def getErrors(output):
# Parse preflight output and return error codes and descriptions as element object
errorsElt=ET.Element("validationErrors")
outputLines=output.split('\n')
noLines=len(outputLines)
# Loop through output, skip 1st line which doesn't contain any useful info
for i in range(1,noLines):
thisLine=outputLines[i]
thisLineItems=thisLine.split(":")
noItems=len(thisLineItems)
if noItems>=2:
thisErrorCode=thisLineItems[0]
thisErrorDescription=thisLineItems[1]
# Some silly looking code but error descriptions may contain colons
if noItems>2:
for j in range (2,noItems):
thisErrorDescription = thisErrorDescription + ":" + thisLineItems[j]
thisErrorCode.strip()
thisErrorDescription.strip()
errorElt=ET.Element("error")
errorElt.appendChildTagWithText("code", thisErrorCode)
errorElt.appendChildTagWithText("description", thisErrorDescription)
errorsElt.append(errorElt)
return(errorsElt)
def checkOneFile(file):
# Process one file and return analysis result as text string (which contains
# formatted XML)
fileData = readFileBytes(file)
isValidJP2, tests, characteristics = BoxValidator(
"JP2", fileData).validate() #validateJP2(fileData)
# Generate property values remap table
remapTable = generatePropertiesRemapTable()
# Create printable version of tests and characteristics tree
tests.makeHumanReadable()
characteristics.makeHumanReadable(remapTable)
# Create output elementtree object
root = ET.Element('jpylyzer')
# Create elements for storing tool and file meta info
toolInfo = ET.Element('toolInfo')
fileInfo = ET.Element('fileInfo')
# File name and path may contain non-ASCII characters, decoding to Latin should
# (hopefully) prevent any Unicode decode errors. Elementtree will then deal with any non-ASCII
# characters by replacing them with numeric entity references
try:
# This works in Python 2.7, but raises error in 3.x (no decode attribute for str type!)
fileName = os.path.basename(file).decode("iso-8859-15", "strict")
filePath = os.path.abspath(file).decode("iso-8859-15", "strict")
except AttributeError:
# This works in Python 3.x, but goes wrong withh non-ASCII chars in 2.7
fileName = os.path.basename(file)
filePath = os.path.abspath(file)
# Produce some general tool and file meta info
toolInfo.appendChildTagWithText("toolName", scriptName)
toolInfo.appendChildTagWithText("toolVersion", __version__)
fileInfo.appendChildTagWithText("fileName", fileName)
fileInfo.appendChildTagWithText("filePath", filePath)
fileInfo.appendChildTagWithText("fileSizeInBytes",
str(os.path.getsize(file)))
fileInfo.appendChildTagWithText("fileLastModified",
time.ctime(os.path.getmtime(file)))
# Append to root
root.append(toolInfo)
root.append(fileInfo)
# Add validation outcome
root.appendChildTagWithText("isValidJP2", str(isValidJP2))
# Append test results and characteristics to root
root.append(tests)
root.append(characteristics)
return (root)
def checkFiles(images):
if len(images) == 0:
warnings.warn("no images to check!")
for image in images:
thisFile = image
isFile = os.path.isfile(thisFile)
if isFile:
# Read and analyse one file
fileData = readFileBytes(thisFile)
isValidJP2, tests, characteristics = BoxValidator(
"JP2", fileData).validate() #validateJP2(fileData)
# Generate property values remap table
remapTable = generatePropertiesRemapTable()
# Create printable version of tests and characteristics tree
tests.makeHumanReadable()
characteristics.makeHumanReadable(remapTable)
# Create output elementtree object
root = ET.Element('jpylyzer')
# Create elements for storing tool and file meta info
toolInfo = ET.Element('toolInfo')
fileInfo = ET.Element('fileInfo')
# Produce some general tool and file meta info
toolInfo.appendChildTagWithText("toolName", scriptName)
toolInfo.appendChildTagWithText("toolVersion", __version__)
fileInfo.appendChildTagWithText("fileName", thisFile)
fileInfo.appendChildTagWithText("filePath",
os.path.abspath(thisFile))
fileInfo.appendChildTagWithText("fileSizeInBytes",
str(os.path.getsize(thisFile)))
fileInfo.appendChildTagWithText(
"fileLastModified", time.ctime(os.path.getmtime(thisFile)))
# Append to root
root.append(toolInfo)
root.append(fileInfo)
# Add validation outcome
root.appendChildTagWithText("isValidJP2", str(isValidJP2))
# Append test results and characteristics to root
root.append(tests)
root.append(characteristics)
# Write output
sys.stdout.write(root.toxml())
def checkOneFile(file):
# Process one file and return analysis result as text string (which contains
# formatted XML)
fileData = readFileBytes(file)
isValidJP2, tests, characteristics = BoxValidator(
"JP2", fileData).validate() #validateJP2(fileData)
# Generate property values remap table
remapTable = generatePropertiesRemapTable()
# Create printable version of tests and characteristics tree
tests.makeHumanReadable()
characteristics.makeHumanReadable(remapTable)
# Create output elementtree object
root = ET.Element('jpylyzer')
# Create elements for storing tool and file meta info
toolInfo = ET.Element('toolInfo')
fileInfo = ET.Element('fileInfo')
# Produce some general tool and file meta info
toolInfo.appendChildTagWithText("toolName", scriptName)
toolInfo.appendChildTagWithText("toolVersion", __version__)
fileInfo.appendChildTagWithText("fileName", os.path.basename(file))
fileInfo.appendChildTagWithText("filePath", os.path.abspath(file))
fileInfo.appendChildTagWithText("fileSizeInBytes",
