From b046014cf60d3e8e04f97af4021a4d182af10589 Mon Sep 17 00:00:00 2001 From: eugenesvk Date: Sun, 12 Jan 2025 16:15:56 +0700 Subject: [PATCH] Add KDL2 grammar in KDL --- SPEC.md | 3 +- examples/kdl-grammar2.kdl | 118 ++++++++++++++++++++++++++++++++++++++ 2 files changed, 120 insertions(+), 1 deletion(-) create mode 100644 examples/kdl-grammar2.kdl diff --git a/SPEC.md b/SPEC.md index 7f2a72c..0a8e9df 100644 --- a/SPEC.md +++ b/SPEC.md @@ -837,7 +837,8 @@ except for non Unicode Scalar Value, which can't be represented even as escapes) This is the full official grammar for KDL and should be considered authoritative if something seems to disagree with the text above. The [grammar -language syntax](#grammar-language) is defined below. +language syntax](#grammar-language) is defined below. For a version converted to KDL, +see [examples/kdl-grammar2.kdl](./examples/kdl-grammar2.kdl). ``` document := bom? version? nodes diff --git a/examples/kdl-grammar2.kdl b/examples/kdl-grammar2.kdl new file mode 100644 index 0000000..605109f --- /dev/null +++ b/examples/kdl-grammar2.kdl @@ -0,0 +1,118 @@ +/-kdl-version 2 +§ "KDL Grammar" spec=(🔗)"github.com/kdl-org/kdl/blob/main/SPEC.md" !="See symbol defs at the bottom"\ +document = bom? version? nodes \ + version =≝ /- u␠* "kdl-version" u␠+ ⦅ "1" | "2" ⦆ u␠* ␤ \ + nodes = ⦅l_␠* node⦆* l_␠* \ + node = node_base node_end \ + node_chlast = node_base node_end? \ + node_base =≝ /-? 🄣? n␠* string \ + ⦅n␠+ /-? node_prop∨arg ⦆* \ + ⦅n␠+ /- node_children ⦆* !="Slashdashed node_children must be after props and args" \ + ⦅n␠+ node_children ⦆? !="Only children may follow a slashdashed child" \ + ⦅n␠+ /- node_children ⦆* \ + n␠* + +§ "Entries" \ +node_prop∨arg = prop | argument \ + prop = string n␠* "=" n␠* 🄣? n␠* value \ + argument = 🄣? n␠* value \ + 🄣 = "(" n␠* string n␠* ")" \ + value = string | number | boolean | #null \ +node_children = "{" nodes node_chlast? "}" \ +node_end = comment_line | ␤ | ； | eof + +§ "Strings" \ +string = str🆔 | string'' | string_raw ¶ \ + str🆔 = clean🆔 | signed🆔 | dotted🆔 \ + clean🆔 = ⦅⦅char🆔 − 〔0–9〕 − ± − .⦆ char🆔*⦆ − keyword🆔🛑 \ + signed🆔 = ± ⦅⦅char🆔 − 〔0–9〕 − .⦆ char🆔*⦆? \ + dotted🆔 = ±? . ⦅⦅char🆔 − 〔0–9〕 ⦆ char🆔*⦆? \ + char🆔 = uchar − u␠ − ␤ − char🆔¬ − char🛑 \ + char🆔¬ = #"(;){"}[#]/=\"# \ + keyword🆔🛑 = true⠀|⠀false⠀|⠀null⠀|⠀inf⠀|⠀-inf⠀|⠀nan \ + string'' = “ ⦅ char_str − ␤⦆* ” \ + |= “““ \ + ⦅␤ de␠e≡ ⦅“?“? char_str − ␤ - u␠⦆ ⦅“?“? char_str − ␤⦆* ⦆* \ + ⦅␤ ⦅ u␠⦆* ⦆* \ + ␤ de␠e* ””” !="de␠e≡ must exactly match de␠e; only required before non-whitespace" \ + char_str = char⎋ | uni⎋ | w␠⎋ | 〔¬⧵“〕 − char🛑 \ + char⎋ = ⧵ 〔“⧵bfnrts〕 \ + uni⎋ = ⧵ "u{" uchar₁₆ "}" \ + uchar₁₆ = 〔0–10FFFF〕 − (surrogate)〔D800–DFFF〕 !="ranges in \u{F} format, but represented as hex₁₆, not actual symbols, so \u{B0} is 'B0', not '°', can be 0-padded up to max len of 6"\ + w␠⎋ = ⧵ ⦅u␠|␤⦆+ \ + de␠e = u␠ | w␠⎋ \ + string_raw = ＃*ℕ string_raw_quoted ＃*ℕ≡ !="ℕ=1–∞ ℕ≡ must match ℕ" \ + string_raw_quoted = “ string_raw_body_line ” \ + |= “““ ␤ string_raw_body☰line ␤ de␠u* ””” \ + string_raw_body_line = "" \ + |= ⦅char_raw − ”⦆ char_raw*? \ + |= “ ⦅char_raw − ”⦆ char_raw*? \ + char_raw = uchar − ␤ − char🛑 \ + string_raw_body☰line = de␠u≡ ⦅uchar − char🛑⦆*? !="←de␠u must exactly match ≡↑" \ + de␠u = u␠ + +§ "Numbers (space in 〔〕 is ignored)" \ +number = float_keyword | hex | octal | binary | decimal \ + hex = ±?⠀0x 🔢₁₆ ⦅_ | 🔢₁₆⦆* \ + octal = ±?⠀0o 〔0–7〕 〔0–7_〕* \ + binary = ±?⠀0b 〔0⠀1〕 〔0⠀1_〕* \ + decimal = ±? integer ⦅. integer⦆? ℯ? \ + integer = 〔0–9〕 〔0–9_〕* \ + ℯ = 〔eE〕 ±? integer \ + \ + ± = 〔+-〕 \ + 🔢₁₆ = 〔0–9a–fA–F〕 + +§ "Keywords and booleans" \ +keyword = boolean | #null \ +float_keyword = #inf | #-inf | #nan \ +boolean = #true | #false + +§ "Specific code points" !