From 81ae77fe9dee2d91d30eee2e375b5cda220c8272 Mon Sep 17 00:00:00 2001
From: "Charles K. Neimog" <charlesneimog@outlook.com>
Date: Sun, 30 Jul 2023 16:14:54 -0300
Subject: [PATCH] Export annotation color as hex RGB in JSON (#78)

* add color when using json
* minor fixes

---------

Co-authored-by: Andrew Baumann <ab@ab.id.au>
---
 pdfannots/__init__.py     | 14 ++++++++++++--
 pdfannots/printer/json.py |  3 +++
 pdfannots/types.py        | 17 ++++++++++++++++-
 3 files changed, 31 insertions(+), 3 deletions(-)

diff --git a/pdfannots/__init__.py b/pdfannots/__init__.py
index da4d6f5..007cf5d 100644
--- a/pdfannots/__init__.py
+++ b/pdfannots/__init__.py
@@ -22,7 +22,7 @@
 import pdfminer.settings
 import pdfminer.utils
 
-from .types import Page, Outline, AnnotationType, Annotation, Document
+from .types import Page, Outline, AnnotationType, Annotation, Document, RGB
 from .utils import cleanup_text, decode_datetime
 
 pdfminer.settings.STRICT = False
@@ -71,6 +71,16 @@ def _mkannotation(
         # decode as string, normalise line endings, replace special characters
         contents = cleanup_text(pdfminer.utils.decode_text(contents))
 
+    rgb: typ.Optional[RGB] = None
+    color = pa.get('C')
+    if color is not None:
+        if (isinstance(color, list)
+                and len(color) == 3
+                and all(isinstance(e, float) and 0 <= e <= 1 for e in color)):
+            rgb = RGB(*color)
+        else:
+            logger.warning("Invalid color %s in annotation on %s", color, page)
+
     # Rect defines the location of the annotation on the page
     rect = pdftypes.resolve1(pa.get('Rect'))
 
@@ -94,7 +104,7 @@ def _mkannotation(
         created = decode_datetime(createds)
 
     return Annotation(page, annot_type, quadpoints, rect,
-                      contents, author=author, created=created)
+                      contents, author=author, created=created, color=rgb)
 
 
 def _get_outlines(doc: PDFDocument) -> typ.Iterator[Outline]:
diff --git a/pdfannots/printer/json.py b/pdfannots/printer/json.py
index 13e128c..fc0b2c5 100644
--- a/pdfannots/printer/json.py
+++ b/pdfannots/printer/json.py
@@ -35,6 +35,9 @@ def annot_to_dict(
     if annot.created:
         result['created'] = annot.created.strftime('%Y-%m-%dT%H:%M:%S')
 
+    if annot.color:
+        result['color'] = annot.color.ashex()
+
     return result
 
 
diff --git a/pdfannots/types.py b/pdfannots/types.py
index 2f2dc82..b957a97 100644
--- a/pdfannots/types.py
+++ b/pdfannots/types.py
@@ -276,6 +276,7 @@ class Annotation(ObjectWithPos):
         text         Text in the order captured (use gettext() for a cleaner form)
         author       Author of the annotation
         created      Timestamp the annotation was created
+        color        RGB color of the annotation
         last_charseq Sequence number of the most recent character in text
 
     Attributes updated only for StrikeOut annotations:
@@ -297,7 +298,8 @@ def __init__(
             rect: typ.Optional[BoxCoords] = None,
             contents: typ.Optional[str] = None,
             author: typ.Optional[str] = None,
-            created: typ.Optional[datetime.datetime] = None):
+            created: typ.Optional[datetime.datetime] = None,
+            color: typ.Optional[RGB] = None):
 
         # Construct boxes from quadpoints
         boxes = []
@@ -324,6 +326,7 @@ def __init__(
         self.author = author
         self.created = created
         self.text = []
+        self.color = color
         self.pre_context = None
         self.post_context = None
         self.boxes = boxes
@@ -479,3 +482,15 @@ def nearest_outline(
                 return page.outlines[idx - 1]
 
         return None
+
+
+class RGB(typ.NamedTuple):
+    red: float
+    green: float
+    blue: float
+
+    def ashex(self) -> str:
+        red_hex = format(int(self.red * 255), '02x')
+        green_hex = format(int(self.green * 255), '02x')
+        blue_hex = format(int(self.blue * 255), '02x')
+        return f"#{str(red_hex)}{str(green_hex)}{(blue_hex)}"