From 28ab3123df1fedd0e9402a5fa5e3c2e13171e5a9 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Guilhem=20Faur=C3=A9?= <pro@gfaure.eu>
Date: Fri, 28 Apr 2023 15:44:50 +0200
Subject: [PATCH] switch to new grammar

---
 spip2md/content.py     |   3 +-
 spip2md/spip.flex.lark |  75 ------------------------
 spip2md/spip.lark      | 129 +++++++++++++++++------------------------
 3 files changed, 55 insertions(+), 152 deletions(-)
 delete mode 100644 spip2md/spip.flex.lark

diff --git a/spip2md/content.py b/spip2md/content.py
index 648f877..815ffb2 100644
--- a/spip2md/content.py
+++ b/spip2md/content.py
@@ -2,8 +2,7 @@ from os import path
 
 from lark import Lark
 
-# spipParser = Lark(open(path.dirname(__file__) + "/spip.lark"))
-spipParser = Lark(open(path.dirname(__file__) + "/spip.flex.lark"))
+spipParser = Lark(open(path.dirname(__file__) + "/spip.lark"))
 
 
 class content:
diff --git a/spip2md/spip.flex.lark b/spip2md/spip.flex.lark
deleted file mode 100644
index 3f75835..0000000
--- a/spip2md/spip.flex.lark
+++ /dev/null
@@ -1,75 +0,0 @@
-// Flexible SPIP Markup grammar for Lark parser
-start: _N* block ( _N+ block )* _N*
-
-?block: paragraph
-      | heading
-      | list
-      | table
-      | tag
-      | HORIZONTAL_RULE -> horizontal_rule
-
-HORIZONTAL_RULE: /----+/
-
-?list: unordered_list
-     | ordered_list
-
-unordered_list: ( _UL list_item _N )+
-ordered_list: ( _OL list_item _N )+
-list_item: _inline{TEXT}+
-_UL: /-\*|-[^#-]/
-_OL: /-#/
-
-table: ( _TBL_META table_metadata "||" _N )? ( table_row _N )+
-table_metadata: table_title "|" table_description
-table_title: _inline{TABLE_TEXT}
-table_description: _inline{TABLE_TEXT}
-table_row: ( _TBL table_cell )+ "|"
-table_cell: _inline{TABLE_TEXT}
-_TBL_META: "||"
-_TBL: "|"
-
-heading: _H _inline{MARKED_TEXT}+ "}}}"
-_H: "{{{"
-
-paragraph: ( _inline{TEXT} _N? )+
-
-_inline{text}: text
-             | emphasis
-             | strong
-             | anchor
-             | tag
-
-TEXT: /(?:[^\r\n\{](?![^\[\n\r]*->))+/
-TABLE_TEXT: /[^\|\r\n\{]+/
-MARKED_TEXT: /[^\}\r\n\{]+/
-
-strong: _B ( _inline{MARKED_TEXT} )+ ( "}}" | _N )
-emphasis: _I ( _inline{MARKED_TEXT} )+ ( "}" | _N )
-_B: /{{(?=[^\{])/
-_I: /{(?=[^\{])/
-
-?anchor: anchor_footnote
-       | anchor_wikipedia
-       | anchor_normal -> anchor
-
-anchor_footnote: _FOOT HREF "]]"
-anchor_wikipedia: _WIKI HREF "]"
-anchor_normal: _A A_TEXT "->" HREF "]"
-_FOOT: /\[\[/
-_WIKI: /\[\?/
-_A: /\[(?=[^\[\n\r]+->)/
-HREF: _PURE_TEXT
-A_TEXT: /[^\r\n\{]+?(?=->)/
-
-tag: end_tag
-   | start_tag
-
-end_tag: _ETAG TAG_NAME ( "|" TAG_OPTION )* ">"
-start_tag: _STAG TAG_NAME ( "|" TAG_OPTION )* ">"
-_STAG: "<"
-_ETAG: "</"
-TAG_NAME: _PURE_TEXT
-TAG_OPTION: _PURE_TEXT
-
-_N: /\r?\n/
-_PURE_TEXT: /[0-9A-Za-z_:\/\-\.]+/
diff --git a/spip2md/spip.lark b/spip2md/spip.lark
index 744f1ea..3f75835 100644
--- a/spip2md/spip.lark
+++ b/spip2md/spip.lark
@@ -1,96 +1,75 @@
-// SPIP Markup grammar for Lark parser
-
-start: _N* block ( _N+ block )+ _N*
+// Flexible SPIP Markup grammar for Lark parser
+start: _N* block ( _N+ block )* _N*
 
 ?block: paragraph
       | heading
       | list
       | table
-      | orphan_tag
-      | SEPARATOR -> hr
+      | tag
+      | HORIZONTAL_RULE -> horizontal_rule
+
+HORIZONTAL_RULE: /----+/
 
 ?list: unordered_list
      | ordered_list
 
-unordered_list: ( _HYPHEN_STAR list_element _N )+ -> ul
-ordered_list: ( _HYPHEN_HASH list_element _N )+ -> ol
-list_element: _inline_format -> li
+unordered_list: ( _UL list_item _N )+
+ordered_list: ( _OL list_item _N )+
+list_item: _inline{TEXT}+
+_UL: /-\*|-[^#-]/
+_OL: /-#/
 