str(os.path.getsize(file)))
fileInfo.appendChildTagWithText("fileLastModified",
time.ctime(os.path.getmtime(file)))
# Append to root
root.append(toolInfo)
root.append(fileInfo)
# Add validation outcome
root.appendChildTagWithText("isValidJP2", str(isValidJP2))
# Append test results and characteristics to root
root.append(tests)
root.append(characteristics)
# Result as XML
result = root.toxml().decode("ascii")
return (result)
def checkOneFileData(fileName, filePath, fileSizeInBytes, fileLastModifiedDate, fileData):
# Process the data from one file and return analysis result as element object
# Create output elementtree object
if config.inputRecursiveFlag or config.inputWrapperFlag:
# Name space already declared in results element, so no need to do it
# here
root = ET.Element('jpylyzer')
else:
root = ET.Element(
'jpylyzer', {'xmlns': 'http://openpreservation.org/ns/jpylyzer/',
'xmlns:xsi': 'http://www.w3.org/2001/XMLSchema-instance',
'xsi:schemaLocation': 'http://openpreservation.org/ns/jpylyzer/ http://jpylyzer.openpreservation.org/jpylyzer-v-1-1.xsd'})
# Create elements for storing tool, file and status meta info
toolInfo = ET.Element('toolInfo')
fileInfo = ET.Element('fileInfo')
statusInfo = ET.Element('statusInfo')
# If file name / path contain any surrogate pairs, remove them to
# avoid problems when writing to XML
fileNameCleaned = stripSurrogatePairs(fileName)
filePathCleaned = stripSurrogatePairs(filePath)
# Produce some general tool and file meta info
toolInfo.appendChildTagWithText("toolName", scriptName)
toolInfo.appendChildTagWithText("toolVersion", __version__)
fileInfo.appendChildTagWithText("fileName", fileNameCleaned)
fileInfo.appendChildTagWithText("filePath", filePathCleaned)
fileInfo.appendChildTagWithText(
"fileSizeInBytes", str(fileSizeInBytes))
fileInfo.appendChildTagWithText(
"fileLastModified", fileLastModifiedDate)
# Initialise success flag
success = True
try:
# Contents of file to memory map object
isValidJP2, tests, characteristics = BoxValidator("JP2", fileData).validate()
# Generate property values remap table
remapTable = generatePropertiesRemapTable()
# Create printable version of tests and characteristics tree
tests.makeHumanReadable()
characteristics.makeHumanReadable(remapTable)
except Exception as ex:
isValidJP2 = False
success = False
exceptionType = type(ex)
if exceptionType == MemoryError:
failureMessage = "memory error (file size too large)"
elif exceptionType == IOError:
failureMessage = "I/O error (cannot open file)"
elif exceptionType == RuntimeError:
failureMessage = "runtime error (please report to developers)"
else:
failureMessage = "unknown error (please report to developers)"
printWarning(failureMessage)
tests = ET.Element("tests")
characteristics = ET.Element('properties')
# Add status info
statusInfo.appendChildTagWithText("success", str(success))
if success == False:
statusInfo.appendChildTagWithText("failureMessage",failureMessage)
# Append all results to root
root.append(toolInfo)
root.append(fileInfo)
root.append(statusInfo)
root.appendChildTagWithText("isValidJP2", str(isValidJP2))
root.append(tests)
root.append(characteristics)
return(root)
def main():
# From where is this script executed?)
applicationPath=os.path.abspath(get_main_dir())
# Path to Preflight app
preflightApp=os.path.normpath(applicationPath+ "/preflight/preflight-1.8.0-20121114.230701-58-jar-with-dependencies.jar")
# Command to launch preflight (may not work on all platforms, ideally move
# this all to separate config file instead of hard-coding it)
preflightCommand="java -jar " + preflightApp
# Get input from command line
args=parseCommandLine()
dirIn=args.dirIn
# Check if dirIn is directory or file, and create list of all input files
if os.path.isfile(dirIn)==True:
myFilesIn=[dirIn]
elif os.path.isdir(dirIn)==True:
myFilesIn=getFilesFromTree(dirIn)
else:
msg=dirIn + " is not a directory or file!"
errorExit(msg)
# Create output elementtree object
root=ET.Element('preflight')
numberOfFiles=len(myFilesIn)
for i in range(0,numberOfFiles):
myFileIn=myFilesIn[i]
fileElt=ET.Element("file")
fileElt.appendChildTagWithText("fileName", myFileIn)
if myFileIn.lower().endswith(".pdf"):
systemString= preflightCommand + ' "' + myFileIn + '"'
p = sub.Popen(systemString,stdout=sub.PIPE,stderr=sub.PIPE)
output, errors = p.communicate()
# Exit status
preflightExitCode=p.returncode
#sys.stderr.write(str(preflightExitCode))
if preflightExitCode<=0:
# Value can be -1 in case of warnings
preflightExitStatus="Success"
else:
preflightExitStatus="Failure"
# Extract error codes and descriptions from output
errorCodesDescs=getErrors(output)
if preflightExitCode>0:
# Preflight raised an exception (we're ignoring warnings here!)
isValidPDFA1b="False"
# Append system errors to output
fileElt.appendChildTagWithText("sysErrors", errors)
elif len(errorCodesDescs)!=0:
# Errors were found, so not valid
isValidPDFA1b="False"
# Add errors codes + descriptions to output
fileElt.append(errorCodesDescs)
else:
# No errors found, so valid
isValidPDFA1b="True"
# Append validation outcome and preflightExitStatus to output
fileElt.appendChildTagWithText("preflightExitStatus", preflightExitStatus)
fileElt.appendChildTagWithText("isValidPDFA1b", isValidPDFA1b)
# Add output to root element
root.append(fileElt)
# Write xml-formatted log to stdout
sys.stdout.write(root.toxml().decode('UTF-8'))