def get_parsed_content(metadata_content):
"""
Parses any of the following types of content:
1. XML string or file object: parses XML content
2. MetadataParser instance: deep copies xml_tree
3. Dictionary with nested objects containing:
- name (required): the name of the element tag
- text: the text contained by element
- tail: text immediately following the element
- attributes: a Dictionary containing element attributes
- children: a List of converted child elements
:raises InvalidContent: if the XML is invalid or does not conform to a supported metadata standard
:raises NoContent: If the content passed in is null or otherwise empty
:return: the XML root along with an XML Tree parsed by and compatible with element_utils
"""
_import_parsers() # Prevents circular dependencies between modules
xml_tree = None
if metadata_content is None:
raise NoContent('Metadata has no data')
else:
if isinstance(metadata_content, MetadataParser):
xml_tree = deepcopy(metadata_content._xml_tree)
elif isinstance(metadata_content, dict):
xml_tree = get_element_tree(metadata_content)
else:
try:
# Strip name spaces from file or XML content
xml_tree = get_element_tree(metadata_content)
except Exception:
xml_tree = None # Several exceptions possible, outcome is the same
if xml_tree is None:
raise InvalidContent(
'Cannot instantiate a {parser_type} parser with invalid content to parse',
parser_type=type(metadata_content).__name__)
xml_root = get_element_name(xml_tree)
if xml_root is None:
raise NoContent('Metadata contains no data')
elif xml_root not in VALID_ROOTS:
content = type(metadata_content).__name__
raise InvalidContent('Invalid root element for {content}: {xml_root}',
content=content,
xml_root=xml_root)
return xml_root, xml_tree
def _init_data_map(self):
""" OVERRIDDEN: Initialize required ISO-19115 data map with XPATHS and specialized functions """
if self._data_map is not None:
return # Initiation happens once
# Parse and validate the ISO metadata root
if self._xml_tree is None:
iso_root = ISO_ROOTS[0]
else:
iso_root = get_element_name(self._xml_tree)
if iso_root not in ISO_ROOTS:
raise InvalidContent('Invalid XML root for ISO-19115 standard: {root}', root=iso_root)
iso_data_map = {'_root': iso_root}
iso_data_map.update(_iso_tag_roots)
iso_data_map.update(_iso_tag_formats)
iso_data_structures = {}
# Capture and format complex XPATHs
ad_format = iso_data_map[ATTRIBUTES]
ft_source = iso_data_map['_attr_src'].replace('/carrierOfCharacteristics/FC_FeatureAttribute', '')
iso_data_structures[ATTRIBUTES] = format_xpaths(
_iso_definitions[ATTRIBUTES],
label=ad_format.format(ad_path='memberName/LocalName'),
aliases=ad_format.format(ad_path='aliases/LocalName'), # Not in spec
definition=ad_format.format(ad_path='definition/CharacterString'),
# First try to populate attribute definition source from FC_FeatureAttribute
definition_src=iso_data_map['_attr_src'] + '/organisationName/CharacterString',
_definition_src=iso_data_map['_attr_src'] + '/individualName/CharacterString',
# Then assume feature type source is the same as attribute: populate from FC_FeatureType
__definition_src=ft_source + '/organisationName/CharacterString',
___definition_src=ft_source + '/individualName/CharacterString'
)
bb_format = iso_data_map[BOUNDING_BOX]
iso_data_structures[BOUNDING_BOX] = format_xpaths(
_iso_definitions[BOUNDING_BOX],
east=bb_format.format(bbox_path='eastBoundLongitude/Decimal'),
south=bb_format.format(bbox_path='southBoundLatitude/Decimal'),
west=bb_format.format(bbox_path='westBoundLongitude/Decimal'),
north=bb_format.format(bbox_path='northBoundLatitude/Decimal')
)
ct_format = iso_data_map[CONTACTS]
iso_data_structures[CONTACTS] = format_xpaths(
_iso_definitions[CONTACTS],
name=ct_format.format(ct_path='individualName/CharacterString'),
organization=ct_format.format(ct_path='organisationName/CharacterString'),
position=ct_format.format(ct_path='positionName/CharacterString'),
email=ct_format.format(
ct_path='contactInfo/CI_Contact/address/CI_Address/electronicMailAddress/CharacterString'
)
)
dt_format = iso_data_map[DATES]
iso_data_structures[DATES] = {
