def test_add_encoding_desc(self):
openn_tei = OPennTEI(open(TestOPennTEI.test_partial_tei))
# print openn_tei.to_string()
desc_string = open(self.encoding_desc).read()
openn_tei.add_encoding_desc(desc_string)
# print openn_tei.to_string()
root = self.assertXmlDocument(openn_tei.to_string())
self.assertXpathsExist(root, ('//ns:catDesc[text()= "Headpiece"]',))
def tei(self):
if getattr(self, 'openn_tei', None) is None:
f = open(self.package_dir.partial_tei_path, 'r')
try:
self.openn_tei = OPennTEI(f.read())
finally:
f.close()
return self.openn_tei
def get_tei(self, document):
try:
return OPennTEI(document.tei_xml)
except OPennException as oex:
msg = "Error processing document: id: %d, base_dir: '%s'" % (
document.id, document.base_dir)
self.logger.error(msg)
raise OPennException(msg, oex, str(oex))
def regen_partial_tei(self, doc, **kwargs):
# validate directory
# Move files:
#
# - pages.xlsx required
# - marc.xml required unless bibid.txt present
# - bibid.txt ignored; BibID should be in existing TEI
# - holdingid.txt optional; may be required for Penn MSS (with BibID in TEI)
data_dir = kwargs.get('METADATA_DIR', None)
if data_dir is None or data_dir.strip() == '':
raise OPennException("Missing required METADATA_DIR")
if not os.path.exists(data_dir):
raise OPennException("Cannot find METADATA_DIR: '%s'" % (data_dir,))
metadata_files = ('pages.xlsx', 'marc.xml', 'holdingid.txt')
for file in metadata_files:
full_path = os.path.abspath(os.path.join(data_dir, file))
if os.path.exists(full_path):
dest = os.path.abspath(os.path.join(self.source_dir, file))
if full_path == dest:
pass
elif os.path.exists(full_path):
shutil.copyfile(full_path, dest)
tei = OPennTEI(doc.tei_xml)
bibid = tei.bibid
# make sure we have the marc.xml file
if os.path.exists(self.marc_xml):
pass
elif bibid is None:
OPennException("Saved TEI lacks BibID; required MARC file missing: '%s'" % (self.marc_xml,))
else:
if not self.NEW_BIBID_RE.match(bibid):
bibid = '99%s3503681' % (str(bibid),)
self.write_xml(bibid, self.marc_xml)
# create pages.xml from the page.xlsx
self.write_openn_xml(self.openn_xml_path)
# fake the pih.xml by merging pages.xml with marc.xml (from above)
self.write_pih_xml()
self.save_rights_data()
partial_tei_xml = self.gen_partial_tei()
self.write_partial_tei(self.source_dir, partial_tei_xml)
self.validate_partial_tei()
self.stage_marc_xml()
self.add_removal(self.pih_filename)
self.add_removal(self.bibid_filename)
self.add_removal(self.holdingid_filename)
self.add_removal(self.openn_xml_path)
self.add_removal(self.xlsx_path)
self.add_removal(os.path.join(self.source_dir, 'sha1manifest.txt'))
def write_partial_tei(self, outdir, xml):
outfile = os.path.join(outdir, 'PARTIAL_TEI.xml')
f = open(outfile, 'w+')
try:
f.write(xml)
# try to read it
f.seek(0)
tei = OPennTEI(f)
tei.validate()
finally:
f.close()
return outfile
def regen_partial_tei(self, doc, **kwargs):
data_dir = kwargs.get('METADATA_DIR', None)
if data_dir:
metadata_files = ['keywords.txt', 'holdingid.txt']
for file in metadata_files:
full_path = os.path.abspath(os.path.join(data_dir, file))
if os.path.exists(full_path):
dest = os.path.abspath(os.path.join(self.source_dir, file))
shutil.copyfile(full_path, dest)
tei = OPennTEI(doc.tei_xml)
bibid = tei.bibid
if bibid is None:
raise OPennException("Whoah now. bibid is none. That ain't right.")