="\u{0} unicode format is implied: 〔u0〕=〔\u{0}〕 〔A–F〕=〔\u{A}–\u{F}〕 (range)" \ +bom = "\u{FEFF}" \ +char🛑 = (¬1st)"bom" (c0a)〔u0–u8〕 (c0b)〔E–1F〕 (cdel)〔7F〕 (c_dir)〔200E–200F⠀202A–202E⠀2066–2069〕 usv¬ (🔗spec)"disallowed-literal-code-points" \ + usv¬ =(surrogate)≝ (high)〔D800–DBFF〕 (low)〔DC00–DFFF〕 !="valid c0: 〔u9–uD〕 9␉\t A␊\n B␋\\v C␌\f D␍\r" \ +uchar = 〔0–D7FF⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀E000–10FFFF〕 !="Any Unicode Scalar Value (no surrogates)" \ +u␠ = 〔9⠀20⠀A0⠀1680⠀2000⠀2001⠀2002⠀2003⠀2004⠀2005⠀2006⠀2007⠀2008⠀2009⠀200A⠀202F⠀205F⠀3000〕 !="Non line-breaking unicode White_Space" (🔗spec)"whitespace"\ + "                               　 "\ +u␤ = 〔A⠀B⠀C⠀D⠀85⠀2028⠀2029〕 !="(85NEL LS PS) Line-breaking unicode White_Space" (🔗spec)"newline" \ +“="\"" ”="\"" ⧵="\\" ；=";" ⁄="/" *="∗" ＃="#" + +§ "Comments (// /* */ after = serve as definition symbols)" \ +/- ≝ /- l_␠* \ +comment_block =≝ /* commented_block + commented_block = */ \ + |= ⦅comment_block | * | ⁄ | 〔¬*⁄〕+⦆ 🗘 \ +comment_line =≝ // ¬␤* ⦅␤ | eof⦆ + +§ "Whitespace" !="\u{F} unicode format is implied in 〔F〕 and ommited for brevity" \ +l_␠ = n␠ | ␤ | comment_line !="Whitespace where newlines are allowed" \ + n␠ = cu␠* escline cu␠* | cu␠+ !="Whitespace within nodes, where newline-ish things must be esclined" \ + cu␠ = u␠ | comment_block \ + escline = ⧵ cu␠* ⦅comment_line | ␤ | eof⦆ \ + ␤ = 〔D〕〔A〕 | u␤ !="〔D〕〔A〕 \r\n ␍␊ is a sequence, but counts as 1 newline" (🔗spec)"newline" + +§ "Grammar language help: ABNF-like with some regex in a " rule="definition" "format" { + • #"\⧵ "“ "” ;； /⁄ *∗ #＃"#="literals represented with unicode alts to avoid quotes" + • #"\ " ; = "#="regular KDL syntax, not grammar"\ + #" "x" "#="double quotes for literals" ##" #"x"# "##="raw+double quotes for literals"\ + ∅=(empty)"" #"\"#=(text)"escape \" \\ or add ‘uchar’ in hex \u{FEFF}"\ + keys=(and_types)"extra group info" ≝="filler to allow /-comments in prop values" + • |="logical OR" !="tip/comment" §="section marker" 🗘="refers to self (rule name left of =)" + •⠀Regex⠀matches \ + ?⠀="0|1 zero or one" \ + +⠀="1+ one or more ( greedy)" \ + *⠀="0+ zero or more ( greedy, match as many instances as possible)" \ + *?="0+ zero or more (non-greedy, match as few instances as possible; used only in raw strings)" \ + *1–6=" inclusive range: at least 1 at most 6" \ + "0–9"="char range (inclusive)" –="en-dash used to not blend with hyphen-minus in search" \ + ¬="not this (^ in regex)" ¬foo="must not match 'foo'" 〔¬⧵“〕=#"[^\"] any char except for \ or " "# \ + 〔set⠀〕="[a b]" #"set"#="[ab]" ≝="regex Char set matches like [], where any char will be a single match"\ + "↑ ignore ↑ contained spaces (for visual separation or groups and alignment)" + ⦅group⦆ "items must be matched together"\ + a|b="'a or b', whichever matches 1st. Multiple items before | are a single group" "a b c | d"="⦅a b c⦆ | d"\ + ≡="defined to be identical to another similarly named match" + • −="minus sign; 'except for' whatever follows it" "uchar⠀−⠀␤"="match ‘uchar’ rule, but not ‘␤’ rule" + • ¶="cut point: always matches and consumes no chars, but once matched, bans backtracking past that point in the source. If a parser would rewind past ¶, it must instead fail the overall parse, as if it had run out of options (only for ‘string_raw’ to ensure the 1st instance of the appropriate closing quote sequence ends it)" + • "A single definition may be split over multiple lines. Newlines are treated as spaces" +}