DATE_TYPE_MULTIPLE: dt_format.format(type_path='TimeInstant/timePosition'),
DATE_TYPE_RANGE_BEGIN: dt_format.format(type_path='TimePeriod/begin/TimeInstant/timePosition'),
DATE_TYPE_RANGE_END: dt_format.format(type_path='TimePeriod/end/TimeInstant/timePosition'),
DATE_TYPE_SINGLE: dt_format.format(type_path='TimeInstant/timePosition') # Same as multiple
}
iso_data_structures[DATES][DATE_TYPE_RANGE] = [
iso_data_structures[DATES][DATE_TYPE_RANGE_BEGIN],
iso_data_structures[DATES][DATE_TYPE_RANGE_END]
]
df_format = iso_data_map[DIGITAL_FORMS]
iso_data_structures[DIGITAL_FORMS] = format_xpaths(
_iso_definitions[DIGITAL_FORMS],
name=df_format.format(df_path='name/CharacterString'),
content='', # Not supported in ISO-19115 (appending to spec)
decompression=df_format.format(df_path='fileDecompressionTechnique/CharacterString'),
version=df_format.format(df_path='version/CharacterString'),
specification=df_format.format(df_path='specification/CharacterString'),
access_desc=iso_data_map['_access_desc'],
access_instrs=iso_data_map['_access_instrs'],
network_resource=iso_data_map['_network_resource']
)
keywords_structure = {
'keyword_root': 'MD_Keywords/keyword',
'keyword_type': 'MD_Keywords/type/MD_KeywordTypeCode',
'keyword': 'MD_Keywords/keyword/CharacterString'
}
for keyword_prop in KEYWORD_PROPS:
iso_data_structures[keyword_prop] = deepcopy(keywords_structure)
lw_format = iso_data_map[LARGER_WORKS]
iso_data_structures[LARGER_WORKS] = format_xpaths(
_iso_definitions[LARGER_WORKS],
title=lw_format.format(lw_path='title/CharacterString'),
edition=lw_format.format(lw_path='edition/CharacterString'),
origin=iso_data_map['_lw_citation'].format(lw_path='individualName/CharacterString'),
online_linkage=iso_data_map['_lw_linkage'].format(lw_path='linkage/URL'),
other_citation=lw_format.format(lw_path='otherCitationDetails/CharacterString'),
date=lw_format.format(lw_path='editionDate/Date'),
place=iso_data_map['_lw_contact'].format(lw_path='address/CI_Address/city/CharacterString'),
info=iso_data_map['_lw_citation'].format(lw_path='organisationName/CharacterString')
)
ps_format = iso_data_map[PROCESS_STEPS]
iso_data_structures[PROCESS_STEPS] = format_xpaths(
_iso_definitions[PROCESS_STEPS],
description=ps_format.format(ps_path='description/CharacterString'),
date=ps_format.format(ps_path='dateTime/DateTime'),
sources=ps_format.format(
ps_path='source/LI_Source/sourceCitation/CI_Citation/alternateTitle/CharacterString'
)
)
ri_format = iso_data_map[RASTER_INFO]
iso_data_structures[RASTER_INFO] = format_xpaths(
_iso_definitions[RASTER_DIMS],
type=ri_format.format(ri_path='dimensionName/MD_DimensionNameTypeCode'),
_type=ri_format.format(ri_path='dimensionName/MD_DimensionNameTypeCode/@codeListValue'),
size=ri_format.format(ri_path='dimensionSize/Integer'),
value=ri_format.format(ri_path='resolution/Measure'),
units=ri_format.format(ri_path='resolution/Measure/@uom')
)
# Assign XPATHS and gis_metadata.utils.ParserProperties to data map
for prop, xpath in iteritems(dict(iso_data_map)):
if prop == ATTRIBUTES:
iso_data_map[prop] = ParserProperty(self._parse_attribute_details, self._update_attribute_details)
elif prop in (CONTACTS, PROCESS_STEPS):
iso_data_map[prop] = ParserProperty(self._parse_complex_list, self._update_complex_list)
elif prop in (BOUNDING_BOX, LARGER_WORKS):
iso_data_map[prop] = ParserProperty(self._parse_complex, self._update_complex)
elif prop == DATES:
iso_data_map[prop] = ParserProperty(self._parse_dates, self._update_dates)
elif prop == DIGITAL_FORMS:
iso_data_map[prop] = ParserProperty(self._parse_digital_forms, self._update_digital_forms)
elif prop in KEYWORD_PROPS:
iso_data_map[prop] = ParserProperty(self._parse_keywords, self._update_keywords)
elif prop == RASTER_INFO:
iso_data_map[prop] = ParserProperty(self._parse_raster_info, self._update_raster_info)
else:
iso_data_map[prop] = xpath
self._data_map = iso_data_map
self._data_structures = iso_data_structures
def _init_data_map(self):
""" OVERRIDDEN: Initialize required FGDC data map with XPATHS and specialized functions """
if self._data_map is not None:
return # Initiation happens once
# Parse and validate the FGDC metadata root
if self._xml_tree is None:
