From 73341c1bc9eadb0dfc99a64a5bfd367487a3e875 Mon Sep 17 00:00:00 2001
From: Caleb Evans <caleb@calebevans.me>
Date: Tue, 5 Mar 2024 20:38:53 -0800
Subject: [PATCH] Add audio message count to attachment totals analyzer

---
 ica/analyzers/attachment_totals.py   | 1 +
 ica/analyzers/totals_by_day.py       | 2 +-
 ica/queries/attachments.sql          | 4 +++-
 ica/queries/messages.sql             | 1 +
 tests/data/dbs/chats/attachment.json | 1 +
 5 files changed, 7 insertions(+), 2 deletions(-)

diff --git a/ica/analyzers/attachment_totals.py b/ica/analyzers/attachment_totals.py
index 78aedca..ae8cade 100644
--- a/ica/analyzers/attachment_totals.py
+++ b/ica/analyzers/attachment_totals.py
@@ -40,6 +40,7 @@ def main() -> None:
             .count()
             .item()
         ),
+        "audio_messages": (dfs.attachments["filename"].str.endswith(".caf").sum()),
         "recorded_videos": (dfs.attachments["mime_type"].eq("video/quicktime").sum()),
     }
     ica.output_results(
diff --git a/ica/analyzers/totals_by_day.py b/ica/analyzers/totals_by_day.py
index 5b86497..c069039 100644
--- a/ica/analyzers/totals_by_day.py
+++ b/ica/analyzers/totals_by_day.py
@@ -19,7 +19,7 @@ def main() -> None:
     )
     ica.output_results(
         (
-            dfs.messages
+            dfs.messages[["text", "is_from_me", "datetime", "is_reaction"]]
             # Count all "text" column values by converting them to integers
             # (always 1), because resampling the DataFrame will remove all
             # non-numeric columns
diff --git a/ica/queries/attachments.sql b/ica/queries/attachments.sql
index b0502e1..f347909 100644
--- a/ica/queries/attachments.sql
+++ b/ica/queries/attachments.sql
@@ -1,4 +1,6 @@
-SELECT "mime_type"
+SELECT "mime_type",
+       "filename",
+       "message_id"
 FROM "attachment"
 INNER JOIN "message_attachment_join"
     ON "attachment"."ROWID" = "attachment_id"
diff --git a/ica/queries/messages.sql b/ica/queries/messages.sql
index d336d38..cdff017 100644
--- a/ica/queries/messages.sql
+++ b/ica/queries/messages.sql
@@ -1,4 +1,5 @@
 SELECT
+    "ROWID",
     "text",
     "attributedBody",
     datetime("message"."date" / 1000000000 + strftime("%s", "2001-01-01") ,"unixepoch") as "datetime",
diff --git a/tests/data/dbs/chats/attachment.json b/tests/data/dbs/chats/attachment.json
index 4298009..cf54031 100644
--- a/tests/data/dbs/chats/attachment.json
+++ b/tests/data/dbs/chats/attachment.json
@@ -1,6 +1,7 @@
 [
   {
     "ROWID": "caa5c5fd-f301-4104-b0c4-934f2e52992b",
+    "filename": "abc.gif",
     "mime_type": "image/gif"
   }
 ]