def test_dump(self):
# Arrange
c = b"abc"
# Temporarily redirect stdout
old_stdout = sys.stdout
sys.stdout = mystdout = StringIO()
# Act
IptcImagePlugin.dump(c)
# Reset stdout
sys.stdout = old_stdout
# Assert
self.assertEqual(mystdout.getvalue(), "61 62 63 \n")
def get_iptc_data(filename):
"""Return a dict with the raw IPTC data."""
logger = logging.getLogger(__name__)
iptc_data = {}
raw_iptc = {}
# PILs IptcImagePlugin issues a SyntaxError in certain circumstances
# with malformed metadata, see PIL/IptcImagePlugin.py", line 71.
# ( https://github.com/python-pillow/Pillow/blob/9dd0348be2751beb2c617e32ff9985aa2f92ae5f/src/PIL/IptcImagePlugin.py#L71 )
try:
img = _read_image(filename)
raw_iptc = IptcImagePlugin.getiptcinfo(img)
except SyntaxError:
logger.info('IPTC Error in %s', filename)
# IPTC fields are catalogued in:
# https://www.iptc.org/std/photometadata/specification/IPTC-PhotoMetadata
# 2:05 is the IPTC title property
if raw_iptc and (2, 5) in raw_iptc:
iptc_data["title"] = raw_iptc[(2, 5)].decode('utf-8', errors='replace')
# 2:120 is the IPTC description property
if raw_iptc and (2, 120) in raw_iptc:
iptc_data["description"] = raw_iptc[(2, 120)].decode('utf-8',
errors='replace')
# 2:105 is the IPTC headline property
if raw_iptc and (2, 105) in raw_iptc:
iptc_data["headline"] = raw_iptc[(2, 105)].decode('utf-8',
errors='replace')
return iptc_data
def read_metadata(self, filename):
try:
image = Image.open(filename)
iptc = IptcImagePlugin.getiptcinfo(image)
metadata = {
k.strip(): v.strip()
for (k, v) in
[kv.decode("utf-8").split(':') for kv in iptc[(2, 25)]]
}
return metadata
except Exception as ex:
try:
tags = None
with open(filename, 'rb') as image_file:
tags = exifread.process_file(image_file, debug=True)
if tags and 'Image ApplicationNotes' in tags:
xml = str(tags['Image ApplicationNotes'])
xml = ET.fromstring(xml)
items = xml.findall(
'.//{http://www.w3.org/1999/02/22-rdf-syntax-ns#}Bag/{http://www.w3.org/1999/02/22-rdf-syntax-ns#}li'
)
items = [item.text.split(':') for item in items]
items = [(a.strip(), b.strip()) for a, b in items]
metadata = dict(items)
return metadata
except Exception as ex2:
pass
return None
def read_exif(self):
"""Reads relevant exif tags from an image provided as PIL.Image object"""
image_handle = Image.open(self.image_path)
exif = image_handle.info["parsed_exif"]
iptc = IptcImagePlugin.getiptcinfo(image_handle)
labeled_exifs = label_exifs(exif)
image_handle = None
# longitude, latitude
geotags = get_geotagging(exif)
self.lon, self.lat = get_coordinates(geotags)
# date taken
self._date_taken = datetime.strptime(
labeled_exifs.get("DateTimeOriginal"),
"%Y:%m:%d %H:%M:%S",
)
# headline
if not self.skip:
headline = iptc.get(IPTC_KEYS["headline"])
if headline is not None:
self.headline = headline.decode("UTF-8")
else:
self.headline = reverse_geocode(self)
# description
description = iptc.get(IPTC_KEYS["description"])
if description is not None:
self.description = description.decode("UTF-8")
else:
self.description = "No description...yet"
def expand(self):
self.fetch_data()
# Get rating
# cursor.execute("SELECT AVG(rating) AS average FROM " + tbl_image_rating + " WHERE image = %s GROUP BY image", [imageId])
# rating = cursor.fetchone()
# if rating != None:
# image['rating'] = rating
# get author name
if self.author:
self.author_name = db.fetch("SELECT name FROM " + db.tbl_author + " WHERE id=%s", [self.author], one=True,
as_list=True)
# get labels
labels = db.fetch(
"SELECT id, name FROM " + db.tbl_label + " INNER JOIN " + db.tbl_image_label + " ON (label = id AND image = %s)",
[self.id])
if labels:
self.labels = labels
# get meta data
self.open()
exif = {}
iptc = {}
exif_info = self.image._getexif() or {}
for tag, value in exif_info.items():
decoded = ExifTags.TAGS.get(tag, str(tag))
exif[decoded] = value
iptc_info = IptcImagePlugin.getiptcinfo(self.image) or {}
for tag, value in iptc_info.items():
decoded = iptc_tags.get(tag, str(tag))
iptc[decoded] = value
self.exif = exif
self.iptc = iptc
def get_iptc_data(filename):
"""Return a dict with the raw IPTC data."""
