adjust docs processor

Author: joshuadavidthomas

docs/processor.py

@@ -13,16 +13,21 @@
 
 import logging
 import re
+from dataclasses import dataclass
 from difflib import Differ
 from functools import reduce
 from itertools import islice
 from pathlib import Path
 from typing import Callable
+from typing import Dict
+from typing import List
+from typing import NamedTuple
 
 from rich.console import Console
 from rich.logging import RichHandler
 from rich.panel import Panel
 from rich.progress import track
+from rich.rule import Rule
 
 console = Console()
 logging.basicConfig(
@@ -66,103 +71,203 @@ def write_file(path: Path, content: str) -> bool:
         return False
 
 
-def preview_changes(original: str, processed: str, context_lines: int = 2) -> None:
-    """Show a preview of the changes made."""
-    console.print("\n[yellow]Preview of changes:[/yellow]")
+@dataclass
+class DiffStats:
+    original_lines: int
+    processed_lines: int
+    difference: int
 
-    # Basic statistics
-    orig_lines = original.count("\n")
-    proc_lines = processed.count("\n")
-    diff_lines = proc_lines - orig_lines
 
-    stats_panel = Panel(
-        f"Original lines: {orig_lines}\n"
-        f"Processed lines: {proc_lines}\n"
-        f"Difference: {diff_lines:+d} lines",
-        title="Statistics",
-        border_style="blue",
+class DiffLine(NamedTuple):
+    orig_line_no: int
+    proc_line_no: int
+    change_type: str
+    content: str
+
+
+def calculate_stats(original: str, processed: str) -> DiffStats:
+    return DiffStats(
+        original_lines=original.count("\n"),
+        processed_lines=processed.count("\n"),
+        difference=processed.count("\n") - original.count("\n"),
     )
-    console.print(stats_panel)
 
-    # Create diff
-    differ = Differ()
-    diff = list(differ.compare(original.splitlines(), processed.splitlines()))
 
-    # Find changed line groups with context
+def create_diff_lines(diff_output: List[str]) -> List[DiffLine]:
+    """Convert raw diff output into structured DiffLine objects with line numbers."""
+    diff_lines = []
+    orig_line_no = proc_line_no = 0
+
+    for line in diff_output:
+        if line.startswith("? "):  # Skip hint lines
+            continue
+
+        change_type = line[0:2]
+        content = line[2:]
+
+        current_orig = orig_line_no if change_type in ("  ", "- ") else 0
+        current_proc = proc_line_no if change_type in ("  ", "+ ") else 0
+
+        diff_lines.append(DiffLine(current_orig, current_proc, change_type, content))
+
+        # Update line numbers
+        if change_type == "  ":
+            orig_line_no += 1
+            proc_line_no += 1
+        elif change_type == "- ":
+            orig_line_no += 1
+        elif change_type == "+ ":
+            proc_line_no += 1
+
+    return diff_lines
+
+
+def group_changes(
+    diff_lines: List[DiffLine], context_lines: int = 5
+) -> List[List[DiffLine]]:
+    """Group changes with their context lines."""
     changes = []
     current_group = []
     in_change = False
-    last_change_line = -1
+    last_change_idx = -1
 
-    for i, line in enumerate(diff):
-        if line.startswith("? "):  # Skip hint lines
-            continue
+    for i, line in enumerate(diff_lines):
+        is_change = line.change_type in ("- ", "+ ")
 
-        is_change = line.startswith(("- ", "+ "))
         if is_change:
-            if not in_change:  # Start of a new change group
-                start = max(0, i - context_lines)
-                # If we're close to previous group, connect them
-                if start <= last_change_line + context_lines:
-                    start = last_change_line + 1
+            if not in_change:
+                # Start of a new change group
+                start_idx = max(0, i - context_lines)
+
+                # Connect nearby groups or start new group
+                if start_idx <= last_change_idx + context_lines:
+                    start_idx = last_change_idx + 1
                 else:
                     if current_group:
                         changes.append(current_group)
                     current_group = []
-                    # Add previous context
-                    current_group.extend(
-                        l for l in diff[start:i] if not l.startswith("? ")
-                    )
+                    # Add leading context
+                    current_group.extend(diff_lines[start_idx:i])
+
             current_group.append(line)
             in_change = True
-            last_change_line = i
-        else:
-            if in_change:
-                # Add following context
-                following_context = list(
-                    islice(
-                        (l for l in diff[i:] if not l.startswith("? ")), context_lines
-                    )
+            last_change_idx = i
+
+        elif in_change:
+            # Add trailing context
+            following_context = list(
+                islice(
+                    (l for l in diff_lines[i:] if l.change_type == "  "), context_lines
                 )
-                if following_context:  # Only extend if we have context to add
-                    current_group.extend(following_context)
-                in_change = False
+            )
+            current_group.extend(following_context)
+            in_change = False
 
     if current_group:
         changes.append(current_group)
 
