Esempio n. 1
0
def index():
    data = 0
    sites = Site(current_app)
    site = sites.totalcount()
    seeds = Seed(current_app)
    seed = seeds.totalcount()
    users = User(current_app)
    uid = session['uid']
    user = users.view(uid)
    lastlogin = user[0]["lastlogintime"]
    return render_template("index.html", site=site, seed=seed, data=data, lastlogin=lastlogin)
Esempio n. 2
0
def copynew(seed_id):
    if seed_id:
        seed = Seed(current_app)
        sid = seed.copynew(seed_id)
        if sid:
            seed_fields = Seed_fields(current_app)
            seed_fields.copynew(seed_id, sid)
            seed_filter = Seed_filter(current_app)
            seed_filter.copynew(seed_id, sid)
    url = request.referrer and request.referrer or url_for("seeds.index")
    return redirect(url)
Esempio n. 3
0
def view(seed_id,guid):
    items = None
    fields = {}
    if guid and seed_id:
        db = Seed_Model();
        r = db.view(seed_id);
        seed = Seed(r.list()[0])
        t = Grab(seed)
        if len(t):
            db_field = Field_Model(current_app)
            items = t[guid]
            field_datas = db_field.list(items["type"])
            for field_data in field_datas:
                fields[field_data["name"]] = field_data["title"]
    return render_template("test_seed/view.html", items=items, fields=fields)
Esempio n. 4
0
def seed_logs(page=1):
    seed = Seed(current_app)
    seed_logs = Seed_log(current_app)
    field = Field(current_app)
    filte = []
    status = None
    start_time = ""
    end_time = ""
    if request.args.get("start_time"):
        start_time = request.args.get("start_time")
        tmp = time.mktime(datetime.datetime.strptime(start_time, "%Y-%m-%d %H:%M").timetuple())
        filte.append("start_time >= "+str(tmp))
    if request.args.get("end_time"):
        end_time = request.args.get("end_time")
        tmp = time.mktime(datetime.datetime.strptime(end_time, "%Y-%m-%d %H:%M").timetuple())
        filte.append("start_time <= "+str(tmp))
    if request.args.get("status"):
        status = request.args.get("status")
        filte.append("status = "+str(status))
    if request.args.get("seed_id"):
        seed_id = request.args.get("seed_id")
        filte.append("sid = "+str(seed_id))
    filte = " and ".join(filte)
    seeds1 = seed_logs.list(page, PER_PAGE, filte, "start_time DESC")
    seeds = []
    if not seeds1 and page != 1:
        abort(404)
    for seed_item in seeds1:
        tmp = seed.view(seed_item["sid"]).list()
        if len(tmp) > 0:
            seed_item["seed_name"] = tmp[0]["seed_name"]
        else:
            seed_item["seed_name"] = u"未知"
        seeds.append(seed_item)
    count = seed_logs.totalcount(filte)
    pagination = Pagination(page, PER_PAGE, count)
    fields = field.getSeedType()
    return render_template("logs/seed.html", pagination=pagination, seeds=seeds, fields=fields, status=status, start_time=start_time, end_time=end_time)
Esempio n. 5
0
def list(seed_id):
    items = {}
    num = 0
    seed_type = None
    if seed_id:
        db = Seed_Model();
        r = db.view(seed_id);
        seed = Seed(r.list()[0])
        t = Grab(seed)
        if len(t):
            num = len(t.keys())
            seed_type = t.seed_type
            for guid in t.keys():
                tmp = {}
                dont_craw_content = [
                'kaifu', 'kaice', "gift"
                ]
                if seed_type in dont_craw_content:
                    for f in t.items[guid].fields:
                        tmp[f] = t.items[guid][f].value
                else:
                    tmp["url"] = t.items[guid]["url"]
                items[guid] = tmp
    return render_template("test_seed/list.html", items=items, num=num, seed_id=seed_id, seed_type=seed_type, dont_craw_content=dont_craw_content)
Esempio n. 6
0
def index(page=1):
    seed = Seed(current_app)
    field = Field(current_app)
    seed_fields = Seed_fields(current_app)
    filte = {}
    seed_type = None
    if request.method == "POST":
        error = None
        action = request.form.get("do")
        if action == "delete":
            seed_filter_model = Seed_filter(current_app)
            sids = request.form.getlist("sid[]")
            if len(sids) > 0:
                for sid in sids:
                    if sid:
                        seed.remove(sid)
                        seed_fields.remove(sid)
