From ef5ce9060080dd00a49587be92e614d05afa46f5 Mon Sep 17 00:00:00 2001
From: Joshua Tauberer <tauberer@govtrack.us>
Date: Sat, 14 May 2022 11:47:03 +0000
Subject: [PATCH] Add a diff option for bills to show file changes before
 writing updates to disk

---
 congress/tasks/bills.py | 23 +++++++++++++++++++----
 congress/tasks/utils.py | 27 +++++++++++----------------
 2 files changed, 30 insertions(+), 20 deletions(-)

diff --git a/congress/tasks/bills.py b/congress/tasks/bills.py
index 694f9cd..92aec56 100644
--- a/congress/tasks/bills.py
+++ b/congress/tasks/bills.py
@@ -135,7 +135,10 @@ def process_bill(bill_id, options):
     # Convert and write out data.json and data.xml.
     utils.write(
         json.dumps(bill_data, indent=2, sort_keys=True),
-        os.path.dirname(fdsys_xml_path) + '/data.json')
+        os.path.dirname(fdsys_xml_path) + '/data.json',
+        {
+            "diff": options.get("diff")
+        })
 
     from bill_info import create_govtrack_xml
     with open(os.path.dirname(fdsys_xml_path) + '/data.xml', 'wb') as xml_file:
@@ -148,7 +151,10 @@ def process_bill(bill_id, options):
     # file under a new path.
     utils.write(
         utils.read(_path_to_billstatus_file(bill_id).replace(".xml", "-lastmod.txt")),
-        os.path.join(os.path.dirname(fdsys_xml_path), "data-fromfdsys-lastmod.txt"))
+        os.path.join(os.path.dirname(fdsys_xml_path), "data-fromfdsys-lastmod.txt"),
+        {
+            "diff": options.get("diff")
+        })
 
     return {
         "ok": True,
@@ -291,9 +297,18 @@ def reparse_actions(bill_id, options):
 
     wrote_any = False
 
+    if options.get("diff"):
+        confirmer = utils.show_diff_ask_ok
+    else:
+        # If no --diff is given, just check that
+        # the content hasn't changed --- don't bother
+        # writing out anything with identical content.
+        def confirmer(source, revised, fn):
+            return source != revised
+
     # Write new data.json file.
     revised = json.dumps(bill_data, indent=2, sort_keys=True)
-    if utils.show_diff_ask_ok(source, revised, data_json_fn):
+    if confirmer(source, revised, data_json_fn):
       utils.write(revised, data_json_fn)
       wrote_any = True
 
@@ -303,7 +318,7 @@ def reparse_actions(bill_id, options):
     with open(data_xml_fn, 'r') as xml_file:
         source = xml_file.read()
     revised = create_govtrack_xml(bill_data, options)
-    if utils.show_diff_ask_ok(source, revised.decode("utf8"), data_xml_fn):
+    if confirmer(source, revised.decode("utf8"), data_xml_fn):
       with open(data_xml_fn, 'wb') as xml_file:
         xml_file.write(revised)
       wrote_any = True
diff --git a/congress/tasks/utils.py b/congress/tasks/utils.py
index d46a9ba..ad0159f 100644
--- a/congress/tasks/utils.py
+++ b/congress/tasks/utils.py
@@ -348,31 +348,26 @@ def write(content, destination, options={}):
     if options.get("diff"):
         # Instead of writing the file, do a comparison with what's on disk
         # to test any changes. But be nice and replace any update date with
-        # what's in the previous file so we avoid spurrious changes. Use
-        # how updated_at appears in the JSON and in the XML.
+        # what's in the previous file so we avoid spurrious changes in the
+        # diff. Use how updated_at appears in the JSON and in the XML.
         if os.path.exists(destination):
             with open(destination) as f:
-                existing_content = f.read()
+                source = f.read()
+            revised = content
             for pattern in ('"updated_at": ".*?"', 'updated=".*?"'):
-                m1 = re.search(pattern, existing_content)
-                m2 = re.search(pattern, content)
+                m1 = re.search(pattern, source)
+                m2 = re.search(pattern, revised)
                 if m1 and m2:
-                    content = content.replace(m2.group(0), m1.group(0))
+                    revised = revised.replace(m2.group(0), m1.group(0))
 
             # Avoid writing to disk and spawning `diff` by checking if
             # the files match in memory.
-            if content == existing_content:
+            if revised == source:
                 return
 
-        # Shell `diff` and let it display output directly to the console.
-        # Write `content` to disk first so diff can see it. Maybe more
-        # efficient to pipe?
-        fn = "/tmp/congress-changed-file"
-        with open(fn, 'w') as f:
-            f.write(content)
-        os.system("diff -u %s %s" % (destination, fn))
-        os.unlink(fn)
-        return
+            if not show_diff_ask_ok(source, revised, destination):
+                # User cancelled save.
+                return
 
     # Save the content to disk.
     mkdir_p(os.path.dirname(destination))