From 96f88f4cbe732d7ac48661ac50d9a843a1e26c2b Mon Sep 17 00:00:00 2001
From: Becky Smith <becky.smith@thedatalab.org>
Date: Wed, 25 Sep 2024 17:00:36 +0100
Subject: [PATCH] Add a formatter for log files that can render ansi colours

Convert the text of the log file to HTML using ansi2html, which
converts ANSI colour codes to css classes so we get the colour
formatting from the original log.

It produces a full HTML file, which isn't what we want to display.
We just extract the <pre></pre> tag which contains the log content,
and the <style> which contains the relevant inline css.
---
 airlock/renderers.py         | 25 ++++++++++++++++++++++++-
 requirements.prod.in         |  1 +
 requirements.prod.txt        |  4 ++++
 tests/unit/test_renderers.py | 26 ++++++++++++++++++++++++++
 4 files changed, 55 insertions(+), 1 deletion(-)

diff --git a/airlock/renderers.py b/airlock/renderers.py
index 8faa73cc..14f6041e 100644
--- a/airlock/renderers.py
+++ b/airlock/renderers.py
@@ -2,6 +2,7 @@
 
 import csv
 import mimetypes
+import re
 from dataclasses import dataclass
 from email.utils import formatdate
 from functools import cached_property
@@ -9,9 +10,11 @@
 from pathlib import Path
 from typing import IO, Any, ClassVar, Self, cast
 
+from ansi2html import Ansi2HTMLConverter
 from django.http import FileResponse, HttpResponseBase
 from django.template import Template, loader
 from django.template.response import SimpleTemplateResponse
+from django.utils.safestring import mark_safe
 
 from airlock.types import UrlPath
 from airlock.utils import is_valid_file_type
@@ -158,9 +161,29 @@ def context(self):
         }
 
 
+class LogRenderer(TextRenderer):
+    def context(self):
+        # Convert the text of the log file to HTML, converting ANSI colour codes to css classes
+        # so we get the colour formatting from the original log.
+        # We don't need the full HTML file that's produced, so just extract the <pre></pre>
+        # tag which contains the log content and the inline styles.
+        conv = Ansi2HTMLConverter()
+        text = conv.convert(self.stream.read())
+        style = re.match(r".*(?P<style><style.*</style>).*", text, flags=re.S).group(
+            "style"
+        )
+        pre_tag = re.match(r".*(?P<pre_tag><pre.*</pre>).*", text, flags=re.S).group(
+            "pre_tag"
+        )
+        return {
+            "text": mark_safe(f"{style}{pre_tag}"),
+            "class": Path(self.filename).suffix.lstrip("."),
+        }
+
+
 FILE_RENDERERS = {
     ".csv": CSVRenderer,
-    ".log": TextRenderer,
+    ".log": LogRenderer,
     ".txt": TextRenderer,
     ".json": TextRenderer,
     ".md": TextRenderer,
diff --git a/requirements.prod.in b/requirements.prod.in
index 80a47b1f..c092513a 100644
--- a/requirements.prod.in
+++ b/requirements.prod.in
@@ -1,3 +1,4 @@
+ansi2html
 Django
 django-vite
 slippers
diff --git a/requirements.prod.txt b/requirements.prod.txt
index 6935e226..9fede944 100644
--- a/requirements.prod.txt
+++ b/requirements.prod.txt
@@ -4,6 +4,10 @@
 #
 #    pip-compile --allow-unsafe --generate-hashes --strip-extras requirements.prod.in
 #
+ansi2html==1.9.2 \
+    --hash=sha256:3453bf87535d37b827b05245faaa756dbab4ec3d69925e352b6319c3c955c0a5 \
+    --hash=sha256:dccb75aa95fb018e5d299be2b45f802952377abfdce0504c17a6ee6ef0a420c5
+    # via -r requirements.prod.in
 asgiref==3.8.1 \
     --hash=sha256:3e1e3ecc849832fe52ccf2cb6686b7a55f82bb1d6aee72a58826471390335e47 \
     --hash=sha256:c343bd80a0bec947a9860adb4c432ffa7db769836c64238fc34bdc3fec84d590
diff --git a/tests/unit/test_renderers.py b/tests/unit/test_renderers.py
index 0d813e06..e4c7dcb4 100644
--- a/tests/unit/test_renderers.py
+++ b/tests/unit/test_renderers.py
@@ -12,10 +12,12 @@
     (".png", "image/png", False, None),
     (".csv", "text/html", False, "airlock/templates/file_browser/csv.html"),
     (".txt", "text/html", False, "airlock/templates/file_browser/text.html"),
+    (".log", "text/html", False, "airlock/templates/file_browser/text.html"),
     (".html", "text/html", True, "airlock/templates/file_browser/plaintext.html"),
     (".png", "text/html", True, "airlock/templates/file_browser/plaintext.html"),
     (".csv", "text/html", True, "airlock/templates/file_browser/plaintext.html"),
     (".txt", "text/html", True, "airlock/templates/file_browser/plaintext.html"),
+    (".log", "text/html", True, "airlock/templates/file_browser/plaintext.html"),
 ]
 
 
@@ -145,3 +147,27 @@ def test_plaintext_renderer_handles_invalid_utf8(tmp_path):
     response.render()
     assert response.status_code == 200
     assert "invalid � continuation byte" in response.rendered_content
+
+
+def test_log_renderer_handles_ansi_colors(tmp_path):
+    log_file = tmp_path / "test.log"
+    # in ansi codes:
+    # \x1B[32m = foregrouund green
+    # \x1b[1m bold
+    # \x1b[0m resets formatting
+    log_file.write_bytes(
+        b"No ansi here \x1b[32m\x1b[1mThis is green and bold.\x1b[0m This is not."
+    )
+    relpath = log_file.relative_to(tmp_path)
+    Renderer = renderers.get_renderer(relpath)
+    renderer = Renderer.from_file(log_file, relpath)
+    response = renderer.get_response()
+    response.render()
+    assert response.status_code == 200
+
+    assert "ansi1 { font-weight: bold; }" in response.rendered_content
+    assert "ansi32 { color: #00aa00; }" in response.rendered_content
+    assert (
+        '<span class="ansi1 ansi32">This is green and bold.</span>'
+        in response.rendered_content
+    )