def runAnnotate(reader, alg, paramDict, outName):
""" annotate all articles in reader
"""
tmpOutFname = makeLocalTempFile()
if outName=="stdout":
outFh = sys.stdout
else:
outFh = pubStore.utf8GzWriter(tmpOutFname)
doSectioning = attributeTrue(alg, "sectioning")
logging.debug("Sectioning activated: %s" % doSectioning)
if "startup" in dir(alg):
logging.debug("Running startup")
alg.startup(paramDict)
addFields = paramDict.get("addFields", [])
writeHeaders(alg, outFh, doSectioning, addFields)
annotIdAdd = getAnnotId(alg, paramDict)
onlyMain, onlyMeta, bestMain = getAlgPrefs(alg, paramDict)
#addSnippet = "snippet" in alg.headers
for articleData, fileDataList in reader.iterArticlesFileList(onlyMeta, bestMain, onlyMain):
logging.debug("Annotating article %s with %d files, %s" % \
(articleData.articleId, len(fileDataList), [x.fileId for x in fileDataList]))
for fileData in fileDataList:
writeAnnotations(alg, articleData, fileData, outFh, \
annotIdAdd, doSectioning, addFields)
if outName!="stdout":
outFh.close()
moveTempToFinal(tmpOutFname, outName)
def newTempOutFile(tmpFnames, outName, alg, addFields):
""" open a new temporary file on local disk and add it to the tmpFnames map
Write headers.
Returns a tuple outFh, tmpFnames where tmpFnames is a list (tempFilename, finalFilename)
"""
if outName=="stdout":
outFh = sys.stdout
return outFh, tmpFnames
tmpOutFname = makeLocalTempFile()
tmpFnames.append( tmpOutFname )
outFh = pubStore.utf8GzWriter(tmpOutFname)
if addFields!=None:
writeHeaders(alg, outFh, addFields)
return outFh, tmpFnames