data[, by = .(year, label), .(freq = .N)]|>
  group_by(label) |>
  e_chart(year) |>
  e_bar(freq) |>
  e_tooltip(trigger = 'axis') |>
  e_x_axis(type = 'category') |>
  e_y_axis(show = FALSE) |>
  e_title(text = '政府公布政策数量',
          subtext = '数据来源：中国政府网',
          left = 'left') |>
  e_legend(top = 20) |>
  e_labels(position = 'top') |>
  e_facet(
    rows = 3,
    # 分面的行数
    cols = 1,
    # 分面的列数
    legend_pos = "top",
    # 图例的位置
    legend_space = 12 # 图例与图形区域之间的距离
  )

data$content_contain <- ifelse(grepl("党", data$content) , "Y", "N")
data$content_contain1 <- ifelse(grepl("毛泽东|邓小平|江泽民|胡锦涛|习近平", data$content) , "Y", "N")

e1 <- data[, by = .(year, label), .(prob1 = round(uniqueN(id[content_contain == 'Y']) /
                                                    (.N), 4))] |>
  group_by(label) |>
  e_charts(year) |>
  e_area(prob1) |>
  e_tooltip(trigger = 'axis') |>
  e_x_axis(type = 'category') |>
  e_y_axis(min = 0, max = 1) |>
  e_title(text = '政府公布政策文件内容中含“党”字的比例',
          subtext = '数据来源：中国政府网',
          left = 'left') |>
  e_legend(top = 20) |>
  e_labels(show = FALSE) |>
  e_facet(
    rows = 1,
    # 分面的行数
    cols = 3,
    # 分面的列数
    legend_pos = "top",
    legend_space = 12
  )

e2 <- data[, by = .(year, label), .(prob2 = round(uniqueN(id[content_contain1 == 'Y']) /
                                                    uniqueN(id[content_contain == 'Y']) , 4))] |>
  group_by(label) |>
  e_charts(year) |>
  e_area(prob2) |>
  e_tooltip(trigger = 'axis') |>
  e_x_axis(type = 'category') |>
  e_y_axis(min = 0, max = 1) |>
  e_title(text = '政府公布政策文件内容中，含历届最高领导人姓名除以含“党”字的比例',
          subtext = '数据来源：中国政府网',
          left = 'left') |>
  e_legend(top = 20) |>
  e_labels(show = FALSE) |>
  e_facet(
    rows = 1,
    # 分面的行数
    cols = 3,
    # 分面的列数
    legend_pos = "top",
    legend_space = 12
  )

e_arrange(e1,e2,cols = 1,rows = 2)

# 区分来源
data$source <-
  ifelse(
    grepl("中共中央", data$title) | grepl("中共中央", data$puborg),
    "中共中央",
    ifelse(
      grepl("中央军委", data$title) | grepl("中央军委", data$puborg),
      "中央军委",
      ifelse(
        grepl("国务院", data$title) | grepl("国务院", data$puborg),
        "国务院",
        "其它"
      )
    )
  )
data$source <- factor(data$source, levels = c('中共中央', '中央军委', '国务院', '其它'))

#table(data$label, data$source)

dcast(data[, by = .(source, label), .(freq = .N)], label ~ source, value.var = 'freq') |>
  e_charts(label) |>
  e_pie(`中共中央`, radius = "30%", center = c("30%", "30%")) |>
  e_pie(`中央军委`, radius = "30%", center = c("70%", "30%")) |>
  e_pie(`国务院`, radius = "30%", center = c("30%", "70%")) |>
  e_pie(`其它`, radius = "30%", center = c("75%", "65%")) |>
  e_labels(position = 'outside', formatter = '{b}\n{c},  {d}%') |>
  e_legend(right='right') |>
  e_title(text = '不同来源下各文件类别的比例', subtext = '左上-中共中央；右上：中央军委；\n\n左下：国务院；右下：其它')

data[, puborg := ifelse(
  puborg %in% c("国务院", "国务院办公厅", "国务院办公厅秘书局"),
  "国务院",
  ifelse(
    puborg %in% c("国务院、中央军委", "国务院办公厅、中央军委办公厅"),
    "国务院、中央军委",
    ifelse(
      puborg %in% c(
        "国务院应对新型冠状病毒感染肺炎疫情联防联控机制",
        "国务院应对新型冠状病毒感染疫情联防联控机制、中央农村工作领导小组",
        "中央应对新型冠状病毒感染肺炎疫情工作领导小组"
      ),
      "应对新冠疫情",
      ifelse(puborg %in% c("国务院抗震救灾总指挥部", "国务院应急管理办公室"), "应急管理", puborg)
    )
  )
)][label == '国务院文件', by = .(year, puborg), .(freq = .N)] |>
  group_by(puborg) |>
  e_charts(year) |>
  e_line(freq) |>
  e_x_axis(
    type = 'value',
    min = 1969,
    max = 2024,
    axisLabel = list(
      formatter = htmlwidgets::JS("function(value) { return value.toString().replace(',', ''); }")
    )
  ) |>
  e_tooltip(trigger = 'axis') |>
  e_y_axis(
    show = TRUE,
    splitNumber = 3,
    axisLabel = list(inside = TRUE, rotate = -90)
  ) |>
  e_title(text = '各发文机关公布政策文件数量',
          subtext = '数据来源：中国政府网',
          left = 'left') |>
  e_legend(top = 20) |>
  e_facet(
    rows = 4,
    # 分面的行数
    cols = 1,
    # 分面的列数
    legend_pos = "top",
    # 图例的位置
    legend_space = 12 # 图例与图形区域之间的距离
  )

