pubLoadSolr

#!/usr/bin/env python

# a command line tool to load a pubtools database into solr

# load default python packages
from __future__ import print_function
import logging, optparse, os, shutil, glob, tempfile, sys
from os.path import *

# add <scriptDir>/lib/ to package search path
sys.path.insert(0, join(dirname(abspath(__file__)),"lib"))

# load our own libraries
import pubConf, pubGeneric, pubStore, maxCommon

#import scorched
from mysolr import Solr

# ===== FUNCTIONS =======

def loadIntoSolr(inDirs):
    " load data from inDirs into solr "
    solrUrl = pubConf.solrUrl
    solr = Solr(solrUrl, version=4)
    i = 0
    docs = []
    for inDir in inDirs:
        for article, fileList in pubStore.iterArticleDirList(inDir):
            i+=1
            articleInfo = article._asdict()
            for fileData in fileList:
                doc = dict(fileData._asdict())
                doc.update(articleInfo)

                newDoc = {}
                for key, val in doc.iteritems():
                    if key=="fileId":
                        newDoc["id"] = val
                    else:
                        newDoc[key+"_t"]=val
                docs.append(newDoc)

            if len(docs)>10000:
                print("COMMIT", i)
                print(solr.update(docs))
                docs = []

def main(args, options):
    pubGeneric.setupLoggingOptions(options)
    inDirs = pubGeneric.resolveDatasetDesc(args[0])
    loadIntoSolr(inDirs)

# === COMMAND LINE INTERFACE, OPTIONS AND HELP ===
parser = optparse.OptionParser("""usage: %prog [options] <inDirOrFiles> - load articles meta info and files into a solr server configured in pubConf.
""")

parser.add_option("-d", "--debug", dest="debug", action="store_true", help="show debug messages")
parser.add_option("-v", "--verbose", dest="verbose", action="store_true", help="show more debug messages")
(options, args) = parser.parse_args()

if args==[]:
    parser.print_help()
    exit(1)

main(args, options)