new approach

app.py

@@ -12,40 +12,48 @@ def insert_points(text):
     text = re.sub(r'\n\s*\n+', '\n', text)
 
     # -----------------------------
-    # 2) INSERT A TAG BEFORE EACH REMAINING NEWLINE
-    #    So every line (including blank ones) is prefixed by exactly one tag
+    # 2) PROCESS HEADINGS FIRST (before dealing with newlines)
+    #    This ensures headings get properly tagged
     # -----------------------------
-    def newline_replacer(_match):
+    def heading_replacer(m):
         nonlocal counter
-        out = f"[INSERT_POINT_{counter:03d}]\n"
+        # Get the full heading match
+        full_heading = m.group(0)
+        # Extract just the hash symbols
+        hashes = m.group(1)
+        # Insert a newline before and after the tag
+        replacement = f"\n[INSERT_POINT_{counter:03d}]\n{full_heading}"
         counter += 1
-        return out
+        return replacement
 
-    text = re.sub(r'\n', newline_replacer, text)
+    # Match headings that are at the start of a line (preceded by newline or start of string)
+    # We capture the full heading pattern including content after the hashes
+    text = re.sub(r'(^|\n)(\#{1,6}[ \t].+?)(?=\n|$)', heading_replacer, text)
 
     # -----------------------------
-    # 3) INSERT A TAG BEFORE EACH HEADING IF IT'S NOT
-    #    (A) ALREADY TAGGED, OR
-    #    (B) IMMEDIATELY AFTER A NEWLINE (WHICH WOULD HAVE A TAG)
-    #
-    # We detect # or ## or ### up to 6 in a row. If there's no
-    # [INSERT_POINT_\d{3}] or newline right before, we insert a new tag.
+    # 3) INSERT A TAG BEFORE EACH PARAGRAPH BREAK
+    #    These are newlines that aren't already tagged
     # -----------------------------
-    def heading_replacer(m):
+    def newline_replacer(_match):
         nonlocal counter
-        # group(1) is the string of hashes, like "##" or "#####"
-        hashes = m.group(1)
-        replacement = f"[INSERT_POINT_{counter:03d}]{hashes}"
+        # Insert tag on its own line
+        out = f"\n[INSERT_POINT_{counter:03d}]\n"
         counter += 1
-        return replacement
+        return out
 
-    # Negative lookbehind means:
-    #   - Not preceded by [INSERT_POINT_###]
-    #   - Not preceded by a newline
-    # So if the line starts with "##" and we've just inserted a tag for the line break,
-    # we won't insert a second tag for the heading.
-    pattern = r'(?<!\[INSERT_POINT_\d{3}\])(?<!\n)(\#{1,6})'
-    text = re.sub(pattern, heading_replacer, text)
+    # Replace newlines that aren't already preceded by a tag
+    text = re.sub(r'(?<!\[INSERT_POINT_\d{3}]\n)\n', newline_replacer, text)
+
+    # -----------------------------
+    # 4) CLEAN UP: Remove any extra newlines before tags
+    #    (Sometimes we might have created double newlines)
+    # -----------------------------
+    text = re.sub(r'\n\n(\[INSERT_POINT_\d{3}])', r'\n\1', text)
+
+    # If text doesn't start with a newline + tag, add one at the beginning
+    if not text.startswith('\n[INSERT_POINT_'):
+        text = f"\n[INSERT_POINT_{counter:03d}]\n" + text
+        counter += 1
 
     return text
 
@@ -64,12 +72,11 @@ demo = gr.Interface(
     title="Insert Point Tagger",
     description=(
         "1) Collapses multiple blank lines into one.\n"
-        "2) Inserts a single [INSERT_POINT_XXX] tag before each remaining newline.\n"
-        "3) Inserts a single [INSERT_POINT_XXX] tag before any '#' heading, unless\n"
-        "   there's already a tag from the preceding newline.\n"
-        "This avoids double tags while keeping heading text on the same line."
+        "2) Inserts [INSERT_POINT_XXX] tags on their own lines.\n"
+        "3) Properly tags headings and paragraph breaks.\n"
+        "4) Maintains sequential numbering for all tags."
     ),
 )
 
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch()