def prepareGoldForEvaluation(corpus,
sourceDir,
tempdir,
goldDir=None,
goldPackage=None):
if os.path.exists(os.path.join(tempdir, "gold")):
shutil.rmtree(os.path.join(tempdir, "gold"))
if goldDir == None:
if not hasattr(Settings, "BIONLP_EVALUATOR_GOLD_DIR"):
print >> sys.stderr, corpus, "BIONLP_EVALUATOR_GOLD_DIR setting not defined"
return None
if goldPackage == None:
goldPackage = Settings.EVALUATOR[corpus + "_DEVEL-gold"]
goldDir = os.path.join(Settings.BIONLP_EVALUATOR_GOLD_DIR, goldPackage)
if not os.path.exists(goldDir):
print >> sys.stderr, corpus, "Evaluator gold data directory", goldDir, "does not exist"
goldDir = None
if goldDir != None and goldDir.endswith(".tar.gz") or goldDir.endswith(
".tgz"):
goldDir = Download.getTopDir(
os.path.join(tempdir, "gold"),
Download.extractPackage(goldDir, os.path.join(tempdir, "gold")))
print >> sys.stderr, "Uncompressed evaluation gold to", goldDir
if goldDir != None and not hasGoldDocuments(sourceDir, goldDir):
print >> sys.stderr, "Evaluation input has no gold documents"
goldDir = None
return goldDir
def install(self, destDir=None, downloadDir=None, redownload=False, updateLocalSettings=False):
url = Settings.URL["BLLIP_SOURCE"]
if downloadDir == None:
downloadDir = os.path.join(Settings.DATAPATH) + "/tools/download"
if destDir == None:
destDir = Settings.DATAPATH + "/tools/BLLIP"
items = Download.downloadAndExtract(url, destDir, downloadDir + "/bllip.zip", None, False)
print >> sys.stderr, "Installing BLLIP parser"
Tool.testPrograms("BLLIP parser", ["make", "flex"], {"flex":"flex --version"})
parserPath = Download.getTopDir(destDir, items)
cwd = os.getcwd()
os.chdir(parserPath)
print >> sys.stderr, "Compiling first-stage parser"
subprocess.call("make", shell=True)
print >> sys.stderr, "Compiling second-stage parser"
subprocess.call("make reranker", shell=True)
os.chdir(cwd)
print >> sys.stderr, "Installing the McClosky biomedical parsing model"
url = "http://bllip.cs.brown.edu/download/bioparsingmodel-rel1.tar.gz"
Download.downloadAndExtract(url, destDir, downloadDir, None)
bioModelDir = os.path.abspath(destDir + "/biomodel")
# Check that everything works
Tool.finalizeInstall(["first-stage/PARSE/parseIt", "second-stage/programs/features/best-parses"],
{"first-stage/PARSE/parseIt":"first-stage/PARSE/parseIt " + bioModelDir + "/parser/ < /dev/null",
"second-stage/programs/features/best-parses":"second-stage/programs/features/best-parses -l " + bioModelDir + "/reranker/features.gz " + bioModelDir + "/reranker/weights.gz < /dev/null"},
parserPath, {"BLLIP_PARSER_DIR":os.path.abspath(parserPath),
"MCCLOSKY_BIOPARSINGMODEL_DIR":bioModelDir}, updateLocalSettings)
def install(destDir=None, downloadDir=None, redownload=False, updateLocalSettings=False):
url = Settings.URL["BLLIP_SOURCE"]
if downloadDir == None:
downloadDir = os.path.join(Settings.DATAPATH) + "/tools/download"
if destDir == None:
destDir = Settings.DATAPATH + "/tools/BLLIP"
items = Download.downloadAndExtract(url, destDir, downloadDir + "/bllip.zip", None, False)
print >> sys.stderr, "Installing BLLIP parser"
Tool.testPrograms("BLLIP parser", ["make", "flex"], {"flex":"flex --version"})
parserPath = Download.getTopDir(destDir, items)
cwd = os.getcwd()
os.chdir(parserPath)
print >> sys.stderr, "Compiling first-stage parser"
subprocess.call("make", shell=True)
print >> sys.stderr, "Compiling second-stage parser"
subprocess.call("make reranker", shell=True)
os.chdir(cwd)
print >> sys.stderr, "Installing the McClosky biomedical parsing model"
url = "http://bllip.cs.brown.edu/download/bioparsingmodel-rel1.tar.gz"
