From 93895b8a01535ee70ccc4d5e3189862a3cae61a2 Mon Sep 17 00:00:00 2001
From: marihachi <marihachi0620@gmail.com>
Date: Thu, 6 Jan 2022 01:05:37 +0900
Subject: [PATCH] Introduce nesting limit (#87)
* support fnDepthLimit
* fix parse option
* rename fnDepthLimit to nestLimit
* support limit nesting: big, bold, small, italic, strike
* improve mention
* fix hashtag
* support limit nesting: hashtag
* support limit nesting: url
* refine link label
* refactor
* fix link
* change default value of nestLimit
* fix link label
* add test
* restore and modify test
---
etc/mfm-js.api.md | 1 +
src/api.ts | 8 ++-
src/internal/parser.pegjs | 147 ++++++++++++++++++++++----------------
test/parser.ts | 144 ++++++++++++++++++++++++++++++++++++-
4 files changed, 235 insertions(+), 65 deletions(-)
diff --git a/etc/mfm-js.api.md b/etc/mfm-js.api.md
index d4e14c3..ea4f238 100644
--- a/etc/mfm-js.api.md
+++ b/etc/mfm-js.api.md
@@ -233,6 +233,7 @@ export type NodeType<T extends MfmNode['type']> = T extends 'quote' ? MfmQuote :
// @public (undocumented)
export function parse(input: string, opts?: Partial<{
fnNameList: string[];
+ nestLimit: number;
}>): MfmNode[];
// Warning: (ae-forgotten-export) The symbol "MfmPlainNode" needs to be exported by the entry point index.d.ts
diff --git a/src/api.ts b/src/api.ts
index 4124061..eaf2b3b 100644
--- a/src/api.ts
+++ b/src/api.ts
@@ -8,8 +8,12 @@ const parser: peg.Parser = require('./internal/parser');
/**
* Generates a MfmNode tree from the MFM string.
*/
-export function parse(input: string, opts: Partial<{ fnNameList: string[]; }> = {}): MfmNode[] {
- const nodes = parser.parse(input, { startRule: 'fullParser', fnNameList: opts.fnNameList });
+export function parse(input: string, opts: Partial<{ fnNameList: string[]; nestLimit: number; }> = {}): MfmNode[] {
+ const nodes = parser.parse(input, {
+ startRule: 'fullParser',
+ fnNameList: opts.fnNameList,
+ nestLimit: opts.nestLimit
+ });
return nodes;
}
diff --git a/src/internal/parser.pegjs b/src/internal/parser.pegjs
index 76cc315..715e02c 100644
--- a/src/internal/parser.pegjs
+++ b/src/internal/parser.pegjs
@@ -64,6 +64,28 @@
}
return options.fnNameList.includes(name);
}
+
+ // nesting control
+
+ const nestLimit = options.nestLimit || 20;
+ let depth = 0;
+ function enterNest() {
+ if (depth + 1 > nestLimit) {
+ return false;
+ }
+ depth++;
+ return true;
+ }
+
+ function leaveNest() {
+ depth--;
+ return true;
+ }
+
+ function fallbackNest() {
+ depth--;
+ return false;
+ }
}
//
@@ -230,19 +252,22 @@ unicodeEmoji
// inline: big
big
- = "***" content:(!"***" @inline)+ "***"
+ = "***" content:bigContent "***"
{
return FN('tada', { }, mergeText(content));
}
+bigContent
+ = &{ return enterNest(); } @(@(!"***" @inline)+ &{ return leaveNest(); } / &{ return fallbackNest(); })
+
// inline: bold
bold
- = "**" content:(!"**" @inline)+ "**"
+ = "**" content:boldContent "**"
{
return BOLD(mergeText(content));
}
- / "<b>" content:(!"</b>" @inline)+ "</b>"
+ / "<b>" content:boldTagContent "</b>"
{
return BOLD(mergeText(content));
}
@@ -252,25 +277,31 @@ bold
return BOLD(parsedContent);
}
+boldContent
+ = &{ return enterNest(); } @(@(!"**" @inline)+ &{ return leaveNest(); } / &{ return fallbackNest(); })
+
+boldTagContent
+ = &{ return enterNest(); } @(@(!"</b>" @inline)+ &{ return leaveNest(); } / &{ return fallbackNest(); })
+
// inline: small
small
- = "<small>" content:(!"</small>" @inline)+ "</small>"
+ = "<small>" content:smallContent "</small>"
{
return SMALL(mergeText(content));
}
+smallContent
+ = &{ return enterNest(); } @(@(!"</small>" @inline)+ &{ return leaveNest(); } / &{ return fallbackNest(); })
+
// inline: italic
italic
- = italicTag
- / italicAlt
-
-italicTag
- = "<i>" content:(!"</i>" @inline)+ "</i>"
+ = "<i>" content:italicContent "</i>"
{
return ITALIC(mergeText(content));
}
+ / italicAlt
italicAlt
= "*" content:$(!"*" ([a-z0-9]i / _))+ "*" &(EOF / LF / _ / ![a-z0-9]i)
@@ -284,18 +315,27 @@ italicAlt
return ITALIC(parsedContent);
}
+italicContent
+ = &{ return enterNest(); } @(@(!"</i>" @inline)+ &{ return leaveNest(); } / &{ return fallbackNest(); })
+
// inline: strike
strike
- = "~~" content:(!("~" / LF) @inline)+ "~~"
+ = "~~" content:strikeContent "~~"
{
return STRIKE(mergeText(content));
}
- / "<s>" content:(!("</s>" / LF) @inline)+ "</s>"
+ / "<s>" content:strikeTagContent "</s>"
{
return STRIKE(mergeText(content));
}
+strikeContent
+ = &{ return enterNest(); } @(@(!("~" / LF) @inline)+ &{ return leaveNest(); } / &{ return fallbackNest(); })
+
+strikeTagContent
+ = &{ return enterNest(); } @(@(!("</s>" / LF) @inline)+ &{ return leaveNest(); } / &{ return fallbackNest(); })
+
// inline: inlineCode
inlineCode
@@ -321,83 +361,59 @@ mention
}
mentionName
- = !"-" mentionNamePart+ // first char is not "-".
