From bd22d26291cdd666b382812e836f0c22f394528e Mon Sep 17 00:00:00 2001
From: Nathan TeBlunthuis <nathanteblunthuis@gmail.com>
Date: Thu, 29 May 2025 18:02:14 -0700
Subject: [PATCH] update deps and add edit_summary to wikiq output.

---
 pyproject.toml | 9 ++++-----
 wikiq          | 8 ++++++--
 2 files changed, 10 insertions(+), 7 deletions(-)

diff --git a/pyproject.toml b/pyproject.toml
index 59e1043..368239a 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -14,12 +14,11 @@ dependencies = [
     "yamlconf",
 ]
 
-# [tool.uv.sources]
-# yamlconf = { git = "https://github.com/groceryheist/yamlconf" }
-# mwxml = { git = "https://github.com/groceryheist/python-mwxml" }
+[tool.uv.sources]
+yamlconf = { git = "https://github.com/groceryheist/yamlconf" }
+mwxml = { git = "https://github.com/groceryheist/python-mwxml" }
 
 [dependency-groups]
 dev = [
-    "pandas>=2.1.0",
-    "pytest>=8.3.5",
+    "pandas>=2.1.0"
 ]
diff --git a/wikiq b/wikiq
index 290225b..5ce219d 100755
--- a/wikiq
+++ b/wikiq
@@ -242,6 +242,7 @@ class RevDataBase:
     title: str
     namespace: int
     deleted: bool
+    edit_summary: str
     text_chars: int = None
     revert: bool = None
     reverteds: list[int] = None
@@ -271,7 +272,8 @@ class RevDataBase:
         pa.field("sha1",pa.string()),
         pa.field("minor",pa.bool_()),
         pa.field("editor",pa.string()),
-        pa.field("anon",pa.bool_())
+        pa.field("anon",pa.bool_()),
+        pa.field("edit_summary",pa.bool_())
     ]
 
     # pyarrow is a columnar format, so most of the work happens in the flush_parquet_buffer function
@@ -535,7 +537,8 @@ class WikiqParser:
                                              editorid="" if rev.deleted.user == True or rev.user.id is None else rev.user.id,
                                              title=page.title,
                                              deleted=rev.deleted.text,
-                                             namespace=namespace
+                                             namespace=namespace,
+                                             edit_summary=rev.comment
                                              )
 
                 rev_data = self.matchmake_revision(rev, rev_data)
@@ -717,6 +720,7 @@ def match_archive_suffix(input_filename):
 
 def open_input_file(input_filename, fandom_2020=False):
     cmd = match_archive_suffix(input_filename)
+
     if fandom_2020:
         cmd.append("*.xml")
     try: