
 """
 __title__ = 'pyessv'
-__version__ = '0.7.0.0'
+__version__ = '0.7.0.1'
 __author__ = 'ES-DOC'
 __license__ = 'GPL'
 __copyright__ = 'Copyright 2017 ES-DOC'
@@ -62,18 +62,28 @@
 from pyessv._governance import reset
 
 from pyessv._initializer import init
+from pyessv._initializer import load_cv
 
 from pyessv._loader import load_random
 from pyessv._loader import load
+from pyessv._loader import all_scopes
 
 from pyessv._model import Authority
 from pyessv._model import Collection
 from pyessv._model import Scope
 from pyessv._model import Term
 
+from pyessv._builders import build_dataset_identifier
+from pyessv._builders import build_directory
+from pyessv._builders import build_filename
+
 from pyessv._parser import parse
-from pyessv._parsers import parse_dataset_identifer
-from pyessv._parsers import parse_dataset_identifers
+from pyessv._parsers import parse_dataset_identifier
+from pyessv._parsers import parse_dataset_identifiers
+from pyessv._parsers import parse_directory
+from pyessv._parsers import parse_directories
+from pyessv._parsers import parse_filename
+from pyessv._parsers import parse_filenames
 
 from pyessv._utils.logger import log
 from pyessv._utils.logger import log_error

pyessv/_builder_template.py

@@ -0,0 +1,91 @@
+# -*- coding: utf-8 -*-
+
+"""
+.. module:: pyessv._model._builder_template.py
+   :copyright: Copyright "December 01, 2016", IPSL
+   :license: GPL/CeCIL
+   :platform: Unix, Windows
+   :synopsis: A vocabulary constrained template builder, e.g. a dataset identifier.
+
+.. moduleauthor:: Mark Conway-Greenslade <[email protected]>
+
+
+"""
+from pyessv._model import Collection, Term
+from pyessv._exceptions import TemplateParsingError, TemplateValueError
+from pyessv._constants import BUILDER_FIELDS
+from pyessv._utils.compat import basestring, str
+
+
+class TemplateBuilder(object):
+    """A vocabulary template builder.
+
+    """
+    def __init__(self, template, collections, strictness, separator='.'):
+        """Instance constructor.
+
+        :param str template: Identifier template.
+        :param tuple collections: pyessv collection identifiers.
+        :param int strictness: Strictness level to apply when applying name matching rules.
+        :param str seprarator: Separator to apply when parsing.
+
+        """
+        from pyessv._loader import load
+
+        self.separator = separator
+        self.template_parts = template.split(separator)
+        self.template = template
+        self.strictness = strictness
+
+        # Inject pyessv collections into template.
+        collection_idx = 0
+        for idx, part in enumerate(self.template_parts):
+            if part == '{}':
+                self.template_parts[idx] = load(collections[collection_idx])
+                collection_idx += 1
+
+    def build(self, terms, att='label', alt_name=0):
+        """Build template instance from a list of pyessv terms.
+
+        :returns: Template instance string.
+
+        """
+        assert isinstance(alt_name, int), 'Invalid alternative name index'
+        assert att in BUILDER_FIELDS, 'Invalid name'
+
+        # Instantiate string parts.
+        string_parts = list()
+
+        # Iterate template.
+        for template_part in self.template_parts:
+
+            # Append constant match.
+            if isinstance(template_part, basestring):
+                string_parts.append(template_part)
+                continue
+
+            # Append term match.
+            collection = template_part
+            term = None
+            for term in terms:
+                if term.collection.namespace == collection.namespace:
+                    break
+
+            # Append term from associations.
+            if not term:
+                for term in [association for association in t.associations for t in terms]:
+                    if term.collection.namespace == collection.namespace:
+                        break
+
+            # Verify collection is found among terms.
+            if not term:
+                raise TemplateValueError('Collection not found among terms :: {}'.format(collection))
+
+            # Get term field.
+            if att == 'alternative_names':
+                string_parts.append(getattr(term, att)[alt_name])
+            else:
+                string_parts.append(getattr(term, att))
+
+
+        return self.separator.join(string_parts)

pyessv/_builders/__init__.py

@@ -0,0 +1,17 @@
+# -*- coding: utf-8 -*-
+
+"""
+.. module:: pyessv._builders.__init__.py
+   :copyright: Copyright "December 01, 2016", IPSL
+   :license: GPL/CeCIL
+   :platform: Unix, Windows
+   :synopsis: Expression builders.
+
+.. moduleauthor:: Mark Conway-Greenslade <[email protected]>
+
+
+"""
+
+from pyessv._builders.dataset_id import build_dataset_identifier
+from pyessv._builders.directory import build_directory
+from pyessv._builders.filename import build_filename

