feat: check preprint not journalArticle

closes: #137

CHANGELOG.md

@@ -7,6 +7,18 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+### Minor Breaking Changes
+
+- 检测重复条目和检测网页条目现在在每一次 lint 都执行，而不是仅在导入条目时执行。
+
+### Added
+
+- 检测是否有预印本被错误存为期刊文章。close: [#137](https://github.com/northword/zotero-format-metadata/issues/137)
+
+### Fixed
+
+- 修复前一版本从 Semantic Scholar 更新预印本时将期刊文章存为会议的问题。
+
 ## [1.14.0] - 2024-03-16
 
 ### Added

addon/content/preferences.xhtml

@@ -43,18 +43,24 @@
     <!-- 条目 -->
     <groupbox>
       <label><html:h2 data-l10n-id="section-item"></html:h2></label>
-      <!-- 添加时检测是否存在重复条目 -->
+      <!-- 检测是否存在重复条目 -->
       <checkbox
         data-l10n-id="item-noDuplication"
         preference="__prefsPrefix__.noDuplicationItems"
         native="true"
       />
-      <!-- 添加时检测网页条目 -->
+      <!-- 检测网页条目 -->
       <checkbox
         data-l10n-id="item-checkWebpage"
         preference="__prefsPrefix__.checkWebpage"
         native="true"
       />
+      <!-- 检测预印本误存为期刊文章 -->
+      <checkbox
+        data-l10n-id="item-NoPreprintJournalArticle"
+        preference="__prefsPrefix__.noPreprintJournalArticle"
+        native="true"
+      />
     </groupbox>
 
     <!-- 字段：标题 -->

addon/locale/en-US/addon.ftl

@@ -54,4 +54,8 @@ unimplemented = This feature is unimplemented.
 ## 重复条目弹窗
 dialog-dup-title = Duplicates found
 dialog-dup-desc = The currently imported item is a duplicate of the item in the library
-dialog-dup-button-merge = Go to merge
+dialog-dup-button-merge = Go to merge
+
+## 规则运行警告
+checkWebpage-warning = The URL of this WebPage item contains the domain of academic publisher, there may be an exception to the import, please confirm!
+NoPreprintJournalArticle-warning = The URL of this journal article item is the domain of the **preprint** server, please confirm that there may be an exception to the import!

addon/locale/en-US/preferences.ftl

@@ -20,6 +20,8 @@ item-noDuplication =
     .label = No duplicate items added
 item-checkWebpage = 
     .label = No webpage items linked to academic publishers
+item-NoPreprintJournalArticle =
+    .label = No journalArticle items have preprint url
 
 ## 标题
 section-title = Title

addon/locale/zh-CN/addon.ftl

@@ -54,4 +54,8 @@ unimplemented = 此功能尚未实现。
 ## 重复条目弹窗
 dialog-dup-title = 发现重复条目
 dialog-dup-desc = 以下条目与已有条目重复
-dialog-dup-button-merge = 前往合并条目窗格
+dialog-dup-button-merge = 前往合并条目窗格
+
+## 规则运行警告
+checkWebpage-warning = 监测到您导入了一个 WebPage 条目，其 URL 中包含了主要学术出版商的域名，\n导入可能存在异常，请确认！
+NoPreprintJournalArticle-warning = 监测到您导入了一个期刊文章条目，其 URL 中包含了预印本服务器的域名，\n导入可能存在异常，请确认！

addon/locale/zh-CN/preferences.ftl

@@ -20,6 +20,8 @@ item-noDuplication =
     .label = 不应添加重复条目
 item-checkWebpage = 
     .label = 网页条目不应连接到学术出版社（应为期刊文章或其他类型）
+item-NoPreprintJournalArticle =
+    .label = 预印本不应保存为期刊文章
 
 ## 标题
 section-title = 标题 Title

addon/prefs.js

@@ -4,6 +4,7 @@ pref("__prefsPrefix__.richtext.toolBar", true);
 pref("__prefsPrefix__.richtext.hotkey", true);
 pref("__prefsPrefix__.noDuplicationItems", true);
 pref("__prefsPrefix__.checkWebpage", true);
+pref("__prefsPrefix__.noPreprintJournalArticle", true);
 pref("__prefsPrefix__.titleSentenceCase", true);
 pref("__prefsPrefix__.creatorsCase", true);
 pref("__prefsPrefix__.lang", true);