                        seed_filter_model.remove(sid)
                url = request.referrer and request.referrer or url_for('seeds.index')
                return redirect(url)
            else:
                error = "请选择要删除的数据"
        if action == "enable" or action == "disable":
            edid = (action == "enable") and 1 or 0
            sids = request.form.getlist("sid[]")
            if len(sids) > 0:
                for sid in sids:
                    if sid:
                        save = {"enabled":edid}
                        seed.edit(sid,**save)
                url = request.referrer and request.referrer or url_for('seeds.index')
                return redirect(url)
            else:
                error = "请选择要启用的数据"
        return error
    if request.args.get("keywords"):
        filte["seed_name"] = request.args.get("keywords")
    if request.args.get("type"):
        seed_type = request.args.get("type")
        filte["type"] = seed_type
    seeds1 = seed.list(page, PER_PAGE, filte)
    seeds = []
    if not seeds1 and page != 1:
        abort(404)
    for seed_item in seeds1:
        seed_item["tags"] = getTagsBySeedId(seed_item["sid"])
        seeds.append(seed_item)
    count = seed.totalcount(filte)
    pagination = Pagination(page, PER_PAGE, count)
    fields = field.getSeedType()
    return render_template("seed/list.html", pagination=pagination, seeds=seeds, fields=fields, seed_type=seed_type)
Esempio n. 7
0
def edit(seed_id):
    tags = []
    alltags = []
    if request.method == "POST" and request.form.get("sid"):
        sid = request.form.get("sid")
        enabled = checkboxVal(request.form.get("enabled"))
        list1 = {
            "urlformat": request.form.get("urlformat"),
            "urltype": request.form.get("urltype"),
            "startpage": request.form.get("startpage"),
            "maxpage": request.form.get("maxpage"),
            "step": request.form.get("step"),
            "zero": request.form.get("zero"),
            "listparent": request.form.get("listparent"),
            "entryparent": request.form.get("entryparent"),
            "contenturl": request.form.get("contenturl"),
            "contentparent": request.form.get("contentparent"),
            "pageparent": request.form.get("pageparent"),
            "filters": request.form.get("filters"),
        }
        time1 = int(time.time())
        save = {
            "type": request.form.get("type"),
            "seed_name": request.form.get("seed_name"),
            "charset": request.form.get("charset"),
            "frequency": float(request.form.get("frequency")) * 3600,
            "timeout": request.form.get("timeout"),
            "tries": request.form.get("tries"),
            "enabled": enabled,
            "listtype": request.form.get("listtype"),
            "lang": request.form.get("lang"),
            "update_time": time1,
            "rule": phpserialize.dumps(list1),
        }
        field = Field(current_app)
        fields = field.list(save["type"]).list()
        guid_rule_datas = request.form.get("guid_rule")
        save["guid_rule"] = getFeildIdByTitle(guid_rule_datas, save["type"])
        seed = Seed(current_app)
        msg = seed.edit(sid, **save)
        seed_field = Seed_fields(current_app)
        # filter_model = Filter(current_app)
        # filters_org_data = filter_model.list()
        seed_filter_model = Seed_filter(current_app)
        # filters_data = {}
        # for filter_item in filters_org_data:
        #     filters_data[str(filter_item["id"])] = filter_item
        for field in fields:
            field_data = seed_field.view(sid, field.id).list()
            if len(field_data) > 0:
                seed_field.edit(
                    sid,
                    field.id,
                    request.form.get(field.name),
                    request.form.get("page_type_" + field.name),
                    request.form.get("fetch_all_" + field.name),
                )
            else:
                seed_value = {}
                seed_value["seed_id"] = sid
                seed_value["field_id"] = field.id
                seed_value["value"] = request.form.get(field.name)
                seed_value["page_type"] = request.form.get("page_type_" + field.name)
                seed_value["fetch_all"] = (
                    request.form.get("fetch_all_" + field.name) and request.form.get("fetch_all_" + field.name) or 0
