ed/Odin
1
0
Fork 0
mirror of https://github.com/Ed94/Odin.git synced 2026-08-01 12:18:15 +00:00
Code Issues Packages Projects Releases Wiki Activity

[xml] Improve XML tests, test core:encoding/entity.

Browse Source
This commit is contained in:
Jeroen van Rijn
2021-12-05 02:52:23 +01:00
parent 3d72e80ccf
commit d65d6edb0e
7 changed files with 291 additions and 176 deletions
Download Patch File Download Diff File Expand all files Collapse all files
core/encoding/entity/entity.odin
+7
View File
@@ -115,7 +115,14 @@ decode_xml :: proc(input: string, options := XML_Decode_Options{}, allocator :=
We don't need to check if we need to write a `<`, because if it isn't CDATA or a comment, We don't need to check if we need to write a `<`, because if it isn't CDATA or a comment,
it couldn't have been part of an XML tag body to be decoded here. it couldn't have been part of an XML tag body to be decoded here.
Keep in mind that we could already *be* inside a CDATA tag.
If so, write `>` as a literal and continue.
*/ */
if in_data {
write_rune(&builder, '<')
continue
}
in_data = _handle_xml_special(&t, &builder, options) or_return in_data = _handle_xml_special(&t, &builder, options) or_return
case ']': case ']':
core/encoding/entity/example/entity_example.odin
+1
View File
@@ -50,6 +50,7 @@ _main :: proc() {
using fmt using fmt
options := xml.Options{ flags = { .Ignore_Unsupported, .Intern_Comments, .Unbox_CDATA, .Decode_SGML_Entities }} options := xml.Options{ flags = { .Ignore_Unsupported, .Intern_Comments, .Unbox_CDATA, .Decode_SGML_Entities }}
doc, _ := xml.parse(#load("test.html"), options) doc, _ := xml.parse(#load("test.html"), options)
defer xml.destroy(doc) defer xml.destroy(doc)
core/encoding/entity/example/test.html
+1 -1
View File
@@ -22,7 +22,7 @@
</div> </div>
<!-- EXPECTED: Foozle]! © BOX ® /BOX42&;1234& --> <!-- EXPECTED: Foozle]! © BOX ® /BOX42&;1234& -->
<div> <div>
&verbar; &vert; &VerticalLine; &fjlig; &grave; &bsol; &reg; &rhov; &CounterClockwiseContourIntegral; &verbar; &vert; &VerticalLine; &fjlig; &grave; &bsol; &reg; &rhov; &CounterClockwiseContourIntegral; &bsemi;
</div> </div>
</body> </body>
</html> </html>
tests/core/assets/XML/entities.html
+29
View File
@@ -0,0 +1,29 @@
<html>
<head>
<title>Entity Reference Test</title>
<style>
body {
background: #000; color: #eee;
width: 40%;
margin-left: auto;
margin-right: auto;
font-size: 14pt;
}
</style>
</head>
<body>
<h1>Entity Reference Test</h1>
<div id="test_cdata_in_comment" foo="">
Foozle]!&#32;&copy;&#x20;<!-- <![CDATA[&#32;&reg;&#x20;]]> -->42&;1234&
</div>
<!-- foo attribute should be empty but present -->
<!-- EXPECTED: Foozle]! © 42&;1234& -->
<div id="test_cdata_unwrap_and_passthrough">
Foozle]!&#32;&copy;&#x20;<![CDATA[BOX&#32;&reg;&#x20;/BOX]]>42&;1234&
</div>
<!-- EXPECTED: Foozle]! © BOX ® /BOX42&;1234& -->
<div>
&verbar; &vert; &VerticalLine; &fjlig; &grave; &bsol; &reg; &rhov; &CounterClockwiseContourIntegral; &bsemi;
</div>
</body>
</html>
tests/core/assets/XML/nl_NL-xliff-1.0.xliff → tests/core/assets/XML/nl_NL-xliff-1.2.xliff
View File
tests/core/assets/XML/utf8.xml
+1 -1
View File
@@ -4,5 +4,5 @@
<부끄러운:barzle> <부끄러운:barzle>
<name foo:bar="birmese">ရှက်စရာ ဇီးကွက်</name> <name foo:bar="birmese">ရှက်စရာ ဇီးကွက်</name>
<nickname>Owl of Shame</nickname> <nickname>Owl of Shame</nickname>
<data>More CDATA <![CDATA[<greeting>Hello, world!</greeting><![CDATA] < ]]> Nonsense.</data> <data>More CDATA <![CDATA[<greeting>Hello, world!</greeting><![CDATA] <$]]> Nonsense.</data>
</부끄러운:barzle> </부끄러운:barzle>
tests/core/encoding/xml/test_core_xml.odin
+252 -174
View File
@@ -3,16 +3,16 @@ package test_core_xml
import "core:encoding/xml" import "core:encoding/xml"
import "core:testing" import "core:testing"
import "core:mem" import "core:mem"
import "core:strings"
import "core:io"
import "core:fmt" import "core:fmt"
import "core:hash"
Silent :: proc(pos: xml.Pos, fmt: string, args: ..any) { Silent :: proc(pos: xml.Pos, fmt: string, args: ..any) {
// Custom (silent) error handler. // Custom (silent) error handler.
