ed/metadesk
1
0
Fork 0
mirror of https://github.com/Ed94/metadesk.git synced 2026-08-06 23:58:48 +00:00
Code Issues Packages Projects Releases Wiki Activity

[grammar test] Remove malformed tagging. Give up on comments.

Browse Source
This commit is contained in:
Miguel Lechon
2021-03-08 20:02:27 +01:00
parent bf9bfae929
commit 2d22fc17d5
2 changed files with 32 additions and 56 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tests/grammar.c
-16
View File
@@ -147,7 +147,6 @@ enum OperationFlags
OperationFlag_Fill = 1<<0, OperationFlag_Fill = 1<<0,
OperationFlag_Markup = 1<<1, OperationFlag_Markup = 1<<1,
OperationFlag_Tag = 1<<2, OperationFlag_Tag = 1<<2,
OperationFlag_PreComment = 1<<3,
}; };
static void Extend(MD_String8 *s, char c) static void Extend(MD_String8 *s, char c)
@@ -161,7 +160,6 @@ static void ExpandProduction(MD_Node *production, MD_String8List *out, MD_Node *
static void ExpandRule(MD_Node *rule, MD_String8List *out_strings, MD_Node *cur_node, OperationFlags op_flags, static void ExpandRule(MD_Node *rule, MD_String8List *out_strings, MD_Node *cur_node, OperationFlags op_flags,
MD_u32 max_depth, MD_u32 depth) MD_u32 max_depth, MD_u32 depth)
{ {
MD_String8 pre_comment = {0};
for(MD_EachNode(rule_element, rule->first_child)) for(MD_EachNode(rule_element, rule->first_child))
{ {
MD_b32 expand = 1; MD_b32 expand = 1;
@@ -181,8 +179,6 @@ static void ExpandRule(MD_Node *rule, MD_String8List *out_strings, MD_Node *cur_
if(expand) if(expand)
{ {
MD_Node *node_to_tag = 0; MD_Node *node_to_tag = 0;
MD_Node *node_to_precomment = 0;
MD_String8 precomment = {0};
OperationFlags old_op_flags = op_flags; OperationFlags old_op_flags = op_flags;
for(MD_EachNode(tag_node, rule_element->first_tag)){ for(MD_EachNode(tag_node, rule_element->first_tag)){
if(MD_StringMatch(tag_node->string, MD_S8Lit("child"), 0)) if(MD_StringMatch(tag_node->string, MD_S8Lit("child"), 0))
@@ -209,12 +205,6 @@ static void ExpandRule(MD_Node *rule, MD_String8List *out_strings, MD_Node *cur_
{ {
op_flags |= OperationFlag_Markup; op_flags |= OperationFlag_Markup;
} }
else if(MD_StringMatch(tag_node->string, MD_S8Lit("pre_comment"), 0))
{
op_flags |= OperationFlag_PreComment;
node_to_precomment = cur_node;
cur_node = NewChild(0); // NOTE(mal): Only used to store the comment, won't be linked as a node
}
else if(MD_StringMatch(tag_node->string, MD_S8Lit(OPTIONAL_TAG), 0)) else if(MD_StringMatch(tag_node->string, MD_S8Lit(OPTIONAL_TAG), 0))
{ {
} }
@@ -274,12 +264,6 @@ static void ExpandRule(MD_Node *rule, MD_String8List *out_strings, MD_Node *cur_
cur_node = node_to_tag; cur_node = node_to_tag;
} }
if(node_to_precomment)
{
node_to_precomment->comment_before = cur_node->string;
cur_node = node_to_precomment;
}
op_flags = old_op_flags; op_flags = old_op_flags;
} }
} }
tests/grammar.md
+32 -40
View File
@@ -36,6 +36,11 @@ string_literal_items : string_literal_item [string_literal_items]
string_literal_item : ascii_no_backslash_no_quotes | '\'' | '\\' ascii string_literal_item : ascii_no_backslash_no_quotes | '\'' | '\\' ascii
symbol_label : '~'|'!'|'%'|'^'|'&'|'*'|'+'|'-'|'/'|'|'|'<'|'>'|'$'|'='|'.'|'?'|'$' symbol_label : '~'|'!'|'%'|'^'|'&'|'*'|'+'|'-'|'/'|'|'|'<'|'>'|'$'|'='|'.'|'?'|'$'
/* What follows is a range of annotated grammars that can be used to generate
* tests of increasing complexity and completeness to check against MetaDesk.