                )
                seed_field.add(**seed_value)
            # 更改过滤规则
            filter_ids = request.form.getlist("filter" + str(field.id) + "[]")
            order_ct = 0
            if len(filter_ids) > 0 and filter_ids[0] != "none":
                seed_filter_model.remove(sid, field.id)
                for filter_id in filter_ids:
                    filter_item = request.form.getlist("profile_" + str(field.id) + "_" + str(filter_id) + "[]")
                    tmp = {}
                    ct = 0
                    for config in filter_item:
                        v = request.form.get(config + "_" + str(field.id) + "_" + filter_id)
                        if not v:
                            ct = ct + 1
                        tmp[config] = v
                    if ct != len(filter_item):
                        order_ct = order_ct + 1
                        filter_save = {
                            "sid": int(sid),
                            "field_id": int(field.id),
                            "filter_id": int(filter_id),
                            "profile": phpserialize.dumps(tmp),
                            "list_order": int(order_ct),
                        }
                        seed_filter_model.add(**filter_save)
        # 更改标签
        tags_data = request.form.get("tags")
        tags_data = tags_data.split(",")
        tags_model = Tags(current_app)
        seed_tag = Seed_tag(current_app)
        current_tags_data = seed_tag.list(sid)
        del_tags_data = {}
        for current_tag in current_tags_data:
            del_tags_data[current_tag["tid"]] = current_tag
        for tag in tags_data:
            tag = tag.strip()
            if tag:
                ishere = tags_model.findByName(tag).list()
                if len(ishere) > 0:
                    tid = ishere[0]["id"]
                else:
                    tid = tags_model.add(name=tag)
                if tid in del_tags_data:
                    del del_tags_data[tid]
                ishere = seed_tag.view(sid, tid).list()
                if len(ishere) == 0:
                    seed_tag.add(sid=sid, tid=tid)
        for k in del_tags_data:
            seed_tag.remove(del_tags_data[k]["sid"], del_tags_data[k]["tid"])
        url = getReferer()
        url = url and url or url_for("seeds.index")
        return redirect(url)
    if request.method == "GET" and seed_id > 0:
        setReferer()
        size = request.cookies.get("editorSize")
        if not size:
            size = 50
        seed = Seed(current_app)
        seed_data = seed.view(seed_id)[0]
        seed_type = seed_data["type"]
        seed_field = getSeedFieldsBySid(seed_id, seed_type)
        seed_fields = Seed_fields(current_app)
        page_types = seed_fields.getpageType()
        seed_data["frequency"] = float(seed_data["frequency"]) / float(3600)
        if seed_data["rule"]:
            seed_data["rule"] = seed_data["rule"].encode("utf-8")
            seed_data["rule"] = phpserialize.loads(seed_data["rule"], decode_strings=True)
        # 获取GUID规则
        if seed_data["guid_rule"]:
            seed_data["guid_rule"] = getFeildTitleById(seed_data["guid_rule"], seed_type)
        # 取出标签
        seed_tag = Seed_tag(current_app)
        tags_model = Tags(current_app)
        tags_data = seed_tag.list(seed_id)
        for tag in tags_data:
            tmp = tags_model.view(tag["tid"])[0]
            tags.append(tmp["name"])
        alltags_data = tags_model.list()
        for tag in alltags_data:
            alltags.append(tag["name"])
    return render_template(
        "seed/add.html",
        seed_type=seed_type,
        fields=seed_field,
        seed_data=seed_data,
        sid=seed_id,
        page_types=page_types,
        tags=tags,
        alltags=alltags,
        size=size,
    )
Esempio n. 8
0
def add():
    field = Field(current_app)
    if request.method == "POST":
        list1 = {
            "urlformat": request.form.get("urlformat"),
            "urltype": request.form.get("urltype"),
            "startpage": request.form.get("startpage"),
            "maxpage": request.form.get("maxpage"),
            "step": request.form.get("step"),
            "zero": request.form.get("zero"),
            "listparent": request.form.get("listparent"),
            "entryparent": request.form.get("entryparent"),