logger = logging.getLogger(__name__)
iptc_data = {}
raw_iptc = {}
# PILs IptcImagePlugin issues a SyntaxError in certain circumstances
# with malformed metadata, see PIL/IptcImagePlugin.py", line 71.
# ( https://github.com/python-pillow/Pillow/blob/9dd0348be2751beb2c617e32ff9985aa2f92ae5f/src/PIL/IptcImagePlugin.py#L71 )
try:
img = _read_image(filename)
raw_iptc = IptcImagePlugin.getiptcinfo(img)
except SyntaxError:
logger.info('IPTC Error in %s', filename)
# IPTC fields are catalogued in:
# https://www.iptc.org/std/photometadata/specification/IPTC-PhotoMetadata
# 2:05 is the IPTC title property
if raw_iptc and (2, 5) in raw_iptc:
iptc_data["title"] = raw_iptc[(2, 5)].decode('utf-8', errors='replace')
# 2:120 is the IPTC description property
if raw_iptc and (2, 120) in raw_iptc:
iptc_data["description"] = raw_iptc[(2, 120)].decode('utf-8',
errors='replace')
# 2:105 is the IPTC headline property
if raw_iptc and (2, 105) in raw_iptc:
iptc_data["headline"] = raw_iptc[(2, 105)].decode('utf-8',
errors='replace')
return iptc_data
def get_meta_iptc(file_stream):
"""Returns the image IPTC metadata in a dictionary of tag:value pairs.
@param file_stream: stream
"""
file_stream.seek(0)
img = Image.open(file_stream)
iptc_raw = IptcImagePlugin.getiptcinfo(img)
metadata = {}
if iptc_raw is None:
return metadata
for code, value in iptc_raw.items():
try:
tag = iim_codes[code]
except KeyError:
continue
if isinstance(value, list):
try:
value = [v.decode('utf-8') for v in value]
except UnicodeDecodeError:
pass
elif isinstance(value, bytes):
try:
value = value.decode('utf-8')
except UnicodeDecodeError:
pass
metadata[tag] = value
return metadata
def test_getiptcinfo_tiff_none(self):
# Arrange
im = Image.open("Tests/images/hopper.tif")
# Act
iptc = IptcImagePlugin.getiptcinfo(im)
# Assert
self.assertIsNone(iptc)
def test_getiptcinfo_tiff_none(self):
# Arrange
im = Image.open("Tests/images/hopper.tif")
# Act
iptc = IptcImagePlugin.getiptcinfo(im)
# Assert
self.assertIsNone(iptc)
def test_getiptcinfo_jpg_none(self):
# Arrange
im = hopper()
# Act
iptc = IptcImagePlugin.getiptcinfo(im)
# Assert
self.assertIsNone(iptc)
def test_getiptcinfo_jpg_none():
# Arrange
with hopper() as im:
# Act
iptc = IptcImagePlugin.getiptcinfo(im)
# Assert
assert iptc is None
def test_getiptcinfo_jpg_none(self):
# Arrange
im = hopper()
# Act
iptc = IptcImagePlugin.getiptcinfo(im)
# Assert
self.assertIsNone(iptc)
def test_i(self):
# Arrange
c = b"a"
# Act
ret = IptcImagePlugin.i(c)
# Assert
self.assertEqual(ret, 97)
def test_i():
# Arrange
c = b"a"
# Act
ret = IptcImagePlugin.i(c)
# Assert
assert ret == 97
def test_getiptcinfo_tiff_none():
# Arrange
with Image.open("Tests/images/hopper.tif") as im:
# Act
iptc = IptcImagePlugin.getiptcinfo(im)
# Assert
assert iptc is None
def test_i(self):
# Arrange
c = b"a"
# Act
ret = IptcImagePlugin.i(c)