fgdc_root = FGDC_ROOT
else:
fgdc_root = get_element_name(self._xml_tree)
if fgdc_root != FGDC_ROOT:
raise InvalidContent('Invalid XML root for ISO-19115 standard: {root}', root=fgdc_root)
fgdc_data_map = {'_root': FGDC_ROOT}
fgdc_data_structures = {}
# Capture and format other complex XPATHs
ad_format = _fgdc_tag_formats[ATTRIBUTES]
fgdc_data_structures[ATTRIBUTES] = format_xpaths(
_fgdc_definitions[ATTRIBUTES],
label=ad_format.format(ad_path='attrlabl'),
aliases=ad_format.format(ad_path='attalias'),
definition=ad_format.format(ad_path='attrdef'),
definition_src=ad_format.format(ad_path='attrdefs')
)
bb_format = _fgdc_tag_formats[BOUNDING_BOX]
fgdc_data_structures[BOUNDING_BOX] = format_xpaths(
_fgdc_definitions[BOUNDING_BOX],
east=bb_format.format(bbox_path='eastbc'),
south=bb_format.format(bbox_path='southbc'),
west=bb_format.format(bbox_path='westbc'),
north=bb_format.format(bbox_path='northbc')
)
ct_format = _fgdc_tag_formats[CONTACTS]
fgdc_data_structures[CONTACTS] = format_xpaths(
_fgdc_definitions[CONTACTS],
name=ct_format.format(ct_path='cntperp/cntper'),
_name=ct_format.format(ct_path='cntorgp/cntper'), # If not in cntperp
organization=ct_format.format(ct_path='cntperp/cntorg'),
_organization=ct_format.format(ct_path='cntorgp/cntorg'), # If not in cntperp
position=ct_format.format(ct_path='cntpos'),
email=ct_format.format(ct_path='cntemail')
)
dt_format = _fgdc_tag_formats[DATES]
fgdc_data_structures[DATES] = {
DATE_TYPE_MULTIPLE: dt_format.format(type_path='mdattim/sngdate/caldate'),
DATE_TYPE_RANGE_BEGIN: dt_format.format(type_path='rngdates/begdate'),
DATE_TYPE_RANGE_END: dt_format.format(type_path='rngdates/enddate'),
DATE_TYPE_SINGLE: dt_format.format(type_path='sngdate/caldate')
}
fgdc_data_structures[DATES][DATE_TYPE_RANGE] = [
fgdc_data_structures[DATES][DATE_TYPE_RANGE_BEGIN],
fgdc_data_structures[DATES][DATE_TYPE_RANGE_END]
]
df_format = _fgdc_tag_formats[DIGITAL_FORMS]
fgdc_data_structures[DIGITAL_FORMS] = format_xpaths(
_fgdc_definitions[DIGITAL_FORMS],
name=df_format.format(df_path='digtinfo/formname'),
content=df_format.format(df_path='digtinfo/formcont'),
decompression=df_format.format(df_path='digtinfo/filedec'),
version=df_format.format(df_path='digtinfo/formvern'),
specification=df_format.format(df_path='digtinfo/formspec'),
access_desc=df_format.format(df_path='digtopt/onlinopt/oncomp'),
access_instrs=df_format.format(df_path='digtopt/onlinopt/accinstr'),
network_resource=df_format.format(df_path='digtopt/onlinopt/computer/networka/networkr')
)
lw_format = _fgdc_tag_formats[LARGER_WORKS]
fgdc_data_structures[LARGER_WORKS] = format_xpaths(
_fgdc_definitions[LARGER_WORKS],
title=lw_format.format(lw_path='title'),
edition=lw_format.format(lw_path='edition'),
origin=lw_format.format(lw_path='origin'),
online_linkage=lw_format.format(lw_path='onlink'),
other_citation=lw_format.format(lw_path='othercit'),
date=lw_format.format(lw_path='pubdate'),
place=lw_format.format(lw_path='pubinfo/pubplace'),
info=lw_format.format(lw_path='pubinfo/publish')
)
ps_format = _fgdc_tag_formats[PROCESS_STEPS]
fgdc_data_structures[PROCESS_STEPS] = format_xpaths(
_fgdc_definitions[PROCESS_STEPS],
description=ps_format.format(ps_path='procdesc'),
date=ps_format.format(ps_path='procdate'),
sources=ps_format.format(ps_path='srcused')
)
# Assign XPATHS and gis_metadata.utils.ParserProperties to fgdc_data_map
fgdc_data_formats = dict(_fgdc_tag_formats)
for prop, xpath in iteritems(fgdc_data_formats):
if prop in (ATTRIBUTES, CONTACTS, DIGITAL_FORMS, PROCESS_STEPS):
fgdc_data_map[prop] = ParserProperty(self._parse_complex_list, self._update_complex_list)
elif prop in (BOUNDING_BOX, LARGER_WORKS):
fgdc_data_map[prop] = ParserProperty(self._parse_complex, self._update_complex)
elif prop == DATES:
fgdc_data_map[prop] = ParserProperty(self._parse_dates, self._update_dates)
else:
fgdc_data_map[prop] = xpath
self._data_map = fgdc_data_map
self._data_structures = fgdc_data_structures
def _init_data_map(self):
""" OVERRIDDEN: Initialize required ISO-19115 data map with XPATHS and specialized functions """
if self._data_map is not None:
