Add translation mode and statisticsHEADmain

8 files changed, 475 insertions, 0 deletions

diff --git a/src/impl/abs.c b/src/impl/abs.c
new file mode 100644
index 0000000..a6aa76d
--- /dev/null
+++ b/src/impl/abs.c
@@ -0,0 +1,81 @@
+// Copyright (c) 2024, Marvin Borner <dev@marvinborner.de>
+// SPDX-License-Identifier: MIT
+
+#include <log.h>
+#include <impl.h>
+
+static void encode(Term *term, FILE *fp)
+{
+	switch (term->type) {
+	case ABS:
+		while (1) {
+			fprintf(fp, "0");
+			if (term->u.abs.body->type != ABS)
+				break;
+			term = term->u.abs.body;
+		}
+		fprintf(fp, "01");
+		encode(term->u.abs.body, fp);
+		break;
+	case APP:
+		fprintf(fp, "01");
+		encode(term->u.app.lhs, fp);
+		encode(term->u.app.rhs, fp);
+		break;
+	case IDX:
+		for (size_t i = 0; i <= term->u.index; i++)
+			fprintf(fp, "1");
+		fprintf(fp, "0");
+		break;
+	default:
+		fatal("invalid type %d\n", term->type);
+	}
+}
+
+static Term *decode(FILE *fp)
+{
+	Term *res = 0;
+
+	char a = getc(fp);
+
+	if (a == '0') {
+		size_t count = 0;
+		while (getc(fp) == '0')
+			count++;
+
+		if (count == 0) {
+			res = term_new(APP);
+			res->u.app.lhs = decode(fp);
+			res->u.app.rhs = decode(fp);
+		} else {
+			Term *head = term_new(ABS);
+
+			res = head;
+			for (size_t i = 0; i < count - 1; i++) {
+				res->u.abs.body = term_new(ABS);
+				res = res->u.abs.body;
+			}
+			res->u.abs.body = decode(fp);
+			res = head;
+		}
+	} else if (a == '1') {
+		res = term_new(IDX);
+		res->u.index = 0;
+		while (getc(fp) == '1')
+			res->u.index++;
+	}
+
+	if (!res)
+		fatal("invalid parsing state!\n");
+
+	return res;
+}
+
+Impl impl_abs(void)
+{
+	return (Impl){
+		.name = "abs",
+		.encode = encode,
+		.decode = decode,
+	};
+}
diff --git a/src/impl/abs_app_right.c b/src/impl/abs_app_right.c
new file mode 100644
index 0000000..a346013
--- /dev/null
+++ b/src/impl/abs_app_right.c
@@ -0,0 +1,104 @@
+// Copyright (c) 2024, Marvin Borner <dev@marvinborner.de>
+// SPDX-License-Identifier: MIT
+
+#include <log.h>
+#include <impl.h>
+
+// my blc 3 for now!
+
+static void encode(Term *term, FILE *fp)
+{
+	switch (term->type) {
+	case ABS:
+		while (1) {
+			fprintf(fp, "0");
+			if (term->u.abs.body->type != ABS)
+				break;
+			term = term->u.abs.body;
+		}
+		fprintf(fp, "10");
+		encode(term->u.abs.body, fp);
+		break;
+	case APP:;
+		Term *stack[128] = { 0 }; // TODO: HEAP!
+		int stacked = 0;
+
+		while (1) {
+			fprintf(fp, "0");
+			stack[stacked++] = term->u.app.rhs;
+			term = term->u.app.lhs;
+			if (term->type != APP)
+				break;
+		}
+
+		fprintf(fp, "11");
+		encode(term, fp);
+		for (int i = stacked - 1; i >= 0; i--)
+			encode(stack[i], fp);
+		break;
+	case IDX:
+		for (size_t i = 0; i <= term->u.index; i++)
+			fprintf(fp, "1");
+		fprintf(fp, "0");
+		break;
+	default:
+		fatal("invalid type %d\n", term->type);
+	}
+}
+
+static Term *decode(FILE *fp)
+{
+	Term *res = 0;
+
+	char a = getc(fp);
+
+	if (a == '0') {
+		size_t count = 0;
+		while (getc(fp) == '0')
+			count++;
+
+		char b = getc(fp);
+
+		if (b == '0') {
+			Term *head = term_new(ABS);
+
+			res = head;
+			for (size_t i = 0; i < count; i++) {
+				res->u.abs.body = term_new(ABS);
+				res = res->u.abs.body;
+			}
+			res->u.abs.body = decode(fp);
+			res = head;
+		} else if (b == '1') {
+			res = term_new(APP);
+			res->u.app.lhs = decode(fp);
+			res->u.app.rhs = decode(fp);
+
+			for (size_t i = 0; i < count; i++) {
+				Term *prev = res;
+				res = term_new(APP);
+				res->u.app.lhs = prev;
+				res->u.app.rhs = decode(fp);
+			}
+		}
+	} else if (a == '1') {
+		res = term_new(IDX);
+		res->u.index = 0;
+		while (getc(fp) == '1')
+			res->u.index++;
+	}
+
+	if (!res)
+		fatal("invalid parsing state!\n");
+
+	return res;
+}
+
+Impl impl_abs_app_left(void)
+{
+	return (Impl){
+		.name = "abs_app_left",
+		.encode = encode,
+		.decode = decode,
+	};
+}
diff --git a/src/impl/app_both.c b/src/impl/app_both.c
new file mode 100644
index 0000000..47f78aa
--- /dev/null
+++ b/src/impl/app_both.c
@@ -0,0 +1,112 @@
+// Copyright (c) 2024, Marvin Borner <dev@marvinborner.de>
+// SPDX-License-Identifier: MIT
+
+#include <log.h>
+#include <impl.h>
+
+static void encode(Term *term, FILE *fp)
+{
+	switch (term->type) {
+	case ABS:
+		fprintf(fp, "01");
+		encode(term->u.abs.body, fp);
+		break;
+	case APP:
+		fprintf(fp, "0");
+
+		Term *stack[128] = { 0 }; // TODO: HEAP!
