From 253c3d6110c3ab268eeafd6ba214003d8e58a46b Mon Sep 17 00:00:00 2001
From: Thomas Wade <gigatech55@gmail.com>
Date: Tue, 21 Apr 2020 04:39:28 +0930
Subject: [PATCH] Parse submissions

---
 Comment.py    |   7 ++-
 Submission.py | 163 +++++++++++++++++++++++++++++++++++++++++++++-----
 User.py       |   5 +-
 3 files changed, 156 insertions(+), 19 deletions(-)

diff --git a/Comment.py b/Comment.py
index ffcbec8..adc86dc 100644
--- a/Comment.py
+++ b/Comment.py
@@ -10,9 +10,12 @@ class Comment:
     A comment on a FurAffinity submission.
     """
 
-    def __init__(self):
+    def __init__(self, submission: Submission, id: int):
+        self.id = id
         self.author: User.User = None
         self.timestamp: datetime = None
-        self.submission: Submission.Submission = None
+        self.submission: Submission.Submission = submission
         self.parent: Comment = None
         self.replies: List[Comment] = None
+        self.hidden_by_page_owner = None
+        self.hidden_by_author = None
diff --git a/Submission.py b/Submission.py
index 6907beb..ddcbefa 100644
--- a/Submission.py
+++ b/Submission.py
@@ -11,20 +11,22 @@ class Submission:
     A FurAffinity submission.
     """
 
-    def __init__(self, id: int):
+    def __init__(self, api, id: int):
+        self._api = api
         self.id: int = id
-        self.title: str = None
-        self.author: User.User = None
-        self.description: str = None
-        self.tags: List[str] = None
-        self.timestamp: datetime = None
-        self.category: str = None
-        self.theme: str = None
-        self.species: str = None
-        self.gender: str = None
-        self.faves: int = None
-        self.comments: List[Comment.Comment] = None
-        self.views: int = None
+        self._title: str = None
+        self._author: User.User = None
+        self._description: str = None
+        self._tags: List[str] = None
+        self._timestamp: datetime = None
+        self._category: str = None
+        self._subcategory: str = None
+        self._theme: str = None
+        self._species: str = None
+        self._gender: str = None
+        self._faves: int = None
+        self._comments: List[Comment.Comment] = None
+        self._views: int = None
 
     @property
     def url(self):
@@ -33,5 +35,138 @@ class Submission:
         else:
             return None
 
+    @property
+    def title(self):
+        if not self._title:
+            self._get_submission()
+
+        return self._title
+
+    @property
+    def author(self):
+        if not self._author:
+            self._get_submission()
+
+        return self._author
+
+    @property
+    def description(self):
+        if not self._description:
+            self._get_submission()
+
+        return self._description
+
+    @property
+    def tags(self):
+        if not self._tags:
+            self._get_submission()
+
+        return self._tags
+
+    @property
+    def timestamp(self):
+        if not self._timestamp:
+            self._get_submission()
+
+        return self._timestamp
+
+    @property
+    def category(self):
+        if not self._category:
+            self._get_submission()
+
+        return self._category
+
+    @property
+    def subcategory(self):
+        if not self._subcategory:
+            self._get_submission()
+
+        return self._subcategory
+
+    @property
+    def species(self):
+        if not self._species:
+            self._get_submission()
+
+        return self._species
+
+    @property
+    def gender(self):
+        if not self._gender:
+            self._get_submission()
+
+        return self._gender
+
+    @property
+    def faves(self):
+        if not self._faves:
+            self._get_submission()
+
+        return self._faves
+
+    @property
+    def comments(self):
+        if not self._comments:
+            self._get_submission()
+
+        return self._comments
+
+    @property
+    def views(self):
+        if not self._views:
+            self._get_submission()
+
+        return self._views
+
+    def _get_submission(self):
+        soup = self._api.get_soup('{}/view/{}'.format(FAAPI.FA_BASE_URL, self.id))
+
+        # Parse all the easy stuff
+        self._title = soup.select('.submission-title')[0].text.strip()
+        self._author = User.User.new_or_cached(self._api, soup.select('.submission-id-sub-container a')[0].text.strip())
+        self._description = soup.select('.submission-description')[0].text.strip()
+        self._tags = [tag.text.strip() for tag in soup.select('.submission-sidebar .tags')]
+        self._timestamp = datetime.datetime.strptime(soup.select('.popup_date')[0].get('title'), '%b %d, %Y %H:%M %p')  # e.g. Nov 26, 2019 02:47 PM
+        self._category = soup.select('.category-name')[0].text.strip()
+        self._subcategory = soup.select('.type-name')[0].text.strip()
+        self._species = soup.select('.info div:nth-of-type(2) span')[0].text.strip()
+        self._gender = soup.select('.info div:nth-of-type(3) span')[0].text.strip()
+        self._faves = int(soup.select('.favorites .font-large')[0].text.strip())
+        self._views = int(soup.select('.views .font-large')[0].text.strip())
+
+        # Parse comments
+        self._comments = []
+        for comment_element in soup.select('.comment_container'):
+            # Parse all the easy stuff
+            comment = Comment.Comment(self, int(comment_element.select('.comment_anchor')[0].get('id')[4:]))  # Skip the 'cid:' in the ID e.g. cid:142785397
+
+            # TODO: Comment ancestry
+
+            # Toggle hidden status
+            if comment_element.select('.comment-deleted'):
+                # Deleted by its owner, set flags and skip remaining attributes
+                comment.hidden_by_author = True
+                comment.hidden_by_page_owner = False
+
+                self._comments.append(comment)
+                continue
+            elif 'collapsed_height' in comment_element.get('class'):
+                # Deleted by the page owner, set flags and skip remaining attributes
+                comment.hidden_by_author = False
+                comment.hidden_by_page_owner = True
+
+                self._comments.append(comment)
+                continue
+            else:
+                comment.hidden_by_author = False
+                comment.hidden_by_page_owner = False
+
+            # Parse content from non-hidden comments
+            comment.author = User.User.new_or_cached(self._api, comment_element.select('.comment_username')[0].text.strip())
+            comment.timestamp = datetime.datetime.fromtimestamp(int(comment_element.get('data-timestamp')))  # FIXME: Needs to account for timezone difference since these are server-local epochs. Seems to be hosted in New York's TZ
+
+            self._comments.append(comment)
+
     def __str__(self):
-        return '{} - {} ({}/view/{})'.format(self.author, self.title, FAAPI.FA_BASE_URL, self.id)
+        return '{} - {} ({}/view/{})'.format(self.author, self.title, self.url)
diff --git a/User.py b/User.py
index 1a6f361..5f0fc23 100644
--- a/User.py
+++ b/User.py
@@ -78,8 +78,7 @@ class User:
 
             # Get all the submissions on this page
             for item in soup.select('.gallery figure'):
-                sub = Submission.Submission(item.get('id')[4:])  # IDs look like 'sid-35908275', so we just skip the 'sid-'
-                sub.author = self
+                sub = Submission.Submission(self._api, int(item.get('id')[4:]))  # IDs look like 'sid-35908275', so we just skip the 'sid-'
                 submissions.append(sub)
 
         return submissions
@@ -100,7 +99,7 @@ class User:
 
             # Get all the submissions on this page
             for item in soup.select('.gallery figure'):
-                sub = Submission.Submission(item.get('id')[4:])  # IDs look like 'sid-35908275', so we just skip the 'sid-'
+                sub = Submission.Submission(self._api, int(item.get('id')[4:]))  # IDs look like 'sid-35908275', so we just skip the 'sid-'
                 submissions.append(sub)
 
         return submissions