almost a good parser

1 files changed, 20 insertions, 10 deletions

diff --git a/code/grammar/markdown.pest b/code/grammar/markdown.pest
index ebdfcd7..e387182 100644
--- a/code/grammar/markdown.pest
+++ b/code/grammar/markdown.pest
@@ -1,17 +1,27 @@
+// WHITESPACE = _{ " " | "\t" | NEWLINE }
 
-file = _{ SOI ~ block* ~ EOI }
+file = _{ SOI ~ NEWLINE* ~ (block ~ NEWLINE*)* ~ EOI }
 
-block = { (code_block | header | list | paragraph | "") ~ NEWLINE }
+block = { code_block | latex_block | header | unordered_list | ordered_list | paragraph }
     header = { "#"+ ~ span }
-    list = { list_item+ }
-        list_item = { "-" ~ span }
+    unordered_list = { unordered_list_item+ }
+        unordered_list_item = { "-" ~ span ~ NEWLINE }
+    ordered_list = { ordered_list_item+ }
+        ordered_list_item = { ASCII_DIGIT+ ~ "." ~ span ~ NEWLINE }
     paragraph = { span }
-    code_block = { "```" ~ (!"```" ~ ANY) ~ "```" }
+    code_block = { "```" ~ code_block_lang ~ NEWLINE ~ code_block_inner ~ "```" }
+        code_block_lang = { ASCII_ALPHANUMERIC* }
+        code_block_inner = { (!"```" ~ ANY)* }
+    latex_block = { "$$" ~ latex_block_inner ~ "$$" }
+        latex_block_inner = { (!"$$" ~ ANY)* }
 
-span = { (style_bold | style_italic | style_code | text)+ }
+span = { (style_bold | style_italic | style_code | inline_latex | text)+ }
     style_bold = { "**" ~ text ~ "**" }
     style_italic = { "_" ~ text ~ "_" }
-    style_code = { "`" ~ text ~ "`" }
-    text = { (!(NEWLINE | "*" | "_" | "`") ~ ANY)+ }
-
-
+    style_code = { "`" ~ code_inner ~ "`" }
+        code_inner = { (!"`" ~ ANY)+ }
+    inline_latex = { "$" ~ latex_inner ~ "$" }
+        latex_inner = { (!"$" ~ ANY)+ }
+    
+    text = { (!("\n" ~ forbidden_span_wrap | "*" | "_" | "`" | "$") ~ ANY)+ }
+    forbidden_span_wrap = _{ "\n" | "-" | ASCII_DIGIT+ ~ "." }