WIP: Lax parser

[kasbah]

No description provided.

[kasbah]

So I found out I can create and UNKNOWN lexer rule that should match everything that hasn't been previously defined.

ignored: UNKNOWN* EOF;
 
UNKNOWN: .+?;

And that kind of works:

$ node bin/electro-grammar.js "akjdkadj 10 asdjkdj  ohm xaksjdkjd"
line 1:0 mismatched input 'a' expecting NUMBER
line 1:9 extraneous input '10' expecting {<EOF>, UNKNOWN}
line 1:21 extraneous input 'ohm' expecting {<EOF>, UNKNOWN}
{ component: { resistance: 10, type: 'resistor' },
  ignored: 'akjdkadj  asdjkdj   xaksjdkjd' }

I simplified the grammar to just work on resistors for the time being, trying to figure out why it ignores the k in 10k.

$ node bin/electro-grammar.js "10k xaksjdkjd"
line 1:2 mismatched input 'k' expecting {OHM, RPREFIX}
line 1:0 extraneous input '10' expecting {<EOF>, UNKNOWN}
{ component: { resistance: 10, type: 'resistor' },
  ignored: 'k xaksjdkjd' }

[kasbah]

We probably should take advantage of mode(M):

mode (M)
After matching this token, switch the lexer to mode M . The next
time the lexer tries to match a token, it will look only at rules in mode M .
M can be a mode name from the same grammar or an integer literal. See
grammar Strings earlier.

[dvc94ch]

That looks like a lexer ambiguity. Try this on the java backend: echo "1k abc" | grun ElectroGrammar resistor -diagnostics -tree -tokens to find out what token k is (since it's not r-prefix).

[kasbah]

Hmm, it seems to match everything to UNKNOWN so it looks like I was mistaken about how this would work. I wonder how come it kinda half works at all.

[kasbah]

SO answers (1 and 2) seem to suggest that the last lexer rule will be the lowest priority. But that's not what I am seeing.

EDIT: Seems the imports re-order things or otherwise mess up the priority. :/ Tracking here: antlr/antlr4#2209