Merge pull request #3677 from masatake/quarto-parser

Quarto: new parser

Tmain/list-subparsers-all.d/stdout-expected.txt

@@ -16,6 +16,7 @@ OpenAPI              Yaml       base <> sub {bidirectional}
 PlistXML             XML        base <> sub {bidirectional}
 PythonLoggingConfig  Iniconf    base <> sub {bidirectional}
 QtMoc                C++        base <> sub {bidirectional}
+Quarto               Markdown   base <= sub {dedicated}
 R6Class              R          base <> sub {bidirectional}
 RMarkdown            Markdown   base <= sub {dedicated}
 RSpec                Ruby       base => sub {shared}

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

@@ -0,0 +1,3 @@
+--sort=no
+--extras=+{guest}
+--fields=+{end}{line}{language}

Units/parser-quarto.r/simple.d/expected.tags

@@ -0,0 +1,2 @@
+defun	input.qmd	/^#| label: defun$/;"	l	line:2	language:Quarto	end:6
+f	input.qmd	/^def f():$/;"	f	line:4	language:Python	end:5

Units/parser-quarto.r/simple.d/input.qmd

@@ -0,0 +1,7 @@
+```{python}
+#| label: defun
+#| fig-cap: "Define a function"
+def f():
+    pass
+```
+

Units/parser-quarto.r/unexecuted-block.d/args.ctags

@@ -0,0 +1,3 @@
+--sort=no
+--extras=+{guest}
+--fields=+{end}{line}{language}

Units/parser-quarto.r/unexecuted-block.d/expected.tags

@@ -0,0 +1,4 @@
+__anon53b260bc0100	input.qmd	/^```{python}$/;"	l	line:1	language:Quarto	end:4
+__anon53b260bc0200	input.qmd	/^```{{python}}$/;"	l	line:6	language:Quarto	end:9
+f	input.qmd	/^def f():$/;"	f	line:2	language:Python	end:3
+g	input.qmd	/^def g():$/;"	f	line:7	language:Python	end:8

Units/parser-quarto.r/unexecuted-block.d/input.qmd

@@ -0,0 +1,9 @@
+```{python}
+def f():
+    pass
+```
+
+```{{python}}
+def g():
+    pass
+```

Units/parser-rmarkdown.r/frontmatter.d/expected.tags

@@ -1,8 +1,8 @@
 S1	input.rmd	/^# S1$/;"	chapter	line:4	language:Markdown	end:17
-xyX	input.rmd	/^```{r xyX}$/;"	chunklabel	line:6	language:RMarkdown	extras:subparser
+xyX	input.rmd	/^```{r xyX}$/;"	chunklabel	line:6	language:RMarkdown	extras:subparser	end:16
 S2	input.rmd	/^# S2$/;"	chapter	line:18	language:Markdown	end:28
-__anon9d4e183a0100	input.rmd	/^```{r, cache = TRUE, dependson = "xyX"}$/;"	chunklabel	line:20	language:RMarkdown	extras:subparser,anonymous
-__anon9d4e183a0200	input.rmd	/^```{python}$/;"	chunklabel	line:24	language:RMarkdown	extras:subparser,anonymous
+__anon9d4e183a0100	input.rmd	/^```{r, cache = TRUE, dependson = "xyX"}$/;"	chunklabel	line:20	language:RMarkdown	extras:subparser,anonymous	end:22
+__anon9d4e183a0200	input.rmd	/^```{python}$/;"	chunklabel	line:24	language:RMarkdown	extras:subparser,anonymous	end:28
 S3	input.rmd	/^# S3$/;"	chapter	line:29	language:Markdown	end:30
 x	input.rmd	/^x <- 1$/;"	globalVar	line:8	language:R	extras:guest	end:8
 foo	input.rmd	/^foo <- function () {$/;"	function	line:9	language:R	extras:guest	end:12

Units/parser-rmarkdown.r/simple-rmarkdown.d/expected.tags

@@ -1,8 +1,8 @@
 S1	input.rmd	/^# S1$/;"	chapter	line:1	language:Markdown	end:14
-xyX	input.rmd	/^```{r xyX}$/;"	chunklabel	line:3	language:RMarkdown	extras:subparser
+xyX	input.rmd	/^```{r xyX}$/;"	chunklabel	line:3	language:RMarkdown	extras:subparser	end:13
 S2	input.rmd	/^# S2$/;"	chapter	line:15	language:Markdown	end:25
-__anon4a45a9700100	input.rmd	/^```{r, cache = TRUE, dependson = "xyX"}$/;"	chunklabel	line:17	language:RMarkdown	extras:subparser,anonymous
-__anon4a45a9700200	input.rmd	/^```{python}$/;"	chunklabel	line:21	language:RMarkdown	extras:subparser,anonymous
+__anon4a45a9700100	input.rmd	/^```{r, cache = TRUE, dependson = "xyX"}$/;"	chunklabel	line:17	language:RMarkdown	extras:subparser,anonymous	end:19
+__anon4a45a9700200	input.rmd	/^```{python}$/;"	chunklabel	line:21	language:RMarkdown	extras:subparser,anonymous	end:25
 S3	input.rmd	/^# S3$/;"	chapter	line:26	language:Markdown	end:27
 x	input.rmd	/^x <- 1$/;"	globalVar	line:5	language:R	extras:guest	end:5
 foo	input.rmd	/^foo <- function () {$/;"	function	line:6	language:R	extras:guest	end:9

docs/news.rst

@@ -454,6 +454,7 @@ The following parsers have been added:
 * PythonLoggingConfig
 * QemuHX *optlib*
 * QtMoc
+* Quarto *Markdown based subparser*
 * R
 * R6Class *R based subparser*
 * Rake *Ruby based subparser*

main/parsers_p.h

@@ -148,6 +148,7 @@
 	PythonLoggingConfigParser, \
 	QemuHXParser, \
 	QtMocParser, \
+	QuartoParser, \
 	RMarkdownParser, \
 	RParser, \
 	RakeParser, \

parsers/markdown.c

@@ -212,26 +212,53 @@ static void getFootnoteMaybe (const char *line)
 	vStringDelete (footnote);
 }
 