-    # Format and display the changes
-    formatted_output = []
-    for i, group in enumerate(changes):
-        if i > 0:
-            formatted_output.append(
-                "[bright_black]⋮ skipped unchanged content ⋮[/bright_black]"
-            )
+    return changes
 
-        # Track the last line to avoid duplicates
-        last_line = None
 
-        for line in group:
-            # Skip if this line is the same as the last one
-            if line == last_line:
-                continue
+def get_changes(
+    original: str, processed: str
+) -> tuple[Dict[str, int], List[List[DiffLine]]]:
+    """Generate diff information and statistics."""
+    # Get basic statistics
+    stats = calculate_stats(original, processed)
+
+    # Create and process diff
+    differ = Differ()
+    diff_output = list(differ.compare(original.splitlines(), processed.splitlines()))
+    diff_lines = create_diff_lines(diff_output)
+    grouped_changes = group_changes(diff_lines)
+
+    return vars(stats), grouped_changes
+
+
+@dataclass
+class ChangeGroup:
+    orig_no: int
+    proc_no: int
+    change_type: str
+    content: str
+
+    def format_line_info(self) -> str:
+        """Format the line numbers and separator based on change type."""
+        if self.change_type == "  ":
+            return f"[bright_black]{self.orig_no:4d}│{self.proc_no:4d}│[/bright_black]"
+        elif self.change_type == "- ":
+            return f"[bright_black]{self.orig_no:4d}│    │[/bright_black]"
+        else:  # "+" case
+            return f"[bright_black]    │{self.proc_no:4d}│[/bright_black]"
+
+    def format_content(self) -> str:
+        """Format the content based on change type."""
+        if self.change_type == "  ":
+            return f"[white]{self.content}[/white]"
+        elif self.change_type == "- ":
+            return f"[red]- {self.content}[/red]"
+        else:  # "+" case
+            return f"[green]+ {self.content}[/green]"
+
+
+def create_stats_panel(stats: dict) -> Panel:
+    """Create a formatted statistics panel."""
+    stats_content = (
+        f"Original lines: {stats['original_lines']}\n"
+        f"Processed lines: {stats['processed_lines']}\n"
+        f"Difference: {stats['difference']:+d} lines"
+    )
+    return Panel(
+        stats_content,
+        title="Statistics",
+        border_style="blue",
+    )
+
 
-            if line.startswith("  "):  # unchanged
-                formatted_output.append(f"[white]{line[2:]}[/white]")
-            elif line.startswith("- "):  # removed
-                formatted_output.append(f"[red]━ {line[2:]}[/red]")
-            elif line.startswith("+ "):  # added
-                formatted_output.append(f"[green]+ {line[2:]}[/green]")
+def create_separator(prev_group: List[tuple], current_group: List[tuple]) -> Rule:
+    """Create a separator between change groups with skip line information."""
+    if not prev_group:
+        return None
 
-            last_line = line
+    last_orig = max(l[0] for l in prev_group if l[0] > 0)
+    next_orig = min(l[0] for l in current_group if l[0] > 0)
+    skipped_lines = next_orig - last_orig - 1
 
-    console.print(
-        Panel(
-            "\n".join(formatted_output),
-            title="Changes with Context",
-            border_style="yellow",
+    if skipped_lines > 0:
+        return Rule(
+            f" {skipped_lines} lines skipped ",
+            style="bright_black",
+            characters="⋮",
         )
-    )
+    return Rule(style="bright_black", characters="⋮")
+
+
+def print_change_group(group: List[tuple]) -> None:
+    """Print a group of changes with formatting."""
+    for orig_no, proc_no, change_type, content in group:
+        change = ChangeGroup(orig_no, proc_no, change_type, content)
+        line_info = change.format_line_info()
+        content_formatted = change.format_content()
+        console.print(f"{line_info} {content_formatted}")
+
+
+def preview_changes(original: str, processed: str) -> None:
+    """Show a preview of the changes made."""
+    console.print("\n[yellow]Preview of changes:[/yellow]")
+
+    # Get diff information and show statistics
+    stats, changes = get_changes(original, processed)
+    console.print(create_stats_panel(stats))
+
+    # Print changes with separators between groups
+    for i, group in enumerate(changes):
+        if i > 0:
+            separator = create_separator(changes[i - 1], group)
+            if separator:
+                console.print(separator)
+
+        print_change_group(group)
 
 
 def process_readme(