+		int stacked = 0;
+
+		// TODO: there's probably something here about finding optimal
+		//       left/right folds but this is ok for now
+
+		int fold_right = term->u.app.rhs->type == APP;
+
+		while (1) {
+			fprintf(fp, "0");
+			stack[stacked++] =
+				fold_right ? term->u.app.lhs : term->u.app.rhs;
+			term = fold_right ? term->u.app.rhs : term->u.app.lhs;
+			if (term->type != APP)
+				break;
+		}
+
+		fprintf(fp, "1%d", fold_right);
+		if (fold_right) {
+			for (int i = 0; i < stacked; i++)
+				encode(stack[i], fp);
+			encode(term, fp);
+		} else {
+			encode(term, fp);
+			for (int i = stacked - 1; i >= 0; i--)
+				encode(stack[i], fp);
+		}
+		break;
+	case IDX:
+		for (size_t i = 0; i <= term->u.index; i++)
+			fprintf(fp, "1");
+		fprintf(fp, "0");
+		break;
+	default:
+		fatal("invalid type %d\n", term->type);
+	}
+}
+
+static Term *decode(FILE *fp)
+{
+	Term *res = 0;
+
+	char a = getc(fp);
+
+	if (a == '0') {
+		size_t count = 0;
+		while (getc(fp) == '0')
+			count++;
+
+		if (count == 0) {
+			res = term_new(ABS);
+			res->u.abs.body = decode(fp);
+		} else if (getc(fp) == '1') { // fold right
+			Term *head = term_new(APP);
+			res = head;
+			for (size_t i = 0; i < count - 1; i++) {
+				res->u.app.lhs = decode(fp);
+				res->u.app.rhs = term_new(APP);
+				res = res->u.app.rhs;
+			}
+			res->u.app.lhs = decode(fp);
+			res->u.app.rhs = decode(fp);
+			res = head;
+		} else { // fold left
+			res = term_new(APP);
+			res->u.app.lhs = decode(fp);
+			res->u.app.rhs = decode(fp);
+
+			for (size_t i = 0; i < count - 1; i++) {
+				Term *prev = res;
+				res = term_new(APP);
+				res->u.app.lhs = prev;
+				res->u.app.rhs = decode(fp);
+			}
+		}
+	} else if (a == '1') {
+		res = term_new(IDX);
+		res->u.index = 0;
+		while (getc(fp) == '1')
+			res->u.index++;
+	}
+
+	if (!res)
+		fatal("invalid parsing state!\n");
+
+	return res;
+}
+
+Impl impl_app_both(void)
+{
+	return (Impl){
+		.name = "app_both",
+		.encode = encode,
+		.decode = decode,
+	};
+}
diff --git a/src/impl/app_left.c b/src/impl/app_left.c
new file mode 100644
index 0000000..dcc62cd
--- /dev/null
+++ b/src/impl/app_left.c
@@ -0,0 +1,89 @@
+// Copyright (c) 2024, Marvin Borner <dev@marvinborner.de>
+// SPDX-License-Identifier: MIT
+
+#include <log.h>
+#include <impl.h>
+
+static void encode(Term *term, FILE *fp)
+{
+	switch (term->type) {
+	case ABS:
+		fprintf(fp, "01");
+		encode(term->u.abs.body, fp);
+		break;
+	case APP:
+		fprintf(fp, "0");
+
+		Term *stack[128] = { 0 }; // TODO: HEAP!