return # Initiation happens once
# Parse and validate the ISO metadata root
if self._xml_tree is None:
iso_root = ISO_ROOTS[0]
else:
iso_root = get_element_name(self._xml_tree)
if iso_root not in ISO_ROOTS:
raise InvalidContent('Invalid XML root for ISO-19115 standard: {root}', root=iso_root)
iso_data_map = {'_root': iso_root}
iso_data_map.update(_iso_tag_roots)
iso_data_map.update(_iso_tag_formats)
iso_data_structures = {}
# Capture and format complex XPATHs
ad_format = iso_data_map[ATTRIBUTES]
ft_source = iso_data_map['_attr_src'].replace('/carrierOfCharacteristics/FC_FeatureAttribute', '')
iso_data_structures[ATTRIBUTES] = format_xpaths(
_iso_definitions[ATTRIBUTES],
label=ad_format.format(ad_path='memberName/LocalName'),
aliases=ad_format.format(ad_path='aliases/LocalName'), # Not in spec
definition=ad_format.format(ad_path='definition/CharacterString'),
# First try to populate attribute definition source from FC_FeatureAttribute
definition_src=iso_data_map['_attr_src'] + '/organisationName/CharacterString',
_definition_src=iso_data_map['_attr_src'] + '/individualName/CharacterString',
# Then assume feature type source is the same as attribute: populate from FC_FeatureType
__definition_src=ft_source + '/organisationName/CharacterString',
___definition_src=ft_source + '/individualName/CharacterString'
)
bb_format = iso_data_map[BOUNDING_BOX]
iso_data_structures[BOUNDING_BOX] = format_xpaths(
_iso_definitions[BOUNDING_BOX],
east=bb_format.format(bbox_path='eastBoundLongitude/Decimal'),
south=bb_format.format(bbox_path='southBoundLatitude/Decimal'),
west=bb_format.format(bbox_path='westBoundLongitude/Decimal'),
north=bb_format.format(bbox_path='northBoundLatitude/Decimal')
)
ct_format = iso_data_map[CONTACTS]
iso_data_structures[CONTACTS] = format_xpaths(
_iso_definitions[CONTACTS],
name=ct_format.format(ct_path='individualName/CharacterString'),
organization=ct_format.format(ct_path='organisationName/CharacterString'),
position=ct_format.format(ct_path='positionName/CharacterString'),
email=ct_format.format(
ct_path='contactInfo/CI_Contact/address/CI_Address/electronicMailAddress/CharacterString'
)
)
dt_format = iso_data_map[DATES]
iso_data_structures[DATES] = {
DATE_TYPE_MULTIPLE: dt_format.format(type_path='TimeInstant/timePosition'),
DATE_TYPE_RANGE_BEGIN: dt_format.format(type_path='TimePeriod/begin/TimeInstant/timePosition'),
DATE_TYPE_RANGE_END: dt_format.format(type_path='TimePeriod/end/TimeInstant/timePosition'),
DATE_TYPE_SINGLE: dt_format.format(type_path='TimeInstant/timePosition') # Same as multiple
}
iso_data_structures[DATES][DATE_TYPE_RANGE] = [
iso_data_structures[DATES][DATE_TYPE_RANGE_BEGIN],
iso_data_structures[DATES][DATE_TYPE_RANGE_END]
]
df_format = iso_data_map[DIGITAL_FORMS]
iso_data_structures[DIGITAL_FORMS] = format_xpaths(
_iso_definitions[DIGITAL_FORMS],
name=df_format.format(df_path='name/CharacterString'),
content='', # Not supported in ISO-19115 (appending to spec)
decompression=df_format.format(df_path='fileDecompressionTechnique/CharacterString'),
version=df_format.format(df_path='version/CharacterString'),
specification=df_format.format(df_path='specification/CharacterString'),
access_desc=iso_data_map['_access_desc'],
access_instrs=iso_data_map['_access_instrs'],
network_resource=iso_data_map['_network_resource']
)
keywords_structure = {
'keyword_root': 'MD_Keywords/keyword',
'keyword_type': 'MD_Keywords/type/MD_KeywordTypeCode',
'keyword': 'MD_Keywords/keyword/CharacterString'
}
iso_data_structures[KEYWORDS_PLACE] = keywords_structure
iso_data_structures[KEYWORDS_THEME] = keywords_structure
lw_format = iso_data_map[LARGER_WORKS]
iso_data_structures[LARGER_WORKS] = format_xpaths(
_iso_definitions[LARGER_WORKS],
title=lw_format.format(lw_path='title/CharacterString'),
edition=lw_format.format(lw_path='edition/CharacterString'),
origin=iso_data_map['_lw_citation'].format(lw_path='individualName/CharacterString'),
online_linkage=iso_data_map['_lw_linkage'].format(lw_path='linkage/URL'),
other_citation=lw_format.format(lw_path='otherCitationDetails/CharacterString'),
date=lw_format.format(lw_path='editionDate/Date'),