-static bool extractLanguageForCodeBlock (const char *langMarker,
-										 vString *codeLang)
+static markdownSubparser * extractLanguageForCodeBlock (const char *langMarker,
+														vString *codeLang)
 {
-	subparser *s;
-	bool r = false;
+	subparser *s = NULL;
+	bool b = false;
+	markdownSubparser *r = NULL;
 
 	foreachSubparser (s, false)
 	{
 		markdownSubparser *m = (markdownSubparser *)s;
 		enterSubparser(s);
 		if (m->extractLanguageForCodeBlock)
-			r = m->extractLanguageForCodeBlock (m, langMarker, codeLang);
+			b = m->extractLanguageForCodeBlock (m, langMarker, codeLang);
 		leaveSubparser();
-		if (r)
+		if (b)
+		{
+			r = m;
 			break;
+		}
 	}
 
 	return r;
 }
 
+static void notifyCodeBlockLine (markdownSubparser *m, const unsigned char *line)
+{
+	subparser *s = (subparser *)m;
+	if (m->notifyCodeBlockLine)
+	{
+		enterSubparser(s);
+		m->notifyCodeBlockLine (m, line);
+		leaveSubparser();
+	}
+}
+
+static void notifyEndOfCodeBlock (markdownSubparser *m)
+{
+	subparser *s = (subparser *)m;
+
+	if (m->notifyEndOfCodeBlock)
+	{
+		enterSubparser(s);
+		m->notifyEndOfCodeBlock (m);
+		leaveSubparser();
+	}
+}
+
 static void findMarkdownTags (void)
 {
 	vString *prevLine = vStringNew ();
@@ -249,6 +276,7 @@ static void findMarkdownTags (void)
 
 	nestingLevels = nestingLevelsNewFull (0, fillEndField);
 
+	markdownSubparser *marksub = NULL;
 	while ((line = readLineFromInputFile ()) != NULL)
 	{
 		int lineLen = strlen ((const char*) line);
@@ -296,7 +324,8 @@ static void findMarkdownTags (void)
 					startLineNumber = startSourceLineNumber = lineNum + 1;
 
 					vStringClear (codeLang);
-					if (! extractLanguageForCodeBlock (langMarker, codeLang))
+					marksub = extractLanguageForCodeBlock (langMarker, codeLang);
+					if (! marksub)
 					{
 						vStringCopyS (codeLang, langMarker);
 						vStringStripLeading (codeLang);
@@ -309,7 +338,12 @@ static void findMarkdownTags (void)
 					if (vStringLength (codeLang) > 0
 						&& startLineNumber < endLineNumber)
 						makePromise (vStringValue (codeLang), startLineNumber, 0,
-							endLineNumber, 0, startSourceLineNumber);
+									 endLineNumber, 0, startSourceLineNumber);
+					if (marksub)
+					{
+						notifyEndOfCodeBlock(marksub);
+						marksub = NULL;
+					}
 				}
 
 				lineProcessed = true;
@@ -330,6 +364,9 @@ static void findMarkdownTags (void)
 			lineProcessed = true;
 		}
 
+		if (marksub)
+			notifyCodeBlockLine (marksub, line);
+
 		/* code block or comment */
 		if (inCodeChar || inComment)
 			lineProcessed = true;

parsers/markdown.h

@@ -21,9 +21,46 @@ typedef struct sMarkdownSubparser markdownSubparser;
 
 struct sMarkdownSubparser {
 	subparser subparser;
+	/* ```something
+	   ---^ The rest of string is passed as LANGMARKER.
+
+	   A sub parser analyses LANGMARKER.
+	   If the sub parser can extract a name of language from LANGMARKER,
+	   the parser puts the name to LANGNAME, and returns true.
+	   If not, return false.
+
+	   e.g.
+
+	   ```{python}
+
+	   Fot this input, ctags pases "{python}" as LANGMARKER. */
 	bool (* extractLanguageForCodeBlock) (markdownSubparser *s,
 										  const char *langMarker,
 										  vString *langName);
+
+	/* ```something
+	   ...
+	   <code block>
+	   ...
+	   ```
+
+	   ctags passes each LINE in the code block to the sub parser
+	   that returns true when ctags calls extractLanguageForCodeBlock()
+	   with the sub parser. */
+	void (* notifyCodeBlockLine) (markdownSubparser *s,
+								  const unsigned char *line);
+
+	/* ```something
+	   ...
+	   codeblock
+	   ...
+	   ```
+	   ^ The end of code block.
+
+	   ctags notifies the sub parser that the end of code block
+	   is found.
+	*/
+	void (* notifyEndOfCodeBlock) (markdownSubparser *s);
 };
 
 #endif