From d0118b56b1a647f066cbc58a43d038a72a0a0ccd Mon Sep 17 00:00:00 2001
From: Michael Sippel <micha@fragmental.art>
Date: Sat, 15 Mar 2025 18:49:06 +0100
Subject: [PATCH] add morphisms
- unicode
- value delimited seq
- zigzag encoding (still lacking any way to get to signed integer yet'..)
---
morphisms/unicode.morphism-base | 93 +++++++++++++++++++++++++++++
morphisms/value_delim.morphism-base | 71 ++++++++++++++++++++++
morphisms/zigzag.morphism-base | 26 ++++++++
3 files changed, 190 insertions(+)
create mode 100644 morphisms/unicode.morphism-base
create mode 100644 morphisms/value_delim.morphism-base
create mode 100644 morphisms/zigzag.morphism-base
diff --git a/morphisms/unicode.morphism-base b/morphisms/unicode.morphism-base
new file mode 100644
index 0000000..8805ace
--- /dev/null
+++ b/morphisms/unicode.morphism-base
@@ -0,0 +1,93 @@
+```
+```
+
+morph_string_as_ascii_to_utf8 ()
+ <Seq ~ <ValueTerminated '\0'> Char~Ascii~x86.UInt8>
+--> <Seq Char~Unicode>
+ ~ UTF-8
+ ~ <Seq~<ValueTerminated 0> x86.UInt8>
+```
+ while( *src ) { *dst++ = *src++; }
+ *dst = 0;
+ return 0;
+```
+
+morph_string_as_utf8_to_ascii ()
+ <Seq Char~Unicode>
+ ~ UTF-8
+ ~ <Seq~<ValueTerminated 0> x86.UInt8>
+--> <Seq ~ <ValueTerminated '\0'> Char~Ascii~x86.UInt8>
+```
+ while( *src ) {
+ if( *src < 128 ) {
+ *dst++ = *src++;
+ } else {
+ fprintf(stderr, "(morph UTF-8 to Ascii) ignore multi-byte character\n");
+ while( *++src >= 128 );
+ }
+ }
+ *dst = 0;
+ return 0;
+```
+
+morph_string_as_ascii_to_utf32 ()
+ <Seq ~ <ValueTerminated '\0'> Char~Ascii~x86.UInt8>
+--> <Seq Char~Unicode>
+ ~ UTF-32LE
+ ~ <Seq~<ValueTerminated 0> x86.UInt32>
+```
+ while( *src ) { *dst++ = *src++; }
+ *dst = 0;
+ return 0;
+```
+
+morph_string_as_utf8_to_utf32 ()
+
+ <Seq Char~Unicode>
+ ~ UTF-8
+ ~ <Seq~<ValueTerminated 0> x86.UInt8>
+
+--> <Seq Char~Unicode>
+ ~ UTF-32LE
+ ~ <Seq~<ValueTerminated 0> x86.UInt32>
+
+```
+ bool has_multibyte = false;
+ uint32_t val = 0;
+ while( *src ) {
+ uint8_t tag = (*src >> 6)&0b11;
+ switch( tag ) {
+ // single byte
+ case 0b00:
+ case 0b01:
+ if( has_multibyte ) {
+ *dst++ = val;
+ has_multibyte = false;
+ }
+ *dst++ = *src++;
+ break;
+
+ // start multibyte
+ case 0b11:
+ if( has_multibyte ) {
+ *dst++ = val;
+ }
+ has_multibyte = true;
+ val = (*src++) & 0b111111;
+ break;
+
+ // continue multibyte
+ case 0b10:
+ val <<= 6;
+ val |= (*src++) & 0b111111;
+ break;
+ }
+ }
+
+ if( has_multibyte )
+ *dst++ = val;
+
+ *dst++ = 0;
+
+ return 0;
+```
diff --git a/morphisms/value_delim.morphism-base b/morphisms/value_delim.morphism-base
new file mode 100644
index 0000000..1870ff7
--- /dev/null
+++ b/morphisms/value_delim.morphism-base
@@ -0,0 +1,71 @@
+```
+#include <array/length-prefix.h>
+#include <stdlib.h>
+```
+
+morph_seqseq_valsep_uint8 (T: Type, SrcDelim: T, DstDelim: T)
+ < Seq <Seq T> >
+ ~ < ValueSep SrcDelim T >
+ ~ < Seq~<LengthPrefix x86.UInt64> T >
+
+--> < Seq <Seq T> >
+ ~ < ValueSep DstDelim T >
+ ~ < Seq~<LengthPrefix x86.UInt64> T >
+```
+ length_prefix_uint8_array_clear( dst );
+
+ uint8_t * dst_items = dst->items;
+ for( uint64_t i = 0; i < src->len; ++i ) {
+ if( src->items[i] == SrcDelim ) {
+ length_prefix_uint8_array_push( dst, DstDelim );
+ } else if( src->items[i] == DstDelim ) {
+ if( DstDelim == '\n' ) {
+ length_prefix_uint8_array_push( dst, '\\' );
+ length_prefix_uint8_array_push( dst, 'n' );
+ }
+ } else {
+ length_prefix_uint8_array_push( dst, src->items[i] );
+ }
+ }
+
+ return 0;
+```
+
+
+morph_seqseq_as_valsep_to_lenpfx (T: Type, Delim: T, EscKey: T)
+ < Seq <Seq T> >
+ ~ < ValueSep T Delim >
+ ~ < Seq~<LengthPrefix x86.UInt64> T >
+
+--> < Seq~<LengthPrefix x86.UInt64>
+ <Seq~<LengthPrefix x86.UInt64> T >
+ ~ <RefMut < Seq~<LengthPrefix x86.UInt64> T>>
+ ~ x86.Address
+ ~ x86.UInt64
+ >
+```
+ length_prefix_uint64_array_clear( dst );
+
+ struct LengthPrefixUInt8Array * cur_item = NULL;
+
+ uint8_t const * start = &src->items[0];
+ uint8_t const * cur = start;
+ uint8_t const * end = &src->items[src->len];
+
+ while( cur < end ) {
+ if( *cur == Delim || cur+1 == end ) {
+ uint64_t len = cur - start;
+
+ cur_item = malloc( sizeof(uint64_t) + sizeof(uint8_t) * len );
+ cur_item->len = len;
+ memcpy( cur_item->items, start, len );
+
+ length_prefix_uint64_array_push( dst, (uint64_t)cur_item );
+ start = ++cur;
+ } else {
+ cur++;
+ }
+ }
+
+ return 0;
+```
diff --git a/morphisms/zigzag.morphism-base b/morphisms/zigzag.morphism-base
new file mode 100644
index 0000000..0d54033
--- /dev/null
+++ b/morphisms/zigzag.morphism-base
@@ -0,0 +1,26 @@
+```
+```
+
+morph_i64_as_twos_complement_to_zigzag ()
+ ℤ ~ x86.Int64
+--> ℤ ~ ZigZagInt ~ ℕ ~ x86.UInt64
+```
+ if( *src >= 0 ) {
+ *dst = (2 * (uint64_t)*src)
+ } else {
+ *dst = (2 * (uint64_t)(- *src)) - 1;
+ }
+
+ return 0;
+```
+
+morph_i64_as_zigzag_to_twos_complement ()
+ ℤ ~ ZigZagInt ~ ℕ ~ x86.UInt64
+--> ℤ ~ x86.Int64
+```
+ if( *src % 2 == 0 ) {
+ *dst = *src / 2;
+ } else {
+ *dst = - ((*src+1) / 2);
+ }
+```