try new approach

app.py

@@ -1,61 +1,75 @@
 import re
 import gradio as gr
 
-def insert_points(text):
-    # --- Step 0: Normalize line endings ---
-    # Replace Windows (\r\n) and old Mac (\r) line endings with Unix (\n)
-    # This simplifies the regex patterns below.
-    text = text.replace('\r\n', '\n').replace('\r', '\n')
 
+def insert_points(text):
     counter = 1
 
-    # Replacer function for the initial insertion pass
-    def replacer(match):
+    # -----------------------------
+    # 1) COLLAPSE MULTIPLE BLANK LINES INTO ONE
+    #    e.g. 2+ newlines in a row => single newline
+    # -----------------------------
+    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
+    # -----------------------------
+    def newline_replacer(_match):
+        nonlocal counter
+        out = f"[INSERT_POINT_{counter:03d}]\n"
+        counter += 1
+        return out
+
+    text = re.sub(r'\n', newline_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.
+    # -----------------------------
+    def heading_replacer(m):
         nonlocal counter
-        tag = f"[INSERT_POINT_{counter:03d}]"
+        # group(1) is the string of hashes, like "##" or "#####"
+        hashes = m.group(1)
+        replacement = f"[INSERT_POINT_{counter:03d}]{hashes}"
         counter += 1
-        # Return the tag followed by the original matched text
-        # group(0) will be one or more newlines OR one or more hashes
-        return tag + match.group(0)
-
-    # --- Step 1: Insert tags before hash sequences and consolidated newline sequences ---
-    # Pattern matches:
-    # 1) '\n+' : One or more consecutive (now normalized) newline characters. Collapses blank lines.
-    # 2) '\#+' : One or more consecutive '#' characters.
-    pattern_initial = r'(\n+|\#+)'
-    processed_text = re.sub(pattern_initial, replacer, text)
-
-    # --- Step 2: Clean up potential heading splits ---
-    # This fixes cases where Step 1 resulted in "[TAG_A]\n[TAG_B]###"
-    # It looks for TAG_A, followed by whitespace (\s* includes the \n),
-    # followed by TAG_B, followed by hashes (#+).
-    cleanup_pattern = r'(\[INSERT_POINT_\d{3}\])\s*(\[INSERT_POINT_\d{3}\])(\#+)'
-    # Replaces the whole match with just TAG_B (\2) and the hashes (\3).
-    processed_text = re.sub(cleanup_pattern, r'\2\3', processed_text)
-
-    return processed_text
-
-# --- Gradio Interface Code (with Copy Button) ---
+        return replacement
+
+    # 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)
+
+    return text
+
+
+# -----------------------------
+#  GRADIO INTERFACE
+# -----------------------------
 demo = gr.Interface(
     fn=insert_points,
     inputs=gr.Textbox(
         lines=10,
-        placeholder="Paste your text here...\n(Line endings will be normalized)",
+        placeholder="Paste your text here...",
         label="Your Input Text"
     ),
-    outputs=gr.Textbox(
-        label="Processed Text with Tags",
-        show_copy_button=True # <--- Added copy button here
-    ),
-    title="Insert Point Tagger v3", # Optional: update title
+    outputs=gr.Textbox(label="Processed Text with Tags"),
+    title="Insert Point Tagger",
     description=(
-        "Paste text to add '[INSERT_POINT_###]' tags:\n"
-        "1) **Before** each '#' sequence (headings).\n"
-        "2) **Before** each sequence of one or more newlines (one tag per line break/blank line).\n"
-        "Normalizes line endings and cleans up heading tag spacing."
+        "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."
     ),
-    allow_flagging='never'
 )
 
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch()