def getSeedAddForm(param_dic, dao_util):
sc_type = param_dic["sc_type"]
field_info_dic = getFieldInfoDic("crawl", "seed", sc_type)
param_dic["_data_type"] = "seed"
load_data = dao_util.getDataList(param_dic, "unique")
data_dic = getDataDic(load_data, field_info_dic["select"])
if data_dic["domain_id"]:
param_dic["domain_id"] = data_dic["domain_id"]
elif param_dic["domain_id"]:
data_dic["domain_id"] = param_dic["domain_id"]
seed_type_dic = getDataMappingDic(param_dic["_service_type"], param_dic["_data_type"], "seed_type")
html_seed_type = "<option value='%s'>%s</option>"%(data_dic["seed_type"], data_dic["seed_type"])
for seedTypeData in seed_type_dic.items():
html_seed_type += "<option value='%s'>%s</option>"%seedTypeData
status_dic = getDataMappingDic(param_dic["_service_type"], param_dic["_data_type"], "crawl_status")
html_status = "<option value='%s'>%s</option>"%(data_dic["crawl_status"], data_dic["crawl_status"])
for statusData in status_dic.items():
html_status += "<option value='%s'>%s</option>"%statusData
added_form = ""
if sc_type == "oi":
added_form += """
<tr> <td>모바일URL</td> <td><input type='text' name='mobile_url' class='textInput' value='%s' style="width:200px;"/></td> </tr>
<tr> <td>토픽</td> <td><input type='text' name='topic' class='textInput' value='%s' style="width:200px;"/></td> </tr>
"""%(data_dic["mobile_url"], data_dic["topic"])
elif sc_type == "blog":
added_form += """
<tr> <td>블로그 ID</td> <td><input type='text' name='blog_id' class='textInput' value='%s' style="width:100px;"/></td> </tr>
<tr> <td>블로그 링크</td> <td><input type='text' name='link' class='textInput' value='%s' style="width:200px;"/></td> </tr>
"""%(data_dic["blog_id"], data_dic["link"])
html_form = """
<form id='seedAddForm' action='/moon/action/sc_common.py' method='post'>
<input type='hidden' name='cmd' value='addUpdate' />
<input type='hidden' name='_data_type' value='seed' />
<input type='hidden' name='sc_type' value='%s' />
<input type='hidden' name='seed_id' value='%s' />
<input type='hidden' name='domain_id' value='%s' />
"""%(sc_type, data_dic["seed_id"], data_dic["domain_id"])+"""
<table>
<tr style='color:red; font-weight:800;'>
<td>씨드ID</td> <td>도메인ID</td> <td>뷰URL 수</td>
<td>신규URL<br/>미발견 횟수</td> <td>마지막방문</td> <td>Generator</td>
</tr>
<tr>
<td>%s</td> <td>%s</td> <td>%s</td> <td>%s</td> <td>%s</td> <td>%s</td>
</tr>
</table>
"""%(data_dic["seed_id"],data_dic["domain_id"],data_dic["view_url_count"],data_dic["no_new_count"],getDateByStamp(data_dic["last_visit"]),data_dic["generator"])+"""
<table class="formTable">
<tr> <td>씨드URL GUID</td> <td><input type='text' name='seed_guid' class='textInput' value='%s' style="width:300px;"/></td> </tr>
<tr> <td>다운로드URL</td> <td><input type='text' name='down_url' class='textInput' value='%s' style="width:300px;"/></td> </tr>
<tr> <td>씨드명</td> <td><input type='text' name='seed_name' class='textInput' value='%s' style="width:200px;"/></td> </tr>
<tr> <td>INT_ID최대값</td> <td><input type='text' name='max_no' class='textInput' value='%s' style="width:200px;"/></td> </tr>
