Porting the facebook submodule to tabular records

medialab · Dec 12, 2023 · 99e8fcd · 99e8fcd
1 parent 719aa4f
commit 99e8fcd
Show file tree

Hide file tree

Showing 8 changed files with 70 additions and 99 deletions.
diff --git a/minet/cli/facebook/comments.py b/minet/cli/facebook/comments.py
@@ -7,13 +7,13 @@
 from minet.cli.utils import with_enricher_and_loading_bar
 from minet.cli.facebook.utils import with_facebook_fatal_errors
 from minet.facebook import FacebookMobileScraper
-from minet.facebook.constants import FACEBOOK_COMMENT_CSV_HEADERS
+from minet.facebook.types import MobileFacebookComment
 from minet.facebook.exceptions import FacebookInvalidTargetError
 
 
 @with_facebook_fatal_errors
 @with_enricher_and_loading_bar(
-    headers=FACEBOOK_COMMENT_CSV_HEADERS,
+    headers=MobileFacebookComment,
     title="Scraping comments",
     unit="posts",
     nested=True,

diff --git a/minet/cli/facebook/post.py b/minet/cli/facebook/post.py
@@ -10,9 +10,7 @@
     print_translation_warning_if_needed,
 )
 from minet.facebook import FacebookMobileScraper
-from minet.facebook.constants import (
-    FACEBOOK_POST_WITH_REACTIONS_CSV_HEADERS,
-)
+from minet.facebook.types import MobileFacebookPostWithReactions
 from minet.facebook.exceptions import (
     FacebookInvalidTargetError,
     FacebookNotPostError,
@@ -28,7 +26,7 @@
 
 @with_facebook_fatal_errors
 @with_enricher_and_loading_bar(
-    headers=FACEBOOK_POST_WITH_REACTIONS_CSV_HEADERS,
+    headers=MobileFacebookPostWithReactions,
     title="Scraping posts",
     unit="posts",
 )

diff --git a/minet/cli/facebook/post_authors.py b/minet/cli/facebook/post_authors.py
@@ -7,13 +7,13 @@
 from minet.cli.utils import with_enricher_and_loading_bar
 from minet.cli.facebook.utils import with_facebook_fatal_errors
 from minet.facebook import FacebookMobileScraper
-from minet.facebook.constants import FACEBOOK_USER_CSV_HEADERS
+from minet.facebook.types import MobileFacebookUser
 from minet.facebook.exceptions import FacebookInvalidTargetError
 
 
 @with_facebook_fatal_errors
 @with_enricher_and_loading_bar(
-    headers=FACEBOOK_USER_CSV_HEADERS, title="Finding authors", unit="posts"
+    headers=MobileFacebookUser, title="Finding authors", unit="posts"
 )
 def action(cli_args, enricher, loading_bar):
     scraper = FacebookMobileScraper(cli_args.cookie, throttle=cli_args.throttle)

diff --git a/minet/cli/facebook/posts.py b/minet/cli/facebook/posts.py
@@ -10,13 +10,13 @@
     print_translation_warning_if_needed,
 )
 from minet.facebook import FacebookMobileScraper
-from minet.facebook.constants import FACEBOOK_POST_CSV_HEADERS
+from minet.facebook.types import MobileFacebookPost
 from minet.facebook.exceptions import FacebookInvalidTargetError
 
 
 @with_facebook_fatal_errors
 @with_enricher_and_loading_bar(
-    headers=FACEBOOK_POST_CSV_HEADERS,
+    headers=MobileFacebookPost,
     title="Scraping group posts",
     unit="groups",
     nested=True,

diff --git a/minet/facebook/constants.py b/minet/facebook/constants.py
@@ -19,56 +19,6 @@
 )
 FACEBOOK_WEB_RATE_LIMITER_STATE = RateLimiterState(1, FACEBOOK_WEB_DEFAULT_THROTTLE)
 
-FACEBOOK_USER_CSV_HEADERS = ["user_label", "user_id", "user_handle", "user_url"]
-
-FACEBOOK_COMMENT_CSV_HEADERS = [
-    "post_id",
-    "id",
-    "user_id",
-    "user_handle",
-    "user_url",
-    "user_label",
-    "text",
-    "html",
-    "formatted_date",
-    "date",
-    "reactions",
-    "replies",
-    "in_reply_to",
-]
-
-FACEBOOK_POST_CSV_HEADERS = [
-    "url",
-    "user_id",
-    "user_handle",
-    "user_url",
-    "user_label",
-    "text",
-    "html",
-    "translated_text",
-    "translated_html",
-    "translated_from",
-    "formatted_date",
-    "date",
-    "reactions",
-    "comments",
-]
-
-FACEBOOK_POST_STATS_CSV_HEADERS = [
-    "error",
-    "canonical",
-    "account_name",
-    "timestamp",
-    "time",
-    "link",
-    "aria_label",
-    "text",
-    "share_count",
-    "comment_count",
-    "reaction_count",
-    "video_view_count",
-]
-
 FACEBOOK_REACTION_KEYS = OrderedDict(
     {
         1: "like",
@@ -82,7 +32,3 @@
         16: "care",
     }
 )
-
-FACEBOOK_POST_WITH_REACTIONS_CSV_HEADERS = FACEBOOK_POST_CSV_HEADERS + [
-    "reactions_types"
-]
diff --git a/minet/facebook/formatters.py b/minet/facebook/formatters.py
diff --git a/minet/facebook/mobile_scraper.py b/minet/facebook/mobile_scraper.py
@@ -37,11 +37,11 @@
 from minet.scrape.utils import BeautifulSoupWithoutXHTMLWarnings
 from minet.scrape.std import get_display_text
 from minet.facebook.utils import grab_facebook_cookie
-from minet.facebook.formatters import (
-    FacebookComment,
-    FacebookPost,
-    FacebookUser,
-    FacebookPostWithReaction,
+from minet.facebook.types import (
+    MobileFacebookComment,
+    MobileFacebookUser,
+    MobileFacebookPost,
+    MobileFacebookPostWithReactions,
 )
 from minet.facebook.exceptions import (
     FacebookInvalidCookieError,
@@ -80,7 +80,7 @@ def resolve_relative_url(url):
     return urljoin(FACEBOOK_MOBILE_URL, url)
 