} }
OPTIONS :: xml.Options{ OPTIONS :: xml.Options{ flags = { .Ignore_Unsupported, .Intern_Comments, },
flags = {
.Ignore_Unsupported, .Intern_Comments,
},
expected_doctype = "", expected_doctype = "",
} }
@@ -22,76 +22,153 @@ TEST_fail := 0
TEST :: struct { TEST :: struct {
filename: string, filename: string,
options: xml.Options, options: xml.Options,
expected: struct { err: xml.Error,
error: xml.Error, crc32: u32,
xml_version: string,
xml_encoding: string,
doctype: string,
},
} }
/*
Relative to ODIN_ROOT
*/
TEST_FILE_PATH_PREFIX :: "tests/core/assets/XML"
TESTS :: []TEST{ TESTS :: []TEST{
/* /*
First we test that certain files parse without error. First we test that certain files parse without error.
*/ */
{ {
filename = "assets/XML/utf8.xml", /*
options = OPTIONS, <?xml version="1.0" encoding="utf-8"?>
expected = { <!DOCTYPE 恥ずべきフクロウ>
error = .None, <恥ずべきフクロウ 올빼미_id="Foozle&#32;<![CDATA[<greeting>Hello, world!"</greeting>]]>Barzle">
xml_version = "1.0", <부끄러운:barzle>
xml_encoding = "utf-8", <name foo:bar="birmese">ရှက်စရာ ဇီးကွက်</name>
doctype = "恥ずべきフクロウ", <nickname>Owl of Shame</nickname>
<data>More CDATA <![CDATA[<greeting>Hello, world!</greeting><![CDATA] <$]]> Nonsense.</data>
</부끄러운:barzle>
*/
/*
Tests UTF-8 idents and values.
Test namespaced ident.
Tests that nested partial CDATA start doesn't trip up parser.
*/
filename = "utf8.xml",
options = {
flags = {
.Ignore_Unsupported, .Intern_Comments,
},
expected_doctype = "恥ずべきフクロウ",
}, },
crc32 = 0x30d82264,
}, },
{ {
filename = "assets/XML/nl_NL-qt-ts.ts", /*
options = OPTIONS, Same as above.
expected = { Unbox CDATA in data tag.
error = .None, */
xml_version = "1.0", filename = "utf8.xml",
xml_encoding = "utf-8", options = {
doctype = "TS", flags = {
.Ignore_Unsupported, .Intern_Comments, .Unbox_CDATA,
},
expected_doctype = "恥ずべきフクロウ",
}, },
crc32 = 0x6d38ac58,
}, },
{ {
filename = "assets/XML/nl_NL-xliff-1.0.xliff", /*
options = OPTIONS, Simple Qt TS translation file.
expected = { `core:i18n` requires it to be parsed properly.
error = .None, */
xml_version = "1.0", filename = "nl_NL-qt-ts.ts",
xml_encoding = "UTF-8", options = {
doctype = "", flags = {
.Ignore_Unsupported, .Intern_Comments, .Unbox_CDATA, .Decode_SGML_Entities,
},
expected_doctype = "TS",
}, },
crc32 = 0x7bce2630,
}, },
{ {
filename = "assets/XML/nl_NL-xliff-2.0.xliff", /*
options = OPTIONS, Simple XLiff 1.2 file.
expected = { `core:i18n` requires it to be parsed properly.
error = .None, */
xml_version = "1.0", filename = "nl_NL-xliff-1.2.xliff",
xml_encoding = "utf-8", options = {
doctype = "", flags = {
.Ignore_Unsupported, .Intern_Comments, .Unbox_CDATA, .Decode_SGML_Entities,
},
expected_doctype = "xliff",
}, },
crc32 = 0x43f19d61,
},
{
/*
Simple XLiff 2.0 file.
`core:i18n` requires it to be parsed properly.
