From 26fa657e4a92a260aa03ca73ceda4fdc6aa30f13 Mon Sep 17 00:00:00 2001
From: Tim Grelka <grelka@indibit.eu>
Date: Sun, 8 Dec 2024 11:33:21 +0100
Subject: [PATCH] feat: render units with non-breaking space

---
 renderer/fifty_ohm_html_renderer.py | 49 ++++++++++++++++++++++++---
 renderer/unit.py                    | 51 +++++++++++++++++++++++++++++
 test/test_unit_token.py             | 18 ++++++++++
 3 files changed, 114 insertions(+), 4 deletions(-)
 create mode 100644 renderer/unit.py
 create mode 100644 test/test_unit_token.py

diff --git a/renderer/fifty_ohm_html_renderer.py b/renderer/fifty_ohm_html_renderer.py
index fa6a6ab..ed2f077 100644
--- a/renderer/fifty_ohm_html_renderer.py
+++ b/renderer/fifty_ohm_html_renderer.py
@@ -1,12 +1,45 @@
 from mistletoe import HtmlRenderer
 
-from renderer.comment import Comment
-from renderer.quote import Quote
+from .comment import Comment
+from .unit import Unit
+from .quote import Quote
+
+
+units = {
+    "A": "A",
+    "Ah": "Ah",
+    "A/mm²": "A/mm²",
+    "baud": "baud",
+    "Bit": "Bit",
+    "Bit/s": "Bit/s",
+    "dB": "dB",
+    "dBi": "dBi",
+    "dBm": "dBm",
+    "dBW": "dBW",
+    "F": "F",
+    "J": "J",
+    "Hz": "Hz",
+    "H": "H",
+    "cm": "cm",
+    "m": "m",
+    "m²": "m²",
+    "Ohm": "Ω",
+    "ppm": "ppm",
+    "pps": "pps",
+    "s": "s",
+    "V": "V",
+    "W": "W",
+    "Wh": "Wh",
+    "°": "°",
+    "%": "%",
+}
+
+no_space_units = ["°", "%"]
 
 
 class FiftyOhmHtmlRenderer(HtmlRenderer):
     def __init__(self):
-        super().__init__(Comment, Quote)
+        super().__init__(Comment, Unit, Quote)
 
     def render_comment():
         return None
@@ -15,4 +48,12 @@ class FiftyOhmHtmlRenderer(HtmlRenderer):
         return f"„{self.render_inner(token)}“"
 
     def render_underline(self, token):
-        return f"<u>{self.render_inner(token)}</u>"
\ No newline at end of file
+        return f"<u>{self.render_inner(token)}</u>"
+
+    @staticmethod
+    def render_unit(token):
+        unit = token.prefix + units[token.unit]
+        if token.unit in no_space_units:
+            return f"{token.value}{unit}"
+        else:
+            return f"{token.value}&#8239;{unit}"
diff --git a/renderer/unit.py b/renderer/unit.py
new file mode 100644
index 0000000..7364ff0
--- /dev/null
+++ b/renderer/unit.py
@@ -0,0 +1,51 @@
+import re
+
+from mistletoe.span_token import SpanToken
+
+
+class Unit(SpanToken):
+    units = [
+        "A",
+        "Ah",
+        "A/mm²",
+        "baud",
+        "Bit",
+        "Bit/s",
+        "dB",
+        "dBi",
+        "dBm",
+        "dBW",
+        "F",
+        "J",
+        "Hz",
+        "H",
+        "cm",
+        "m",
+        "m²",
+        "Ohm",
+        "ppm",
+        "pps",
+        "s",
+        "V",
+        "W",
+        "Wh",
+        "°",
+        "%",
+    ]
+    prefixes = ["p", "n", "μ", "m", "k", "M", "G"]
+
+    pattern = re.compile(
+        r"([0-9,]+) ?("
+        # Group all prefixes together in a capture group
+        # Empty string at end of list is required so the group always exists in the resulting match
+        + "|".join(prefixes + [""])
+        + ")("
+        # Group all units together in a capture group.
+        + "|".join(units)
+        + r")(?!\w)"
+    )
+
+    def __init__(self, match):
+        self.value = match.group(1)
+        self.prefix = match.group(2)
+        self.unit = match.group(3)
diff --git a/test/test_unit_token.py b/test/test_unit_token.py
new file mode 100644
index 0000000..47c929e
--- /dev/null
+++ b/test/test_unit_token.py
@@ -0,0 +1,18 @@
+import mistletoe
+
+from renderer.fifty_ohm_html_renderer import FiftyOhmHtmlRenderer
+
+from .util import paragraph
+
+
+def test_unit_html():
+    assertions = {
+        "0,1 V": "0,1&#8239;V",
+        "33 A": "33&#8239;A",
+        "1 Ah": "1&#8239;Ah",
+        "100 mOhm": "100&#8239;mΩ",
+        "25 Ohm": "25&#8239;Ω",
+    }
+
+    for input, output in assertions.items():
+        assert mistletoe.markdown(input, FiftyOhmHtmlRenderer) == paragraph(output)