def checkRqrdNamespace(xml_file):
logger.info(
'verifying that this "body" namespace url in document.xml matches our template'
)
namespace_check = True
ns_url = compareElementNamespace(cfg.doc_xml, cfg.template_document_xml,
'body')
# log alert as needed
if ns_url != 'expected' and ns_url != 'unavailable':
namespace_check = False
setup_cleanup.setAlert('error', 'unexpected_namespace')
# send notification to wf team so we can keep an eye on these
subject = usertext_templates.subjects()["ns_notify"]
bodytxt = usertext_templates.emailtxt()["ns_notify"].format(
inputfilename=cfg.inputfilename,
ns_url=ns_url,
submitter=cfg.submitter_email,
alert_text=usertext_templates.alerts()['unexpected_namespace'],
support_email_address=cfg.support_email_address,
helpurl=cfg.helpurl)
sendmail.sendMail([cfg.alert_email_address], subject, bodytxt)
return namespace_check
os_utils.writeXMLtoFile(doc_root, doc_xml)
### zip ziproot up as a docx
logger.info(
"Zipping updated xml to replace workingfile .docx in the tmpfolder"
)
# os_utils.rm_existing_os_object(newdocxfile, 'validated_docx') <-- shouldnt be necessary, zipfile.py should overwrite
zipDOCX.zipDOCX(ziproot, workingfile) # < for prod
# zipDOCX.zipDOCX(ziproot, newdocxfile) # < for testing
########## SKIP RUNNING STUFF, LOG ALERTS
# Doc is not styled or has protection enabled, skip python validation
else:
logger.warn("* * Skipping ISBN_check:")
if protection:
errstring = usertext_templates.alerts()["protected"].format(
protection=protection)
os_utils.logAlerttoJSON(alerts_json, "error", errstring)
logger.warn("* {}".format(errstring))
########## CLEANUP
# write our json for style report to tmpdir
logger.debug("Writing isbn_check.json")
isbn_dict["completed"] = True
os_utils.dumpJSON(isbn_dict, cfg.isbn_check_json)
setup_cleanup.cleanupforValidator(this_outfolder, workingfile,
cfg.inputfilename, '',
cfg.stylereport_txt, alerts_json,
cfg.script_name)
except:
cfg.defineLogger(logfile, cfg.loglevel)
logger = logging.getLogger(__name__)
# w_logger = logging.getLogger('w_logger')
######### RUN!
# only run if this script is being invoked directly
if __name__ == '__main__':
try:
########## SETUP
# get file submitter via api, copy infile to tmpdir, setup outfolder
submitter_email, display_name, notdocx = setup_cleanup.setupforReporterOrConverter(inputfile, inputfilename, workingfile, this_outfolder, inputfile_ext)
if notdocx == True:
errstring = usertext_templates.alerts()["notdocx"].format(scriptname=cfg.script_name)
os_utils.logAlerttoJSON(alerts_json, "error", errstring)
logger.warn("* {}".format(errstring))
else:
# copy template to tmpdir, unzip infile and tmpdir
setup_cleanup.copyTemplateandUnzipFiles(macmillan_template, tmpdir, workingfile, ziproot, template_ziproot)
########## CHECK DOCUMENT
### check and compare versions, styling percentage, doc protection
logger.info('Comparing docx version to template, checking percent styled, checking if protection, trackchanges...')
version_result, current_version, template_version = check_docx.version_test(cfg.customprops_xml, cfg.template_customprops_xml, doc_version_min, doc_version_max)
percent_styled, macmillan_styled_paras, total_paras = check_docx.macmillanStyleCount(cfg.doc_xml, cfg.styles_xml)
protection, tc_marker_found, trackchange_status = check_docx.getProtectionAndTrackChangesStatus(cfg.doc_xml, cfg.settings_xml, cfg.footnotes_xml, cfg.endnotes_xml)
# create warnings re: track changes:
if tc_marker_found == True:
cfg.doc_xml, cfg.styles_xml)
protection, tc_marker_found, trackchange_status = check_docx.getProtectionAndTrackChangesStatus(
cfg.doc_xml, cfg.settings_xml, cfg.footnotes_xml, cfg.endnotes_xml)
# log for the rest o the validator suite:
report_dict["percent_styled"] = percent_styled
########## RUN STUFF
# Basic requirements passed, proceed with validation & cleanup
if percent_styled >= percent_styled_min and protection == "" and version_result != "docversion_above_maximum":
logger.info("Proceeding! (percent_styled='%s', protection='%s')" %
(percent_styled, protection))
# note and accept all track changes
if tc_marker_found == True:
errstring = usertext_templates.alerts()["v_unaccepted_tcs"]
os_utils.logAlerttoJSON(alerts_json, "warning", errstring)
logger.warn("* {}".format(errstring))
check_docx.acceptTrackChanges(cfg.doc_xml)
if os.path.exists(cfg.footnotes_xml):
check_docx.acceptTrackChanges(cfg.footnotes_xml)
if os.path.exists(cfg.endnotes_xml):
check_docx.acceptTrackChanges(cfg.endnotes_xml)
# # # Attach the template as needed
if version_result == "no_version" or version_result == "docversion_below_minimum":
errstring = usertext_templates.alerts()["v_no_template"]
os_utils.logAlerttoJSON(alerts_json, "warning", errstring)
logger.warn("* {}".format(errstring))
# logger.info("'version_result' = '%s'. Attaching macmillan template (updating styles in styles.xml, etc)" % version_result)
# # print "* This .docx did not have section start styles, attaching up-to-date template"
time.strftime("%y%m%d-%H%M%S")))
cfg.defineLogger(logfile, cfg.loglevel)
logger = logging.getLogger(__name__)
######### RUN!
# only run if this script is being invoked directly
if __name__ == '__main__':
try:
########## SETUP
# get file submitter via api, copy infile to tmpdir, setup outfolder
submitter_email, display_name, notdocx = setup_cleanup.setupforReporterOrConverter(
inputfile, inputfilename, workingfile, this_outfolder,
inputfile_ext)
if notdocx == True:
errstring = usertext_templates.alerts()["notdocx"].format(
scriptname=cfg.script_name)
os_utils.logAlerttoJSON(alerts_json, "error", errstring)
logger.warn("* {}".format(errstring))
else:
# copy template to tmpdir, unzip infile and tmpdir
setup_cleanup.copyTemplateandUnzipFiles(macmillan_template, tmpdir,
workingfile, ziproot,
template_ziproot)
########## CHECK DOCUMENT
### check and compare versions, styling percentage, doc protection
logger.info(
'Comparing docx version to template, checking percent styled, checking if protection, trackchanges...'
)
version_result, current_version, template_version = check_docx.version_test(
cfg.customprops_xml, cfg.template_customprops_xml,