            "contenturl": request.form.get("contenturl"),
            "contentparent": request.form.get("contentparent"),
            "pageparent": request.form.get("pageparent"),
            "filters": request.form.get("filters"),
        }
        time1 = int(time.time())
        save = {
            "type": request.form.get("type"),
            "seed_name": request.form.get("seed_name"),
            "charset": request.form.get("charset"),
            "frequency": float(request.form.get("frequency")) * 3600,
            "timeout": request.form.get("timeout"),
            "tries": request.form.get("tries"),
            "enabled": request.form.get("enabled") is 1 and 1 or 0,
            "listtype": request.form.get("listtype"),
            "lang": request.form.get("lang"),
            "created_time": time1,
            "update_time": time1,
            "rule": phpserialize.dumps(list1),
        }
        fields = field.list(save["type"]).list()
        guid_rule_datas = request.form.get("guid_rule")
        save["guid_rule"] = getFeildIdByTitle(guid_rule_datas, save["type"])
        seed = Seed(current_app)
        sid = seed.add(**save)
        if sid:
            seed_field = Seed_fields(current_app)
            # page_types = request.form.getlist("page_type[]")
            for field in fields:
                seed_value = {}
                seed_value["seed_id"] = sid
                seed_value["field_id"] = field.id
                seed_value["value"] = request.form.get(field.name)
                seed_value["page_type"] = request.form.get("page_type_" + field.name)
                seed_value["fetch_all"] = (
                    request.form.get("fetch_all_" + field.name) and request.form.get("fetch_all_" + field.name) or 0
                )
                seed_field.add(**seed_value)
            # 插入标签
            tags_data = request.form.get("tags")
            tags_data = tags_data.split(",")
            tags_model = Tags(current_app)
            seed_tag = Seed_tag(current_app)
            for tag in tags_data:
                tag = tag.strip()
                if tag:
                    ishere = tags_model.findByName(tag).list()
                    if len(ishere) > 0:
                        tid = ishere[0]["id"]
                    else:
                        tid = tags_model.add(name=tag)
                    ishere = seed_tag.view(sid, tid).list()
                    if len(ishere) == 0:
                        seed_tag.add(sid=sid, tid=tid)
        url = getReferer()
        url = url and url or url_for("seeds.index")
        return redirect(url)
    fields = field.getSeedType()
    if request.method == "GET" and request.args.get("type"):
        setReferer()
        size = request.cookies.get("editorSize")
        if not size:
            size = 50
        seed_data = {}
        seed_data["rule"] = {}
        seed_type = request.args.get("type")
        fields = field.list(seed_type).list()
        seed_field = Seed_fields(current_app)
        page_types = seed_field.getpageType()
        alltags = []
        tags_model = Tags(current_app)
        alltags_data = tags_model.list()
        for tag in alltags_data:
            alltags.append(tag["name"])
        return render_template(
            "seed/add.html",
            seed_type=seed_type,
            fields=fields,
            seed_data=seed_data,
            page_types=page_types,
            alltags=alltags,
            size=size,
        )
    return render_template("seed/select_type.html", fields=fields)
Esempio n. 9
0
	extrarules = self.parent.extrarules
	for i, rule in enumerate(extrarules):
	    if rule and rule["page_type"] == "content":
		field_id = rule["field_id"]
		value = rule["value"]
		fetch_all = rule["fetch_all"]
		self.extrarules.append((field_id, value, fetch_all))

        if "filters" in parent.rule:
	    self.filters = []
            if parent.rule["filters"]:
                parent.rule["filters"] = parent.rule["filters"].split("|");
	        for f in parent.rule["filters"]:
		      self.filters.append(f)
        else:
            self.filters = []


if __name__ == "__main__":
    from web.models import Seed as Seed_Model
    db = Seed_Model();
    r = db.view(8);
    t = Seed(r.list()[0])
    
    # test Seed info
    rule = t.getRule();
    #test rule
    rule.getListRule();
    #test article
    rule.getArticleRule();