Dev/easonlee

.github/workflows/weread.yml

@@ -3,7 +3,7 @@ name: weread note sync
 on:
   workflow_dispatch:
   schedule:
-    - cron: "0 0 * * *"
+    - cron: "0 * * * *"
 concurrency:
   group: ${{ github.workflow }}-${{ github.ref }}
   cancel-in-progress: true
@@ -33,6 +33,8 @@ jobs:
     steps:
       - name: Checkout
         uses: actions/checkout@v3
+        with:
+          ref: dev/easonlee
       - name: Set up Python
         uses: actions/setup-python@v4
         with:
@@ -47,4 +49,4 @@ jobs:
       - name: weread sync
         run: |
           python -u scripts/weread.py
-
+

README.md

@@ -20,7 +20,7 @@
 
 | 微信群 | QQ群 |
 | --- | --- |
-| <div align="center"><img src="https://github.com/malinkang/weread2notion/assets/3365208/e28d1fb8-a861-47e1-b5e8-b25f17a6a285" ></div> | <div align="center"><img src="https://images.malinkang.com/2024/04/b225b17d60670e4a6ff3459bbde80d28.jpg" width="50%"></div> |
+| <div align="center"><img src="https://images.malinkang.com/2024/04/2dba3fdad0dbe53663c4de076b36baed.jpg" ></div> | <div align="center"><img src="https://images.malinkang.com/2024/04/b225b17d60670e4a6ff3459bbde80d28.jpg" width="50%"></div> |
 
 
 ## 捐赠

scripts/book.py

@@ -1,4 +1,5 @@
 import argparse
+import traceback
 from datetime import datetime, timedelta
 import os
 
@@ -10,6 +11,8 @@
 import utils
 from config import book_properties_type_dict, tz
 from retrying import retry
+import json
+from datetime import datetime
 