place=iso_data_map['_lw_contact'].format(lw_path='address/CI_Address/city/CharacterString'),
info=iso_data_map['_lw_citation'].format(lw_path='organisationName/CharacterString')
)
ps_format = iso_data_map[PROCESS_STEPS]
iso_data_structures[PROCESS_STEPS] = format_xpaths(
_iso_definitions[PROCESS_STEPS],
description=ps_format.format(ps_path='description/CharacterString'),
date=ps_format.format(ps_path='dateTime/DateTime'),
sources=ps_format.format(
ps_path='source/LI_Source/sourceCitation/CI_Citation/alternateTitle/CharacterString'
)
)
# Assign XPATHS and gis_metadata.utils.ParserProperties to data map
for prop, xpath in iteritems(dict(iso_data_map)):
if prop == ATTRIBUTES:
iso_data_map[prop] = ParserProperty(self._parse_attribute_details, self._update_attribute_details)
elif prop in (CONTACTS, PROCESS_STEPS):
iso_data_map[prop] = ParserProperty(self._parse_complex_list, self._update_complex_list)
elif prop in (BOUNDING_BOX, LARGER_WORKS):
iso_data_map[prop] = ParserProperty(self._parse_complex, self._update_complex)
elif prop == DATES:
iso_data_map[prop] = ParserProperty(self._parse_dates, self._update_dates)
elif prop == DIGITAL_FORMS:
iso_data_map[prop] = ParserProperty(self._parse_digital_forms, self._update_digital_forms)
elif prop in [KEYWORDS_PLACE, KEYWORDS_THEME]:
iso_data_map[prop] = ParserProperty(self._parse_keywords, self._update_keywords)
else:
iso_data_map[prop] = xpath
self._data_map = iso_data_map
self._data_structures = iso_data_structures
def _init_data_map(self):
""" OVERRIDDEN: Initialize required FGDC data map with XPATHS and specialized functions """
if self._data_map is not None:
return # Initiation happens once
# Parse and validate the ArcGIS metadata root
if self._xml_tree is None:
agis_root = ARCGIS_ROOTS[0] # Default to uncapitalized
else:
agis_root = get_element_name(self._xml_tree)
if agis_root not in ARCGIS_ROOTS:
raise InvalidContent(
'Invalid XML root for ArcGIS metadata: {root}', root=agis_root)
agis_data_map = {'_root': agis_root}
agis_data_map.update(_agis_tag_formats)
agis_data_structures = {}
# Capture and format complex XPATHs
ad_format = agis_data_map[ATTRIBUTES]
agis_data_structures[ATTRIBUTES] = format_xpaths(
_agis_definitions[ATTRIBUTES],
label=ad_format.format(ad_path='attrlabl'),
aliases=ad_format.format(ad_path='attalias'),
definition=ad_format.format(ad_path='attrdef'),
definition_src=ad_format.format(ad_path='attrdefs'))
bb_format = agis_data_map[BOUNDING_BOX]
agis_data_structures[BOUNDING_BOX] = format_xpaths(
_agis_definitions[BOUNDING_BOX],
east=bb_format.format(bbox_path='eastBL'),
south=bb_format.format(bbox_path='southBL'),
west=bb_format.format(bbox_path='westBL'),
north=bb_format.format(bbox_path='northBL'))
ct_format = agis_data_map[CONTACTS]
agis_data_structures[CONTACTS] = format_xpaths(
_agis_definitions[CONTACTS],
name=ct_format.format(ct_path='rpIndName'),
organization=ct_format.format(ct_path='rpOrgName'),
position=ct_format.format(ct_path='rpPosName'),
email=ct_format.format(ct_path='rpCntInfo/cntAddress/eMailAdd'))
dt_format = agis_data_map[DATES]
agis_data_structures[DATES] = {
DATE_TYPE_MULTIPLE:
dt_format.format(type_path='TM_Instant/tmPosition'),
'_' + DATE_TYPE_MULTIPLE:
dt_format.format(type_path='TM_Instant/tmPosition/@date'),
DATE_TYPE_RANGE_BEGIN:
dt_format.format(type_path='TM_Period/tmBegin'),
'_' + DATE_TYPE_RANGE_BEGIN:
dt_format.format(type_path='TM_Period/tmBegin/@date'),
DATE_TYPE_RANGE_END:
dt_format.format(type_path='TM_Period/tmEnd'),
'_' + DATE_TYPE_RANGE_END:
dt_format.format(type_path='TM_Period/tmEnd/@date'),
# Same as multiple dates, but will contain only one
DATE_TYPE_SINGLE:
dt_format.format(type_path='TM_Instant/tmPosition'),
'_' + DATE_TYPE_SINGLE:
dt_format.format(type_path='TM_Instant/tmPosition/@date')
}
agis_data_structures[DATES][DATE_TYPE_RANGE] = [
agis_data_structures[DATES][DATE_TYPE_RANGE_BEGIN],
agis_data_structures[DATES][DATE_TYPE_RANGE_END]
]
agis_data_structures[DATES]['_' + DATE_TYPE_RANGE] = [
agis_data_structures[DATES]['_' + DATE_TYPE_RANGE_BEGIN],
agis_data_structures[DATES]['_' + DATE_TYPE_RANGE_END]
]
df_format = agis_data_map[DIGITAL_FORMS]