Download.downloadAndExtract(url, destDir, downloadDir, None)
bioModelDir = os.path.abspath(destDir + "/biomodel")
# Check that everything works
Tool.finalizeInstall(["first-stage/PARSE/parseIt", "second-stage/programs/features/best-parses"],
{"first-stage/PARSE/parseIt":"first-stage/PARSE/parseIt " + bioModelDir + "/parser/ < /dev/null",
"second-stage/programs/features/best-parses":"second-stage/programs/features/best-parses -l " + bioModelDir + "/reranker/features.gz " + bioModelDir + "/reranker/weights.gz < /dev/null"},
parserPath, {"BLLIP_PARSER_DIR":os.path.abspath(parserPath),
"MCCLOSKY_BIOPARSINGMODEL_DIR":bioModelDir}, updateLocalSettings)
def install(destDir=None, downloadDir=None, redownload=False, updateLocalSettings=False):
print >> sys.stderr, "Installing Stanford Parser"
if downloadDir == None:
downloadDir = os.path.join(Settings.DATAPATH, "tools/download/")
if destDir == None:
destDir = os.path.join(Settings.DATAPATH, "tools/")
items = Download.downloadAndExtract(Settings.URL["STANFORD_PARSER"], destDir, downloadDir)
stanfordPath = Download.getTopDir(destDir, items)
Tool.finalizeInstall(["stanford-parser.jar"],
{"stanford-parser.jar":"java -cp stanford-parser.jar:slf4j-api.jar edu.stanford.nlp.trees.EnglishGrammaticalStructure"},
stanfordPath, {"STANFORD_PARSER_DIR":stanfordPath}, updateLocalSettings)
def install(self, destDir=None, downloadDir=None, redownload=False, updateLocalSettings=False):
print >> sys.stderr, "Installing Stanford Parser"
if downloadDir == None:
downloadDir = os.path.join(Settings.DATAPATH, "tools/download/")
if destDir == None:
destDir = os.path.join(Settings.DATAPATH, "tools/")
items = Download.downloadAndExtract(Settings.URL["STANFORD_PARSER"], destDir, downloadDir)
stanfordPath = Download.getTopDir(destDir, items)
Tool.finalizeInstall(["stanford-parser.jar"],
{"stanford-parser.jar":"java -cp stanford-parser.jar edu.stanford.nlp.trees.EnglishGrammaticalStructure"},
stanfordPath, {"STANFORD_PARSER_DIR":stanfordPath}, updateLocalSettings)
def install(destDir=None, downloadDir=None, redownload=False):
print >> sys.stderr, "Installing BioNLP'11 evaluators"
settings = {}
if downloadDir == None:
downloadDir = Settings.DATAPATH
if destDir == None:
destDir = Settings.DATAPATH
for corpus in ["GE", "BB", "BI", "CO"]:
print >> sys.stderr, "Installing BioNLP'11", corpus, "evaluator"
settings[corpus + "_EVALUATOR"] = Download.getTopDir(destDir + "/tools/evaluators/", Download.downloadAndExtract(Settings.URL[corpus + "_EVALUATOR"], destDir + "/tools/evaluators/", downloadDir + "/tools/download/"))
print >> sys.stderr, "Installing BioNLP'11", corpus, "evaluator gold data"
Download.downloadAndExtract(Settings.URL[corpus + "_DEVEL"], destDir + "/tools/evaluators/gold/" + corpus + "-devel", downloadDir + "/corpora/BioNLP11-original/corpus/", os.path.basename(Settings.URL[corpus + "_DEVEL"])[:-len(".tar.gz")])
return settings
def checkEvaluator(corpus, sourceDir, goldDir=None):
# Check evaluator
if not hasattr(Settings, "BIONLP_EVALUATOR_DIR"):
print >> sys.stderr, corpus, "BIONLP_EVALUATOR_DIR setting not defined"
evaluatorDir = None
else:
evaluatorDir = os.path.join(Settings.BIONLP_EVALUATOR_DIR,
Settings.EVALUATOR[corpus])
# Check source data
tempdir = None
if sourceDir.endswith(".tar.gz"):
tempdir = tempfile.mkdtemp()
Download.extractPackage(sourceDir, os.path.join(tempdir, "source"))
sourceDir = os.path.join(tempdir, "source")
else: #if corpus in ("GE09", "BB11", "BI11"):
# GE09 a2 files have to be renamed and relation identifier "R" has to be replaced with "E" for the BB11 and BI11 relations.
