def save_to_file(self, data, file_path):
create_or_get_directory(os.path.dirname(file_path))
if os.path.exists(file_path):
with open(file_path, 'a') as f:
dict_writer = csv.DictWriter(f, data.keys())
dict_writer.writerow(data)
logger.info(f'保存数据至:{file_path}')
else:
with open(file_path, 'w') as f:
dict_writer = csv.DictWriter(f, data.keys())
dict_writer.writeheader()
dict_writer.writerow(data)
logger.info(f'保存数据至:{file_path}')
def requirement_word_cloud(self):
"""招聘要求词云"""
jieba.analyse.set_stop_words('configures/stop_words.txt')
data, word = [], ''
for content in self.frame['requirement']:
try:
word = "".join(content)
except Exception as error:
print(f"获取职位诱惑发生错误:{error}")
data.append(word)
tags = jieba.analyse.extract_tags("".join(data).lower(),
topK=100,
withWeight=True)
advantage_jieba = []
for wd, weight in tags:
advantage_jieba.append([wd, weight])
html_path = os.path.join(
create_or_get_directory(os.path.join('report', 'single')),
'requirement_word_cloud.html')
(WordCloud().add(series_name="招聘要求",
data_pair=advantage_jieba,
word_size_range=[16, 166]).set_global_opts(
tooltip_opts=opts.TooltipOpts(
is_show=True), ).render(html_path))
script_string = self.get_javascript_string(html_path)
self.set_report_javascript(script_string, 'requirement_word_cloud')
self.middle_charts.append({
'title': '招聘要求关键字',
'name': 'requirement_word_cloud'
})
def city_company_scale_chart(self):
"""前10城市的企业阶段"""
city_company_scale = pd.crosstab(self.frame.city,
self.frame.company_scale,
margins=True).sort_values(
by='All', ascending=False)[:10]
city_company_scale = city_company_scale.drop('All',
axis=0).drop('All',
axis=1)
html_path = os.path.join(
create_or_get_directory(os.path.join('report', 'single')),
'city_company_scale_chart.html')
bar = Bar(init_opts=opts.InitOpts(width="60%"))
bar.add_xaxis(list(city_company_scale.index))
for i in range(len(list(city_company_scale.T.index))):
bar.add_yaxis(city_company_scale.T.index[i],
[str(v) for v in city_company_scale.T.values[i]])
bar.set_global_opts(
legend_opts=opts.LegendOpts(is_show=False),
xaxis_opts=opts.AxisOpts(
axislabel_opts=opts.LabelOpts(rotate=-20, color='red')),
yaxis_opts=opts.AxisOpts(axislabel_opts=opts.LabelOpts(
color='red')),
)
bar.render(html_path)
script_string = self.get_javascript_string(html_path)
self.set_report_javascript(script_string, 'city_company_scale_chart')
self.right_charts.append({
'title': '前10城市的企业阶段',
'name': 'city_company_scale_chart'
})
def company_field_chart(self):
"""企业领域统计"""
kd = self.frame.company_field.value_counts()
field_count = {}
for field_key, field_value in dict(kd).items():
field_keys = re.split(r',|，|、|\.|\|\t| |丨', field_key)
for key in field_keys:
if field_count.get(key, None):
field_count[key] = field_count[key] + field_value
else:
field_count[key] = field_value
html_path = os.path.join(
create_or_get_directory(os.path.join('report', 'single')),
'company_field_chart.html')
c = (Pie().add("企业领域", [
[k, str(v)] for k, v in field_count.items()
]).set_global_opts(legend_opts=opts.LegendOpts(
is_show=False), ).set_series_opts(tooltip_opts=opts.TooltipOpts(
trigger="item", formatter="{a} <br/>{b}: {c} ({d}%)")).render(
html_path))
script_string = self.get_javascript_string(html_path)
self.set_report_javascript(script_string, 'company_field_chart')
self.right_charts.append({
'title': '企业领域统计',
'name': 'company_field_chart'
})
def company_size_bar_chart(self):
"""企业规模统计"""
kd = self.frame.company_size.value_counts()
results = [[k, str(v)] for k, v in dict(kd).items()]
html_path = os.path.join(
create_or_get_directory(os.path.join('report', 'single')),
'company_size_bar_chart.html')
c = (Pie(init_opts=opts.InitOpts()).add(
series_name="工作经验",
data_pair=results,
rosetype="radius",
radius="55%",
center=["50%", "50%"],
label_opts=opts.LabelOpts(is_show=False, position="center"),
).set_global_opts(
legend_opts=opts.LegendOpts(is_show=False), ).set_series_opts(
tooltip_opts=opts.TooltipOpts(
trigger="item", formatter="{a} <br/>{b}: {c} ({d}%)"),
label_opts=opts.LabelOpts(color="rgba(255, 255, 255, 0.3)"),
).render(html_path))
script_string = self.get_javascript_string(html_path)
self.set_report_javascript(script_string, 'company_size_bar_chart')
self.right_charts.append({
'title': '企业规模统计',
'name': 'company_size_bar_chart'
})
def address_pie_chart(self):
"""招聘数前10城市"""
kd = self.frame.city.value_counts()[:10]
results = [[k, str(v)] for k, v in dict(kd).items()]
html_path = os.path.join(
create_or_get_directory(os.path.join('report', 'single')),
'address_pie_chart.html')
c = (Pie().add("招聘数前10城市", results).set_global_opts(
legend_opts=opts.LegendOpts(is_show=False), ).set_series_opts(
tooltip_opts=opts.TooltipOpts(
trigger="item",
formatter="{a} <br/>{b}: {c} ({d}%)")).render(html_path))
script_string = self.get_javascript_string(html_path)
self.set_report_javascript(script_string, 'address_pie_chart')
self.left_charts.append({
'title': '招聘数前10城市',
'name': 'address_pie_chart'
})
def company_scale_bar_chart(self):
"""企业发展阶段统计"""
kd = self.frame.company_scale.value_counts()
results = [[k, str(v)] for k, v in dict(kd).items()]
html_path = os.path.join(
create_or_get_directory(os.path.join('report', 'single')),
'company_scale_bar_chart.html')
c = (Pie().add("企业发展阶段", results).set_colors([
"green", "yellow", "red", "pink", "orange", "purple"
]).set_global_opts(legend_opts=opts.LegendOpts(
is_show=False), ).set_series_opts(tooltip_opts=opts.TooltipOpts(
trigger="item", formatter="{a} <br/>{b}: {c} ({d}%)")).render(
html_path))
script_string = self.get_javascript_string(html_path)
self.set_report_javascript(script_string, 'company_scale_bar_chart')
self.right_charts.append({
'title': '企业发展阶段统计',
'name': 'company_scale_bar_chart'
})
def salary_pie_chart(self):
"""薪酬分布"""
salary_xaxis = [
'0', '5000', '10000', '15000', '20000', '30000', '40000', '50000',
'60000', '70000', '80000', '90000', '100000'
]
salary_yaxis = []
kd = self.frame.salary.value_counts()
salary_dict = {
0: 0,
5000: 0,
10000: 0,
15000: 0,
20000: 0,
30000: 0,
40000: 0,
50000: 0,
60000: 0,
70000: 0,
80000: 0,
90000: 0,
100000: 0
}
for k, v in dict(kd).items():
min_salary = float(
k.split('-')[0].split('·')[0].replace(
'以上', '').rstrip('k').rstrip('K')) * 1000
try:
max_salary = float(
k.split('-')[1].split('·')[0].replace(
'以上', '').rstrip('k').rstrip('K')) * 1000
except:
max_salary = min_salary
salary = 0.5 * (min_salary + max_salary)
if salary in range(0, 5000): salary = 0
if salary in range(5000, 10000): salary = 5000
if salary in range(10000, 15000): salary = 10000
if salary in range(15000, 20000): salary = 15000
if salary in range(20000, 30000): salary = 20000
if salary in range(30000, 40000): salary = 30000
if salary in range(40000, 50000): salary = 40000
if salary in range(50000, 60000): salary = 50000
if salary in range(60000, 70000): salary = 60000
if salary in range(70000, 80000): salary = 70000
if salary in range(80000, 90000): salary = 80000
if salary in range(90000, 100000): salary = 90000
if salary in range(100000, 10000000000): salary = 100000
if salary_dict.get(salary, ''):
salary_dict[salary] = salary_dict[salary] + v
else:
salary_dict[salary] = v
salary_dict = dict(collections.OrderedDict(sorted(
salary_dict.items())))
for salary_key, salary_value in salary_dict.items():
salary_yaxis.append(
opts.BarItem(
name=str(salary_key),
value=int(salary_value),
itemstyle_opts=opts.ItemStyleOpts(color="#d48265"),
))
html_path = os.path.join(
create_or_get_directory(os.path.join('report', 'single')),
'salary_pie_chart.html')
c = (Bar(init_opts=opts.InitOpts(
width="60%")).add_xaxis(salary_xaxis).add_yaxis(
"", salary_yaxis, category_gap=-1).set_series_opts(
label_opts=opts.LabelOpts(is_show=False)).set_global_opts(
yaxis_opts=opts.AxisOpts(
name="岗位数量",
axislabel_opts=opts.LabelOpts(color='red'),
name_textstyle_opts=opts.TextStyleOpts(
color='#d48265')),
xaxis_opts=opts.AxisOpts(
name="薪酬",
axislabel_opts=opts.LabelOpts(color='red'),
name_textstyle_opts=opts.TextStyleOpts(
color='#d48265')),
tooltip_opts=opts.TooltipOpts(
trigger="item",
formatter=JsCode("""function (params) {
var name='';
if (params.name < 5000){
name='小于5K'
}
else if (params.name < 10000) {
name='5K-10K'
}
else if (params.name < 15000) {
name='10K-15K'
}
else if (params.name < 20000) {
name='15K-20K'
}
else if (params.name < 30000) {
name='20K-30K'
}
else if (params.name < 40000) {
name='30K-40K'
}
else if (params.name < 50000) {
name='40K-50K'
}
else if (params.name < 60000) {
name='50K-60K'
}
else if (params.name < 70000) {
name='60K-70K'
}
else if (params.name < 80000) {
name='70K-80K'
}
else if (params.name < 90000) {
name='80K-90K'
}
else if (params.name < 100000) {
name='90K-100K'
}
else {
name='大于100K'
}
return name+':'+params.value;
}
""")),
).render(html_path))
script_string = self.get_javascript_string(html_path)
self.set_report_javascript(script_string, 'salary_pie_chart')
self.middle_charts.append({
'title': '薪资分布',
'name': 'salary_pie_chart'
})