 
-def scrape_comments(html, direction=None, in_reply_to=None):
+def scrape_comments(html, direction=None, in_reply_to=None) -> MobileFacebookComment:
     soup = BeautifulSoupWithoutXHTMLWarnings(html, "lxml")
 
     data = {
@@ -139,7 +139,7 @@ def scrape_comments(html, direction=None, in_reply_to=None):
     )
 
     for item in valid_items:
-        item_id = item.get("id")
+        item_id = item["id"]
 
         # Skipping comment if same as commented
         if item_id == in_reply_to:
@@ -206,7 +206,7 @@ def scrape_comments(html, direction=None, in_reply_to=None):
                     data["replies"].append((resolve_relative_url(replies_url), item_id))
 
         data["comments"].append(
-            FacebookComment(
+            MobileFacebookComment(
                 post_id=post_id,
                 id=item_id,
                 user_id=getattr(user, "id", ""),
@@ -302,7 +302,7 @@ def scrape_posts(html):
             else None
         )
 
-        post = FacebookPost(
+        post = MobileFacebookPost(
             url=post_url,
             user_id=getattr(user, "id", ""),
             user_handle=getattr(user, "handle", ""),
@@ -401,7 +401,7 @@ def scrape_video(soup):
         else None
     )
 
-    post = FacebookPostWithReaction(
+    post = MobileFacebookPostWithReactions(
         url=video_url,
         user_id=getattr(user, "id", ""),
         user_handle=getattr(user, "handle", ""),
@@ -492,7 +492,7 @@ def scrape_photo(soup):
         else None
     )
 
-    post = FacebookPostWithReaction(
+    post = MobileFacebookPostWithReactions(
         url=photo_url,
         user_id=getattr(user, "id", ""),
         user_handle=getattr(user, "handle", ""),
@@ -597,7 +597,7 @@ def scrape_post(html):
         else None
     )
 
-    post = FacebookPostWithReaction(
+    post = MobileFacebookPostWithReactions(
         url=post_url,
         user_id=getattr(user, "id", ""),
         user_handle=getattr(user, "handle", ""),
@@ -765,8 +765,8 @@ def post_author(self, url):
         user_label = user_item.get_text().strip()
 
         if isinstance(parsed, ParsedFacebookHandle):
-            return FacebookUser(user_label, None, parsed.handle, parsed.url)
+            return MobileFacebookUser(user_label, None, parsed.handle, parsed.url)
         elif isinstance(parsed, ParsedFacebookUser):
-            return FacebookUser(user_label, parsed.id, parsed.handle, parsed.url)
+            return MobileFacebookUser(user_label, parsed.id, parsed.handle, parsed.url)
         else:
             raise TypeError
diff --git a/minet/facebook/types.py b/minet/facebook/types.py
@@ -111,3 +111,51 @@ def walk(entries: Iterable[Tuple["FacebookComment", List["FacebookComment"]]]):
         walk(filter(lambda entry: entry[0].depth == 0, index.values()))
 
         return sorted_comments
+
+
+@dataclass
+class MobileFacebookComment(TabularRecord):
+    post_id: str
+    id: str
+    user_id: str
+    user_handle: str
+    user_url: str
+    user_label: str
+    text: str
+    html: str
+    formatted_date: str
+    date: Optional[str]
+    reactions: str
+    replies: str
+    in_reply_to: Optional[str]
+
+
+@dataclass
+class MobileFacebookUser(TabularRecord):
+    label: str
+    id: Optional[str]
+    handle: Optional[str]
+    url: str
+
+
+@dataclass
+class MobileFacebookPost(TabularRecord):
+    url: str
+    user_id: str
+    user_handle: str
+    user_url: str
+    user_label: str
+    text: str
+    html: str
+    translated_text: Optional[str]
+    translated_html: Optional[str]
+    translated_from: Optional[str]
+    formatted_date: str
+    date: Optional[str]
+    reactions: str
+    comments: str
+
+
+@dataclass
+class MobileFacebookPostWithReactions(MobileFacebookPost):
+    reactions_types: Optional[str]