# X-lines have to be removed from all tasks
tempdir = tempfile.mkdtemp()
shutil.copytree(sourceDir, os.path.join(tempdir, "source"))
sourceDir = os.path.join(tempdir, "source")
# Filter extra data
removeXLines(sourceDir)
# Check gold data
if goldDir == None:
if not hasattr(Settings, "BIONLP_EVALUATOR_GOLD_DIR"):
print >> sys.stderr, corpus, "BIONLP_EVALUATOR_GOLD_DIR setting not defined"
return evaluatorDir, None
goldDir = os.path.join(Settings.BIONLP_EVALUATOR_GOLD_DIR,
Settings.EVALUATOR[corpus + "-gold"])
if not os.path.exists(goldDir):
print >> sys.stderr, corpus, "Evaluator gold data directory", goldDir, "does not exist"
goldDir = None
if goldDir != None and goldDir.endswith(".tar.gz"):
if tempdir == None:
tempdir = tempfile.mkdtemp()
goldDir = Download.getTopDir(
os.path.join(tempdir, "gold"),
Download.extractPackage(goldDir, os.path.join(tempdir, "gold")))
print >> sys.stderr, "Uncompressed evaluation gold to", goldDir
if goldDir != None and not hasGoldDocuments(sourceDir, goldDir):
print >> sys.stderr, "Evaluation input has no gold documents"
goldDir = None
# Use absolute paths
sourceDir = os.path.abspath(sourceDir)
if evaluatorDir != None:
evaluatorDir = os.path.abspath(evaluatorDir)
if goldDir != None:
goldDir = os.path.abspath(goldDir)
if tempdir != None:
tempdir = os.path.abspath(tempdir)
return evaluatorDir, sourceDir, goldDir, tempdir
def checkEvaluator(corpus, sourceDir, goldDir = None):
# Check evaluator
if not hasattr(Settings, "BIONLP_EVALUATOR_DIR"):
print >> sys.stderr, corpus, "BIONLP_EVALUATOR_DIR setting not defined"
evaluatorDir = None
else:
evaluatorDir = os.path.join(Settings.BIONLP_EVALUATOR_DIR, Settings.EVALUATOR[corpus])
# Check source data
tempdir = None
if sourceDir.endswith(".tar.gz"):
tempdir = tempfile.mkdtemp()
Download.extractPackage(sourceDir, os.path.join(tempdir, "source"))
sourceDir = os.path.join(tempdir, "source")
else: #if corpus in ("GE09", "BB11", "BI11"):
# GE09 a2 files have to be renamed and relation identifier "R" has to be replaced with "E" for the BB11 and BI11 relations.
# X-lines have to be removed from all tasks
tempdir = tempfile.mkdtemp()
shutil.copytree(sourceDir, os.path.join(tempdir, "source"))
sourceDir = os.path.join(tempdir, "source")
# Filter extra data
removeXLines(sourceDir)
# Check gold data
if goldDir == None:
if not hasattr(Settings, "BIONLP_EVALUATOR_GOLD_DIR"):
print >> sys.stderr, corpus, "BIONLP_EVALUATOR_GOLD_DIR setting not defined"
return evaluatorDir, None
goldDir = os.path.join(Settings.BIONLP_EVALUATOR_GOLD_DIR, Settings.EVALUATOR[corpus + "-gold"])
print >> sys.stderr, "Found gold data directory", goldDir
if not os.path.exists(goldDir):
print >> sys.stderr, corpus, "Evaluator gold data directory", goldDir, "does not exist"
goldDir = None
if goldDir != None and goldDir.endswith(".tar.gz"):
if tempdir == None:
tempdir = tempfile.mkdtemp()
goldDir = Download.getTopDir(os.path.join(tempdir, "gold"), Download.extractPackage(goldDir, os.path.join(tempdir, "gold")))
print >> sys.stderr, "Uncompressed evaluation gold to", goldDir
if goldDir != None and not hasGoldDocuments(sourceDir, goldDir):
print >> sys.stderr, "Evaluation input has no gold documents"
goldDir = None
# Use absolute paths
sourceDir = os.path.abspath(sourceDir)
if evaluatorDir != None:
evaluatorDir = os.path.abspath(evaluatorDir)
if goldDir != None:
goldDir = os.path.abspath(goldDir)
if tempdir != None:
tempdir = os.path.abspath(tempdir)
return evaluatorDir, sourceDir, goldDir, tempdir
def checkEvaluator(corpus, sourceDir, goldDir=None):
# Check evaluator
if not hasattr(Settings, "BIONLP_EVALUATOR_DIR"):
print >> sys.stderr, corpus, "BIONLP_EVALUATOR_DIR setting not defined"
