Readline-based parser of markdown #3233
Conversation
This parser is based on the asciidoc parser and tries to preserve all features of the regex-based parser (all kinds, full scope, sectionMarker field, running subparsers for code).
Codecov Report
@@ Coverage Diff @@
## master #3233 +/- ##
==========================================
+ Coverage 85.01% 85.28% +0.27%
==========================================
Files 206 206
Lines 49127 49084 -43
==========================================
+ Hits 41765 41862 +97
+ Misses 7362 7222 -140
Continue to review full report at Codecov.
|
|
Thank you. Of course, I will merge this C implementation. Let's fill the diff --git a/parsers/markdown.c b/parsers/markdown.c
index 173130e06..89358d89a 100644
--- a/parsers/markdown.c
+++ b/parsers/markdown.c
@@ -69,17 +69,22 @@ static NestingLevels *nestingLevels = NULL;
* FUNCTION DEFINITIONS
*/
-static NestingLevel *getNestingLevel(const int kind)
+static NestingLevel *getNestingLevel(const int kind, int adjustment_when_pop)
{
NestingLevel *nl;
tagEntryInfo *e;
+ unsigned long line = getInputLineNumber();
while (1)
{
nl = nestingLevelsGetCurrent(nestingLevels);
e = getEntryOfNestingLevel (nl);
if ((nl && (e == NULL)) || (e && (e->kindIndex >= kind)))
+ {
+ if (e && line > adjustment_when_pop)
+ e->extensionFields.endLine = line - adjustment_when_pop;
nestingLevelsPop(nestingLevels);
+ }
else
break;
}
@@ -88,7 +93,7 @@ static NestingLevel *getNestingLevel(const int kind)
static int makeMarkdownTag (const vString* const name, const int kind, const bool two_line)
{
- const NestingLevel *const nl = getNestingLevel(kind);
+ const NestingLevel *const nl = getNestingLevel(kind, two_line? 2: 1);
int r = CORK_NIL;
if (vStringLength (name) > 0)Could you include this? |
|
Instead of nestingLevelsNew, let's use nestingLevelsNewFull. We can add a callback function called when a level is popped. We can fill the deleteBlockData is an example of such a callback function. I will work on this topic tonight, JST. |
|
nesting level API must be extended to pass the line adjustment data. So we can pass |
|
@techee, let me take over this pull request. |
Sure, no problem, less work for me :-). |
|
What's the |
|
$ cat -n /tmp/foo.c
1 struct point
2 {
3 int x;
4 int y;
5 };
6
7 int
8 main(void)
9 {
10 return 0;
11 }
12 $ ctags -o - --fields=+ne /tmp/foo.c
main /tmp/foo.c /^main(void)$/;" f line:8 typeref:typename:int end:11
point /tmp/foo.c /^struct point $/;" s line:1 file: end:5
x /tmp/foo.c /^ int x;$/;" m line:3 struct:point typeref:typename:int file: end:3
y /tmp/foo.c /^ int y;$/;" m line:4 struct:point typeref:typename:int file: end:4 |
|
Nice, I can imagine this information could be interesting for Geany too. I assume this is currently available only for some parsers, not all parsers reporting scope, right? |
No, not all parsers. `grep endLine parsers/*.c' may report what you want to know:-). |
|
See #3235. |
|
|
||
| static int makeMarkdownTag (const vString* const name, const int kind, const bool two_line) | ||
| { | ||
| const NestingLevel *const nl = getNestingLevel(kind); |
There was a problem hiding this comment.
In #3235, I move the getNestingLevel() to...
| int r = CORK_NIL; | ||
|
|
||
| if (vStringLength (name) > 0) | ||
| { |
There was a problem hiding this comment.
...here. The nesting level should be popped when making a tag for the name.
|
The changes were merged via #3236. |
We have a rather bad markdown parser in Geany and when looking at uctags, I realized there's just a regex-based parser. From the past experience, these tend to be rather slow so for us it would be better to have a hand-written parser.
I created a simple readline-based parser based on the asciidoc parser and tried to preserve all the features of the regex-based parser (all kinds, full scope, sectionMarker field, running subparsers for code). Would such a parser be interesting for uctags or is the regex-based one the preferred solution?