redrafted the design

This commit is contained in:
anon 2023-09-18 21:33:12 +02:00
parent d743b4a4c8
commit 349b5bfa4f

View File

@ -1,30 +1,20 @@
# hl
General purpose highlighter.
# libhl
// it would be lovely to have a different name the "library" part and the cli
## API
int hl_init(void);
int hl_deinit(void);
These functions are responsible for the library's "life time".
`hl_init()` must be called before any other library function.
`hl_deinit()` will ensure all occupied memory is freed.
# Usage
hl will read from stdin and write to stdout.
hl < source/main.c
### Cli Options
-h : display help message
-F <dir> : syntax file look up directory
-s <syntax> : specify syntax to load
### Environment variables
HL_HOME : default directory to load syntax files from
# API
void render_string(const char * const string, const char * const mode);
This function matches _string_ against all known highlighting rules and dispatches the appropriate callback defending on mode.
This function matches _string_ against all known highlighting rules and dispatches the appropriate callback depending on _mode_.
typedef void (*attribute_callback_t)(const char * const string, const int length, void * const attributes);
The type used for defining appropriate callbacks for render_string().
string - string to be outputed
length - number of characters that matched a highlighting rule;
0 if rule passed, in such a case the user is expected still want 1 character outputed
attributes - arbitrary data associated with the matched rule; intended to hold color/font information for example
+ string - string to be outputed
+ length - number of characters that matched a highlighting rule
+ attributes - arbitrary data associated with the matched rule; intended to hold color/font information for example
typedef struct {
char * key;
@ -43,30 +33,49 @@ This is how you append a display mode that render_string() will search based on
} token_type_t;
These are the valid type of distinct token types.
KEYSYMBOL - a string which is contextless, the surounding text is ignored
"mysymbol" will match inside all of these:
"something mysymbol something"
"somethingmysymbolsomething"
it is intended to match such thing as programming language operators,
so both "var a = 'a'" and "var a='a'" are recognized
KEYWORD - a string which is recognized when surounded by word bundaries such as ' ' or '\t'
MATCH - a Vim style regular expression to be recognized
REGION - a Vim style regular expression where the starting and ending patters are to be distinguished from the contents
+ KEYSYMBOL - a string which is contextless, the surounding text is ignored
"mysymbol" will match inside all of these:
"something mysymbol something"
"somethingmysymbolsomething"
it is intended to match such thing as programming language operators
+ KEYWORD - a string which is recognized when surounded by word bundaries such as ' ' or '\t'
+ MATCH - a regular expression to be recognized
+ REGION - a regular expression where the starting and ending patters are to be distinguished from the contents
The universal way to add a new pattern to be recognized is with:
token * new_token(const char * const syntax, const token_type_t t, const hl_group_t * const g);
This wraps one of the following:
// ?!
There are also convinience functions:
// NOTE: the return value is the number tokens successfully inserted
int new_keyword_tokens(const char * const * words, hl_group_t * const g);
int new_keyword_tokens(const char * const * words, hl_group_t * const g); // _words_ must be NULL terminated
int new_syntax_character_tokens(const char * const chars, hl_group_t * const g);
The regex engine used for MATCHes is Jeger by default, emulating Vim regex.
However the regex engine can be overridden:
// ?!
---
#hl
General purpose highlighter (and demo program for libhl).
## Usage
hl will read from stdin and write to stdout.
hl < source/main.c
### Cli Options
-h : display help message
-F <dir> : syntax file look up directory
-s <syntax> : specify syntax to load
### Environment variables
HL_HOME : default directory to load syntax files from
---
# Scripting
hl can parse a small subset of VimScript: the few instructions related to highlighing, and it ignores everything else.
All Vim highlighing scripts should be valid hl scripts.