*/
filename = "nl_NL-xliff-2.0.xliff",
options = {
flags = {
.Ignore_Unsupported, .Intern_Comments, .Unbox_CDATA, .Decode_SGML_Entities,
},
expected_doctype = "xliff",
},
crc32 = 0x961e7635,
},
{
filename = "entities.html",
options = {
flags = {
.Ignore_Unsupported, .Intern_Comments,
},
expected_doctype = "html",
},
crc32 = 0xdb4a1e79,
},
{
filename = "entities.html",
options = {
flags = {
.Ignore_Unsupported, .Intern_Comments, .Unbox_CDATA,
},
expected_doctype = "html",
},
crc32 = 0x82588917,
},
{
filename = "entities.html",
options = {
flags = {
.Ignore_Unsupported, .Intern_Comments, .Unbox_CDATA, .Decode_SGML_Entities,
},
expected_doctype = "html",
},
crc32 = 0x5e74d8a6,
}, },
/* /*
Then we test that certain errors are returned as expected. Then we test that certain errors are returned as expected.
*/ */
{ {
filename = "assets/XML/utf8.xml", filename = "utf8.xml",
options = { options = {
flags = { flags = {
.Ignore_Unsupported, .Intern_Comments, .Ignore_Unsupported, .Intern_Comments,
}, },
expected_doctype = "Odin", expected_doctype = "Odin",
}, },
expected = { err = .Invalid_DocType,
error = .Invalid_DocType, crc32 = 0x49b83d0a,
xml_version = "1.0",
xml_encoding = "utf-8",
doctype = "恥ずべきフクロウ",
},
}, },
} }
@@ -115,6 +192,136 @@ when ODIN_TEST {
} }
} }
test_file_path :: proc(filename: string) -> (path: string) {
path = fmt.tprintf("%v%v/%v", ODIN_ROOT, TEST_FILE_PATH_PREFIX, filename)
temp := transmute([]u8)path
for r, i in path {
if r == '\\' {
temp[i] = '/'
}
}
return path
}
doc_to_string :: proc(doc: ^xml.Document) -> (result: string) {
/*
Effectively a clone of the debug printer in the xml package.
We duplicate it here so that the way it prints an XML document to a string is stable.
This way we can hash the output. If it changes, it means that the document or how it was parsed changed,
not how it was printed. One less source of variability.
*/
print :: proc(writer: io.Writer, doc: ^xml.Document) -> (written: int, err: io.Error) {
if doc == nil { return }
using fmt
written += wprintf(writer, "[XML Prolog]\n")
for attr in doc.prolog {
written += wprintf(writer, "\t%v: %v\n", attr.key, attr.val)
}
written += wprintf(writer, "[Encoding] %v\n", doc.encoding)
if len(doc.doctype.ident) > 0 {
written += wprintf(writer, "[DOCTYPE] %v\n", doc.doctype.ident)
if len(doc.doctype.rest) > 0 {
wprintf(writer, "\t%v\n", doc.doctype.rest)
}
}
for comment in doc.comments {
written += wprintf(writer, "[Pre-root comment] %v\n", comment)
}
if doc.root != nil {
wprintln(writer, " --- ")
print_element(writer, doc.root)
wprintln(writer, " --- ")
}
return written, .None
}
print_element :: proc(writer: io.Writer, element: ^xml.Element, indent := 0) -> (written: int, err: io.Error) {
if element == nil { return }
using fmt
tab :: proc(writer: io.Writer, indent: int) {
for _ in 0..=indent {
wprintf(writer, "\t")
}
}
tab(writer, indent)
if element.kind == .Element {
wprintf(writer, "<%v>\n", element.ident)
if len(element.value) > 0 {
tab(writer, indent + 1)
wprintf(writer, "[Value] %v\n", element.value)
}
for attr in element.attribs {
tab(writer, indent + 1)
wprintf(writer, "[Attr] %v: %v\n", attr.key, attr.val)
}
for child in element.children {
print_element(writer, child, indent + 1)
}
} else if element.kind == .Comment {
wprintf(writer, "[COMMENT] %v\n", element.value)
}
return written, .None
}
buf: strings.Builder
defer strings.destroy_builder(&buf)
print(strings.to_writer(&buf), doc)
return strings.clone(strings.to_string(buf))
}
@test
run_tests :: proc(t: ^testing.T) {
using fmt
for test in TESTS {
path := test_file_path(test.filename)
printf("\nTrying to parse %v\n\n", path)
doc, err := xml.parse(path, test.options, Silent)
defer xml.destroy(doc)
tree_string := doc_to_string(doc)
tree_bytes := transmute([]u8)tree_string
defer delete(tree_bytes)
crc32 := hash.crc32(tree_bytes)
failed := err != test.err
err_msg := tprintf("Expected return value %v, got %v", test.err, err)
expect(t, err == test.err, err_msg)
failed |= crc32 != test.crc32
err_msg = tprintf("Expected CRC 0x%08x, got 0x%08x", test.crc32, crc32)
expect(t, crc32 == test.crc32, err_msg)
if failed {
/*
Don't fully print big trees.
