Pascal: Add parsing of function/proc signatures #3205

eht16 · 2021-11-29T13:38:22Z

Parse the signatures and return type for functions and procedures in Pascal.

This features is ported from the Geany Pascal parser where we use this code already for many years and it might be useful for others too.

parsers/pascal.c

Units/parser-pascal.r/simple-pascal.d/args.ctags

codecov · 2021-11-29T13:51:47Z

Codecov Report

Merging #3205 (6205ac2) into master (1d9e6fe) will increase coverage by 0.10%.
The diff coverage is 97.50%.

❗ Current head 6205ac2 differs from pull request most recent head bd60584. Consider uploading reports for the commit bd60584 to get more accurate results

@@            Coverage Diff             @@
##           master    #3205      +/-   ##
==========================================
+ Coverage   84.84%   84.94%   +0.10%     
==========================================
  Files         205      206       +1     
  Lines       48369    49030     +661     
==========================================
+ Hits        41040    41650     +610     
- Misses       7329     7380      +51

Impacted Files	Coverage Δ
parsers/pascal.c	`90.24% <97.50%> (+12.64%)`	⬆️
parsers/gdscript.c	`89.82% <0.00%> (ø)`
main/entry.c	`88.37% <0.00%> (+0.04%)`	⬆️

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 1d9e6fe...bd60584. Read the comment docs.

masatake · 2021-11-29T14:36:01Z

If you are thinking about your hoidays:-), I would like you to see #2241, too.
I need helps from a person who knows pascal.

eht16 · 2021-11-29T14:57:25Z

If you are thinking about your hoidays:-), I would like you to see #2241, too. I need helps from a person who knows pascal.

Ha, nice try.
Actually, the "type" desclaration parsing is, apart from this PR, the only difference between the ctags and Geany Pascal parser. I was about to submit a PR for this too (not realizing #2241 is already there). But I noticed the current implementation is incomplete and buggy and so rather needs to be rewritten or removed. I personally do not know Pascal any good, learned it a bit in school but this was a few decades ago... :).
I don't want and can't spend time on implementing this properly. Sorry.

masatake · 2021-11-29T21:06:27Z

@eht16, I see.
After merging GDScript, I will take more time for this pull request.

masatake · 2021-12-01T13:03:32Z

parsers/pascal.c

+	}
+
+	*end = '\0';
+	*arglist = strdup(start);


My understanding:

To do *end = '\0', you have to do c = strdup(buf).
To do strdup(start), you have to do *end = '\0', terminating the C string.

I think you can reduce strdup() with eStrndup(const char* str, size_t len); declared in main/routines.h.
With it, you can make a new C string without doing *end = '\0'.

Thanks for the hint, did it and then rewrote it with your vString suggestion below :D.

masatake · 2021-12-01T13:04:31Z

parsers/pascal.c

+	*end = '\0';
+	*arglist = strdup(start);
+
+	eFree(c);


If eStrndup helps you expectedly, you can remove this eFree().

Fortunately, that code vanished by using vString.

parsers/pascal.c

masatake · 2021-12-01T13:27:57Z

parsers/pascal.c

@@ -74,6 +80,68 @@ static bool tail (const char *cp)
 	return result;
 }

+static void parseArglist(const char *buf, char **arglist, char **vartype)


Using vString() for storing arglist and vartype is an alternative approach.

In a micro view, an operation on vString tasks higher const than using char * directly.
However, in a macro view, vString is not so bad. Using char *, you must call strdup twice in every calling parseArglist.
vString can be reused once it is new'ed in findPascalTags.
A vString can clear its internal buffer keeping char * with vStringClear.
Though, it is cleared, the buffer is not deallocated.

Writing English is hard for me. Let's show me in pseudo-code:

/* Your code using raw C string */ static void parseArglist(const char *buf, char **arglist, char **vartype) { *arglist = strdup(); *vartype = strdup(); } static void findPascalTags (void) { char *arglist = NULL; char *vartype = NULL; // ... loop() { createPascalTag(); // ... free (arglist); free (vartype); parseArglist(, &arglist, &vartype); } free (arglist); free (vartype); } /* What we can do with vString */ static void parseArglist(const char *buf, vString *arglist, vString *vartype) { char c, *str; /* ... */ vStringPut(arglist, c); /* or */ vStringCatS(arglist, str); /* or */ vStringNCatS(arglist, str, len); vStringPut(vartype, c); /* or */ vStringCatS(vartype, str); /* or */ vStringNCatS(vartype, str, len); } static void findPascalTags (void) { vString *arglist = vStringNew(); vString *vartype = vStringNew(); // ... loop() { createPascalTag(); // ... vStringClear (arglist); vStringClear (vartype); parseArglist(, arglist, vartype); } vStringDelete (arglist); vStringDelete (vartype); }

Of course, I don't care about the performance of this level.
However, vString helps me understand the logic.