agis_data_structures[DIGITAL_FORMS] = format_xpaths(
_agis_definitions[DIGITAL_FORMS],
name=df_format.format(df_path='formatName'),
content=df_format.format(df_path='formatInfo'),
decompression=df_format.format(df_path='fileDecmTech'),
version=df_format.format(df_path='formatVer'),
specification=df_format.format(df_path='formatSpec'),
access_desc=agis_data_map['_access_desc'],
access_instrs=agis_data_map['_access_instrs'],
network_resource=agis_data_map['_network_resource'])
lw_format = agis_data_map[LARGER_WORKS]
agis_data_structures[LARGER_WORKS] = format_xpaths(
_agis_definitions[LARGER_WORKS],
title=lw_format.format(lw_path='resTitle'),
edition=lw_format.format(lw_path='resEd'),
origin=lw_format.format(lw_path='citRespParty/rpIndName'),
online_linkage=lw_format.format(
lw_path='citRespParty/rpCntInfo/cntOnlineRes/linkage'),
other_citation=lw_format.format(lw_path='otherCitDet'),
date=lw_format.format(lw_path='date/pubDate'),
place=lw_format.format(
lw_path='citRespParty/rpCntInfo/cntAddress/city'),
info=lw_format.format(lw_path='citRespParty/rpOrgName'))
ps_format = agis_data_map[PROCESS_STEPS]
agis_data_structures[PROCESS_STEPS] = format_xpaths(
_agis_definitions[PROCESS_STEPS],
description=ps_format.format(ps_path='stepDesc'),
date=ps_format.format(ps_path='stepDateTm'),
sources=ps_format.format(ps_path='stepSrc/srcDesc'))
ri_format = agis_data_map[RASTER_INFO]
agis_data_structures[RASTER_INFO] = format_xpaths(
_agis_definitions[RASTER_DIMS],
type=ri_format.format(ri_path='@type'),
size=ri_format.format(ri_path='dimSize'),
value=ri_format.format(ri_path='dimResol/value'),
units=ri_format.format(ri_path='dimResol/value/@uom'))
# Assign XPATHS and gis_metadata.utils.ParserProperties to data map
for prop, xpath in iteritems(dict(agis_data_map)):
if prop in (ATTRIBUTES, CONTACTS, PROCESS_STEPS):
agis_data_map[prop] = ParserProperty(self._parse_complex_list,
self._update_complex_list)
elif prop in (BOUNDING_BOX, LARGER_WORKS):
agis_data_map[prop] = ParserProperty(self._parse_complex,
self._update_complex)
elif prop in ('attribute_accuracy', 'dataset_completeness'):
agis_data_map[prop] = ParserProperty(self._parse_report_item,
self._update_report_item)
elif prop == DATES:
agis_data_map[prop] = ParserProperty(self._parse_dates,
self._update_dates)
elif prop == DIGITAL_FORMS:
agis_data_map[prop] = ParserProperty(
self._parse_digital_forms, self._update_digital_forms)
elif prop == RASTER_INFO:
agis_data_map[prop] = ParserProperty(self._parse_raster_info,
self._update_raster_info)
else:
agis_data_map[prop] = xpath
self._data_map = agis_data_map
self._data_structures = agis_data_structures
def _init_data_map(self):
""" OVERRIDDEN: Initialize required FGDC data map with XPATHS and specialized functions """
if self._data_map is not None:
return # Initiation happens once
# Parse and validate the ArcGIS metadata root
if self._xml_tree is None:
agis_root = ARCGIS_ROOTS[0] # Default to uncapitalized
else:
agis_root = get_element_name(self._xml_tree)
if agis_root not in ARCGIS_ROOTS:
raise InvalidContent('Invalid XML root for ArcGIS metadata: {root}', root=agis_root)
agis_data_map = {'_root': agis_root}
agis_data_map.update(_agis_tag_formats)
agis_data_structures = {}
# Capture and format complex XPATHs
ad_format = agis_data_map[ATTRIBUTES]
agis_data_structures[ATTRIBUTES] = format_xpaths(
_agis_definitions[ATTRIBUTES],
label=ad_format.format(ad_path='attrlabl'),
aliases=ad_format.format(ad_path='attalias'),
definition=ad_format.format(ad_path='attrdef'),
definition_src=ad_format.format(ad_path='attrdefs')
)
bb_format = agis_data_map[BOUNDING_BOX]
agis_data_structures[BOUNDING_BOX] = format_xpaths(
_agis_definitions[BOUNDING_BOX],
east=bb_format.format(bbox_path='eastBL'),
south=bb_format.format(bbox_path='southBL'),
west=bb_format.format(bbox_path='westBL'),
north=bb_format.format(bbox_path='northBL')
)
ct_format = agis_data_map[CONTACTS]
agis_data_structures[CONTACTS] = format_xpaths(
_agis_definitions[CONTACTS],
name=ct_format.format(ct_path='rpIndName'),
organization=ct_format.format(ct_path='rpOrgName'),
position=ct_format.format(ct_path='rpPosName'),