# 替换不规则的分隔符为普通空格
data[, puborg := gsub("\u2002|、|　|，", " ", puborg)]

# 去掉一些字词，便于后面合并
chars_to_remove <- c("办公厅", "等", "办公室", "和", "综合司", "员会", "国家")
pattern <- paste(chars_to_remove, collapse = "|")
data[, puborg := gsub(pattern, "", puborg)]

# 定义 col_split_count 函数
col_split_count <- function(dt, year) {
  dt[, col_split := strsplit(puborg, " ")]
  dt_long <- dt[, .(col_split = unlist(col_split)), by = .(id)]
  dt_long[, .(count = .N), by = .(col_split)]
}

# 应用 col_split_count 函数并合并结果
years <- 2018:2024
puborg_data_list <- lapply(years, function(y) {
  dt <- col_split_count(data[label == '国务院部门文件' & year == y], y)
  dt[, year := y]
  dt
})

puborg_data <- rbindlist(puborg_data_list)
rm(puborg_data_list)

quantile(puborg_data[, by = .(col_split), .(freq = sum(count))]$freq, probs = seq(0, 1, 0.1))

puborg_data[, rank := rank(-count, ties.method = "first"), by = year][rank <= 5] |>
  e_charts(year) |>
  e_scatter(
    serie = count,
    size =  count,
    bind = col_split,
    itemStyle = list(opacity = 0.8),
    scale_js = "function(data){ return 12*(Math.sqrt(data[2]) + 0.1);}"
  ) |>
  e_x_axis(
    type = 'category',
    name = '发文年份',
    nameLocation = 'center',
    nameGap = 20
  ) |>
  e_y_axis(name = '发文数量',
           nameLocation = 'center',
           nameGap = 30) |>
  e_labels(
    position = 'inside',
    formatter = '{b}',
    labelLayout = list(moveOverlap = 'shiftY')
  ) |>
  e_title(text = '各发文机关每年公布政策文件数量-排名前五',
          subtext = '数据来源：中国政府网',
          left = 'left') |>
  e_legend(show = FALSE)

# 每年发文数量最多的机关
result <- puborg_data[, .SD[which.max(count)], by = year]

# 计算每个 str 数量最大的年份及其对应的数量
max_year_count <- puborg_data[, .(year = year[which.max(count)], max_count = max(count)), by = col_split]

# 计算总数
sum_all <- puborg_data[, by = .(col_split), .(freq = sum(count))]

sum_all[max_year_count, on = .(col_split)][freq >= 60][order(year)] |>
  e_charts(max_count) |>
  e_scatter(
    serie = year,
    size =  freq,
    bind = col_split,
    itemStyle = list(opacity = 0.8),
    scale_js = "function(data){ return 5*(Math.sqrt(data[2]/10) + 0.1);}",
    label = list(show = TRUE, formatter = '{b}'),
    labelLayout = list(x = "85%", moveOverlap = "shiftY"),
    labelLine = list(show = TRUE, lineStyle = list(color = "gray")),
    emphasis = list(focus = "self")
  ) |>
  e_y_axis(
    type = 'value',
    min = 2018,
    max = 2024,
    name = '各机关发文数量最多的年份',
    axisLabel = list(
      formatter = htmlwidgets::JS("function(value) { return value.toString().replace(',', ''); }")
    )
  ) |>
  e_x_axis(name = '按年汇总的最多发文数量',
           nameLocation = 'center',
           nameGap = 35) |>
  e_tooltip(
    padding = 5,
    borderWidth = 1,
    trigger = "item",
    formatter = htmlwidgets::JS(
      "function(params){
        return('<strong>发文机关:' +  params.name +
               '</strong><br />按年汇总的最多发文数量: ' + params.value[0] +'次'+
               '<br />发文数量最多年份: ' + params.value[1] +'年'+
      '<br />总发文数量: ' + params.value[2] +'次')}"
    )
  ) |>
  e_title(text = '发文数量最多的年份',
          subtext = '气泡大小：总发文数量',
          left = 'center') |>
  e_legend(show = FALSE) |>
  e_grid(left = '10%', right = '20%')