This comment is not for "must-fix". Your code is already works fine. So this is a just comment.

Great idea and thanks for the detailed explanation.
I rewrote the code to use vString (back in 2009 when I wrote the code initially I wasn't so much aware of the vString helper and didn't touch the since then).

It's simpler now and we don't need any more string copying than what is done implicitly by vStringCat*().

masatake · 2021-12-01T13:30:14Z

NOTE for myself: following semi-fuzz tests are passed.

make chop LANGUAGES=Pascal
make slap LANGUAGES=Pascal
make noise LANGUAGES=Pascal

eht16 · 2021-12-02T08:42:17Z

The chop, noise and slap make targets do not output anything for me and quit pretty fast. I'm not sure if they actually did test anything.

masatake · 2021-12-02T09:33:28Z

parsers/pascal.c

-			parseArglist((const char*) cp, &arglist, (kind == K_FUNCTION) ? &vartype : NULL);
+			vStringClear (arglist);
+			vStringClear (vartype);
+   			parseArglist((const char*) cp, arglist, (kind == K_FUNCTION) ? vartype : NULL);


A minor issue. The line 282 and the line 283 are not aligned.

Line 282 uses only tag characters.
However, the line 283 uses tabs and spaces mixed way.

Oops, will fix.

Another thing I noted (and probably was introduced by me in 2009), there are some inconsistencies in using a space between function name and the opening brace of the argument list. Should I fix them as well, to consistently use a space between?

Should I fix them as well, to consistently use a space between?

Thank you for the offering. If possible, could you fix them in ANOTHER commit?
In a commit, let's do one thing.

Ok, the wrong indentation is fixed.

Did you mean another commit in this PR to fix the argument list space problem or a separate PR?
I don't mind.

I also don't mind. However, you already submitted a good commit that aligns with the goal of this pull request.
So I will merge this.

masatake · 2021-12-02T18:46:39Z

Thank you.

eht16 commented Nov 29, 2021

View reviewed changes

parsers/pascal.c Outdated Show resolved Hide resolved

masatake reviewed Nov 29, 2021

View reviewed changes

Units/parser-pascal.r/simple-pascal.d/args.ctags Show resolved Hide resolved

eht16 force-pushed the pascal_add_signature_parsing branch from 98c81b0 to 8c02ce0 Compare November 29, 2021 14:26

masatake reviewed Dec 1, 2021

View reviewed changes

parsers/pascal.c Outdated Show resolved Hide resolved

masatake reviewed Dec 1, 2021

View reviewed changes

masatake reviewed Dec 2, 2021

View reviewed changes

Pascal: Add parsing of function/proc signatures

bd60584

eht16 force-pushed the pascal_add_signature_parsing branch from 6205ac2 to bd60584 Compare December 2, 2021 11:46

masatake merged commit 7e06c76 into universal-ctags:master Dec 2, 2021

eht16 mentioned this pull request Dec 2, 2021

Pascal: Consistently use spaces before opening parentheses #3213

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Pascal: Add parsing of function/proc signatures #3205

Pascal: Add parsing of function/proc signatures #3205

eht16 commented Nov 29, 2021

codecov bot commented Nov 29, 2021 •

edited

Loading

masatake commented Nov 29, 2021

eht16 commented Nov 29, 2021

masatake commented Nov 29, 2021

masatake Dec 1, 2021

eht16 Dec 2, 2021

masatake Dec 1, 2021

eht16 Dec 2, 2021

masatake Dec 1, 2021

eht16 Dec 2, 2021

masatake commented Dec 1, 2021

eht16 commented Dec 2, 2021

masatake Dec 2, 2021

masatake Dec 2, 2021

eht16 Dec 2, 2021

masatake Dec 2, 2021

eht16 Dec 2, 2021

masatake Dec 2, 2021

masatake commented Dec 2, 2021

Pascal: Add parsing of function/proc signatures #3205

Pascal: Add parsing of function/proc signatures #3205

Conversation

eht16 commented Nov 29, 2021

codecov bot commented Nov 29, 2021 • edited Loading

Codecov Report

masatake commented Nov 29, 2021

eht16 commented Nov 29, 2021

masatake commented Nov 29, 2021

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

masatake commented Dec 1, 2021

eht16 commented Dec 2, 2021

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

masatake commented Dec 2, 2021

codecov bot commented Nov 29, 2021 •

edited

Loading