fenced_code: Optimize FENCE_RE to fix cubic worst-case complexity.

Signed-off-by: Anders Kaseorg <anders@zulip.com>

zerver/lib/email_notifications.py

@@ -165,7 +165,7 @@ def fix_spoilers_in_text(content: str, language: str) -> str:
         m = FENCE_RE.match(line)
         if m:
             fence = m.group("fence")
-            lang = m.group("lang")
+            lang: Optional[str] = m.group("lang")
             if lang == "spoiler":
                 open_fence = fence
                 output.append(line)

zerver/lib/markdown/__init__.py

@@ -1747,12 +1747,13 @@ class MarkdownListPreprocessor(markdown.preprocessors.Preprocessor):
             m = FENCE_RE.match(lines[i])
             if m:
                 fence_str = m.group("fence")
-                is_code = not m.group("lang") in ("quote", "quoted")
+                lang: Optional[str] = m.group("lang")
+                is_code = lang not in ("quote", "quoted")
                 has_open_fences = not len(open_fences) == 0
                 matches_last_fence = (
                     fence_str == open_fences[-1].fence_str if has_open_fences else False
                 )
-                closes_last_fence = not m.group("lang") and matches_last_fence
+                closes_last_fence = not lang and matches_last_fence
 
                 if closes_last_fence:
                     open_fences.pop()

zerver/lib/markdown/fenced_code.py

@@ -76,7 +76,7 @@ Dependencies:
 
 """
 import re
-from typing import Any, Iterable, List, Mapping, MutableSequence, Optional, Sequence
+from typing import Any, Callable, Dict, Iterable, List, Mapping, MutableSequence, Optional, Sequence
 
 import lxml.html
 from django.utils.html import escape
@@ -100,21 +100,21 @@ FENCE_RE = re.compile(
 
     [ ]* # spaces
 
-    (
+    (?:
+        # language, like ".py" or "{javascript}"
         \{?\.?
         (?P<lang>
-            [a-zA-Z0-9_+-./#]*
+            [a-zA-Z0-9_+-./#]+
         ) # "py" or "javascript"
-        \}?
-    ) # language, like ".py" or "{javascript}"
-    [ ]* # spaces
-    (
-        \{?\.?
+
+        [ ]* # spaces
+
+        # header for features that use fenced block header syntax (like spoilers)
         (?P<header>
-            [^~`]*
-        )
+            [^ ~`][^~`]*
+        )?
         \}?
-    ) # header for features that use fenced block header syntax (like spoilers)
+    )?
     $
     """,
     re.VERBOSE,
@@ -139,7 +139,7 @@ Missing required -X argument in curl command:
                 raise MarkdownRenderingException(error_msg.format(command=line.strip()))
 
 
-CODE_VALIDATORS = {
+CODE_VALIDATORS: Dict[Optional[str], Callable[[List[str]], None]] = {
     "curl": validate_curl_content,
 }
 
@@ -177,12 +177,13 @@ def generic_handler(
     processor: Any,
     output: MutableSequence[str],
     fence: str,
-    lang: str,
-    header: str,
+    lang: Optional[str],
+    header: Optional[str],
     run_content_validators: bool = False,
     default_language: Optional[str] = None,
 ) -> BaseHandler:
-    lang = lang.lower()
+    if lang is not None:
+        lang = lang.lower()
     if lang in ("quote", "quoted"):
         return QuoteHandler(processor, output, fence, default_language)
     elif lang == "math":
@@ -203,8 +204,8 @@ def check_for_new_fence(
     m = FENCE_RE.match(line)
     if m:
         fence = m.group("fence")
-        lang = m.group("lang")
-        header = m.group("header")
+        lang: Optional[str] = m.group("lang")
+        header: Optional[str] = m.group("header")
         if not lang and default_language:
             lang = default_language
         handler = generic_handler(
@@ -243,7 +244,7 @@ class CodeHandler(BaseHandler):
         processor: Any,
         output: MutableSequence[str],
         fence: str,
-        lang: str,
+        lang: Optional[str],
         run_content_validators: bool = False,
     ) -> None:
         self.processor = processor
@@ -310,7 +311,11 @@ class QuoteHandler(BaseHandler):
 
 class SpoilerHandler(BaseHandler):
     def __init__(
-        self, processor: Any, output: MutableSequence[str], fence: str, spoiler_header: str
+        self,
+        processor: Any,
+        output: MutableSequence[str],
+        fence: str,
+        spoiler_header: Optional[str],
     ) -> None:
         self.processor = processor
         self.output = output
@@ -407,7 +412,7 @@ class FencedBlockPreprocessor(Preprocessor):
             output.append("")
         return output
 
-    def format_code(self, lang: str, text: str) -> str:
+    def format_code(self, lang: Optional[str], text: str) -> str:
         if lang:
             langclass = LANG_TAG.format(lang)
         else:
@@ -474,14 +479,15 @@ class FencedBlockPreprocessor(Preprocessor):
             quoted_paragraphs.append("\n".join("> " + line for line in lines))
         return "\n".join(quoted_paragraphs)
 
-    def format_spoiler(self, header: str, text: str) -> str:
+    def format_spoiler(self, header: Optional[str], text: str) -> str:
         output = []
         header_div_open_html = '<div class="spoiler-block"><div class="spoiler-header">'
         end_header_start_content_html = '</div><div class="spoiler-content" aria-hidden="true">'
         footer_html = "</div></div>"
 
         output.append(self.placeholder(header_div_open_html))
-        output.append(header)
+        if header is not None:
+            output.append(header)
         output.append(self.placeholder(end_header_start_content_html))
         output.append(text)
         output.append(self.placeholder(footer_html))

zerver/tests/test_markdown.py

@@ -69,8 +69,8 @@ from zerver.models import (
 class SimulatedFencedBlockPreprocessor(FencedBlockPreprocessor):
     # Simulate code formatting.
 
-    def format_code(self, lang: str, code: str) -> str:
-        return lang + ":" + code
+    def format_code(self, lang: Optional[str], code: str) -> str:
+        return (lang or "") + ":" + code
 
     def placeholder(self, s: str) -> str:
         return "**" + s.strip("\n") + "**"