|
| 1 | +defmodule GroupherServer.CMS.Delegate.CiteTasks do |
| 2 | + @moduledoc """ |
| 3 | + run tasks in every article blocks if need |
| 4 | +
|
| 5 | + current task: "cite link" and "mention" |
| 6 | +
|
| 7 | + ## cite link |
| 8 | +
|
| 9 | + 我被站内哪些文章或评论引用了，是值得关注的事 |
| 10 | + 我引用了谁不重要，帖子里链接已经表明了, 这和 github issue 的双向链接不一样，因为一般不需要关注这个 |
| 11 | + 帖子是否解决，是否被 merge 等状态。 |
| 12 | +
|
| 13 | + 基本结构： |
| 14 | +
|
| 15 | + cited_thread, cited_article_id, [xxx_article]_id, [block_id, block2_id], |
| 16 | +
|
| 17 | + POST post_333 -> cited_article_333, [block_id, block2_id]] |
| 18 | +
|
| 19 | + cited_type, cited_content_id, [contents]_id, [block_id, cited_block_id], |
| 20 | +
|
| 21 | + cited_type: thread or comment |
| 22 | + content: article or comment |
| 23 | + # cited_article_comment_id, [xxx_article]_id, [block_id, block2_id, ...], |
| 24 | + """ |
| 25 | + |
| 26 | + import Ecto.Query, warn: false |
| 27 | + import Helper.Utils, only: [get_config: 2, thread_of_article: 1, done: 1] |
| 28 | + import GroupherServer.CMS.Helper.Matcher |
| 29 | + import Helper.ErrorCode |
| 30 | + |
| 31 | + alias GroupherServer.{CMS, Repo} |
| 32 | + alias CMS.Model.CitedContent |
| 33 | + alias Helper.ORM |
| 34 | + |
| 35 | + alias Ecto.Multi |
| 36 | + |
| 37 | + @site_host get_config(:general, :site_host) |
| 38 | + @article_threads get_config(:article, :threads) |
| 39 | + @valid_article_prefix Enum.map(@article_threads, &"#{@site_host}/#{&1}/") |
| 40 | + |
| 41 | + def handle(%{body: body} = article) do |
| 42 | + with {:ok, %{"blocks" => blocks}} <- Jason.decode(body), |
| 43 | + article <- Repo.preload(article, author: :user) do |
| 44 | + Multi.new() |
| 45 | + |> Multi.run(:delete_all_cited_contents, fn _, _ -> |
| 46 | + delete_all_cited_contents(article) |
| 47 | + end) |
| 48 | + |> Multi.run(:update_cited_info, fn _, _ -> |
| 49 | + blocks |
| 50 | + |> Enum.reduce([], &(&2 ++ parse_cited_info_per_block(article, &1))) |
| 51 | + |> merge_same_cited_article_block |
| 52 | + |> update_cited_info |
| 53 | + end) |
| 54 | + |> Repo.transaction() |
| 55 | + |> result() |
| 56 | + end |
| 57 | + end |
| 58 | + |
| 59 | + # delete all records before insert_all, this will dynamiclly update |
| 60 | + # those cited info when update article |
| 61 | + # 插入引用记录之前先全部清除，这样可以在更新文章的时候自动计算引用信息 |
| 62 | + defp delete_all_cited_contents(article) do |
| 63 | + with {:ok, thread} <- thread_of_article(article), |
| 64 | + {:ok, info} <- match(thread) do |
| 65 | + query = from(c in CitedContent, where: field(c, ^info.foreign_key) == ^article.id) |
| 66 | + |
| 67 | + ORM.delete_all(query, :if_exist) |
| 68 | + end |
| 69 | + end |
| 70 | + |
| 71 | + # defp batch_done |
| 72 | + |
| 73 | + defp update_cited_info(cited_contents) do |
| 74 | + clean_cited_contents = Enum.map(cited_contents, &Map.delete(&1, :cited_article)) |
| 75 | + # IO.inspect(clean_cited_contents, label: "clean_cited_contents") |
| 76 | + with true <- {0, nil} !== Repo.insert_all(CitedContent, clean_cited_contents) do |
| 77 | + update_citing_count(cited_contents) |
| 78 | + else |
| 79 | + _ -> {:error, "insert cited content error"} |
| 80 | + end |
| 81 | + end |
| 82 | + |
| 83 | + defp update_citing_count(cited_contents) do |
| 84 | + Enum.all?(cited_contents, fn content -> |
| 85 | + count_query = from(c in CitedContent, where: c.cited_by_id == ^content.cited_by_id) |
| 86 | + count = Repo.aggregate(count_query, :count) |
| 87 | + |
| 88 | + cited_article = content.cited_article |
| 89 | + meta = Map.merge(cited_article.meta, %{citing_count: count}) |
| 90 | + |
| 91 | + case cited_article |> ORM.update_meta(meta) do |
| 92 | + {:ok, _} -> true |
| 93 | + {:error, _} -> false |
| 94 | + end |
| 95 | + end) |
| 96 | + |> done |
| 97 | + end |
| 98 | + |
| 99 | + @doc """ |
| 100 | + merge same cited article in different blocks |
| 101 | + e.g: |
| 102 | + [ |
| 103 | + %{ |
| 104 | + block_linker: ["block-zByQI"], |
| 105 | + cited_by_id: 190058, |
| 106 | + cited_by_type: "POST", |
| 107 | + post_id: 190059, |
| 108 | + user_id: 1413053 |
| 109 | + }, |