<tr> <td>씨드타입</td> <td><select form='seedAddForm' name='seed_type'>%s</select> 기본값 : 숫자(N)</td>
<tr> <td>크롤상태</td> <td><select form='seedAddForm' name='crawl_status'>%s</select> 기본값 : 준비중(W)</td>
%s
</table>
<input type="submit" value="등록 (혹은 수정)" style=" width:420px; font-weight:800; color:red;"/>
</form>
"""%(data_dic["seed_guid"], data_dic["down_url"], data_dic["seed_name"], data_dic["max_no"], html_seed_type, html_status, added_form)
return html_form
def getTableAndPagingHtml(param_dic, dao_util):
if not dao_util:
return "", ""
if not param_dic["sc_type"]:
return "", ""
if param_dic["page"]:
page = int(param_dic["page"])
param_dic["page"] = page+1
next_page_get_params = html.getGetParamStr(param_dic)
param_dic["page"] = page-1
if (page-1) < 1:
param_dic["page"] = 1
prev_page_get_params = html.getGetParamStr(param_dic)
param_dic["page"] = page
else:
param_dic["page"] = 2
next_page_get_params = html.getGetParamStr(param_dic)
param_dic["page"] = 1
prev_page_get_params = html.getGetParamStr(param_dic)
sc_type = param_dic["sc_type"]
status_dic = getDataMappingDic(param_dic["_service_type"], param_dic["_data_type"], "status")
data_list = dao_util.getDataList(param_dic, "web")
col_name_list = ["도메인ID<br/>(클릭 시 수정)", "도메인", "사이트명<br/>(메모)", "상태", "서버", "PID",
"씨드URL수%s"%(html.getOrderingLinkHtml("/moon/crawl/domain_list.py", param_dic, "seed_url_count")),
"신규URL수%s<br/>(최근7일 현황)"%(html.getOrderingLinkHtml("/moon/crawl/domain_list.py", param_dic, "daily_count")),
"마지막 크롤일%s"%(html.getOrderingLinkHtml("/moon/crawl/domain_list.py", param_dic, "last_visit"))]
if param_dic["sc_type"] == "user": # TODO SC별 테이블 CUSTOMIZING
col_name_list.append("새컬럼명")
col_name_list.append("관리")
html_title_row = html.getTableTR(col_name_list, "title")
html_table = "<table style='table-layout:fixed; word-break:break-all;'>"+html_title_row
row_count = 0
# MAKE TABLE HTML
for row_data in data_list:
domain_id = row_data[0]
domain = row_data[1]
name = row_data[2]
memo = row_data[3]
if memo:
name = "%s<br/>(%s)"%(name, memo)
status = row_data[4]
if status:
status = status_dic[status]
server = row_data[5]
pid = row_data[6]
seed_url_count = row_data[7]
daily_count = row_data[8]
crawl_count_history = row_data[9]
last_visit = row_data[10]
if last_visit:
last_visit = getDateByStamp(last_visit)
else:
last_visit = ""
added_td_tags = ""
if sc_type == "user": # TODO 표시 데이터 SC별 CUSTOMIZING
added_td_tags += "<td>%s</td>"%row_data[11]
if (row_count % 2) == 0:
html_row = "<tr class='evenRow'>"
else:
html_row = "<tr>"
document_view_link = "http://topicplus.zuminternet.com/recent_documents.py?sc=%s&channelId=%s"%(sc_type, domain_id)
if sc_type == "bbs":
document_view_link = "http://topicplus.zuminternet.com/recent_documents.py?sc=%s&siteId=%s"%(sc_type, domain_id)
html_row += """
<td><a href="/moon/domanager/domain_add.py?domainIdx=%s" target="_blank">%s</a></td>
<td><a href="%s" target="_blank">%s</a><br/>
<a href="/moon/domanager/urlpt_list.py?domainLike=%s" target='_blank'>URL패턴</a> |
<a href="/moon/crawl/seedurl_list.py?sc_type=%s&domain_id=%s" target='_blank'>씨드URL</a><br/>
<a href="/moon/domanager/sitemapurl_list.py?domain_id=%s" target='_blank'>사이트맵URL</a>
</td>
"""%(domain_id, domain_id, "http://"+domain, domain, domain, sc_type, domain_id, domain_id)+"""
<td>%s</td>
<td>%s</td>
<td>%s</td>
<td>%s</td>
<td>%s</td>
<td>
<a href="%s" target="_blank" >%s</a>
<br/>%s
</td>
<td>%s</td>
"""%(name, status, server, pid, seed_url_count, document_view_link, daily_count, crawl_count_history, last_visit)+ added_td_tags +"""
<td>
<a href="/moon/crawl/data_add.py?sc_type=%s&domain_id=%s" target="_blank" >수정</a> |
<a href="#" onclick="confirmPopup('%s','/moon/action/sc_common.py?cmd=del&sc_type=%s&_data_type=domain&domain_id=%s')">삭제</a><br/>
<a href="/moon/crawl/data_add.py?sc_type=%s&domain_id=%s#seedAddForm" target="_blank" >씨드등록</a> |
<a href="#" onclick="confirmPopup2('%s','/moon/action/sc_common.py?cmd=del&sc_type=%s&_data_type=seed&domain_id=%s')">씨드삭제</a><br/>
<a href="/moon/action/sc_common.py?cmd=resetNoNewCount&sc_type=%s&domain_id=%s" target="_blank" >씨드전체 신규문서<br/>미발견 횟수 초기화</a>
</td>
</tr>
"""%(sc_type, domain_id, domain, sc_type, domain_id, sc_type, domain_id, domain, sc_type, domain_id, sc_type, domain_id)
html_table += html_row
row_count += 1
# MAKE PAGING HTML
total_count = dao_util.getDataList(param_dic, "count")
total_page = total_count / 15
if (total_count % 15) > 0:
total_page += 1
html_paging = """
<p>
<b>페이지 [%s/%s]</b> : <b style='color:red;'>%s</b> 개중 <b style='color:red;'>%s</b> 개 표시
<a href='/moon/crawl/domain_list.py?%s#tableDiv'><<이전</a> || <a href='/moon/crawl/domain_list.py?%s#tableDiv'>다음>></a>
</p>
"""%( param_dic["page"], total_page, total_count, len(data_list), prev_page_get_params, next_page_get_params)
return html_table+"</table>", html_paging
def getTableAndPagingHtml(param_dic, dao_util):
if not dao_util:
return "", ""
if not param_dic["sc_type"]:
return "", ""
if param_dic["page"]:
page = int(param_dic["page"])
param_dic["page"] = page + 1
next_page_get_params = html.getGetParamStr(param_dic)
param_dic["page"] = page - 1
if (page - 1) < 1:
param_dic["page"] = 1
prev_page_get_params = html.getGetParamStr(param_dic)
param_dic["page"] = page
else:
param_dic["page"] = 2
next_page_get_params = html.getGetParamStr(param_dic)
param_dic["page"] = 1
prev_page_get_params = html.getGetParamStr(param_dic)
data_list = dao_util.getDataList(param_dic, "web")
sc_type = param_dic["sc_type"]
col_name_list = [
"ID",
"씨드 URL 정보",
"씨드명<br/>(앵커값)",
"Generator",
"크롤상태<br/>(씨드타입)",
"view URL수<br/><b style='color:#6495ed;'>신규URL미발견 횟수</b><br/>INT_ID최대값",
"마지막 방문일",
]
if sc_type == "blog": # TODO : SC별 TITLE 행 표시 데이터 CUSTOMIZING
col_name_list[2] = "채널명<br/>(작성자)" # 씨드명 대신 표시
col_name_list[3] = "Generator<br/>(블로그ID)" # 앵커TEXT 대신 표시
# col_name_list.append("새컬럼명")
col_name_list.append("관리")
html_title_row = html.getTableTR(col_name_list, "title")
html_table = "<table style='table-layout:fixed; word-break:break-all;'>" + html_title_row
# MAKE TABLE HTML
type_dic = getDataMappingDic(param_dic["_service_type"], param_dic["_data_type"], "seed_type")