 TAG_ICON_URL = "https://www.notion.so/icons/tag_gray.svg"
 USER_ICON_URL = "https://www.notion.so/icons/user-circle-filled_gray.svg"
@@ -48,6 +51,7 @@ def insert_book_to_notion(books, index, bookId):
         book["书架分类"] = archive_dict.get(bookId)
     if bookId in notion_books:
         book.update(notion_books.get(bookId))
+
     bookInfo = weread_api.get_bookinfo(bookId)
     if bookInfo != None:
         book.update(bookInfo)
@@ -62,13 +66,13 @@ def insert_book_to_notion(books, index, bookId):
     markedStatus = book.get("markedStatus")
     status = "想读"
     if markedStatus == 4:
-        status = "已读"
+        status = "阅读完"
     elif book.get("readingTime", 0) >= 60:
         status = "在读"
     book["阅读状态"] = status
     book["阅读时长"] = book.get("readingTime")
     book["阅读天数"] = book.get("totalReadDay")
-    book["评分"] = book.get("newRating")
+    #book["评分"] = book.get("newRating")
     if book.get("newRatingDetail") and book.get("newRatingDetail").get("myRating"):
         book["我的评分"] = rating.get(book.get("newRatingDetail").get("myRating"))
     elif status == "已读":
@@ -81,43 +85,53 @@ def insert_book_to_notion(books, index, bookId):
     elif book.get("readingBookDate"):
         date = book.get("readingBookDate")
     book["时间"] = date
-    book["开始阅读时间"] = book.get("beginReadingDate")
-    book["最后阅读时间"] = book.get("lastReadingDate")
+    book["类型"] = "书籍"
+    #book["开始阅读时间"] = book.get("beginReadingDate")
+    #book["最后阅读时间"] = book.get("lastReadingDate")
     cover = book.get("cover").replace("/s_", "/t7_")
     if not cover and not cover.strip() and not cover.startswith("http"):
         cover = BOOK_ICON_URL
     if bookId not in notion_books:
         isbn = book.get("isbn")
-        if isbn and isbn.strip():
-            douban_url = get_douban_url(isbn)
-            if douban_url:
-                book["douban_url"] = douban_url
-        book["书名"] = book.get("title")
+        #链接timeout，先注释
+        #if isbn and isbn.strip():
+        #    douban_url = get_douban_url(isbn)
+        #    if douban_url:
+        #        book["douban_url"] = douban_url
+        book["标题"] = book.get("title")
         book["BookId"] = book.get("bookId")
         book["ISBN"] = book.get("isbn")
         book["链接"] = utils.get_weread_url(bookId)
-        book["简介"] = book.get("intro")
+        #book["简介"] = book.get("intro")
         book["作者"] = [
             notion_helper.get_relation_id(
                 x, notion_helper.author_database_id, USER_ICON_URL
             )
             for x in book.get("author").split(" ")
         ]
-        if book.get("categories"):
-            book["分类"] = [
-                notion_helper.get_relation_id(
-                    x.get("title"), notion_helper.category_database_id, TAG_ICON_URL
-                )
-                for x in book.get("categories")
-            ]
+        #if book.get("categories"):
+        #    book["分类"] = [
+        #        notion_helper.get_relation_id(
+        #            x.get("title"), notion_helper.category_database_id, TAG_ICON_URL
+        #        )
+        #        for x in book.get("categories")
+        #    ]
     properties = utils.get_properties(book, book_properties_type_dict)
+
+    #print(f"正在插入《{book.get('title')}》,一共{len(books)}本，当前是第{index+1}本。")
+    if not book.get("readDetail") or not book.get("readDetail").get("data"):
+        print(f"《{book.get('title')}》没有阅读记录，跳过")
+        return
+
     if book.get("时间"):
-        notion_helper.get_date_relation(
+        #取书籍的阅读记录，可能有多天
+        book['时间'] = [x['readDate'] for x in book.get('readDetail').get('data')]
+        notion_helper.get_date_relations(
             properties,
-            pendulum.from_timestamp(book.get("时间"), tz="Asia/Shanghai"),
+            #时间戳转换为北京时间
+            [pendulum.from_timestamp(x, tz="Asia/Shanghai") for x in book.get("时间")]
         )
 
-    print(f"正在插入《{book.get('title')}》,一共{len(books)}本，当前是第{index+1}本。")
     parent = {"database_id": notion_helper.book_database_id, "type": "database_id"}
     result = None
     if bookId in notion_books:
@@ -142,6 +156,7 @@ def insert_book_to_notion(books, index, bookId):
 def insert_read_data(page_id, readTimes):
     readTimes = dict(sorted(readTimes.items()))
     filter = {"property": "书架", "relation": {"contains": page_id}}
+    #拿到当前书籍$page_id的所有阅读记录
     results = notion_helper.query_all_by_book(notion_helper.read_database_id, filter)
     for result in results:
         timestamp = result.get("properties").get("时间戳").get("number")
@@ -160,6 +175,7 @@ def insert_read_data(page_id, readTimes):
         insert_to_notion(None, int(key), value, page_id)
 
 
+#插入或更新阅读记录
 def insert_to_notion(page_id, timestamp, duration, book_database_id):
     parent = {"database_id": notion_helper.read_database_id, "type": "database_id"}
     properties = {
@@ -186,36 +202,78 @@ def insert_to_notion(page_id, timestamp, duration, book_database_id):
 
