From 2e7fefc7f7a50e8b6edd23667eeb884cc627872b Mon Sep 17 00:00:00 2001
From: Laria Carolin Chabowski <laria@laria.me>
Date: Tue, 7 Mar 2023 21:51:40 +0100
Subject: [PATCH] Implement some string manipulation functions

---
 src/CMakeLists.txt                   |  1 +
 src/builtins.c                       | 31 ++++++++++++++
 src/functional-tests/string-manip.at | 42 ++++++++++++++++++
 src/globals.apfl                     | 64 ++++++++++++++++++++++++++++
 src/strings.c                        |  5 +++
 5 files changed, 143 insertions(+)
 create mode 100644 src/functional-tests/string-manip.at

diff --git a/src/CMakeLists.txt b/src/CMakeLists.txt
index 130b2c8..aaa4b3d 100644
--- a/src/CMakeLists.txt
+++ b/src/CMakeLists.txt
@@ -112,6 +112,7 @@ functionaltest("compare")
 functionaltest("concat")
 functionaltest("join")
 functionaltest("quine")
+functionaltest("string-manip")
 
 install(TARGETS apfl DESTINATION lib)
 install(TARGETS apfl-bin DESTINATION bin)
diff --git a/src/builtins.c b/src/builtins.c
index 77787cd..58355de 100644
--- a/src/builtins.c
+++ b/src/builtins.c
@@ -576,6 +576,35 @@ set_func_name(apfl_ctx ctx)
     apfl_set_func_name(ctx, -2, -1);
 }
 
+static void
+substring(apfl_ctx ctx)
+{
+    apfl_get_list_member_by_index(ctx, 0, 0);
+    struct apfl_string_view sv = apfl_get_string(ctx, -1);
+    apfl_get_list_member_by_index(ctx, 0, 1);
+    apfl_number start = apfl_get_number(ctx, -1);
+    apfl_get_list_member_by_index(ctx, 0, 2);
+    apfl_number len = apfl_get_number(ctx, -1);
+
+    apfl_push_string_view_copy(ctx, apfl_string_view_substr(sv, (size_t)start, (size_t)len));
+}
+
+static void
+stringsearch(apfl_ctx ctx)
+{
+    apfl_get_list_member_by_index(ctx, 0, 0);
+    struct apfl_string_view haystack = apfl_get_string(ctx, -1);
+    apfl_get_list_member_by_index(ctx, 0, 1);
+    struct apfl_string_view needle = apfl_get_string(ctx, -1);
+
+    ptrdiff_t off = apfl_string_view_search(haystack, needle);
+    if (off < 0) {
+        apfl_push_nil(ctx);
+    } else {
+        apfl_push_number(ctx, (apfl_number)off);
+    }
+}
+
 static bool
 iterate_dict_callback(apfl_ctx ctx, void *opaque)
 {
@@ -646,5 +675,7 @@ apfl_builtins(apfl_ctx ctx)
     add_builtin(ctx, "-serialize-bytecode", serialize_bytecode);
     add_builtin(ctx, "-unserialize-bytecode", unserialize_bytecode);
     add_builtin(ctx, "set-func-name", set_func_name);
+    add_builtin(ctx, "substring", substring);
+    add_builtin(ctx, "stringsearch", stringsearch);
     add_builtin(ctx, "iterate-dict", iterate_dict);
 }
diff --git a/src/functional-tests/string-manip.at b/src/functional-tests/string-manip.at
new file mode 100644
index 0000000..6e6c31a
--- /dev/null
+++ b/src/functional-tests/string-manip.at
@@ -0,0 +1,42 @@
+===== script =====
+print (substr 0 3 "foobar")
+print (substr 3 "foobar")
+print (substr 1 3 "foobar")
+print (substr -1 "foobar")
+print (substr 6 "foobar")
+print (substr 10 "foobar")
+
+print (strsearch "a" "aaababcabcde")
+print (strsearch "ab" "aaababcabcde")
+print (strsearch "x" "aaababcabcde")
+print (strsearch "ab" 3 "aaababcabcde")
+
+print (split "::" "foo:bar:::baz::::xxx::::y::z::")
+print (split "::" 3 "foo:bar:::baz::::xxx::::y::z::")
+
+===== output =====
+foo
+bar
+oob
+r
+
+
+0
+2
+nil
+4
+[
+  "foo:bar"
+  ":baz"
+  ""
+  "xxx"
+  ""
+  "y"
+  "z"
+  ""
+]
+[
+  "foo:bar"
+  ":baz"
+  "::xxx::::y::z::"
+]
diff --git a/src/globals.apfl b/src/globals.apfl
index 94b28a4..9bc6ede 100644
--- a/src/globals.apfl
+++ b/src/globals.apfl
@@ -99,6 +99,65 @@
         has pred == y
     }
 
+    identity := { x -> x }
+
+    is := -named 'is (partial has identity)
+
+    -raw-substring := builtins.substring
+    substr := {
+    start s ->
+        substr start (len s) s
+    start?(is < 0) end s ->
+        substr (+ start (len s)) end s
+    start end?(is < 0) s ->
+        substr start (+ end (- (len s) start)) s
+    start end s ->
+        -raw-substring s start end
+    }
+
+    -raw-stringsearch := builtins.stringsearch
+    strsearch := {
+    needle haystack ->
+        -raw-stringsearch haystack needle
+    needle start?(is < 0) haystack ->
+        strsearch needle (+ start (len s)) haystack
+    needle start haystack ->
+        off := (strsearch needle (substr start haystack))
+        if (== off nil) { nil } { + start off }
+    }
+
+    split := ({
+        split-aux := { maxlen-reached sep s ->
+            sep-len := len sep
+            parts := []
+            loop {
+                if (maxlen-reached (+ 1 (len parts))) {
+                    parts = [~parts s]
+                    false
+                } {
+                    off := strsearch sep s
+                    if (== nil off) {
+                        parts = [~parts s]
+                        false
+                    } {
+                        parts = [~parts (substr 0 off s)]
+                        s = substr (+ off sep-len) s
+                        true
+                    }
+                }
+            }
+
+            parts
+        }
+
+        split := {
+        sep s ->
+            split-aux {false} sep s
+        sep maxlen s ->
+            split-aux (is >= maxlen) sep s
+        }
+    })
+
     keach := {
     d?(has type 'dict) body ->
         out := nil
@@ -155,9 +214,14 @@
         '-serialize-bytecode -> -serialize-bytecode
         '-unserialize-bytecode -> -unserialize-bytecode
         '& -> &
+        'substr -> substr
+        'strsearch -> strsearch
+        'split -> split
         'partial -> partial
         'compose -> compose
         'has -> has
+        'identity -> identity
+        'is -> is
         '!= -> !=
         '!> -> !>
         '!< -> !<
diff --git a/src/strings.c b/src/strings.c
index 0abd0f9..b2eba29 100644
--- a/src/strings.c
+++ b/src/strings.c
@@ -80,6 +80,11 @@ bool
 apfl_string_copy(struct apfl_allocator allocator, struct apfl_string *dst, struct apfl_string_view src)
 {
     apfl_string_deinit(allocator, dst);
+
+    if (src.len == 0) {
+        return true;
+    }
+
     if ((dst->bytes = ALLOC_BYTES(allocator, src.len)) == NULL) {
         return false;
     }