status_dic = getDataMappingDic(param_dic["_service_type"], param_dic["_data_type"], "crawl_status")
row_count = 0
for row_data in data_list:
seed_id = row_data[0]
domain_id = row_data[1]
seed_guid = row_data[2]
down_url = row_data[3]
seed_name = "<b>%s</b>" % row_data[4]
seed_type = row_data[5]
if seed_type:
seed_type = type_dic[seed_type]
anchor_name = row_data[6]
if anchor_name:
seed_name = "%s<br/>(%s)" % (seed_name, anchor_name)
crawl_status = row_data[7]
status = ""
if crawl_status:
status = "크롤:<b>%s</b>" % status_dic[crawl_status]
status += "<br/>타입:<b>%s</b>" % seed_type
view_url_count = "뷰수:<b>%s</b><br/>" % row_data[8]
no_new_count = "<b style='color:blue;'>노뉴:%s</b><br/>" % row_data[9]
max_no = "맥스:<b>%s</b>" % row_data[10]
numbers = view_url_count + no_new_count + max_no
last_visit = row_data[11]
generator = row_data[12]
if last_visit:
last_visit = getDateByStamp(last_visit)
else:
last_visit = ""
# TODO : SC별 데이터 행 표시 데이터 CUSTOMIZING
added_url_info = ""
added_td_tags = ""
if sc_type == "blog":
if row_data[13]: # 블로그 ID
generator += "<br/>(%s)" % row_data[13]
if row_data[14]: # 블로그 링크
added_url_info = "<a href='%s' target='_blank'>해당 블로그 바로가기</a><br/>" % row_data[14]
if row_data[15]: # author
seed_name += "<br/>작성자 : %s" % row_data[15]
# ROW 데이터 표시 시작
if (row_count % 2) == 0:
html_row = "<tr class='evenRow'>"
else:
html_row = "<tr>"
html_row += (
"""
<td>%s</td>
<td>
<b style="color:red;">%s</b><br/>
%s
도메인ID:<a href="/moon/domanager/domain_add.py?domainIdx=%s">%s</a>
<a href="%s" target="_blank" >다운로드URL</a>
</td>
"""
% (seed_id, seed_guid.replace("&", "&"), added_url_info, domain_id, domain_id, down_url)
+ """
<td>%s</td>
<td>%s</td>
<td>%s</td>
<td>%s</td>
<td>%s</td>
"""
% (seed_name, generator, status, numbers, last_visit)
+ added_td_tags
+ """
<td>
<a href="/moon/crawl/data_add.py?sc_type=%s&seed_id=%s" target="_blank" >수정</a> |
<a href="/moon/action/sc_common.py?sc_type=%s&cmd=blockSeed&seed_id=%s" target="_blank" >차단</a> |
<a href="#" onclick="confirmPopup('%s','/moon/action/sc_common.py?cmd=del&sc_type=%s&_data_type=seed&seed_id=%s')">삭제</a>
<br/><a href="/moon/action/sc_common.py?cmd=resetNoNewCount&sc_type=%s&seed_id=%s" target="_blank" >신규URL수 초기화</a>
<br/><a href="/moon/domanager/pattern_add.py?urlTypeTestUrl=%s" target="_blank">URL추출 TEST</a>
</td>
</tr>
"""
% (
sc_type,
seed_id,
sc_type,
seed_id,
seed_guid,
sc_type,
seed_id,
sc_type,
seed_id,
urllib2.quote(down_url),
)
)
html_table += html_row
row_count += 1
# MAKE PAGING HTML
total_count = dao_util.getDataList(param_dic, "count")
total_page = total_count / 15
if (total_count % 15) > 0:
total_page += 1
html_paging = """
<p>
<b>페이지 [%s/%s]</b> : <b style='color:red;'>%s</b> 개중 <b style='color:red;'>%s</b> 개 표시
<a href='/moon/crawl/seedurl_list.py?%s#tableDiv'><<이전</a> || <a href='/moon/crawl/seedurl_list.py?%s#tableDiv'>다음>></a>
</p>
""" % (
param_dic["page"],
total_page,
total_count,
len(data_list),
prev_page_get_params,
next_page_get_params,
)
return html_table + "</table>", html_paging