*/
tree_string = tree_string[:min(2_048, len(tree_string))]
println(tree_string)
}
}
}
main :: proc() { main :: proc() {
t := testing.T{} t := testing.T{}
@@ -133,132 +340,3 @@ main :: proc() {
fmt.printf("%v/%v tests successful.\n", TEST_count - TEST_fail, TEST_count) fmt.printf("%v/%v tests successful.\n", TEST_count - TEST_fail, TEST_count)
} }
@test
run_tests :: proc(t: ^testing.T) {
using fmt
count := 0
for test in TESTS {
printf("Trying to parse %v\n\n", test.filename)
doc, err := xml.parse(test.filename, test.options, Silent)
defer xml.destroy(doc)
err_msg := tprintf("Expected return value %v, got %v", test.expected.error, err)
expect(t, err == test.expected.error, err_msg)
if len(test.expected.xml_version) > 0 {
xml_version := ""
for attr in doc.prolog {
if attr.key == "version" {
xml_version = attr.val
}
}
err_msg = tprintf("Expected XML version %v, got %v", test.expected.xml_version, xml_version)
expect(t, xml_version == test.expected.xml_version, err_msg)
}
if len(test.expected.xml_encoding) > 0 {
xml_encoding := ""
for attr in doc.prolog {
if attr.key == "encoding" {
xml_encoding = attr.val
}
}
err_msg = tprintf("Expected XML encoding %v, got %v", test.expected.xml_encoding, xml_encoding)
expect(t, xml_encoding == test.expected.xml_encoding, err_msg)
}
err_msg = tprintf("Expected DOCTYPE %v, got %v", test.expected.doctype, doc.doctype.ident)
expect(t, doc.doctype.ident == test.expected.doctype, err_msg)
/*
File-specific tests.
*/
switch count {
case 0:
expect(t, len(doc.root.attribs) > 0, "Expected the root tag to have an attribute.")
attr := doc.root.attribs[0]
attr_key_expected := "올빼미_id"
attr_val_expected := "Foozle&#32;<![CDATA[<greeting>Hello, world!\"</greeting>]]>Barzle"
attr_err := tprintf("Expected %v, got %v", attr_key_expected, attr.key)
expect(t, attr.key == attr_key_expected, attr_err)
attr_err = tprintf("Expected %v, got %v", attr_val_expected, attr.val)
expect(t, attr.val == attr_val_expected, attr_err)
expect(t, len(doc.root.children) > 0, "Expected the root tag to have children.")
child := doc.root.children[0]
first_child_ident := "부끄러운:barzle"
attr_err = tprintf("Expected first child tag's ident to be %v, got %v", first_child_ident, child.ident)
expect(t, child.ident == first_child_ident, attr_err)
case 2:
expect(t, len(doc.root.attribs) > 0, "Expected the root tag to have an attribute.")
{
attr := doc.root.attribs[0]
attr_key_expected := "version"
attr_val_expected := "1.2"
attr_err := tprintf("Expected %v, got %v", attr_key_expected, attr.key)
expect(t, attr.key == attr_key_expected, attr_err)
attr_err = tprintf("Expected %v, got %v", attr_val_expected, attr.val)
expect(t, attr.val == attr_val_expected, attr_err)
}
{
attr := doc.root.attribs[1]
attr_key_expected := "xmlns"
attr_val_expected := "urn:oasis:names:tc:xliff:document:1.2"
attr_err := tprintf("Expected %v, got %v", attr_key_expected, attr.key)
expect(t, attr.key == attr_key_expected, attr_err)
attr_err = tprintf("Expected %v, got %v", attr_val_expected, attr.val)
expect(t, attr.val == attr_val_expected, attr_err)
}
case 3:
expect(t, len(doc.root.attribs) > 0, "Expected the root tag to have an attribute.")
{
attr := doc.root.attribs[0]
attr_key_expected := "xmlns"
attr_val_expected := "urn:oasis:names:tc:xliff:document:2.0"
attr_err := tprintf("Expected %v, got %v", attr_key_expected, attr.key)
expect(t, attr.key == attr_key_expected, attr_err)
attr_err = tprintf("Expected %v, got %v", attr_val_expected, attr.val)
expect(t, attr.val == attr_val_expected, attr_err)
}
{
attr := doc.root.attribs[1]
attr_key_expected := "version"
attr_val_expected := "2.0"
attr_err := tprintf("Expected %v, got %v", attr_key_expected, attr.key)
expect(t, attr.key == attr_key_expected, attr_err)
attr_err = tprintf("Expected %v, got %v", attr_val_expected, attr.val)
expect(t, attr.val == attr_val_expected, attr_err)
}
}
count += 1
}
}