* To run them, uncomment them one by one and run the build/grammar test.
*/
//// Arbitrarily deep tree, possibly empty //// Arbitrarily deep tree, possibly empty
// file : [@child set_list] // file : [@child set_list]
// set_list : '{' [@child set_list] '}' [separator @sibling set_list] // set_list : '{' [@child set_list] '}' [separator @sibling set_list]
@@ -60,18 +65,12 @@ symbol_label : '~'|'!'|'%'|'^'|'&'|'*'|'+'|'-'|'/'|'|'|'<'|'>
//// Tags //// Tags
// file : [@child set_list] // file : [@child set_list]
// set_list : scoped_set [separator @sibling set_list] | unscoped_set [unscoped_separator @sibling set_list] // set_list : scoped_set [separator @sibling set_list] | unscoped_set [unscoped_separator @sibling set_list]
// scoped_set : {[tag_list] '{' [@child set_list] '}'} // scoped_set : {[tag_list] untagged_scoped_set}
// untagged_scoped_set : '{' [@child set_list] '}'
// unscoped_set : {[tag_list] @fill 'A' [unscoped_set_tail]} // unscoped_set : {[tag_list] @fill 'A' [unscoped_set_tail]}
// tag_list : '@' @tag tag ' ' [tag_list] // tag_list : '@' @tag tag ' ' [tag_list]
// tag : @fill 'T'['(' [@child set_list] ')'] // tag : @fill 'T'['(' [@child set_list] ')']
// // unscoped_set_tail : ':' @child unscoped_set | ' ' @sibling unscoped_set | ':' @child scoped_set | ' ' @sibling scoped_set // unscoped_set_tail : {':' @child unscoped_set | ' ' @sibling unscoped_set | ':' untagged_scoped_set | ' ' @sibling scoped_set}
// unscoped_set_tail : {':' @child unscoped_set | ' ' @sibling unscoped_set | set_tail_hack | ' ' @sibling scoped_set}
// // NOTE(mal): Ideally the set_tail_hack should be captured by the simpler
// // ':' @child scoped_set
// // but there's a quirk in the grammar.
// // In "A:{}", A has no children but in "A:@T{}", it has one tagged children,
// // which means that *the presence of tags* introduces the child
// set_tail_hack : ':' [@child tag_list] '{' '}' | ':' @child [tag_list] '{' @child set_list '}'
//// Alternative scope markers //// Alternative scope markers
// file : [@child set_list] // file : [@child set_list]
@@ -81,12 +80,7 @@ symbol_label : '~'|'!'|'%'|'^'|'&'|'*'|'+'|'-'|'/'|'|'|'<'|'>
// unscoped_set : {[tag_list] @fill 'A' [unscoped_set_tail]} // unscoped_set : {[tag_list] @fill 'A' [unscoped_set_tail]}
// tag_list : '@' @tag tag ' ' [tag_list] // tag_list : '@' @tag tag ' ' [tag_list]
// tag : @fill 'T'['(' [@child set_list] ')'] // tag : @fill 'T'['(' [@child set_list] ')']
// unscoped_set_tail : {':' @child unscoped_set | ' ' @sibling unscoped_set | set_tail_hack | ' ' @sibling scoped_set} // unscoped_set_tail : {':' @child unscoped_set | ' ' @sibling unscoped_set | ':' untagged_scoped_set | ' ' @sibling scoped_set}
// set_tail_hack : {':' [@child tag_list] '{' '}' |
// ':' @child [tag_list] '{' @child set_list '}' |
// ':' [@child tag_list] alt_scope_beg alt_scope_end |
// ':' @child [tag_list] alt_scope_beg @child set_list alt_scope_end
// }
//// General tags and labels //// General tags and labels
file : [@child set_list] file : [@child set_list]
@@ -96,29 +90,27 @@ untagged_scoped_set : '{' [@child set_list] '}' | alt_scope_beg [@child set_list