-table: ( table_meta _N )? ( row _N )+ -> table
-table_meta: _PIPE _PIPE TEXT _PIPE TEXT _PIPE _PIPE -> title_description
-row: ( _PIPE cell )+ _PIPE -> tr
-cell: _inline_format -> td
+table: ( _TBL_META table_metadata "||" _N )? ( table_row _N )+
+table_metadata: table_title "|" table_description
+table_title: _inline{TABLE_TEXT}
+table_description: _inline{TABLE_TEXT}
+table_row: ( _TBL table_cell )+ "|"
+table_cell: _inline{TABLE_TEXT}
+_TBL_META: "||"
+_TBL: "|"
 
-heading: _O_CURLY_3 ( TEXT | link | nested_italic | nested_bold ) _C_CURLY_3 -> h2
+heading: _H _inline{MARKED_TEXT}+ "}}}"
+_H: "{{{"
 
-paragraph: ( _inline_format _N? )+ -> p
+paragraph: ( _inline{TEXT} _N? )+
 
-_inline_format: TEXT
-              | italic
-              | bold
-              | link
-              | orphan_tag
-              | orphan_quote
-              | tag
+_inline{text}: text
+             | emphasis
+             | strong
+             | anchor
+             | tag
 
-bold: _O_CURLY_2 ( TEXT | link | nested_italic )+ _C_CURLY_2 -> strong
-italic: _O_CURLY ( TEXT | link | nested_bold )+ _C_CURLY -> em
-nested_bold: TEXT+ _O_CURLY_2 ( TEXT | link )+ _C_CURLY_2 -> strong
-nested_italic: TEXT+ _O_CURLY ( TEXT | link )+ _C_CURLY -> em
+TEXT: /(?:[^\r\n\{](?![^\[\n\r]*->))+/
+TABLE_TEXT: /[^\|\r\n\{]+/
+MARKED_TEXT: /[^\}\r\n\{]+/
 
-?link: footnote
-     | wikipedia_link
-     | a
+strong: _B ( _inline{MARKED_TEXT} )+ ( "}}" | _N )
+emphasis: _I ( _inline{MARKED_TEXT} )+ ( "}" | _N )
+_B: /{{(?=[^\{])/
+_I: /{(?=[^\{])/
 
-footnote: _O_SQUARE_2 HREF _C_SQUARE_2 -> footnote
-wikipedia_link: _O_SQUARE_INTERO HREF _C_SQUARE -> a_wikipedia
-a: _LINK_OPENING LINK_TEXT _ARROW HREF _C_SQUARE -> a
+?anchor: anchor_footnote
+       | anchor_wikipedia
+       | anchor_normal -> anchor
 
-tag: closing_quote
-   | opening_quote
-   | closing_tag
-   | opening_tag
+anchor_footnote: _FOOT HREF "]]"
+anchor_wikipedia: _WIKI HREF "]"
+anchor_normal: _A A_TEXT "->" HREF "]"
+_FOOT: /\[\[/
+_WIKI: /\[\?/
+_A: /\[(?=[^\[\n\r]+->)/
+HREF: _PURE_TEXT
+A_TEXT: /[^\r\n\{]+?(?=->)/
 
-orphan_quote: _ORPHAN_OPENING "quote" _C_ANGLE -> orphan_quote
-orphan_tag: _ORPHAN_OPENING TEXT ( _PIPE TEXT )* _C_ANGLE -> orphan_tag
+tag: end_tag
+   | start_tag
 
-closing_quote: _O_ANGLE _SLASH "quote" _C_ANGLE -> closing_quote
-opening_quote: _O_ANGLE "quote" _C_ANGLE -> opening_quote
-closing_tag: _O_ANGLE _SLASH TEXT ( _PIPE TEXT )* _C_ANGLE -> closing_tag
-opening_tag: _O_ANGLE TEXT ( _PIPE TEXT )* _C_ANGLE -> opening_tag
+end_tag: _ETAG TAG_NAME ( "|" TAG_OPTION )* ">"
+start_tag: _STAG TAG_NAME ( "|" TAG_OPTION )* ">"
+_STAG: "<"
+_ETAG: "</"
+TAG_NAME: _PURE_TEXT
+TAG_OPTION: _PURE_TEXT
 
-// Terminals
 _N: /\r?\n/
-SEPARATOR: "----" "-"*
-TEXT: ( PURE_TEXT | AMBIGUOUS_TEXT )+
-
-_HYPHEN_STAR: "-*"
-_HYPHEN_HASH: "-#"
-_PIPE: "|"
-_SLASH: "/"
-_O_ANGLE: "<"
-_C_ANGLE: ">"
-_ORPHAN_OPENING: /<(?=([^>\/]+?)>)(?!.*<\/\1>)/
-
-_O_CURLY_3: "{{{"
-_C_CURLY_3: "}}}"
-_O_CURLY_2: "{{"
-_C_CURLY_2: "}}"
-_O_CURLY: "{"
-_C_CURLY: "}"
-
-_O_SQUARE_2: "[["
-_C_SQUARE_2: "]]"
-_O_SQUARE_INTERO: "[?"
-_C_SQUARE: "]"
-_ARROW: "->"
-_LINK_OPENING: /\[(?=/ PURE_TEXT+ /]*->)/
-HREF: PURE_TEXT+
-LINK_TEXT: PURE_TEXT+ /(?=->)/
-
-/// Characters that could be markup but aren’t in this situation
-AMBIGUOUS_TEXT: /\[(?!.*->.*\])/
-              | /\]/
-              | /\\}/
-/// Every characters that have no markup meaning
-PURE_TEXT: /[^\r\n\|\{\}\[\]\<\>]/
+_PURE_TEXT: /[0-9A-Za-z_:\/\-\.]+/