def add_category(request):
'''
添加标注类型
:param request:
:return:
'''
if request.method == "POST":
new_ctg = request.POST.get("newctg", "")
utils.get_database(request.session.get(utils.SESSION_DB, "")).update_categories(new_ctg)
sug_category = {}
sug_category["category"] = utils.get_database(request.session.get(utils.SESSION_DB, "")).get_categories()
count = {}
for s in sug_category["category"]:
count[s] = 0
sug_data = utils.get_database(request.session.get(utils.SESSION_DB, "")).get_suggests()
for line in sug_data:
count[line["sug"]] += 1
sug_category["counts"] = count
utils.log_category_info(request.session.get(utils.SESSION_USER, ""), "添加标注类型",
request.session.get(utils.SESSION_DB, ""), new_ctg)
# 更新服务
requests.post(utils.update_sug_url(request.session.get(utils.SESSION_DB, "")),
data=request.session.get(utils.SESSION_DB, ""), headers=utils.headers)
return HttpResponse(json.dumps(sug_category), content_type='application/json')
def suggest(request):
if not utils.check_user_session(request):
return HttpResponseRedirect("/login/")
# 进入标注界面,更新服务,按所选类型切换标注术语集
requests.post(utils.update_sug_url(request.session.get(utils.SESSION_DB, "")),
data=request.session.get(utils.SESSION_DB, ""), headers=utils.headers)
return render_to_response("suggest.html")
def delete_selected_sugs(sugs, db):
'''
数据库删除分词标注，更新服务数据
:param sugs:
:param db: 术语集名称
:return:
'''
utils.get_database(db).delete_suggests(sugs)
requests.post(utils.update_sug_url(db), data=db, headers=utils.headers)
def delete_sug_category(request):
'''
删除某个标注和该标注下的分词
:param request:
:return:
'''
if request.method == "POST":
category = request.POST.get("category", "")
utils.get_database(request.session.get(utils.SESSION_DB, "")).delete_category(category)
utils.get_database(request.session.get(utils.SESSION_DB, "")).delete_items_by_sug(category)
# 更新服务
requests.post(utils.update_sug_url(request.session.get(utils.SESSION_DB, "")),
data=request.session.get(utils.SESSION_DB, ""), headers=utils.headers)
utils.log_category_info(request.session.get(utils.SESSION_USER, ""), "删除标注类型",
request.session.get(utils.SESSION_DB, ""), category)
return HttpResponse("", content_type='application/test')
def add_segs_sugs(request):
'''
从新增数据添加分词标注
:param request: msgs:{sug:"人工-中心词,髋关节-部位"}
:return:
'''
if request.method == "POST":
data = eval(request.POST.get("msg", ""))
res = add_sugs(data["sug"], request.session.get(utils.SESSION_DB, ""))
add_segs(request.session.get(utils.SESSION_DB, ""))
logger = logging.getLogger(utils.SUGGEST_LOG)
for sug, old_sug in res.iteritems():
if old_sug == "":
op = "确认标注"
else:
op = "更新标注"
old_sug += "=>"
utils.log_data_info(logger,
request.session.get(utils.SESSION_USER,
""), "手动",
request.session.get(utils.SESSION_DB, ""), op,
old_sug + sug)
d = refresh_datafile_sug({}, request.session.get(utils.SESSION_DB, ""))
# 更新服务
requests.post(utils.update_seg_url(
request.session.get(utils.SESSION_DB, "")),
data="",
headers=utils.headers)
requests.post(utils.update_sug_url(
request.session.get(utils.SESSION_DB, "")),
data="",
headers=utils.headers)
return HttpResponse(json.dumps(d), content_type='application/json')
def update_segs_sugs(request):
'''
修改原始数据
:param request: source:origin/new
msgs:dict,[seg:sug]
:return:
'''
if request.method == "POST":
msgs = eval(request.POST.get("msg", "")) # {"高血压":"中心词"}
logger = logging.getLogger(utils.SUGGEST_LOG)
for seg, sug in msgs.iteritems():
old_sug = utils.get_database(
request.session.get(utils.SESSION_DB,
"")).get_suggest_from_seg(seg)[0]["sug"]
utils.log_data_info(logger,
request.session.get(utils.SESSION_USER,
""), "手动",
request.session.get(utils.SESSION_DB,
""), "更新原始数据",
seg + ":" + old_sug + "=>" + seg + ":" + sug)
# 先获得更新的词原先的标注，用于写入日志，再更新数据库
utils.get_database(request.session.get(
utils.SESSION_DB, "")).update_single_sug_category(msgs)
requests.post(utils.update_sug_url(
request.session.get(utils.SESSION_DB, "")),
data="",
headers=utils.headers)
requests.post(utils.update_seg_url(
request.session.get(utils.SESSION_DB, "")),
data="",
headers=utils.headers)
data = init_origin_data({}, request.session.get(utils.SESSION_DB, ""))
return HttpResponse(json.dumps(data), content_type='application/json')
def upload_data_file(request):
'''
上传分词/标注数据,txt/csv格式
分词-标注数据库已有，跳过
分词-标注数据库没有，写入
分词-标注和数据库的不一样，提示是否覆盖
？只有分词，没有标注：未知
错误数据（格式不对，标注类型不对），不能写入
:param request:
:return:
'''
# 上传文件信息写入log
# upload_filename = request.FILES.get("myfile", None)
# name = upload_filename.name
# utils.logger_file_info(request.session.get(utils.SESSION_USER, ""), "上传分词标注数据",
# request.session.get(utils.SESSION_DB, ""), name)
checked = request.POST.get("checked", "") # 同一分词，不同标注，是否覆盖原始数据
if checked == '0':
checked = False
else:
checked = True
all_categories = utils.get_database(
request.session.get(utils.SESSION_DB, "")).get_categories()
i = 0
# 上传标注
for line in open("tmp.csv").readlines():
line = line.strip()
if len(line) > 1:
if len(line.split("\t")) != 2:
pass
else:
seg, sug = line.split("\t")
if sug in all_categories: # 添加的标注需属于已有标注
# 添加数据
utils.get_database(
request.session.get(utils.SESSION_DB,
"")).insert_suggests(
{
dbinfo.SUG_SEG: seg,
dbinfo.SUG_SUG: sug,
dbinfo.SEG_SOURCE: "",
dbinfo.SUG_SOURCE: "",
dbinfo.SUG_STATE: "已存",
dbinfo.SUG_COUNT: 1
},
cover=checked)
utils.log_sug_info(
request.session.get(utils.SESSION_USER, ""), "从数据文件",
request.session.get(utils.SESSION_DB, ""), "添加标注",
seg + ":" + sug)
i += 1
requests.post(utils.update_seg_url(
request.session.get(utils.SESSION_DB, "")),
data="",
headers=utils.headers)
requests.post(utils.update_sug_url(
request.session.get(utils.SESSION_DB, "")),
data="",
headers=utils.headers)
# 更新数据
data = init_origin_data({}, request.session.get(utils.SESSION_DB, ""))
# 这里按sug排序
request.session[utils.SESSION_ALLDATA] = build_sug_dict(data["items"])
return HttpResponse(json.dumps(data), content_type='application/json')
def delete_segs_sugs(request):
'''
新增数据的删除
:param request: msg:{sug:"多根-其他,双坐骨-部位,..."} {0:{seg:,sug:},1:{seg:,sug:}}
:return:
'''
if request.method == "POST":
msgs = eval(request.POST.get("msg", ""))
dic = {}
# 删除新增数据
if "sug" in msgs.keys():
sugs = msgs["sug"]
for s in sugs.split(","):
if s:
idx = s.rfind("-") # 最后一个"-"是分隔分词与标注，分词有可能带有"-"
dic['seg'], dic['sug'] = s[:idx], s[idx + 1:len(s)]
utils.get_database(
request.session.get(utils.SESSION_DB,
"")).delete_suggests(dic)
logger = logging.getLogger(utils.SUGGEST_LOG)
utils.log_data_info(
logger, request.session.get(utils.SESSION_USER,
""), "手动",
request.session.get(utils.SESSION_DB, ""), "删除新增数据",
dic['seg'] + ":" + dic['sug'])
# 删除原始数据
else:
for k, v in msgs.iteritems():
utils.get_database(request.session.get(utils.SESSION_DB,
"")).delete_suggests(v)
logger = logging.getLogger(utils.SUGGEST_LOG)
utils.log_data_info(
logger, request.session.get(utils.SESSION_USER, ""), "手动",
request.session.get(utils.SESSION_DB, ""), "删除原始数据",
v["seg"] + ":" + v["sug"])
data = init_origin_data({}, request.session.get(utils.SESSION_DB, ""))
data = refresh_datafile_sug(data,
request.session.get(utils.SESSION_DB, ""))
# 这里按sug排序
request.session[utils.SESSION_ALLDATA] = build_sug_dict(data["items"])
# 更新服务
requests.post(utils.update_seg_url(
request.session.get(utils.SESSION_DB, "")),
data="",
headers=utils.headers)
requests.post(utils.update_sug_url(
request.session.get(utils.SESSION_DB, "")),
data="",
headers=utils.headers)
if request.session.get(utils.SESSION_DB, "") == "zhenduan":
url = utils.seg_service_url_zd
elif request.session.get(utils.SESSION_DB, "") == "shoushu":
url = utils.seg_service_url_ss
requests.post(url,
data=request.session.get(utils.SESSION_DB, ""),
headers=utils.headers)
if request.session.get(utils.SESSION_DB, "") == "zhenduan":
url = utils.sug_service_url_zd
elif request.session.get(utils.SESSION_DB, "") == "shoushu":
url = utils.sug_service_url_ss
requests.post(url,
data=request.session.get(utils.SESSION_DB, ""),
headers=utils.headers)
return HttpResponse(json.dumps(data), content_type='application/json')