From d2f65d217a9f0be624f84edc734f0413ddc38b82 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Erik=20Bj=C3=A4reholt?= <erik@bjareho.lt>
Date: Sun, 5 Nov 2023 17:10:36 +0100
Subject: [PATCH] fix: added some comments, and strip.sh

---
 gptme/tools/reduce.py   | 1 +
 scripts/describe_api.py | 2 ++
 scripts/strip.sh        | 6 ++++++
 3 files changed, 9 insertions(+)
 create mode 100644 scripts/strip.sh

diff --git a/gptme/tools/reduce.py b/gptme/tools/reduce.py
index 1f4fe1c47..e292dc4de 100644
--- a/gptme/tools/reduce.py
+++ b/gptme/tools/reduce.py
@@ -62,6 +62,7 @@ def reduce_log(
 
 def truncate_msg(msg: Message, lines_pre=10, lines_post=10) -> Message | None:
     """Truncates message codeblocks to the first and last `lines_pre` and `lines_post` lines, keeping the rest as `[...]`."""
+    # TODO: also truncate long <details> (as can be found in GitHub issue comments)
     content_staged = msg.content
 
     # Truncate long codeblocks
diff --git a/scripts/describe_api.py b/scripts/describe_api.py
index ba63640ed..30025ae41 100644
--- a/scripts/describe_api.py
+++ b/scripts/describe_api.py
@@ -1,5 +1,7 @@
 """
 Describes the API of a given Python file such that a LLM can easily get a summary of the API in a codefile (and later an entire project).
+
+NOTE: this is easier to do with ctags
 """
 import click
 import jedi
diff --git a/scripts/strip.sh b/scripts/strip.sh
new file mode 100644
index 000000000..a51939428
--- /dev/null
+++ b/scripts/strip.sh
@@ -0,0 +1,6 @@
+#gh issue view $ISSUE_NUMBER > issue.md
+#gh issue view $ISSUE_NUMBER -c > comments.md
+
+# strip long <details>...</details> from issue.md and comments.md
+perl -0777 -i -pe 's/\n<details>.*?<\/details>//sg' issue.md
+perl -0777 -i -pe 's/\n<details>.*?<\/details>//sg' comments.md