+ = [a-z0-9_]i (&("-"+ [a-z0-9_]i) . / [a-z0-9_]i)*
{
+ // NOTE: first char and last char are not "-".
return text();
}
-mentionNamePart
- = "-" &mentionNamePart // last char is not "-".
- / [a-z0-9_]i
-
mentionHost
- = ![.-] mentionHostPart+ // first char is neither "." nor "-".
+ = [a-z0-9_]i (&([.-]i+ [a-z0-9_]i) . / [a-z0-9_]i)*
{
+ // NOTE: first char and last char are neither "." nor "-".
return text();
}
-mentionHostPart
- = [.-] &mentionHostPart // last char is neither "." nor "-".
- / [a-z0-9_]i
-
// inline: hashtag
hashtag
- = "#" !("\uFE0F"? "\u20E3") content:hashtagContent
+ = "#" !("\uFE0F"? "\u20E3") !(invalidHashtagContent !hashtagContentPart) content:$hashtagContentPart+
{
return HASHTAG(content);
}
-hashtagContent
- = !(invalidHashtagContent !hashtagContentPart) hashtagContentPart+ { return text(); }
-
invalidHashtagContent
= [0-9]+
hashtagContentPart
- = hashtagBracketPair
- / hashtagChar
+ = "(" hashPairInner ")"
+ / "[" hashPairInner "]"
+ / "「" hashPairInner "ã€"
+ / ![ 　\t.,!?'"#:\/\[\]ã€ã€‘()「ã€<>] CHAR
-hashtagBracketPair
- = "(" hashtagContent* ")"
- / "[" hashtagContent* "]"
- / "「" hashtagContent* "ã€"
-
-hashtagChar
- = ![ 　\t.,!?'"#:\/\[\]ã€ã€‘()「ã€<>] CHAR
+hashPairInner
+ = &{ return enterNest(); } @(@hashtagContentPart* &{ return leaveNest(); } / &{ return fallbackNest(); })
// inline: URL
url
- = "<" url:altUrlFormat ">"
+ = "<" url:$("http" "s"? "://" (!(">" / _) CHAR)+) ">"
{
return N_URL(url, true);
}
- / url:urlFormat
-{
- return N_URL(url);
-}
-
-urlFormat
- = "http" "s"? "://" urlContentPart+
+ / "http" "s"? "://" (&([.,]+ urlContentPart) . / urlContentPart)+
{
- return text();
+ // NOTE: last char is neither "." nor ",".
+ return N_URL(text());
}
urlContentPart
- = urlBracketPair
- / [.,] &urlContentPart // last char is neither "." nor ",".
+ = "(" urlPairInner ")"
+ / "[" urlPairInner "]"
/ [a-z0-9_/:%#@$&?!~=+-]i
-urlBracketPair
- = "(" urlContentPart* ")"
- / "[" urlContentPart* "]"
-
-altUrlFormat
- = "http" "s"? "://" (!(">" / _) CHAR)+
-{
- return text();
-}
+urlPairInner
+ = &{ return enterNest(); } @(@(urlContentPart / [.,])* &{ return leaveNest(); } / &{ return fallbackNest(); })
// inline: link
@@ -408,23 +424,34 @@ link
}
linkLabel
- = linkLabelPart+
+ = (!"]" @linkLabelPart)+
linkLabelPart
- = url { return text(); /* text node */ }
- / link { return text(); /* text node */ }
- / mention { return text(); /* text node */ }
- / !"]" @inline
+ = emojiCode
+ / unicodeEmoji
+ / big
+ / bold
+ / small
+ / italic
+ / strike
+ / inlineCode
+ / mathInline
+ / hashtag
+ / fn
+ / inlineText
// inline: fn
fn
- = "$[" name:$([a-z0-9_]i)+ &{ return ensureFnName(name); } args:fnArgs? _ content:(!("]") @inline)+ "]"
+ = "$[" name:$([a-z0-9_]i)+ &{ return ensureFnName(name); } args:fnArgs? _ content:fnContent "]"
{
args = args || {};
return FN(name, args, mergeText(content));
}
+fnContent
+ = &{ return enterNest(); } @(@(!"]" @inline)+ &{ return leaveNest(); } / &{ return fallbackNest(); })
+
fnArgs
= "." head:fnArg tails:("," @fnArg)*
{
diff --git a/test/parser.ts b/test/parser.ts
index 11d00e2..bcc5757 100644
--- a/test/parser.ts
+++ b/test/parser.ts
@@ -984,14 +984,16 @@ hoge`;
assert.deepStrictEqual(mfm.parse(input), output);
});
- it('do not yield link node even if label is recognisable as a link', () => {
+ it('cannot nest a link in a link label', () => {
const input = 'official instance: [[https://misskey.io/@ai](https://misskey.io/@ai)](https://misskey.io/@ai).';
const output = [
TEXT('official instance: '),
LINK(false, 'https://misskey.io/@ai', [
- TEXT('[https://misskey.io/@ai](https://misskey.io/@ai)')
+ TEXT('[https://misskey.io/@ai')
]),
- TEXT('.')