if len(bibid) < 8:
bibid = '99%s3503681' % (str(bibid),)
self.write_xml(bibid,self.pih_filename)
partial_tei = self.build_partial_tei()
self.write_partial_tei(self.source_dir, partial_tei)
self.add_removal(self.pih_filename)
self.add_removal(self.keywords_filename())
def test_scripts(self):
openn_tei = OPennTEI(codecs.open(self.mscodex906_tei))
self.assertIn(u'Written in Gothic cursive, with the first words or first line at the beginning of sections in bâtarde script; f. 306-322v (a single gathering) in a second hand.', openn_tei.scripts)
def test_collation(self):
openn_tei = OPennTEI(codecs.open(self.mscodex906_tei))
self.assertEqual(u'Paper, 342; 1-18¹⁶ 19¹⁸ 20-22¹⁶; [ii], 1-317, 319-340, [i]; misnumbered at 318, no loss of text. Foliation and line numbering in a later hand. Signatures at bottom right; catchwords at bottom center, often cropped.', openn_tei.collation)
def test_colophon(self):
openn_tei = OPennTEI(open(self.mscodex83_tei))
self.assertEqual('Colophon (f. 32v): Explicit hic liber; de pena sum modo liber/ Explicit hoc totum; pro pena da michi potum/ Explicit expliceat; ludere scriptor eat/ Finito libro sit laus et gloria Christo.', openn_tei.colophon)
def test_layouts(self):
openn_tei = OPennTEI(open(self.ljs270_tei))
self.assertIn('Inscribed in 23 lines on 2 sides of the tablet.', openn_tei.layouts)
def __init__(self, document):
self._document = document
self._tei = OPennTEI(self.document.tei_xml)
def test_no_title(self):
openn_tei = OPennTEI(open(self.mscodex1589_tei_no_title))
with self.assertRaises(OPennException) as oe:
openn_tei.validate()
self.assertIn('title', str(oe.exception))
def test_origin(self):
openn_tei = OPennTEI(open(self.mscodex906_tei))
self.assertEqual(u'Written in France after 1474, based on primary watermark.', openn_tei.origin)
import sys
# Don't run this stuff
sys.exit(1)
# UPDATE TEI from file system
from openn.xml.openn_tei import OPennTEI
doc = Document.objects.get(pk=6853)
tei = OPennTEI(open('/mnt/scratch02/openn/site/Data/0032/ms_or_044/data/ms_or_044_TEI.xml').read())
doc.tei_xml = tei.to_string()
doc.save()
names = (('0007', 'lehigh_002', 'lehigh_codex_002'),
('0007', 'lehigh_003', 'lehigh_codex_003'),
('0007', 'lehigh_006', 'lehigh_codex_006'),
('0007', 'lehigh_007', 'lehigh_codex_007'),
('0007', 'Antiphon_25', 'lehigh_codex_025'),
('0007', 'BookofHoursoftheRomanuse_18', 'lehigh_codex_018'),
('0003', 'BMCMS2', 'BMC_MS02'),
('0003', 'BMCMS5', 'BMC_MS05'),
('0003', 'BMCMS31', 'BMC_MS31'),
('0003', 'BMCMS32', 'BMC_MS32'),
('0003', 'BMCMS33', 'BMC_MS33'),
('0003', 'BMCMS34', 'BMC_MS34'),
('0003', 'BMC_MS8', 'BMC_MS08'),
('0003', 'BMC_MS9', 'BMC_MS09'),
('0003', 'BMC_MS7', 'BMC_MS07'),
('0003', 'BMC_MS4', 'BMC_MS04'),
('0003', 'BMC_MS6', 'BMC_MS06'),
('0003', 'BMC_MS3', 'BMC_MS03'),
('0012', 'ms2_2224q', 'lcp_ms02'),
def build_partial_tei(self):
xml_string = self.gen_partial_tei()
tei = OPennTEI(xml_string)
self.add_keywords(tei)
self.add_encoding_desc(tei)
return tei.to_string()
def test_decoration(self):
openn_tei = OPennTEI(codecs.open(self.mscodex906_tei))
self.assertEqual(u'Many 2- and 3-line initials in red; capital at beginning of each line stroked with yellow; rubrication in red, often with cropped notes for rubrication or illustration barely visible at edge of page; spaces for illumination, approximately 9 lines in height, are frequent at the beginning of the manuscript but disappear in the second half (after f. 171v).', openn_tei.decoration)
def test_signatures(self):
openn_tei = OPennTEI(open(self.mscodex75_tei))
self.assertEqual(u'Some signatures visible in red in the first half of some quires.', openn_tei.signatures)
def test_binding(self):
openn_tei = OPennTEI(open(self.mscodex906_tei))
self.assertEqual(u'Contemporary blind-stamped calf, rebacked; wormholes in boards, leather, and text near spine.', openn_tei.binding)
def _openn_tei(self):
return OPennTEI(self._read_tei())
def test_watermark(self):
openn_tei = OPennTEI(open(self.ljs498_tei))
self.assertEqual(u'Similar to Briquet, Aigle à deux têtes 285 (f. 45, 48; Gratz, 1594-1598; Millstatt (Carinthie), 1597-1600; Vienna, 1599) and Aigle à deux têtes 282 (Gratz, 1580; Osnabruck, 1588); similar to Briquet, Aigle à deux têtes 291 (f. 189; Gratz, 1598; Kempten, 1607-1627).', openn_tei.watermark)
def test_get_call_number(self):
openn_tei = OPennTEI(open(TestOPennTEI.test_partial_tei))
self.assertEqual('Ms. Codex 1223', openn_tei.call_number)
def test_resource(self):
openn_tei = OPennTEI(open(self.mscodex906_tei))
self.assertEqual(u'http://hdl.library.upenn.edu/1017/d/medren/3559152', openn_tei.resource)
def test_get_title(self):
openn_tei = OPennTEI(open(TestOPennTEI.test_partial_tei))
self.assertEqual('Fragments of the Digests of Justinian, Book 37, Titles 7-9', openn_tei.title)
def test_no_idno(self):
openn_tei = OPennTEI(open(self.mscodex1589_tei_no_idno))
with self.assertRaises(OPennException) as oe:
openn_tei.validate()
self.assertIn('Call number', str(oe.exception))
def test_foliation(self):
openn_tei = OPennTEI(open(self.mscodex52_tei))
self.assertEqual('Paper, i (contemporary paper) + 119; 1-94, [95-119]; contemporary foliation in ink, modern foliation in pencil, upper right recto.', openn_tei.foliation)