#!/usr/bin/python import cgi import time import sys import glob import os import re sys.path.append("py") import json def getDataFile(filename, dir=None): if dir == None: dir = filename return "data/" + dir + '/' + filename def getHeadDB(file): return getDataFile(file)+".head.json" def getLatestDBFilename(file): ver = getNextVersion(file) if ver.find('.')>=0: # old version format return getDBFilename(file, ver) ver = int(ver) while ver > 1: ver-=1 filename = getDBFilename(file, ver) if os.path.exists(filename): return filename return getOCRDBFilename(file) def getOCRDBFilename(file): return getDataFile(file)+".ocr.json" def getDBFilename(file, ver): if ver == 'ocr': return getOCRDBFilename(file) return getDataFile(file) + '-' + str(ver) + '.json' def getVersionFromFilename(filename): if filename.rfind('ocr.json')>=0: return 'ocr' dotpos = filename.rfind('.') if dotpos < 0: return filename[filename.rfind('-')+1:] else: return filename[filename.rfind('-')+1:filename.rfind('.')] def _patch(latestDB, file, logs): f = open(latestDB, "r") rects = json.read(f.read()) f.close() if 'rects' in rects: rects = rects['rects'] logs = json.read(logs) offX = logs['visArea']['x'] offY = logs['visArea']['y'] for log in logs['log']: r = {'x':offX+log['x'], 'y':offY+log['y'], 'width':log['w'], 'height':log['h'], 'text':log['val']} if log['id'] == 'new': if log['val'] != "": id = len(rects) r['new'] = 1 rects[str(id)] = r else: if log['val'] == "": r['del'] = 1 rects[str(log['id'])] = r return rects def patch(file, logs, newfile): latestDB = getLatestDBFilename(file) rects = _patch(latestDB, file, logs) out = {} out['parent'] = getVersionFromFilename(latestDB) out['ver'] = getVersionFromFilename(newfile) out['rects'] = rects f = open(newfile, "w") f.write(json.write(out)) f.close() def readLatestDB(file): f = open(getLatestDBFilename(file), "r") ret = f.read() f.close() return ret def getVisArea(file,ver): filename = glob.glob(getDataFile(file)+"-"+ ver +'-*.log')[0] f = open(filename,"r") logs = json.read(f.read()) f.close() return json.write(logs['visArea']) def getLogFromVersion(file, ver): filename = glob.glob(getDataFile(file)+"-"+ ver +'-*.log')[0] f = open(filename,"r") logs = f.read() f.close() return logs def getLogFromAssignmentId(hitId, dir = '*'): flist = glob.glob('data/'+dir+'/*-*.log') #fixme for file in flist: f = open(file,"r") data = json.read(f.read()) f.close() if 'assignment' in data and data['assignment'] == hitId: m = re.search('.*/(.*)-(\d+)-\d+\.\d+\.log$', file) fname = m.group(1) ver = m.group(2) return json.write({'file':fname, 'ver':ver}) def getNextVersion(dir, increment=False): file = getDataFile("latest-version",dir) if os.path.exists(file): f = open(file, 'r') line = f.readline().rstrip() f.close() ver = line else: ver = "1" if increment: f = open(file, 'w') f.write(str(int(ver)+1)) f.close() return ver def writelog(dir, txtlog, oldfile, newfile): log = json.read(txtlog) log['parent'] = getVersionFromFilename(oldfile) log['ver'] = getVersionFromFilename(newfile) filename = getDataFile( newfile, dir)+ '-' + str(time.time()) + '.log' f = open(filename,"w") f.write(json.write(log)) f.close() def readDB(file, ver): filename = getDBFilename(file, ver) if os.path.exists(filename): f = open(filename,"r") ret = f.read() f.close() return ret else: txtlog = getLogFromVersion(file, ver) log = json.read(txtlog) if 'parent' in log: latestDB = getDBFilename(file, log['parent']) else: latestDB = getLatestDBFilename(file) rects = _patch(latestDB, file, txtlog) out = {} out['parent'] = getVersionFromFilename(latestDB) out['ver'] = ver out['rects'] = rects return json.write(out) def getReviewLog(file): filename = getDataFile('reviewLog',file) if os.path.exists(filename): f = open(filename,"r") ret = f.read() f.close() return json.read(ret) else: return {'approve':{}, 'reject':{}} def writeReviewLog(file, log): filename = getDataFile('reviewLog',file) f = open(filename,"w") f.write(json.write(log)) f.close() def getDimension(file): filename = getDataFile('dimension',file) f = open(filename,"r") ret = f.read() f.close() return ret def reviewSubmission(file, ver, approve, turkLog=None): reviewLog = getReviewLog(file) if approve == 'yes': action = 'approve' else: action = 'reject' if not turkLog: turkLog = ver reviewLog[action][ver] = turkLog writeReviewLog(file, reviewLog) if approve == 'no': return log = getLogFromVersion(file, ver) ver = getNextVersion(file, True) headDB = getDBFilename(file, ver) patch(file, log, headDB) def main(): form = cgi.FieldStorage() FromTurk = False ApproveAll = False if 'fromTurk' in form: FromTurk = True if 'key' in form and form['key'].value == 'UIDER': ApproveAll = True if 'approve' in form: if 'assignment' in form: jFileVer = getLogFromAssignmentId(form['assignment'].value, form['db'].value) FileVer = json.read(jFileVer) ver = FileVer['ver'] else: ver = form['ver'].value reviewSubmission(form['db'].value, ver, form['approve'].value) print "You answer \"" + form['approve'].value + '\" has been saved.' return if "assignment" in form: file = '*' if 'db' in form: file = form['db'].value print getLogFromAssignmentId(form['assignment'].value, file) return if "db" in form: if 'dimension' in form: print getDimension(form['db'].value) return if "ver" in form: if 'visArea' in form: print getVisArea(form['db'].value,form['ver'].value) else: print readDB(form['db'].value, form['ver'].value) else: print readLatestDB(form['db'].value) return if not "file" in form: print "error" return dir = form['file'].value newVersion = str(getNextVersion(dir,True)) newfile = dir + '-' + newVersion latestDB = getLatestDBFilename(dir) writelog(dir, form['log'].value, latestDB, newfile) if not FromTurk and ApproveAll: print "approved" reviewSubmission(dir, newVersion, "yes") print "saved!" print "Content-type: text/html\n\n" #try: # main() #except Exception,e: # print "failed " + str(e) main()