From 5e385d0744764cc22c66af898ed11859df0b6225 Mon Sep 17 00:00:00 2001
From: Quentin Carbonneaux
Date: Thu, 7 May 2020 12:26:37 +0200
Subject: Cleanup formatting in .. coqtop:: directives

---
 doc/tools/coqrst/coqdoc/main.py | 22 +++++-----------------
 doc/tools/coqrst/coqdomain.py   | 14 +++++++++++---
 2 files changed, 16 insertions(+), 20 deletions(-)

diff --git a/doc/tools/coqrst/coqdoc/main.py b/doc/tools/coqrst/coqdoc/main.py
index a3fc069e6c..9a12d5e6f8 100644
--- a/doc/tools/coqrst/coqdoc/main.py
+++ b/doc/tools/coqrst/coqdoc/main.py
@@ -48,28 +48,16 @@ def coqdoc(coq_code, coqdoc_bin=None):
     finally:
         os.remove(filename)
 
-def is_whitespace_string(elem):
-    return isinstance(elem, NavigableString) and elem.strip() == ""
-
-def strip_soup(soup, pred):
-    """Strip elements matching pred from front and tail of soup."""
-    while soup.contents and pred(soup.contents[-1]):
-        soup.contents.pop()
-
-    skip = 0
-    for elem in soup.contents:
-        if not pred(elem):
-            break
-        skip += 1
-
-    soup.contents[:] = soup.contents[skip:]
-
 def lex(source):
     """Convert source into a stream of (css_classes, token_string)."""
     coqdoc_output = coqdoc(source)
     soup = BeautifulSoup(coqdoc_output, "html.parser")
     root = soup.find(class_='code')
-    strip_soup(root, is_whitespace_string)
+    if root.children:
+        # strip the leading '\n'
+        first = next(root.children)
+        if isinstance(first, NavigableString) and first.string[0] == '\n':
+            first.string.replace_with(first.string[1:])
     for elem in root.children:
         if isinstance(elem, NavigableString):
             yield [], elem
diff --git a/doc/tools/coqrst/coqdomain.py b/doc/tools/coqrst/coqdomain.py
index 9d51d2198a..253fc5283f 100644
--- a/doc/tools/coqrst/coqdomain.py
+++ b/doc/tools/coqrst/coqdomain.py
@@ -905,9 +905,17 @@ class CoqtopBlocksTransform(Transform):
         return isinstance(node, nodes.Element) and 'coqtop_options' in node
 
     @staticmethod
-    def split_sentences(source):
+    def split_sentences(node):
         """Split Coq sentences in source. Could be improved."""
-        return re.split(r"(?<=(?<!\.)\.)\s+", source)
+        lines = map(lambda s: s.rstrip(), node.rawsource.splitlines())
+        out = [""]
+        for l in lines:
+            out[-1] = out[-1] + l + "\n"
+            if l.endswith("."):
+                out.append("")
+        if out[-1] == "":
+            out.pop()
+        return out
 
     @staticmethod
     def parse_options(node):
@@ -986,7 +994,7 @@ class CoqtopBlocksTransform(Transform):
             repl.sendone('Unset Coqtop Exit On Error.')
         if options['warn']:
             repl.sendone('Set Warnings "default".')
-        for sentence in self.split_sentences(node.rawsource):
+        for sentence in self.split_sentences(node):
             pairs.append((sentence, repl.sendone(sentence)))
         if options['abort']:
             repl.sendone('Abort All.')
-- 
cgit v1.2.3


From 24497500d30a6719b2eb92c3795ab067e2cdf14f Mon Sep 17 00:00:00 2001
From: Quentin Carbonneaux
Date: Fri, 8 May 2020 13:29:50 +0200
Subject: Simplify splitting

---
 doc/tools/coqrst/coqdomain.py | 20 ++++++++------------
 1 file changed, 8 insertions(+), 12 deletions(-)

diff --git a/doc/tools/coqrst/coqdomain.py b/doc/tools/coqrst/coqdomain.py
index 253fc5283f..df11960403 100644
--- a/doc/tools/coqrst/coqdomain.py
+++ b/doc/tools/coqrst/coqdomain.py
@@ -905,17 +905,13 @@ class CoqtopBlocksTransform(Transform):
         return isinstance(node, nodes.Element) and 'coqtop_options' in node
 
     @staticmethod
-    def split_sentences(node):
-        """Split Coq sentences in source. Could be improved."""
-        lines = map(lambda s: s.rstrip(), node.rawsource.splitlines())
-        out = [""]
-        for l in lines:
-            out[-1] = out[-1] + l + "\n"
-            if l.endswith("."):
-                out.append("")
-        if out[-1] == "":
-            out.pop()
-        return out
+    def split_lines(source):
+        """Split Coq input in chunks
+
+        A chunk is a minimal sequence of consecutive lines of the input that
+        ends with a '.'
+        """
+        return re.split(r"(?<=(?<!\.)\.)\s+\n", source)
 
     @staticmethod
     def parse_options(node):
@@ -994,7 +990,7 @@ class CoqtopBlocksTransform(Transform):
             repl.sendone('Unset Coqtop Exit On Error.')
         if options['warn']:
             repl.sendone('Set Warnings "default".')
-        for sentence in self.split_sentences(node):
+        for sentence in self.split_lines(node.rawsource):
             pairs.append((sentence, repl.sendone(sentence)))
         if options['abort']:
             repl.sendone('Abort All.')
-- 
cgit v1.2.3


From 71d0315f2559dca43a5ea733f91199965b8b7879 Mon Sep 17 00:00:00 2001
From: Quentin Carbonneaux
Date: Fri, 8 May 2020 19:12:50 +0200
Subject: Recursively look for the first string node

---
 doc/tools/coqrst/coqdoc/main.py | 16 +++++++++++-----
 1 file changed, 11 insertions(+), 5 deletions(-)

diff --git a/doc/tools/coqrst/coqdoc/main.py b/doc/tools/coqrst/coqdoc/main.py
index 9a12d5e6f8..de0d912c03 100644
--- a/doc/tools/coqrst/coqdoc/main.py
+++ b/doc/tools/coqrst/coqdoc/main.py
@@ -48,16 +48,22 @@ def coqdoc(coq_code, coqdoc_bin=None):
     finally:
         os.remove(filename)
 
+def first_string_node(node):
+    """Return the first string node, or None if does not exist"""
+    while node.children:
+        node = next(node.children)
+        if isinstance(node, NavigableString):
+            return node
+
 def lex(source):
     """Convert source into a stream of (css_classes, token_string)."""
     coqdoc_output = coqdoc(source)
     soup = BeautifulSoup(coqdoc_output, "html.parser")
     root = soup.find(class_='code')
-    if root.children:
-        # strip the leading '\n'
-        first = next(root.children)
-        if isinstance(first, NavigableString) and first.string[0] == '\n':
-            first.string.replace_with(first.string[1:])
+    # strip the leading '\n'
+    first = first_string_node(root)
+    if first and first.string[0] == '\n':
+        first.string.replace_with(first.string[1:])
     for elem in root.children:
         if isinstance(elem, NavigableString):
             yield [], elem
-- 
cgit v1.2.3