Attempting something new

app/email_new_converter.py

@@ -575,14 +575,16 @@ class EmailNewConverter:
     def _fix_section_header_tables(self, soup: BeautifulSoup) -> None:
         """Restore section header tables to the simplest, most compatible 2-column layout.
         
-        Section headers like "Hot Polymarkets" and "Volume" are in a table with two <td>s.
-        Desktop clients handle many layout hints, but mobile Gmail can mis-render when we add
-        `table-layout:fixed` or explicit widths/min-widths. The most reliable pattern is:
-        - header table: width="100%"
-        - second <td>: align="right" (and optional text-align:right), no forced widths
+        Based on analysis of working emails (original_msg.eml), the most reliable pattern is:
+        - Simple 2-column table with width="100%"
+        - Second cell has align="right" attribute
+        - NO table-layout:fixed
+        - NO explicit widths on cells
+        - NO spacer cells or nested tables
+        
+        This matches the original template structure that worked on mobile.
         """
         # Find all section header tables - they're inside .wrap_h divs and have two <td>s
-        # One with the section title, one with align="right" for the label
         wrap_h_divs = soup.find_all("div", class_="wrap_h")
         
         for wrap_h in wrap_h_divs:
@@ -591,17 +593,14 @@ class EmailNewConverter:
             for table in tables:
                 tds = table.find_all("td")
                 if len(tds) == 2:
-                    # This is likely a section header table
-                    # First td: section title (left)
-                    # Second td: label like "Volume" (right)
-                    
+                    # This is a section header table
                     first_td = tds[0]
                     second_td = tds[1]
                     
-                    # Ensure the header table is full width using HTML attribute.
+                    # Ensure the header table is full width using HTML attribute
                     table["width"] = "100%"
 
-                    # Remove table-layout constraints (common cause of mobile misalignment).
+                    # Remove table-layout constraints (causes mobile misalignment)
                     table_style = table.get("style", "")
                     if table_style and "table-layout" in table_style.lower():
                         table["style"] = re.sub(
@@ -611,7 +610,8 @@ class EmailNewConverter:
                             flags=re.IGNORECASE,
                         ).strip("; ")
 
-                    # Remove width/min-width constraints on both header cells (let table algorithm work).
+                    # Remove width/min-width/white-space constraints on both cells
+                    # Let the table algorithm handle spacing naturally
                     for td in (first_td, second_td):
                         if td.get("width"):
                             del td["width"]
@@ -619,71 +619,22 @@ class EmailNewConverter:
                         if td_style:
                             td_style = re.sub(r'width\s*:\s*[^;]+;?\s*', '', td_style, flags=re.IGNORECASE)
                             td_style = re.sub(r'min-width\s*:\s*[^;]+;?\s*', '', td_style, flags=re.IGNORECASE)
-                            # We previously added white-space:nowrap; remove it to match legacy behavior.
                             td_style = re.sub(r'white-space\s*:\s*[^;]+;?\s*', '', td_style, flags=re.IGNORECASE)
+                            td_style = re.sub(r'padding-right\s*:\s*[^;]+;?\s*', '', td_style, flags=re.IGNORECASE)
                             td["style"] = td_style.strip("; ")
 
-                    # Use 3-column layout with spacer cell for reliable right-alignment on mobile.
-                    # This is more reliable than 2-column layouts on mobile email clients.
-                    # Structure: [Title] [Spacer] [Subtitle]
-                    
-                    # Get the current content of both cells
-                    first_td_contents = list(first_td.children)
+                    # Ensure second cell has align="right" attribute
+                    second_td["align"] = "right"
                     
