Moving on with the host program

host/host.py

@@ -1,32 +0,0 @@
-from pyusb import usb
-import colorsys
-import numpy as np
-
-CRATE_WIDTH = 5
-CRATE_HEIGHT = 4
-CRATES_X = 16
-CRATES_Y = 2
-DISPLAY_WIDTH = CRATES_X * CRATE_WIDTH
-DISPLAY_HEIGHT = CRATES_Y * CRATE_HEIGHT
-FRAME_SIZE = CRATE_WIDTH*CRATE_HEIGHT*3
-
-dev = usb.core.find(idVendor=0x1cbe, idProduct=0x0003)
-
-def sendframe(framedata):
-    """ Send a frame to the display
-
-    The argument contains a h * w array of 3-tuples of (r, g, b)-data
-    """
-    def chunks(l, n):
-        for i in xrange(0, len(l), n):
-            yield l[i:i+n]
-
-    for cx, cy in itertools.product(range(16), range(2)):
-        data = [ v for x in range(CRATE_WIDTH) for y in range(CRATE_HEIGHT) for v in framedata[cy*CRATE_HEIGHT + y][cx*CRATE_WIDTH + x] ]
-        if len(data) != FRAME_SIZE:
-            raise ValueError('Invalid frame data. Expected {} bytes, got {}.'.format(FRAME_SIZE, len(data)))
-        # Send framebuffer data
-        dev.write(0x01, bytes([0, x, y])+bytes(data))
-    # Send latch command
-    dev.write(0x01, b'\x01')
-

host/matelight/config.py

@@ -0,0 +1,27 @@
+
+# Hard timeout in seconds after which (approximately) the rendering of a single item will be cut off
+RENDERER_TIMEOUT = 20.0
+# How long to show an image by default
+DEFAULT_IMAGE_DURATION = 10.0
+# Default scrolling speed in pixels/second
+DEFAULT_SCROLL_SPEED = 4
+# Pixels to leave blank between two letters
+LETTER_SPACING = 1
+
+# Display geometry
+# ┌─────────┐ ┌──┬──┬──┬ ⋯ ┬──┬──┬──┐
+# │1 o o o 5│ │ 1│  │  │   │  │  │16│
+# │6 o o o o│ ├──┼──┼──┼ ⋯ ┼──┼──┼──┤
+# │o o o o o│ │17│  │  │   │  │  │32│
+# │o o o o20│ └──┴──┴──┴ ⋯ ┴──┴──┴──┘
+# └─────────┘
+CRATE_WIDTH = 5
+CRATE_HEIGHT = 4
+CRATES_X = 16
+CRATES_Y = 2
+
+# Computed values
+DISPLAY_WIDTH = CRATES_X * CRATE_WIDTH
+DISPLAY_HEIGHT = CRATES_Y * CRATE_HEIGHT
+FRAME_SIZE = DISPLAY_WIDTH*DISPLAY_HEIGHT*3
+

host/matelight/host.py

@@ -0,0 +1,16 @@
+from pyusb import usb
+import colorsys
+import numpy as np
+
+dev = usb.core.find(idVendor=0x1cbe, idProduct=0x0003)
+
+def sendframe(framedata):
+    if not isinstance(framedata, np.array) or framedata.shape != (DISPLAY_WIDTH, DISPLAY_HEIGHT, 3) or framedata.dtype != np.int8:
+        raise ValueError('framedata must be a ({}, {}, 3)-numpy array of int8s'.format(DISPLAY_WIDTH, DISPLAY_HEIGHT))
+
+    for cx, cy in itertools.product(range(16), range(2)):
+        cratedata = framedata[cx*CRATE_WIDTH:(cx+1)*CRATE_WIDTH, cy*CRATE_HEIGHT:(cy+1)*CRATE_HEIGHT]
+        # Send framebuffer data
+        dev.write(0x01, bytes([0, x, y])+bytes(list(cratedata.flatten())))
+    # Send latch command
+    dev.write(0x01, b'\x01')

