def validate_doc_xml(doc_id, rules):
doc_info = get_doc_info(doc_id)
doc_name = doc_info[0]
doc_corpus = doc_info[1]
doc_content = get_doc_content(doc_id)
cache_hit = cache.get_report(doc_id, "xml")
if cache_hit:
return cache_hit
report = ""
xml_rule_fired = False
for rule in rules:
if not rule.applies(doc_name, doc_corpus):
continue
xml_rule_fired = True
res = rule.validate(doc_content)
report += res
if not xml_rule_fired:
report = "<strong>No applicable XML schemas</strong><br>"
elif report:
report = "<strong>XML problems:</strong><br>" + report
else:
report = "<strong>XML is valid</strong><br>"
cache.cache_validation_result(doc_id, "xml", report)
return report
def validate_doc_export(doc_id, rules, timestamps=None): doc_info = get_doc_info(doc_id) doc_name = doc_info[0] doc_corpus = doc_info[1] doc_content = get_doc_content(doc_id) ether_doc_name = "gd_" + doc_corpus + "_" + doc_name if not timestamps: timestamps = get_timestamps(ether_url) last_edit = int(timestamps[ether_doc_name]) if last_edit <= int(cache.get_timestamp(doc_id, "export")): return cache.get_report(doc_id, "export") socialcalc = get_socialcalc(ether_url, ether_doc_name) report = "" export_rule_fired = False for rule in rules: if not rule.applies(doc_name, doc_corpus): continue export_rule_fired = True res = rule.validate(socialcalc, doc_id) report += res if not export_rule_fired: report = "<strong>No applicable export schemas</strong><br>" elif report: report = "<strong>Export problems:</strong><br>" + report else: report = "<strong>Export is valid</strong><br>" cache.cache_timestamped_validation_result(doc_id, "export", report, last_edit) return report
def validate_all_export_bulk(docs):
cached_reports = {}
reports = []
rules = [BulkExportValidator(x) for x in get_export_rules()]
timestamps = get_timestamps(ether_url)
doc_ids = []
for doc in docs:
doc_id, doc_name, doc_corpus, doc_mode, doc_schema, validation, timestamp = doc
if doc_mode != "ether":
continue
ether_doc_name = "gd_" + doc_corpus + "_" + doc_name
last_edit = int(timestamps[ether_doc_name])
if last_edit <= int(cache.get_timestamp(doc_id, "export")):
cached_reports[doc_id] = cache.get_report(doc_id, "export")
continue
doc_ids.append(doc_id)
for rule in rules:
report, fired = rule.validate(doc_ids)
if fired:
reports.append(report)
def merge_dicts(dictlist):
keys = apply(set().union, dictlist)
ret_dict = {}
for k in keys:
for d in dictlist:
ret_dict[k] = "".join(d.get(k, ''))
#return {k: "".join(d.get(k, '') for d in dictlist) for k in keys}
return ret_dict
reports = merge_dicts([cached_reports] + reports)
for doc_id in doc_ids:
if doc_id not in reports:
reports[doc_id] = "No applicable export schemas"
for doc_id, report in reports.items():
if doc_id in cached_reports:
continue
doc_name, doc_corpus, _, _, _, _, _ = get_doc_info(doc_id)
ether_doc_name = "gd_" + doc_corpus + "_" + doc_name
last_edit = int(timestamps[ether_doc_name])
cache.cache_timestamped_validation_result(doc_id, "export", report,
last_edit)
return json.dumps(reports)
def validate_doc_ether(doc_id, rules, timestamps=None, editor=False):
doc_info = get_doc_info(doc_id)
doc_name = doc_info[0]
doc_corpus = doc_info[1]
ether_doc_name = "gd_" + doc_corpus + "_" + doc_name
if not timestamps:
timestamps = get_timestamps(ether_url)
last_edit = int(timestamps[ether_doc_name])
if last_edit <= int(cache.get_timestamp(doc_id, "ether")):
return cache.get_report(doc_id, "ether")
socialcalc = get_socialcalc(ether_url, ether_doc_name)
parsed_ether = parse_ether(socialcalc, doc_id=doc_id)
report = ''
cells = []
ether_rule_fired = False
for rule in rules:
if not rule.applies(doc_name, doc_corpus):
continue
ether_rule_fired = True
res = rule.validate(parsed_ether)
if len(res['tooltip']) > 0:
report += ("""<div class="tooltip">""" + res['report'][:-5] +
""" <i class="fa fa-ellipsis-h"></i>""" +
"<span class=\"msg\">" + res['tooltip'] + "</span>" +
"</div>")
else:
report += res['report']
cells += res['cells']
if not ether_rule_fired:
report = "<strong>No applicable spreadsheet validation rules</strong><br>"
elif report:
report = "<strong>Spreadsheet Problems:</strong><br>" + report
else:
report = "<strong>Spreadsheet is valid</strong><br>"
cache.cache_timestamped_validation_result(doc_id, "ether", report,
last_edit)
if editor:
highlight_cells(cells, ether_url, ether_doc_name)
return report
def validate_doc_meta(doc_id, rules):
# metadata validation
report = ""
meta = get_doc_meta(doc_id)
doc_info = get_doc_info(doc_id)
doc_name = doc_info[0]
doc_corpus = doc_info[1]
cache_hit = cache.get_report(doc_id, "meta")
if cache_hit:
return cache_hit
meta_rule_fired = False
for rule in rules:
if not rule.applies(doc_name, doc_corpus):
continue
meta_rule_fired = True
res = rule.validate(meta)
if len(res['tooltip']) > 0:
report += ("""<div class="tooltip">"""
+ res['report'][:-5]
+ """ <i class="fa fa-ellipsis-h"></i>"""
+ "<span class=\"msg\">" + res['tooltip'] + "</span>"
+ "</div>")
else:
report += res['report']
if not meta_rule_fired:
report = "<strong>No applicable metadata rules</strong><br>"
elif len(report) == 0:
report = "<strong>Metadata is valid<br></strong>"
else:
report = "<strong>Metadata Problems:</strong><br>" + report
cache.cache_validation_result(doc_id, "meta", report)
return report