-                    # Extract content from second_td, handling nested tables if present
+                    # Add text-align:right as fallback in inline style
                     second_td_style = second_td.get("style", "")
-                    nested_table = second_td.find("table")
-                    if nested_table:
-                        # If there's a nested table, get content from its td
-                        nested_td = nested_table.find("td")
-                        if nested_td:
-                            second_td_contents = list(nested_td.children)
-                            # Use nested_td style if available, otherwise use second_td style
-                            nested_style = nested_td.get("style", "")
-                            if nested_style:
-                                second_td_style = nested_style
-                        else:
-                            second_td_contents = list(second_td.children)
-                    else:
-                        second_td_contents = list(second_td.children)
-                    
-                    # Get the row containing these cells
-                    tr = first_td.find_parent("tr")
-                    if not tr:
-                        continue
-                    
-                    # Clear the row
-                    tr.clear()
-                    
-                    # First cell: title (left)
-                    new_first_td = soup.new_tag("td")
-                    first_td_style = first_td.get("style", "")
-                    if first_td_style:
-                        # Remove padding-right if it exists (we'll use spacer instead)
-                        first_td_style = re.sub(r'padding-right\s*:\s*[^;]+;?\s*', '', first_td_style, flags=re.IGNORECASE)
-                        new_first_td["style"] = first_td_style.strip("; ")
-                    for content in first_td_contents:
-                        new_first_td.append(content)
-                    tr.append(new_first_td)
-                    
-                    # Second cell: spacer (flexible width, pushes content apart)
-                    # No width specified - table algorithm will give it remaining space
-                    spacer_td = soup.new_tag("td")
-                    spacer_td["style"] = "font-size:1px;line-height:1px;"
-                    spacer_td.append("\u00A0")  # Non-breaking space to prevent collapse
-                    tr.append(spacer_td)
-                    
-                    # Third cell: subtitle (right)
-                    new_second_td = soup.new_tag("td")
-                    if second_td_style:
-                        new_second_td["style"] = second_td_style
-                    new_second_td["align"] = "right"
-                    new_second_td["valign"] = "bottom"
-                    for content in second_td_contents:
-                        new_second_td.append(content)
-                    tr.append(new_second_td)
+                    if "text-align" not in second_td_style.lower():
+                        if second_td_style and not second_td_style.endswith(";"):
+                            second_td_style += ";"
+                        second_td_style += "text-align:right;"
+                        second_td["style"] = second_td_style.strip("; ")
 
-                    logger.info("Fixed section header table layout using nested table for mobile compatibility")
+                    logger.info("Fixed section header table to simple 2-column layout for mobile compatibility")
 
     def _normalize_style_attributes(self, soup: BeautifulSoup) -> None:
         """Normalize style attributes for consistent email client parsing.

app/post_process.py

@@ -354,14 +354,16 @@ def fix_cloudinary_image_transformations(html_content: str) -> str:
 
 
 def fix_section_header_alignment(html_content: str) -> str:
-    """Restore section header tables to the simplest, most compatible layout.
+    """Restore section header tables to the simplest, most compatible 2-column layout.
     
-    Insight from prior working emails: the most reliable pattern for mobile clients is the
-    default table algorithm (no table-layout:fixed, no forced widths/min-widths on the header cells).
+    Based on analysis of working emails (original_msg.eml), the most reliable pattern is:
+    - Simple 2-column table with width="100%"
+    - Second cell has align="right" attribute
+    - NO table-layout:fixed
+    - NO explicit widths on cells
+    - NO spacer cells or nested tables
     
-    This function strips those constraints while preserving the core semantics:
-    - header table remains width="100%"
-    - second cell remains align="right" (with optional text-align:right fallback)
+    This matches the original template structure that worked on mobile.
     """
     from bs4 import BeautifulSoup
     
@@ -372,91 +374,64 @@ def fix_section_header_alignment(html_content: str) -> str:
     for wrap_h in wrap_h_divs:
         tables = wrap_h.find_all("table", role="presentation")
         for table in tables:
-            # Identify candidate header table: exactly one row with 2 or 3 tds.
+            # Identify candidate header table: exactly one row with 2 tds
             tr = table.find("tr")
             if not tr:
                 continue
             tds = tr.find_all("td", recursive=False)
             
-            # If already 3 columns, skip (already converted)
-            if len(tds) == 3:
-                continue
-            
-            # Must have exactly 2 columns to convert
+            # Must have exactly 2 columns
             if len(tds) != 2:
                 continue
             
             first_td, second_td = tds[0], tds[1]
             
-            # Keep header table full width via attribute (legacy behavior).
+            # Ensure header table is full width via attribute
             if table.get("width") != "100%":
                 table["width"] = "100%"
                 modified = True
             