+		int stacked = 0;
+
+		while (1) {
+			fprintf(fp, "0");
+			stack[stacked++] = term->u.app.rhs;
+			term = term->u.app.lhs;
+			if (term->type != APP)
+				break;
+		}
+
+		fprintf(fp, "1");
+		encode(term, fp);
+		for (int i = stacked - 1; i >= 0; i--)
+			encode(stack[i], fp);
+		break;
+	case IDX:
+		for (size_t i = 0; i <= term->u.index; i++)
+			fprintf(fp, "1");
+		fprintf(fp, "0");
+		break;
+	default:
+		fatal("invalid type %d\n", term->type);
+	}
+}
+
+static Term *decode(FILE *fp)
+{
+	Term *res = 0;
+
+	char a = getc(fp);
+
+	if (a == '0') {
+		size_t count = 0;
+		while (getc(fp) == '0')
+			count++;
+
+		if (count == 0) {
+			res = term_new(ABS);
+			res->u.abs.body = decode(fp);
+		} else { // foldl1
+			res = term_new(APP);
+			res->u.app.lhs = decode(fp);
+			res->u.app.rhs = decode(fp);
+
+			for (size_t i = 0; i < count - 1; i++) {
+				Term *prev = res;
+				res = term_new(APP);
+				res->u.app.lhs = prev;
+				res->u.app.rhs = decode(fp);
+			}
+		}
+	} else if (a == '1') {
+		res = term_new(IDX);
+		res->u.index = 0;
+		while (getc(fp) == '1')
+			res->u.index++;
+	}
+
+	if (!res)
+		fatal("invalid parsing state!\n");
+
+	return res;
+}
+
+Impl impl_app_left(void)
+{
+	return (Impl){
+		.name = "app_left",
+		.encode = encode,
+		.decode = decode,
+	};
+}
diff --git a/src/impl/app_right.c b/src/impl/app_right.c
new file mode 100644
index 0000000..2583741
--- /dev/null
+++ b/src/impl/app_right.c
@@ -0,0 +1,89 @@
+// Copyright (c) 2024, Marvin Borner <dev@marvinborner.de>
+// SPDX-License-Identifier: MIT
+
+#include <log.h>
+#include <impl.h>
+
+static void encode(Term *term, FILE *fp)
+{
+	switch (term->type) {
+	case ABS:
+		fprintf(fp, "01");
+		encode(term->u.abs.body, fp);
+		break;
+	case APP:
+		fprintf(fp, "0");
+
+		Term *stack[128] = { 0 }; // TODO: HEAP!
+		int stacked = 0;
+
+		while (1) {
+			fprintf(fp, "0");
+			stack[stacked++] = term->u.app.lhs;
+			term = term->u.app.rhs;
+			if (term->type != APP)
+				break;
+		}
+
+		fprintf(fp, "1");
+		for (int i = 0; i < stacked; i++)
+			encode(stack[i], fp);
+		encode(term, fp);
+		break;
+	case IDX:
+		for (size_t i = 0; i <= term->u.index; i++)
+			fprintf(fp, "1");
+		fprintf(fp, "0");
+		break;
+	default:
+		fatal("invalid type %d\n", term->type);
+	}
+}
+
+static Term *decode(FILE *fp)
+{
+	Term *res = 0;
+
+	char a = getc(fp);
+
+	if (a == '0') {
+		size_t count = 0;
+		while (getc(fp) == '0')
+			count++;
+
+		if (count == 0) {
+			res = term_new(ABS);
+			res->u.abs.body = decode(fp);
+		} else {
+			Term *head = term_new(APP);
+			res = head;
+			for (size_t i = 0; i < count - 1; i++) {
+				res->u.app.lhs = decode(fp);
+				res->u.app.rhs = term_new(APP);
+				res = res->u.app.rhs;
+			}
+			res->u.app.lhs = decode(fp);
+			res->u.app.rhs = decode(fp);
+			res = head;
+		}
+	} else if (a == '1') {
+		res = term_new(IDX);
+		res->u.index = 0;
+		while (getc(fp) == '1')
+			res->u.index++;
+	}
+
+	if (!res)
+		fatal("invalid parsing state!\n");
+
+	return res;
+}
+
+Impl impl_app_right(void)
+{
+	return (Impl){
+		.name = "app_right",
+		.encode = encode,
+		.decode = decode,
+	};
+}
diff --git a/src/impl/merged_abs.c b/src/impl/merged_abs.c
deleted file mode 100644
index e69de29..0000000
--- a/src/impl/merged_abs.c
+++ /dev/null
diff --git a/src/impl/merged_app.c b/src/impl/merged_app.c
deleted file mode 100644
index e69de29..0000000
--- a/src/impl/merged_app.c
+++ /dev/null
diff --git a/src/impl/merged_both.c b/src/impl/merged_both.c
deleted file mode 100644
index e69de29..0000000
--- a/src/impl/merged_both.c
+++ /dev/null