things!

Common/lzw.h

@@ -0,0 +1,55 @@
+#ifndef COM_LZW_H
+#define COM_LZW_H
+
+#include <stdbool.h>
+#include <stdint.h>
+
+/* Alphabet table for use in LZW algorithms */
+struct com_lzw_table {
+  /* table_size   indexed entry is reserved for clear code */
+  /* table_size+1 indexed entry is reserved for terminator */
+  struct com_lzw_table_entry *table;
+  uint32_t init_size; /* Alphabet size, table clear is done over it */
+  uint32_t size;
+  uint32_t cap;
+  bool continuous; /* If true, first order search is simple indexing, where
+                     codepoint = index, up to table_init_size */
+};
+
+struct com_lzw_table_entry {
+  uint16_t child_chain; /* Start of linked list inside table, or 0 if none */
+  uint16_t next_child;  /* Index of next child, or 0 if none */
+  char code;            /* Single byte, children bytes are appended to it */
+};
+
+/* Returns table that is optimized for continuous range of codepoints */
+/* Useful for binary compression, such as .GIF and .SAM formats */
+struct com_lzw_table com_lzw_continous_table(uint8_t codepoints);
+
+/* Returns table with alphabet inferred from the incoming data */
+/* Useful for alphanumeric compression, where not all codepoints are in use */
+/* TODO: Sort table by frequency? */
+struct com_lzw_table com_lzw_infer_table(const char *datain, uint32_t sizein);
+
+/* Each bit corresponds to a byte value, position-wise */
+struct com_lzw_encoded_alphabet {
+  uint64_t b0;
+  uint64_t b1;
+  uint64_t b2;
+  uint64_t b3;
+};
+struct com_lzw_encoded_alphabet
+com_lzw_encode_alphabet(const struct com_lzw_table *table);
+
+struct com_lzw_table
+com_lzw_decode_table(struct com_lzw_encoded_alphabet alphabet);
+
+/* Return table to its original form */
+void com_lzw_clear_table(struct com_lzw_table *table);
+
+void com_lzw_free_table(struct com_lzw_table *table);
+
+bool com_lzw_compress(const struct com_lzw_table *table, const char *datain,
+                      uint32_t sizein, char **dataout, uint32_t *sizeout);
+
+#endif