+ TEXT(']('),
+ N_URL('https://misskey.io/@ai'),
+ TEXT(').'),
];
assert.deepStrictEqual(mfm.parse(input), output);
});
@@ -1081,6 +1083,142 @@ hoge`;
});
});
+ describe('nesting limit', () => {
+ it('big', () => {
+ const input = '<b><b>***abc***</b></b>';
+ const output = [
+ BOLD([
+ BOLD([
+ TEXT('**'),
+ ITALIC([
+ TEXT('abc'),
+ ]),
+ TEXT('**'),
+ ]),
+ ]),
+ ];
+ assert.deepStrictEqual(mfm.parse(input, { nestLimit: 2 }), output);
+ });
+
+ describe('bold', () => {
+ it('basic', () => {
+ const input = '<i><i>**abc**</i></i>';
+ const output = [
+ ITALIC([
+ ITALIC([
+ TEXT('*'),
+ ITALIC([
+ TEXT('abc'),
+ ]),
+ TEXT('*'),
+ ]),
+ ]),
+ ];
+ assert.deepStrictEqual(mfm.parse(input, { nestLimit: 2 }), output);
+ });
+
+ it('tag', () => {
+ const input = '<i><i><b>abc</b></i></i>';
+ const output = [
+ ITALIC([
+ ITALIC([
+ TEXT('<b>abc</b>'),
+ ]),
+ ]),
+ ];
+ assert.deepStrictEqual(mfm.parse(input, { nestLimit: 2 }), output);
+ });
+ });
+
+ it('small', () => {
+ const input = '<i><i><small>abc</small></i></i>';
+ const output = [
+ ITALIC([
+ ITALIC([
+ TEXT('<small>abc</small>'),
+ ]),
+ ]),
+ ];
+ assert.deepStrictEqual(mfm.parse(input, { nestLimit: 2 }), output);
+ });
+
+ it('italic', () => {
+ const input = '<b><b><i>abc</i></b></b>';
+ const output = [
+ BOLD([
+ BOLD([
+ TEXT('<i>abc</i>'),
+ ]),
+ ]),
+ ];
+ assert.deepStrictEqual(mfm.parse(input, { nestLimit: 2 }), output);
+ });
+
+ describe('strike', () => {
+ it('basic', () => {
+ const input = '<b><b>~~abc~~</b></b>';
+ const output = [
+ BOLD([
+ BOLD([
+ TEXT('~~abc~~'),
+ ]),
+ ]),
+ ];
+ assert.deepStrictEqual(mfm.parse(input, { nestLimit: 2 }), output);
+ });
+
+ it('tag', () => {
+ const input = '<b><b><s>abc</s></b></b>';
+ const output = [
+ BOLD([
+ BOLD([
+ TEXT('<s>abc</s>'),
+ ]),
+ ]),
+ ];
+ assert.deepStrictEqual(mfm.parse(input, { nestLimit: 2 }), output);
+ });
+ });
+
+ it('hashtag', () => {
+ const input = '<b><b>#abc(xyz)</b></b>';
+ const output = [
+ BOLD([
+ BOLD([
+ HASHTAG('abc'),
+ TEXT('(xyz)'),
+ ]),
+ ]),
+ ];
+ assert.deepStrictEqual(mfm.parse(input, { nestLimit: 2 }), output);
+ });
+
+ it('url', () => {
+ const input = '<b><b>https://example.com/abc(xyz)</b></b>';
+ const output = [
+ BOLD([
+ BOLD([
+ N_URL('https://example.com/abc'),
+ TEXT('(xyz)'),
+ ]),
+ ]),
+ ];
+ assert.deepStrictEqual(mfm.parse(input, { nestLimit: 2 }), output);
+ });
+
+ it('fn', () => {
+ const input = '<b><b>$[a b]</b></b>';
+ const output = [
+ BOLD([
+ BOLD([
+ TEXT('$[a b]'),
+ ]),
+ ]),
+ ];
+ assert.deepStrictEqual(mfm.parse(input, { nestLimit: 2 }), output);
+ });
+ });
+
it('composite', () => {
const input =
`before
--
GitLab