diff options
| author | Patrick Kingston <patrick@pkingston.xyz> | 2026-01-28 19:23:13 -0500 |
|---|---|---|
| committer | Patrick Kingston <patrick@pkingston.xyz> | 2026-01-28 19:23:13 -0500 |
| commit | fc8569a88200ec88c3773c338fdcf16b16ea51d5 (patch) | |
| tree | 32b7616811d5248ef53dc8d98bd996652f25b967 /bible/huffman.clj | |
| parent | 0e17cdb8d895589dbf1470f5b06a49b1b94e2e4d (diff) | |
Get compressor working
Diffstat (limited to 'bible/huffman.clj')
| -rw-r--r-- | bible/huffman.clj | 10 |
1 files changed, 2 insertions, 8 deletions
diff --git a/bible/huffman.clj b/bible/huffman.clj index c219b7e..fccdae1 100644 --- a/bible/huffman.clj +++ b/bible/huffman.clj @@ -149,14 +149,8 @@ (if (= basecount len) base (str (apply str (take (- len basecount) (repeat "0"))) base)))) - ;; --- little class for Huffman Codewords -(def sorted-huffman-tree-codewords - (->> huffman-tree-syms - (sort-by (juxt (comp count val) key)) - (map #(->HuffmanCodeword (first %1) (Long/parseUnsignedLong (second %1) 2) (int (count (second %1))))))) - (defn build-canonical-encodings "Build canonical huffman encodings from a huffman tree takes [symbols] a list of huffman codes derived from a code-tree" @@ -197,7 +191,7 @@ (build-canonical-encodings huffman-tree)) -(assert +#_(assert (= (map #(.length %1) sorted-huffman-tree-codewords) (map #(.length %1) sorted-huffman-tree-codewords)) "Some of the codes changed length when canonicalizing") @@ -228,7 +222,7 @@ {:bits totalbits :bytes totalbytes :kb totalkb - :mb totalmb})) + :mb totalmb})) (def stage1-optimized-token-encoding-length (get-encoded-length canonical-encodings tokens))) |