email=ct_format.format(ct_path='rpCntInfo/cntAddress/eMailAdd')
)
dt_format = agis_data_map[DATES]
agis_data_structures[DATES] = {
DATE_TYPE_MULTIPLE: dt_format.format(type_path='TM_Instant/tmPosition'),
'_' + DATE_TYPE_MULTIPLE: dt_format.format(type_path='TM_Instant/tmPosition/@date'),
DATE_TYPE_RANGE_BEGIN: dt_format.format(type_path='TM_Period/tmBegin'),
'_' + DATE_TYPE_RANGE_BEGIN: dt_format.format(type_path='TM_Period/tmBegin/@date'),
DATE_TYPE_RANGE_END: dt_format.format(type_path='TM_Period/tmEnd'),
'_' + DATE_TYPE_RANGE_END: dt_format.format(type_path='TM_Period/tmEnd/@date'),
# Same as multiple dates, but will contain only one
DATE_TYPE_SINGLE: dt_format.format(type_path='TM_Instant/tmPosition'),
'_' + DATE_TYPE_SINGLE: dt_format.format(type_path='TM_Instant/tmPosition/@date')
}
agis_data_structures[DATES][DATE_TYPE_RANGE] = [
agis_data_structures[DATES][DATE_TYPE_RANGE_BEGIN],
agis_data_structures[DATES][DATE_TYPE_RANGE_END]
]
agis_data_structures[DATES]['_' + DATE_TYPE_RANGE] = [
agis_data_structures[DATES]['_' + DATE_TYPE_RANGE_BEGIN],
agis_data_structures[DATES]['_' + DATE_TYPE_RANGE_END]
]
df_format = agis_data_map[DIGITAL_FORMS]
agis_data_structures[DIGITAL_FORMS] = format_xpaths(
_agis_definitions[DIGITAL_FORMS],
name=df_format.format(df_path='formatName'),
content=df_format.format(df_path='formatInfo'),
decompression=df_format.format(df_path='fileDecmTech'),
version=df_format.format(df_path='formatVer'),
specification=df_format.format(df_path='formatSpec'),
access_desc=agis_data_map['_access_desc'],
access_instrs=agis_data_map['_access_instrs'],
network_resource=agis_data_map['_network_resource']
)
lw_format = agis_data_map[LARGER_WORKS]
agis_data_structures[LARGER_WORKS] = format_xpaths(
_agis_definitions[LARGER_WORKS],
title=lw_format.format(lw_path='resTitle'),
edition=lw_format.format(lw_path='resEd'),
origin=lw_format.format(lw_path='citRespParty/rpIndName'),
online_linkage=lw_format.format(lw_path='citRespParty/rpCntInfo/cntOnlineRes/linkage'),
other_citation=lw_format.format(lw_path='otherCitDet'),
date=lw_format.format(lw_path='date/pubDate'),
place=lw_format.format(lw_path='citRespParty/rpCntInfo/cntAddress/city'),
info=lw_format.format(lw_path='citRespParty/rpOrgName')
)
ps_format = agis_data_map[PROCESS_STEPS]
agis_data_structures[PROCESS_STEPS] = format_xpaths(
_agis_definitions[PROCESS_STEPS],
description=ps_format.format(ps_path='stepDesc'),
date=ps_format.format(ps_path='stepDateTm'),
sources=ps_format.format(ps_path='stepSrc/srcDesc')
)
# Assign XPATHS and gis_metadata.utils.ParserProperties to data map
for prop, xpath in iteritems(dict(agis_data_map)):
if prop in (ATTRIBUTES, CONTACTS, PROCESS_STEPS):
agis_data_map[prop] = ParserProperty(self._parse_complex_list, self._update_complex_list)
elif prop in (BOUNDING_BOX, LARGER_WORKS):
agis_data_map[prop] = ParserProperty(self._parse_complex, self._update_complex)
elif prop in ('attribute_accuracy', 'dataset_completeness'):
agis_data_map[prop] = ParserProperty(self._parse_report_item, self._update_report_item)
elif prop == DATES:
agis_data_map[prop] = ParserProperty(self._parse_dates, self._update_dates)
elif prop == DIGITAL_FORMS:
agis_data_map[prop] = ParserProperty(self._parse_digital_forms, self._update_digital_forms)
else:
agis_data_map[prop] = xpath
self._data_map = agis_data_map
self._data_structures = agis_data_structures
def _init_data_map(self):
""" OVERRIDDEN: Initialize required FGDC data map with XPATHS and specialized functions """
if self._data_map is not None:
return # Initiation happens once
# Parse and validate the FGDC metadata root
if self._xml_tree is None:
fgdc_root = FGDC_ROOT
else:
fgdc_root = get_element_name(self._xml_tree)
if fgdc_root != FGDC_ROOT:
raise InvalidContent('Invalid XML root for ISO-19115 standard: {root}', root=fgdc_root)
fgdc_data_map = {'_root': FGDC_ROOT}
fgdc_data_structures = {}
# Capture and format other complex XPATHs
ad_format = _fgdc_tag_formats[ATTRIBUTES]
fgdc_data_structures[ATTRIBUTES] = format_xpaths(
_fgdc_definitions[ATTRIBUTES],
label=ad_format.format(ad_path='attrlabl'),
aliases=ad_format.format(ad_path='attalias'),