pyessv/_builders/dataset_id.py

@@ -0,0 +1,77 @@
+# -*- coding: utf-8 -*-
+
+"""
+.. module:: pyessv._builders.dataset_id.py
+   :copyright: Copyright "December 01, 2016", IPSL
+   :license: GPL/CeCIL
+   :platform: Unix, Windows
+   :synopsis: Encapsulates building of an ESGF dataset identifier.
+
+.. moduleauthor:: Mark Conway-Greenslade <[email protected]>
+
+"""
+from pyessv._model.term import Term
+from pyessv import all_scopes
+from pyessv._constants import PARSING_STRICTNESS_1
+from pyessv._factory import create_template_builder
+from pyessv._utils.compat import basestring
+
+# Instantiated template
+_TEMPLATE = None
+
+# Instantiated template collections
+_COLLECTIONS = None
+
+# Instantiated project.
+_PROJECT = None
+
+# Instantiated builder.
+_BUILDER = None
+
+
+def build_dataset_identifier(project, terms):
+    """Builds a dataset identifier.
+
+    :param str project: Project code.
+    :param set terms: Dataset identifier terms.
+
+    :returns: Dataset identifier.
+    :rtype: str
+
+    """
+    assert isinstance(project, basestring), 'Invalid project'
+    assert isinstance(terms, set), 'Invalid terms'
+
+    global _PROJECT, _BUILDER, _TEMPLATE, _COLLECTIONS
+
+    if _PROJECT != project:
+
+        # Get scope corresponding to the project code.
+        scopes = all_scopes()
+        assert project in [scope.name for scope in scopes], 'Unsupported project'
+        scope = [scope for scope in scopes if scope.name == project][0]
+
+        assert 'dataset_id' in scope.data.keys(), 'Dataset ID parser not found'
+        assert 'template' in scope.data['dataset_id'].keys(), 'Dataset ID parser template not found'
+        assert 'collections' in scope.data['dataset_id'].keys(), 'Dataset ID parser template collections not found'
+
+        # Get template from data scope.
+        _TEMPLATE = scope.data['dataset_id']['template']
+        assert isinstance(_TEMPLATE, basestring), 'Invalid template'
+
+        # Get template collections from data scope.
+        _COLLECTIONS = list()
+        for name in scope.data['dataset_id']['collections']:
+            _COLLECTIONS.append([collection.namespace for collection in scope.collections if collection.name == name.replace('_','-')][0])
+        assert _COLLECTIONS, 'Invalid collections'
+
+        # Instantiate parser JIT.
+        _BUILDER = create_template_builder(_TEMPLATE, tuple(_COLLECTIONS), PARSING_STRICTNESS_1)
+
+        # Cached project.
+        _PROJECT = project
+
+    for term in terms:
+        assert isinstance(term, Term), 'Invalid term :: {}'.format(term)
+
+    return _BUILDER.build(terms)