host/matelight/listeners.py

@@ -0,0 +1,19 @@
+from socketserver import *
+import zlib
+import struct
+
+class ThreadingTCPServer(ThreadingMixIn, TCPServer): pass
+class ThreadingUDPServer(ThreadingMixIn, UDPServer): pass
+
+class MateLightUDPHandler(BaseRequestHandler):
+    def handle(self):
+        data = self.request[0].strip()
+        if len(data) != FRAME_SIZE+4:
+            raise ValueError('Invalid frame size: Expected {}, got {}'.format(FRAME_SIZE+4, len(frame)))
+        frame = data[:-4]
+        crc1, = struct.unpack('!I', data[-4:])
+        crc2 = zlib.crc32(frame),
+        if crc1 != crc2:
+            raise ValueError('Invalid frame CRC checksum: Expected {}, got {}'.format(crc2, crc1))
+        socket.sendto(b'ACK', self.client_address)
+

host/matelight/queuemgr.py

@@ -0,0 +1,52 @@
+#!/usr/bin/env python3
+
+from renderers import TextRenderer, ImageRenderer
+import host, config
+import time, math
+
+score = lambda now, last, lifetime, priority, item: priority*math.log(now-last)/(10.0+item.duration)
+
+class FuzzyQueue:
+	def __init__(self, default):
+		self._default = default
+		self.put(default, 0.0, 0)
+		self._l = []
+
+	def put(self, item, priority=1.0, lifetime=0):
+		lifetime += time.time()
+		self._l.append((0, lifetime, priority, item))
+
+	def pop(self):
+		""" Get an item from the queue
+
+		NOTE: This is *not* a regular pop, as it does not necessarily remove the item from the queue.
+		"""
+		now = time.time()
+		# Choose item based on last used and priority
+		_, index, (_, lifetime, priority, item) = max(sorted([(score(now, *v), i, v) for i, v in self._l]))
+		# If item's lifetime is exceeded, remove
+		if lifetime < now and item is not self._default:
+			del self._l[index]
+		# Otherwise, set item's last played time
+		self._l[index] = (now, lifetime, prioity, item)
+		# Finally, return
+		return item
+
+q = FuzzyQueue()
+
+def insert_text(text, priority=1.0, lifetime=0, escapes=True):
+	q.put(TextRenderer(text, escapes), priority, lifetime)
+
+def insert_image(image, priority=1.0, lifetime=0):
+	q.put(ImageRenderer(image), priority, lifetime)
+
+def render_thread():
+	while True:
+		start = time.time()
+		for frame, delay in q.pop().frames(start):
+			then = time.time()
+			if then-start+delay > RENDERER_TIMEOUT:
+				break
+			sendframe(frame)
+			now = time.time()
+			time.sleep(min(RENDERER_TIMEOUT, delay - (now-then)))

