hare

commit 59e3dfbef782f8b8640a119bc2aecf64a645a72c
parent 69c15619b04c2c992110ac93e5ae370ba12c1be9
Author: Drew DeVault <sir@cmpwn.com>
Date:   Mon, 19 Apr 2021 11:42:58 -0400

format::html: new module

Diffstat:
A
format/html/docs.ha
 | 
2
++
A
format/html/escape.ha
 | 
40
++++++++++++++++++++++++++++++++++++++++
M
scripts/gen-stdlib
 | 
8
++++++++
M
stdlib.mk
 | 
28
++++++++++++++++++++++++++++

4 files changed, 78 insertions(+), 0 deletions(-)
diff --git a/format/html/docs.ha b/format/html/docs.ha
@@ -0,0 +1,2 @@
+// This module provides an HTML parser. Or rather, it will provide one. Right
+// now all it does is escape HTML strings for safe printing.
diff --git a/format/html/escape.ha b/format/html/escape.ha
@@ -0,0 +1,40 @@
+use encoding::utf8;
+use io;
+use strings;
+use strio;
+
+// Prints a string to an output stream, escaping any of HTML's reserved
+// characters.
+export fn escape(in: str, out: *io::stream) (size | io::error) = {
+	let z = 0z;
+	let iter = strings::iter(in);
+	for (true) match (strings::next(&iter)) {
+		void => break,
+		rn: rune => z += io::write(out, switch (rn) {
+			'&' => strings::toutf8("&"),
+			'<' => strings::toutf8("&lt;"),
+			'>' => strings::toutf8("&gt;"),
+			'"' => strings::toutf8("&quot;"),
+			'\'' => strings::toutf8("&apos;"),
+			*   => utf8::encoderune(rn),
+		})?,
+	};
+	return z;
+};
+
+@test fn escape() void = {
+	let sink = strio::dynamic();
+	defer io::close(sink);
+	escape("hello world!", sink);
+	assert(strio::string(sink) == "hello world!");
+
+	let sink = strio::dynamic();
+	defer io::close(sink);
+	escape("\"hello world!\"", sink);
+	assert(strio::string(sink) == "&quot;hello world!&quot;");
+
+	let sink = strio::dynamic();
+	defer io::close(sink);
+	escape("<hello & 'world'!>", sink);
+	assert(strio::string(sink) == "&lt;hello &amp; &apos;world&apos;!&gt;");
+};
diff --git a/scripts/gen-stdlib b/scripts/gen-stdlib
@@ -267,6 +267,13 @@ format_elf() {
 	gen_ssa format::elf
 }
 
+format_html() {
+	gen_srcs format::html \
+		escape.ha \
+		docs.ha
+	gen_ssa format::html encoding::utf8 io strings strio
+}
+
 gensrcs_format_xml() {
 	gen_srcs format::xml \
 		types.ha \
@@ -674,6 +681,7 @@ endian
 errors
 fmt
 format_elf
+format_html
 format_xml
 fs
 fs_mem
diff --git a/stdlib.mk b/stdlib.mk
@@ -123,6 +123,9 @@ hare_stdlib_deps+=$(stdlib_fmt)
 stdlib_format_elf=$(HARECACHE)/format/elf/format_elf.o
 hare_stdlib_deps+=$(stdlib_format_elf)
 
+stdlib_format_html=$(HARECACHE)/format/html/format_html.o
+hare_stdlib_deps+=$(stdlib_format_html)
+
 stdlib_format_xml=$(HARECACHE)/format/xml/format_xml.o
 hare_stdlib_deps+=$(stdlib_format_xml)
 
@@ -427,6 +430,17 @@ $(HARECACHE)/format/elf/format_elf.ssa: $(stdlib_format_elf_srcs) $(stdlib_rt)
 	@HARECACHE=$(HARECACHE) $(HAREC) $(HAREFLAGS) -o $@ -Nformat::elf \
 		-t$(HARECACHE)/format/elf/format_elf.td $(stdlib_format_elf_srcs)
 
+# format::html
+stdlib_format_html_srcs= \
+	$(STDLIB)/format/html/escape.ha \
+	$(STDLIB)/format/html/docs.ha
+
+$(HARECACHE)/format/html/format_html.ssa: $(stdlib_format_html_srcs) $(stdlib_rt) $(stdlib_encoding_utf8) $(stdlib_io) $(stdlib_strings) $(stdlib_strio)
+	@printf 'HAREC \t$@\n'
+	@mkdir -p $(HARECACHE)/format/html
+	@HARECACHE=$(HARECACHE) $(HAREC) $(HAREFLAGS) -o $@ -Nformat::html \
+		-t$(HARECACHE)/format/html/format_html.td $(stdlib_format_html_srcs)
+
 # format::xml
 stdlib_format_xml_srcs= \
 	$(STDLIB)/format/xml/types.ha \
@@ -977,6 +991,9 @@ hare_testlib_deps+=$(testlib_fmt)
 testlib_format_elf=$(TESTCACHE)/format/elf/format_elf.o
 hare_testlib_deps+=$(testlib_format_elf)
 
+testlib_format_html=$(TESTCACHE)/format/html/format_html.o
+hare_testlib_deps+=$(testlib_format_html)
+
 testlib_format_xml=$(TESTCACHE)/format/xml/format_xml.o
 hare_testlib_deps+=$(testlib_format_xml)
 
@@ -1286,6 +1303,17 @@ $(TESTCACHE)/format/elf/format_elf.ssa: $(testlib_format_elf_srcs) $(testlib_rt)
 	@HARECACHE=$(TESTCACHE) $(HAREC) $(TESTHAREFLAGS) -o $@ -Nformat::elf \
 		-t$(TESTCACHE)/format/elf/format_elf.td $(testlib_format_elf_srcs)
 
+# format::html
+testlib_format_html_srcs= \
+	$(STDLIB)/format/html/escape.ha \
+	$(STDLIB)/format/html/docs.ha
+
+$(TESTCACHE)/format/html/format_html.ssa: $(testlib_format_html_srcs) $(testlib_rt) $(testlib_encoding_utf8) $(testlib_io) $(testlib_strings) $(testlib_strio)
+	@printf 'HAREC \t$@\n'
+	@mkdir -p $(TESTCACHE)/format/html
+	@HARECACHE=$(TESTCACHE) $(HAREC) $(TESTHAREFLAGS) -o $@ -Nformat::html \
+		-t$(TESTCACHE)/format/html/format_html.td $(testlib_format_html_srcs)
+
 # format::xml
 testlib_format_xml_srcs= \
 	$(STDLIB)/format/xml/types.ha \