pyessv/_builders/directory.py

@@ -0,0 +1,77 @@
+# -*- coding: utf-8 -*-
+
+"""
+.. module:: pyessv._builders.dataset_id.py
+   :copyright: Copyright "December 01, 2016", IPSL
+   :license: GPL/CeCIL
+   :platform: Unix, Windows
+   :synopsis: Encapsulates building of an ESGF dataset identifier.
+
+.. moduleauthor:: Mark Conway-Greenslade <[email protected]>
+
+"""
+from pyessv._model.term import Term
+from pyessv import all_scopes
+from pyessv._constants import PARSING_STRICTNESS_1
+from pyessv._factory import create_template_builder
+from pyessv._utils.compat import basestring
+
+# Instantiated template
+_TEMPLATE = None
+
+# Instantiated template collections
+_COLLECTIONS = None
+
+# Instantiated project.
+_PROJECT = None
+
+# Instantiated builder.
+_BUILDER = None
+
+
+def build_directory(project, terms):
+    """Builds a directory.
+
+    :param str project: Project code.
+    :param set terms: Directory terms.
+
+    :returns: Directory string.
+    :rtype: str
+
+    """
+    assert isinstance(project, basestring), 'Invalid project'
+    assert isinstance(terms, set), 'Invalid terms'
+
+    global _PROJECT, _BUILDER, _TEMPLATE, _COLLECTIONS
+
+    if _PROJECT != project:
+
+        # Get scope corresponding to the project code.
+        scopes = all_scopes()
+        assert project in [scope.name for scope in scopes], 'Unsupported project'
+        scope = [scope for scope in scopes if scope.name == project][0]
+
+        assert 'directory_structure' in scope.data.keys(), 'Directory parser not found'
+        assert 'template' in scope.data['directory_structure'].keys(), 'Directory parser template not found'
+        assert 'collections' in scope.data['directory_structure'].keys(), 'Directory parser template collections not found'
+
+        # Get template from data scope.
+        _TEMPLATE = scope.data['directory_structure']['template']
+        assert isinstance(_TEMPLATE, basestring), 'Invalid template'
+
+        # Get template collections from data scope.
+        _COLLECTIONS = list()
+        for name in scope.data['directory_structure']['collections']:
+            _COLLECTIONS.append([collection.namespace for collection in scope.collections if collection.name == name.replace('_','-')][0])
+        assert _COLLECTIONS, 'Invalid collections'
+
+        # Instantiate parser JIT.
+        _BUILDER = create_template_builder(_TEMPLATE, tuple(_COLLECTIONS), PARSING_STRICTNESS_1, separator='/')
+
+        # Cached project.
+        _PROJECT = project
+
+    for term in terms:
+        assert isinstance(term, Term), 'Invalid term :: {}'.format(term)
+
+    return _BUILDER.build(terms)

pyessv/_builders/filename.py

@@ -0,0 +1,77 @@
+# -*- coding: utf-8 -*-
+
+"""
+.. module:: pyessv._builders.dataset_id.py
+   :copyright: Copyright "December 01, 2016", IPSL
+   :license: GPL/CeCIL
+   :platform: Unix, Windows
+   :synopsis: Encapsulates building of an ESGF dataset identifier.
+
+.. moduleauthor:: Mark Conway-Greenslade <[email protected]>
+
+"""
+from pyessv._model.term import Term
+from pyessv import all_scopes
+from pyessv._constants import PARSING_STRICTNESS_1
+from pyessv._factory import create_template_builder
+from pyessv._utils.compat import basestring
+
+# Instantiated template
+_TEMPLATE = None
+
+# Instantiated template collections
+_COLLECTIONS = None
+
+# Instantiated project.
+_PROJECT = None
+
+# Instantiated builder.
+_BUILDER = None
+
+
+def build_filename(project, terms):
+    """Builds a filename.
+
+    :param str project: Project code.
+    :param set terms: Filename terms.
+
+    :returns: Filename string.
+    :rtype: str
+
+    """
+    assert isinstance(project, basestring), 'Invalid project'
+    assert isinstance(terms, set), 'Invalid terms'
+
+    global _PROJECT, _BUILDER, _TEMPLATE, _COLLECTIONS
+
+    if _PROJECT != project:
+
+        # Get scope corresponding to the project code.
+        scopes = all_scopes()
+        assert project in [scope.name for scope in scopes], 'Unsupported project'
+        scope = [scope for scope in scopes if scope.name == project][0]
+
+        assert 'filename' in scope.data.keys(), 'Filename parser not found'
+        assert 'template' in scope.data['filename'].keys(), 'Filename parser template not found'
+        assert 'collections' in scope.data['filename'].keys(), 'Filename parser template collections not found'
+
+        # Get template from data scope.
+        _TEMPLATE = scope.data['filename']['template']
+        assert isinstance(_TEMPLATE, basestring), 'Invalid template'
+
+        # Get template collections from data scope.
+        _COLLECTIONS = list()
+        for name in scope.data['filename']['collections']:
+            _COLLECTIONS.append([collection.namespace for collection in scope.collections if collection.name == name.replace('_','-')][0])
+        assert _COLLECTIONS, 'Invalid collections'
+
+        # Instantiate parser JIT.
+        _BUILDER = create_template_builder(_TEMPLATE, tuple(_COLLECTIONS), PARSING_STRICTNESS_1, separator='_')
+
+        # Cached project.
+        _PROJECT = project
+
+    for term in terms:
+        assert isinstance(term, Term), 'Invalid term :: {}'.format(term)
+
+    return _BUILDER.build(terms)

pyessv/_codecs/json_codec/encoder.py

@@ -20,7 +20,7 @@
 from pyessv._utils import convert
 from pyessv._utils.compat import json
 from pyessv._utils.compat import numeric_types
-from pyessv._utils.compat import str
+from pyessv._utils.compat import str, basestring