host/matelight/renderers.py

@@ -0,0 +1,143 @@
+import numpy as np
+try:
+	import re2 as re
+except ImportError:
+	import re
+from PIL import Image
+from pixelterm import xtermcolors
+
+default_palette = [
+		(0x00, 0x00, 0x00), # 0 normal colors
+		(0xcd, 0x00, 0x00), # 1
+		(0x00, 0xcd, 0x00), # 2
+		(0xcd, 0xcd, 0x00), # 3
+		(0x00, 0x00, 0xee), # 4
+		(0xcd, 0x00, 0xcd), # 5
+		(0x00, 0xcd, 0xcd), # 6
+		(0xe5, 0xe5, 0xe5), # 7
+		(0x7f, 0x7f, 0x7f), # 8 bright colors
+		(0xff, 0x00, 0x00), # 9
+		(0x00, 0xff, 0x00), # 10
+		(0xff, 0xff, 0x00), # 11
+		(0x5c, 0x5c, 0xff), # 12
+		(0xff, 0x00, 0xff), # 13
+		(0x00, 0xff, 0xff), # 14
+		(0xff, 0xff, 0xff)] # 15
+
+class CharGenerator:
+	def __init__(self, seq=None, lg=None, text=''):
+		settings = False, False, False, default_palette[8], default_palette[0]
+		if lg:
+			settings = lg,bold, lg.blink, lg.underscore, lg.fg, lg.bg
+		self.bold, self.blink, self.underscore, self.fg, self.bg = settings
+		self.text = text
+		self.parse_escape_sequence(seq)
+
+	def parse_escape_sequence(seq):
+		codes = list(map(int, seq[2:-1].split(';')))
+		fg, bg, reverse, i = self.fg, self.bg, False, 0
+		while i<len(codes):
+			a = codes[i]
+			if a in [38, 48]:
+				if codes[i+1] == 5:
+					c = xtermcolors.xterm_colors[codes[i+2]]
+					fg, bg = (c, bg) if a == 38 else (fg, c)
+					i += 2
+			elif a == 39:
+				fg = (0,0,0)
+			elif a == 49:
+				bg = (0,0,0)
+			elif a == 0:
+				fg, bg = (0,0,0), (0,0,0)
+				self.bold, self.blink, self.underscore = False, False, False
+			elif a in range(30, 38):
+				fg = default_palette[a-30]
+			elif a in range(90, 98):
+				fg = default_palette[a-90+8]
+			elif a in range(40, 48):
+				bg = default_palette[a-40]
+			elif a in range(100, 108):
+				bg = default_palette[a-100+8]
+			elif a == 7:
+				reverse = True
+			elif a == 5:
+				self.blink = True
+			elif a == 4:
+				self.underscore = True
+			elif a == 1: # Literally "bright", not bold.
+				self.bold = True
+			i += 1
+		fg, bg = bg, fg if reverse else fg, bg
+		self.fg, self.bg = fg, bg
+
+	def generate_char(self, c, now):
+		fg, bg = self.bg, self.bg if self.blink and now%1.0 < 0.3 else self.fg, self.bg
+		...
+
+	def generate(self, now):
+		chars = [self.generate_char(c, now) for c in self.text]
+		# This refers to inter-letter spacing
+		space = np.zeros((LETTER_SPACING, DISPLAY_HEIGHT, 3))
+		spaces = [space]*(len(chars)-1)
+		everything = chars + spaces
+		everything[::2] = chars
+		everything[1::2] = spaces
+		return np.concatenate(everything)
+
+class TextRenderer:
+	def __init__(self, text, escapes=True):
+		"""Renders text into a frame buffer
+
+		"escapes" tells the renderer whether to interpret escape sequences (True) or not (False).
+		"""
+		generators = []
+		current_generator = CharGenerator()
+		for esc, char in r'(\x1B\[[0-9;]+m)|(.)'.finditer(text):
+			if esc:
+				if current_generator.text != '':
+					generators.append(current_generator)
+				current_generator = CharGenerator(esc, current_generator)
+			elif char:
+				current_generator.text += char
+		self.generators = generators + [current_generator]
+
+	def frames(self, start):
+		now = time.time()
+		zeros = [np.zeros((DISPLAY_WIDTH, DISPLAY_HEIGHT, 3))]
+		# Pad the array with one screen's worth of zeros on both sides so the text fully scrolls through.
+		raw = np.concatenate([zeros]+[g.generate(now) for g in generators]+[zeros])
+		w,h,_ = raw.size
+		for i in range(DISPLAY_WIDTH+w, 0, -1):
+			frame = raw[i:i+DISPLAY_WIDTH, :, :]
+			yield frame, 1/DEFAULT_SCROLL_SPEED
+			raw = np.concatenate([zeros]+[g.generate(now) for g in generators]+[zeros])
+
+class ImageRenderer:
+	def __new__(cls, image_data):
+		img = Image.open(io.BytesIO(image_data))
+		self.img = img
+
+	def frames(self):
+		img = self.img
+		palette = img.getpalette()
+		last_frame = Image.new("RGB", img.size)
+		# FIXME set delay to 1/10s if the image is animated, only use DEFAULT_IMAGE_DURATION for static images.
+		delay = img.info.get('duration', DEFAULT_IMAGE_DURATION*1000.0)/1000.0
+
+		for frame in ImageSequence.Iterator(img):
+			#This works around a known bug in Pillow
+			#See also: http://stackoverflow.com/questions/4904940/python-converting-gif-frames-to-png
+			frame.putpalette(palette)
+			c = frame.convert("RGB")
+
+			if img.info['background'] != img.info['transparency']:
+				last_frame.paste(c, c)
+			else:
+				last_frame = c
+
+			im = last_frame.copy()
+			im.thumbnail((DISPLAY_WIDTH, DISPLAY_HEIGHT), Image.NEAREST)
+			data = np.array(im.getdata(), dtype=np.int8)
+			data.reshape((DISPLAY_WIDTH, DISPLAY_HEIGHT, 3))
+			yield data, delay
+