-            # Strip table-layout from table style if present.
+            # Strip table-layout from table style if present
             table_style = table.get("style", "")
             if table_style and "table-layout" in table_style.lower():
                 new_table_style = re.sub(r"table-layout\s*:\s*[^;]+;?\s*", "", table_style, flags=re.IGNORECASE).strip("; ")
                 table["style"] = new_table_style
                 modified = True
             
-            # Convert 2-column to 3-column layout with spacer cell
-            # Get the current content of both cells
-            first_td_contents = list(first_td.children)
-            
-            # Get styles first
-            first_td_style = first_td.get("style", "")
-            second_td_style = second_td.get("style", "")
-            
-            # Extract content from second_td, handling nested tables if present
-            nested_table = second_td.find("table")
-            if nested_table:
-                # If there's a nested table, get content from its td
-                nested_td = nested_table.find("td")
-                if nested_td:
-                    second_td_contents = list(nested_td.children)
-                    # Use nested_td style if available, otherwise use second_td style
-                    nested_style = nested_td.get("style", "")
-                    if nested_style:
-                        second_td_style = nested_style
-                else:
-                    second_td_contents = list(second_td.children)
-            else:
-                second_td_contents = list(second_td.children)
-            
-            # Clear the row
-            tr.clear()
-            
-            # First cell: title (left)
-            new_first_td = soup.new_tag("td")
-            if first_td_style:
-                # Remove padding-right if it exists (we'll use spacer instead)
-                first_td_style = re.sub(r"padding-right\s*:\s*[^;]+;?\s*", "", first_td_style, flags=re.IGNORECASE)
-                new_first_td["style"] = first_td_style.strip("; ")
-            for content in first_td_contents:
-                new_first_td.append(content)
-            tr.append(new_first_td)
-            
-            # Second cell: spacer (flexible width, pushes content apart)
-            spacer_td = soup.new_tag("td")
-            spacer_td["style"] = "font-size:1px;line-height:1px;"
-            spacer_td.append("\u00A0")  # Non-breaking space to prevent collapse
-            tr.append(spacer_td)
+            # Remove width/min-width/white-space/padding-right constraints on both cells
+            for td in (first_td, second_td):
+                if td.get("width"):
+                    del td["width"]
+                    modified = True
+                
+                td_style = td.get("style", "")
+                if td_style:
+                    original_style = td_style
+                    td_style = re.sub(r"width\s*:\s*[^;]+;?\s*", "", td_style, flags=re.IGNORECASE)
+                    td_style = re.sub(r"min-width\s*:\s*[^;]+;?\s*", "", td_style, flags=re.IGNORECASE)
+                    td_style = re.sub(r"white-space\s*:\s*[^;]+;?\s*", "", td_style, flags=re.IGNORECASE)
+                    td_style = re.sub(r"padding-right\s*:\s*[^;]+;?\s*", "", td_style, flags=re.IGNORECASE)
+                    td_style = td_style.strip("; ")
+                    if td_style != original_style:
+                        td["style"] = td_style
+                        modified = True
             
-            # Third cell: subtitle (right)
-            new_second_td = soup.new_tag("td")
-            if second_td_style:
-                new_second_td["style"] = second_td_style
-            new_second_td["align"] = "right"
-            new_second_td["valign"] = "bottom"
-            for content in second_td_contents:
-                new_second_td.append(content)
-            tr.append(new_second_td)
+            # Ensure second cell has align="right" attribute
+            if second_td.get("align") != "right":
+                second_td["align"] = "right"
+                modified = True
             
-            modified = True
+            # Add text-align:right as fallback in inline style
+            second_td_style = second_td.get("style", "")
+            if "text-align" not in second_td_style.lower():
+                if second_td_style and not second_td_style.endswith(";"):
+                    second_td_style += ";"
+                second_td_style += "text-align:right;"
+                second_td["style"] = second_td_style.strip("; ")
+                modified = True
     
     if modified:
-        print("  Restored section header table layout for mobile")
+        print("  Restored section header tables to simple 2-column layout for mobile")
         return str(soup)
     
     print("  No section header tables found that need alignment fixes")