 
 if __name__ == "__main__":
+    current_time = datetime.now()
+    print("开始同步阅读记录，当前时间: ", current_time)
     weread_api = WeReadApi()
     notion_helper = NotionHelper()
     notion_books = notion_helper.get_all_book()
     bookshelf_books = weread_api.get_bookshelf()
+    # 有阅读记录的图书信息
+    # 样例数据：{'$bookId': {'bookId': '26062915', 'progress': 1, 'chapterUid': 8, 'chapterOffset': 0,
+    #   'chapterIdx': 8, 'appId'： '11413501'， 'updateTime': 1691221509, 'readingTime': 875,
+    #   'syncKey': 720606794}
+    # }
     bookProgress = bookshelf_books.get("bookProgress")
     bookProgress = {book.get("bookId"): book for book in bookProgress}
     archive_dict = {}
+    #acchive是书单的名字
     for archive in bookshelf_books.get("archive"):
         name = archive.get("name")
         bookIds = archive.get("bookIds")
         archive_dict.update({bookId: name for bookId in bookIds})
     not_need_sync = []
+    # Notion里的文献笔记
+    # 样例数据：{'$bookId': {'pageId': '10c911dc-da56-8106-8b58-ddaacc287c74', 'readingTime': 2325,
+    # 'category': None, 'Sort': None, 'douban_url': None,
+    # 'cover': {'type': 'external', 'external': {'url': 'https://bts-image.xyzcdn.net/aHR0cHM6Ly9pbWFnZS54eXpjZG4ubmV0L0ZxUWs2VThtWDU0YnZ3MFBsbm5HemtHMVpEajkuanBn.jpg'}},
+    # 'myRating': None, 'comment': None, 'status': '阅读完'}
+    # }
     for key, value in notion_books.items():
         if (
             (
                 key not in bookProgress
                 or value.get("readingTime") == bookProgress.get(key).get("readingTime")
             )
             and (archive_dict.get(key) == value.get("category"))
-            and (value.get("cover") is not None)
-            and (
-                value.get("status") != "已读"
-                or (value.get("status") == "已读" and value.get("myRating"))
-            )
+            #and (value.get("cover") is not None)
+            #and (
+            #    value.get("status") != "已读"
+            #    or (value.get("status") == "已读" and value.get("myRating"))
+            #)
         ):
+            #这里判定Notion中的书籍和微信中的书籍是否有属性发现变化，没有的话就不需要同步了
             not_need_sync.append(key)
+            #continue
     notebooks = weread_api.get_notebooklist()
+    notebooks_map = {d["bookId"] : d for d in notebooks if "bookId" in d}
     notebooks = [d["bookId"] for d in notebooks if "bookId" in d]
+
     books = bookshelf_books.get("books")
+    bookshelf_map = {d["bookId"] : d for d in books if "bookId" in d}
     books = [d["bookId"] for d in books if "bookId" in d]
+
+
     books = list((set(notebooks) | set(books)) - set(not_need_sync))
+    len = len(books)
     for index, bookId in enumerate(books):
-        insert_book_to_notion(books, index, bookId)
+        try:
+            bp = bookProgress[bookId] if bookId in bookProgress  else None
+            bs = bookshelf_map[bookId] if bookId in bookshelf_map else None
+            nt = notebooks_map[bookId] if bookId in notebooks_map else None
+
+            title = bs['title'] if bs and 'title' in bs else ""
+            print(f"正在插入《{title}》,一共{len}本，当前是第{index + 1}本。")
+
+            if bp and bp['readingTime'] < 600:
+                print(f"《{title}》阅读不超时10分钟，跳过")
+                continue
+
+            if not nt:
+                print(f"《{title}》没有阅读笔记，跳过")
+                continue
+
+            insert_book_to_notion(books, index, bookId)
+        except Exception as e:
+            print("处理book: " + bookId + "出现异常，跳过:")
+            traceback.print_exc()
+            continue

scripts/config.py

@@ -10,7 +10,7 @@
 SELECT = "select"
 
 book_properties_type_dict = {
-    "书名":TITLE,
+    "标题":TITLE,
     "BookId":RICH_TEXT,
     "ISBN":RICH_TEXT,
     "链接":URL,
@@ -30,5 +30,6 @@
     "书架分类":SELECT,
     "我的评分":SELECT,
     "豆瓣链接":URL,
+    "类型":SELECT
 }
 tz='Asia/Shanghai'