unscoped_set : {[tag_list] @fill label [unscoped_set_tail]} unscoped_set : {[tag_list] @fill label [unscoped_set_tail]}
tag_list : '@' @tag tag ' ' [tag_list] tag_list : '@' @tag tag ' ' [tag_list]
tag : @fill id['(' [@child set_list] ')'] tag : @fill id['(' [@child set_list] ')']
unscoped_set_tail : {':' @child unscoped_set | ' ' @sibling unscoped_set | set_tail_hack | ' ' @sibling scoped_set} unscoped_set_tail : {':' @child unscoped_set | ' ' @sibling unscoped_set | ':' untagged_scoped_set | ' ' @sibling scoped_set}
set_tail_hack : {':' [@child tag_list] '{' '}' |
':' @child [tag_list] '{' @child set_list '}' |
':' [@child tag_list] alt_scope_beg alt_scope_end |
':' @child [tag_list] alt_scope_beg @child set_list alt_scope_end
}
//// Comments before /* Comments
//// TODO: This needs some work to make sure that "a /* comment_before_b */ b" can't be generated * Comments around nodes are accessible to the user. Here's how they behave:
//// but that "a /* comment_after_a */ b" can * - The text inside a comment immediatly following a node is stored as the
// file : [@child set_list] * comment_after member of that node. No newlines can happen between a
// set_list : scoped_set [separator @sibling set_list] | unscoped_set [unscoped_separator @sibling set_list] * node and its after_comment.
// scoped_set : {[tag_list] [@pre_comment pre_comment] untagged_scoped_set} * - The text inside a comment preceding a node is stored as the
// pre_comment : '/' '/' [' '] [@fill c_code_content] '\n' | '/' '*' [@fill cpp_code_content] '*' '/' * comment_before of that node _unless_ it is already the comment_after
// c_code_content : 'c''o''m''m''e''n''t' // TODO: Arbitrary strings, including C-style comments, as long as /* */ pairs are balanced * of another node. One newline between the comment and the node is
// cpp_code_content : 'c''o''m''m''e''n''t' // TODO: Arbitrary strings that don't start with space * obviously necessary in the case of C++-style comments and it's also
// untagged_scoped_set : '{' [@child set_list] '}' | alt_scope_beg [@child set_list] alt_scope_end * allowed for C-style comments.
// unscoped_set : {[tag_list] @fill label [unscoped_set_tail]} * - If the first character inside a C++-style comment is a space, it's
// tag_list : '@' @tag tag ' ' [tag_list] * omitted from the stored string
// tag : @fill id['(' [@child set_list] ')'] *
// unscoped_set_tail : {':' @child unscoped_set | ' ' @sibling unscoped_set | set_tail_hack | ' ' @sibling scoped_set} * The semantically annotated Backus-Naur form that we're using is not a
// set_tail_hack : {':' [@child tag_list] '{' '}' | * good fit to describe the grammar of comments.
// ':' @child [tag_list] '{' @child set_list '}' | * To prevent the comment in "a /* comment */ b" from being interpreted as
// ':' [@child tag_list] alt_scope_beg alt_scope_end | * a comment_before of "b", instead of what it is (a comment_after of "a"),
// ':' @child [tag_list] alt_scope_beg @child set_list alt_scope_end * we would have to complicate the grammar by introducing several extra
// } * productions with this specific purpose in mind.
* The sensitivity of whitespace in the attachment of comments to nodes is
* also cumbersome to express in BNF.
*/