commit ac3a118010b8fad81abffbb0f1bb077303c88eaa
Author: Natty <natty.sh.git@gmail.com>
Date:   Fri Dec 22 23:13:07 2023 +0100

    Initial proof of concept

diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000..381b8e3
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1,3 @@
+/.idea
+/cmake-build-*
+/build
\ No newline at end of file
diff --git a/CMakeLists.txt b/CMakeLists.txt
new file mode 100644
index 0000000..b24fe51
--- /dev/null
+++ b/CMakeLists.txt
@@ -0,0 +1,6 @@
+cmake_minimum_required(VERSION 3.26)
+project(rle C)
+
+set(CMAKE_C_STANDARD 23)
+
+add_executable(rle main.c)
diff --git a/main.c b/main.c
new file mode 100644
index 0000000..94b1695
--- /dev/null
+++ b/main.c
@@ -0,0 +1,110 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include <limits.h>
+
+#define GLYPHS_IN_LOOKUP 4
+#define GLYPH_SIZE (CHAR_BIT)
+
+unsigned char* encode(const unsigned char* data, const size_t len, size_t* out_len)
+{
+    // This can be further optimized for size, I was just lazy
+    unsigned char* rle = calloc(len + 1, sizeof(unsigned char));
+
+    if (len == 0) {
+        *out_len = 0;
+        return rle;
+    }
+
+    *out_len = 1;
+    unsigned char bit = 0;
+    for (size_t i = 0; i < len; ++i) {
+        unsigned char val = data[i >> 3] & ((unsigned char) (1 << (7 - i & 7)));
+        // A bit flip has occurred in the stream
+        if (!val ^ !bit) {
+            bit = !bit;
+            (*out_len)++;
+        }
+
+        (*(rle + *out_len - 1))++;
+    }
+
+    return rle;
+}
+
+struct lookup_val {
+    unsigned char position : 5;
+    unsigned char offset : 3;
+};
+
+// Just assume lookup is the correct length
+void fill_lookup(struct lookup_val* lookup, const unsigned char* rle, const size_t rle_len) {
+    size_t current = 0;
+    size_t rest = GLYPH_SIZE;
+    for (size_t i = 0; i < rle_len; ++i) {
+        while (rest >= GLYPH_SIZE) {
+            rest -= GLYPH_SIZE;
+            lookup[current].position = i;
+            lookup[current].offset = rest;
+            current++;
+        }
+
+        rest += rle[i];
+    }
+}
+
+void lookup_glyph(const unsigned char* rle,
+                  const struct lookup_val* lookup,
+                  const char glyph,
+                  unsigned char* glyph_data_out,
+                  const size_t glyph_data_len) {
+    const unsigned char* rle_current = &rle[lookup[glyph].position];
+    unsigned char current = *rle_current - lookup[glyph].offset;
+    unsigned char bit = 0;
+    for (size_t i = 0; i < glyph_data_len; ++i) {
+        if (!current) {
+            rle_current++;
+            current = *rle_current;
+            bit = !bit;
+        }
+
+        glyph_data_out[i >> 3] |= bit << (unsigned char) (7 - i & 7);
+
+        current--;
+    }
+}
+
+int main()
+{
+    // The encoding can be done externally, just like building the lookup table
+    unsigned char data[] = { 0b01000101, 0b1110001, 0b00010010, 0b000100010 };
+    size_t rle_len;
+    unsigned char* rle = encode(data, sizeof(data) * CHAR_BIT, &rle_len);
+
+    printf("RLE-encoded data: \n");
+    for (int i = 0; i < rle_len; ++i)
+        printf("%dx%s, ", rle[i], i % 2 ? "1" : "0");
+
+    printf("\n\n");
+
+    // Index into the lookup table is the ASCII char value we want, here for simplicity only 4 chars
+    struct lookup_val lookup[GLYPHS_IN_LOOKUP];
+    fill_lookup(lookup, rle, rle_len);
+
+    printf("Lookup table: \n");
+    for (int i = 0; i < GLYPHS_IN_LOOKUP; ++i)
+        printf("%d, %d \n", lookup[i].position, lookup[i].offset);
+
+    putchar('\n');
+
+    unsigned char glyph;
+    lookup_glyph(rle, lookup, 2, &glyph, GLYPH_SIZE);
+
+    printf("Lookup result: ");
+
+    for (unsigned char i = GLYPH_SIZE; i > 0; --i) {
+        bool val = glyph & (1 << (i - 1));
+        putchar('1' * val + '0' * !val);
+    }
+
+    putchar('\n');
+}