definition=ad_format.format(ad_path='attrdef'),
definition_src=ad_format.format(ad_path='attrdefs')
)
bb_format = _fgdc_tag_formats[BOUNDING_BOX]
fgdc_data_structures[BOUNDING_BOX] = format_xpaths(
_fgdc_definitions[BOUNDING_BOX],
east=bb_format.format(bbox_path='eastbc'),
south=bb_format.format(bbox_path='southbc'),
west=bb_format.format(bbox_path='westbc'),
north=bb_format.format(bbox_path='northbc')
)
ct_format = _fgdc_tag_formats[CONTACTS]
fgdc_data_structures[CONTACTS] = format_xpaths(
_fgdc_definitions[CONTACTS],
name=ct_format.format(ct_path='cntperp/cntper'),
_name=ct_format.format(ct_path='cntorgp/cntper'), # If not in cntperp
organization=ct_format.format(ct_path='cntperp/cntorg'),
_organization=ct_format.format(ct_path='cntorgp/cntorg'), # If not in cntperp
position=ct_format.format(ct_path='cntpos'),
email=ct_format.format(ct_path='cntemail')
)
dt_format = _fgdc_tag_formats[DATES]
fgdc_data_structures[DATES] = {
DATE_TYPE_MULTIPLE: dt_format.format(type_path='mdattim/sngdate/caldate'),
DATE_TYPE_RANGE_BEGIN: dt_format.format(type_path='rngdates/begdate'),
DATE_TYPE_RANGE_END: dt_format.format(type_path='rngdates/enddate'),
DATE_TYPE_SINGLE: dt_format.format(type_path='sngdate/caldate')
}
fgdc_data_structures[DATES][DATE_TYPE_RANGE] = [
fgdc_data_structures[DATES][DATE_TYPE_RANGE_BEGIN],
fgdc_data_structures[DATES][DATE_TYPE_RANGE_END]
]
df_format = _fgdc_tag_formats[DIGITAL_FORMS]
fgdc_data_structures[DIGITAL_FORMS] = format_xpaths(
_fgdc_definitions[DIGITAL_FORMS],
name=df_format.format(df_path='digtinfo/formname'),
content=df_format.format(df_path='digtinfo/formcont'),
decompression=df_format.format(df_path='digtinfo/filedec'),
version=df_format.format(df_path='digtinfo/formvern'),
specification=df_format.format(df_path='digtinfo/formspec'),
access_desc=df_format.format(df_path='digtopt/onlinopt/oncomp'),
access_instrs=df_format.format(df_path='digtopt/onlinopt/accinstr'),
network_resource=df_format.format(df_path='digtopt/onlinopt/computer/networka/networkr')
)
lw_format = _fgdc_tag_formats[LARGER_WORKS]
fgdc_data_structures[LARGER_WORKS] = format_xpaths(
_fgdc_definitions[LARGER_WORKS],
title=lw_format.format(lw_path='title'),
edition=lw_format.format(lw_path='edition'),
origin=lw_format.format(lw_path='origin'),
online_linkage=lw_format.format(lw_path='onlink'),
other_citation=lw_format.format(lw_path='othercit'),
date=lw_format.format(lw_path='pubdate'),
place=lw_format.format(lw_path='pubinfo/pubplace'),
info=lw_format.format(lw_path='pubinfo/publish')
)
ps_format = _fgdc_tag_formats[PROCESS_STEPS]
fgdc_data_structures[PROCESS_STEPS] = format_xpaths(
_fgdc_definitions[PROCESS_STEPS],
description=ps_format.format(ps_path='procdesc'),
date=ps_format.format(ps_path='procdate'),
sources=ps_format.format(ps_path='srcused')
)
ri_format = _fgdc_tag_formats[RASTER_INFO]
fgdc_data_structures[RASTER_INFO] = format_xpaths(
_fgdc_definitions[RASTER_INFO],
dimensions=ri_format.format(ri_path='rasttype'),
row_count=ri_format.format(ri_path='rowcount'),
column_count=ri_format.format(ri_path='colcount'),
vertical_count=ri_format.format(ri_path='vrtcount'),
x_resolution=_fgdc_tag_formats['_raster_resolution'] + '/absres',
_x_resolution=_fgdc_tag_formats['__raster_resolution'] + '/longres',
y_resolution=_fgdc_tag_formats['_raster_resolution'] + '/ordres',
_y_resolution=_fgdc_tag_formats['__raster_resolution'] + '/latres',
)
# Assign XPATHS and gis_metadata.utils.ParserProperties to fgdc_data_map
fgdc_data_formats = dict(_fgdc_tag_formats)
for prop, xpath in iteritems(fgdc_data_formats):
if prop in (ATTRIBUTES, CONTACTS, DIGITAL_FORMS, PROCESS_STEPS):
fgdc_data_map[prop] = ParserProperty(self._parse_complex_list, self._update_complex_list)
elif prop in (BOUNDING_BOX, LARGER_WORKS):
fgdc_data_map[prop] = ParserProperty(self._parse_complex, self._update_complex)
elif prop == DATES:
fgdc_data_map[prop] = ParserProperty(self._parse_dates, self._update_dates)
elif prop == RASTER_INFO:
fgdc_data_map[prop] = ParserProperty(self._parse_complex, self._update_raster_info)
else:
fgdc_data_map[prop] = xpath
self._data_map = fgdc_data_map
self._data_structures = fgdc_data_structures
