Allow multiple node return (#2713)

* Pack ParseConfig into Stream. Allow multiple nodes to be returned at once. Break WholeElement into multiple ParseNodes.

* Whoops, don't override the existing node whenever there's text.

* Make Result a generic type so I actually know what I'm returning in each spot.

* Keep escaping markdown code spans

* Properly clear the last node when I stop seeing nodes.

* Move curlifying over to a final step I do on text nodes.

* Handle *all* the places where Text gets generated, and move curlify to a Text method.

* Argh, still need to check for a word character after the apos.

* Cache the length of the stream

* Allow for nodes to *decrement* the line count as well, if they print to more lines than they occupied in the source.

* Update tests

* lint

bikeshed/InputSource.py

@@ -28,19 +28,27 @@ def lines(self) -> list[line.Line]:
         offset = 0
         for i, text in enumerate(self.rawLines, 1):
             lineNo = i + offset
-            # The early HTML parser runs before Markdown,
-            # and in some cases removes linebreaks that were present
-            # in the source. When properly invoked, it inserts
-            # a special PUA char for each of these omitted linebreaks,
-            # so I can remove them here and properly increment the
-            # line number.
+            # The early HTML parser can change how nodes print,
+            # so they occupy a different number of lines than they
+            # had in the source. Markdown parser needs to know
+            # the correct source lines, tho, so when this happens,
+            # the nodes will insert special PUA chars to indicate that.
+            # I can remove them here and properly adjust the line number.
             # Current known causes of this:
             # * line-ending -- turned into em dashes
             # * multi-line start tags
+            # * multi-line markdown code spans;
+            #     - the text loses its newlines
+            #     - the original text goes into an attribute on the start
+            #       tag now
             ilcc = constants.incrementLineCountChar
+            dlcc = constants.decrementLineCountChar
             if ilcc in text:
                 offset += text.count(ilcc)
                 text = text.replace(ilcc, "")
+            if dlcc in text:
+                offset -= text.count(dlcc)
+                text = text.replace(dlcc, "")
 
             ret.append(line.Line(lineNo, text))
 

bikeshed/constants.py

@@ -11,4 +11,5 @@
 macroStartChar = "\uebbb"
 macroEndChar = "\uebbc"
 incrementLineCountChar = "\uebbd"
+decrementLineCountChar = "\uebbf"
 bsComment = "<!--\uebbe-->"