# 需要计算总发文数量排名前90%的发文机关
departments <- sum_all[freq >= 60, col_split]

# 对每个部门进行计算，但不直接修改 puborg 列，最后合并结果时通过 dept 区分
result_list <- lapply(departments, function(dept) {
  data[label == '国务院部门文件' &
         puborg %like% dept, by = year, # 只按年份分组
       .(
         freq1 = .N, # 各部门发文数量
         freq2 = uniqueN(id[puborg %like% '财政']), # 联合财政的发文数量
         prob = round(uniqueN(id[puborg %like% '财政']) / (.N), 2)
       )][, department := dept]  # 计算后添加部门列
})

# 将所有部门的结果合并
combined_data <- rbindlist(result_list)

quantile(combined_data$prob, probs = seq(0, 1, 0.1))


datatable(
  dcast(combined_data[year >= 2018 &
                        !(department == '财政部')], department ~ year, value.var = 'prob'),
  rownames = FALSE,
  colnames = c('发文机关', 2018:2024),
  caption = '各机关发文数量中“联合财政部发文”的比例',
  options = list(
    dom = 'tp',
    order = list(7, 'desc')
  )
) |> 
  formatStyle('2018', backgroundColor = styleInterval(
    c(0.1, 0.2, 0.3,0.4,0.5, 0.6),
    c('#c0e0b0',"#c9ecb4",'#44A57C','#35856A','#2B6857','#284E3F','#264234')
  ))|>
  formatStyle('2019', backgroundColor = styleInterval(
    c(0.1, 0.2, 0.3,0.4,0.5, 0.6),
    c('#c0e0b0',"#c9ecb4",'#44A57C','#35856A','#2B6857','#284E3F','#264234')
  ))|>
  formatStyle('2020', backgroundColor = styleInterval(
    c(0.1, 0.2, 0.3,0.4,0.5, 0.6),
    c('#c0e0b0',"#c9ecb4",'#44A57C','#35856A','#2B6857','#284E3F','#264234')
  ))|>
  formatStyle('2021', backgroundColor = styleInterval(
    c(0.1, 0.2, 0.3,0.4,0.5, 0.6),
    c('#c0e0b0',"#c9ecb4",'#44A57C','#35856A','#2B6857','#284E3F','#264234')
  ))|>
  formatStyle('2022', backgroundColor = styleInterval(
    c(0.1, 0.2, 0.3,0.4,0.5, 0.6),
    c('#c0e0b0',"#c9ecb4",'#44A57C','#35856A','#2B6857','#284E3F','#264234')
  ))|>
  formatStyle('2023', backgroundColor = styleInterval(
    c(0.1, 0.2, 0.3,0.4,0.5, 0.6),
    c('#c0e0b0',"#c9ecb4",'#44A57C','#35856A','#2B6857','#284E3F','#264234')
  ))|>
  formatStyle('2024', backgroundColor = styleInterval(
    c(0.1, 0.2, 0.3,0.4,0.5, 0.6),
    c('#c0e0b0',"#c9ecb4",'#44A57C','#35856A','#2B6857','#284E3F','#264234')
  ))

data[label == '国务院部门文件',by=.(childtype_1),.(freq=.N)][is.na(childtype_1)==FALSE][order(freq)]|>
  e_charts(childtype_1) |>
  e_polar() |>
  e_angle_axis(childtype_1) |>
  e_radius_axis() |>
  e_line(freq, name = "发文数量", coord_system = "polar") 

# 拆分发文机关列
n <- nrow(data[label == '国务院部门文件'])
childtype_data <- data[label == '国务院部门文件'][, list(
  childtype_1 = rep(childtype_1, lengths(puborg)),
  year = rep(year, lengths(puborg)),
  puborg = unlist(strsplit(puborg, " "))
), by = 1:n]

# 计算每个主题大类和年份的发文机关数量
dt_count <- childtype_data[year>=2018, .N, by = .(childtype_1, year, puborg)]

# 选择每个主题大类和年份中数量最多的发文机关
dt_max <- dt_count[dt_count[, .I[N == max(N)], by = .(childtype_1, year)]$V1]

dt_max_dcast <- dcast(
  dt_max,
  childtype_1 ~ year,
  value.var = c('puborg'),
  fun.aggregate = max
)
colnames(dt_max_dcast)[1]  <-"主题大类"

datatable(
  dt_max_dcast[is.na(`主题大类`) == FALSE],
  rownames = FALSE,
  options = list(pageLength = 20, dom = 't'),
  caption = '每个主题大类每年发文数量最多的机关名称'
)