From 518ea8524128db7bebe42ecbe2347a85e2c362aa Mon Sep 17 00:00:00 2001
From: Tom de Geus <tdegeus@users.noreply.github.com>
Date: Fri, 2 Feb 2024 11:51:43 +0100
Subject: [PATCH] Better command-argument detection (#113)

---
 tests/test_indent_texindent.py | 102 +++++++++++++++++++++++++++++++++
 texplain/__init__.py           |  51 ++++++++++++++++-
 2 files changed, 150 insertions(+), 3 deletions(-)

diff --git a/tests/test_indent_texindent.py b/tests/test_indent_texindent.py
index d399f97..9c02ef3 100644
--- a/tests/test_indent_texindent.py
+++ b/tests/test_indent_texindent.py
@@ -1,3 +1,5 @@
+import pytest
+
 import texplain
 
 
@@ -59,3 +61,103 @@ def test_sentence_squash():
 
     ret = texplain.indent(text)
     assert ret.strip() == formatted.strip()
+
+
+def test_arguments():
+    text = r"""
+\hypersetup{
+    pdftitle=\@title,
+    citecolor=NavyBlue,
+    filecolor=NavyBlue,
+    linkcolor=NavyBlue,
+    urlcolor=NavyBlue,
+breaklinks,bookmarksopen=true,
+}
+"""
+
+    formatted = r"""
+\hypersetup{
+    pdftitle=\@title,
+    citecolor=NavyBlue,
+    filecolor=NavyBlue,
+    linkcolor=NavyBlue,
+    urlcolor=NavyBlue,
+    breaklinks,
+    bookmarksopen=true,
+}
+"""
+
+    ret = texplain.indent(text)
+    assert ret.strip() == formatted.strip()
+
+
+@pytest.mark.skip(reason="TODO: decide how to handle this case.")
+def test_arguments_comment():
+    text = r"""
+\hypersetup{%
+    pdftitle=\@title,
+    citecolor=NavyBlue,
+    filecolor=NavyBlue,
+    linkcolor=NavyBlue,
+    urlcolor=NavyBlue,
+breaklinks,bookmarksopen=true,
+}
+"""
+
+    formatted = r"""
+\hypersetup{%
+    pdftitle=\@title,
+    citecolor=NavyBlue,
+    filecolor=NavyBlue,
+    linkcolor=NavyBlue,
+    urlcolor=NavyBlue,
+    breaklinks,
+    bookmarksopen=true,
+}
+"""
+    ret = texplain.indent(text)
+    assert ret.strip() == formatted.strip()
+
+
+def test_arguments_nargs_below():
+    text = r"""
+\hypersetup{
+    pdftitle=\@title,citecolor=NavyBlue}
+"""
+
+    formatted = r"""
+\hypersetup{
+    pdftitle=\@title,
+    citecolor=NavyBlue
+}
+"""
+    ret = texplain.indent(text)
+    assert ret.strip() == formatted.strip()
+
+
+def test_arguments_nargs_above():
+    text = r"""
+\hypersetup{
+    pdftitle=\@title,citecolor=NavyBlue,bookmarksopen=true}
+"""
+
+    formatted = r"""
+\hypersetup{
+    pdftitle=\@title,
+    citecolor=NavyBlue,
+    bookmarksopen=true
+}
+"""
+    ret = texplain.indent(text)
+    assert ret.strip() == formatted.strip()
+
+
+def test_arguments_false_detection():
+    text = r"""
+\footnote{
+    This is a text: A = B, C = D, E = F.
+}.
+"""
+    formatted = text
+    ret = texplain.indent(text)
+    assert ret.strip() == formatted.strip()
diff --git a/texplain/__init__.py b/texplain/__init__.py
index 28e9d15..fe68367 100644
--- a/texplain/__init__.py
+++ b/texplain/__init__.py
@@ -1698,6 +1698,48 @@ def indent(
     return _rstrip_lines(text)
 
 
+def _argument_block_one_per_line(text: str) -> str:
+    r"""
+    Detect is text is a(n) (list of) arguments.
+    The text should not contain the argument itself:
+    ``\foo{a, b}`` is not allowed, but should be ``a, b``.
+
+    For example::
+
+        a=b, c=d, e=f
+
+    is formatted to::
+
+        a=b,
+        c=d,
+        e=f
+
+    Argument structure:
+
+        *   ``\w*\s*\=.*``: word followed by ``=`` and anything.
+        *   ``\w*``: word.
+
+    :param text: Text.
+    :return: Formatted text.
+    """
+    arguments = re.split(r"(?<!\\)(\,)", text)
+    arguments = [i.strip() for i in arguments]
+    arguments = list(filter(None, arguments))
+    closing_comma = arguments[-1] == ","
+    arguments = list(filter(lambda i: i != ",", arguments))
+
+    for arg in arguments:
+        if not (re.match(r"(^\w*\s*\=.*$)+", arg) or re.match(r"(^\w*$)", arg)):
+            return text
+
+    ret = ",\n".join(arguments)
+
+    if closing_comma:
+        ret += ","
+
+    return ret
+
+
 def _detail_one_sentence_per_line(text: str) -> str:
     """
     Split text into sentences.
@@ -1730,6 +1772,11 @@ def _one_sentence_per_line(
     :param text: Text.
     :param fold: List of placeholder types to fold before formatting (and restore after).
     :param base: Base name for placeholders.
+
+    :param command:
+        Check if the text are arguments of a command, and format one argument per line.
+        TODO: rename argument: ``command`` is not intuitive.
+
     :return: Formatted text.
     """
 
@@ -1772,9 +1819,7 @@ def _one_sentence_per_line(
         ret += _detail_one_sentence_per_line(text[start:])
 
     if command:
-        match = [i for i in re.finditer(r"(\w*\=.*\,\ )+", ret)]
-        if len(match) > 0:
-            ret = ",\n".join(ret.split(", "))
+        ret = _argument_block_one_per_line(ret)
 
     return text_from_placeholders(ret, placeholders)