# Assert
self.assertEqual(ret, 97)
def test_getiptcinfo_jpg_found(self):
# Arrange
im = Image.open(TEST_FILE)
# Act
iptc = IptcImagePlugin.getiptcinfo(im)
# Assert
self.assertIsInstance(iptc, dict)
self.assertEqual(iptc[(2, 90)], b"Budapest")
self.assertEqual(iptc[(2, 101)], b"Hungary")
def test_getiptcinfo_jpg_found(self):
# Arrange
im = Image.open(TEST_FILE)
# Act
iptc = IptcImagePlugin.getiptcinfo(im)
# Assert
self.assertIsInstance(iptc, dict)
self.assertEqual(iptc[(2, 90)], b"Budapest")
self.assertEqual(iptc[(2, 101)], b"Hungary")
def test_dump(self):
# Arrange
c = b"abc"
# Temporarily redirect stdout
try:
from cStringIO import StringIO
except ImportError:
from io import StringIO
import sys
old_stdout = sys.stdout
sys.stdout = mystdout = StringIO()
# Act
IptcImagePlugin.dump(c)
# Reset stdout
sys.stdout = old_stdout
# Assert
self.assertEqual(mystdout.getvalue(), "61 62 63 \n")
def test_getiptcinfo_jpg_found():
# Arrange
with Image.open(TEST_FILE) as im:
# Act
iptc = IptcImagePlugin.getiptcinfo(im)
# Assert
assert isinstance(iptc, dict)
assert iptc[(2, 90)] == b"Budapest"
assert iptc[(2, 101)] == b"Hungary"
def get_iptc_data(filename):
"""Return a dict with the raw IPTC data."""
img = _read_image(filename)
raw_iptc = IptcImagePlugin.getiptcinfo(img)
# IPTC fields are catalogued in:
# https://www.iptc.org/std/photometadata/specification/IPTC-PhotoMetadata
iptc_data = {}
# 2:05 is the IPTC title property
if raw_iptc and (2, 5) in raw_iptc:
iptc_data["title"] = raw_iptc[(2, 5)].decode('utf-8')
# 2:120 is the IPTC description property
if raw_iptc and (2, 120) in raw_iptc:
iptc_data["description"] = raw_iptc[(2, 120)].decode('utf-8')
return iptc_data
def extractIptc(self, image):
iptc = {}
iptc_tags = {
(1, 90): 'CodedCharacterSet',
(2, 0): 'Whatever',
(2, 5): 'ObjectName',
(2, 10): 'Urgency',
(2, 15): 'Category',
(2, 20): 'Subcategories',
(2, 25): 'Keywords',
(2, 40): 'SecialInstructions',
(2, 55): 'CreationDate',
(2, 60): 'CreationTime',
(2, 62): 'DigitizationDate',
(2, 80): 'AuthorByline',
(2, 85): 'AuthorTitle',
(2, 90): 'City',
(2, 92): 'SubLocation',
(2, 95): 'State',
(2, 100): 'CountryCode',
(2, 101): 'Country',
(2, 103): 'OTR',
(2, 105): 'Headline',
(2, 110): 'Source',
(2, 115): 'PhotoSource',
(2, 116): 'Copyright',
(2, 120): 'Caption',
(2, 122): 'CaptionWriter',
}
iptc_info = IptcImagePlugin.getiptcinfo(image) or {}
for tag, value in iptc_info.items():
decoded = iptc_tags.get(tag, str(tag))
if isinstance(value, (bytes, bytearray)):
iptc[decoded] = value.decode('utf-8')
elif isinstance(value, (list)):
for k, v in enumerate(value):
value[k] = v.decode('utf-8')
iptc[decoded] = value
else:
iptc[decoded] = str(value)
self.metaData[self.currentFile]['iptc'] = iptc
def get_iptc_metadata(self, image_file_path):
"""
Reads the IPTC metadata of an image given by 'image_file_path' and returns it.
Returns empty array if IPTC metadata is not available.
:param image_file_path:
:return iptc:
"""
logger.debug('get_iptc_metadata()')
try:
im = PIL.Image.open(image_file_path)
iptc = IptcImagePlugin.getiptcinfo(im)
im.close()
if not iptc is None:
return iptc
else:
return []
except:
e = sys.exc_info()[0]
logger.debug('is_image_landscape() - {}'.format(e))
def get_image_metadata(path):
metadata = {'Exif.Image.Rating': None, 'Iptc.Application2.Keywords': []}
exif, iptc = None, None
try:
image = Image.open(path)
exif = image._getexif()
iptc = IptcImagePlugin.getiptcinfo(image)
image = None
except:
log.error("Error loading metadata: %s" % path)
if exif and exif.has_key(EXIF_IMAGE_RATING_RAW):
metadata[RATING] = exif[EXIF_IMAGE_RATING_RAW]
if iptc and iptc.has_key(IPTC_APPLICATION2_KEYWORDS_RAW):
metadata[TAGS] = iptc[IPTC_APPLICATION2_KEYWORDS_RAW]
if isinstance(metadata[TAGS], str):
metadata[TAGS] = [metadata[TAGS]]
metadata[TAGS] = set(metadata[TAGS])
return metadata
def get_image_metadata(path):
metadata = {'Exif.Image.Rating': None,
'Iptc.Application2.Keywords': []}
exif, iptc = None, None
try:
image = Image.open(path)
exif = image._getexif()
iptc = IptcImagePlugin.getiptcinfo(image)
image = None
except:
log.error("Error loading metadata: %s" % path)
if exif and exif.has_key(EXIF_IMAGE_RATING_RAW):
metadata[RATING] = exif[EXIF_IMAGE_RATING_RAW]
if iptc and iptc.has_key(IPTC_APPLICATION2_KEYWORDS_RAW):
metadata[TAGS] = iptc[IPTC_APPLICATION2_KEYWORDS_RAW]
if isinstance(metadata[TAGS], str):
metadata[TAGS] = [metadata[TAGS]]
metadata[TAGS] = set(metadata[TAGS])
return metadata
def get_meta_iptc(file_stream):
"""Returns the image IPTC metadata in a dictionary of tag:value pairs.