evaluatorDir = None
else:
evaluatorDir = os.path.join(Settings.BIONLP_EVALUATOR_DIR, Settings.EVALUATOR[corpus])
# Check source data
tempdir = None
if sourceDir.endswith(".tar.gz"):
tempdir = tempfile.mkdtemp()
Download.extractPackage(sourceDir, os.path.join(tempdir, "source"))
sourceDir = os.path.join(tempdir, "source")
elif corpus == "GE09": # a2 files have to be renamed
tempdir = tempfile.mkdtemp()
shutil.copytree(sourceDir, os.path.join(tempdir, "source"))
sourceDir = os.path.join(tempdir, "source")
# Check gold data
if goldDir == None:
if not hasattr(Settings, "BIONLP_EVALUATOR_GOLD_DIR"):
print >> sys.stderr, corpus, "BIONLP_EVALUATOR_GOLD_DIR setting not defined"
return evaluatorDir, None
goldDir = os.path.join(Settings.BIONLP_EVALUATOR_GOLD_DIR, Settings.EVALUATOR[corpus + "-gold"])
if not os.path.exists(goldDir):
print >> sys.stderr, corpus, "Evaluator gold data directory", goldDir, "does not exist"
goldDir = None
if goldDir != None and goldDir.endswith(".tar.gz"):
if tempdir == None:
tempdir = tempfile.mkdtemp()
goldDir = Download.getTopDir(
os.path.join(tempdir, "gold"), Download.extractPackage(goldDir, os.path.join(tempdir, "gold"))
)
print >> sys.stderr, "Uncompressed evaluation gold to", goldDir
if goldDir != None and not hasGoldDocuments(sourceDir, goldDir):
print >> sys.stderr, "Evaluation input has no gold documents"
goldDir = None
# Use absolute paths
sourceDir = os.path.abspath(sourceDir)
if evaluatorDir != None:
evaluatorDir = os.path.abspath(evaluatorDir)
if goldDir != None:
goldDir = os.path.abspath(goldDir)
if tempdir != None:
tempdir = os.path.abspath(tempdir)
return evaluatorDir, sourceDir, goldDir, tempdir
def prepareGoldForEvaluation(corpus, sourceDir, tempdir, goldDir=None, goldPackage=None):
if os.path.exists(os.path.join(tempdir, "gold")):
shutil.rmtree(os.path.join(tempdir, "gold"))
if goldDir == None:
if not hasattr(Settings, "BIONLP_EVALUATOR_GOLD_DIR"):
print >> sys.stderr, corpus, "BIONLP_EVALUATOR_GOLD_DIR setting not defined"
return None
if goldPackage == None:
goldPackage = Settings.EVALUATOR[corpus + "_DEVEL-gold"]
goldDir = os.path.join(Settings.BIONLP_EVALUATOR_GOLD_DIR, goldPackage)
if not os.path.exists(goldDir):
print >> sys.stderr, corpus, "Evaluator gold data directory", goldDir, "does not exist"
goldDir = None
if goldDir != None and goldDir.endswith(".tar.gz") or goldDir.endswith(".tgz"):
goldDir = Download.getTopDir(os.path.join(tempdir, "gold"), Download.extractPackage(goldDir, os.path.join(tempdir, "gold")))
print >> sys.stderr, "Uncompressed evaluation gold to", goldDir
if goldDir != None and not hasGoldDocuments(sourceDir, goldDir):
print >> sys.stderr, "Evaluation input has no gold documents"
goldDir = None
return goldDir
def install(destDir=None, downloadDir=None, redownload=False):
print >> sys.stderr, "Installing BioNLP'11 evaluators"
settings = {}
if downloadDir == None:
downloadDir = Settings.DATAPATH
if destDir == None:
destDir = Settings.DATAPATH
for corpus in ["GE11", "BB11", "BI11", "CO11"]:
print >> sys.stderr, "Installing BioNLP'11", corpus, "evaluator"
settings[corpus + "_EVALUATOR"] = Download.getTopDir(
destDir + "/tools/evaluators/",
Download.downloadAndExtract(Settings.URL[corpus + "_EVALUATOR"],
destDir + "/tools/evaluators/",
downloadDir + "/tools/download/"))
print >> sys.stderr, "Installing BioNLP'11", corpus, "evaluator gold data"
Download.downloadAndExtract(
Settings.URL[corpus + "_DEVEL"],
destDir + "/tools/evaluators/gold/" + corpus + "-devel",
downloadDir + "/corpora/BioNLP11-original/corpus/",
os.path.basename(Settings.URL[corpus +
"_DEVEL"])[:-len(".tar.gz")])
return settings