| 110 | + %{ |
| 111 | + block_linker: ["block-zByQI", "block-ZgKJs"], |
| 112 | + cited_by_id: 190057, |
| 113 | + cited_by_type: "POST", |
| 114 | + post_id: 190059, |
| 115 | + user_id: 1413053 |
| 116 | + }, |
| 117 | + ] |
| 118 | + """ |
| 119 | + defp merge_same_cited_article_block(cited_contents) do |
| 120 | + cited_contents |
| 121 | + |> Enum.reduce([], fn content, acc -> |
| 122 | + case Enum.find_index(acc, &(&1.cited_by_id == content.cited_by_id)) do |
| 123 | + nil -> |
| 124 | + acc ++ [content] |
| 125 | + |
| 126 | + index -> |
| 127 | + List.update_at( |
| 128 | + acc, |
| 129 | + index, |
| 130 | + &Map.merge(&1, %{block_linker: &1.block_linker ++ content.block_linker}) |
| 131 | + ) |
| 132 | + end |
| 133 | + end) |
| 134 | + end |
| 135 | + |
| 136 | + @doc """ |
| 137 | + return fmt like: |
| 138 | + [ |
| 139 | + %{ |
| 140 | + block_linker: ["block-ZgKJs"], |
| 141 | + cited_by_id: 190057, |
| 142 | + cited_by_type: "POST", |
| 143 | + cited_article: #loaded, |
| 144 | + post_id: 190059, |
| 145 | + user_id: 1413053 |
| 146 | + } |
| 147 | + ... |
| 148 | + ] |
| 149 | + """ |
| 150 | + defp parse_cited_info_per_block(article, %{"id" => block_id, "data" => %{"text" => text}}) do |
| 151 | + links_in_block = Floki.find(text, "a[href]") |
| 152 | + |
| 153 | + Enum.reduce(links_in_block, [], fn link, acc -> |
| 154 | + with {:ok, cited_article} <- parse_cited_article(link), |
| 155 | + # do not cite artilce itself |
| 156 | + true <- article.id !== cited_article.id do |
| 157 | + List.insert_at(acc, 0, shape_cited_content(article, cited_article, block_id)) |
| 158 | + else |
| 159 | + _ -> acc |
| 160 | + end |
| 161 | + end) |
| 162 | + |> Enum.uniq() |
| 163 | + end |
| 164 | + |
| 165 | + defp shape_cited_content(article, cited_article, block_id) do |
| 166 | + {:ok, thread} = thread_of_article(article) |
| 167 | + {:ok, info} = match(thread) |
| 168 | + |
| 169 | + %{ |
| 170 | + cited_by_id: cited_article.id, |
| 171 | + cited_by_type: cited_article.meta.thread, |
| 172 | + # used for updating citing_count, avoid load again |
| 173 | + cited_article: cited_article, |
| 174 | + block_linker: [block_id], |
| 175 | + user_id: article.author.user.id |
| 176 | + } |
| 177 | + |> Map.put(info.foreign_key, article.id) |
| 178 | + end |
| 179 | + |
| 180 | + defp parse_cited_article({"a", attrs, _}) do |
| 181 | + with {:ok, link} <- parse_link(attrs), |
| 182 | + true <- is_site_article_link?(link) do |
| 183 | + load_cited_article_from_url(link) |
| 184 | + end |
| 185 | + end |
| 186 | + |
| 187 | + @doc """ |
| 188 | + parse link from Floki parse result |
| 189 | +
|
| 190 | + e.g: |
| 191 | + [{"href", "https://coderplanets.com/post/190220", "bla", "bla"}] -> |
| 192 | + {:ok, "https://coderplanets.com/post/190220"} |
| 193 | + """ |
| 194 | + defp parse_link(attrs) do |
| 195 | + with {"href", link} <- Enum.find(attrs, fn {a, _v} -> a == "href" end) do |
| 196 | + {:ok, link} |
| 197 | + else |
| 198 | + _ -> {:error, "invalid fmt"} |
| 199 | + end |
| 200 | + end |
| 201 | + |
| 202 | + # 检测是否是站内文章的链接 |
| 203 | + defp is_site_article_link?(url) do |
| 204 | + Enum.any?(@valid_article_prefix, &String.starts_with?(url, &1)) |
| 205 | + end |
| 206 | + |
| 207 | + # get cited article from url |
| 208 | + # e.g: https://coderplanets.com/post/189993 -> ORM.find(Post, 189993) |
| 209 | + defp load_cited_article_from_url(url) do |
| 210 | + %{path: path} = URI.parse(url) |
| 211 | + path_list = path |> String.split("/") |
| 212 | + thread = path_list |> Enum.at(1) |> String.downcase() |> String.to_atom() |
| 213 | + article_id = path_list |> Enum.at(2) |
| 214 | + |
| 215 | + with {:ok, info} <- match(thread) do |
| 216 | + ORM.find(info.model, article_id) |
| 217 | + end |
| 218 | + end |
| 219 | + |
| 220 | + defp result({:ok, %{update_cited_info: result}}), do: {:ok, result} |
| 221 | + |
| 222 | + defp result({:error, :update_cited_info, _result, _steps}) do |
| 223 | + {:error, [message: "cited article", code: ecode(:cite_artilce)]} |
| 224 | + end |
| 225 | +end |
0 commit comments