src/modules/rules-presets.ts

@@ -8,6 +8,9 @@ function getStdLintRules() {
     // 作者、期刊、年、期、卷、页 -> 判断语言 -> 作者大小写 -> 匹配缩写 -> 匹配地点 -> 格式化日期 -> 格式化DOI
     const rules = [];
     // getPref("isEnableOtherFields") ? rules.push(new Rules.UpdateMetadata({ mode: "blank" })) : "";
+    getPref("noDuplicationItems") ? rules.push(new Rules.NoDuplicatItem({})) : "skip";
+    getPref("checkWebpage") ? rules.push(new Rules.NoWebPageItem({})) : "skip";
+    getPref("noPreprintJournalArticle") ? rules.push(new Rules.NoPreprintJournalArticle({})) : "skip";
     getPref("lang") ? rules.push(new Rules.UpdateItemLanguage({})) : "";
     getPref("creatorsCase") ? rules.push(new Rules.CapitalizeCreators({})) : "";
     getPref("titleSentenceCase") ? rules.push(new Rules.TitleSentenceCase({})) : "";
@@ -24,8 +27,6 @@ function getStdLintRules() {
 
 function getNewItemLintRules() {
     const rules = [];
-    getPref("checkWebpage") ? rules.push(new Rules.NoWebPageItem({})) : "skip";
-    getPref("noDuplicationItems") ? rules.push(new Rules.NoDuplicatItem({})) : "skip";
     getPref("lint.onAdded") ? rules.push(getStdLintRules()) : "";
     return rules.flat();
 }

src/modules/rules/index.ts

@@ -13,6 +13,7 @@ import TitleSentenceCase from "./field-title-sentence-case";
 import University from "./field-university";
 import NoDuplicatItem from "./no-duplicate-item";
 import NoExtraZeros from "./no-extra-zeros";
+import NoPreprintJournalArticle from "./no-journalArticle-preprint";
 import NoWebPageItem from "./no-webpage-item";
 import UpdateMetadata from "./retrive-metadata";
 import SetFieldValue from "./set-field-value";
@@ -36,4 +37,5 @@ export default {
     NoWebPageItem,
     ThesisType,
     University,
+    NoPreprintJournalArticle,
 };

src/modules/rules/no-journalArticle-preprint.ts

@@ -0,0 +1,27 @@
+import { getString } from "../../utils/locale";
+import { progressWindow } from "../../utils/logger";
+import { isStringMatchStringInArray } from "../../utils/str";
+import { RuleBase, RuleBaseOptions } from "./rule-base";
+
+// 当条目为 webpage ，且 url 为各期刊出版社时，警告
+
+const publisherUrlKeyWords = ["arxiv.org", "biorxiv.org", "medrxiv.org", "chinaxiv.org"];
+
+class NoPreprintJournalArticleOptions implements RuleBaseOptions {}
+
+export default class NoPreprintJournalArticle extends RuleBase<NoPreprintJournalArticleOptions> {
+    constructor(options: NoPreprintJournalArticleOptions) {
+        super(options);
+    }
+
+    apply(item: Zotero.Item): Zotero.Item | Promise<Zotero.Item> {
+        if (item.itemType !== "journalArticle") return item;
+        const url = item.getField("url");
+        if (typeof url == "string" && url !== "" && isStringMatchStringInArray(url, publisherUrlKeyWords)) {
+            ztoolkit.log("The url of this journalArticle item is match with domin of preprint publisher.");
+            // show alart todo: 对话框完善，通过 URL 获取 DOI 并通过 DOI 强制更新条目类别
+            progressWindow(getString("NoPreprintJournalArticle-warning"), "fail").startCloseTimer(100000);
+        }
+        return item;
+    }
+}

src/modules/rules/no-webpage-item.ts

@@ -1,3 +1,4 @@
+import { getString } from "../../utils/locale";
 import { progressWindow } from "../../utils/logger";
 import { isStringMatchStringInArray } from "../../utils/str";
 import { RuleBase, RuleBaseOptions } from "./rule-base";
@@ -102,10 +103,7 @@ export default class NoWebPageItem extends RuleBase<NoWebPageItemOptions> {
         if (typeof url == "string" && url !== "" && isStringMatchStringInArray(url, publisherUrlKeyWords)) {
             ztoolkit.log("The url of this webpage item is match with domin of publisher.");
             // show alart todo: 对话框完善，通过 URL 获取 DOI 并通过 DOI 强制更新条目类别
-            progressWindow(
-                "监测到您导入了一个 WebPage 条目，其 URL 中包含了主要学术出版商的域名，\n导入可能存在异常，请确认！",
-                "fail",
-            ).startCloseTimer(10000);
+            progressWindow(getString("checkWebpage-warning"), "fail").startCloseTimer(100000);
         }
         ztoolkit.log(`The url of this webpage item is not belong to publisher, maybe a normal webpage.`);
         return item;