@param file_stream: stream
"""
file_stream.seek(0)
img = Image.open(file_stream)
iptc_raw = IptcImagePlugin.getiptcinfo(img)
metadata = {}
if iptc_raw is None:
return metadata
for code, value in iptc_raw.items():
try:
tag = iim_codes[code]
except KeyError:
continue
if isinstance(value, list):
value = [decode(v) for v in value]
elif isinstance(value, bytes):
value = decode(value)
metadata[tag] = value
return metadata
def get_caption(image):
iptc = IptcImagePlugin.getiptcinfo(image)
return iptc.get((2, 5)).decode()
@staticmethod
def mdget(phile):
try:
im = Image.open(phile)
except IOError, why:
sys.stderr.write("Can't find metadata for %s (%s).\n" %
(phile, why)
)
return None
out = {}
exif_info = im._getexif() or {}
for tag, value in exif_info.items():
decoded = ExifTags.TAGS.get(tag, "unknown")
out["Exif." + decoded] = value
iptc_info = IptcImagePlugin.getiptcinfo(im) or {}
for tag, value in iptc_info.items():
decoded = PhotoWebber._iptc_tags.get(tag, "unknown")
out["Iptc." + decoded] = value
return out, im.size
@staticmethod
def error(msg):
"Something has gone horribly wrong."
sys.stderr.write("FATAL: %s\n" % msg)
sys.exit(1)
def photoweb_cli():
"Run command-line photoweb."
usage = "Usage: %prog [options] <dir>"
def pil(file_path: str):
from PIL import Image, IptcImagePlugin
im = Image.open(file_path)
meta = IptcImagePlugin.getiptcinfo(im)
pprint(meta)
def update_metadata(self):
self.open()
iptc_info = IptcImagePlugin.getiptcinfo(self.image) or {}
exif_info = self.image._getexif() or {}
fields = []
values = []
if IPTC_OBJECT_NAME in iptc_info:
title = iptc_info[IPTC_OBJECT_NAME].decode('utf-8')
fields.append("description")
values.append(title)
timestamp = None
if IPTC_DATE_CREATED in iptc_info and IPTC_TIME_CREATED in iptc_info:
timestamp_str = iptc_info[IPTC_DATE_CREATED].decode('utf-8') + iptc_info[IPTC_TIME_CREATED].decode('utf-8')
if len(timestamp_str) == 14:
timestamp_str = timestamp_str + '+0000'
timestamp = datetime.strptime(timestamp_str, '%Y%m%d%H%M%S%z')
elif EXIF_DATE_TIME_ORIGINAL in exif_info:
timestamp_str = exif_info[EXIF_DATE_TIME_ORIGINAL]
if len(timestamp_str) == 19:
timestamp_str = timestamp_str + '+0000'
elif timestamp_str[-3] == ':':
timestamp_str = kill_char(timestamp_str, len(timestamp_str)-3)
timestamp = datetime.strptime(timestamp_str, '%Y:%m:%d %H:%M:%S%z')
elif EXIF_DATE_TIME_DIGITIZED in exif_info:
timestamp_str = exif_info[EXIF_DATE_TIME_DIGITIZED]
timestamp = datetime.strptime(timestamp_str, '%Y:%m:%d %H:%M:%S')
if timestamp:
fields.append("stime")
values.append(timestamp)
if IPTC_BYLINE in iptc_info:
author_name = iptc_info[IPTC_BYLINE].decode('utf-8')
author_id = db.fetch("SELECT id FROM " + db.tbl_author + " WHERE name=%s", [author_name], one=True,
as_list=True)
if not author_id:
author_id = db.execute("INSERT INTO " + db.tbl_author + "(name) VALUES (%s) RETURNING id",
[author_name])
fields.append("author")
values.append(author_id)
if IPTC_KEYWORDS in iptc_info:
labels = []
if isinstance(iptc_info[IPTC_KEYWORDS], list):
for label in iptc_info[IPTC_KEYWORDS]:
labels.append(label.decode('utf-8'))
else:
labels.append(iptc_info[IPTC_KEYWORDS].decode('utf-8'))
self.set_labels(labels)
fields.append("width")
values.append(self.image.width)
fields.append("height")
values.append(self.image.height)
if fields:
values.append(self.id)
fields_str = ", ".join(map(lambda x: "%s=%%s" % x, fields))
db.execute("UPDATE " + db.tbl_image + " SET " + fields_str + " WHERE id=%s", values)
db.commit()
