Split pattern code into its own file

Makefile

@@ -31,7 +31,7 @@ LDLIBS=-lgc -lcord -lm -lunistring -lgmp -ldl
 BUILTIN_OBJS=builtins/siphash.o builtins/array.o builtins/bool.o builtins/channel.o builtins/nums.o builtins/functions.o builtins/integers.o \
 						 builtins/pointer.o builtins/memory.o builtins/text.o builtins/thread.o builtins/c_string.o builtins/table.o \
 						 builtins/types.o builtins/util.o builtins/files.o builtins/range.o builtins/shell.o builtins/path.o \
-						 builtins/optionals.o
+						 builtins/optionals.o builtins/pattern.o
 TESTS=$(patsubst %.tm,%.tm.testresult,$(wildcard test/*.tm))
 
 all: libtomo.so tomo

builtins/functions.c

@@ -18,6 +18,7 @@
 #include "functions.h"
 #include "integers.h"
 #include "optionals.h"
+#include "pattern.h"
 #include "pointer.h"
 #include "siphash.h"
 #include "string.h"

builtins/path.c

@@ -18,6 +18,7 @@
 #include "integers.h"
 #include "optionals.h"
 #include "path.h"
+#include "pattern.h"
 #include "text.h"
 #include "types.h"
 #include "util.h"

builtins/pattern.h

@@ -0,0 +1,33 @@
+#pragma once
+
+// The type representing text patterns for pattern matching.
+
+#include <stdbool.h>
+#include <printf.h>
+#include <stdint.h>
+
+#include "datatypes.h"
+#include "integers.h"
+#include "types.h"
+
+#define Pattern(text) ((Pattern_t)Text(text))
+#define Patterns(...) ((Pattern_t)Texts(__VA_ARGS__))
+
+Text_t Text$replace(Text_t str, Pattern_t pat, Text_t replacement, Pattern_t backref_pat, bool recursive);
+Pattern_t Pattern$escape_text(Text_t text);
+Text_t Text$replace_all(Text_t text, Table_t replacements, Pattern_t backref_pat, bool recursive);
+Array_t Text$split(Text_t text, Pattern_t pattern);
+Text_t Text$trim(Text_t text, Pattern_t pattern, bool trim_left, bool trim_right);
+Int_t Text$find(Text_t text, Pattern_t pattern, Int_t i, int64_t *match_length);
+Array_t Text$find_all(Text_t text, Pattern_t pattern);
+PUREFUNC bool Text$has(Text_t text, Pattern_t pattern);
+PUREFUNC bool Text$matches(Text_t text, Pattern_t pattern);
+Text_t Text$map(Text_t text, Pattern_t pattern, Closure_t fn);
+
+#define Pattern$hash Text$hash
+#define Pattern$compare Text$compare
+#define Pattern$equal Text$equal
+
+extern const TypeInfo Pattern$info;
+
+// vim: ts=4 sw=0 et cino=L2,l1,(0,W4,m1,\:0

builtins/shell.c

@@ -5,6 +5,7 @@
 #include "array.h"
 #include "functions.h"
 #include "integers.h"
+#include "pattern.h"
 #include "shell.h"
 #include "text.h"
 #include "types.h"

builtins/text.h

@@ -6,11 +6,16 @@
 #include <stdbool.h>
 #include <printf.h>
 #include <stdint.h>
+#include <unistr.h>
 
 #include "datatypes.h"
 #include "integers.h"
 #include "types.h"
 
+typedef struct {
+    int64_t subtext, sum_of_previous_subtexts;
+} TextIter_t;
+
 int printf_text(FILE *stream, const struct printf_info *info, const void *const args[]);
 int printf_text_size(const struct printf_info *info, size_t n, int argtypes[n], int sizes[n]);
 
@@ -34,16 +39,8 @@ Text_t Text$lower(Text_t text);
 Text_t Text$title(Text_t text);
 Text_t Text$as_text(const void *text, bool colorize, const TypeInfo *info);
 Text_t Text$quoted(Text_t str, bool colorize);
-Text_t Text$replace(Text_t str, Pattern_t pat, Text_t replacement, Pattern_t backref_pat, bool recursive);
-Text_t Text$replace_all(Text_t text, Table_t replacements, Pattern_t backref_pat, bool recursive);
-Array_t Text$split(Text_t text, Pattern_t pattern);
-Text_t Text$trim(Text_t text, Pattern_t pattern, bool trim_left, bool trim_right);
-Int_t Text$find(Text_t text, Pattern_t pattern, Int_t i, int64_t *match_length);
-Array_t Text$find_all(Text_t text, Pattern_t pattern);
-PUREFUNC bool Text$has(Text_t text, Pattern_t pattern);
 PUREFUNC bool Text$starts_with(Text_t text, Text_t prefix);
 PUREFUNC bool Text$ends_with(Text_t text, Text_t suffix);
-PUREFUNC bool Text$matches(Text_t text, Pattern_t pattern);
 char *Text$as_c_string(Text_t text);
 __attribute__((format(printf, 1, 2)))
 public Text_t Text$format(const char *fmt, ...);
@@ -56,19 +53,16 @@ Text_t Text$from_codepoint_names(Array_t codepoint_names);
 Text_t Text$from_bytes(Array_t bytes);
 Array_t Text$lines(Text_t text);
 Text_t Text$join(Text_t glue, Array_t pieces);
-Text_t Text$map(Text_t text, Pattern_t pattern, Closure_t fn);
 Text_t Text$repeat(Text_t text, Int_t count);
+int32_t Text$get_grapheme_fast(Text_t text, TextIter_t *state, int64_t index);
+ucs4_t Text$get_main_grapheme_fast(Text_t text, TextIter_t *state, int64_t index);
+
+static inline int32_t Text$get_grapheme(Text_t text, int64_t index)
+{
+    TextIter_t state = {0, 0};
+    return Text$get_grapheme_fast(text, &state, index);
+}
 
 extern const TypeInfo Text$info;
 
-#define Pattern(text) ((Pattern_t)Text(text))
-#define Patterns(...) ((Pattern_t)Texts(__VA_ARGS__))
-Pattern_t Pattern$escape_text(Text_t text);
-
-#define Pattern$hash Text$hash
-#define Pattern$compare Text$compare
-#define Pattern$equal Text$equal
-
-extern const TypeInfo Pattern$info;
-
 // vim: ts=4 sw=0 et cino=L2,l1,(0,W4,m1,\:0

builtins/tomo.h

@@ -22,6 +22,7 @@
 #include "nums.h"
 #include "optionals.h"
 #include "path.h"
+#include "pattern.h"
 #include "pointer.h"
 #include "range.h"
 #include "shell.h"

docs/text.md

@@ -255,9 +255,11 @@ finding the value because the two texts are equivalent under normalization.
 # Patterns
 
 As an alternative to full regular expressions, Tomo provides a limited string
-matching pattern syntax that is intended to solve 80% of use cases in 2% of the
-code size (PCRE's codebase is roughly 150k lines of code, and Tomo's entire
-Text codebase is around 1.8K lines of code).
+matching pattern syntax that is intended to solve 80% of use cases in under 1%
+of the code size (PCRE's codebase is roughly 150k lines of code, and Tomo's
+pattern matching code is a bit under 1k lines of code). Tomo's pattern matching
+syntax is highly readable and works well for matching literal text without
+getting [leaning toothpick syndrome](https://en.wikipedia.org/wiki/Leaning_toothpick_syndrome).
 
 For more advanced use cases, consider linking against a C library for regular
 expressions or pattern matching.