I can load one file and traverse it with babel, it goes something like this:
var babylon = require("babylon");
let contents = fs.readFileSync("example.js","utf-8");
let ast = babylon.parse(contents);
Now the question is, how can I get the AST (Abstract Syntax Tree) if I have multiple files in my program.
main.js
export const getFoo(){
return "a"
}
example.js
import {getFoo} from './main'
let bar = getFoo() + "baz";
Obviously I would like to see the function declaration and the function call expression into the same AST, but also at the same time getting the line numbers and columns (node.loc) information to also show the specific file.
You can concatenate the AST from several files if you know their paths and can load them.
import {parse} from '#babel/parser';
const a = 'var a = 1;'; // or fs.readFileSync("a.js","utf-8");
const b = 'var b = 2;'; // or fs.readFileSync("b.js","utf-8");
const astA = parse(a, { sourceFilename: 'a.js' });
const astB = parse(b, { sourceFilename: 'b.js' });
const ast = {
type: 'Program',
body: [].concat(astA.program.body, astB.program.body)
};
Source example
But I can't find out how to get AST from several files without loading them directly. I tried to write a babel plugin to analyze code from an imported file and I haven't realized how to do that.
I am developing an extension for visual studio code using language server protocol, and I am including the support for "Go to symbol in workspace". My problem is that I don't know how to select the matches...
Actually I use this function I wrote:
function IsInside(word1, word2)
{
var ret = "";
var i1 = 0;
var lenMatch =0, maxLenMatch = 0, minLenMatch = word1.length;
for(var i2=0;i2<word2.length;i2++)
{
if(word1[i1]==word2[i2])
{
lenMatch++;
if(lenMatch>maxLenMatch) maxLenMatch = lenMatch;
ret+=word1[i1];
i1++;
if(i1==word1.length)
{
if(lenMatch<minLenMatch) minLenMatch = lenMatch;
// Trying to filter like VSCode does.
return maxLenMatch>=word1.length/2 && minLenMatch>=2? ret : undefined;
}
} else
{
ret+="Z";
if(lenMatch>0 && lenMatch<minLenMatch)
minLenMatch = lenMatch;
lenMatch=0;
}
}
return undefined;
}
That return the sortText if the word1 is inside the word2, undefined otherwise. My problem are cases like this:
My algorithm see that 'aller' is inside CallServer, but the interface does not mark it like expected.
There is a library or something that I must use for this? the code of VSCode is big and complex and I don't know where start looking for this information...
VSCode's API docs for provideWorkspaceSymbols() provide the following guidance (which I don't think your example violates):
The query-parameter should be interpreted in a relaxed way as the editor will apply its own highlighting and scoring on the results. A good rule of thumb is to match case-insensitive and to simply check that the characters of query appear in their order in a candidate symbol. Don't use prefix, substring, or similar strict matching.
These docs were added in response to this discussion, where somebody had very much the same issue as you.
Having a brief look at VSCode sources, internally it seems to use filters.matchFuzzy2() for the highlighting (see here and here). I don't think it's exposed in the API, so you would probably have to copy it if you wanted the behavior to match exactly.
I'm trying to take something like var a = 5; and transpile it to something like thing.a = 5.
Using this code below in my visitor, it tells me unexpected token .
VariableDeclarator: {
enter: function (path, state) {
path.replaceWith(
t.assignmentExpression(
'=',
t.memberExpression(
t.identifier('abc'),
t.identifier('def')
),
t.stringLiteral('xyz')
)
)
}
}
What am I not taking into account here?
What's the canonical way to accomplish this?
Turns out I was operating on a Declarator rather than a Declaration. So what I was doing was causing it to compile to something like var a.4 = 'def'. Naturally, that fails.
"abc" is read-only while parsing file:
export const abc = 20;
abc = 30; // Reassigning the value
Will this fail across all browsers?
If the browsers are on 'use strict mode' declaring a const will work but the export will fail.
Just found out the answer. Browsers do throw error if we try to change the value of a const and they don't allow to rewrite. But safari doesn't handle this, it doesn't throws any error.
I wish to do the following in Specman:
my_task() is {
var my_var : int;
my_var = 5;
message(LOW,appendf("%s=[%d]",my_var.to_name(),my_var));
};
Currently, I'm in search of the internal task to_name(). I do not want to create a struct for this. I wish to only use Specman internals.
I'm not sure how you'd do this unless you somehow have the collection of all the fields, which would give you the name of all the fields.
So, I'm going to branch predict and assume that you want all the fields of a given struct/unit:
extend sys {
A : list of uint;
B : int;
cee : string;
run() is also {
var rf_sys: rf_struct = rf_manager.get_exact_subtype_of_instance(sys);
for each (field) in rf_sys.get_declared_fields() {
print field;
print field.get_long_name(); // <-- Here's your "get_name()" function
};
};
};
On version 8.2, this yields:
Usage: . env.sh [-32bit|-64bit] [-v] [[VAR=value]...]
Welcome to Specman Elite(64) (09.20.482-d) - Linked on Wed Mar 2 13:32:19
2011
Protected by U.S. Patents 6,141,630 ;6,182,258; 6,219,809; 6,347,388;
6,487,704; 6,499,132; 6,502,232; 6,519,727; 6,530,054; 6,675,138; 6,684,359;
6,687,662; 6,907,599; 6,918,076; 6,920,583; Other Patents Pending.
1 notification was modified by command 'set notify -severity=WARNING
DEPR_START_TCM_ARG_BY_REF'
Checking license ... OK
Loading /nfs/pdx/home/rbroger1/tmp.e ...
read...parse...update...patch...h code...code...clean...GC(sys)...
Doing setup ...
Generating the test using seed 1...
Starting the test ...
Running the test ...
field = rf_field 'time', Specman's private modules
field.get_long_name() = "time"
field = rf_field 'logger', Specman's private modules
field.get_long_name() = "logger"
field = rf_field 'A', line 5 in #tmp
field.get_long_name() = "A"
field = rf_field 'B', line 6 in #tmp
field.get_long_name() = "B"
field = rf_field 'cee', line 7 in #tmp
field.get_long_name() = "cee"
No actual running requested.
Checking the test ...
Checking is complete - 0 DUT errors, 0 DUT warnings.
If that doesn't quite answer your question, look more into Specman's introspection or reflection interface in the documentation for your version of Specman. Warning, details are a bit scarce from Cadence. Also, see my answer to "Specman: how to retrieve values of var which is stored in another var".. Finally, in specview you can use the data browser to browse the rf_manger itself ( introspection at its best). Then you can find all the functions that Cadence doesn't tell you about in their documentation.
You can't get the string name of a variable. Although you can get the name of a field using get_name().
This makes some sense, because the variable is only known at the location it is declared. You can't even access the variable in a later extension of my_task(). So if you are already editing the code at the location where the variable was declared, just say "my_var" rather than my_var.to_name(). (Yes, typos and cut and paste errors can happen.)
I think this is what you are looking for:
define <dump'action> "dump <exp>" as {
out("<exp>","=[",<exp>,"]");
};
you can use this macro in the specman command line or inside functions, for example:
foo() is {
var a : int = 5;
dump a;
};
will give:
a=[5]